From a4b17b86304f938994dcc99f353ebdd191e9aca4 Mon Sep 17 00:00:00 2001 From: rociojoo Date: Tue, 10 May 2022 20:45:26 -0600 Subject: [PATCH] update docs folder --- docs/analyses.html | 297 ++++++++++++++---- docs/data-collection-and-processing.html | 28 +- docs/index.html | 8 +- .../Barplots_flip_topics_taxa.png | Bin 0 -> 166047 bytes docs/r-session-information.html | 2 +- docs/references.html | 2 +- docs/search_index.json | 2 +- docs/survey-about-movement-ecology.html | 24 +- 8 files changed, 277 insertions(+), 86 deletions(-) create mode 100644 docs/posts/moveco-images/Barplots_flip_topics_taxa.png diff --git a/docs/analyses.html b/docs/analyses.html index 9c6292b..e0a2801 100644 --- a/docs/analyses.html +++ b/docs/analyses.html @@ -24,7 +24,7 @@ - + @@ -320,13 +320,6 @@

3 Data analysis

proportion of articles in that studied movement.


-
-

-Fig. 3.1. Proportion of scientific articles extracted from WoS that were about -animal or human movement for each year between 2009 and 2018. The solid line and -grey area represent the fitted local polynomial regression (estimates and error bounds, respectively). -
-


Several dimensions of the mov-eco literature were analyzed: research topics, taxonomical groups studied, components of the movement ecology framework studied, tracking devices used, software tools used, and statistical @@ -392,8 +385,10 @@

3 Data analysis

3.1 Topic analysis

-

-Fig. 3.2. Stages of topic analysis. +
+ +

Fig. 3.2. Stages of topic analysis.

+


The topics were not defined a priori. @@ -434,16 +429,20 @@

3.1.1 The model

reasonable value that would not be too large than we could not interpret them, or too small that the topics were too general.

-

-Fig. 3.3. Schematic representation of the links between words, documents and topics. +
+ +

Fig. 3.3. Schematic representation of the links between words, documents and topics. Each document is a mixture of topics. Each topic is modeled as a distribution of words. Each word comes out of one of these topics. -Source of the image: Blei, D.M. 2012. Probabilistic topic models. Communications of the ACM, 55(4), 77-84. +Source of the image: Blei, D.M. 2012. Probabilistic topic models. Communications of the ACM, 55(4), 77-84.

+


-

-Fig. 3.4. Schematic representation of the Latent Dirichlet model described above. +
+ +

Fig. 3.4. Schematic representation of the Latent Dirichlet model described above.

+


@@ -480,7 +479,7 @@

3.1.4 Model outputs

wordclouds for each topic, where the area occupied by each word was proportional to its \(\hat{\beta}\) value.

-

+

Fig. 3.5. Wordclouds of each topic based on \(\hat{\beta}\) values. Download the codes for the plot here.
@@ -490,31 +489,31 @@

3.1.4 Model outputs

download) of the 5 most associated abstracts to each topic, to aid the interpretation of the topics.

-

Based on these outputs, the topics were interpreted as: 1) Social interactions -and dispersal, 2) Movement models, 3) Habitat selection, 4) Detection and +

Based on these outputs, the topics were interpreted as: 1) Dispersal, +2) Movement models, 3) Habitat selection, 4) Detection and data, 5) Home ranges, 6) Aquatic systems, 7) Foraging in marine megafauna, 8) Biomechanics, 9) Acoustic telemetry, 10) Experimental designs, 11) Activity -budgets, 12) Avian migration, 13) Sports, 14) Human activity patterns, -15) Breeding ecology. For an extended description of these topics, see the main text of the manuscript.

-

The sum of \(\gamma\) values for each topic (\(\sum_d E(\theta_d | z_k)\) for each \(k\)) served as proxies of the +budgets, 12) Migration, 13) Sports, 14) Human activity patterns, +15) Breeding ecology. For an extended description of these topics, see the main text of the manuscript. +The sum of \(\gamma\) values for each topic (\(\sum_d E(\theta_d | z_k)\) for each \(k\)) served as proxies of the “prevalence” of the topic relative to all other topics and were used to rank them.

-
-

-Fig. 3.6. Measure of prevalence of each topic. -Download the codes for the plot here. -
-


-
-

-Fig. 3.7. Time series of the relative prevalence of each topic every year. -The code can be downloaded from here. -
-


+ + + + + + + + + + + +

A heatmap of the \(\gamma\) values also showed that most papers were evidently more associated to one topic and few were split into several topics.

-

-Fig. 3.8. Heatmap of \(\gamma\) values per abstract and topic. +

+Fig. 3.6. Heatmap of \(\gamma\) values per abstract and topic. The corresponding code can be downloaded clicking here.


@@ -544,8 +543,8 @@

3.1.5.1 Consistency

information to interpret the topics and understand to which types of abstracts they were strongly associated to.

-

-Fig. 3.9. Wordclouds of each topic based on most strongly associated abstracts. +

+Fig. 3.7. Wordclouds of each topic based on most strongly associated abstracts. The code for this plot can be downloaded clicking here.


@@ -724,21 +723,21 @@

3.2 Taxonomical identification

3.2.1 Outputs

-

-Fig. 3.10. Number of species studied in each year for the five classes with most studied species. +

+Fig. 3.8. Number of species studied in each year for the five classes with most studied species. More species have been studied in the last years. The code for plot can be downloaded here.


-

-Fig. 3.11. Proportion of papers in each year studying each of the five most commonly studied taxonomic groups. +

+Fig. 3.9. Proportion of papers in each year studying each of the five most commonly studied taxonomic groups. A paper can study several taxonomic groups, hence the proportions for each year can sum up to more than one. The code for plot can be downloaded here.


-

-Fig. 3.12. For each topic, relative frequencies of papers studying each taxonomical group. Only papers with more than 50% of association to each topic are used for this graph. The code for plot can be downloaded here. +

+Fig. 3.10. For each topic, relative frequencies of papers studying each taxonomical group. Only papers with more than 50% of association to each topic are used for this graph. The code for plot can be downloaded here.


@@ -805,27 +804,207 @@

3.3.1 Outputs

information on the framework was gathered) that use terms related to each component.

-

-Fig. 3.13. Representation of the components of the movement ecology framework and how much they were studied in the last decade: external factors, internal state, motion and navigation capacities, whose interactions result in the observed movement path. +

+Fig. 3.11. Representation of the components of the movement ecology framework and how much they were studied in the last decade: external factors, internal state, motion and navigation capacities, whose interactions result in the observed movement path. The size of each component box is proportional to the percentage of papers (in parentheses) tackling them irrespectful of whether they are only about this component or in combination with another one. The latter is specified through the segments that join the components to the observed movement path. One fill color corresponds to papers that only studied one component, while two or more colors correspond to papers that tackled two or three components, respectively (the ones from those colors). The width of the segment is proportional to the percentage of papers that studied that combination (or single component). Only combinations corresponding to \(>5\%\) of papers are shown; e.g. combinations involving navigation and papers studying navigation on its own had \(<5\%\) of papers each therefore they are not shown in the graph. This graph was made using Inkscape. The percentages in it and an alternative graphical representation of them can be found in this downloadable code.


+ ++++++++++ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
ExternalInternalMotionNavigation2009-2018 Count2009-2018 Percentage1999-2008 Count1999-2008 Percentage
X---237133.341833.8
XX--176824.828723.2
-X--6639.3967.8
X-X-4856.81048.4
XXX-4246.0695.6
--X-3835.4564.5
-XX-3735.2635.1
X--X1762.5443.6
XX-X1361.9151.2
---X831.2231.9
X-XX731.0121.0
-X-X500.7161.3
--XX490.7151.2
XXXX450.6110.9
-XXX410.680.6
+

Table 3.5. Number and percentage of articles in movement ecology of animals and +human mobility studying each combination of components of the Movement +Ecology Framework for the decades 2009-2018 and 1999-2008. In each row, an X +in the column indicates the studied component.

-

-Fig. 3.14. Representation of the components of the movement ecology framework and how much they were studied between 1999 and 2008: external factors, internal state, motion and navigation capacities, whose interactions result in the observed movement path. +

+Fig. 3.12. Representation of the components of the movement ecology framework and how much they were studied between 1999 and 2008: external factors, internal state, motion and navigation capacities, whose interactions result in the observed movement path. The size of each component box is proportional to the percentage of papers (in parentheses) tackling them irrespectful of whether they are only about this component or in combination with another one. The latter is specified through the segments that join the components to the observed movement path. One fill color corresponds to papers that only studied one component, while two or more colors correspond to papers that tackled two or three components, respectively (the ones from those colors). The width of the segment is proportional to the percentage of papers that studied that combination (or single component). Only combinations corresponding to \(>5\%\) of papers are shown; e.g. combinations involving navigation and papers studying navigation on its own had \(<5\%\) of papers each therefore they are not shown in the graph. This graph was made using Inkscape. The percentages in it and an alternative graphical representation of them can be found in this downloadable code.


-

-Fig. 3.15. Proportion of papers in each year focusing on each component of the MEF. +

+Fig. 3.13. Proportion of papers in each year focusing on each component of the MEF. A study can focus on more than one component, hence the proportions for each year can sum up to more than one. The code for plot can be downloaded here.


-

-Fig. 3.16. Proportion of papers studying each taxonomic class for each component of the MEF. +

+Fig. 3.14. Proportion of papers studying each taxonomic class for each component of the MEF. Papers associated to several components were accounted for in several frames. The code for plot can be downloaded here.
@@ -874,8 +1053,8 @@

3.4 Tracking devices

3.4.1 Output

-

-Fig. 3.17. Proportion of papers in each year using the five most commonly used tracking devices. +

+Fig. 3.15. Proportion of papers in each year using the five most commonly used tracking devices. A study can use more than one device, hence the proportions for each year can sum up to more than one. The code for plot can be downloaded here.
@@ -902,8 +1081,8 @@

3.5 Software

3.5.1 Output

-

-Fig. 3.18. Proportion of papers in each year using the five most commonly used software. +

+Fig. 3.16. Proportion of papers in each year using the five most commonly used software. A study can use more than one software, hence the proportions for each year can sum up to more than one. The code for plot can be downloaded here.


@@ -927,8 +1106,8 @@

3.6 Statistical methods

3.6.1 Outputs

-

-Fig. 3.19. Proportion of papers in each year mentioning each type of statistical method. +

+Fig. 3.17. Proportion of papers in each year mentioning each type of statistical method. A study can use more than one type of method, hence the proportions for each year can sum up to more than one. The code for the plot can be downloaded here.


@@ -954,7 +1133,7 @@

3.6.1 Outputs

-

Table 3.5. Percentage of papers using each type of statistical method. The code for this table can be downloaded here.

+

Table 3.6. Percentage of papers using each type of statistical method. The code for this table can be downloaded here.

@@ -1025,7 +1204,7 @@

3.6.1 Outputs

-

Table 3.6. Most common statistical trigrams in M&M sections of papers (with more than 100 mentions in papers). To reconstruct this table, readers can refer to this downloadable code.

+

Table 3.7. Most common statistical trigrams in M&M sections of papers (with more than 100 mentions in papers). To reconstruct this table, readers can refer to this downloadable code.

*The code to process the dictionaries of the framework, tracking devices, software, statistical methods, and taxonomy is downloadable here.

diff --git a/docs/data-collection-and-processing.html b/docs/data-collection-and-processing.html index 8c9b5a8..8c46b96 100644 --- a/docs/data-collection-and-processing.html +++ b/docs/data-collection-and-processing.html @@ -24,7 +24,7 @@ - + @@ -240,8 +240,10 @@

2 Data collection and processing<

2.1 Identification of movement ecology (mov-eco) papers

-

-Fig. 2.1. Workflow to identify movement ecology papers. +
+ +

Fig. 2.1. Workflow to identify movement papers.

+


@@ -358,8 +360,10 @@

2.1.6 Differences with other appr

2.2 Downloading whole manuscripts

-

-Fig. 2.2. Downloading movement ecology papers. +
+ +

Fig. 2.2. Downloading movement ecology papers.

+


We used the fulltext package (Chamberlain (2019)) in R, using Elsevier, @@ -375,8 +379,10 @@

2.2 Downloading whole manuscripts

2.3 Extracting the Material and Methods (M&M) sections

-

-Fig. 2.3. Summary of procedure to extract Material and Methods section from each paper. +
+ +

Fig. 2.3. Summary of procedure to extract Material and Methods section from each paper.

+


For some analyses (section 3), we needed the Material and methods @@ -395,12 +401,14 @@

2.3 Extracting the Material and M institutions we accessed them from.

-

-Fig. 2.4. Number of articles about movement ecology of animals and human +
+ +

Fig. 2.4. Number of articles about movement ecology of animals and human mobility identified by our algorithm and per journal. In blue, the number of articles that were downloaded and a methods section was identified. In yellow, the number of articles that were not downloaded. Only the journals -with more than 20 identified articles are shown in the graphs. +with more than 20 identified articles are shown in the graphs.

+


diff --git a/docs/index.html b/docs/index.html index a70726e..c265410 100644 --- a/docs/index.html +++ b/docs/index.html @@ -24,7 +24,7 @@ - + @@ -238,12 +238,12 @@

1 Introduction

- +

This is the companion website for the manuscript “Recent trends in movement ecology of animals and human mobility,” a quantitative review of animal and human movement literature in 2009-2018 from Joo et al., serving as the manuscript’s supplementary information page. The R codes for the analyses are available in this GitHub repository.

@@ -275,7 +275,7 @@

1.1 Abstract of the manuscript
-

+


diff --git a/docs/posts/moveco-images/Barplots_flip_topics_taxa.png b/docs/posts/moveco-images/Barplots_flip_topics_taxa.png new file mode 100644 index 0000000000000000000000000000000000000000..ce3da2da48509de3248c1743c2321b1c691fa1cf GIT binary patch literal 166047 zcmagG1yok;)-}8V3F#7$78MZz326|dZAAp61qo?TIu$8t6_Az^kd|(cmJ;c1P&%aR zU)%GZ=WxF9egAtq9Nfwsdtb5ETyxHK`98QOPlQj4k01!qO$8Z61i?u{5UgK#IPgEk z)^R=vg7N0LwDg0U($dV9))ofOP4y9kEyN{6T%lHyzFu4Xw(J#RTT-j0A0ki4&RBJ` zQ^rfBo>8;mF8Mg9Z&wee!6 zm1XheQH&!w&oqNRp-;wPA@^9{n3_Mvt6=fbLU{adwsz0S@L+eW*^n@4uS|-y><`B! z-fxZ~Ts>WvcSn$8l4$bx-P?2J_vxQrym{{WB}V>Vuj_&|!Zlr|G+y?1@ST@q6~9k^ zj_NG~haLBov+Qe>R!R@~zi%6ADhpXyDPQ65ExqLSR46-><3-FY_dC@(av8kfhtg-y zsQPq$3DIS_yMLPGICzS>`QrO@Lb*U}rFbdpgK@XjGW9%e;R#kx3dz;_LJ3_-BWj~w zv;F=4!l+A<<7bP14jvg}ecZiJG1{9TFL^9rX*JS2Ek16IOXK|WqO^4qa@Q>$DUa(E zoN<4p{>iXyDYGD`3zu11GuhnTLs}BD?MQf5n2hSZcl4FB8xo44;4xJn*2DPFFBtBTP{ z!fgDoF54*Mt!&uaPa!#{@V+)hV&h$S^M+Ik@9fvR2E0QnB$1b;SzAn-TcdMlt3LGi3%X^2u)M~2X953(B+5E^zTIcN%yNnAg3B=ES{fIB# z9(9U}ZpXN6T!4j3ESaf%UPL4^^oC3~hUr)(!~0A3&8;Qp|NQQ1`u<&NtkUJl6Znqm zN|$|Z)BYS*+$1jTN`Gl@;-_^!WEmXrk>fjSQ$GIwrq+^Q5)v>a4*yUvy6!wxrXqWY z8ypoy5dWYFa3m5+0guB1LKh@mt&O}@EJpds|4nwtQquv`cz&{57UB&ho|F7YV&}f zpP%jWWensLHugC&vGSghCl|=$<%1YAWH1p*N=nJ>zC@L5Ez>9;XH^#$aTu-PF9M&6 zi)m;B8qzB5pFTbJQiQ>AYw>h^j!j>)ZIT{x|9h0!7Rx$kdu-5}L95~N?}ajJ;cgyF zSUrOuI*sZ!IwZyx#VVY(8@ldG1mH5-SI_=28!DVw$gXO5ckwp2?V_skJcH}b1g`Vm z?-bQy%X6%EBkS#TKEOd?QpL8)Q5@mYGdja6`sSx8V5U5 zLBTCEnS1x-MQ~P=%A57BvW!i<-QXGZ-)$9laZ%y&7QwmxHOObM?agV5)u!`7<00~l zHAy$$|EO|v+u7YcCn!jrU3DZf>UQj+OimW^9;W|+vU2aD*ze!JGuWuVq^4H4M)0k* zi>~DG`uLPHkJlLF;=N?>z0&^%$uICf~NiM^Yg~#2kR~Uxq7&j z$45@^4rUP%TG;$duE+ah((w)~tsVxA?Ce?J(-f@Zxs8k-O6&!# zb~~$27l(>SZr;4fO+#pZe7GGF9_~#T+9p)^}GA}O8(rb~1$47^VJ1&zmX_@WP*OKT&`JhQ} zCRa%DE$!Dxnc(L; z5xREfhcn^Y!C_%uDY1_Dp`oEWNBb)w@84sE-tydEsbXqwZVqmKwlf=LvD$LAYJ7dB z;>o|Xz@GS4C>ZSBs ze0+SIE!vgN!sxDPWO6GXP%SBS&h&kP^}hKHr+VQy%nlnHn`Dvs2)9l(a&WX=Wnmko zuI_d>H-+;8OUIl-T+cJ!r~6w=JFt_i)+S}&czY+pRPEF<9sT&_xnakImH)v)y-YlC z)=b{|W1d4-i2mYO6_eA}V$JTxT)bl1k1{)pJF#K|*7QA!yJied14HfoU4i5whH*j9 z%@~dEn%i8sB=Wv1^1I>I^fi?^KBvKZ@_kE2+GA&0tRAbkZs=BeLavz(kzpZYbY8vT zywN5I8KY<(nVFe6$H_^QC>J25p+UdCv9UJkcU3CAC*j3h`>)Xo(Ob7}<-C}?M=$ac zFV1x@?eWyh#eSaQVyjwfWN~iJdv(=vdUkelaUf5~dKynTL-kdAjHr5Lp3}B@hFSr> zS|$|)%3s4JeCVQwk9@i^T44{14Q>fR!mnz}sj1Q85YzjRNF0#%^z?+`P`YalnOvh? z`+g;RygTg?%pH0auRS~-s;c@}RPN%OJ$qKQ$UH`WL`KGrPSBiwYh^6x&!3lKo86BB zKCi$+J$LS0@k-^ssjJ)J-X=QnIs#F!s81&=`qp}rOwGl)HFPG*=Nfgg_{gxZQQ{{h zC&y=Hwa>bDhjFT}LTuH8lt?_&3(FaPbt6qlw=s|bO_5}Dbbeha_osd)--faCzQbAz z5v4O*TcWO^VG_dA4urv{0#S6KBOx}8_4N~Z9oIb+(`4X|t(FG)qeX1+Q|`w#3|UkK zoRjwYl$we+*Aa({ba!_%U%5i?>eVaGhUU;26SG!CGQDSVvroGbHUWY>eE1Nu%ME2^ zDhP_7x_+CQXNzwd`c;?-k~s*-?`N3wX53Iyi%gbqg>@Az;l}7q%mP9K=o(@xvV^#{8fJkNh7QMI_) zRSd!M8v4)jx?Cjmjv}V*dEt6Qg>YNwZtG}aSBQ=1U|&_#`G2o3Z*+a-(q?gRT)zC= zwlYZKhz4f=YUNJC%oiv4nlo3Fb+(sB(0F{b(S*)U@G``|?T&90i@?uUcb1E_D zx|EUu(+B9}o_Jq` zkMmH@(^}^6mR#6~avr?YTpE%$Z=PnoA8WcnoB%J>n`#KaBP6VY=|(5HrRInB@tz8V zaB2WM^^m+)tFcJNN9@MxVV)kSs2JL$JimLFD3C!E2VxJ8UV~7mde#22D2G-#4SW=q zR4<7R!IyhAH7Rgt zSJLzF@F4CUxFos*U-bn@xEL}e42R5FtR*~Z+zF_;u8H2))YMGupkK=SoSZx{U>H{q zSc2%4r2F^px5|(97KC2R682_k`W212mNhRKag7b;i3d)bIlcIJW&LwSi;(jUw}8c{ zKZNwj7`xFM&z@ayAF{(J-UqO3pxz#>Zwn9e%rvm1y2muAN$f&+J`v-re0r+~t{Ey4NEVE__3Q z4n*GJuhF6rZ3cu)*XgVFM^>0ANAG!JU|^^gnx0i+Kl8Jz>!F4QA+iG@40g)u!A6HV z|02W#)pC0tN_=eF<>Awr%0outE9D!+5X%j1lD))-)So=@9j$QEn{AB_WL0q@X#-8Aq zAmb`kduJdn-@1J}Gi>5lsi@8TGn;YTVb}KS?s7H@Y=$j{_geuX3YZV$A@+xxeVn*q z2)98ijo9IzE`g%X2;0tx!YH5N>@Q#NJ0y;s^V|w?GGJZ+u=ZwADy%)0AU*u;RFO@PZEsV8w&J=&)oOR$Cs|C2du8QMkZj1a){Lf+15QLN@c1FaR)A zV@5~8!K8ndhN$Hd$x2E}l)KPu6;(tylvm;F>kENjGx3Ues{!A=7?JvzH$G~~Pu=yP zBvFn_Nl7^mAold>(*c!w0#PAQXn0CJXf)S*iTYDk*+++62+?By`3+vty(zdCfPICupFO zrEvqc_}$sQawLUt$y?;MPkCjZJ!QBOG`sViP}}JzxuM7Kd(ZU{ARI$;--O+mW*vhd!Rm8H{rSOf%mXQvL3 zPmpBTI5=dUKfnC?^=od}K&eXX^upGsU}agC(4RST1`RK|dU~AzYY-5gYn^2Aqj!motYpsF zKYh;~!VblY$SWHW6=si3`bK_u4R-a}HCk)*P;HcdTp5S5+abeCJs^s<|2enO4`%11 z-MRi!n=8D$yq0@^o=0DIq1ATXy|`R5$9S-0;g;uE*BOI5T)4BpuP!2CV`IZj8$jwS zb64Oc1RuQ9r=Pz}$Vr=lHEdi!dpnHdhJgVqQr*#kE=(GJW6D#Q7=;#N?F^niY8OIp z@mNnQERI&tVPj+a1O&)BI*Pb|X6jSX)FeV4u!r8|>Vr2{FEz4V%P?&H$fI!gZau&m z9;o?Hl+OCTc|KRJeOlR2IL|z{Kd-EPtKm*!=d~d%{#LiSn&fodIisPwGO^8_*G(Sn zeuo^dEW=J`v^Tdj{C$c#E(Z+Pq?DIYfFE8KPl zwo2ibf(6^8?3M5T?EF!L%;7?SoCh!sSnfS9oTGl0yfAEy()%8J|5>$~WxNG^|H@Ku(VttX9kB`Uk zkI{uom(*jThbo)|>9kmk3!vC}ib9DZ3no}Yxo*d#lXU0hR6w3=8YRSFaW-L1CPYQ&HZD3KdUu! z%ecd)?0p4~iT&bDW#zE8uLk=1HAdG;%a5Ybf!Yx2GBit*7|m2K5*Uf?wY=b6k}#5M zcviFYB4RSk3e0bhzvhK9=pH6x76-kQE z{q*zgoMc%-+TPbgV3Igk#j>MiG3L)z^@kmm2n-ECy>8csq(I5LNN4m1X@hmZHJ*%w89}m;!7kT`DRnuNxjX@jxLR z@AXw>%<|I^D%`wjb^B1b%a$U<&^o_kZ>eZZz>)ChM6I{geD~#-zkgx_ar3a-~YFb(iz`)E|e%b)E_C}b`ovT@17yT*7%rYj#G-bq9CaTh3 zMz7uWeEIs=6cmAv{|BUBw>N+NSW;$b#_ed%jhXVuPNO^{qjkOY3EZ;7#%w$w;Wn;X zh%bOiCGN#t51`hv#{eiSAI_!uNjW{&Y)IfkXz0`FrVxQR8+Cx|)Y{Hu=mU6nzR}8u;t~NMQQo~H_XjWq zlRQ*pA#ZCd0AFnhgiCmD93NVSF@`8iKZtvrtQcQfI`|IvTWzh?Z0lLm!TjcTacYjH zhL3i8N0)LhS%|)?)#zOAFY`Byb->x*UTI8;bA1bOKnDgCGRDq!)o}n|bDi<(S7Gmq zwmyt?R$`Dj9PiRNZ9I5XQXdli%S5=Xv#?7m@Xw}ZF1`Hsxi11SVobP6KB*&=R8(q* zS3?v?0a(Cln*ih`15{;tSf1sC6k38neS;3NKED~TF*^&|$Hp5@$J=hl-ZL%XQ&7be z7+*8&hz+>>=<|W#!?*Q2;~peAFMlgS+4_CHlLgAtk9>y25MBo7ts#+Z%(m5D$u55_ z?zE-T5i1_n!Q?PacCZ%YRtMFa)pQe?*zv(en1>xp!e>H8F`A>JqtvXd2B4v$oHewu z+}zyIZK(S7*=fkA!66|}BdFlI0T-z=PFSYw?AXi5$t5HuoeC0L_vxw0Dzi@4WY7`J z78Merfyhz==w)>=uVZq;o2el|)^G4a!&XBEEMC=IUF;)ZU(WOMp8<|2Y-=?+T-yywXp@t%E={`Rq4t|07rVeH@wT!#TahHtD-FRPxpdS7-AJg~p~7 z{O915l^y$-Ia4Cs%Eto~wr@jLh z2elkCDINf1kAnkeG$gp}{kfs4`mxRC;~m!Nxw)yydcV{!U!?l7GOx99ZDnQU6BvkNx&J*xuQBi(+%;sC=|D-Y0^=U`46TaNW*2AJ5~)D1 z!qhx@a=l%L8{K|QEiE2e*^L$3qkd2sNwdBet}s=ZRlcug2T*WpX$Ut{+0XTGFW&9= zh>;-=f?>3{6BTTr2W*ryJi0P|S1oJ{j~1U3H1mN%fYEV%p946(=P&k9-o>zk@lK3L zXlM@x+9>JXU=#+S=4Xi`Ql&OMev{PLI~dm6yW)*S4kApD7S}S1hv`y z8B}+*+ahXm`db$ZS?atGdc_5-wbM$4H!wW0p)=Bp&C+duCnS3&zfk1wg#4Qf8=d+Q zFAZ43em*TFG8&kRZevI~+E#P!luWoX-Pcg1pfl#LpkVh`f*DEU$5ND1ruZ*0h5|O; ziAPgr9q2#vMj$$%Y=FbW#s9pzgc{wXwNGHdUwQSy>HLVRX*d)vY~QikQ#9ag=v>O(Jm9pU2p zHlM8P8XE;gW6&XlApfg3lMV{psQa zpD*Sfu15_LdaE9uS7?#M+?(5+v3$%?dAK!XAbK7`Erg6;V^tC_$X<}?)o4t=&jt4R zhj9TwU{ZWfhU-+Yzqups+`T7A8vsA;$6w)o^mEoZ&d$!32bJaT+pJV#)6>&Eq~hHb zF3hi11)P@;1TjWXc*Im7@#|OepzC{V|9t!!Zc zMphOv2n<@<#47bH=g%kBJKzw}z5x~^y2PghCWLFfS@qXT5tzmELPGuNBwLzi{2=`2 z3*C2i7K^o?JUz?8)PKxRO9W&2ulvER&srykHQR2*cz zDfE?1)p%p%#z?aNjwJw*75wP;(Xadpm30W>B9(N2K`~N)Ui}wD#SGP4%<82pi=(dl zO9pdQFMpE*;=vu?+UHPIR1|U6>H7I|6?lJLs8rF4Ud*8Ej{yd8SOj8JUhd^7={Xnn zRKZXfHeHj+frhzPXRB?^3vQ|{`<$rPz*BPBu7@F^@a3w+VewYN`Tnw3hN;=7+&=_0 zr%`=+vxNB~B)pzQi3|3A$ddLaLs7sC-&M>G) z7=!|uH}b@mmSJp4#7$bRI(Ah^{%HVWsb=XLsE(iZ=REai{ZD`(3&CCg9G98|nXikA}-6Uitk-vG6sv(TF<u9x9WVo)~trg0Mys z1tJ)8;3aTL7)`G+hsMPP{rssY2F$Cn9mUA+)rs1s)>awNgiyEuvO40foL(PgQR$nJ z!GW^+pq+f(8wCZ!T}8_QTd%<%1JNk4#=cs0zze!HDJiM@<8~ZNY4(oz z7FRtwjh=Ccy{XYz^|dbkLdg0)tv=eEnxzcd^Ci2;XjVz$fwH6%4ULb z5k(H#J&v}SKfQmX)zxsxLdt@?W7-}%R*YWS%Wdy{U!!AGs0rcvUSndsw#>BIutThO zNmJhB;_Z^}1FHE3agFN<`+~e??d^Y|GaP~@>=t3 z`@4vU=LJ*4m$|sIIt?M^AnpKo)KwBxAx(8YJHNR2m7cS9M8szPRDoK+hYyq&E?#`X zxJC`E?~^jI>^V_)yQCCrZVA=^9=cg$A^_^2HMOG`x>FpJ`T^^#}P z1k+wKBBjLl>~UABo!Vn`TI9hy^_nW&>i5r!&kPNx=6llFI5}mZ+y_pyXZnCQBBS}a zb;)qnl3mZ&MYv&u_GsVa+u^A41q72UD1&={etX1AdnW*`7BE`{`r2xT+fmoL8|Zfj z2M45-l+%_RE#|uoH&5Tc^uC#t@`8zNEZh&oE712HY@S$*l+JWt|9u*A8L;A*n3yzv zwx7Fx15)}|;=e$CGtVqT;lgXhw3o)3Mfca&xh}73;d^4dYBnE&4Wjeup}fTeCl&K;>J;`aANB4Ik4N(}smG7ce7JQHr~&UFPmu0h zTsDM5aHwl-MDtWmz8PT^bg-#R1qW&td4y(zhj7GBgY34bI{a-6yf!8urP~C*mw1$nxsiML8%u zfk{ahWaQ=5KPRW8IMusli+AQi^epV-?AvEKRU4p{;a6b*ozV%I|vw=*Zw@3f*8AM-^jKOW~DON|Q0o^`E6 zQ0DS?Dn6T2bp3t_*SGQZYgqyzx$h~6#UMLYwB5+?vH^s(u5#nV=WxwD?l3U1*9J*C z1x>jSB?PsF;z`%m)>6~cYs*$D{mRS5pqA4`1@7O;;SV5fAt=`XNm!gul8fCVE{=(u zocw{NW&kz;l@7d2$Z3lMMd5u~PDCKXqfs3e-~V~}o1Zfwwp5ReF@ZkLc|9w;hTv+s zJF1gbzrqr-9Cx4D0A}O-<;$n@3_BX&DUaQ<29#Yj=UbM(@Ip0?+Y)|N-qrPbZ^jcn zezVo_nvbG(TyA71Zf@F|QJ9$@XlJl{RdmilwAXgo3~gJ1^bs}Y>*G@mI)K%|juq-2 zfpM;I+>`>V1A;&v2YT%iN6?qsckamO>L#?=aH!>{Hk*wWTj?csXcTZbwIB!(m$muN zyE4KkfCz&`kFNgu`g-64}eEk-SkFMSC9-soXW~0D38F4wc5W#@B#mZ%rvKKk1UdWtKK#~28OIg zhZ_bjE;mQoN2893C&W*j)Y$7=P)TaV@0TI~zq7A;U#*h@j~1YH*uhX-_{$O;Tohy6 z;+KBi7M!aIIQ^@t!W+yZdtX!1Jp5G7)7X+96cIN!UzPC0@=^aE$Z4!wGSiqNA23hh zW%!b!Oy!mJntD3y@^rsAb9lvj*{XQA=B2|00mLuX1v@(;N2+ywyAyuLQ1a z%zF_OyFB)hu`XMCON7_Z;?ZE)$jV~<$ zg&Qqf9$4(UJ$pv@d2_kr?KzhDPnE?VeG@Wh<3zMRlD@dgSavD4=3)Xd6vj(Bqvdc- z+2Hv??G3qL1u-2hKMQ3_9hhF0w8max;i4*^4ybAN(;@PM1&ea;PtWxTIJ=J|^9 zlZ8;UkFk5#Ob?ahyQxy+MEazLay@=L9ywceiLng#RGuPKxj9z~9ydJ<-&KkrmpV_Q z$}%4;#M~3u5XeA~Tl||ee|zN4iPb8TRgIBmd(EVmgDvzs)!aVlFMhBJS=iWoLSeZ6 zd1%4qYoYrF(s~Bo2^h70i#~WA6SYYI9})8LqYsbeaRG*!9u7lxU+Iq+?PObKcB%$c5PyGM$OsMZ1OT|4$9pazEpPRlVA$GN$=u;Fcs zTj8qbCc!2a^X2`PE;kKQeWmEod{heu$bh_oR#p3^m53_828V=_y`RPf_cGIyZFb&5 zMIAJjjh2cuexLWVpr%%K`4vb@P@y$u>Nfg}S2$+Hw15Ec;g#)++P@Ncsd=b*s0#G} zE`X~<^uR(orR&Ka=iQ$0JmF!LAnNjo9d~rstr>{}v$VIjN6bV3G>d5&Vk!AWGk`&q zN;h|j^x7tIDo%`u&1=|%w{GF1thN$62GZHt`H|lwtZfC%VM!0);p5>UmQFd%y~Hm! z=O3rrfTxqn?n_{UseqGy%(c2<>dD_r`T*Fp%9wU_Id;m7qAx{8ne#^cg}XzOdRmJbiorU;M{Y20t&?NAkyJ`LhXnhA)b<8L!&c zu3?~WoqBq_c5TQrY~dAbRwIbY9># zBdZK3UzeRZiR0(H$EZ;WvJ+?~^jt3C)Z>BjVC#y9``e9> zyHIfeRi6h22GFAN^WCT)6;5`jT@%X8-Wn>t;SQ>kmzWC05ibEo#W%9 zc>eskklivpSj4<+7J7HA+WPAoh!#sQ?L~HP!C=3?4DD z9K;r==i9;B{Ldr7wox>xJXnV12CS}w!$Tg^er`y(>AxZUV!oJ-aJe1UpTSGOuB|uM9s_JtBcQZ+C3e)F{m-L&S%`Ti8LTZHC`gi!U;?{X zZlnO%QCnj~!$bdh9Xw$~R}ji@+qOyQH4uWfTsxl!=>~XWS{OMtE-pD68-9>ZAZn7- z)zzuCMM3OD9UU!UEi<+jL42n$B`=0ZF_YTc5VjfQB@^Dt{qQ#d99-o6)y5k4oZ7Uo zh{d+V$4Bk<#jtPp+4n7o&+vJZli57O&+Y%Bx#~HEeKv^1C)GoDYeV;+WWO3|wiVvA z+OhU%d}U>&$3|OAEAYpU2g`3{WVlwF`x`5cD!fs4RKsu}*qfA`{M%b*78azs+_ko% zGP`XGn$ct0I+#hx9D;vqx7F)!_SmYs4Xh`U z637ysA`9{Un5QS~$|W|FH=};dit6a#2h&+2mU&W`Z7Dzzs`*9_^EMzOh@>uCTd7oaD9 zx$~_Cc{3Xj)h0mnRk(Af4)h8FWCv(L%hmB4ccKJ{z>NfD7!N)^J}LlX#mC13ZLqUk zwqglFum*^Jnq{{2Fgo5~!H{IhZ4N%)kz z5}l~)fe>5-q{!3YG}=7?Yq7%p`;9;sabhV#oe2Ms2cL5;q(Go}v5-5g6yV|-bzI;? zEzHP|0uvRWrv6%IA+Wp@(M+_Vrli!Vd5sN}5KwLMU~2?VSguJQ7rnS6;*{CL|;v2oQ=5UmmAF05^z{@PD&X<6BOdfG5OJ8wCU!;1G3@lM|Q>YYrur zcQ)dVOKKu5wOAjpe6hQUOODUM?j8g2w8OV+`mq}E378f^u87|8w2>mrXx zV(X~C*Qn+Ya(Fv^hwHc4v>zxtONax$T6-Ip-N3V1kBSH9kYY1eidDN+2zCsPm5-A9?HN z_wdmp56H$*SnBUt?|v0#?n!%e2`&n3AIz68H-ByY^~(r+GwGcNkeHPJt|Fi}z!wuJ zs5Z!+{We*ZqwvWn&?+hOH55h#YKXfA*Y>>ijuZ#1EPuJgo|WiaBPL+vHa9o_$_`SX z=rEIB6V_y-lwuDRDai8D=+**);r9Gn^nOd41ohxP0X>3ZuhwI+y{YO`KWVT~b7tvv zrhHHlf9S25Kkgr{E*`@N{#e*-m2H*vuV;`g?$6rAOoZM|%sQ|IQa{`_A?1ejSvCEkn;Lp~;+Zhki^T%t? z75!QkhD-8#j)AYmLefPOmnrG#2_d5g$mwnFLM@vMiZobxuz=@;4hZSE>n_|ce?9mE zd{(D5NGv=rX798d$GL2sea0KnQkwT(Nj4@BbwX!YxO3Cp(xzeu8vY>y){5|ibc-%|b*z<5vsRu(rk$t?>qeDX{ z>xWK{*eJQq2Cr`&j+qKbXZP{_g`Resngwc{&v?%Qsf7;Fg?9!W8+fNM!Ec_@+ciu< z=j4F)!g7)$p9GJ+RvXVNCWO2-6n`7SkjeU=9aI7 z(q*w)nAiR6%?H@7w=`JrQOtoCa&idtgx!r2m;h3a?+8?VYN%HTNl(Pi68aS=FefG^ z5O>gPy(`Bo8%`TrK$V?q-RC5fPdtfB}{Uvi|4KpO(wR!W`;_guv8q%nzBL z-0~Zi)A0-%r7urQb_ETaK+D-FWE>2!0&&UEu`nPPRQe#jr~j)y0eJH*b~(hv=)sdw z4e^nTp57mVtVWe<%zwwzNrUXKx{! z&P5*Te!t$xJIcKZ~P7n`1!z|GL)-Gn1*DSXxSptp1QD zd5XIGvw@{Q3#@*finfajNe%kuQi-!*$ zASD`s&~yMTJM^gP30j&VzZ-Noq5gP?Sd`S%==h*R4o!Hvy1Ii&ft2KDV;O@%dYL@K z!!Ua2y4(*4W$Sq4nC(=;mc!fCil;n;2mq0r{!HHN9-l{guTtXCoMK@iSsxddwz8O0 z&F1n@P6s`;NI282PY5T%M}&}A!9f_11#kws;Bm=})Mh3H?-B-J+NG}*w|p@z4jmZ* z3ZE5&)P$7ttX-_E4fgTj&|*Av0!bC^6~seUy>C>>{!4c)FCSiLk(8Onzs}G6SX(;? zIyPX7)insX&D25F`cOs18{{_0WJRm{M(H%a{dd;N@*oVE6}EruC-t{y4hy|6GaX+! z^Z_ePX|_inMn420HUI z=spB=0;bjC+ru6}5?~?sV#R!6EuxVJ_-)i-`aexb&;|hl$PM;Tw8%$IbZ_6jC51ih z&FF{+AL;eLNSx6~AnGq0x-lOq;VfRm=%OtsIb*(u2hzVtny&}cC@G+ zF~lFVrsmbhhZp@w^#>%;U+%yM>lT%y(ZPf*1RSC?kiB-8C&7;}vBJnXFol5HiMkx= zh!ma|r~yhz1jjiqvklba@DJ%UB^Gc4i97z01oQ(Oi`DMBKJc_4pfdvz3H&X}=L4aQ z+WH|k!0xI4>I^j{Fs$dH?gPgOTpk`i{Cn$NK6ZR0sQBnn zCQcG)?Ek-v68Zl@QePKlehj=eSO^OS_R!vsv;LUwAie%s9-#xd_TzPzQWOB8Ts;)e z@!&Z$J_}v5zXtLyp*?)?YBWHh{mIY?1*X-2UIudV*U*~KJHO@zokT)MhbwM^kyZb9 z66*hhJpX^pfd(UuD0n>NQIOo#c$Tv6NvknZWK`UaW?W!md zbS6UY;(VScZvHZ85+^|9`cBn9C=&Jj0m9<^3!hO>=a=+!hOG20g?B)frJ0$(6%89; zbdIM7hflS2aS9`c6`HwiETrBZeJsU)v`2k#_MHaYTLZQ{g<`?nVW@=1(ihmuXtrob3e zl44rRSU|FJdc~28I;=n_2g7O}Skr}a*$40 zQ~aP;F!WYR@8d_0Y7?7mL5KXoX8sos%jy9*hTI`5DJhxgS1qE;@r?IL-X^qH0H}^u z3%?YtdYYKn)=~01_0}=$Y8YM=RW?i#Lk$7~1sJRTT8^rSKyubCV}9QptrnCWJ#9lR zAyFl;ulaW&G6qE4VO3IAmP%HH@N2uHnyU1LYUS&CGsa~*%k9KjP?>Ud>+z?jr&Vb` z0A@n{B_L(&tTi&3;r<9r>Vih=49zks7`|$-cLxJpM2)zE`KQ6O-v|tJCgWAM_swWg z+Hwq52yj)W$)%J~L@l$B;tl_N;P_(7$&jPG0+x#2#4VyA1|1Hy@H}kHS1N@Br=qO% z{t|tWL7qk7$(e)LhWn$jKe&bii-P}$!0c40)l6vLdgXGFQ7S&?v6%R;5s^V&PI2)HxLqetZ}0Q}KUh^y z@BWFL$Wb1_rdO{j8Yxx zc9%;$6fS&q5RJN*A`P9GT`l1X7r@|M107~&#FqRSV?$$fC$+>_@1VspY0%UW9WY;_ zrsFEtLuhyRaywkmhD8{xaKZdZ_5(H!&wohatXa)1&{N!&mKf(_y{dPKu@ z04ptXC?A5JAYj-|UzEA;Nv%~4Y=YHzH3q$qW%c?lv`&V6`0xh$&CxDOR7V7R^!N{v z|K=B;D{X=1Ykdi5xPKnu8)+D6(AcL7{9}uzTN68dfNTZ{E zv#LVtQD3{d3OTGqlN;q~y z264xbE_LkD3FHJ>7%p0$+t)|q(|sw!e?ah$M3qa~D<&KEuY7(L5=$^YhPFr0GXA37 z$(U=)1_ICHEB83R$gpH{dDrN;XAKuf#AQ_+Y6*Ac9(}r_BI@?cBqsf|M?dww>T&PQu6*l(rS6N6!tcN5U91%h0-~(T| ziE2#fxeb+j?C7x_;@{d@4JZf))46vfpZJSV_Yol&c4RE`?*66ZG_7*s94*hfbEM!C zqvMZ6?l>0Q4!;~?qgxN3cKLlJCXQK%^ca8CU9mnXnduw$qFzBIHVafhUOS({;0_kHrlMM$Z zfKh7_x_nW$DY$l_HY25`#!XW)Ryl$uZ&_t!DlkujPXe`o!vO(j;Q*IY2-*n-xgSAL ziyt@xfc`_-m8gaPDwEk+ljw`Brf_(n$qIVhP+kQhKNb$5yIkT8K*RvSbHTU~m81k` zl!S3S!Tf93^@IjZ(C?l-dxqj?sKUJg35RunCIQVEnKM{eSZL-3&%qQ3#AstKX!HdK z5=Uz!XzMhv`X~z@B3Wu8CWY>EQWDyKfchZrO0fgbhnxyMfj1y@x{I#7*aPE!a6|-J z&AEgFsh*)B@h}W z#-Z)w>t%J+AOa*--aQIm89>M)K&fG3VP$qn>@V`s1jxZHHbw$41`iY3z5!<^$b#z# zzK;mHOc_LN5k$*=0^5WaA3@#8Xmb^irNFz`1N9!I8Q0rQhM))YfM;!CjLHui&O(p| zpG4G9x2eF3T4U7dbn3O!wpx*w7tVrpnbeSt0TB^Vjw*-0EH3or?vz>nxPPrr3;N2q zIMb8$^Yx=kwAb{s>neR*9`^akPF0#fBkU>68%05~0GIRjH$dc-OU#uHAh7E@)UpWD ze5#>?zB<>E9&}oZ7g?!g4YfMls@v+3##dGLwN)BD0Y&1ykuyJD^y{bQVO;tG1`POw zp+>-7Sp!xWFyBw0Q1XJ@8Zo=Du#l4hz!-Q=n5B^5U>%oXEnopQ@1@8=Q>9cqDddN| zs30(HLGDAF*q~_wJ!Jz*29WY;L|&$9?t-4Kd$6;H_F4EXbVZ5?m`e(p59jPwHa|1A zN=zsOFC8f!$F`{hm0x}7aMlmo7#I0(y1Cp`3`Qw|xj9RhBKL<)1!or~CT6#+s zy>goh)!~Y&?t%N|#pkIwH8o}ft>P=Vbjp|x)l~+M2B>qd>T;Osbrz3xkGRpb{~5^I zs>&+N1In*jcXF!(hLdc{>V9B~FjxLC@f`(zaOw(c+JcUY+R9KC4w5h2k`5fHLc?c3 z2oCaR6qF#7LdO=rNv}|}3fO>=PtaQievwn|tQ7aahHh&oP@>i-2d7;$g4^8m&gs;L7B);CzgM`>wj$$G56nEQhp%6MW-f41;;-C3=tpl*yPF^7k9m|A1|LVxn_N51xRNu+Q`?Yf_{M zeeOmZ1voT8yoCciCZ?O;!?r*fN$8bBTeoHXsLpY45CTmO+yg{r)V^8{7z(I-2&_P9 z;3MwPPTsOGWy)}Z_4FC`J$+daT8}|$0<#kf)F^Pyj0dVe0Z3AXZdJf7r>KSDc;LD2nr?@^l4NA9PXV2H;4OL%)g z*BK%Su`M3h`5t42b_;mk55*1CuAtO|v%v}&4jMw7vae_>85&yvy_FOMoGNk@XxSo;3t=Pog|4EEm3iNQ0MB~e8W6M>wJ2?m4jfPs;?gBtAUvR}6554i zLG6WJAHzqz&v?&)dlrG#vfsTTCvJkQgn(&a4hJL@l-qy+3Q6KM#7-74i2>&UybQ3( zwI6x({`Lcc;_(gG3*V{6AXL|fA80~D4K;AxH`pjAf2YJ1wz~nA+}RqkScA$1H4&l_ z08ZoqI}MyoLk6=7G$!J{Y2^gq3|h?J1O<^mg{Q9he@5Q#ceNaUUqq!C0P5(eMP}FO zpw*g+Lv__HfQw3RbG`>UjTxZE0Or-?u(h!f7wu_-oR!%H@je%TBPx8OLkmoMW*iY5 zIYB_h@%J zW^Qg$I7|r^Br2&x90oHidJYyGoHP)s!XBzx=`4&-Ko9~pbZ{D>IS2?S;uS?&=sH7t zPr+vh-Cxq|q3B^xwn6h0c&F^)Kqo6Wmj_k@v=0=w*_bk6;>qHhU}ok43DR5EIo!u9 zBQx?|lFj>pj#W$f8AH|LS$Xmt{0nWl3ni#pbE7IM8sz(@-5uS>&VV zX4Vom~CN?peLQRTTNZQXhqs*dUd4=866#s z5$V;}*9VenTn;E!obRq~Q=){vK26`tR4~EuSK|e>0piYLiQiE}N@hcLN1;<(KlfvN zU$`_|hK++3Zv>MJ%VoMypVas1nVcS~rj}naB$?kx)p$H)Nyuu|DGrE?#0j4}PwpVe zI)zOlBNG$Tsp=a^o7XjS<}HGEN=kqp4x8)Nr%;3@dhRF&5@kpbaNHNHREDfkAo^i- zLRRSA0}CLaF3_ZH_C)>s8*VL*h}Lp$L<;eoe)_a(GW#Q3!0p?&-3K&Ql%7Pil5~h| zoS_TwNhR2yjlxdNs+?GN-rHgbDlXoPs7t*!m{@UEaBZyA=~q?`9ZF9z_pI-j0B2`7_bBY%%s0-%VlT{u<9Ivq|3 z`}ZT^UUG}X*A?oj9c)my+VX))K#gi}hp60(c4`2Y8tZZ#0YG!S8+Qz$3lBX*W|;g$ z5KC>DUq77&We%EW4i?92JWzuG+A9n0J~%*QI<}|f8E-t)OW>Ll=!%0bq`LZgI0r9C z{nCE`FPY8hkv$?eXeN{a1{4`~m{XD;Us?(PxvH$J^xv)+^ry%bD>&f+IDfQj7fLYO zIN0n&=y5|pr$P^UAiM>7B6S7}r2E(}H_HqDt&nHRT zc0Bl-^DV}%YZO~zAPdh-f;XdlD|n%9r1wYx z70k;^4rihmt>5GQ_pqWrRbDQ<`2fN2zQo}v1PZQ#T@D~3p)(iS7U1X!DIFbV@DNx6 zbBx-EplFAlvYappIKu*t-~gfzJ%YgA(Gi;FN#Tf)8q_TTZQQ7M2H6<(Bm+iA5a643 z06mURO<^G@$pyH34H`m=J_Z|`F%S?CL?1OlBCGqn<9xQBb$WEh-8%Y zf4+3ixjMi5{@wTQ?{Ph@>pG|5>+^Y!*Xy}nv(VZeJanjqGJJ4=!x?mm5N240JYaGG zr$>@RNQaWs#w{WY0K7rGPu)HaG=_bn;Xp3yvgJU@anwVW+x&lg&SHUw{vt=&(Tr~_ z(G$G~TM?>roco3KskSR{35B>ph6v9KoW;Hp$`Gw?K*o=yMhqR z^V=(b;!Zo(SoKji9z?-xvg=xW*Y)QKK+JV*1E#y_0WvXQc&++|D;+Km4+I|Q>D?<4 zqs&86pHyYY_rTsxt!7U&CNT1NGEE}hJOGReT~bbgX^&C&Y}2U0qqt_jl>_LlsPlvt9YOI48p}JkIZt{ z&crZZY*~CZGHfU+o*%Gw#do)}cV;5fj*22O6sL>bo^|)`U5jV-E-qnp`kkga+6|x6 zjuzVm;4CqdwKLLL!GSjfiDvt}AnzAr`;Q)-6I%ZMb9KH>ESwgSoaUp!`_^1MxbE}` z4Ih{5@hJa2yyuMY6`V&;;Jr;ri3O_zZ46Wo$yeRg1uc_!>Ng1EQc8|W-nr^S-+4AH zTS!5E6~#SoZ|PkxI*z}%VLW-o%|Eo3a;}C07=$he!Mpx@X)>%u1JL<)Un%<_SL1>u0?)SjC+d$?l&Cu8*tusUV?cMAgrM*DQfRW40Y5LzEps0QWdRI@}m z>=P0U>Ks0l-!!bOu1iOmfgcpof%bv@#Rq#A{ zEVa<4P5#~FX(c@IkFeg}-cN&*%7UPhl4S?(c`cafZg(N{;JR(N(@_&m29p7M#}6=f z`V{;bbm(|7DvjsApeRNQx-(2I{Y_xnllOIX+-IFF!_3WiLPJAIBAJjN{)@lhLLWwg zV*v0Ju7`5wKTc;)&zE1t}uWPo$CNo{;qu|G_Wbj{fF#IFIDAkPF|uRgKp2`3JOXLbAkcIC+2wG44I$+-Cp zSqw5!0QImX;2UPohCPB828V_gTXS1mTOL=^T~|4#E17jZhI$&oj4Ri#FV@^1rgjY3 zRO*SQ#+6k zOXQ~x*2wvPh4PkcAZJ`$C`c^1B0cf+4dBDz?7+hFWGQ6zy%>P+ZS>G@YK`k!m(!hD zhJXHifHMGqy!%@gyF{i|@k z^UU6?^HW=!JJ>qlu=)spCFdwAV6+|NLyU_PLptB1B742xuK#(7hm!a1MBMIj(67L4 zx8W$lz6W$mf&x%{D+B}z+c%t%mzTdKWyOmas?R4~ac$<4J)aDaGAgU)(@O>YT6F@) zKHjw=!uNE~^ld{mpx2V;UdsAx_USXwe#IcFfYJ_#Fa2Q-MJ8W+ka`kdhfiN+Uv`m>67OzTgZD2v|a5B{wmpVCVAq>Zff>WFAHf`Y!> zyJ*fl1hf>+&N*Pg`d|$<0N7>zFQ0ZV``Tp3fyyIH3>5#6C4h4?4*hMz%{yoRuc+?* zvd9uvm2vqgo;Ci{49hR{{(UCA1_IX*18OD||KAs(lKp82nzQ@b=z}f|bR8d3@W}s! z7sn$!0xA0YbbAq^@j=Uwv;jUJlq}a82j&mMU`skyTszaaxVU=J4&2A$MGdM9-8@4f zv|JCOxD*)SfjZ)3BDEXJH{`0pQuCj$dfypq1<>syKPT(#3;nI=u(2_bmgf%3CPh!? zA}@uogR!5dKen|oF-%QOQG;&DDJ(2B(%mvXyfVqWh8YL`cjV?VD6ShP2!$}A2bFA# zhqA%Cbvtf!d|!Cv9spf~*`)xlmn?ob&>8lB>XG^YnhMC;u{RhD0D}y;;XIK$t=_R{ z{(NR6g`vCyEko}Kc@DES{dTmZE?S@_gx;7e-Yu=I7pU9=e9btENRy0#DlM{cQzjs~ zBmMEwA2``>$vI2nX~drSVUzxY)DL))RJo$UWH7L{Um*4lf5S1qYjBjHWo214e+FfX zcJdrVELB0Egc)hC;5&aEB=yE<2@Y@ogq)zB=5cTssi=oNdX%Rps)vIGejru`7=SYn zn~)3qS`w&uJY`0{`X#h|A?mAv)zBl#7XXQB7B(p<3L z2N=Pxg7?I~3SJdQ5J3o-Ytz)1{VK`$1PETCVoBe(b5A z1D|Hb;;skiTV7UHwz?``r?bq#!@}0ZNlEF#?qXBRC3l3_QG&4zZqpIrK#2o+Uioy` zkGc)Y-vb~H58A^Y;BP2)d#P%tcisHIYQg%Wrrx`DkGyJF!3$8i#Fq_#YMXpUtcdAs z{ApX(`YIs499>>%x|_L1a+1Cu(%HaKYVNmZXJX7&%3?yDN~vgF@>jjRm62LTSqKzr z@$+ZWrZ=!3)Sa&NxjHcNWBIegPZ4Ch!Loi+`VZnxl$D#&TfiqlUWllWv2Ov{XIuY3 zt@mq2K*H|rOlSXvA`30x;xe@lSj5oAw*{5;WZ>((mD>>}!W$z$Ge)TPsh{}mJ;Iew zg?a|QrSN)Btv=Uj9bJ7tjAaibLTCZ(%%V#ohPs6%`@o@L$fJS>!LMh`D%&yHof2=%+3lh+;MG7nF#^)4m52r?=v zD-+c(&4@4}Y6A5*qJdD0u@@5*L_Eo^U6&8cSPN-xMO=x|MwBSHcL5l;#_ahE3(hlg z+PCjo{FQJ`8nF5C^t$?252LkY5d2LI{775vFc~`Ho1kp`z$jyk_5>JDjan8 z4H4C$>nEL(u>VBJUyL<53U&a(hwks4Ezd|O5hUy6rV7CK{=O6fe82E->D*%t7T~W! zm&*sqW#_J5-RoD#(y$f|8Qzj4Fz4sBh3!vh_^L*twmy|+5wYHU0uC<-$0pf^*p%Jq zD{zqD6U?UK-;LWQHRFYKZl!#Nkbol|@ru5I_P=rA9#q}y)~##){CP&i!CEv5pFh`p z&^ZiTJ+yDtZEazzIPZV}PR2a|*b$C0b-PFpyoVp6K_IanhQBaiR#kJe9N&Xrp>OSZq+*$(v)NIcvTWpM{a=*9(i0iGD=J-@pUt0hf8VEy_sy?t zYQLGJ`2N)Tt8;MbGwEVk!q!nZ6_|IqgVX^^J~X6^(|e!O|EvW%&W0umI|*2$rA|0} zI;;5ksh)%&YtyMubE*6R2BVMb>CU)x%x|c1MXYFlnRh|m6?pdDfkLOh>HP9#A4H_* z6q*MwE6Io5J*)98A7Kx5FfHXRX#a{10LEM_G-WX zTvDDd7h$XGfr>%DG3h}Y(jP)}?NUNWUWmAs3%~L7{tcd<@AwZqT^M1A(F}jX*++%> zC*f?FG3S-P{q>JqTUL%}idxi)gQaYS^ld9lYG6(gOiRX(GbjYmX#uc^@wYltuLEli z&Qe@9$R(SQ6mkZbH2ketgdESf!ih10Y_R|*G3go+7(ltZswHUV+0f`yYCb6@kpGVh z9Zo~?T>|?AS+mFQYshbR1D`2f4g`~dxS7n2hS-v#F*}WcMN$@;)4&(QJMB32@7-HY z@@3@zQ|t(bu=e|;X3rPuh@7R==-}X`DAhH$^!N4U!fGK3<92w*(A(W2$MCr6>_hIo zReudeG{hj&Y+CSdPiC?_AXOe1Ati`OfNJ5`KMT_m^ulF5){fw3Q78xlhJz#XEO0>< z#)sBcfNEb)j$4w4P%5lJU-<0qvDoj^*=ElCSazfK4bJps!d7ea4gxl{h{X7`${t zT|8z0&}kywYYx7gD>kttc?;LRRRPv3SMz;&=rs1@goC5g_?aC2$qMvIIIl;*9?LPU z51l33VQntB;a8YnnE#_+USA~3_q^xhP<_QKJ!{YHo9>uB*TJ#>9!8_|TpQ|U5jlEP za>~4UOpJ$_=bI}-S-F9^^?0*wW$K+dT~7(cEtAC`JCQw27X>Ov^vI;#L03_pY;`ng zjYP3}*pAx%uS%=tOCK7)-Vu^>FJ*%{9A=YW4=g{`NlcvYQQN?1PJY_4;^fIS-0lFE zO;0PXgNB0CPN0*z<+%~tfy!e8qAQaGrH`A6wp7hIaz{A2`fyD7->)$(;)>bovXAX7 z4DX*ZlG9N)S3S8{0iw?v&?*yyf*DZgBLJ!u4d$fT3AklOUjes6#~pi7@QLiP=c(|a z3`l%9W|53u=9Q|y?f+9(rSDvO1#$eu_}2+Ve!pX|B$2n9XGlLLT(&C~eQ@{-&|MSyI zieCT?jOvl16==W(4Ox)vPj?xSgrt?yDZozEo_SPm5`(Qw1_rbPh@T*R1nO?YcQf<= zU!V5#rQAHU zY0x5WfH~9S>Y5C61LVDit&N7pV4p*myH#6zDUc(bA5|3{DC)?D0FYn>;+SEiZw8kL z*ajf<3kpr>ZmP9|5$>Na8-eioYH(q(Fi-%)YALQ|2Om1r(^pCoAb&q@WWMw`Pd5j}GuV-p;0YjMNiJ~6pul=?ERtJb3rd+sX=9b zR+%r?EIRWNS|;Wc+rsL9wI9(X11Ay?vI?bqb}pI#G-^@Z8NjVbe#gI}%{=|LpLbrW z-+>tsPyjNJwsozTjb?h(1$&wv{O^S&TEWOBk{}^I9E}=&TNn_Uin@+E(D)7#q^LGw zMhrAFWu|)|Rl(&UimG}k+9|+#&6>~C^B~3B38W&K7vjjOAADJX&uUf(l)FuG2B;E#8^UH{M6-vPwQg;iw@=4;JaU1 zfi5qZxB{2zz`#I`kcH(u`oO50y2dl$HG?kOku)c*QwfjWan)TnI$3(oJDpa%^T-h) zmBb@1E}2@soLrv=eyXPPkS+sdc?o!o(qr5{1(wUa2+}W{np1S~pqbxvl^<;)Zl7)N zV}M^KFbo9B#fwuqH!C4LwnxUa6f1)at?8d}GGJL&O|U4S%)pIGjtJ0IwJyeBN)~T% z-HdyEDg&ny1i6c)EQ574u(@zVlluYSRMn5fNAmVC35vs*SCWvvK=WAwK{-gP>x4izLcx2>zNV4q?`Mspn(K}ip z*NSY$XFS1w?EEo2tPuY0J~CJ#2s5#DuebJC#Gp*?^@k%B5c)*3i10Jeh#YvSa0T(&f9qM!{MEBOP#H)K`XkIV z*^XQO%qIEqVS~;lH1F;x*HM<&E=||e>5Pk<>7IwR5&(wojlzA7A0P zXmzK>xZOa;!oSFhIrgC3q1N+2?4*as4iF)rYM_RXH^ftQzj(13C{We&J+|O|FM%k` zd-rYuYLj;82Ka{#cGZDNvgfNy$JstR=og5HlVCHv@M=oRbjl)53hU2 z(+!mHpRWo{Rl|+wct}M+n;u<(*Nm0hz(;*Tp(&sqp*E*v zs1!^Lz>Jp5@E=Mz4>UpydJ;3#C`%V5wgQQ?;vs_!nzFA|XQ#^QTRrV7@x28JqXM%frt0LWXZZ!RXWj;6lQbIwIns2gF#*>fGzq! z%fPw3z@Z|DT!lcA1`{FKkoS@59bn_< zQN&RM4Ymm@gQmwJf%t56vRYBaF2LBoOGCjFJwZyTnzX4e9#6c?Tqw@CuLyVK`*s z$Fk1vxZn+-S|>MpS62W3(QHX&BhpKKVz9qo3Bi|OjCz5;@&VkXoFse)vMob7@VwnO z8Z(5du`O~iXX+|gY6H~PxE2YgCF%sqL<1~G$*-rla0-H$)62}?h1B2zp5}NCg^7!y5qs7*aN!!nr7bB<2k%reGc@Lj$&TT!{< za$*GMP;W23GYbuAOFr~4v7g!sDA!=8@G`vlM_B$xA5>>6uWtU(EBkA3hjmI1v=#Yq zZw5ax_)Gqz8J<)To~ZfGvO9i3%Vp#2Kh7s7OS%1%<6IH>{^pw%%a@;>jRBfdd*hhX z!UZ|lOsy&v9)Hyy;F#%tjnzr)?Bv1Q{d%9SSK!YlM#Y}1_+C2ojwMh2>pQ#B`5)5t z0NNMcap%FEDOD8FjV@~L;>9NYCRyzeEdj9t;E<0E$v6oxADBFX2Ad*wluHNhu-|{w zP`Rq!<<7IlZ;ZX8J@0OJZJ0b5YHMnecr|WqiY^V!#P|#(@Z)=aI=S8rnT3=Yf{^}`KV>2GK;GpFoH>#2C=`NzW!i=NCH-JXhcM71 z9}=31W1Yrgw>1wnB4+u~qemshPott<=lpO*VgfxsqlCew{U_ zxWctRsHqvOOTB`pXR9Ks%%?lQ$OOm{XfzG=sZhvY(y76#$oTeZ{o>6I#Ha8&eOf{1 z7kfwN<1Ww(7IT=Th3SW5ml76Fv}ziTwrI2AV(y9#`Ur zqt+W)iMzH>A|o>d3fZ~`UuNNgWaMC$*%dne&`E+7unBKc_>cbTYgQ?y%`VkSw-eed zC)bzWH1;gN;xnpiia#b$N#L#nGx(Hz(sChoV>w_28aEL*YKdwM%G8g&$l1OM?lT{E zD$w4l&md+Xu}yo2lkXfVkq?v|083o|}a0cK~fJ#GIyU82H#U)?-g`>68jU z{1Q$^RJ7pHSWk9*DL|$LwVqIMVSiHXa$v6<`RYl3R`o;j&>;QYM3@`?dQQF)!s5Un z9^)}d6mpc5yfc?1FUw<=t8Dh`A>_bp#`DOZksZy5>bMgq(@}A^RWZGO#c3>bbc;FF z)tnG(B3GgS=1@;WbE#&Zb8>d35J<7xU6W6$>(G%Stxz>l!i4O>OkJ0OLqMp{K#E7! zEN-ieb5NGjh)Jk|js32k6k5ABHw2swf74!dK%ER4Qb-2m&TO3F%|KEJ0f{sWacp0L zM5vaKiBSQmg?t}UPKHDws+tntPyzr7^`IvS@+}9Q-*hlt_(#p+9#|{TX43saiJB1l zd_ZEP5|rv_Dv$NQ?|t+N^>o=Y!NRHdE6Za4SwoW#ryoB8ts(3d2W`e#Bm;kdgR%fY zoSry9`HK_VG7jwS2(C5LHVKe@$o18iOV&>CFTFu9@`>q^2f$$Q>Mm0H>0`ib0Fa!N z<#hM?IOc3QEjFaZ?~XJw4-q8t1EJ`-C8l288cx;!PnT{QHG#_~{* zCuHRaQP=0IGMc;>VXyr!f22d#Xz54wn6`5gOjN^7iD8(R>^q; z58%^dC>`k@5Hha|9y$>e9{>r;Sq(=yPceL`sZx;i1DQS4MHCT=>1X42D?0JH^0DvD z3l;)~q_MGB?SLIlJKWr@ug?dWI?b@92v55ZkaR2za<1$1fLElcp+pZ7xtAI{cYiX# z{=E0}q=AFzBT7;7-2^zPW=wZXb`+;&&U{Wm&+p+kK)Jg`GB0M2wuua*5_89dkViFf z(0#oC9%S+Aufx(<+oUAR#mra8Ap>)e9Sf&|ghF{Mb0|ru-zNO6QP z^k$c0ssjy_F=KACgwsI*=5)jYQR{|AVI?&1=CvBM({`nV$q1;xM;k}K)3>qjXq*h*Y6}R)=4ndG4K~^-O$K<*=P(fu8~Ri4 zy#*<@VMd`a=hBcv0EUu=*ay=YG{`h^*P}f`bQ!7Q)#|0g zv)`8uOfWdws_dXI*>yIJMYnBLyCsBEGAoJf!EsDt$v|@?h0MZQh0)<8=8}3ejnLc~ zm?DlL2tqyW)LL*1tgx3L2$R4=Fa`O*Zu+3fr#aB93=fe?n#YV4L~%Q#NPT(^r)#Xs zh#kT$TkwA&M-FPRXJ01ojbUHO!y`=-zQp3ADkN1>q`jc@dVrA7QqO{f^_Wf%TFCvH znpZGLSA3)anqqPx;DG`|96yH+k;+EO`?+!@4+$IQx_UJxSeuk@fVI+lbYkErm6CK! z;Z%Wu@BIxG6%}{%QUK6Mm_;&v0HOrE!efFNku_}d_J@a6n)X3bL<6A{N4oGMJHUuR zL4lwINmh-(N}z&PLs@U~^I5>%Un!}S#Dz(-n9<;t0rTJdy-Z26=?8oYePD}PZywRL zPLcQFigu_+@(qqOcv2?0f6avHxayMDZ$USMG-|qn>h(MyYh7gUq7+)bDYFN|1+JCw=m9!*X!Vgo$h{|IcyxTIH1ea;^I>9{riV79 zcVHkK+7Ojrbwt7muOsE6pe2#7DpKnNXs!el0~x$;ea&01LjZE<>x6t!|Aex z|3%}gwehJ%xkDv1CXzwjRdqM*KJ84)b7KCcasO{EfGCGHD+90C1BV;{dy6k*V1gP7 z9K#}VzhJGP0pbwR^`@!8P=HCkQMLLyK&EH9=3$$oh#ka2O0r%+oI7X;YAvXZUIhBD z%H9<P0Ir4JMH_?I!k%y+O9X{+oP+GekS0BkcA~oVQBcGO= z5Tf<1AAddYtn{$VlrdYm6n5_ynOvN;Vv?UurP)P)MGsGDuLLTi%H4q&rvZI=cg2<% zA(t~}AgXWF*SNm(_<^LTRz(7yM*1Pb0ADq4vMs&C4QP6rZC?^wn=gGTV9v__E$V)3*8|# z1o@X}<|4$w23426!C^&WIxOK`JDGAs+!9EOorkjfgz^qIODF=Q0^9lt6!c&kC&U6I0}M{$G{;4UzquZE&6 zDg8w$)Jj|?#uD<9p%HB$Or>sL3*2P=ApfhDAV=r!!ljjlm?1%m<##=*CJiOX00v== zGlVGLo>QBgT;)!79L`E#Krwv?1VRO5?z}@or)jP1c3j8tjUY&b6P0$rQpU|CKitfMbag@^B#Xa7@rFY@_{za%*!#+-AEfiq>wM${L9b7Qhwpp?`*5p;+n~UtyXIsJ$N=+!gnKD8gI#q7IvNr}ywmaoLyB3W|SM zuf82{MnxSK&8X1LKl*U(#lC3Uv3Ohe|5E$b*P`|Ejw2sR$!=;;{=JB^;DJkl zdzCjOWP<2#*K4}dnnx5e4g3S{Nb}*Ypk=Sq)~*`9;R*AmM?YQqpz&4&6#sis^eMU* zc(Lky_6mj+OZL0f!}Z^{`5bA96G@x+dqv=MKo}S6oH=vQgC+a2sdU$6L(2#2kzI;a zQ+@2V-Tfd9LW>KrCLm~6ZY?_S_lIp6-t@DBJhE;Dh$5baQBm-|msq8YY*sTHIv2#n zP2IhubLoteqj#U+*nxc!vv__O%(S1j0aZrMosaKOsRRiI+`QS=>$}GO#Ftl%1?!bv zU1eJvG!%}AFS=3Ch9F4(Y%QGq7%|Q|>6=GcdLJkl_14%_LFR9&D&+oWVBTi8XBbk?#l@;iMpf|*TLJ8T7%AC=Y;SG$$;Pcxs$s_Bk zDk>N$w)$$z=QiVAlSkRF_(suj1c4^YZyfJnJ$2`YsU&RhK5P~|+}b>xT(Gpb<%1{f zdwO(iCODQP=XJ`h))!F|^9nXaL2-df5L~2XtT|?qsAvFUd2W5b)rU>xdT|?74NJ8W z#NPmS`-a+alkllL>z0mS9&V6bL2@19kg7xJl5VRCp<2EFwYQxEq$HS({7-ORHjhCA z-!Q@=4;2$#MTuIHnJ{bq5%$*7sZ}-{{?uB-R8k1Fe&Y(L!Xi66`mOgRt)0=?+YvB8q5!xjSRjqUeFy8a#IM~Cn!NGrd7Ezlg4wh2M5_!o>4flJzCJ%&Ig{WFOH z&hkeaA*q*!DkF7L*RgF;y0T8|h=ewH%ag;s$EHk~qK6H}=vfs`SQ##wl@Q;9ql1OD zyGmYJxjUjy?+ezFWy5ud50Kn@#H)=SjEC&}Ih%VKIEB$8ciGLtxk9OWth^2w4RkDT z%AQ-Ixt5@LTbEr2Ml{su{Klz;l=DCkXd*E%q1xcro1vK?d*a0K)rpv(Y83HHc^&4@ zv(`?LYCT_LP72(6Y7sGzt{M65_c6%9^0)p}wt@!q$c&MWSz)@ z;eZ6c>{57T2p*>p3K+udcJIXC>^Y!1qAhw3lk%IOF!24!Q_>>Z&~)V$7IsX;+IJk= z?>{n;(-gmfZMm1=y6GR7g-_O7zghk6-P*UWx*i8d3%ni;4EL)odL_fVH=^*>o{c4u z2Kyfw2DYj3Us-c4kejc4z85p!ni0v3vo`jRPUafDKAXPd5Ndv;C{r+PTxc zbz8P6f1J+i(9DF*JBm&gRN1YPs?HA|&O@@GHTnaZ6$IQ(ZE~BCBd^6~3{%8iOY9X2 z-C0snQiS^&MOUL@BJn8ffC$UEK#o95OG_vzo&$|&0odp^-XAg`e*gtM?%9@Tx3Tqe z7W3(+mu>G>#FnC|SCD^gnfUQ$d%z4t=b8coKqX2&+UPO+N3H>()JK zyjNZkhlM}epJp4Hb9RO5m^`tXKx}hek`VJ$- zrfnv$0Kq_UG%y71q31SvUkHeVNOV^Ar3xU#pi@_ock}Xm+34uxq>m(0T!EG+x9`*% z!DwfSfFeN()L`dY2H*q5M=boXQH6%O1N_#<6C+jp>}U^n_hV4v_4f5eB#%PTv`sMV8a|;o6;uGoo#%uv zJ{M81v!c^eQkp4c3Sb-8*eJrG$woj9K8)^`${0s_3(1)v4V?_aB8&oX1V?u(j)++J zf*|jtsbvF=+#m*lu0O||5OyJaHxeLWTmxzqCI$^^N7!WXwBkV^{pGAjYX3xN03L_+ zV3!6$4?u8J90ahy5}93j2NBvkgH)DuNfQvdW{m<~CTtz1-z8&7n1&g+rr&wL;m{1z zG3iV?2818dDI*sn^nWy70JQyeM9zTp#G~QP%s(G%D2w@FM~)yWjh-=ko%^GogE-A2 zhpsu<7I64P#b;y$V~A(0PE4y^iz_bCkIq%2JqM3eK`eRr1m~2WNl$V)VjTMP!GzL9 zb_p5Vc*f-6fXoj#USv^`m=-)Rv<7iR6I=?uQflN1KM-gbn8>CDWEF}niTZ3Yz$@TR z5TnLgaCMng-#Bs$k&{7LPNZ_^>)T2hSxcAJe>A2j8VC=npX^@jHS@}~tTFw}(iv5v^s2AoPfFsb?&KMl^eLU|1wj&`w{z<1f?{K~wl2jvKsPAb7t?J>In1an6|9rKj1ny>hw}=R8JaFz6zKt8xXCV<6sd3XnDvwn?zt ziVR;^m+@opf3+*STkB3-DT0#t3)zA0(5Pl~RK|yd(*{-6O;Opry}Gh#_WgI`Ijz}K zK?NY*HZb)i`zG7TzbdqRagX-O>pO$Lq#xUrK9`%@3o2NrBW;cA@t|nvH;|>jzWue0 z9eg{?IH5*_?60M#haI}|o;r6p#t_9vS>^lz^YV8UL7>JZ{`ke+1BO@j@aF>v;BwKu zu@)W9@s*9ZO)(^;^2->}jhM^|zAfnawpEMne~rIDeU#RB^Nr&j(*&PHKK>(@wwf(Zbc=<9nWsyX%jszC)e<4wP8VpP{>WV zOp|uwID~f(&EfW=X1gE^2hJfi;=jxy3Msz`ST!1Z`s9f|$MOe8KhRpUIub+;nk*9B!9yYInr#p;~Qv&+4&bsKm(VQb`Y3E>ZT1Gye< z2POvI5Qd5-I@H10M9Bn@FMC1@e(OxsY730VK)sTW$6JQbN)ON5h|D{5=w{7=Jy7<1 za`}4Mvqxb5Wg3&_5fo{)3#ko2lSoOi{(|Rw;cE@aY{Bq+(IfH1KW{QlZx#=iBa{+7 z8C?4J$~{Jy71Y$`U*+d}V0RCG>Bx)DD7)E(FMD<~$LcJ{DVAdgTWYsrP!gN!h81Br z_i>4#@RItO6N$i>{J2R9_OJh{khxg8&l4C`&Wjgep+5#*HKW5A3(+dm>Gk_${aY7A2w9dcixSi!jKb8x?&`r60?2S&p1 z?S05Urr2p3`FaYkzR{U7a8UqalCJL8SI;1ppbL}mb4b;}J`k)(sbY}ys90{qcBW$q zU=@|RInbpWb@KoA;lu*UG-irnNGIb_$by?UZcN#`cQ4#sZ$8SvyxLc#Wwionia8CJ zpOG8^rzwSrpl*EK82u|X>t0GqO0F|BqWGAQ`aEb!gm{yPJ#rDiiH4)`3YH;7SJ0p% zT*X`sy0Gcev`n}TP-9mGdB@o`{QhA5=gRKhdTG`YW{80Bdk9+`Xe@^M6l@SLCW%jX zFX(7S$r12LC}$KrcDAQwU{_iJce2E3P;+dU&nL+Upv<@K;&BgH=uYfLvXs+^J_DV(8t z4$(bV!ckBJjgI2}WiZffOG@{_F7yl6&Gdcv=ldaJSQuU<_~y#Q}`-nj>IubhXxZ={u^%81HdDv-|4m(@$`c zW-Tq9i;+ad>hCXOrG&5Xh+m{0lKR2vd`fSC6e`z;v@=G+V9^vB*i~cF=byJ+0JCafqJHe4a0H+9bGz4iryqjP~ynvIA zWzOcZ<}na9#Qu>`RzqXS=FOWmBWEF^Oc^i}));KJh&E*+LpKe*@^R(Il7*mn1>j{u zJIX&Dzp=Wr6c{KHjc2*d0jx@fTtqWe9#z}TC1;jYH^0lmt=1M3tTD&I@>1Pda#xO<&^??U))+WKk}AC6ZD1Y2efqezBaYS`0=!>Ao|BvVOGVL_ z({rGQAy0S9S48{eH8f;4+D5!Sq^W6I-hupsLv@LX=Nv3BEIospA zp!R?uMp}kaBrTExqiPRB2n<`7<7gcdqNZpDH)bhu43TIOsCRus!wL?sx-omwIo1f{ z_XAS0dd%zN&l9rZ-Be9_{4`9)!sauC^p3GRw$;nAM`ZS@av~)W($u0i2 zJ`K;JE!+1B2Lx#sl$6Y70G+msACs`Fa-`8H@RDMX<1#wqQE$_Jhtd&8;A}cb;6d+n z0W{VemE5R^JU`OW(};r-sip*{23VjQqvjZjEdk(h$Vb&+i@D^0w6j^)($v08DGpy< z_JZ@VF=1(S?4lec#Wl?5a*Vo}J27}0W4Ggm)HgY4Xg<}6Nop#cEHdze#iu=0j_PKS z-MyJcXVy+L`{!qt&f64-XP&2)I8PAyQw)pi~dQtU); zeAfoG+1P|;_&usX99>2ZB~XB(8+wi%)(p2lB(>8RCTYjI3Ggs zcydVN@DL2Q19#s{sm~nHg^~cyV*T$-df+IT&)q9GmL>NDs%}bkFgK4>ss7zMa(qVY zMh+T#O+)yQYeZJyn3x#;E&XVeK7cVKHvujmU>2uA}605971}#V(k1`$Ex|a+a>g(bCK1Uf6f8`xuBFRPUI>_u9kS)XWfmLO4DhMJI zC1l>Xj{h(Q(!qrNnpE?z7xcRA#sW2z@)y(7WdYX&X6PK{vA4I+80kqTw+&6%MmWaX zUR`(uC=>qF=eIAqT$Z3)!NhIv5V!Fq0MW!Ucu}yUpf^D7!GQl!=&O(d=MF9gGgF`C zrs1xs_$UddBk`@~4QY!Aj!HY-iS2A2XZhPr&u1(B{bB{km00+Tr$r;OnyhZyx2*Il z@^+n;BD?E5WwwI-h6sc8jRIt;MzXXAr#+c;Bk7f+=j8Tv%l^ZxGKxA&fh!vEB zU6{<$7>8SpsB!F7a!r#dcX;?7h&mvR&yu@96suH~Adq|m{BU0XexvL459hip;^Uxx zMI-sg#vSCH0yJnOw0`b%dLN*If&!vfS2xsmJ3BcAx|k_k-}(H)q;K#s(fg5@E`cwB z=QGxt9)eK~ZLZjDZwk!C*X-Wo;*-DJfnOFYM$d5;!-Xn8ym?w_iU99v$qJ{$(OI>; zsXzF%k^!a5EA@m}!1QGN40#MuJVqxWg?Q5$Y4b&1z=QY$q>)ex4Sw!lCxo=2Ae{GL z#2_XP9MIM@zWM_46L|2ww+H-yo&yPA`Pv_+6lNnl6^y^_Jlw$6bMC0JBF-n`AS5*J zd$PGfCZGwkICUFvi~*E|lL6vWZLc3F!0;I0pt%q`x9oiqw@nZjN^cI!9Pu=7h3M+e ztrfFyG?SdZa>fFNU)UXSU2pd>GULX#=nHm7w>GCnA{XU}-$I8`%&I{R;qBh~f^?zC+}z&u&O{uYg0b+2zBlTk z9I8BU!xXF6hvI+$MJE13Uf|)Ritw-hdc19h#?;@dx{5zHA*A8AFQF?7)PDO%<%$RY z>u-1-$!z)zJ^~1)<1z1}_BvJts=+r$(d2D~K4kr6&&NEU8`7n~Zj=up@Bt@C?jI<@ zci=E($KDJ5iV20}CyHE!9$*!;tEfaKKRbp$BbWhCKEJTg1Iwm@om@IiXZmMrqX)=B za^+PBkLTfXj}*ioB_|eHE-~Qi#P>1aT*pvWVl0(`;kwJ~ za5kYjC=c7snUb0s`wgnYJ%|~Gum;EoK2-37Jvbxb_cN+fRvyq!w~Nv!mNu17NUpEe zwcSv6FNxxPa75(*W{2rBa_)n}4-T#aUTqL3*Il2w5$X8pZvt>RrPy|`ljZvJ=c2(Y zwLGdj=j{e_Nx9GsmEa60)2ryG@U5o%~P`o&;lZi9&8Wzj0qJKSpup*rlGwU`n5BNaHEhiOig>QyFWc2IVJeYLtlr9 z84QXSLH$E^0v@9YjB}tQB#P45?Jsyj%GxZtjX&-tgyo8Yl5&Dc0UHA@KXxRo5NcHF z8gL>QxX(>w0uYP&^|igpy3sozrK5Wv?L5KkNOM!Uuxl-!>9W6GE^Lu7g2F8Tnmr$% zVyUU(9{{uv&5Vvxwi-RDSFYF$LF2`zPoKWKS=?w765pZkUnfsC z-GvJ&nG)J~J$!_DK)y-vI$O8eG?C&nDFhxb?>yU3nkJzFhVUzMk2ZH5lksGc8dVG#*1$x@gbGS$KnF z7t0^Upcb*CahVV7Jda=@JS9Mfv=cx#(713QHu*U=%!U(hA5JU2i){;Vfh?OqTw^GT z1y)X+U?cz|pEp=koXD*ZJyO9XW7p*g*=gSVL?QT3x7HI6X<9tEEm`+Ai*n8FFtA1 z;2-kG{N`rpR|rA=?j(mQmU3E<(pT}xm7Fy+X%ChQlI*TM@TeJ`#tQ^_`~1oTFZmGH z(bh)b?r?W<)5(rf9=cwSX1h|L3Tg<4xeKIYff`_3-kw_wH1d@GhmWg(`q-!>(D0ir zq7}CR&c~7|hp?3-Dg@+1F+%ZX&`>a^I1f}p(%6U5uS&&dK zuTO&)2D}5;w?-6<)MwRtJlJt0dDwoH*9X*5PULB+!jsyVbhB|4yF}Hn zW`;x!g|9B0(_Ro3yXk1E&5K8ENp{)>o!VAK^0#dTZ!hqoBuOAnk_lTehKB}`i=r%w zH!!u#e(~om&2M7@Hq*fFp~?jRD0u&k_G6um4al@G{Ck>1Js}r!bqyA(3aa>V1Eg>=M{7xGx3NZEP>KD55*)^b5L>T zXTB9$>kJMD%DwEZC`iBr0TWUJbxN{PAY`K;UEA(?zT5@anAY>;P&Fn7pyaDmh2y<% z7kCKGE{!8uDr+~3K~xYfbR595@!Bq-Qif*3;MO%tN&=NdGub4DkL>^njOPi}+W`N= z*F{m-T!dy*RL6Wy5nVv(tc>M|R-mUMVR|rg>pT49e+4p7dya>L@QLl%2b7(P0yeeW z$;`6xJ1jY{TO%0MM|GzKgjY~%KMTkXO7AInWnn3+SDX}JF*A0r*~oZks>@b`)ab;Y z_X^(IxOvAOVvk=X?DL|=zb)xSN`ZfBDAxT zjN7YR`T^7~GKo+o;PIOz(Au-;oLqZCLbGuxUJMTxprP6jSAu#Vrl0aTDb?G$R8&et zL)J)tBII>0ZWU91oIP~AmK*kD)~Bh$=>@qY6Brayr9gCYsh}^hT2~1_F&3Q9c<&Iz ze0mTDK|_~uEJg}68(o}8Ss_6P?rjDG$4WkwI9O*^pOncg52qR$BI;CT&9;Ely75fE z3jHCiBy-A%3e=gi%;oz$0{F1#kPbTqXDn@G8a)Dj3d+o?-d;0!c1Wm;1x5%bPHFDZ z$9>^j#%ACy5j}N}ocZV-7T~)-m71;sa{$@+;rj-&HVr5#<17j^nuIcY@LheJ8?H_u zW5ab#ydRz8c12q_lNpU_=3ezaxTJ_Gml_NKUgX_EM)4-?T8GOL4pa_Bo0XpwKrR_ylC{h$i zlKeHJpFQo=XXfUfngD_yi4#H_Tpd)OdepYx2-_Mehm(ll#s{gEl@$eZKrR5BM{^M= zG?>P{J~_dBrwm12~Hjfc#X?xAQXu;(cBphE>F~7ZQ?}5mYW9Q2g7X#xUBgP@J>d(SBiNYnVPw&gs)X3~8<)r2#8^X6#+RreotMg|pKsn@ti;}8XOKj8_8rzqAP=DCGzM^f ziL?@2mE1CO8D)6R54 z@D-AP;~Ga!qot!kbc~#AcpUSOAJl2QD{cEoWw7{deeF^jB+6)p*FthuBts80JB^vU zuNHCb&p+ROl6hSebo$74o-MBRCJjwnJd?JmCMC>HNa{8WQuH-KC(l2U1OnX6{B3yR zwu$okEIk9wxH+aXYzM>FR<(} zT(d+$)6HG9s8UY@3lcX^kao4_yp>$_p~K@ckG1yV976=?Y~z>pW^muYD-{2v3-Vy} z6O*nVfM!ZdloSm2LX6(x6zt#!D31o7 zeTnEt8VT1lk*XN+WXUk_T-g)jve>}-;zNaql>GLw2yx^lr{qwd(69{XN=Q_aSvQ{- zml#A-TN~>51hi_Wp&F?*ow%%741fO$4(??f)jH)S79mx&-c{8LJVc_ z<9+0li)xdS6K%CoLI`plU0l56^pSlHBWHxnY%YaaM5Y={zCW;D4~(o&pMHT^fNwvE zp@NT;T5(8CjTyin4r0GM-e_^?#ew&q&9&|J9wQqqAf)9k<)(=#7^H)3NQ0+Ci|O|; z_GxO?b~$|lo>MgBIx%KXM?NSL`i){@VqYpfB9=)9LQlw?sM*jw-u9@@K>Ytm{!=uwB+Ah?b@H#j(WDLFY= zn>!B4z(yPRcwQ#0|ZudD4;vUtRkAR+azuL)$oVN$%8ngHM-*D+|&LwFF zO|YYRJt!T?r3U#;S*Ei+J@(mmS12j_il9QuX($pnkUBHPdG+3giwBVfS9C_b5G+E? z0=1#LW5!yP}<$O}vg5BpG7CEdU=a=a+^_*2)jQ6mc;{S<0{o)GqPld>M9gR+J!Gpb=aPgBQH4+(F2orarEnnny4;!Jm`au_Y0lk&U#2%qq=gv{ z4GqzZx(#mQXDBQN5q~t;?m9XesYVBuqi6cui#GaszU-qUbBL8-HP%XINB=w9Um8L# zy19|uF$%pT;^xEzj|60Y4X-riE*TjD-)~h}uW9dpPpDKuQozcWTZ)}deFzkLpD)VpC%k-RjDr`Y8l+0RTS=HeSM^h-M2^4l*ylO*1|N^dRWeCD3jt z;>|Yg{-lfjlk}tu3mZ2kD{3e0=#`W-PwX!8VWET6z2I?X{&j=hVEBOkfUd#;?uJf1 zEH@kS6(xQ@R9FkaQBGwjWo3c$Eg!w@?>`OFWijJMk&PSU580Z5$46ulWCoVsTfS1D zKcKKa%tF1b?XlW%BM?RO7E~w0tl@zU6A(pGE`a`yB)RP{W)3nJXKduzuP{j72LLIrKu2|`)TqP+GH4bM)j1zPP=o zS*97x`I2eBd=qrC3>v67>q0y_ctoAGf}M^?J_7`pLKH$fQ|)^lzcM<v|LlB8QpgB|fBKNZ{az#0T0(Nj7*~1Z&VVK!^$Mx(wEE)C|J9(2G;4 z1&@`6xq#y?1FitQ5P9xkv2bX=Z5+u9-IX$_rH0}2y#RDU2^a`E4TuRvFr{G%gUo^u z27r1pEVo1M*f=Ze9Ny=gN6(x?#Y&_h(CuMN!5|&p;ltD5H{D+YmWGV_Xp>08=Kvs! z5Dhw2K^y=Imqe`eiTI8wbNAPgWtiGxlnJh*!`lPyV2}>!I-wGv*;#1DLEY!l99?+m z`B16TS&UbYv$zl7%T|8BN^b88IX6@ED8R9l@my$#8O2(GNS)miV!6vQs`wm*Z=yBE zM#BsTgM>X~VFSB%X#(sXftmO~xC$tIl^67QJ~DB*ctkY=r2w2^w5k#(O=s zsApOpXtVT@Eb~9DRB`ubg8#=&wj<}0-X5;AV2i4ZPOi3A+5Oc|y712^{bQe^xbL(Ntuia*9K5v=H`O_XobfHU-2uE;JOGv3 z2Y~4qT|T8kJELJFhd+ACsK!!2;FIPy ze3MJ-r=LwX&U%B%YuHMpa-&&kIGEl&umDa>=0UP}kz%8(F_W7y7myIv*ark&9LcPc zzpXV$&MFE4hugFl{^{_>Bw%5_7{unoy7Bq~Y1k&Qi(#eJ`(Cquh)dFZF@Pn)k^w7F zQA$Z{$o=XqmHpaGHcQL~VJJYe@hb*?0b2adw(D@Q(C|Bm5hOB5c$Lv#Dhtcjv&z&; zBqe|^Ex-!s?TUZ%0f{+zn4I9_9|1szVu5Lvf&t8I13l^T7^a6|vlX8-Jcr^|SxuWV zg~mG*2&gni?zjzidt}|gRIR7KpJVk|_oDvWb1aT`_fMhHlfm<9u7DYW=c_zP9EB6j zeDB4XEL-_?SJL96@3l%r!+}8gNl*FtsK-q4yBiIYa&vR@64?+o7zG3rD@qB8!+_xt zC*>X!Bf2x}Ek{!l1XNg(RK-!i2l`fWr+lti0&zFd>cH{|{=`Bjk|(DjzRGQE;0R`w z(8!d-7D!v9z8!oZ^k&a-15@HDjUgp$m2g?F;%C*JG&=;?1tr+wQY8ZG7%o^l`Ce>PMnMAo}X8M|GI)%KJ z_BUU|w!7K@Fph@rMq64yc#2|_)vygO4z=C7qUK-(d`-@4esBAPe zGSUaA@d;)x%yxP?bA?C#jEKQ%&g08&I0)`Hm<0KUOiM4`xK^pG{Bg1GvMs&@-C<-| zZ!J!|KE%k-OUHSpzk1%h8UTY7yO6)C0$BzEhiA@{-gESt%f1h>#KI-z86)Ie`Yimv z&48(>k|&Wg4g|_NgdO5yP!KoC!owZ!iO;(puYqF#Q0e3(hCKWMjkU@j=elei+;!(6dnB4Mwxqp6=x;Maa^7}{5;)^o_No};c-T)5@;r{IF z&wDP7`|Z!P43yE2_xIO-Loz0Jg&Nt~+ZO_8y^1nFaU=~UGgxmouUX8dFBmKX{a9N0 z0OHF@6pptVY0Ep|J#_;|_d}hzGgU&~t+~FF;B%ZRv89fq8#$h*#>7yIxDjh&fV&36DzK!y-FZGrV?0T3Tg)LrD3!Upy-^o{6bl6_{~yj+5p$}M zVgR$+XX(= z?cNL?pDW5EZ3yhw>wk#u&EPGNg^5ka)h&=5pnlxYP$G4fz{8h<2)oQhR)paA8ANxY zsl)}tlg9fiJ-4;YkYdlESprsTP+ou{44gO%fP#WNcCIE|^d2G}QTpYg1C9*IjQ=y1-fhti1TBb>6TXhN8cPc&Mk&qp|oRsAG0@)}yV7PQHEl{gSJE|sk z?P!z-b0WKruh>mSesZx&x!|8%E?Y2v1*^5P^5BP0JWQ{?ZPGpY&}I+&|Hs&O zfOEaK|9=rGWmKrlv`aJ*Nk-{(8qyL%q@+m6PDnJnWb=62LKW$H780K!O$(VTUJ2)GDTM|qYBy?l* zk9dN5grOGMIm5sb$zG7>u%noMi+Yu+OpQJ?iNpbge+B6(Kqw^DxtWCpCEmd3t{$2g zoDdUs@p*Z!uo)y}6iy4JWpko);y)}`HUMx*0|xCRTV9l>#D2sIBh5MT|GNK?=q&W#FUI*7FpG3+Fd_XBa$FUKmk#?fZquMuW6d zX1OTysZS1wRPqL3Jj{mUA zc{z+f$VwXN-x#e^oI4g@Bk`p$PlbBe`S_mGlZdZ88+@zxY@Z1eN0>V!VQ1!&CqRJU z$>pgkZ1be1SfyVUlP`Q|9(Y>vt~^LHeFgyAF3^T4^%B-;n-D*Hpwpb(KXfh*8j?Vk z;RSITihCZ6M^pejcM zO4@$zE|h@ye(FzV;zVRpSD z$iViE3Ct#P+yGUZ?GfxOV~;XVW6%@@RU6$i$i5@#Kau4chIWnM&~58atFSn%={k(| zlZmlp*S!hh%u#?gfM=mcLg9qQmA1FJW5P0Kq6krR0%2_>=MJXtBZY=V)4@JdQ#VKu z5qPz6N&A&`x17VxUavJ;e{Y-IEN~LQ(v&LmdA_UdCHi`OCQkw3adGxsJx%Pl22Ke} zUu}@uUThpi-a54sUV|Nd>@V-&xTx4Aj+ZSfWk5=E|1>{_Co>!JYQ1-Y?yT~gN!!B@ z6Wl24^{C}w2Jq5S#N%s`QyI?+eem``qyA6i;ft04W69vkx%8*4JQml?@f2MY-U5kV zp@v{!b{4FAMSl&D|92woua5Wz8OG|O+1#tHPKq}bRaF7gZvHuQE~EaD*4PZl4&{N= zh&_y{u;pYVYXv+FxR=BlZzs6?^k@H9&!2~VyA#QdS`w)OW=MV-xZxF;P3NX@R7HL6 z(O?a5+on!=dcqjz86c?xmFg>QGb;K{TLapzN)c|kpWXT37&FGlFN=ft?41+7G~a{=!sQAhc?k2M zfewx~@9$66T%tJ}Zemr{n#~%=B=jZpH_eC87TCaLZ3#EX?fb*Am~hrzmuTdUS`UF$ zPSX$D+3FTCZFb0^5f}}BwHuo#j&L&osWcD6d>hF9b*Y!v54?(-8M&&&&~#phG$tuD zGNjNy9Dw9&Mq?WArl^t+yMRke-G4?cAYe=bulqb(e(8Pmc)fQ;O?6|ARO_>Vfb^FJ zpLE#nkf1lWeg*~L;idL+dOQ^GNPFw!#!E5$lHM6>e z{K!my_TlKOAVV9U|GpT-$*e1{DoU!E#Fsu1ydE{mM8B>2O|tm<$;CL%&pf*?>xASZ zjcSRiBxaA<>QAdi?Sp>a*gEkn2iBK;tQl4cna6QO<%#P+8#J#^@O9wVH5&FQk$B#!L;yvI`aqSkGHhw(A;kLl~I^+Y=@ zhA}NjdN2iBN>!|{j_r$@c9zMU+|y}CUoRxLUSNb&K4UaT0t)bfkwCO3@KGgNNtD!= zF)LgM?Hbr*uKoB``kX$7@Gw$_qdzz*X5gfPVL_H@KH$ERK?rqxI50r0J z^BH1wVB(hvcEq*a`o$*DQ9IX8|GNSZi&*X!FRfMG6F|~g(CH8Xw?Rk7pBo}eSfsy2 z59A4OlSUA5i6f^$Gk7jAQo)A75q}Y9ss}2GiFIuKPj-|@f} zz@E^x|9QJGA6d_Vk6Mdf%%*VOe~9h2LMXz}1H278`d9S_Eq>(~tc|~parwQfX5uAU ze>~W>Hf9D1QO~1*O{%elaa*`$yXOC@$XgL^!NVMQ!3F*qoTHg904IURrR?k9QF49l z-B-mi?aO(QaFEw^gL!+UDF3`6{*N+gF7ED{`HVeV7Tq>aEXV?Y{J7BQ-hmqo=`~#1 zV#aIitA4h$HjTkG3Db}MNG zzb^5QkI$;lRaChVdS9utD=ueV1uJ^Kx|2IiKSR6q8kN0umzVC%i%+f z4kp4xu7}gN?V-Ir&bIZPZoLMD&o`KgOBT+t4~GV4?sp~0E7o%=CPnj^`3V|{pDLMW>!|vmB``! z|Nap~H!ny+a4JTDq9OG%x62m04dv(JfDOP+!Q6mS)<6{JgzO=(E&gmH8<8#{Cl|ti zkMuqG?pI*Q79KEZ%Y43d>S5NtPGBNbkOPC85sE4YkH1bTN_f;3g(y>zv_iCg^4C>U zSO26ZPeHn5iHG7Bg9%n4+4HI3TX!QwP5>4idRu4z{k8Ub)bXmDECcNd9d(Nx;9e}+ z4TI&V7SQXG;22LHg8^t3P|QM<*QCZifBt+@KFk0Kr63R+9>@#^&3@<`K{-eV!;I;F zzX?yd;L;}*1(4sG{z!t8qJDct)RJ1*zcDk&si5l~{65fZU@8M^hswHjpuUoeit(Lm z%#J9grywdv5rRGErk)=^$iQkq4uV^G+g?Y^X=6VC(0OeDn(Tt-#Zq>MRbh4E7a*9Y z<_ZoFI)60KKpQSfUi)$->zwVbUDpp(8XF6!sF)y}j18y(0}0El{srM@B#wGm`oGTq zq>(_H@ZxAJ`1Y5c$9EkrlhhX}_Q@7A_qM+th*IxJQ*iWEw{3wZd#!{)#>8Z&*z(&s zp2NF?)D@RU?^19iB~iE09e8_k_KhW2oq8Lv@n`C^7gFmnS8>nGQr7!6PtAg$MO2|^ zze%QzLyCdSNvg{PMnB+Y3ap=t+xEP_DUM7ZfE%!7)S^ebj6`LKJ|XV1I{;NvQml=i z(5Y`}4`r195_lO!RV;C^e)ZzT*?Y0Gd8&WMrwTbJm+8qDZOo7FBp{?GtCfI_?9~d6 z;jQA%A#3={4+*4oT76d7^zFQCu=(=xC06^Vy_Gs5LOlG&y^l0Qq^=27n2X=r$Ns>} zSQ)eg>79%2p2EE*@S6^Qbcg|vO%M_X>>C)BaqqfayUtagfnXlS8rSMO!8x9R-HqnZ zu^)qc>=11*gOMwofL%CxTvRhaoHi3PojI&anvFPKaJ0gK~3_!a4QV&qns$ z+c&}yR~pCp1g#$(G7JN+{tt5ololG`OFqktmi4CJ}N{WXyJhJo09 zkYLb1f67_0k2^skbT4WI%(m6=IIs;Jgo!%#39ww0fyASk!r;gmEPGKInb?|$KkETg z@HP6zCV$|&&a>4ycOu;r`8;?7&NulTv-nTXlV*hthh?0@xhou8kSH9CKKUWg48Uf~ zT1*}@K=Jm^MWQq-GLJzG&I-5=nIJ$jxeFTw<@tmq*0ulzd{UX3!tUvIU)4>jmg`XX zulHBoG9C(si*2bYD8{aeEIpOZ@+Y_8?a4YcJnBzr{ll;M#5W{QQWwt;4})L zNrHfP!_Y&FRsl~yNpme(7b))8L(W4Rx#SR+;fWLQjAjEe6B#VKVy{o5|DK01Ks0O7 zXge>9y4_8}Stm55a=CGKc$uvEhOFMX>0h=MWx1ZQo*y-o<x8`-7UsUs#E=bXZEvo2alf4-_CZ~ zJ!kp@uLytCsTcr778;U7k1Xiwxg4JN4}q>m$w7%j--1Cqd{R6df$jnTR7G`<|9M-X ztHHU}VO~cHj5a{4B&QZ2uDok@t*V8|%#~N&pZhI4;1l@#TKoxFE%q)t`!dQkva1_P|_fYiZ%q2m8pVyFu@X%j@l?5RC zOt#QdQJNRf71BI+Vrq?)-L9HOPlX$xzHc0Hemm)bgUJ3EZ=^5;ADGOrFrY30q%)&4 z4m%I1+e;p;nJnV@^4!{G|H_-WP1ugWYz1)cT6**5O|q4R-={wSX@(;jpM`6>$sOnH z-s%&z0}5}_K7`#!UH#lqQAKT9%)xj1`+Vmn+W*iy-5qYf_2Hx2u3rsL&))m{&9B8( zPbFfqwm16BzpuaD;HIsCw%CF2U3(3R;umoSzGdCUW3ebT49mZhOXTTpS9>;>`kpb> zjm>h+4?6p0&HXl-f85wiR(kBg#a98LaRKW9c~ub@pCBhY`7}biMAH};;mj4b@7+5B zV~`HBQS?sOv1v=fX@ZB|3-oCWc@&{RAW_4gKVK}wx6A8Mm-X>D5p4kvnB&Dt35tMQ zTGe4Ma6U-o46+n~W>23kVK`Qsy1hF#Vg1zz&I-d30H@-!&-Iu!5@!AG*g373gTY&| z9Uvqex)WfLP}{3%YhNS%J#sfdH=i+iW3l1oP`6J-h7x+jyc8i$c1=x9;+@j00;nCN zvg$FuzIb;OQ^arwrUuN2dLjQG_1Tz^u6J9qq@=$Ow+3@ys z0vSW_^J9z`OAQQSD&w5hQVOf8z88*tUuA-MEmvj33#S^lXu(X4D)^Dg`kjf=gkR?- zP2Z2l+YTo9`1o8D@pVeB{e}H^RG5ZWmz7O*+OoRE;ON@BvZI^MyeZdPKVKy0tW!?u z1}UlZ+9k&&zCFu!`TF%MlL*I|w}AkMJ|~*QRKic%{;S=hJ^8;4Pze;5Bi+cvfJX##)K%RYZj5qFHNlzRVuV9s6!3 zqZd=jK4O<^95tkJQBQmPJ|&2U=M4Px<`$r`h-kLXX641=#O@Wb@v>{_&o^ zBwml#c}EgqvNgUjLB+Xzv~%&-fH5m=WDr#kZO$%jEc9E03D{}QA2X{L5N;BAaEduup%q9c7jWcnjyl4+5I5;1U zzo{Gn-yu$Zzj|mXN$jAber|pb=nN2_Fo^8{p<+Z>m1O5j&Hi6Mq56v7nDzx1)2C+; znKnst8#Y;dJ0uKy;<%3e@J2IaAXj&+p>I9}*(CU=O-`yFmWh`4TD@q|WyQZ<3V#S~ zST#J93awr@{=z52dl~hPP`d5gH=Ch{(FT@E`%hUIC30aZh;~s4yJvW|Fa;xHi=$}{ z#FFqBlBt9hxfrlFV4ma)0>5XRV_2+vQBt7n2D}z8AA0q#tLyHbvyV11*hGWP5W9;D zq4T6?f~7weXhxs~mTE)FSX)XLd0mT)53r+wcVNU?1$_ps$(({80ZT}P9!B*Dxqvey zL_{x<=pld&p`2`C>=q+l4OM!@qpy{x>>jR7lHoKg<>o8y&!Rx=p zrs`j8jLAL3ZHqwVqtFT1aKHF{7}8f{bSnlH(k=p1YsiTx0v-Y6!&|bpi_I|@30s3x znMJnbU@)(Kck0FANcj_|S7(c<@V$MzwTNpZyI@fn>y6!&GrZl-J$83zyAtjl$Su77 zax(|lT^v)@rXpRVbDxKMxi8wurZS$Nv!HOb+$%17IVHoi8Eib>ZW>+nSV;q>RP z)y~Sy^sk#YK@u5JiQz_CNwz+j;uXSSn*=^e?pmyE)O zPkrqioS}?p>o>%4a1oHtA>eE?ZH)i z=!B`F#AhoRhu*U7moLAUQBzTIejD^DVf)S>%CT#`oKcM~OHuc@n)^{l*R}mSN*&#f+|b>bA8=2FK9X24T--0^@whb4=q#-K;PNBaV;j^$@q#0#U%Rj2%nVS!LMQDb^;^=y2qN$~o z6rG#`2iAe&(2FhP3rw2^0MQbxFHnLhg&bT6%OVml zLPX?f=Mn}0oPwDFC-cG<;(J5~`JxfI5aSHLL)!2QB3>FUlh^NxFJUbOQ*98eFb!a5 zl%1ABjwTN-By@aAs3+bep{fHy)N=+&>j~AI_hJ7SeojUZmri(9V9MMUdp+@?L*f_w zMUo5H6RPtxBHx^`l&TfTcIkL)DEjvDghPtHxV1KY&$xd$H)x8c^zM}S70I! zS3-47h=KFcwvz%q$yb}5uPyv_Ms{X8hqnx#O??E^Bza{Y5fj;PNQAb8tbK`*7C*n zJ4fa1RyUCnp!*&1n*HpB`Y{alFxu-qh{LM3Hn!iO^YsmmB{yHxW>srti~kyXWd#NW zv94OR3iV?3JeEh|+c9;Co8w+#4)4~`_U{-Xn|{Ut$zXuVU@2rqE*2l!Ig*Kxi)Y(3 z_pFQ{&iA}=hf=U@Ve+B1$tZOopc`(dm(A_on`ZkS_gHzbp*79-PMPlZ=WAnR4 zyAFzD_%nrB(XD5z7!PKds%_u$cylpNsLlSh8h(7F&^i2NY@f06;})y4l*GKx%feQI z8}dEjkX)nn%#PXF{2k*D*WKNT?qY(Wlxbga(QC19ZgsN9yldU?#pygRh+!!#j#S&$ z+}&v}#r9D64NaLrOX70NaKaiuSKyx6peVutzYLGePbUR%?O3NwsYA3vYve~DuN_gB zZh;E&OtIw(CPdgQAT#GcFNgL5`=e0wp%u5Qb^6Fg66TNu&VziQq^~aqn?Pr{)h$}I zC^m|VgChgA&I5;9q&&cG`tz7&P*ha0O@M4c?PHcCuC91xn?Jb#c+fcAu&$Sie8WO9 zDCELch`ji$hUD2fd4QQPIrc0m_M5$PkEeXq_ylEuV-GGA2qH6EZrY#8#35ucq|EBC zHr0Fkm%{jp-Fou!$!6pI+LeB*CcZgGzu;o;?Fe6C8?_4z773d+h1t`S!`_DD6}!!= zqg2&P{P<`cSsiiav4iE$gW}*j`Px+CA+z_F%a83+fvmwd`vX~jnm~JA&W>a68 z0jhlPWA)tF^E-AxNCE`}@$*Q%3|=Vyp`istS>(3qTgmZ{Zj*&Wy}GEh|4S8AqeNmr zpDPF;0<`0^c`KJMZw(VI-Usg5VWb_uf3F0G7*I_ReRv561-SvD6cs4&t)BFfd&e{0 zE6Y_#L?zUsbsj!%j}6&*|B6hw#-8%kpVQSj{M-&kCjY${m19ggM5!b0KWSHk$osh5 z^Y=fxg46f2HC`@WBmtc_1~Hp-l7S3DL4>gC8dIlWlTnBfs(2$-^l4m>FaW z0u9wQNIy_n6J}3~?*!MC+0$LXng$n&zY8vj!gP4zRY959JosgHQE@R7BYE(v0FK|@ zmqv4-zys&$mXD7ZV14@49YL2+(}LZFksEo_#%bpX!Q$qxpTw)IFNO)^Bq0rk#oah0 zWS?%zx^A{WULMQ^pjxc)U_e`fHGw& z{P^M30KSlCDKr2yF2=oJyXWpctb%d`{u7;CmK${dO}`uitf zbLPq?6=!jHk@f7^n3S@bpV1=!QE6Q0&-^tJ&?sAI$YBH4~h?NJ5cs3dh-4^uFueIYY;SoHF7eKk1 z*GJra`YJd9P#C8ke^ZZfZ#Edu;n0**oeeJaTc~207^nhB`T`RWyQJ%XU19v8CiT+( zrIhBesprEw*yXvZS$B50l52B=$*f75|3NJtW72zY#iS2B1sgjDSv**jI7t z4m2+e0BFL*phJy|LV;!!j8xPW^kFHS3D$WSzA;a@_hY7;McK8q1!14rhVeMfIVf;R zJML4Ef_(Y5zt@16*qNh>FJ~*e&Pd~sOsqpqM;7m}8Plmswyz8dIT*~!0Apd}vnwiM zG;hU~$GypGCCwW0o~e$_T*d(3D3hY3pb83zjh#zAG&l$H{N$Jzk|`q*R2D+(K@?63 z^hJPB9e9-ABk7|sNbrMWreP!jy)daxPn|jynq6cX6C;-dfQUIQwL(aKp4%Ai#-NJA z1%EEI#1m&XJ5Zro{`N0d^3KJVpYAg{A`}8BCr!er>5p1&Q*S>j1BbS+`K7e^9?_c$ z%g?46N7+Wcj$U)d6IGHPqkSKIvf$Fy`B3%#&ZoP3iuJA_HK?`APMCqRI;{M(aLBV& z{NDYDLYb;EK|0LJU@>`g{o`@-XM<*8Zv*5S>2`zFHw9(D+Et4dJ*kfV2MX9@X$KfE zZ>3)aiW-bLVA7sxvJnGS6pIrM5;x?ODm^V z*<}?23IA33B9RDdPdMeblw7^l4DOlFM>B4o)l(>A&$=`^cd9);^O1|)MpE~Ok0qkG zJsKTlmU2{KE=Ql!(s!fF_zng~eV+Xxd6NXAKbi4nx@PS78xdTBOyv{n=MRo;E7jdp z(Qb*F@X?ioP_U((wiNY97W`sNTA^kFY*&ySAZI{d`qmD^m`aKN1_|b-X*~O^h>%;@ zYqd+jn0Pt^Zp@BQL^8Ixu|C6Y9kcTOrs^!IJ)63xlahcBq#ZXu{tg`(%7(ECP1}B`|UL0*gy!fpM(cfdh_14p&Lesb<>c>NHA=oBJhu>8T#nGeD(g;RlZLxIF5mC;D|5dT5pMA+5+*HVt30doLM<2)-WDlC0tu!WGN7PgEfVQM9$AG6YNi_h7Fh+!%gHHfF!~;ooRY%{H zYggz%*N!a}%@jajN(&_e4D=cB9IXus!AJ-DK0|c*@^`_Nld;n;s}c<#@uN-p@7SV} zL`whTRC4Bk8$kw4y|C)s@kGKk#~)cJm@lCUC%TzmF1pgqWHT`WKZtD*+|o{fqZ6YR z5^X_%1s6dkMss@~xWvE$iEIt%UHG2|Tuhi~6HMvF(~y?|8`T?d%*ZyZfs_K)YCWBb z{2t+^)v}}=#T^79nrDOXKq9e$*IxJF=|sSWEa{?2d5Np**F*C#XzRC#ax)l^08{p3 zf>mb&x~()gI3lz9)4zX$8AF$fF#oE48Z_ZKo>bijhx7Il8wO_APKk+|uz8`P@PF}Q z8nABiWr6!|b;Zg#)5-Aw=Pt@qBV%LaEzbdPT#>Y6IrP=!HG{KquZ|9piimU8e$XEC zjWbC<`H}}3IM8pw3oX9%pSNgcqSAX05Ym33kyY(|U>Lv(t?i&4G;>tSH)dULi}X|S zZwY%`^4X+U%743SSnMHq5e@jYH@x1ivR=a_EBP|)E=n6+WE@$PRn~83-S*KD(=B+!-3>Xh@=K!AtmLagY4y?dE}os3Ppp*A&0z18yeZU=C5 zI6%I+y`!mW^TY$83A1dmWkz};i94Teo5TCjYIBdVXiiSqmAX2k;~ZiEnW1Z8N@bK8+6l^)D_}(<;To`J1rr;=EJ`X0T`&X!*3@hImrZI!7 zJNB3!x4V1d+vYRn(#X(+ov`mooKBb=(RhMNAdCswQGr~1TzLSqgxZm>(ioS$MM3yM ze)3**!6%>p2}%@Pz4DGkI4Bx`j3EP)&g1zHQTwx7X#wSNh0F*$YTV9o`eNXYA$Rp$ z=b}lK%p+034}Om}nlFKA+w{7cbHngCagHNRP zpiX3F(8w90O^`EL7?dXuczP51!6ST6Y~!#0DyZ}NFk{A7CSaK$*pr`^02@wTIq90_ z1hbhazDME}t`(f~ve445F|CqmP6Uoj4s=LVp^!VWwqraG4b8?@L&2b!C1I|bV|W(> zJ5v2(GDM0@MN#g@*l7TLZAH0D3u)xrp_Fu}QqX*OplcU{lLquk-@E(zj|*^2pN=cy zg%Dz_c1Hh0!zh?fqoE<`BjAk-hbGZ)ybldtP&9_F7{`pnXN>6~kjEH#HDdlK9|vR8 zyunG8t&8ygQrVujDY=${B1KEOHW&k&(S^W#scsb6yW*SNZn1q|N8J`ECjz z;`$3190wM*L^WOlC|EGeozFj)0P#iDEF2Py{m{^!#X{tV%V@3M>Queezlaz8;TqybMn;a>5>^QU`Fhvf2&^4` zqPDD828(7Tr;3<&wm~K#=-OZIcN-r)_va#UCb5lTWCf{I8H>Bz$B`2Ld0b_;*hlDF6f)d6{xIz8{ z3*>o-$#~y6vQI%Ec3DTyGh`xE+JD6E^&q#;^`U7XbdK=aJlor{_k;irBWSltPymh) zw%S;`eu-M%Nnhm40q11VV#^I`r%nZ%SHR4(-b|(osgVitm)n42Vu0wT8DiS+V-GEm zRl=%T(7(>Yj|#Eagx~0tci;{G22MNd1+gEY$bwJS!0X9V&Cpp~eN5DW;aWiJ^5w$C zPe32LNcoNp&*=2&w_rkJ7EiPK^mHXOw3}eg(k>B-{W@i+IJ7H6WQ-dEB)TgYQZn=~ zK7^)$F?M8XFw{?=M`3~n6S(-KYWeCcSOp^UG!XM3;5H-xhc+3L5LmY&=mEHX5$7YP z$+Nm5EnJ1$kmlg%(#NSsI1}bWz%|JUXr_SFmXL!9Rj!gyBzo|zLe%l>iX$~R87EL) zhk*={2cPj&&~1JAu;T%(GYFBvh~P&I^mL+)V$f@;V3t&tMR{Sk8({rr%XKC;t~~=s zGpyKpxyj7>#Jm$6>M)i*8&d&xBa9F#P2!dOQ^Dp!sgJVv27pn3GYb|k4*DGIPJ>XO zSOMrD$ZiB-5;Qpj;lbDPe-y>@8LQ!l^x=_0ZBTonO%E9-d&aB*aEe9^mlh&wTFo%P zUjx-jbMaC8UCVB5d9l-J`=a?WXJ&=px(l%g*Yew*N}^`x%XdqNq^AL3oZ>p;Hx)It zMM_Hg8|SL1T=w!(6`eIrxM9d1i&SOQu5#ipfx-jhN#R)-!+3dWtaO{+F%0PB6AzR%(M0_A~)uyv1Lx^b}CRH1i? zSb)T^HBDV`Gp_TR6O&gqXQL<&&&;*1!J1Q#$ve3EI~yNUd-0W;-K)u=K^{f(_-&rDm%OW?xO=*^3jwm;e%`^^muI(6@(BvKu z&5UolPL*1_o(~xe$LyOSyl#4+&fq*NKCuABuX#7doidEuy4zeg`?@|K-{8lxUFC~` zv8Bjczxhd(y#u-_n~wam+R+XxxT&2z>3~g?0y%(C z;nv`Np%o+XnHhrYNFWK2BQ;4(qfQY$VkCDe>42N-tIG&tf0(xOPJjQJeK zEEasPfCK-d@U{78;hWcl$`B^Ip9Zb4S76Hd>L74c931Cm^KTj|7_pqxlAGV@B=t2J z*l55SgEWoa7>)BjZz{i5w)bt8lQ_RrXLR{Ng3IP|)Ovs?Yd+TYTk?tUU!6$8d*c|>=fXEWavFDqf)xX$iGwhKG2`I+_@Ld{T`@_R-LZYbFN ztnYw${7n&DO^PL+!5Apdzxb74XY^GReF|5hQW7~GF|97W3)W;k{AOSL9%0>Aej@xt zzAz|kzR`b|=toSO`LQZ6muN)otpeDG`p6gizx}9?uD$U2u)8rdF-#0VnRzFn>jP&P zvPw*^F8^)^3Yso*wenKP#Odu@6=Yrazhj>kT>PytLL|~DG9 zbIrn2%H~|}f8Dpx==-7YhOzw3VIruw@jPWQgC^S%p;zP^7YB-fpOF6hP+I zFYO2-PBVX_kP!ce<2{Zy64LY-iIJm!To(Z@=QJ}(-5EbL&Yzfp%W@gb1=cQEojZ$+ zq?(Y~0I0-g!9u}A5+;gIat(LiTR#oOl%Zy&$!PL$(4%dEPI!0#uUGj5VsM8O5*^8W zfHJVle`Q_+V`uvq{W%OmBtbw2(FBcAp43FieV7Sunamh9w~Y*g)r>I+YNc;K??^@6 zlYOvs)|E|~eRLA9*NcvN!*zqxEwI|&P_)}c<3Cdl&OS?to^?e9$3E&wuy0F4fA;jW zN&i_+Pv#uhyNX;3$Ex?cfrtNcN=k}q+uHfph4OB`JGM#J8YI%$m@yCy#o+9tXpv(G z<8xODP#mxMAEW~=iwB$JYyTLlw-ZNvy|gsk?tngX4QT+H&=B|8U5N?z73{?{==XVE z{wEgzA}7m%ugKI(r`-q4NmJe@1a6n09Rq<|l+Ppl;^>fNB>*rc27q7fr|=4kuDl5E zGO~st%oeJ$1#8yu0PIslstky*RTUKF33ftW3ZlrTT~T3na2VI|JZZjE_jI&hD(5Zf zmM!Had7Fzi3f^F<<;K_f)*Qyr%EcrFR~_zW*m5 zRtP8ys&$s3W3^i0nA!P_sVOd-14e>P?hH&A@Sdn}+oPw=oVk@y^^pjYNxYNl^G%_B%fXDrrbmwanju*r0Z{P?V9@}K1eT!)tw|^3HwO)y z7OW!$vMNO}fZDFqB)QNCRT%3OW~WjP_0*(Frf)xOkqHwm=K${20GzroL`cx~`MVx( zUjOg)czfSJ{7zF|1c>W87UC`dN3W{*d zeTjilW790;YlF|Of}lZ`-IFRMo$O2drO8oHKKAZd1t4cd|T&*nLId9%P zP!L}>u{pZu zc;JF|Y`Ky-SKc=AO={6R7p20Z-7lokmBBk7I@0FbhHxTTxN+m4|Fu9hv2871u3%?w z*o}sW;pml}-80V>0**NYA_l850ppy)!^6b*@>@5k#LZi{c?KPIc%OW?l%okZN)!x_ za4AJste6g8wmNszNP9g6b$fVJ+DB$3%QZZ6;0kOa9skU7B?KNCm|ocrh>9DMP6e2Z66j#Mfc(|Pq>Qd2dK7N`PY=$L z($YpyS!T|fJ)2Ao*ELL9HY!{iF&yfkQ4TyB9~8k!fWZi9Sj_IGAP>HVhB6pri9=^9fS-}xX9`=vganv~2*r)@ zli$54Y2_0d5aHQ^rjZ-g<0r~-C)UQj4IGR* zGa1q);nM`*i$VcI1a$CNyQ|FeL6nG;<0GsYd+dUn?^YWZEhFZ7drd$WbrCUVL|z|B zaE*_b#g;+19*QeW;ml9f-fZ9aH+dP?g$4|yjH)(O1*H-8Lk}8jKcAi4(foLdvlgDx zrx>ZgG{EqYH5tZ7sE=!YXnyw~G`4`FVG~MQC!SRX1EV#d!W6rWDXGQE?U>6G(hF!Z zNHnO}*!%u{f<3W&gE53kkz8Bxy2O&ffj~+1|Au+Xm5zdj1=Jt|pcKj>;xV8I=!Q*f z1N}ufe!#58Jn8@9g;bYQ0(Du0p@+EuzMmQr$7z)U1}Bg=H^MGn9`ON6Xh29LgD6BW zVZe)pOA*x1gns~*H(S2L!m4T#sP3NlLW}hUx;(~~{rhjJuFTEPC!am<+qYdX4)(pY za}O+&As?eiAkeuO9r7!U@gy$$Wj%)Dfk-Pj==9NY5Zw*9gkP?V&5uRk^DOz#o7E@#=Onx_pPL(tNI)emupw%Z1e!UMq55p10pyV(H2xZI>15q&`EU4@# ze*yswI(F?FC`B(dX=O4U5AZ3(CnNvJVs?!_dN44np$HmE0c-`mfxop{WsVo16})>bgJA)8d> zYKBHX`}hY%S;k{8gMtg-5K7-%f|PNwMusXfii#0|DB-+mDT1m zTqVTHMVGZ(q0R1|?<&4rWxHA%JBObxjm#6ZaN$L9@a@q`Ja#y9=i}Gio*bNR8I2JihrXU{Z-u-gh-t|?Tfma2}MTyTHl14UbP3b96RCMa;M7s>K4isR;jdN`c=5gJZ z>?Qg(Xe?>p>xRL0-6e&|gXSOQ}=5TwPF9=SyG=ZP9Vb>6kvWiyfZ7BM!_J>< zHjUF)GBm>Um?)Bw@MJt&G@gX(Cs>_V@n{II3PvtDNVWLXmmo0!dTDQ;^!t4xusFV{ z>pg0NW#zoPxT;9LFhY#ZRLC?~ujv->vojbk`vLac-9 zhs)Ig`6bbFz?j9%7)=(8gdH`ix(|7XdFM2a7_Xi?^kkgNR(HbBinz;|F#<1Ifbkc| z;-JKXmb-fXb-(D=gNiFK>;SA}#&=xS98N&wbB|X`#VLCvIpy zYC=ek@m%5`io-^Qk_tOugtu3=MhMKGPohHYL~EXjydjLs8c8ySb{ABKBNfp?MhWJR zBmcLa#5PL4R=%}O;>HqrKyT>B1V6>-iF*&rrV!vLzE{uX;1olLI zIi|w#)uj(ldoJB1Bt5BR33&MJU{gK2!T=NWSj46!?Lb+caTJL%86^7oX{20{(EJ_r znd+yAW`YNEr6C4Ti;sV}MRc*2F{G< z6QkXlj--uT!kgT&5(C4+xF?f!N>lzI>*U$3=sRy`n<4Ozjk!nwkI%bfYcc9pU}6wS z7}&e{DT6oVVHZbj0ijW3#Hv*xS^@wqm>7T^OmndZKhm|P z)iVTd5})zdJhc6`9NoGBOest&4{K}BVBoZZ_X_BifC}Ncy%{6P--rKdp(>hu?B&ZJ z)^wG`{&iRJhoAdnM7H#zdho|XJ3i8^&`n}Lya`R#TxSZ9 z&Lm`OgiP|qzgkXcwhi2$O zj5q*GYcj)xaD|c{a2lheBAFKMDg_vxc!XGE&zNV=Y-LR+9i}QZVjaIVLc2&TP(p1f zstlJf_JgN~M+PFGR>Nq}RMMPIM(p+!!(>%!U%JsAH4ho5_=q2F=)3oCW02@XykK)J z7-wYJjf}bO_bWs7fEGg)v>bk;*C@1s)<#@I8%n5ccqAyZ>w&x-A1vr-NCj(lxUYqQ z&jflxLDSvcl+6d5f(M6Wu4L*Hi3!~u)L&>!U%WBFtcc>+5wQ|`+GozME+FYRC<}{wQ=ki^=|-o16*j+ zm`;?26yp;gb^T99)g#+C^76$usUo-K9_y_CNyuOEsl{s@I7LA|vN6pLo;r>5V%X9w zbp_5%XaaxsR-C@fX4iSQxm`ZP4>bRRX444~r2NiL1zU2IFKy&jQRy9jW7T|g@9oj? z)Y{b%`ZkD2#l(Fz*E0otcMgLfaZvj#9sPeo4KJ84KQt8B>o`G9|2d!G2Jxk zfKnX6lSf)RDO?Z#Tjj!q#0wRu$|?OBt1zT_ukZG&JNB-S*gL#*xp$ZEZz{4jKfX)@ zh!NaAgg*P=$#V3>n3$Spi;8`9k@ewc0M>hqpKVPp>Sw1d3i>{qdvEnej(cz6gRSc| z=eIA-*PhH;Zlx!tud~{WbImcG)n$%FnCM`iZt{DK;b%uV3}MXB9aF+BR14beoTe1( z24ka)qM|u4r6J2x;Z4?$_xHi)f?_xY1qJm?O*thcCB?O4xy8Jns~aAArQxquQdTw> zWfU&f6f17L;qxWyul$7X62-q1h8t`>B)NHdlzx7|_poD-F1;e!gaaqCjI69a_A6^-ZwC$W9k|Uuqc0~J z{t}EWu+uf``+gYbP$QV|ue#-0D0vfhWodbN9k=tKkk}74;yk23uSVvT@tfV~S?Rk_ zONc5~3`XmCD6ss~QI~>(;Ru2f3xiY~Xm!)Gve@XMhU4#9nuLJ`4{wxAc%!X)6dE6) zFg~&l#EmAyj~_nJp!NcC?kM1J=dQs9OI!!*+qe$E+u`-P6{Z{{d_dG6_4}CkWMFhi zB#@}wjnK-XfDB#lXXE64hyUdvoq|QuEx~7qR%XL_REO^PH(&nsSNa*GfuntlC{@Wr z-&SJxqv;XKT2&A*ckbNzzFiAus&w|T<#S}YzQn$2npRT6hk(uHzlwny2xur75Guk~ zgJ;!hX=#U`9`FDR$?p7@cIa?O$th|ackj-kWI>F)z|1-d-+zr5Qz0~$lo4T5j_bm= zWXTfpAeKTc1h)i~Hq3C`%EDvFE8PF`^(#yf)yt6>$Gzv_rZn(bVgqq^(^o$I?bSFA zz^<0Ug{0`9N)Yls!rGwI83AkxWKKQq8a}5!2KmOd)$j3yf8Z$o-O$`Qd5CX(Y*=!5 zV&z=#xwNEv47)#_W3DsiT|z-d>F)&VP!>B5y*ku+FT&0c*TN;)#s9}8#SX83RSOlv zFtx0rA^-{yI3Zf9qQN4{~izt-*MY~AU)v(_hj0BBk0WcU?p#mZ$;WN#0k?%1R zXWD@X9lV`0YzH@w_U?0;P=+KAcgVSDJvPwSFL<^`#AU?b*s;qz3%At)Vxnwb00vR# za+BeF1iiklrEN&ZM)9WgIA)C$DS~_y?&;%aa+URmTLbir&eSSFl9Y`~G zq*BfxRHwgvI|Z502wtM>Qvx5c;ln|s0kJx_^V_grVyl`4fCRU1`pT6nv42q}2hH0- z10se8{>zG28K~^T!C5RTFHigYc{YPAfFNkC!$(#D#?WL4D8m`s-1)YWT_?8~LtalI zq)|~D=C3z$g!mPD&e^nJP*N5?-@7oX2e%n(Ni@nRWp4nnrn zISozC6-D>qzy=*9CkF?|`ZF~XT2Pq40QI9dE*HfYsNoH8G~$FbJ9kbA^@I1_yI5+nwCK}DrM=I7bF>kX7jgcGe^x9;YiVQ3xA;MYua2ar#35N`wM#i+xstYT&smYzA& zHOFK3?%fno_vu80uj*IC$1%_Z5Y-p$;mX^$<$FpqG6K^hz)C)_OskJ!03Hqq;25z9 z#b$8t_4~jPDl2U?tBvU~;u0s=wVytHvFEMJWVMLM*NsNANL4-^xTrLXUUQ&ASR=XL zbF(rw4*wy(Uuz@@YouG4VcAhN4KW5mcRIg*eOe$=5P9aPd*b7MO-!=V(T5@5Z%Hi! zVRCow+^Iv-%8%=I<(LF?Vc*h6@(x{OphlejYD7q)TMNBqHg>TtsKyKsalv@}BS;gK zFRWOWky07Ju^vz~QhH!dbpwfc1Z>eAwUlL6W8F9tCx^&H3?IHk;%Hyx4>_L|E74Mj zTpU<{8zrG@3T6WBm~aLNuj<^Us=5$7d;Ac>3G0Ii1|0fv7}I574@L(a+?#~+X*OiE znZRzl@q9NTh}aqBBkd|ea!=Ky@ca7Vb8jx$m5$#axhWBL@fWtDbV6J3VVR1`h7wG( zU{2xzgS%z+eFNPFkTJ1hd8mS%fd07vM))Xd%Id=Nb8@D_#8YjM)06f+Ozz;D$JW&_ zFA089*VAur1QVEU4U`TAPkx(y9-lNl0-h~sY8o3GS#Tv$7@T!m5FqQvuVZLhu&9J3 z6`RJ<w3!LFDQ3Sw!K4aa{TV0T#9ysi+^iai4+) zF>oN0Q|Q`m?RgU(o`xpN4YW5*$;e9`uH`o!I*=lXvSZ~Y-BCOyZ$aU~b9T7Zq!X!S z=|Z$%GSJpzvvbGtB25Vp>O+IoP-W0lOXtRE6O*RTFhW9#xf`uSdHEx)cAT7?cVx~6 zq`Sj8Ac+93Ww%oxwPWZNJXZX15=<{j=z*5ylb->M#qKCLB#+cypTFS}wlb*tqG}Yt z40SOC$>hA|wgWGwwQ1aqe%@NSSE~X$T*RvL=b<3Lmy0wm-f@%|dz1`7(88J8WK7R) zTIG@M5;!&NW`BRE{L)wsCQ45D=0(AWxGd;I6~C zt!Ty(Ss#{eA&%;#zXwwQKLaTP9s{0A<`teN1`}3m)BVkH{9?+ro@>Gb@`1vtCw z$0bk&biH5fz-Sb#R-GS8IODF5&kRdRjV`d#S$#}gU#CD^60D*5k*n6XvKv*#kbg(s zs0(WIK_QlJ4elD^C2clO5ClE<#C<@K;`&yk(jQ@6e8$qIu)W^#S$x&1!k9C< z#AL-8q&C;}ybyGIu0`kBQ7PZiOk@08W4&RCxyp^Q61SJC273x9F&V~B!E5rla*1l9i zsU4-8eHF79WyI7Dv_D^E0xPaEK3O_m9r^j!-#Wr|_<{(+6w-SE(=c0~ExvNv7X81v z7T}Py?CfXhP2@ZV+0a3kCS@Ibk1?1kq6dGumYq2@)rqCPrNw<!8!zVKE1&#+Xco1r+UQj1=7Y|Y+U8~=a- zAvl{7M-4}EHfo~+^ev!==#vr;#_+fuO`}i2_~b#95PEVik+$6EVvHvZ?Tr4#f9?A9 z+>bN2ZQHh5N~&a|WX{lq4TI8Z0{VNgzyS)*^>}$$=lQ#(!l#sV2ozee;*#33TMnwV z_y`7eM~3+Dx}?t!+UEcP#I50+74~0TeAyF?jo}Tauhnw%rN-Mn>kT%L7?P(9v(uJ~ zH)^-~`SDjy!a4!fD>2w?AzO|}oNnKbrlwXlWuQ+JSj{xD=SdF7{ zcw!FJln+NBcWr;{D5chTuqYUMr ztMCF(pFZW~+s)*v~~^YW6J8gDiB3sfi_5ZAX$yq74NFuMr%esi*aCq8Q; z@@-)|R9=1*tNn+73EGQr5Z%`s7fr>(MyK$~>X-s4$vgM&D=%6x?m8!lb35+WJGny4 ziVtW>Y}%yV&MPn^TDeANi9mJkyLWrzoLk?grw3Fuy%+ZN-+{f3AVNGXgPfmHr{e7A z&C1F$ZrgX$0*8h?jq_3D)#I!KmPTbZ`FB82SqH(`ep~;=s5KYJZBes25gdJJChSy@ z{>dd9T8p7y7rOoWpQXCl*gtM{jNKNhYi?>vLvQ=^=~YjEVu|CxcR}kiJ(rD*ts!N6 zG&J$_VOmYPid$8lcqTqJF)i32iJ~$TbP1duM^U^@7uv>{yTN!4`Dmb+BDY^U{!z}O z*L>Dh);==A&wts*+By|A4^Dq15E)K%>Bj%!U;pv{QT8VASf}m#@QtWJN!d!4v?HQY zNeGo`CtG$&i!3R7wla!{Hfd8*vX!N>Bx{nQhC;IM*^=x#?{S)EW_o__=l6f#znRZG z&otcb`}@7V*L9x9ah%67`+S>0Le_N0DuEa7#_T>9qYbT>+nSwk-@pG2MJ>f8ihnnN zL!D}E%9_K*#Z_KfsHZ?8Fc^@pYr2YXGsv2b#{RKP;6-;V;Zr}d|K9CjE4-R(zOeILjZ4jqjbcH09Y6HSsBE7^d8DKFBh;yOD>^)F59}uL;{es^f}$^Y zIM_o;gcRIB4-A6%_Q;rV_^_umozNU?FO?ZHgvk;DbEbS$mVcwt*7 z3<;;7<}H*|4Y-=*Y+?$FEBhYd>1sWFbdt~@6L$`Wpepm}i>iB{ppWF_WLVtmrv0{K zIkG;{=|&~knps;v?LEKHy_Gd%w)(amJ1ArTS63BII9fN_%i~3;DnK7klX9W*4zln5 zv`2%5Qb6VSD5OA)>q=`L`QzbShwb6GEBZ~wN<%-r8){-Xz-EAS1=?ngY(E|r7hY8@ zPGN_U`|fW?h#)n#3~9DV#9A^i-bzEo=<>Nl6xZoZ%;jEiH8OIA7k?4sOY7M`@KUgq z7ndwY%4zV-c_@Li#(Oyc3??Au_tgEQ8kmC%ELhNh3%N7&Ejm4DCje7DcuhXi>=uPM zD9@(H`4TFmNQ<=U#c`JT^XJn!j!sp}S)R-eF?UP7IwKo7T$HXF(c%tYwN!)}!J9j} zw{g>SH|(552VPiFrjMTF5zlF$={QQM+UY7g5v~$YB{U_K8(-U}bY@VJN@eV>+Fw&( z(vX~PYxXWZEQM)RMbqeEZ=o1$1i9tq{GeVyXm0rGr3JWtbf`PNxFOBCM-0LvW&?w& zQ#*tmjBK(zCU|Tvy#mOBrnKSv_b0NS3QJHmzOWceR`)vA z+5ik>Z*Dmw=P};1gL*OYB13E`!ErRahmH9ycB&*}{=cs4z~9#uFwtMQ)E>g6gq?ft zWn*(UC0Yd!C;95*o|%Hl!Iqs3wa<$V0|7DVHVb8OtSy+DtdW&fL1KKt>MEd908>Rl zrQ$JJ|L)K0ckcL;te3FJyP=^3Q%;Nx!Jfx+$~_`Sz&+u`3rYpWR>z7_X#yMm3PfN~ z=F9f3CG7B3rO-|sa;6x8B)rPc?>9h~i7>uCOzBiK0E#E&DcdD%{-5HEHXCNRYG($b z1_x;V7UgMA^ZWV3qoXnIQ+(VS4o8n3or%g2;>uM)jI zQ&YJ3grbpy6k3xT`-b}asm0-q+o_@ZrKZLxT9QG2h&a{6E54Ubyt*le_G{G)c1Sm*E4c=3AR-uHH$LJl0n9MnxiZ)K>M zfrT+^&YT-cWlz&HsRs@oDUK)`E*fLDirAgrQPxNny1BkP0L3+`dhuTpL>9mv#5CEn+jfK(#oEh~9SG zoqU>v#dsv~*GMyG1xWg&O#aWCGAtCvOxGS*u1^t6^@xu?hpA}G7if}bg<)PBsy9}K zDc<4NWa~Sc6}Yp(V~soA7pyxr7smb)Mwvjk2?3!Z4kjY-i=j>wR?9p2wW1`do*Z0@ z{N7a`$FSVb!cH9UWnrj~Z-l;|Z`j zFtZG?#?}i_YeYL#)j23#c^*I2OH`Sh@|+x5K?{{I?T;U40}R#d7>~Yv0h|t%5I}#? z{|XHb4kFj@s#4Y|Kz?xQx$0)z^7Ivs=>i8Qd;2#S3L|?fK&DIJ>P3(#WV7`R4I0ao zPMJTD>B#yPK4>GuNU(o)WZhD+r({t{n*r@8_*$5;8t%1m{o38qA@04+ILKi5Fb;5q zU3dFDpiQNB_3BXJWGGf;%D1u_#E&LK6|L9+YUaE2v8#giw^|EXG|M(ge|4_R!KGoC z9>eWTE^K0r#!vGTYYiR1haB7SrGHila0Ja>>U=e)73smU`K zR87}eq~2D7)k=tAz@kl8kpl36L+fduO6MxYlP6D_!hxxexevZW6>S&qzf&WMW^L~| znOL}jSY88*mW!eXzu3ZJ+!IA0xlzCulkp&7)W}~GCcT*Kj!^Pbd;N9^_EY(r{OEtQ zzOg!Z;xV_%EwDj;7mRNKW2+TRG9Rskk z|9nNUmtACvM&p%TTKb}zDH>1AVtBn`aahFu60*as{XoCxog%F;AXVi5hT%3kq_wuo z=>ZLvhSNe!XgLHj6b%3lA{_clIBIPJ>-NH9g%+q1V#?rJGIP)EbsEE%PdiG=z}s@x z>^O)<`agWwjX#@7xHoV_Gh16OKs@Py>;WssoV*2qU>LwTCa`;QW{PCj)*9RPEe9F% z_t)p^cGeY_@kFrfpSl+X(h@z&1N?r=R>B}yIrvD{ob7)$jt*AfCc}Vd;k&rbI3BU* z*h1l~y{mmKG7(T&(SfGe1;+lSob^INqyXBiv#AyUUxV{x1}Sx+Qw({P4r~^C_)HWx z;O(>p0zNzZ*VV6_`{l-Lc;$0^`1ldua~80kZo_kc@o?C0Br~qW#15HMax1HO#ZL$x zzF{UlH6t5`LJ!aLPcQJdov$x849n0gdn!bU%)W zXn+;5Djp6RY>3F8Zz#~!h6c2mX zSaBq8b2tHyBmIU}jOMhahXuxMx-h0Rp za4=dvVD`#zX_^?*1ryy?&tNWW{*1akYALj>P!YN=|Qv8;~GBJ-sv({)CXf0PP zz0I@=o%`_cF!)4f2#&n&+_^KpwnAuMhU}5yslJ#hBS}78 zfA^q7U;Gs{m`n&b?q7$Sf!5QagsN8y*n=P+1HN47{yXv1>q=dI{w)HF2m(7z~g&Zlmpy;#t>_ zR!uqVAi&7CMr+RY<7>M?oHPOI)Q#E!mMJt@3B3_I>t2g;t@E{9G2;7oV*csi!Nk?Z ziCFFc;vNDMvhvs6Ey<_4bY9tTSFx_aET7k>;*;mint`&k`1rvVp(lC<2Du+T1dJ(7 z~% zX)smb*&h9w;u8^J;ju z?12>?D;9YV953Nyg9`a9pp8|9J##Vc2Flpig@wCXoG95By@-IIAZf*^6|HY<4DUxvdr@hR(}S;PpR5eT zONI8GtMuj(Iz8v_rzh^P6U%S7!243mBGUI8SAWS*%#7ifB@uo?s&O)2A5)9d*Esyf zqR53YTQI1U(|1miul;sbjOrN~fR!?QWtoGQC!}IJdyY92axhE-^xuGk#zvk85mrmL z;>Cb~8T%*EmZfaa*VBtyZEkHnTRAHwBF2?d6RaP_4i1lvJ;Y;0G9}c~B=t6fsTPa| z;tGT;Ak`K?MS(+v=ZEG0*y_?`5A;@56@=rna_0_^nN7Np{{xFciU$0ABUE)T0-+ z;oJ{=o%Q_rw2L8F-NHyvaAxR>hoMQMpTtW=I3Oi%#?Quz4Ak??4De~(@Dr0Sy7SNB z>FcYmn`EO5gY|Hnn*?ma_JNJjh70afhvW$|x&_93T{wcA&XCdi5L*x}=VR_T6HvW` zXdlGBN)!a|W1w)N;MqvqroL43cF6Yy0WBH1F({yLy5wP3gADjlxy!GI@^Mu?!3{AG znoxlcxjvc{7L>=&UU#R1WTnFj&R*Eee8Eo<5e=OeY5O7XfGtsMPa1dt9Ew|4V@r8q zUB!p9`ZdUwOO1_(eE5A18W=pp>LD)vKkXv;2T?SjkXyXPR7ibiC&#W`g{ML`!Pto) z0ib_vfKss2Mx^}r+loi|c&t3n`AB`(7>US zkL>DES`iM^uHjB9Q}odj?2$~_@D-5({Q=yXg_V`{SZNeDI2vqSiBO69JPQmABz+|2 zI8IJYx`PG^$5RCb*#ZRx1)c(MF(%eEc-C_3>P$N6N@??APaaufQz(9hUxu5zObEEb z!YwbiYR_3L7G6mrD@R91ea|{VFD6Q?g-xo$tIvp~ew~y7PZ@Rs~);RhZ z7C4G}Auz|0LtSf|>4U%M68#dZ8^jA&E{5@4=ffx=#TGo+|Bl3P28ZNhz19)v0bjUd z6B8)OvB#AU#|j{)(A3)6=wS-w(ZHwW_3LxiJr|+B#4JsALBU1*3oqCjNxt&wqX8zc zT6*6o#jn!YL{NdxFrS%4& z^PA+AH06R$0mqA%vS?Ia)+GSJpjv|szF|9(Z(*ITGg!?v9F_;~L=nz~dxkSfzK#z) z8*l9_^QFN%BSd5VK*uS%Y}vL}uiwJn&fJx#XV_#ra>PI@J{dL7dPTdk^U1`mK#bfI zbg|^90sIl8FU`l^4P`%=yT9IhC4Do!{rxMSl%6?`ss~i8YFAS~P`dNzA7!WcAp!rd zd=61t2(dvK9y|U7cm(jcfNhvY;Is1AuM)Q6ut4>H9!ag0MVtOR>w(SUgE;QRzhA12 zUOXri|M^QvzR-$56{h8}{E9lNjNF}KVlk@d9KfnS9y~oq zT_XZe&2QL1qNpeF_KliDsxRpBO+iNV_5u8E#Pf0f;lm{o{`ilnr(5SxeG2v&2?jt8 zUIS#~L~F3E9d<{Ti6z0hP{Ih<7Cq_#73*&*_IZEbm|a-t-=(g$ zJ~`;Y39IOiul7Sfj8L!;g^xNY{)u~x_#n{EBv4zAA|qKC-@kvaZ*JCVR}2RJ1aQK# zI#D9%d9)nXk;dviTt^ai14_mw2L~Q5)R>(gFz%NnLSzj|fq+w6rg*^dGY7LeKQX3s z2$!zf%uoF((m{07zT=k8K}YiT`@-G#N6aS2M^c@-g>j}-1gdG7j|rr`y%7_*hs2FG z)u}|zXI}MJi;H`6p`efZtwQ&WIx4{8KlfaD}DNTbZ| zy4{_D-_?>-4;>N%C#hv~>$bgOK+pA#d9{nTX%*7+8Usa)H>^m=0UyAqUuKul&) z9m8Y$-&%mwj|#Y7_W|f4rji@NL=1@Tvwn##(xsxq<9m$Mz21#?ESJz*@dp$<6RQ1$ z=g%pE0EfjD0LqZh$sd@B?$0|m_QWN(5OzQmxi>#LE)^S;;p8 zEggez#D$>JaD|rjVIRxs1!ddQRWC;IZGya#5eAJRUf&x)9@#;R_8*^yb%ze?ZZgjD z5Yo}nIR*&?0Jl#u+7c1ZND+ole;K6{G>EtQ!`Qwjzm_`w~g&sZT=RqAE zKAPLZZzly{2qGr;p;F#VjxY)y;{pVaL3p$V0kJ3(7!1_f-XKgE(#F+e$Ypr|D_Q5x zrzs2uls#L~@6Zf)#6bDs&4n??Zar(Ks&)%Mx^i({t zk7SQz#S9PjKHIZQ1U@|2+BR@c~JPfvT;ObHFYu7AFTWYc6kmFD)v`={G)Va>14@g@*G!omo&BK1O;FQ#;Yn1DSUH(G)C zls9SN=|S;CDhdoY5ori~+neM`>6eY~OkR~sG%0a5tZ>uN7?asP|H=ah2f|Pvg8_mm z2HQe8t0>u-Omwwjis!de=#}L>X^D>3W1mG?P-{#aYs;Q~`eV4M$RAUMn* zt-nOm7})xz^#p3Bh`ffs*_xB}$`Nzmxj>mb7{Y26dV<2z4R$gZOFZf5xE&1uoms%V zmHmP4#7Stpcz`L?A@DrrR^l^%$Ipn2FXCM(xOa04jtJXqs{cqSwNlgj94i83l zMa2SKGc~~3qOHNb5yn=P{rU4eN-Y6UUfU{jiZAfz=r}YOd(qY+qy{?*`6~g6pgjhPKS&Gx@d+eh+<=AK@-`Q{yW7M- zIkdSxw{9)Cz3wmXmRemd16JeA z{MU*`AGlaSVG*YP+4HAlG3x|4Jo`j^#^e55X?@#ozbU*py6JZR%mV+&jxXBR`cB^# zK01HFLR52>e!%;H|Am8c>Rnp)a0eI3MuBG$M~UG9-pu(}qgX%Lpe=BO(!%xP*FESq zRO3V+5OP&EbZap5GDFLhkd%~%J?~SpwSmi_JK!F$re>R%)Sa5?SkMOr1FQ;WA=LcV zc+6W|dr9MT0pjAptjQpLj-~@hZhxjb;{wVnGjJfhe;Qot(uzp8P;}{?L2g^%{%%Ma ziSMh2x5_X?kajHvyks(sZp)`~pu|F$B=*=vXX3dW+4>3S$aoBvh z<|i{vr^tB>xCoB;J;7~WfX*5)O6`6tsov3Mw(UPZS{R3AAgkV1E)KEeXz+*N~G{I}2a|K$?j6 zcnHZ4I8<2}SXPHXO+jd!3!odtlgQISKd4|A!<&_qgn$k}4>2)-4`f3zaeePjOMU*} zkMCQAuHoSlf^d>p!DjEQhJntcXAed!yuF8j zF|y?AcY5Wh=+hpR6`WW-v%a>#{@R##hi{T)@lmfHShe9sUMuKZ}bB z74PM3vBR=4X0Wmz&zQu{#qa`fdq8mya%0g!Q4P?XqHsd>T32Z)AW6tBWdI-eGw_m$ii84+&dBo2cT%gqBWcPpPC-G>5?4P8Hjfo;LX_>!OBR*XspzwxkcOw1f*o4);MG1_V~ zHe-+^+TP80(9S(LqOd16#v^d?7)#~7`z1+1s*j`>T^5UBCE@78h3pWHatf?bwrIuJ z=7C;=3rrXvW6)8L-Ka9fHhx_Z_-8PLi$JX6bj^b|wyBxfS!^s4!6@`r1Lh!WB5bqA zlV+nbWH3mzj{Y-bn%akh!dbK2^T*el>`6W#2w^d9rg&#wP7V`#%&mafQ3;w|b^&(3 z{kT3D0W_sN?7=^t73Yt}b`;0TH%M0T1KYhWKZ>qs=a>|C28mfX7s#r0)fGBb*n^}e z|9DD@0klYG2U-)I1=5VDfOr&*G}Hx!7EoJt!Q|?IueQ?YYXuW86@s6hQ56D* zTQ(rGt_*s2MuGSCXHJG07;V?Lja3XfP*@ntSSKI%W!d0~`9T9^8%zV1hsTJ%a$db| zotBv^78(k*VhC7ve%uk#r0X!TNOfSQ{U)Wz6CxMXO{QBVnyaAPLO?2{1O0graJv`3 z^s6k#uG`~FYq5PX7?8Gzx3+)(E}1316zX~$0{8CSv$S<`#3l)763X#tZIv#;&q97Z9wo<`_x(akfal|fum;7K~o{!M4HRashr#Lc$U)uN$L~Zpg zEk2M9MX#=cHa@uI4vK;Lyrzrxm;SyZCFO-*V2kF5{ZJHhDFn|TL|8}X#@@c`NjoZH zn+i{zJn0o296UZFOwQrKuDzgq&+AF*xQgtw=uCd*U>nuMCKi*Xc>Xa$+~)vaBbX7F z0uKYC%U(m(MHLnhG|P;qqY8VUS{biD{Jng+aOZQ|Y=J#$+7D1&l3E5r5pox2Obw-G zg1+1_Im7p2+o^GV4W%7*6K6p;KP@V1kHr%t0kEKq8G^aqKr@W7RGz|Fd)=`)+icfV5h%a<<<vZ93&3e`t}n)t@|+mpN7r=`g`9Ew zNPqBy_l}k^fR^y{cvLL8R$`gd*Vl&)Hb*Od>k%(E3GUr{cc`n1ACZZ0OPPVEx0`=Dx+g~IZub9FA(`l5VBxI#Q5HlEeZ<7 z*ij%e3%&zk-say7jYIVf3<~${v^>C{)_k~PGW^HQ2jT7ye!aT8_w4xu6%tLrp`F}u zU(tOob`U?c@Sn#vTDNy@)HWKwR`$Dw#+A1d%l4RO$@S>ncV~HfOe}^S5-ni0Gxy$8 zSmhb{;QLoct9BA_{n=H_{C9(Fft_+G7HS1dtFs{13x(KI#nN%w!gCd~(B94|uWDrq$GQioN zwES_f1bZxw2*7e*g+VcJU_?=&uJ^R)a}Mr|-?PILl@O+>a$)l8jYLBU#-Lm3 zq|!&;7~xtpV1?TjQu*o;Pza3q&x~k?&~L_1Pu+=(0%z@Oha?y@@YK`OLwbBPt-O2F zLKR#zBbinmG``kS6N2Gg@#YiOohns#CC^CvrlrRgxmo%CXA#RmG#7*p4|!dj)T6^y z4FJMGZ`)CKp-4rKalMxLOAdE$k{mOBS}+*0?$Si%8*}&8z0=njvkBBe;q-67ipMch z8gG*IdPs4>=y2Ko2`2aZ_m{Zx4$oN?zEBFJB!EJQVpPx>G<0;t^lt#l4G~`x_%faCXFa8x?=b$qhaDszx!P%i2Z3XWEd=IG~N^`_luhtg##QZ`; zb0#!BJ|SY8H*aoIni}DPs|p-fW&+ESNPm*~Y+c1(Ylx9Bbz{(#GtEF@yzTm#WselU zJ$M>poQ$axny&@oK(im%*>ELR4Sh0Hdvpt?Rmer6oeQq&Td=B0mW*yJN;3XOUhgwE z4Y{~1+=ZGZgNk$~gqe|+9q80fT`x$$JJOA&2kL@5JzV)pb{G0C3pmBhz$u?f+$#NAp@6+{b*;o%_RxN{lcM;7k<-s zp?Uf)?1bWqiV18dgma*$>BsYmgFE=%h0Gw6*%h?n7il=;FyTJzbqtH|usClqUk8SP zOyhOCJQ3;%2gS<}q@y=jxvU7CD2vzlvddt{&Othg0xY_JmLKSGG4Q&$EN4-9x!ub4 zZ*9p`=ne@vYH4{9geT!1;+x|jz@vjxccD(dfVg$?bNtK!9chqmcQjZ01cp!3yHIOL z#`A0U%Wzq@Kl@%#nk9~kp7<{dR95?*?h@)ev~y5dZPv78DIfv zBn&)D%Il+$0%73#bF|cpkME_W;~~{w)%^|LY`EO=o3 zneoEa@P1lLd(*}NTo?`4-I_puh#teF1^~rvP5)`=o>>2oG@uGuD;Sg*Dn1|FY;yoT z*q_%&?5W#&KR>>5aHjY*uJv1#c4&2Zq`zm21W17!g;m@#V_20P2S$m#+WKcpap zL&2Xs+7A6Q!fRZ^P;Abevq%|dB8GRDc1vQ^b0NC?*O-%}8CfP4Hi&Ns`KK8` zyb~1nD^2cw-_%P3ETN%7D7k_Ng2Gls6Cr4X)1LsYm$z^8LitVhY7~hE;SebTpdD#K zXMihoVZ0*!8UKM=z>ZiA=oMEFZOgrF6)E3wV-vuj?^P*Wc%$U855$kASO7!O>@1EL zO86@XO~G@w+WN-R;}cmQ;C+Vj2Xjt}c(@4D>dD`&6^Ly}XIP~cAPfy))4G8T0GWRk znB4o()a1Qb^*kAzVVE)Yv+XDj8geRGgDESb5^?5C_Kwo5GzI|l-^9lF+TPaW=c>Zg zBRby($bH^IK*~YPdnprXOMdoMC}PFDdo;DMcn(jabt<^!zM!PDgYa*W-uAB zNJyv$Q;1JHR7&nXDS*NL%DOs$w4ahr=#oMVtltF`Q7D$hC$1jX?oR{~wgu1jB3&Tl zA^?Z;fC3PqI>HE&vNJHrk4EW;$}2d3InWS7U@+Z#Bj2{dyklLv^zr#)Nq+gazwLDx zcu&;>1sIT3AC9SdENXN*Clt8cg|paaFkc#Dx!tU~r>KZqdjl2SIeSCpS%IiQgB9f# zcjh29YyZKM0R<3q3qTakmpPtOVE@cqxMkzFN;OPT}@w-Uqs|0Xy^U$gDeCftWL>}bD6q$(!J!@%qoN-r@q&d904dGk6-;~|eCsbkUa zCIHs~k3Zko*toZdq!%=2122>al@*p4x^M;-61yio=Hl?h&dFR`Uc=--RW zj|NXmOPsEZkK*O%=vAQ3eZZ1g)|h-=M@9eokNh1S*JN({#%O{SGc`B&22ux2$&T35 zoTG{!WiOP%epf8*r=!4pY1!bdU0SWnMh*td%Xsio*F@pwHQw@(TdNOJ!XZghBzlPH(q>8XHHlm3su+_K&ny$4{ zKN|}^!k!h|D9Hhk@LJyK7^7J$yKEjwz{y>jz8|-g8lLUClkMihm7nb^DoZ^N0&J$D zA7szt{zdLFisY1s6&F>*HTa`O!ZVAxOo1G^>FmP*4>(C0hoOod+ZKdRou~GxWEMDE zv`ydB1Pm-;Bg(~IMiNVsDPT;6VQe@eFn>M^4iV0U>(o0fFu>CQrW>yUY@HQlzVVMj z|Mk%vdrm!*$Da^R1MamL>QxMwa)1IPoMi6Yxn4m^lX{V9<@xegkB}n;_N#DU58_Hn zIHvG4`oKgW;aSw@{Ms~O-3|$x&c2OFsJ`h9|Ni?{piQLIN^*K*`Pt<2^68uUe)^`S z*-$_c88i&q6krEw0G^Wv1peSbttD-J!`dZ4Sw{Ak^qp4IUJMu+od^ui_8XjM(bPJQ zoj}Z;GzXBqd11RS2QK;KvSk2^W|Bb#(<()|3;oAYhZ4O(9K*(q8&|pwA{4t3Q;meo zQ4Ox6Gdq`QRimAFlZ?heyr6K_JdK0F3ogqZSr0E)Z7dxcW=%gm1wYFKO%O}bKub(f z>4AIeJ&rSNqsREh>gsB00(9na_iPOAx#{cMb2Tb(nmLWyNpe-}APh$8t5@b=3ZL+Z zh3l_`F$&!*>{N5%FALI>wpSAM;f$j9dZ~6qMeR+K)?0HivA4UzE6=jPUIm>+5{C zz^y6AHFRDNw}{x-YFvBs(X%!FiRe~s+b?AxLKlU$Jmcz%6gHc}GvB3xr%z|GC`TAYlIbB`?IpMK3Xsj@Ji=XadJtuZSq zI=cPTEnwL2Ew~%HO^t^cPtvCrS+ZcozE*KiO>Hs7zd|k3ZGQ`3FGMHLp8NW$D4Uxv z!5n;XyR4+TM#?H4cDRbJ3@#9k&pq-SBc#ep=ds$GAR+>qPA?nSQNFr{p}Co7ychF1 zzJj&dlKLuc&P2*Nt?CNMqXG+pi@o`*aL*q|&893Ly*2zJXZ{l+|1MuelFunk_||SV zKefg(`UX2VHcIs#-5WVy?#}9Z^M+VwOdz~#>D`+@J5le0-C2YUnM$!&I={#Zk1g7# zvWyR9H=MFUvmzOv+@0GaLe{|ENH{AKF4bt>Q6n+ca3Sp-9K!52IXJ@Qw%dz|i(gD^ zJIdzG#OKcn<`5xR2jPZ|)H3Ewq0EbUa8G(=xExg*WeQK5W9At0Tu(&kWO3&JywaE^ ze8f%i8o_@5ICbmjJr_3V^qPjx+AAdaiNFay% z&gFbw4`*wf`N9z5A0C;*edWSX`DE$=d}<`HG2rb|^R%EE>IkN4mvj_!xhjeYaQ#JS zn!v`BVJ0<~N-bwjI_I&4$z-3=n^8KrS8&qen1H7FqYr!4cw{H_y1Z)M+vO@(!4i24X>{;OkGm#MnnM#1MO^+?bYArs32e6?fzxHVIm{)sQ*0K@aipr3e8=%V&2erkeEwzK zN7W5a=-QG6-RS7(-jGd>-vyRT`)d8?8EuXnPZN#M5^u1Veu;ABd&h>C16Rc zSJlJlvTO5|9oSbkuU1D|N&}tu*nt=9`Bx+(CZ@D>ur(C+Y80A0?EsJ79o`34DZ334 zJIIKW%>9Bo?c_r-sr#0ybVSO*-;}1jL4#2m1slw<^anrOQvMt~|8Fh8TF8ZRNL2x^ zHq3e$B7YPF5v19YaEQcPqG$@Tbb?;b2Of7Qs2~7lLg<)8XWD%^`|a=>2h7UL%S$0z zuw&o3lN~b-aEo@?j*a#2uPa4<6Yeu3VXaEOO5pTzQNdMyRKn!~<=<1R%H`sg0aJN0 zbih6U`vv6Oi(x>F$U8z_aZ(?`p&Pej(g>0ObXZ8jMqBaPmwO2!31=WP1&~`94BkO< z!`Bpw1ym4kK>PhfgiLn3eOlMU_%9Aw)@|y@e?s{Thg`fzOqwy^Fpv7G6N2XRNeb^YOj$I3lU0QhYEsMe1Lp!MqW|cv)e#8~z zyZ7g%xmKP(1NDW&0hZvolEvk)xdx@g}8Rfcl*+W7f+^k<(OtQbu9i_DtUFSjvt@7r@-KMUnx zOIhC2?nit1^Yq#qbJZrFAa7*$?A0H(Emn|%N2DxYYM;+W#E~|*!7LjKJ3iUrXf?M? z&wlmK?e+BUXi)}-M?Vu!(P--g8PYwIlRw<66#s(Zr!X)@Ps#8W5_=^_+OH~*1Rf}SK! zGp4#gQ(wf@c$ZmPNKTT(QM7G%7j#nfBAa&oDC^nBBVc~I&RzLW!)7H|_ae+0R?ipq z9j;HUSCj&+6Ud>qW$Cr|q9Lo~M?>AZ<0M+e*toTRkA@1ATfha1rA*Uj#(bJp^Y%zR$K>7U0Op}C;g zc8xnPho9Q#_6x!QmJkUXYh5!E9vbX-LFKAuwAxLrcZSXRPjn=35#e$R!5z&(}NtH)o$cq{R#s@JYDBksAZG^uTK%nEKq_ zNi((d3lBWry%l>1_#`7T_GJziZ7GyP8X%A_ngw@oaDdh{j||Ze_lA2)nIue}u`B-G z(eb|{Azsl60U}Dmh^C?~9p@wNd=DrZr_of!3%SIgt4f75@n;SD7%ZNt9Fh_=!eWxN zHL}DE2qA<*2`6qN_XOL^W!T#X?_|YRj6JybiRdr;+I{w$=dNRh2t4`8iBSid6=8V6 zTEk%^EzrZc+)Azqk}IZ9jwUbg^pKrCAd(@yHdT^ zsgI{%K}k?F>Cy)I>Q;c=MO}|aERBXR9su524XWBI*T%+1Qh&kH8V>v_@#RYp zObNp!9ATPui%v3|+CO5<^IICWZxJ50Tf4tVbn3kS9?k9BW{EbY7I!1C>$~Hb z?39Na4splM?NM9Sp_==eaYwxXb@$@tLyLLYZ!&xp1fHBcq^wup5-zYwscx}YTF6G8 zg{#n^d^Hbpxl3^zK+`WD2Zf?+X(^6dG8QSMQoo`{3<+4Q{W97nbr-*WU2|lvn3#yO z=4xdvo_wFlBWnaq6TixsGR(KCO`HfXcyHJ|6L#|f+fQVtcsDU?)Sub-`F-C_-<9gV z6|C;vpj>M7_hL~YlNS{!!Yj##(%E^~-IhWj$?)CK@NSin)R9M*=xp_K z|C?3AkFWWzu+Adv~AHKX9w1l>(O{Jq#V%lIjEAp#L0Cyd;qz!b4QRR$x9W zfeb1_)Yi z+t2q1VoFx3@oL@A58BSMIh@NPU{T?@X({)BJl~_yPuKP-{+5{9G`D}KX7?}aOf56} zoR*S4e320)6;k*ybTdD+6Wak+Z~oQImw66{CKDcHxY%-kpY_WF&8{w~&gzA#qzNT1 z3Y&bW7us&e;m820jNE5Lfhk95E`}Qzu%P|1Wy0m!h|C0mK=as$tKN<+6ylu=c<7!# z{IwF-omIg)*^R!5Dq27&h9!tC>G`}%nw(_NpkY7=N4_#}NI$7ZOtEAqaK2EZ)NhAL-eDzxvNP(H>ezB=S*du)Bw-H7M{d#D9eIbA{F0tIKvl zX*Y*~ev=?{?5F!NzXHG9lycfi2#`c_9#TLU3=%N|c_v8*CNbH3{@J>Hecb|Cqdoe+ z-LZAtXxsem%2#mA7y6*VXJX*pIPRJ6rv@)GMdW?p5v_)4ZFuDQ%Y17``A|*;4d1@y7y6Mo{%vUNYskd+lN-}MwBui6 ztsP#>w#^q5t{N(`NkMnoJ4`tPKi+R+m?r9hvr|>%vvvu8e%B_YDx>m7jZJHwrqU&* zp|_W&+sw9&J7B6R6(6j+SXl1P(-OO!c6-+?Qh$YXeb_enWcAGCmYDrP+nfS_|7T@= zJv~c)baoVFwc*i%%wh+2KWv@TSUJ$%f#KU|Vv+^d03NQ{d+(JIj0P}T>^PxhQ@xQK zu+ax^u{phF>F;8_HerMKT*1m;^D~dB>e{a9a@>qqi1>cw*>R?BoRMpn+jPInR8Zl> z@#AKaX+3Pn3e2{Wo4Kc26|*7jYiy<)VV~9$+QuiUN9vzFm!5~5HQ)>P--J$oKmr4> zf&y3>*WQ=50F)Sb2n&@o_TPt2KyE_}0?p5(eIm>sc+<^w$LB$w*wVA!i1+1>0h~8Y zAafliJll#FV{U7SynI@#w~p!-mVG?0QNQVKdX#eO;}IM}qXUYLJIu`=?yE%ERsAC- zBLssen$i6>hHdp915%pygbliQX5C7xu!@t8{bRJJ2v7#YN%7H!g_j`8=4`WN9uwz- z{ZEMpa5E&{8Cy%}DeC>UPWILlqiTfC;qPfI!Hyr{+1@!`u>G8qWx8l*56}zzL}(y~ zW^`$6*FSgeL&M3imoFFKfY*#(*KGbkOWZ6`Pu}P}YcvDQ?|D3B(@-9iws?l{cLEGI zNUw_-CRn#zJ(JMAb}K+lJk5CnQ#2`jMZH-&yz{&F9x?XWlWvKngR{;lXNC1?3#JuR zK(e9oAVK#n>t8D|zH862=~A@uer4vs>Un&0;J1Fk0Phc-cyapvaehWAC+fh}*BEXS4`XJWsEZ+KKbbcV6 zAcH_pgK2#pQm-O-i=^UE0`<%m`lZSl*sLSLe0gtVE`%g@!#~Z+*CAgK17C|_uYj~q zG=tC>9X|TYPiSN{328E^|5c9Q&jIJlB5bE>JlUWLNX%EjUjKZRcI#bcR|ca+s4}9= z&{dEK2_ke#`GL_F;4$cUu?GIfuQ1;#u1rH;m`j8?{2S;YTjgwt&cX=S=3ll&-*ILR z%?Z<~$%&z)amX$c0rICDhbJ_)mnDceu{?1C)eKo+0zbx4_uje!H2`?2GlnUKQd!*- zi+|a>(?qa$uYv6@gvJ<`XZxTikvA`Wy%`%1`!?kvcQjvkSXt6$1A+o}1{?uFkteLJ z(X$|#QVnO{Ql2#~2R6qC-C>LPyoG;D)J2Q@x!gWcQ^B%KpBbLx&S6Fm=Wuegp4~wg zC?k+{RYw)e-j#2bB=#^(<($b#6;4b%J7IPlzv%AX6}Xmkw$iDtM_Q`4=TtQHce`bi zQ*&={svo7X){71!apWbN?;er z?w@xKvjwBtBJCenJ2&{Jp!G2<8C1-gE%xB@%(>hezN1r0MQ0Dx_ge~0)je!?{_)A) zaXGifT<4^pLe-XKBUbJjy|df~l{O@p`}s1L+U`tlYV49=s$oO4yPw*;V(Y(9NoayW zSmNrOk`G0v$|T}ndoe%Ir(0BRcV2UQq~7J;2e4H@8CKh7e+yVLmE>@mPN?b!nhIS# zapEwmiYXuKg-Y-tK3KPhNWr0$##gtW|M;C}udo83)$g{tyUY>|GW7H3=dMUt;^%wr zO`WY_HgAo?{io8)C2xoBRO87Ox6y8@x-IgRTdj08-<{yl`h29fvEtT|GP2V%vc&Jk z4L$@!Qf^q*c#AY9ZYv$P{W`!EZu;h&mNXUA67lJuZPIFN@&SDK0#qUWC{(X@9}z4g zMLxJ~$2`si{c_@+pKghnNGuk2mX0Z@PE!;bW$IX z@lj#AqCNvrl97G)vqJ4*TTajoyi=XwWWvP2BKn$yG^6@W~v?c^AqW6VS%URN2b0vQM>XD{En^;?quN5=Fb`s*K8iV{W3 z@`?Kd27!g5@?d}JV-wJ-KxOEH#fmyWi~+?XslQ;ofHR80xRA^Aydt1({M&|&2%ut^ z;@~c}oVHv;Mgk%ltE^uEPJtp8SOp3ms#6figE|EwBR^Q8szzwasQ+>$nh)RrV#Z5i zqAzXINj{X)8F3yZ0IFNb!=66e=wD-O0u`>Xb3GIFb0}c3gX81T@5geVZdPDw$nb*a zu_<6R;0s+}VBpDx7lnRY6nf_D(7Gh>kH6G1Y`->|ZWmdjjl zfR}nRzl_A1Ot9nkQR(m#*|iJ?HZ{Y%l?>I_0(=sIx>GOW07%WigH|ZoUuz1 z8-aPlOyC_w99$?-CS`B!Hnxt9wYCqz(LR()kx?@ItR|9`aYhIl0@?v61EEz-1LHLl zE!K*IWvM4qCJU_HvWOVqNlyfJga&-Er_IfeN6H;`?U=Ag(x0#~6b-M(P7}s2K5Bnw z`L*{=*bwBTZti{zTEPpKf~Kg@l_s5;Z;#m=G)E*j2M~!urt1sqhxz*y8pO9VDO+2w z=44|5?YV_78BlUlq9`x8LF6;T@B&6vFTY@(+84Jn*&q|>7xHG}rY@jMdsXwJ&z}iL z_N-8*F}BtFlP1m|Tw_?x>MDpREV=+WeXqL@*$vqWIj*Z~IO&}9G%ndrsa3->f4V?> zxiZz9yG?|7OpL0y4f#HQvKyaulmvGF7a8Il=u>0Uo#e)=-qU7j_%qMJbxr?T?t!PZ z!{ce2tC~|!Mv9~@IW+ z*X)VVL{i`DW9!oPn7~8_AjSp2)dX^a-lY&27$B0lS&~5!&CJ=>DV9&6I97IcPSwp? z>+mY+ebXdcrI03?YNn|(1~36WDIW^!eD*fS8LE=FijsW0vQw4_9QssQc@cZm+eQ(< zG_CyHhh(nrjL@yJw6Pid{Z7j%m#n|4)R!wan&V-k!;Y7i^@=)AZ{p?mheHRwK_kkG z2y(LBPcadl+dq;hz?q-*>DV8ij!ox5N&YyLq)*^DwynP^Ty`JkLw5NhtJ!^2aHzy5 zpYA@2Y<43*_|%BU*uE76@Bs{h0&`?-+ z?8iz*T-bRZABTnGXTU<3K`?@DLD|Ym5@$2ES|42#cV3{!vWU_V_!SrF-Bi1{j0=Hx zUiUo5;=2N)k--zxS0(;j$Qo3$AjBq7v9YRl_}Gf z2O^aJ+ctdVjFDdu1vV;q);0x2gu*wuCg?)Dy*zT6Xnfvr&w<#oo9j>T0g8B_8I6aZ zxOKK)wsT+F6%=)#q=FoTi6QJc-Xr{!@Bv7Xk6Bs0LElsTt-?xfdIK0*2>A$>Jr0SQ zrqhcsKQn`QYxdN=Uswa~5fQ~_@%%KzBc0KIwQ5~t$#0>1XJQ**+(aY-7Eo!EH6dKM za$jGbUV+pZK{&d*`xbhg zA44&}evtG$9BB#qPr+aw->?N;EC_{RY+Dym__-uqYHmI}-s(BI!ei)FT?6>+x8TfY z*ZcdAEZ>MYhMPJq;mBXClO_GX)f#WJK`Zh^hP^4;0+=Mqr(rKIZJgyQ5FWPnIb z@C-7Blb{MdQL%L;2oi*Dq|Y%~>oM+3eLbB`ATIz%LTWV&MJTa;U_I;&6rsL_tL-1# zA8y44)9<LQvehO~L$cah&%RsQ&$7v)CWapqahKT&TboV+!U?^Ri>gZj34j!U^U zF3-9qK3A%*=9-|}OQ~w zqXRB1;nrAbX}K=PEM$h8pqh%ga>|2<2wsq>%mNEIq<@bP`@^Tu`PBS@%b6A5OAMFt zu+QavYMMWfo3+CsCg<~y(ZF&%wP#mFi8Jss#BKpG0|O~A!wW7gJr(3inu`t6bj zLF?8hADVoklzeA0>-2q+ z4fzwXvx^{Makx|d^(!5&G^-AEI>QTMiZEc)B*g+84WE>g4X;@Ny_;c4LUK-n5_oK( zS6VXfaYlkk`l|t8QKabk^KAHbnnPT>cD`lO?U_u!)LbKrI5eGJx{b&J zR>5dQza#4^JmDDqeevQ&D!V4eKX*V%1mpm3o?=e0>bE}pR1Y8sPw!bZv9owAz)Ip* z+d=y3*86@14iY3B+&eLW6gC23!QjD@hX$LGu4;&pQZh88;tT3iocWM8nnRx zd)Xeih(f7BQi8rS!-%-ewrEr`*jIoks6N$M0&7atlJI6C`7(q$-*x(bxsPJKK4FsT z*aHnS-CmYhK&bAXobi9b4l<#*?a;X_|Nr70vqM$BLRO@K1aWc^$1$8*m8>+i>7_Vl>`0BKn>fzI;}cyaB$CggQZuj=3~}oWcqM*x0xHkU(W*mzHDt(=uaXaq-}Wn^ie_3A3$sbic?s zlnC0_abw+)=d6b(mv>n6`RA=@Wj%dm3LAd_>kqesaSuC3lI?Pits6C}7Mimo>-+q8 z|F3M{!s{=QWTMEHf#lO^7BQu&3%QdS^PEhKY=nB-2V}ZasMU z>dpuoRK8hv~tW~j_*s!JH#9Bmi4z5 zfJq~P+fU6nlpmN~E^xp@iOS2bC5=t!26A)(_sm+vk0biNh=0z>af7Ei=&wM2fhnBz zmq6y~-m$GKdD0YT_ATchcssy1joqB<`N@h?8`;ItT9rT6j_#k%?Cj^TQohXh2ol<# z$l0?``+!8;_UGrcj*7I^bbj7n@r4L`#Hao1Zs~WE< z^I_LO-~ac8u#%$-5v|46FSy$zFMP9z0!YUAup zNW^B$LF0kHF`)b!+`>An4-u|35fg^Z$4U z=Sq!89l7cBxBA%l%-1tT-0E*DLYV>DlMNV8capXA|35C!I4GG$^2wM9A{00(7{5x6 z*GoM2--o5)enbcZ*?6@)05(G`A$lpaL>cF922g%hh-O!}@i) zOtXH2_c5sISVE?t4xKv0L}Iswu!pqHYakkCWOsWEZN97MUg_YZb+iih1Hxl8{q?`r zp8<1)f zaKsUUUsARO_2FEC2X6g<+zbr;95{f|zf#&gcE9@1O3y#BJZ5mMLq*S!^fcGco;q)a z4gmEsP|IJmc)n34tBvJl>QCd58Tqs5zF=YJX0dQl?q1etR5}2upyizfkW4|$^-jt8 zFK^eLe33jgn54JRC!sCz`x3G6x#pJgvv0o2l-MDBgKOOFFH2?woEy{zmu9J1^a9w4QU|3k?o&N0-v09^1RmN z#Hy~Sr%$)xY}$;t!GFHSnY*F>{r4-}7_O}fgfLN^2ABsY4Vh73%YZsU`9G{36z)l9 zLIH1W-O2=WO2jfVf_M6Wfj@lYh{{u)6a#pmv*CFrF#)-p00%UO5e``tL(KB3g-rGq zA!QOWbfiFMgFE!4Hi%1cSd>8^_w+7V-%%Dz@HQ~KpAg{>Ig9)Xh@-c=KJ!0SBS;z_ z%pU>9GX>)YR1TsCg5godH0fNAiHqBgaEA{^)?zQjqC(570p5!~*oKC7y&&5I-rU-0>+C=8)ci))$c%$nrD=OLmRmjEzT=EyfeXvuh>i_XPUNrPdA zY4*P{ia&oohbnxDtn9P=vHybmI3(2_8o|cq?DHHwK*kg*4WW^O$QS$Hin)`mLwAu0 zLNzid1Tlz68I*qwLpl-2aq*&xN4&D>hGXHdAg?wRq z?Kex*p)zyMI!mGg;QO#G$0ehb@7mbQz}a7%@tQdOBTL!`Qj2f+tN2w|omZ~+Y?UV- zuBl;D5$l1tUOB#+P0J5n?s)TeHHdje7ehG=nm`Q;+YU5hr?%KgqquR1JRo>!JH?41 zcllY3E0AiSi9OJ+2w|#NK+$buOiavN5s{z^!md#@y)X`bgIRlGx9uzLjaBSmq$0rp zxh!3yo`6-HAv+B#?>ysi)$x>aWtZcBu8VhuPylOHwqyTZ!8BLz<0(~;fsH=N`f%hq zf=Kd}@Ly)wNNc}R(w-!WhHy8auLz2sE>OEp+IgOQi?0u;(n$6hh2_k z65F2YFbGBW-nD43NmT*Y+7KsoA{9mDhZX0k#y`n71cVYY`v?6JX z_z%n8(ycmfZN2TJTXU_ucyIyq$lSYZyuL(-o|~O$u<%)CKMqv7!zg58f7;_ z)+nJOOG@^rNRbjDE@O$YW=oQy(4vycPP9rXjiNwcc=?vHuR4Apu5 z&hK&@pY1?992gvXxTIl~U#_!8TUVX&^<}eGN6U!5>RFw@ncsH--GZHkyYRbsu?6TJ za~E1cmIQCCabB7krb->g=}Lhu8i@l%k#q_yg2=+1F3sa(kN$H-d3iav{H5h^L90o) zuP7=eZt~b$Qsn>n+7ooC^DAXeR(2}3;)ujfZ^Ksy1jdvfZVp(o<>{MSooe50 zU29m2k!2Q{%fdL5(!gv&T$>0{V~C*0+8>%(f^fUBf@1X3ntxqb8$5i9KWxGR4dr^0 z_8F!~Fk&=au$RqLRHj;5`d~5vZEv7By0K~rES+mC@|dLO&)NHnuqPN#0%-1;o4DcPah9 z-gZ|?*l|RNQKXjva$bLMm;;&goU%ug+Ip8g99k;$($`DX*Ukd7A0pW8ytarl@rj9f zsO@9Lpe{j-M0j0vg5%WVyK}38F9F3g?=Yy9BpbmiZYD%W23J%@`FvB~kG|2GiVR)B z)vPOjv(^^cd#N5D2=ezb^fJsk@o#CxPo^P=TO60!V zy?=jxS?|--lJmv0gm0WT#pYn*`s1}eh2LW^Em`jBjY2pURW^-3`*5oSeK&*HFUCIt`DUeed3az7lxNx0b9KyzjS^c_T_TRwfm={P=qp+jpzA zyqArnjK}TNbi1RK(#`O)+h)NN;~#2kH&eQMZE-8LZtH4vXSA4q7U)*bxA50n9uufm zISI0Avns;Q;vs6(6>Zn#?;2{4kMkU%rWX6Kv$nRvZS(YvbY^=uKKCSvTh*B;%D#u)V;2Q-`HVs{eJiA(eO)oP(bH4L$b~c zt4Qy8D}Lv^0$%v>=$ev&m&}$`rwrezQ64mY($RZ4t<=ihpFHug(lc=wRF}K;vfi54 z&DXY3o2qE~Y;!V@0MzsTNe3WSZeog*_#gZ^ZL8-Ztzr+#+!YFwMhZ!;BQL7GnD>A> z8?Twup@rtMtS4)(Ki}PDy3^X$Y93xyRjXGFy0Gl2j=Him4x^dqqSiRhA^O?b5$;+G zt-QRdeqH_HhvbwaB`a#Z)K;fs;ZV|k-6^9IF)I?*2;+Xf6UDGLb;b-2$pevV2U0op zY;_8F>cDJAWk@ZlNt}NCrVN_;xLMWaWRzc64Te3ehAO>8PbI9GBZV>MgU5@j7hYxZ z^FU z)nMPC{q}Y515ENzM!?0RS5gY!wck(-o=@MjygGOY6;)o?{CcCn>azQ8ZxFIcN zcrB6~d>pf9ilH66n{yv$LHuQQ_dj17dcIbfhWpTCwD!uwXile^nj(3ZJ0QoStp(}C zD;^M8`G}kC5ipHDk zV=K>oeb&XJ#n!M<3M-t~7hh2yeZ%0}%TOTH56epMwzT``?BY|InY&_*-Qdd^0AN*{ zqf0GKmkyXc<5 zUq4}jcV@N6l${g+pOKrv3UJ<8xEvcEeb#F zR|}aGHN3ZD_0cVfhF?qWUVxF_@vYwn1L`7pGiuuA6y$372J4l&-&EEHw53AxfeDR+ zz|(Sm-R0`%$-~M`BGP*7`%NYbS#JFN6D{n9X}b&W3{SXq%joP=86^&kb}{kuPu;qO zMPvq)D8e_^gcsGZz0*r7dbjLp3HRU<(sB8x>RN}hKJF(v#y)PmHBPNn;)Zj#HW!Sk zHb~0Y!1_)J8}iKaYSF8|tEVdGpC21KmW|N~uUh4+o8a+wc)_v&j5Qxo^Tq(e-d zsz&?KXVLw36A$(O&Ts5abRD9pIkvINzo-rvRsc)#mxpDqyL>%(_Psl=n8olkv#_YB z#x=04xj;>*`66E~6Ukx$k_|M~*8+3})-7?B&ZVrovABPieqnpKmU&J#|e#-yM}1xMrQ9#(X9# zLz{3wz58?sqUNQK>leSo7sn`&K72ZaIypSGM5Qvqhi;X837Mxh_a0sniC00N$SnY} za*cs~|F%W=gKvdqsU|;rR(NOtlZL?KzvQ-bwW%CsK_Hecih^fnnv!-AZeLSf54TXUUUOEnJoV zWidQE@(Ptj}!?rkJX zHIIzGH6jdWirrXe=Wsi7S~_$%Uf5_}?lZNB<-!bJ_3G~1CKsDjU;7t$d^9sseCMP@ z8g%{V0`c;?U9dT$C6m0%{^}aLK^Ar>D)s*^iQLEJq>wg2t*ETVP)&m1=AB zncq;HB|JOYq(&%wWRwAN(xR|Gj5^0Gcdec+AzOEq&|fG3EZQ zF51nePFpf)LCE+z=RBSL{q#nUuG#Otk0V`M%@EB&vx9#;?EU&6ui~qd{#~{65g@b5 zx8DnKQ6ZG%{~zr8>Ojej*TyIawjolTJ*oSJoewUhcbMC|XRB?Uw>4jUjG}z-`pZiD zU7M?Ddvwqq9GwgB{UmR5Y*TNeWZb0v>K}Ro6_sf&r~Yx1%o*>h@;v*=_Fsy!OH)JM zZwhu~HpTlK-svjS>{`X7!yOf^>XO$qKU&>psE1dNEj6E#Cd~;iu-f376NGHmv$VB_J5W){N;kQ=2{E_+bnhTKd>JK}_1 zRuqmsjP~A8##z$qDl6FgqCZHBuGJ?^@cDk)!2d^O`(4f?wF`(aKyXNGbhij+tKX8|i7H(odc%!1MV9i69MQtc2l0QQx6jrV zRa4=K?3_@BQJq^U6l;sqH{rNBe(hR%VCR&7LC4`mEzTz-jE!Ve3*O;TWVF!H7l!Rb zQzG4o97G8J1FXvH(jz`x>UJ@zWzLJ;9$Px@n`;!?$YhRMsO{Wn77^ov_NP{v z2Zh<%tqZa^FgxJdl`(;7)BMft1Lj_^NHaHizAnaF^Q4;yDzoDpn;S`EFPOKQ5r|6OwKK}O8J~OEE z3~xeoloS^0OZ4Nc2;O3}Bo4Sl$rR=f+7Am4>PDnF(DLSadd(CqkH|hRjr`C;l#F6fV)PK90s2+sr><62o;m{uy+Xm7=r{5EW*#DWgy13| zc*osN;QuoE^h*}=j2kv=c$}|?CH_xsd-djKFABZjnR+m-WMyWHgw2^TB{m1`zoHaD zYj2=sVSv1GWW9|?lVm^yXd-fNe2=Ud(~yQFJ-AJ%5T4{IO&tp#X4`Iq&7 zSkL=abA{3Q#?Db%2Xrmja`Dg*OZpm8)NR_di4d+4G!4YX+Xb>Cj}j-XcW7q0S+t?+ z8@z-qnlx#Gt>IX>V#eLf~QUJfP*~I@22dBD_Xu^qaLmIf{ z(foLg@B;D*=NpV#Lp*&IFhtyr^W%lIFii>-5Czc$Nj`Gcax_!QK8O*w)H zj!MybPX3@gJp7$wl&2B@r`~J!*{36V9{`!1SN1Y~(LG4s26`P;E6d9I%-#r~Eyk~~ z|@00AB?^v?t)zA z!^@^qlwh)x4xVC8cq=M7vrAS=FemnDSF^q;lc(mz?b)iqJGuGv>G$9%@}b`Of&^4_ zWDPdOT!U;48dNz+cGlx#x zKo~$aF?YaX+3xOq~tnhQ0e*5~jP~o9G zZU2DZ$8VwY*KB;y_T`&5vr2NN&t(9Cq>=vkl{XB=nAqE15u;e$u3C z;BPxW21dv67#2A+u6)z5w8isEFR$Vm&(6F^iag!&=CL>T68M#KvNyIF8Jhj_y+w)p z^JJk5Bz8!8lum6Lt=}qjnRxH3ZpV2;`$tl*1c-~>E58tN{>&X6OyEF;Y z@txs+Atv`zsl9o@rJR0<81y&`gTUFbHffgKXYfl<8)Uj$d6eh*Cv@2AaqRc?kou- zhYsCM{pp*tLi^s^-9x}l=_Adv-m8KTFB%-g`3P;;TcNU4@+9m zU-on^CzMQqL96Ianxz(L>3z=!aeI*@N9f>)i4%7u8KC7v^&pOH8RmWVY{EE#cNpD16`|jh%{+a5WgfY0> zJMX00xH^bI6|=N1)tONa`yw7Ceny$YOljHABu*Eo?)P+SZt zpqN!+fy$s!tzpj|*QSPPetXvSn$^Fv1F|cG344!Q$ZWo!9A_Zkm%%*P2)2OqUBQ|!{EZ-zI2s3q=G{{BC%qb=~aB^ES!&K05 z$&uqM*9HkNy*NinI)hvWTH&z8y6hR>s!f}QjDkTzoI^i!6GTp}Q>T{X zb;hlVW^ln5aw1Xh6DccvC0yZ3T*M;SgEbrQ`!LvYnH&>0#M#NI9{HwKlH^PTXYEq# zORAoe69qXT1NNlP_MPa7G(^;qG`%z&5^9rl{P)u(L~#G}wP1p}>iC6=7Bx_i-&+`Z z>DH~$O#YK07Su0emHVc0mrx?zM6Li?p_)@0aZ)B`fS01LQ>ORA3)c*UTCinpibeKO zzvPv~k-%VTlfR74Wh=SJZq@Eq{Lv^Tnm#-_%Tp(Z+Z=**A*wR063HHIATFeA9 zu_?aX*8LJfE7r|?1_^n__E2ELju93V4tE|a#}R%A+Og=$@EKg!o>)t=^w{!}11GVf zd*7#q%?3`AC-$itVg}0~2%E#MBNjndKTo=lWK$SQplgr)OC|uvQ6;>+>i6@lkpU@v z)pz0$S<*UlUVuFXS7^nX@)Gxe{U@HjDZY`F0exTWeaJ$G;#qM#B;?8K4b>}mj+tRN zb^0-~8Hz%60%y)X`>20{%g(<_=cBqa;!lOGewEloOJnDmgh+HGMOoua%g5xe+7(qY znSxNIVvMn_)(jUb2bYvQb(IMfzj`-5wKUYalAQ8v_o<4<%)mn~Jwzm6k#>V>3PQd$%NQ!deBn@4+W9>0~OC7{45QPy3m5hnv`R)y2 zUVtQ?^V#9tTypNwp>n4wJWoo;oo9v(IzHpr+6eZBFg9^7(3y-PTTInw&INDJ*FVna zIU_37|DslW-|I+U=&8!|uAT@}naH6`DR)?B!r7-8E)lEi3I?ngdT5yr>`prsmEL** zH#%%cUuhbUvU8QC_1~Qq1*_AByxPh>`Zc@Z1oX{11t!@PFLV8?(F!j^5w}=pMctC+ zp}{srcPXJLfM82(Py-w>y{*2^Oi4hhK1k@qKr3U0n6V(a$>+}XxMG*~@O(JK7CfQd zWNWXw-hpliNq^s-tI^9T^qLQ<@jY}$by&Mrs}#JO$9Ypx$`5F>|vSfe(&pBd?LE~HA-=^nziXY)#nFaOISA!Bz@4}!6|k!BGxm!`1;=~Ymp8;8MSI- zV(0gE+B9reFkRAx;Er2H&Wn2Vpy+-4c0WIBzs$5lH*2x3?&1f3Z7Q0WsBXR%9zOEN z*25|iJtuYx(AQ1A>T~h!1dCv^rKZ*2S8FG1cA5Hghk&d!qvSkuqH0x}N!RFnVBeVa zOSvRvG)KFG{&z?^lh88$44X#mYi}L9(p~4@nr2XnjerJ}4vQt$g$j(FDj6D&nGbcJ zkwYAN`l)l#{>k-^9+)suLO&UxBC+@^z)M8L`-RLNw9aSo(?tkS`RZ-w@-NNR=;4NP_q_VK@#WN%CVkcS7c}x7+b?|0o6TTEax4MOUJQQ1 zm2Grl%5jRDa;**L-l+HCKKpC51Hu;a3E~;&LxsGF!Sv!k*B}1qqchah&ahFOE9`x$u<=2yKVQCVz)eOQPsVxjOZ(78E`@$K{(af>@*b-t8b0}0 zC?}}_&V~I!k{E#tN(x4Uh1ia{Kg|0JkU_Vnlr+H2wpX<1nq{!Fdha<+fftAiqxes` zc0nSidhUD45WRjHUz8>ub9kMXrwkA=CwgO6aj(TP!$>qd6fcZ-`SRuXlpfqBHfuje ze|fJWsfrxSsY_f6Z>s<6C7QJ5J|tXEtknvVhZ>@hdVFdJ*Rk-mRQ2S^%Qc}M=w{Nh zk%9v2P;zBXwfA{a*Ert9CbX!DExHQ8?7&jPb3H2;g9Wht^bd7wz3nm1Kkk$HRF&O| zuGe!6#7uqli5HcdMXU|HbFU)<-F>)*xj;3Bg>ba_tIQ7lS4|VDvr#Dtk+!(#aSL`b z3t8|TDaPLTKRXk4OA3dGs3hZn{06F!QuS!vo?rjxl33QHomXG;PS*j*;s z^x>3El0inN)e5JwrKKe&wp{YK$uZ;h)rl@FTaDRNtYU``YckJw7eXZgN72xXLNg4J zA?+16p3F>BtSEaqhB@tV%n%i$7-E}C6y2g206^kaqK$9D-$kXtM6>OEB-V>>dc?g|XC@z8_9>H{ z&3HJ_+Bk`ruZ z&!zwW9jKLG2tba=Y>@mOpXT4(eIc_Uh$@KTt*u~826gU9&?%Xsr?4Wj64?AoQkLk~ zV-HwvtRIL2<%390`U!*cr6ahlG2v-}k#bvaL#oU~;}eizi%Lg^taU+{KxJtkHY|RA z@EyNKf%XHx?QdqQrMwzN0ytNuDj@~dmwyNj8_u$Zx8a#2o2Nc~`m{J_r3B#du{tBzPFNIzVB^ubZLL>lie<)&Gzf!pjX>$dj6XP&o0^_ zqqGMo4V&0PyGsM@xE#7@X>t{AmX%=+&W74*YWl_&9cQ1wLA;batOknycA9RS8<)Cx zc^z_V+;MmK%Dl!@U!Z1w8-AR_T}DczvaYVKcXZQVXtGA0AgEq<&K&L@1b|e0O;uI( zBgfr+HiSY}%e1>;M$1R350b2__AZ4B8dg15u504uLU(+*lEd1N(gD*9=fA(c!T*WZVINUIl4v461fVvw*G?;Hs=JNE3PBjh{DG%2e4jy!&s z9!@7>>!;i)DI~p7L8GZ`Jim#{+v^3FP&KV7&92##3Urn)q%1?D9P@n_Ydje8>d!xq zFMn}^c5EcKM;WxdGE;ttUTRI(F_oLG$6+pPwuMxZ(m_bFKfqpORbc?9+KH)AV8JjH zaguKg%QcVLX%Zi-xjb?ca)#$q9#XD=?brX0yLW$k`=h*}aZa&MLNJ(Nq8M1Mi==?F zJ041eaE*g)y}X=G?0My3c<;f3cslX>4=?9C$37Y2D}NBX@x$i9gd-$k(0K50_KgjL z1r|gq9r@wmP_YYO(k*-Gt&1LfpreJ&ZYBRe{^%e;Dna^k3=$L~28&38>gdo^225}S zF=^H3N(*i6zF^1Or8;AxgN;d%KGm}*!%(h>A5~6eG%4TFL~fqbI}wqc>cZ~EhZ>%q zCOnx)35Dh7Ef7h)lLkWD(D22SwcJoLFPx?{82zv~FS<|q!-xkhwYlH8JGJORM*&;j zsD3;2ZI6GR4Cp?CF~c%k#rFXOh{wXt@n?e%4b=MR_8-nv2IXv*%DLcvajy>)whZ~# zCTKEj@f*J23^)59AgX@8BoWc#-Sc(2!x+g7>gLILBq2w%4|8 zo805-eEjX7UsI!$mtQH9Cb=ZhLw2R0K-_bS^3SZK^JB{DcQ`llkl4wnZAcuk1@DbG z*aCM9+yr@N8kCj#ht8%I1yd>i0iFhwDifRNY%HAL{6saKY`zs8u6WJC3B=?U8QGzi z`4uDl8|1_G;B_6c9|#GHCtDI6uyR~_%X{sYG3kGRQJ33lYe(!SAezeoZZPf&?@+WE zU=mCnOAtMjNl%#^jGycEi3@au5uhGj)onm9m7A%5ZU532b+rT0W$f3P)nL!-x~-nWYd=}lJVsYuS z+wT@Z0}YM_^&CW|1)$?gmw#VJOry}Sa@bwBrlwvjn5guD4)t>0A;cTcy z^c9kdB%Kxdl5Md4?e5BaZ@{V7SrJ$7xFr+C$VzSAwryJ(2O?lQF-^NV8SVnhl4dI< z6|h1YII7^0D1UnNXv$B9NsDEJY>|ZqMn4H;t|ay1+@VWZxqW?`Q-N%n(|a5#p~5;K zqV*f$Pcoxulc>RIBbqjEzJ)VERvu|m8)`onjCHpdJ-VV>pFV2nC${+c`OSMK(gIq; zc~mwM410WL0D2&Wf#Rsa0VH?kDx4VHGoJEc_RGdGLnfLQnZ}dmAe2;NQG%wsy zN&P+WA(ByVK0?XO>{b#JX~ey_Gw+a+g3=&&>aPkw(}{P@*8lh3iFV3;glnzd`(WVM zinFh;Z}gI<-FP{EWI>F5s@$M~Oht}++D4e9*C3-vuSej>68l^j1>L}mhvy}u6jH;+q{?x~k-h7D-5)+l`P2 zR81go@;4~Mw$NM&-Nhv>J;-{>x!KY)t*^L{o8{3moDnA38$KAzqJr;`!-J2Lp**`9 zmsVplMfbO)A|p!FZyb?HbaZh@CNE!qPuB|XiY%hB0_q}<7Me*<1AZuT23f93@q<9xq<=K5zjvANJ#$JNFuHhWi@RO^)&jJhPW3F+k<2+jEFSw> zt=rCAvDOTr*gWqL1<_bN0|4I{}D?ly<7eMsHX+^jELAp*q8_ zYR8y7uSP!JRhd!i){h708AiL5E3y7J!AI@dhzAAXU&Kn;My)z zDg(*DmPQTg5EVu9b%k&mKk%{XeYrF$AVp*2D>FV2I-KV>Trzj()kfX`Oy=(kXy=;- zhb2SJmPeL(#@CE?K#EQoT$o}p;~t{PwXV%C>CV?#`f6Uq7$0kkOWg+Ip;s8eJY+1s zG;j^#gABs={V)ab=`Rai$Vp)v`s{JHlp{r^GiN^gQP}qxksGkO%uNche<+CC+JrL-j>6WS3Z6D#-H%SDII3W^m!gIsb$> zb*B@*jTzGj5U-Q7#W@rh+ZfBBhAL1@tW&2=n}03hJiaF8%6fL{$L#i&=u1Rwb?jIt z)RNqruDJ*jLD@z|y&s9aai^2gr7}0bvN2?{kbyeYNO9aUdZW-L!d{qKIK`jJgF@)% z|A<0Q@%V{icpwD zG+#|$Z;`jEU%!3<9d~OC-*7GC5rPG7kC_p@VWPH7>CyjCMUGplJGey)kCAwTexzg9 z(5hSf*7xI#p^xFVcMeA@W0&P`J1tbT+}|(QW!8eAcHRjN+F+vXy`7PFqp(d(Otcwx zzM`B6&34AiQ!n`DHaKhRobqdy`zsz8Gq;bK9}xe>-o-5r6SL>TeI~>QlxZA+_yRJ< z^*x|7YcuD`d=4#<3ZbegV~`xh)~IfKN9d?)mdDRrdgr_IP_*}+k-pC#QoXGZdIIn!GsxUx-Ez5CWkIXPs?o}W`o!lFQT8B zneA339S6d0@SX{$A-`|5&4`z>_Mrk%aM zD(t>W*q}#sAN02k>-0$Q{n?2xV{dqsIritGNqg#}t*eNE<(5mo!=M!kmO{6*zVsKI zmx>j1lVTc?6dhPUB_6i8W?%xw(T43@_R)WW#X0+9SKhcKbHsktitymQW+hT7M0Mf9XMM>s2l;+4$tilJ}1!(C(X9SzdSTq zd$hJ-@xQ!9?b0uK;iIq1WB>i~F*`g;-fz~w&J}n9hLf|%6G*-|a>{`ag&Q*(j(Rxp@t$8v+>Nahx77P=2?XiShr)Won@x*Qrp&&p9wj*dAzpn*%h zw{C4B)+9VxV$G*pnRevE=DmBr=Pb3Y?R{teh>4RX4VW{!^!fG6&sK@vX8pL|15L6% ze^@-jz4MHrhfk=D4qM_`XrVJR^!cG>nZM8j!^Li|r?J6Y9)~QPAmiA0;VEHf^^3BF z9%fPs9o`^j*f3BJd)h7uUgD#dAvq6jc!REJnH=eCQaNEqcy>bXHqE3x_scFC=ul`P z<%lCmRKfK2dm~&sB?F+tFKdR4>D)@XG#F-qcm7o8b{5^xh8vu%+(tCK-9BKb*?aBu zF|TtsoK9T4o;#JDez!rTYw&fz41>>q?&Uj)R~Qkxu)gO$v64SGTNO3M^eY-+>68n* z1k@^-3M4mLAGCm#QBak5A`t+PRVbvbYCd*GSw+QIipF=;XIE)6A2dD%W=Qjg@j5-7=m7LvyqgtA{Y(|W1U!+I!i~!~!vzZ#SkY|BZ<%c1b-VWr)f+sJ znd=j6=aRB{|H#m*sjEOLA~V`1C`*?6J4DR$nPme~uWP#@GL~W*Sr=ZhMJ5*u8|@X& ze4-ttjAXd(9Bchi*#k_leg5HlGx8e~0SJz>*y7TerTaw81h>Wlxbj-VvmW8mvbVV8 zsh_6e5L;h+SL|ym97^5iV|9IenRXa0{jVcsE~>w#%yn(psFCmU3)4@t-&eGqF)F{& zVb$>+5q{#+mrsa61sQ#LISB(_Y;Y$mS1@x*$K|_^{M@V8gLggrnjGQ(7(GxPRW8|u zfR&fKt0+Y7RI!@kg*!#`m}piP_nl&zX}@eg5#H&P)S{_SQfA7K;UImA76~fhp~^9M z@1EVIiZ0r=UZ z2}jyqc+h2rXkHZgF9PpC)g`OlHIg0U_==XPjk=Bphyy)7t zMUrZb_KpQ1o%*N@nl6DG9Si->Y8dFIFEu(}xYA{Yd&hhZLo$AH`59yg7`dyC2Xs&c z73>o}`wcSmK@3XdTD5Qex^TaAUVgK)#lMI&jPbt5(G;a@oY|wzp{N)Wf0J}RiKsMV z9d!xMg&KwDJ8)-ppMd?#OH04kKAD&%(-Gj6fHH2r_Pt*I)1L8YPNo(-TD)kH1a3eu z2|8jEHgU?7R%j!X6pl6JX(D+CWSWoJ57u7BdeBM%AAE{AcWf5{%98oG^5tj|unW9` zv3V;c$>?$84u2Ux(%=6q6)T&Ya>Tv_dk4~}XsdIGA zyzAF&v97L}H4zGC%xe&{A9=hoURAQ`&P+PejNvrQ$Odb>bjR&?bhvcMG&5XCrq&tY zNithS#`~U`==-Fz>TIf5UQ*jf!zw!!q=jViJN(J}ARb1=@t5 z{sK9pkg29Ks=Y??wCFtZP$im$Ep5+V>a6aKA5s#I^EmGnZU{I|a_3I!-nMt|-hkRh z=H`dyb#!rYLEU_bnLxnIoqP5?|E!WGQhzO-|7#j2ftXmM0}ho#%5NqupH`kr^1=Q$ z>A`-vb#&WeIuPYp+!|IUxp?hFjZ*c;fdg%sl;U&tsB<%cbMJ<}aH%|UXBBWm3(kd# z1N}xzf6e1NJj$(lvNsz@J!-t~*N5uEDUdo})~&~x2k37JI&C7YtQA5g&o}bfgM3ch znEt$#Q!IKR?i)7@=%Bv$hcA9N0*t_IFA}4nXP4T8y2;Z3dR=(1TbpKTZQE{r{rYv| zS!a;Ph@hR;Rty}TZzv`>sKkh5_vH&FR)IlaAzNWA?LV;-nNH}Z%jmsdve2aw0ZA>o zvWsbJ1E5H~P#x-6-gs^!4txl#hh!uan((gfk4*!mEH?S;zUs&Iz z>3WEJ1BPD)@Rzw_^oWAdvGlh|i`ENMrYRW>qP-PaBWix=+=5redIMe#O{ z*_DgCoJ<^`Yf$*y%Gl!EgcgYLFF9(~1I4*X1s;VWAo%m&5OIqAC|>AA%jjJYW6Wk( zW+RU=G5I~CN>f`meSuL=l2M2vR^LaaM|z~w<)kZxjkgW|@-EJ{DC5eK>7gG+jv9r% z+Kn(Th~JJdb#$*$cdTZuS+gee-u~ZCg9#tL^O1$=Yv8kv;2Y1nARLWtbuDh8ot0s$ z6oz|;9=e|*Vl((&$-$B!8I*1!x^~?7iinyORUem(nyVieVWu-6;w^xeY%(bY4j=ZW z+Y=1y#EBDJtczNX9=*0y{8*%axrqo#;%w(GpTOJ!c32bdyLT7#m4libiUtan`N7|T zsBVb^MXt6a`+)QZYV%FUy^R55e%DIBFfvDblV5e6Vam-BItyY1ao0*ysW{qp^d#*; z;S;*F8nU8$H#$8{YN(Nz=j!2zD~2t{g9o)oTyo*#lZ5dbi-%w=kd zh2ysPu|9L0r=XaivM!3+2;~Q;!dA5M88zBCq_mDs7;V}kU9GwwLC0jFtLR2Pew}I` zb7=AB7NaoySX@OU;kn9xVWQC#Ow@y^hnbJjJe({Au#p zr@<$@&P|fJc{08 zG51N(AZrE+;Y_)cuqrb7d}1drVElX&NK^@7@;y-o_ z`R6aF!Jj4n*!WcV+7hM|Ov`giW*e+PPDIV?J+=Ec)T^`Cp<7+6Vaec^Ky%bkUBuG; zo|>3qH6was%{xX{n#@k)Kz>I-)D^}Tny5>cE~0uX{4(hru$~N)6sN3tsO`f3|JIex z3+M3W(qHlB$xjq{qNvB%D)Q9Gs3^O+i^$^I_@!)|@I_PzC`PZ$%KDZJPhDlxQ}YSf zn<6E;dr#mexg$87LLEeefu~GV1a8&JbGuJIdJA$g81R#VCm^keMI%fP^_Ow}3=v7lAy5FSnBI91C6(2hS)8p`eQ#>przS@bEXv!gIYx z8!sHi!$Q$lPst!hwXs-KZy}W^*(OasSg!yQm-I%Yjqj4ibrNzy!9+A~P>o}v*>DE& zEEsxPXCFKUu5=0rnT)9*jV$}5Z($BykSV&<Bhkc5}-% zP&C&Za9_{3N1OQ+e{^&ZIen|IH)-NG{8NrH6;1{Vj{dw)7cS6$h>;EX}puR`<@6xr^ zX^8ZX`q}l@VDlc1k3x>Dy~hjplY3^&jdMvq_wJovrYh00ND-fX26N!=M_OQh5^K}@ zG7;1N`s*F*PAU_g9lz6jyi#S8KZMVL`-^;cQ^oTQ&OnyaXwGA_4HJjMQ?2~`ej)Xw z^i$nF&}yr-Yv;w&{i}@J;VO|NA!91c+(91wlhGaYvC9kNcamPy@&t>IqrOi3@>SK4ez#7BiiTX{$8MS!I3}h7+QvC=~uCf2r*11J+d&f3z-p z%^i#^=~=zV4&<=GrsK~(|J6a8t5<7(m4zit{M*SiqxzIb;?L^Zd(I`Sy)?$c@tw0V z2$J@m6?LwofhiF0WwZ?jRMeWHAEUkRH{_9>s;x(9?@2z?wy?UU#eaVw3AY6JqvF5l zI(}h*86KC^Poo+g-=;dg{B@Et$SQCfl{d3cCx%*{=ZOO+bQv`0+B5fb>s7<)QDxSuC4&9+3zZ8kd()b8+Us_$dKihph06^p^=E% z|3RCWk*Z!xoDgGaW)^vXpq!VjhgO^Jb?n!gkbGTXxLoYNbjajv3&6Uhf zI{iIrEP3aFGrwBnzPx!JN|*^^6K1yf)C3bjX4l~$dTfS7l$(F^w7Ag-3DDky|EVn4 zhX{^MlgRSYI3Ru&l#*^RGT>#@i zQ)7f~O4>N7pvl~GEI6`G4}gfkwz%T&B$ngd{eF)S>9!{Qy?;bx_`kQfWea0__w$X9 zvcMbW)K>d(ku3UXan2Bf5nW%@`m4|S;!{#m(DbA)og1+J%*m@)2jZp>oq>>b$WUd> zp;TWe{SNNv3yWDT%750myei16T)2+`{{K4XZX9~!+9E{*I*ieK02CnYAPrCMn>XKG zpR+7v7X=mfRvY9p=k1GOql3P{n5_ieeLn{2kxbP zY!NWrrHb~W8Q3+OK4i(iZXSm&w%avRf-DV1P6-4JQQr&5y4u{ZA51uMgC^1EeYjyR;Ikm_4Q}qvVDBL1uWRiJV*<*E&uKs^|rgHxL(#1)z~TTx90CFH*K3|J3H5 zhhiD%Z^vB`+WRSNu#Bvr1(710B+pSj+B`e8?3*$|n`m0MWTv8QqO$j$hy{!!H@}QW zoVb^dDZOpknshL!9%PkZc}1@GEX>~S{ka>0kRv|~ouuscKecQ3&J(G<4{}BcUGX!@C(k#UFz$iVHV(?Xo)!=(%aQvpvFxfy{miPSzGPd)#b~R ztZOeMoiWx#Hd5}3nJu&@+KIm7hi|kwHUu= z2IiRD{QQ{(j}H9)JDC&DY9h{4iLNMk*fAp8lSodY5NL+=)O`+4^GCPupB5Z%mhOqg zj1w?x`PRVs@)_Fs%BYd9UEN*Oz6zoD?vQa4T(J}r_{QF)Cqe-xo96>;m)}(>_SPUH zxfysitP+=_fstR*aVi@6?X|9L&-Qw`2XX>jJSF^(sy`dm3$PLK@ARE?HV#vtEVH8T zW3{%LVR8sa?X-WT(RQ6wAejz>S2bL9J<_7%8Kcf*%v~u8W94#Ou00m1NOyT@+bkF= z3&ip384coO9*9r(_=_NaA3?&XhUnN6|2m^F**2e_6teK1Mw9K7eYtPnS_DNRV6HZ_ zjy3DQ2n33fQhU$sla>1hmOHXws9#KmzCaE^_i)kFLLBt){ZP*(;SUbqtaYt?J6EQD z7#KXi811;cSui@XIFb+4TDP8GIr&6Lr6c$27XmcLSXmvtZ@-y3D$hxf~*)x9i|_%JTp)9f4FA%tM<|UM+Ap7|999?bw{B z?+vgTGD0Wj=pXe<;IDsZv}r$^ksebXP>wiPvixEZseuinm<(v`9o>_~Vw&07sR{5ztzF?rbkeMjy|Z`Sx+ z(3;)raPTMB`seDoGh_TNmG!d+SeH3|AN<`vFYNYyxebu{P{=BSU)p4b^^%d3#EYDb zEecKvX|Ez7+;PIg@tQ&KV>tsj?h}aaYd~MM_q}1w$L}W{8$Q?Gg_R`JLapa5)G2&+ z(V+Lz7h1}@3p^0|Sn(-B6d-6knJ8I9pbq57+pewU146nN zZd>rZ+SI4)gUu)q=;;n0amCfe{FGKySW9*IF1Km^w|JFFzkIervwF)>CIB+^$BE80m0_R24#Xm-ix zV26(m2z!XpEa7B};Bylq-e0phSi#TSIx`?JbPifEr#W9bvESo)$y>!w$MgUCerQT} zcAts1s-1zOUgPHNViL5qotq;OM*U_w)Dmh2NK&XkO>eB8?>K8uz~wH$<~3N%R=f!b zr%J@sN2JEZIrZ2A&+IsYO>S%d;M)Jb7f@^~-V2XX^2Q!RUMP$>CU7(Fjmcs?-D-c+ltYFlr^(hn`4pigI7 ziL4HWSHHXV$#BEg|JrvmZgF);V)E_Vh#J~Kk!WD+2phJ>t1gyc!0u)XJeetC{Pab z#xXVP-kaRv5Set=9l>qOW||2ndzakJdzpJx=8VWQqp}jD&!D$f%s2aKk1_8-S@Ag7 zdfZ+o)Kn|Dw_!s#i|G!Qnty$>nur|PL>Sx_uSEu+u=;YY4<;SOyUNO~%pK<_#BQU8 zA&~0#vQJNI9b2CsgH4{trPo3yEM{KKqe=Yr!174w6w{G3#4~QkE zqJ^Y}xMG`f_e^=ZZN?6dV9uYJL!rM`3$QLS{l|}|#C_!UP*#YfQgo=|*m2@9i<^}y zFIT1FFkuNe`fFE3OCGO3(BMv2K5tk)JmL# zqPzQ?f0!7ZV^iDX2*n22ynWw>fQO!M1jpt?*@BNx2JldLJ1*6FvDn)zIkf&wfBYW6 zQ{H&Ou8t}gG!MrI{haI8MlGWw35BfAm;4^YvYv2{Vh0a*up7`qP2ZODQASsw%1X-9 zxqpYGSuSvpB51^P$Ck5L5^uAkW*1Zhl@@IC(wN3}M!0|f;rHJ0=mF4+hTgNk2b66d zI2NsDdZ*broE#F~8Wa?i6k+LANbZ;fB=BiTD#&Qzc-0 zT!v6&Yl_p=7*sCuLgTxzcfrq`c|S64MH{mXpyo(QI_r7!bS~8IYu7mjuYnG>#?`?r zteiUYbEg=?esM8+Zi8+cw`%oM@1DN-H6i&w{k(UT=~MlT=RNE~=PW<$yYKqxMQ0ZI zE(%+6&G(|AZs2cwPUZHBZ_{du(FjH*ZOR?qFXsN2+INXxIxN0+UvExh!_u^Sre$*; zIz9Y6{L$bK1_jrfdPv^Nn3rj4R25zqev@qClPA}du8*>WnW~Q<95m?Q4`W7`JYm_C ziHd=Tz@=liB7Aa z*?;$NupW)MC;W{Zz?&%EMb1n*NAiUdX+x6ynn+CJ;lr7^O-`BHy|@r$;WX#UBptEu zfyhfZicPpd@3*-FPkG(870m>@U<`+hApa5&Nq=_>_H~`y_?%I3BIDxaUX8HVg`ft5 z68^aHf4p9Iu|F0VrF(99bw;AvEM1yqgYx9*;jUQIB@*1gz(70}l4?vK#DXzudwwnK z{W8CxK=c0e`WWR)moLl0bF6<)kIzo?U*kR%w`(jCpLrElCLNh57gdLGEMwr#lwW+j zQ@6!~;5AnWGUwynSPr5KBk8A@cGMiZp-Ar-RwH?s&>)_MwPikdqm()Jz#hu(bU$__ zq9o0xKOXRSN`!V3wY7h-Id<0guqXc|+T%s1?UuUOm8ukGZmd{2Iu#+gs+@UmU+|F# zkwd)B_fl5j7=ZJ)Nm;mb+4@PN&lJ6gHdJ)R*JovcbO0xX%%4;LHF26#KlQ4ORd2dW zK$(c^NveGD+HUTI-+rr4-lYHG@ECgl7FtPx6z)i{&3v8vfpW8ytvwkQrY^Jx@iCPH zoA`LIAH>vznO!J@LPJ9(Qb|oi!*-1?3VW0m68T1S!v)4m^qozy0SmUUVxY;P2v*;0 z7KAu^m7YiFot)A`>siGgUK0n>9D-&dmg8!lGmr+f9@%@xC0%%iBp`7D4Igf?+-dsL zKI=LroGdd-9`;!T@&kT;!GTX)jCd8f@bwS6zP$CF+&^4Ai=v(t&)ey7z5H(8$%VVT zt2rrTdMlUFrQqP)l}U&t68oA|qZz#lxs$0X4Dy zAytiM;QV$Gd{qDiLY?I@mek>5?2ASkOSZ^4sf;3%=|-?4n9E0&+a zQ7h$-DE0(|9<9A6*5?B`2yzxOc4)=w)p1a4Wu&BPezCe?xSS~|t^{b|e)PcCNl^`Z zwhNu)0RQJ_^5+(yq!4}-(g%aN8&x36?_VneV>w+@Glw5T=PotpvD%N_sZOn-Vx-!n zGf#?mg%*qE(_Oyjs6mERP1XDN_`p#+_JMXYD7*E2<2OfPop2n)P!kg{_T@zOkiJ)g z>}<1Auhx)Uty;0t;b|RX2qk6^E#YPuO)#dd;KLdpi&V?cqLEotVp_a-+x@P?a!Z31r4B1N-Wn>hmVdYUN&V{1HN#toEa7{F9spJ259kuwbFnHE@(GlI zjL(C)l~cpoXPD=qLmGhkxrlTieQz>@@uAs4$2kY52O0D>@1|ZEus(hzVj+$DhZ7r7 z2HyS+wCnXGEuRj9ryv_biV+$0BP0NJk&m7XfMe%27s zBB|aNi`pZ$dCUg3Mbp$8jhHk-|Wr5P}tC0R7zlUxA{PpqQ2RkXp zz}a0wd_X`x&b<+YPSf2afX|44jJPlT^x>j~VE2*&IF6rJTv9?Vq>9JLd25Vp5&emO zLUMxqiqa04+Sv5qO>$M*eLVSGZf@cw$Lg*)^fb+P+SPLV=-*DCzuKax_oR!^v^+h` zgOZv+8^4D!^r*0NS(mBab4S(q53c^9dY1NwBP4|)nBvfLelf-7>za4yh~I}F-ihwx z>5=Z}vAfbO%P5)OPCuDY&AK1#l8nbgS8f>XJxzJG?5mrSCvxDx*`lt8`gUG8siXdB z%YKz}6536Fs&{%BdXoATUvV|MMiuMMClU$EU=EAUu(UdgBpb`G_i1Z?CQGcRLkS$3 zO`ewfJ=^{oS#>R;e1DmFA%<=TnAb1|6XxRwk^_ZCf({|k?BKQKhi1!)4jYbYSz$D= zv)!9d4nIm)AMlP+`DJjT)+o$gqPP}QuGkUCsCI|Zq8^y2qaPi=9{3zd|0q$SsvV|B^Zm4JZ|(&I}{1wF*R`5oa(_m2+K6yqFh zD5xynGB4?7A-ECH2QK>@bfZd!K8uWbnVw8pKpg%FE!K*h7^O&bAkUS&h8!f}lS}w? z>_VN-&Zp{XE2BmGFxE#)H77g%7RJb2syLi^GOrtoTyoj?m|+b~HdMViE2D(kH1o)QIu>d}{(~=>XXsZn@b?3BhPWlB zOqp`L^428OYIsW?N3CB6LpVTV@${-8(~H7Rlq|QYLM;GeAZ6Oz3*O$|WpwGU!iWVn7>$Q`*7v*!XbZ!XNs<7 zM!?KBy`Ra{bAkWc?Y8y$*~J9mj>MCLiAYjwxKb^oz6=vKe)#l{A-1;B`NhT0FRb>h z);;h0A5VXWJY;0SH+-Jy!cE|OY{?A*kTJ^MKJcx-@`)*RD%!AANK2Nk9r9q-*5Gr7 z>YNec54e99i$k__S9|vEb*G3W?;R(bGM;HVyB!|Bi5`RN>yG(T7xN(*^giT8;L-wp zS)pGT7cdfPL!fznoM6g)&+DgWTS}RT;(`*Y{RPUNJ~*dkh((fdQL_K{GcV5B2KOjK z#*j_4MlkJ(tVLZRUY@AL73hP?@ReBOmc@{MzVgRqXb&2mk?}7~ zg+j;0-G5Q@ibAn|{dy@Ne*N`VRB)b2kr8X&oLqOJrGiUwGss_OUEQE0`+?m{X;hRI z;u^^sGgKick2GHFQOqiLUMtL9WI-X(6FZzN&4Ycc*L)DW6We6sLA?`6nm_JvlDHVO zPyC&cm1{Itl(ukIs9N;zPaUvq&f)nap+RRkGP&(z@51nF*Q`*M(U%+1Jz;sY|CrgE zI>0!qex2zMPEOa;XUJqKJ~S@`3Y}k9)5K@`>Rs!nr>Zq=n9fO{$l(A>i*R?uso|LG z(RPiC(e#5)#<^5xty8!`2fZFBLS%-H&;!A=xbGd(tyAtvUJ{->amc6z8Qw zkvb-f9Y@Xd9rzgGI+PUR69S)bcv@3a6KoaRz%O&+=+s9KAGVM@clSD-&lNRU5vm&5U7hwQiU9ps+G?iI6 z=TZv0bv&Ejq4Lay)tL4q=~lcux;dz|l1{xJ7}_hzHk507Ge=L-@}&o#Ogd~n?#xHK z2UKjo1pK_@;Ov3>iv`HJ?fmY$pS44VgoTAU0%<$Ecn4)3i;P3^!i3I7q!_~}au2Bd zsPe)aqsos{5h-6=xAokd-T+tN8?I5MGb*{eAl;@;nPRo#CP$y?-n;JscV3bxUmsk> zZA9HauJVehb?K3xw2NL*G+)hj4v1YyPv(ON*89)J#c%cwIb)Etn#3)^h?o!&1F%nV zlk$2%cAdu2d>h+rhfZ^7LXrXBlexK`K|tw7)uR9OfF<9#5!x zdgiUv(HeM_W|sc)nFgeO;udm&k&p(lR*&2h{RbyT`4^$#Z(tPlVkQ+cp@;6-iYq@o zYYo1~o&T<8EM45$T@&A=@%@J#dQUfVG7Zzzt}~SA-GLZ>VurP z%>tssIFvER&RSv>(6EYZk3+-jj^A*k zj)Nr9L?}tR5HW>QM7-m(I6)ib$c`?zI6Y>x$+2*sJ69V*EdZ8u9XxneT8paBpXZSh z0m-d;V{&Ff!kHN#E=IU5E-nvzu+Yv`LjhBLIdgW<;K zRz`)fdC7m|WhXlnKD*@ktm(LMiQUk!h4%8lclE~VEoY7@ADsTzYVOlrb@uGpmx~gn zxpbm+oYk_$c(pw<9E$?4%F~npPf1%0-$ARu!(3VW#mR4-8`k~A*S}n%Dx?1RKb*Y@ zT+aFW|9@9xjj?6VGJ_e)60(I5vV>5SC`*<$YX~W_PO^R~ip~U+Zxt!ZFPo{8f!dzOF+%%U!#6Enad76Ny5P zysZ4e8=M}%>XQFBYZiFCOn)5LOF_zAtuR6p&jQl>X~?g(J9%Q?rG}0B@**?xk4Q$b zR%qE=$Wmphaj@#x(SPo{85&#`oAXzg3}-K1EO=aNT|@8>v(Q7Qb26w9z#qirAZ~7T zcPN<-G=uKk@?K$vbPuk4t3St6!aHXJBGDY$o0xLszbGu+^(WYmLy!A#s%!9!KBwhU z-AqP203Dg><|RrkkZpI?QL8_5mS!u9kwNKOrOnV8FE949Ivknu2u=5^s_^Dd^El!tcA=-3O4w?k>%!qGzdtQ%QDT7I)CViK$ibA!*9Q zgXR{71^Hf0Qp&^th4wZKttY zT}c5NHF)5v_d8I3yrgLGB-dhvHA8^bku%6W7<9${vTI0LZS>rmr)$PRG*+qGn>ABX zNMFNo!3Mz`$7u2b#11J7LnS*&j($FSY`iWs{>PjvD5Bnq0Ln0FbYY?c?@NCHG&ki|Bc1~#)S)X!`tZg7u1?nuJi}=L{^jTuQ zfTs~Xe0VgjN7&PkFA<%t-|s8yd`On>b70ogsVCPOT9Dg7)Bfb-r@I5*8_vvSGy8D* z7)kzwcVg*T8)iY~2f~NDa#JjUgLg|w#t^ZM@+XrjB#AV{C>|t2fX1BLA}@8sxRgD= zefR2g&uYA(&j5$H+pn^UVOwA=L2$Z4AW3$;jE(0B$r1?-=UDQLI4YxaX@bR1jT+7F z<1SWW!JN3o{Jja+k|_az&9t9*MP(wCHDO%&7XTiSjw0F)w>|jYp@SiqmALK&>)=MH zD}iK0mNjJoMvTnNByotca!tO_MiY*f z*>mF5CNN%1`S=w4$24#6z1I`9ma?+40Q%(7K6>+bockhciv<$|6)Z7~IDb9{5U@%Z z+PQOH)Px7%PQ-2AV!@r+rvtNxj!QZRbT)6ng4W|#p$;b>sV6I4Gra41t=JH=Lk`-|U~R{J0<=W%*l4!r z;-?a|fT**(dbHB46i|iHRAjr@Qe-~_$qn-O5pqM^nD@hV;Ll6>XgESnusC>^Zc%O0 zM=UAfPgc@jyrTHSzUlKWA4Dv+P<)xSbRVl$kei8R8rvh;4}HEKsbypX#@804!J_c7 zO4VD+KWdi3gDFa6D4y%+89@;?yBj8CE|@xHiYt*g9eebU2@5H_N`rq^pO3yH%7zBKW^o__USQe@EL z^448Fkvma�HNDh&){%9`vBk^BR|%COiX1pM~X0#0Q6gWOj+l7Oz`i;MPM`?I)#H z&`SwpqvUOA(3UA?mwFFxRPc0vLOTgSKtK`cTzw=6Qp679pB<9kLk?-om(~WMG<<0-K9wj($q|#7_H`d(YPA@!?3etwO_7Tx@I!Y= zI`h1)9Ri5*+P2NNfNq|&>^I$S7#`@k!DG07QiH<`%cS_BhfldseVKxuSiN(xAfkrb_No45HMJ_u7FeZ`+)8ZyKoE&-OP19L`@ zMev=hrt>n+6|y~c&2*t72kLTH0#2`Q@5%Q(1r8zrFR5bU<1o!qnQJjxKj{L?Vuem7 zXXelkXn74lcX)k7Y7`uYV=kPGjiMjUi% z+sJ6p0#0!$&q7dl^Evo;9c&og0o&j|Us}IUG}tqS2XqOl7n2%@_@7fh7+m0jcp@%b zAoEMvA#3d-18OtWh^eW*e#1MZVhJ&2jZM*118fN*i_p4bb3khP<<{l)w*x9S|@$)kmxvT z-N%tE9enz(50g7b4??qf$qqn_wPAw-x{rWbL}9>T+|o_Q#K>p^QKSKtKP?-*Z8JBv zmDQsW>b=x$-3HE+1rf+E*$@_3cH+vZIp<0F zQZ26uh5(}7COGPItF#jh&Q&tjLzVq**C?%H+{+DF(NJf!oLi=Au79)3o%mI@rjZ5b z>nE*qwJ&R9%$=%c8yuV$>-#hsc3^Ro)#@i%O5K&;yXhwlX%pE= z&rzd%bNK0G>*L9-_Qx7&Ub7f6aOvo(-0wfa5|kt|B@o6#@e|Q@j42mms_Z5-^2!Qu ze#mr=5JC@&*o#OhqczOg}jQBlT@sO;>zls2+!2^ z4KqWErJuC3Y;%08&-Dtz+6=$CaI-vEcaz>bk2!tjjL*4q=VGkT*4#9xwiKaJ*pz}b zdSknK@R;Zfe7*vsl)w zw5G&A*v)+O=Ozdt+s1bu!Qm_(N?0Qeo~FOQ<}KAJa8en~))?^+3;!o{Q_fMyJle%k zG>PtKS*WJf+E8x3o)eXWxS(G-BfP@=8s8?dC}IMOS~_8pmz^}h@vpG88Wa&xyf>t<5L zXt8MKBjJ%9;ZZoo5gT00TqE8;K|P6YC-?zeO0t1)SV-lN7%AR_LOKhmGTRlcwf5}4 zWVM=bm8LwR5>E%zr*NUs6{Q|>O5Cs1;f@G!PQxYo`V2*5#g5kf z#19?AC$+Eb)hTXrz*3dVan6#v7>hKaEQ|h7_aK>eWLAU)6H>`x&VzwrKqvtK)C33 zjd&kxhW`4VBxBp2(kCkjK|5)&T}Ov^+Rtler$=%bh~TJ4lnaG~+I@WO&7*K20y3Yw zqTlghYekZE7|0p_$*C1G>hYZN?P!4wk_L7OMktXOuyy6Kj&-i(C9oR?9+m8@7H%t7 z*!EPf4A2ag8;`EE*E+@O+GcbZ-7RsD-9gebpU#t zZMX(%G6uGETPYdK;o7vv|-HtEXAb2;t6}?#rhizU_R^$wG4X z@85rV$;heWZnF8K-|CM>h`*jZ33tq`*Q}Ys`f1OIxrEO7|F{u4B_}0`KoIJN1w6+K z&lJ3SH}Hl#{YBE6z&U#HR2ne`fH5swzFds0+z(0%zdAT6q#{boFU!po9L^#J;E!%$ znF7gbN$(RD`)P{?iQTi;WrPN2l`_Nz0T7m9bUEkYe4|ooRX+kuH@b(}z5&7d2~RWS zL^w!kJJC>aLOpuE1_1VkJb92k6dxABOzVfL>0VdWlbWWg6jfQPNURix=`dz_VTTz& zM3qYd;=0#fati(Vr5SHMN&}xqL+EPWlSrn?X`vXcb`8#dys(et*U_A1uF2Pm+R#P- zlY*=1`@13&zr~a;sc#v5_<{25GdfCXN2CWVnCMe|yDmS4_|7-n8z9ntCla|>H*PS_>Y zFHV-2vBzDa%qIKo`*+pc=ib-O7SS91)uy7NqEQwp({d57-;(*3mQgsSeojp1T-IyY zG?xIA%H{<`K{lS*;Zs`*NDuqifqVmiz6aj@h~4K8xQwn^3&V^jU#8}l+w5+|-d|#z z7WSM;#e{7im9mGOO(`ES0U;ZZZw&d4>%0R!G_;9|LX1b`sy~SG?+(j>>vL2wZt6u* zX;)C$yDr*4HX%t{agCY17wIWcE5H;aH(G8&PJ4LwB@PZFnW;INQo5dRR!7nA^4jyE zeGVSpJ-Sb$Jsw42np2r)Pjg6brf1(Ms~DrXKt&s_WW)C1j~Z&Xj0b3 zQCE+(X)x!=DtRw^9BnnW4FCP+9MSZ>(PiX9MxxutO4e~;Twoun_r6$@+PZ}cR0wL( z09LJ`IPhA_C1iOQ+E$Fm<}-sX*KgJ=z4>-urxv+qH@nrNg)?D^IJ?#ZdQ`a!!rJKs zDq9;`2dPebeRH>eS0%-++vc;oytFsa|I~9_kKZsKF#Tymw9!SV%b)%uYWRjkBbhdz_XLs<9d;8&XFou5|azgJue6C6@V}piQh&& z3D%nWBN>6!K&*KE@oTG`FIZ>v92n$)fF)iGYvqbbbhtRqvM4d4GhALgOvaS=tQQstNc1z2v-i} z`KRP_`SWW}@(Y)J&8WGO^wX{Svl`gA*2%zY4BWWYchOMj7Tx|i@-0PyT#qW^o@+zE z6rn2P4-agHP-@g;N%9hfGs4~zfCHj~Mu<*Tx&M1L7u5;%4)dCYI%95>W)fK3;W2%P zq1D2XhDZBz@ z7-$iFHM5k^D1{3+WB%#w`We9x zo`JRQ)4OMn-#ngnx?**}6!ye9202u$UNm9wxi*?uh?h+t(UL9Rt#o6i1|<9X`&&OQ zvz~bVWyO8%b8Xu8Q|FTIsQEtQ+>-}4#}RJ7dHBQGFEh=_D>>Jw_NBse^%)T@HJJ}f zXQ;RgKmO(?mF041_HUoOWt4lXnyp*E>fl@0dXj_SQxIhgg51V>T|JZ)KtH`m;_%6r zf;;?pMI9p=enpGuCcUU^+)7WrAKtwiS8rYFLr8+T>50m@w_;+Bo4yUp-ZjxC{e3Ox z{?=h;8CT~fc2Rz!Q4f6*>^IM=wz^#)SZ-I3_`HPKe$$^meflQj#s@;9Yi!ql6nLd+ z-H{>T=c8^^O{z@Tz17*F%DE;h7{9jt-ZM_Rh3O5k)uQxDj30;P^+54pPWEk^I|gK| z=VoAr4j;*U?pB_ZvAa{1K2W1*xh`%LKqy7QHX~l7Zj_YT5KNP02 z^+NXDV2$y~@1t|O)_2h2Z}vVUlL;L^+ObYtie zO)SJqOKSt#5O<<2NCVyIj+>Q>-fi4U&*<)M7*aIZZ))O@Iyn{h<*u9ia4#kb2DG5E zjozo`FjTj?Vl?d6rsUT5?G{Xz-C+4u5WfY>8 zgi0IEU^ak0d)TU@=aW-7hiY4 zz=0hJ?`680&VoD^k=CHzQ+1e{7|nkoP-NrOu2%o?GhFt0c(G8D(3#EPsU#N>NCcN%a;ktX`Yl)+&-Pv#6BV3QIb;HIM%c;ktaKTh~Mdt8p%Ru$aH zcGPgzH|s%}kmnLK-`j3;$MHoCH&EL;gm-Gk$J#SsuLqAH`#2>xOmzB=aKG> zkA5psv$ZSCs1v3OkaxVCC`QGp&-X$N)GO z(XYVnr`a8T?p3y@13EXpd}Cj_?5oKJtvLv7)Vec3^t%FK=@7HAxB0x0~lYlB)FijK;8? z-*QGwv!6D)Ik2nTH(TSB{M7M#x^x&(QK=8=ast{%90TY=KM)H00(oJd7W@9bq4}4* zZIm}<^fA2_XX?7X<;f9tL@*TLAE9aNk~{r6qUzXUcry4%`{TF-1KUVZDZ$hRU;L_N!`^6~c1yA%q^ zz7<6q-TO#(7FVsJBUp%6;ew5pp}C?wt5V{x2z%bZOse7Z$!NT{PEh8zdUflH3KiY% zF#tAj8`4HZHpHjhvUxq2(=P}N*>=i1LxeqwgM@V(BOVc5h zNJVxM07{5<4O~c}z>X@DqwvmXNvO^33sFu>egR3+G!Y*@eiX;opBBM&M8RPlsbBqx zGR5#|=ECMQ=9m@E9KetZ;w&VuOu;nS3mI7FGZFmw)sBuG7=92m5pXX}3vzL;?RF9C z(PAiE0J<#coVjy#47yM)cm;gwQ{(D)!-m_-T%m}>FD#0ge;aKOJRXxXvosQ-Te4LwDy8C9XQOblq zNj<<}K@1`XI=5iv4GoyPrK&eMQIU0)r4%DVrzN_NHZ*jPanR`4@x&H37}w-Tor9!p z&13E*f^FjWdx)$Q@_r@bIk}1T%qR=u!b`YZQ*bt z`??zvv56Em&9W30X)YTABBRSo@@sK$vmPNNKPRTiQwVA~@&X+$51KQeMZNtSvKJoI z*8Lh4TPs=3z+vCTekXtX-t3Dr|Nond=ik(B{#vV%TWV^mY7gBZGbVUwD8O zdWCm<%UM|jwF=w)*6`^6?GSfe0ax@p562H&;|J|+y)t88&a@V`Hp?H`bR!Dep&lcQq-p9@9uHHcnbJ z_j|Lnj?#7voc|9X58w~HW$?>h)qG*$C;Nq(I$~tj4b9z?7&K^HfWQB6s6qLAJ8nAB z;{jIG<+R^+^X1lxr!+h+?rJ@>8sh`5V_Gw4T<1Rop;NfXMxD|duh%=QdI^Nm=c~{2 zVUnN9^x1RKqhH|3|3oihuqQRFGSb4~{zt{gjC-I;9Z@>q{KX=p$rO-?UIQ&~%;U?? zyn^e#5#b1MQIZzKREdB#R&;q-Qv#g3^p(2&=T}&dHjWEbD90+3D!YAY^@>Cbh4@_n zgw%2@uC9Cb{O6Ad3LqY1skRP}`I)`!)2oc+6f_=H)=B87RC|W?)!AcgvL7lFxv4@S z+wx3W3E-U;(E|!aG}lp%JYHW>qN_b=9(Q~US6}V)AUssR z{89rB{}&piltA5$uCAh{pNpnh)KWV&2G@O_`Nsq(%dW_;sc>f^?!jf8$54dtoZxS@ zHN_yx*@t~#Nt$-`Ce>`T$Az^s#9iF4wCkS z^bog4u=FN6Hu1URD4(?0k;M)SG8%qJ>JrWB>M6*>GeJ zqMe;T6=p?`NvOxZ-7Q<_)Itrzr_Il*b0+I)&04i;;XI!3zJ5nJ}r5OE!HSA|=m9bW8-55&b%WU&*qb4WWQh5{8z({63BT+x<~P7h{XW4AOv5G3FgH)N$iIWTDv$?c0)Y08W%fA+7VSeTAD~DKlTdA z{{Z|2wwA*e2#9q}zsm+4;6w7Gxp8sOZveIzC533{3CaSQeRAr5K}SJCB_|AGs1@=Z zaRCyPN%KzC(EdvcKwm_4!02RIT9m!Ubuz=gOeC~VRI?YaUb#Y_2}FWw^egVV64j_5 zpXZPN);vCelM1RkOv9y1r+m!)+x$a^WZs=aP$B zYtP#z&H-{AWQPoG!)&BzRoxS!b(UsrH)rwXk zwj?v-#Fw&Y_%C&q1-Zr){eJ)_k7o?wZS8XBnSm|g^LK2Xtx2wS;1Kow+qMor{=CqU&ScD1Dr{I-a_jcm zok4AuT}tm*N{MSde)+Ox%8D8_YrdU!(Opd#lc96+u!nWf)V%v<*J19NsCGlXwC=y8 zJtFwb(DGWL?^7^n$_P7hWCs?ReJLCISIgzNU0)fkN0c2bk!mSpPx;SH>w0I*>bG*Q z)=#y)4mVspwf%*P@DoEg`b;V>8xkiOlma&2((M{l>{q(bAv;PN#fw8j|Z} z|6ymdRja;qJ~Nn+C~*3D<*g0dje_Ll1bc|JMAAJ;rBF?%rK&1>Ga{pz{Tt9XXIOd< z_`Pno)k#d1qiQXl`#A7(WOw&1$YSY;=4VeaF^RKm6xLzTpv&_knw1#&P1V}J(uQ9+ zGeu!;Y@r`ERg*k4*XHdu5AOg-LbLUQYZ`Rp>f8b=urO(#s@~KD^vY;V*3U`uy^)_l2;h%qh-EPdNsSir8JbUZ~=!sJ9I z_kdM;@l%DR9Rc#F3GPcRLPg*`NJ5+b`jXM?{gf|JXMjYa zWE^Hirt^$Gg=Iq~_*YRoY41MW{D;?3F?T2hl370bUnTlwxAOO>W9lo!SxXzu>W6ly z!fJ^cmMQ1)oOXc}H>z-qcrzh-<^bAafFK>v7YJEMlPV33rw#~d^+Ajlqm7(<( zoze`=LW%l-LW{-CE38k(CrBnrn=;;@7?z2M+LnVcF;Ms4bWhJ;%Lz}o*rfVTyUEf1 z8&6~#-PPyDB;cxe^yZU%du9{ENlcfMJ8IXi4LGD1klRh&O{B&?#gZrc z>eb_Sx~LL?rpN$D?d*{MW}chjpw6*(?nL%jFY%?cTK)->2-goXu_?(?%ddT2Vk8_K z`a*%WaML->?pagQMJ3ti!l$n*3(kG2eje6)pygkhj1@y07#IaSvQG9r8}F)|;~X&d z<#Uw{lkUd<(lM{SwMV~WKdB$Qazx>B!>?Ai-P?R!jRNe48O>5|eg5YgjaQk}miyYM z*2f184S4W~CX|E~rlsc^53h%|YBsj`>WD?XAH8{Eqhwn!c=|zo{kHng5@-IJ>~Cmc z)LwsAMZok8{D0%KGh%RFCxUXF`T=d*l4`H_g392?Y`h@)_~mQ_r6d^ z4C>9k@~oRI0)RKz92(lEt=f-I=*NGjY%nr@Z~@o(sn46DG_@D|ujiXUR1cR42o}i?;S& z81TpBNk!a;RX|5(%~Ud$01S+CQu;R4)7QuY97 zBR-*CAHYt{=M*O>9uXdEtotxb@^`PHs}{kl0xW1Yt4e50WJk%GT%&3)`sDGZyheh} z>32?}T@5t9)EIUWequP{QP5z>CT)fs&tp@`ufa(dYaQ0ORjaW$mMAqHfBS7amGR?~ z8H;D1pMK~qsr@nRqdoL?#PH#g43Dc&0-7Y*pQCp{Su|!$nS5Mh^eE0c-tHJ1L*QkJ zDY;D%60hQoeJ5?jbjVOLF-D|i;tLrb8JYUk^%V~hw z*cqXlmW6K8ep;|HY-HOq+pV!5U3b+p^b1OA?$=|o#mVpo7ax4c@|;%abE(oK`D1YS zq#2><;S0N_bXx!B^SQJF_hpCCsM>4PHAp@N1|dD?8>BWu(w8 zzT2B`s>eINM0)2l#43y)=?*C~`|dx}-EmX5Db2}bdVDV$zSfYBn!Kxk$~rlf-&fmg zL%=`2Q%<{YUl$q-?$b%@cpK|_9>cOP4zWUQ5d5*mdCF$WgR_+ITc>Y zasI6(F4{m1_2Rzfg~|jwK`W+|1_B06>PO z{2fR%!O*&&&5Q$yIj!p zDQ@<0pHU`Te=ecJ=(k^aa$kCZAcs`81opIfqKR+dHcb)8%4u8>pbSTmb`N;`tY9S@ zmB~i1q=z!L?}m=#V$m-TIM#ort+8kwEW-r@Pg)tl%B4%H_unq;bD$AKf-uY@ins< zY%&3EKLFt%doSSL0Nq}ZO>+X`04CpKXWK!O$(u`@A3IOmxP0LXOxqyxq`L1tdbFXr zV^`nf*50G#Wc>4MPuxkDsXy0MQ|r{B!v=Ex;qo#1%trCTthxySLdd*3jlX^W&Yg2J zEzO(DXtCUz=S8K};v?Cqug?#ds8qfGGu+j&p*#9hiE0y1PwoX$)&$=WdH2mZ2xr`6 zqYy3BFt%|~hm7qP1LDi_VeSBy)NFaU@HHYA;|VsQjI|nT-gF>^B*WM7pG>_*jTULv zHR!7x@@)K^IUN}?9USW{J{5oIiHeGfQj0zde`>JRl~tLuL!OyG(ipg68;-j_*gC>z z$~~KoI;d&0X6I+es(OY}$MFox2;HjhzpM4gDWCf)W12;CHOA1;iYb=B?;Xe}Xt|{C zhc@AcYscB305^`uh9a?wqo6*(%zn|;J%lLg90w!MtO${pm*Z~}M_Vr5a(M8q>%W)H zj!0uuLlXj0Fhhzv;(1n&V?)uGV)x+uctF)3`0o?%{IBhk4Insga4pBLSu>DKv(pKY z=(n`N@0_PaZmmTFD({@FMcttlOX?J_!0WC?#V}eh^c5 zOV=2M9`^}7tNFzx8+ar8l~h!FTE*}C&liyXNy(2RH&V-O*RC!@hFp#RG(Y5-U7oHB zG7@^_i7dk;>xj;;aho>dxk+RL$&58_H?M#>Z_`NPcuzD8R&$R z1Oy^%EW6%in-)y|H21ff>f^$c!Xk6df9!TgrCmZnsP(C1d$(;ff6*jqVqs5izAh6h zXVnO|5Bj~T`(HHM%vLWC1nX(R=dLL&-AB;_fBf=g@v_o;iHV)K{kc^ZRB*F^GfKJuKzyLmlu(Rn359Ep3FWh`FZn%`dRKHy1I*3NqhN%50rY<*>U z*7B5@tCY^cqDkK-w+v0nilpDS8My5|a<$Ios5UMiKJ>47l0$geg8S>o8a*6fyeWN2 zqef0mcMDl}X{Xt@eNMQWBGNJ@A+jP|-zj$y*)DvRByMt6}G(N$Bm=wucFD3ss$}m^k788hve?J-rvJJaTMMlPQ0b26>y+( z+GAZ>vErXlM_#Yxm4o#vucSHIUwL#7E) zj^tbm5X(NFu#`m1AOnzKu^K`iO1Z1~A#CTG`G`?%?}-x~aVp)17I?<;0W#=AZBll! z1-FP`nSrk3QTRRo@FG?Ltht4{*Q804SR6s65Q{R~a@p}=m3Wl~kJC`%#F@^1F3zD& zXy@H-O2`K#An@}T?P{-a&z6gRDw8Qsf;_q)r0{g01GXe7ZhDXjcMt&V3+^dGqjn(l zlI|68&EdDv--^bO+C}5UliL#QIQVFP-W5O13L^_msw>YQcAf;L96{k2g>sFEN0UBXAk z;An?tIf-ogp2Pp53w?eekroPZ`(0lnA8|LSx^CdPpXF3yjLyUn-@PNu@P9R;rn%oz zy6i$k&(#1V84W$YBzvYwJ#|g^rGwnv;w1Wy7a8`|C5htF%R8}q0?o5vu0bZ`blzsH z=+&bBy%OZCh}m|q=w^V*U|CS0`z3ujQo18le$43kMQ6KgjUC2CC?#&?t@#r+QHmJ{ zviT`{P$fZt3EXDMrMy4n;wiS&Z~1ckV(o)1dd;Z;3%!YU;@vAcYby@Sd(o`fe2a~N zCaUCKM23B>g%5PRe$Sbu>YaExd|^i#sU`i}Si5yTM$}!FTi3GfUS*&76jTfH%ggz6 z6T>T#IsK(=D6FZ+O2tt=A5SlKi(h}@^in+1Cx(^#2QJ^!Zupe%ZBJbFeS3KHjG!m# z4=wtJ4o0_iIK`1q&)Vl^ElCNQ`KpYMF2ahIw}tzt_tBN>BOI8{Aj3sifE@4bvyd|h zWkyzIQh7Hg;_B^6nQQ*jr9EC{g+G;yG!)2Zz0HzrFE8cE9Xn2Lt~I26 z*owE4c`h<0oMdh-Ek0U#S1b{C=DlP|^7144G)(AZV}O?e(&tl3mRS9MwJ@Oj(0wP;}gLJ(EF zp4v9Pa}!n>T0gmEj1I{)@Az-x9eJ^YKjRzzbpsr^k%`GD3R|+e;qzZAnf!h<#++i^ zC;oK#`!>?K*fKh-yz8&OE-6j3)#;Z|HjUY6l0}ks+I8!&$>c}foD+jF=aU_IOf%;W zUd=AY{QR_IPJHjTnlsn;G%Q)Qt?a~UXvjl6PsEpI%zt&MWm+sIBSE^mFp3Rc0t@qxdDdY*m+zZp_&9Ve+({Tt!Wg#PW<3*pBPj>qoJKbjR zc-iC88n}QJDSqp(%Ea~O59#CYJ*2&HX`ctlJ!-Z6cL&+-?+$Wu_C6<$ zgeBhl3mLFThTMku536v-ucc>gG!p0c=2^in9UAYs*lT>}`l=&BoR3z=fcp&Y_uab| zP3oO{_i~8qTX}WpqBjqYujA7aiUs3b&n3_G5Cjb%FBsuu+`>LZL2PQ&(A5cbO%n7`EH} z`_7#?R8G+QjU1LBe69q5{L@ul93T5nPdU*e&AZoJRr1bMInTnE1BP{Eh9DX|_X3f5 zyi#!;NWvv(tR<3!u*DbGyj}|2D`{wquElU$k%9Rz_Cmq+LtOIrSF4$BY6jm~nH*#s zN$HkpKO&nC?_lo*dwHWYi^wtf)9~NZUQN*#Su{EwnOGQum?B2;v5+5@M zI{N+H2xVn6CCffu#@xL9E^ScBYTvWxF;h>8dJAP_HT7fz?Iy6`t8`db!M;fTl-4qskm}wJV=FU7jcU!Crm zUjcr|xq0SeUHhQ;s(xir+z$VZaiF?mkz^d+u)i>kt;JJLj~CZ|zf~ijwS78n?D=M% zmfgC?%nr$aUiEkOI3LyCS5dz1cu~8RTe~7Q<>efF++?!*YgP*kR-oCDeMrbML1ymz z&XM2$^Xv8LGt8hN&up>mK4PPU@s?_ry|&kdiGTe;`Son?!5xgp5k2vojJPh0Pka=7 zA~$EQ%D73FduS2Hd`Y=bl3nVQpe~&|Z=v6ew9UtUj>%#_`6E*%O%farHWV4&5aajd z@>>?sM+mKwXh5F9mRK!`!-H;ke9)dI3+t4m{!BJ9ib2yo0!R^=Y{&lnTewh9m{1d$ z;0U_DIp;hfpRu?1{9H#(&3HuhBIYwi&tJW9lP0A+(@Xoe0hxRC>QyXoT`5vS$rd?? zW^a_w9_XZH_;D27Ew0{$NifvYb4du71Pri_QA}wO>C5UVAe*>bKb{%-eLDt0&rkpF zHgZF^VdrQC5dlmGbmZw#0AMJPkdSrmwR}d~fh*cGo6{7Lqj77-;;79x=NP&E>-A68 z%WDPRFg;|x8olKe*2@CSD)3v}k#tvfX56%k$cW=3IPhNxV|WyJX>VdeB>pGHQek=X z_h#}@@Z&sPrh*)U3EZ&NDZXB>NB^V8z;OSKYCgXpu{n>K05~*QXXew}(t}kL$<|f? zsU!reOv`i(gR}v_l#v=X*~I6+Ub^J&D}Nw>FSX?kWtJfhrs(9XiY{*}y8D@@-v^8# zY+Saq9g6wYg)N1<5svQl@SslogE)G={lkCEX_e zr3DywPjBj}zxqY<`Tyz{Pfi410h_uPEj<>uAy0m_`!+aIWX98pfs2uv`N3Do`8vY7BO;eD3ShS!oG*YFc>%8 z9U0fqU9vHzi`0UIfPpc#@3R-yv3s$kKRq(t^QVkSGt8F;{bB1G^Wgb< zbEiFJ9m6)$CS^9L>T{9;d+6$z%ve{co3my|y{w^K{F5FO|J1g!WO2PJP)_~5-RwTC zf}eHCtY7djJs+mqUbp5+YstA_fM9V*;L6=Qw{3fl@*gzhxQUYAg9i`HDsSI`pkxbL z_HU;GP&F%)NK$w~S5q@E{u4fF(Mymaw1<6=jKhZHwZB141+cYzTvpJGelq#wW=YWL zty^2^mXa?|NGPf|v*L4caqJnTwDRNAgPy1+kDx3~a*2Jh@Uy&Mqp4H%(F~@Zr~{Tv z`#;>ysPyL;pPZw0I{#5;@!~?&f^mPp^;uicV4!yKDjAs+XI&5E>($NCJ`cB@IkUj^ zGfKj#*~=IkuXsKDnACVDLZCSaK&$SA8Rn;cG?(Jy@J{UaeN`+t_{j9yzi?FTRGu8x zzh%+KkA6Dt(a}SfWRu9>HQi^W@ho%mcUM%p0*1%;VWL9Xz>2E37q0l9@li;DLG^-K zWgfVV9I5;F$Bl01q>pGtRLNZ6sj>g@aQdryOzb`L%Ifo}eI!t6()TH2O0tQ!LZ^cZ z=(MAdTcJJ03K3tgr4|GeA4cMsbd`6}h=55!wR1FE(GwRgs5o&acbe-!MHTe)@ z?3|dU#0YHJv(Fm6x@X7NI^gL^PJ2Y+9h&V4f$D=s`iv%QjB}UoZ!x^XS+=&gr{ln5!yx ztt7nacqdwAa0*H7g_{aXjd*6;xmT|}xIZVGndOCFEz0%{b8NW3+PP;ib7A~d8!Vjt zc|awNl^Bqo*bPC3;gMfBSAKE+2*@o@(-sQ%HQ>_ zTeq&fP-Z3Rkifg<_G#rv$e~S`*8ieS{4()hZK9|0<~#1f(O0>)GX1!OLd? zF!GaqSZ8(5in{`Z%hDIuYaVg>Y$~f=NFB&a>93)|u)dYWf$V`EZf)(qd?r0>lg{Ih zaSVrLW+FpdO=R4di4$`PZdzfQLS&5?*j1bVgQDyiMYEtKNR50OHf6wm;B{v)+3!eF zA`LzNyXC{}KM}&(BaL*%`6wif&xTCbynATCA6zhI;omehx!pH%kVc$2DC-jdeFC&b z6K){wgO-*%A+KyPpr1Sq*~T`Bk6V&X#mu{LX3AP-L`CNw6Wdx=byF{2CwFfXd`;nL zHELqWUmXy~et5I&7vd2e?-P&74qzzd4u7zN+&6_uA+?dv|2Clow?mwV4gQ$W7AG?<>xDa5T^6JIf?MEv^ zyZ&%cw)9`F*A+HVn8(8f>o-22c*&Xj&}XZ)N%bzwWuWRwiO-RU@M}R4nX;{d+2aKQ zi*>Bn76?nW9$YEQ3jiex8`Jay&gyS38alVfiMim2<0!4}RhR6?e=T!O9@yRksxymK z570>m@g1(pbOfP*>k;>cq@xniA4~7$dS_l)OudC6&u-tndlFE{+U}owqtTvrF6VW& zb8^TgjH=WIj-G4tvYGa?ki!ef%`xqWl56m*XORr6O2|9GeU;beS|F+SAXot15Xqmj+iYN zC*JPA+P!oD0fR&xNJ1mDDqT*MO*5?kYmWB1uS*tuPYKJc{xP?j5~MuY!l|U00w2|N z%brOGK9hBk!#^ucSMrMKxVF2w%|-AC=&x+aZ3iyN50C^vt@g%dW@9m~gH^ArV_-FJ zo@6j^#`3nI^Ec2qGxP|Q%5DpFhP(fv^Ie{Wh@cBHSpW;RXVj@%_a!x7c8yIy9bLFwBZ>-mq zz!y)IG5WJ4M3|<_tPgXc@DFC7@B&b(>|Y0`(o@P z0<`@5>)4nDFCFHzCv~{|^J@6q&u_z?jb6T}VcSaUdSN4Bz_(PW+W%Gn_Z9qXFT-#S`>dw>Bixst+*b_pR zTu%23J(4d~{g&5-_~PT3?E+0~+?)5{cR`6$!ZKKNTb5`2-Mj(*w=i{aO;hKv4W1rBr zK`ri~dzI4~oTKFt-h(O<=`?O6sWls0HoM4ygQUi%{Z+ zc3|hI(h2@rtppq<$_3&$S$*g^uxQ?PNRQET!!~c-I@*0o2?$x0Y@o-Fo;=UZ>jEt- z%FK_&#mg@#y_;_*UF_V!`~)s^tI%L3Pn(^Fi7zf%*SQs)Y<(&)aC}w!d(*h<;@d?f zB@UO>m-i0e>ghQxM0OTlI&a--@!!G&9S(P(I`Fe3?H)o1FN6#Ro_%ufAFZ`S_0ZKGXi!R5>_G-a1Oy-~UC<1x zG8zO0^HILX(SXfcw#MQrUqkJ0uaVH5Wg{P-Q z?}k6FjdS*0^&cvN^zvUrp6z&Md-ZK}>y*e1EWy5E_s-Ywgv@iC6ALU)cLx2K&t43K zsoAVL z&@{bpGH%dif;4XP*Dk%_qAYR1pr&K2LxJj1$| zF50%`-}QZFa@V3I%p)XfNQNOth_B8}gvHPXHIG4sIm8zKrvv_7CPbXjvuC z-m9oDnS?b4&MNiW73bTpdbd;NYpfT~NXdNF)CG-Q2uY$5eULRk(hE+C7akDH&sddb zO^JzoRaz#DXvQP&#QB^#B0%-n;1%zf2qO;gS3jmbm#0(%h3)>dxp;d@x}cBTXOGCvJm_$E6&?y>%DB593M2JIt4aLV(?>2oN!8VZ6?_F>Q31hBn>lMAlV zfL9>5AEI{3m4@@Tr3()37&Z5B@sw!}LI3)^;$9kg-T-$xjePtH#FLN+bbFTq*-1Nx zLs_H_Pts^@dNkDipx{Q!cCN^EK9rCaW^-#t%iah2EZUp4;6D@~X`h*p-hCT{5{`Qc zYU3JF`3WvcDhIT!>PiE|1CC|warXA$cHiHOhRyR5i6}aio{mNHq)p}J|Dz0fmVI~v z^3iU*%9TT2yoO$b{yxxevBnQc<-o_PHEKw>Am8=X-onV7Gehzw|LYTXzv*+f-9=~_ z@soi7Wz*T1Us=6}7l+d%K`SEt<+_eSnhu{z*JKmIpi3yUAkovu^>{Se0fal zklc4DjW$T*G}EJ^c(IO&{IgcvZg~8DaJw7x0@lpT_YL=NFHh&5`SCtO+VKdVK%7J) zdFr9|@|nBq9QY#Z0(wmiBEz%-Z(&a)X0&I znl&>H&`$0$)WO(uO_Qp}{?fz9Z=Y70Mf@W}(LI|R$Lw6NuHn=v>y*228)LHCiOGbh%#m`xd+*+t=AQ|kMl#ftb3Q79Z{sd9 zC*g(e{Qq@+ptV~hehn2>%a2dfd=&(J-qlpIj-jAjLes}HouEHIM23X z%e&xvZ=W~YH~;CR8OiP)1`=$%r*(E@&6Bd=CHBRV2CHepOzhmY-Bug0wEW?& z^*0#5ylpk8rtw@z8{J_VC{HkfA$xlSym(~w%Q|%X(*i=NK%wKzd>-E3<@{|f!`vx6 zrG2{hiCxO}xkFjs->;GY3}nRi(cf2s(;3^9p_UZ)3bzUB3<;QF7sNvIpBgRd{OWpG z+!mnkGwF`Ghm9-BQ=%)f_kQMXhg8~26@AJ6B32C~!o=k79rcQEp7dW+v}DY!e+{yt z<~4>8GyTnj3Fd(|+=n&u+c7?sJ^oP@W%(S*A486Q4;+)zgWI0{ipNq*)hzf~zdCsAS3e!3K@xI`4?9$zX)hU-&RN z$hj7M$O6>;{B*7eK?PV@2>Vp)YD9{4j<$K30A2eAc}f(N3SpXSpF0>;sWgXewW5G|Ll-xZ$$dlfs z>Y|KrG~eZJGNi2o+o&2NK;R5`#*oLcI0xZ#_eBjgbPbrIe;&1?J&!jC^9Qi8=QJgV z*QB{Ya3EX?z4T)m>T@{dIEy_02>ZWjoJ?BRV3kcD8mnwbUkGSc1>OPt$~ZDd z;uHQ+IVD>ij(uTCk0RKPh>n4j0DN{0{@j9vSSjgSVqdV)XFD1p=pf+AIuNMrF!W}f zR|~0iQ3tC#2v+4pF+UaKkJ_~m2hL&)OPI(%>`?K z5F5`qPA9&A#uQ*{IbJ7iKAF~gk&?wBHL~_tiIQHh3=SyI?C8?O#jAP2GfzMC?OxDi zqOSM~U+#SZ(e!lgE^+xtVQ8z_aoOXY^$cg*`X*c?jk%p=&lcwP+E`vJdD_3_^oxG6 z*txuQbXjt+;wuuA9kDN-OJrQ^tE@9?phTArYWV14`;M)ed)>P@$@9FdqjW*nm~@~! zqsPq4D?hqfe7l~Vx*sAPfFlRr zR;f-pORXCKydvTp-jqz7)Qi19O@vuWis1Z^gEZzKNjSz{^P!2k0O>aP<@@v}1S#X) zXt46~VT7KqxK{u(*`p}(9+sgTyr)+Ifh2INa*_xk6;@F>RTbgc| z0ZB?6Qt@@J(26Y8#O~(c$Dtf~p2#4^dYR^#vwkR6x@M>by4DV`yhp^C+0ANWP$_Xp!)$?Y>n z@=lX?0ze&4UzuiSeB{`5kIMt|y}}>iFp&6Ufx0R`wyQj@GD_b*e(CYwcBqB5ZLW|P zFCgdY)fW?rYH6%ixJW2u%iU$riDS7Zu~07=L$+@Wm8T%wj>r1M(&MFvdy^-q{B@O{ zKW{HSl;*MKlFKV=$`;5NOIhrG&t+fLEzqcyQ=h?zJ|H*1weJmhQ-Zb_n+IwKKIs)6 zN6QH+iQ%(2NFPR3TY$zFu@D;Hc#uA6&JcjSmfGWY=7_Ryk(C>DUFtJSNk4+s6=8Gv z2#U(`!33&r#;o+(=bXw{t}lChv?zCImeIN z|HnNT@G*q{(VueYU9;@jzsGoX-|VG)@irJYUxDwwh_1+w{uhZ9hx-Q<=xfDfY)%E2bSs=-|k*H4tV4KbEWy04{-nOlD?`x{T|(zEA-OX(>5H>lLw%ZVD@L~~d&na`{SV+=`jIx0j0`Ths&uJ3Vhrf|e5Z9N4B9jp?Hc}YYM(FK zn9ssSHZhtt>umVK8k`Z-Dq7v_y^KC${4Pa(c$@Bo4UNlp)+4+R)ORxk+JMKIzRRB4 zhMN^G;1;AnY+TzbXQSU79*9w)U&2){`#vd7Lh}a_Ma|- zng&EGaB75Il9jmxj@n10e4P;uE;X!ReaT}bnv}${wGSG{)kMXZn_e)gWaRER+j@U# z0a`V;?$;e2ezJ+lXV*i$=~wEzpf!|$NR(;MP$n@Zh<;SEwV`Qg?-MD;*C8UM4Lj7& z9r0l96eC2(mWFRpF?=VToptdeo~-em|46tHpkg*n057I+dQoje zy1Ad6jia1voE=~9z<~pAsvuKF4?xXU>jL}4>8xL26S|d``j>?TRCe#y?G@4GmvSmVsqWoh|2f1gJNd;DIEG+T3!Y!$@1nmKEQYR- zAaQ2P8QHnk-a49^JAkN?{e!x?j_m7!cGCRSMloPzFEn{I+$l~qAi2zSM?FOCiWcs1 zUGpEo=|5|jo13d5YgjZ^KUcz6^yb_5p|lWWD2Ot+C4xMM*3G-BOmf8|7 zMSb^zysHLw4=>!>>=n0#e!qx*eFx9{KnpbjRpt@oeG|Lz>THi`(u61-=)ItUG)z#^ z?S|rpMocje$c)ey$%eQy&-y{>^)P!$S6Dh|>!Q2v2$DS+LRlLXOHXT_y=x~8BSy|01=Uq)oj{<|J6@67x8eSZBl zMlB}IO6Zqkqra|{T(D@7*q%MY9Ab7=|0)fluP%d@}^!L9w zHgGNwn_3pS3bSYmNiVayO>Wg9(jZomjk1H(o#@i}RNpuqfH$f);X373hwg7u5}&IB zKL$4bmxSle&dwPP%FP2mp~M|~f9MO9?DZ&~EQ@2y0YkFbNAV~i0-Tn=^;$!f7(9gw zt=du{h=XSQ2`GkW5BMI0+zh2^Vw}LK6!T?-|8PeaMOa06jTxhVWaLuj>W>pG*OQWL zJ?f^qM@5~ToMh_z_8;r?gppz^jizT@IaddU3Msds*-G+@TNj3k8f!+Z#RW$o`}pHC z1(HcmU}#pHnPhgDxd9gvMUr%DfQ}Z1Rs%Uk4)diWag6C2wZs*s1LWTXY(z!Qt%{9O zwYPsdDCP_lUKLZ1lk47zLoxvc;_fe!b}i5v^Jwl~{wVikgMhf?s6j=`I~=A&d|iVP zj}(afpjAq65d4O@52=bhzM~Mk(NX^fHhYNo+kyFroeabI%Hp)eNp@|U9NM_CXn))7 zv#iTDHJ6>2LoOtigT~UQ`_MMIjh=D~c+(rSdWOPQm&6n_L{-IH-m|92EuOmUZtf~- z_g>j|uk9Y7p!+;N?#pKPq;fEs5F9_&>YC)U4KMEr+`pF-d-RLdG8okt9VN0k!#ua; zfI5I7)Ny;oAN9I^+e@ZP)t<6}l^N5S031-sTOk476^SpKqlH40)Fc!#kgV7B5Cx+e zO)*vwmz}fGlr@miQ0s@AvS7|MFd8d{z3NHN(?-m1l<4c{yd$ZD8*T{Lhe;Lf_#k4Y zK=mMB1Kq1;Ra1iJ|8G++Te4)CMS#Ti*KN2IapF%)bQbB5JEIJDHH`L?Di9$BHWHtm$={BuHHJ0&=P6GU@%()3~cPCr!apQzAbeJhm=Cz?|>zIp?&35 zs7yA>9gX6T24h2R5y9E8Sync+K$;YZaN%JHH51px1CcNd7h+z2x67%PD)3QrE7z>D zyoEe3@#qM`DCw%?;H2Iwo3ugLVVdC%>vJ33o2G2o_U@A)#!Kem<6w_yhoCf{Gxrg} zEwD%vj{Zx(A0urXmCk*M6VI`DHL>DBI8|LmvOtnaPqFQ@DsRnZXNWnS7u-RA7+Cq)(PH0)2l6rqDtRNC%e@ zRyf+%cmgc}_q2~%AXdBtip&_W$-}M}Kxl0M+(714ybIFJM0VnyKm|ngtaAbPE;_Y5 zL%EAEBKiRA!e7dC8M<=3ea4Fdlf$~8(+RstN_2FJgVf=IeQ+C!66Nhnbii8T1s*an zF^LLYrI*$S5W_ouHyQ2$JRKbHElSfZO7T;V^G)Li&P2k*h@)SynEk~N1k%Y4h%K`y zkina2pI*FV8P5qMk}Hb&7oVF2z)Ov&)^pIy2*R1S;T$-9aj@KY{e#G0dcm8~oDY?` zGAI=ToS~VU)g8qSK@teUyrcv3?~n~PuVv0l)Pcbmhle_v+)|C{5q_{R`M`2<6ZY*u zk!I2sg-E;Y?QMpcRiu;$$lh(&dQ^`{Lxay}C*BchG=o#b5sf`45X}IE zNuv$?%^}1GQal98Tnu29w7x(8lq-VXU}FFFg~tJ=pLZNSvP^e-5;f;&0Y;!+LKU8l z&Th-&Sg+BiXthTG{{fjG>DU8!aY+A=Q|sNsc%2wT=>@Ci9C9&rn#u9As4yQQ(_qTM zPJ*xw93XyqIPoZVgk|FEwl4Hi{Run+Vcu~My$1%Sot9<8aAXGN3^et!t0;+c#|GYH z44}-Fl??#JWws6&0VTm%*qgxqm1euoST<&F+E`u*S0x!?MLbrxExGIbDPqAwwSfr( zL|@d6L7R)m3X=^HW-{&*Z;(qXTj^1vo1U zH@eBx&lePq#suLdl+;?EcW5RG277=rN%UcZj5 zwUK2Hnhswh#wr~A z&X5&McdWu5dj!D*_X_D^_u4unzY)P9<1o~zxpKt{*oQUHmtj7zJp|J_{r&u+KrFyk zd=#Qr0*_;-sd*Q->?WZ}5sg#DEj1nXJbyMj0nj@COmZyzGTrDZ#)cPm{U~Dl`~i|2 zxARdiVD&CxFU!UkS;pa3!&AvM;TTDxMhh(L+g`D10T;I9r}e#PGKmo~`dnTO(24uM z7)U)&4wqn7!r%lmo(lL`)-1kB_}CrPom5SM z?Iq`~$y3J+XqX{ea3*;R+;r~~)}}mUfXaEW4l~W1vx&rZ1Tedz#aropB+oo=$lG-9 zS>4HW#f-5N%lFLKE@c;O27Ab;TsU`t(Q3;(iG^mi3H#-*IC=IDKL&QP7UZaRZaD%J z<%8FM81}+4T0po`%sip%iuz3kR;GD^0v3~^716n7yT*;Z6aH6lQ1;r&?(9s z`;Xyrw)X?JooX(lzEKVc4vt%=;M@tCJaRY1N5H&_=d;WGx%{QhEm8SL9^l^~D%M|} zTgFehs!X~-Asr>6JE8b~8|66l$TV(*!(9{=UAEV7WV?rFBETPITpL)AL%r zTNT&&oyxLCU1W~KT;8tvUx-sE>@3KgjxHx+!H)I-5M`jWLDt-!mItHDJwyQqvH=jjHt z8S(rsL=knZsnOebtUx%>Vj+pu)T`?01x$mgGY1y+rEPa`1y-hKIK%<_KgN~7=d(E| zRjeCd7~;-W*Tg)mRSJl~wLjZOpRb2fp9rP+C>gIu5~Qf;_JpcPp7M6&q#Ez54;$kp+L<_@p$%%m6tMA+D zAkW2dM4lZ@d6?GH>>1>bBSng@v|aO` zA=au2UmiYF38CCSB&4o66^czP=<&1{7(!|0>87S@XejllV)fGO zT!4y;ZB+NlNLJMuj|#O)1&!qP`DHIwBQK?d_9kFkI{Yn)xMX@B+h)<^cI z89k3h_7Bgx35hQ*IvULgrx3Q<;K0S=AVoT(Ri@iT1Ah}#JO}P1mmCY9);M8s9?>gRt(qL-h8BtMm8i>NvDBp|$Ch}%k5@r^gvV5q7bAUq30unAb1=J&w{#0rg1L;_ ze@P0Ci!A*ni(|v)+i-Drx}LJs5!{V;07ETo1hn*NHJ5RhWTl8NPN?|Dk97_lxK1n@ z6PS*pj)WwCW#sV*YBU+Do`tbVXca;q3R~6!ZPCKg=031ZRwWr{%ZER6{L~z>gL*%)W`Ood0Ff zEx83YY1*E8O$FKgQ9?5lKBcdzI#Fe@=rj>p>&d&tyG7qY<&KQvFnS1gsdKh?%0qh4 z@tAot6p)HKn-)t(qi|oXV{;NLsBG`9wKZc_U&VA_KPd>_QGs{ zY(;TCN{#W!^;xT*)jozQ&;`#+8I1Fr45wNCmHlnw9G9*i1q2c*GF_Gkwz|TU72hF@L#fFuL46~xhJ2b(W0U}M^XM=J z+34bot3;P92N7(JsFc{YtrF0yfs;T}!%BuCyJCV+3?8Znxf?3)BmMvEb@RBZ20p@y zF6kf}Na(k9bYQRl0$#_;qb5c0#~;yL<4kl7HCQCq17)ZR>hUJEw1!ItQ6-?>La{+0srAwW%MG&KvXI~$AZQk#QM_juqmqwi zxNJ{I_&H3&_49|DHua$1(9xa(Ne2Bqm6nC((nt)_j!P2Yq>_}8(M4S2|1e5xa@CNn|Pj zP7O;HRaJsx@!yAnt9^uP0BC(I zA~6+QY*pvZv#5U{#vVXh1c~VYnxO<-&-ebbU?GJ`)My)Cn~*E+x#p&dc&TQAOkgRB zV`_DA$B{tlZ9uq4c&cJC>*4LuNrHI3oh z>&e-7HGrTri1lm9mYlmvD7thT23VozaNzTpkf#a`GQjhs2|&8a@v2az7xM!PtD=w5 zwFDX^BqjYCZ1e#%x#>B81~`L+mbTQ3^yPfGC>>Wq4-t5y@*Avw+q)OwArBJ~RTbza^(lp074~Q;3EK5nNQZ=*xZ9fhPOEy zD|+Q!E?gU?yMn<1x7E0!hiY>A*Z}TvZMElB@|b3pb{^RK@lT&>L0^9;Bo~B_#(TVQ z;j!^{AIAhC^gRepmqmd`g_aUV?D4ao9^^UasSv(9C^Mbo$L|dXtwp3&iV zYlFDRLqakb+O+aLI#w3pWVHaF;E*er3f-vp9ZwtkJ$kpz6SNdeLFg(4oSvF4+{IiL z6;Ij;y;zT&qZD4$mdXql%1bc$35$a$;7=iFwW|KVckgGBQS zJ7}HD?XS@rUpxBG z)4>9!Kk&ZP9n$~i-zqI3E`Auj;95XsnElIvH~~6X0=pZex-*1W*Dqi99r|BLXP$2~ z7&hPr2_+?C1ZWk!`6&>u!O~Uz<=)@4pJ4CzBe>}{|84)1bw#}fxi@Sr@3PsFk1A?A z-O!(8tvS_cYyW`zx94)0_7j+9|JQTHLivKYqmAbx8iN&VmcG%j@gIfPvkdfzG+3BT zn>RaWN!hMr)6*xAleloY1{a-In3){R=!#*Bso_A!#n2(CHVyHcjs&5t#vP$G7|JGfA=F6&Af0i7V7LzY;Vx4C~BZ1PxA&}QF-+I>Xr z>k6YiMXekLn!=FnfxdW-VKBd2_+mzVCDGm$8F4STWyV0X&y zi#^Po2(bH*Nw2IL)t`o5eFy#(4k)8nd3jeDO5<#XdE)ST>ZBZX71n+I@Xx%aY@V_# zsm6H}z-c6@d8qJ?WVooeSQ*w`)vG@dL->BJyB*NFE`f`ilFc=(*h-*@(bV|2Xxf@?D%#!VAdfD{b0{b<1h_*{ zKUH5x$B{c5>(UKm(;+c~S5bib5zxnXdL}7@JPhzr55PayZIIn)4mOXJswgS%vDt-h z`y0K9KjEG_ zKm&8w!$bBt*%yrv%PBqcQX-1d1klZiza&Sh8`{4V88+8w+(qN`y@hWNMA3jWyo)R# zyH~>k(Il#-$4<~bAEq%}x8>pmZgJJ1eKw={2qL8rToVX@gB* z?FA0E3zlX5fozN*HO8G4zv1A3%<~v_FA!eut+T1@#`Ze%(%ZgcmhsR-74n?DvJ6_hYZ>KHXGxI|+Q@9zGfhTw47D=UUG z*BmtsCVp0t>1^2H?AZAv)xCVXo;oM41k3S;|LHXli>pvaORvPKU7Bv!VhTbK)X=GI z0U^Ce@8Dc~#2R1}uqm;V4@1El4h^bfB(y!XEk8Ln z-Qo1DMB_;cW-USh>bgNSGT$wCXC!*vu$O21h*Zn(m$_glc94np(x`VJCd)bgmAXgv z^^KQ%NwgQwM*9)60c~k1ZTO0_oGm4U29R@>^az1s?1c99|mdWHPVi~)7JN@QK- zAAT>+F+N|q54VWVEIF$fHyeqmWDkUY-vj+6f6t_gIO8-^0Y(d~YIr|>e|$KJaFRx+ zFuu59;FnnZIkSO-NJEWX*k->Kdk;7hDfNz*Q1$F*?vQ=WUpynyAl`bQr{Mcnzc{Te zV1LbykD}ak`tP81 - + diff --git a/docs/references.html b/docs/references.html index 9d0372b..93c5d99 100644 --- a/docs/references.html +++ b/docs/references.html @@ -24,7 +24,7 @@ - + diff --git a/docs/search_index.json b/docs/search_index.json index 4ea2f48..69f330a 100644 --- a/docs/search_index.json +++ b/docs/search_index.json @@ -1 +1 @@ -[["index.html", "Recent trends in movement ecology of animals and human mobility 1 Introduction 1.1 Abstract of the manuscript", " Recent trends in movement ecology of animals and human mobility Rocío Joo, Simona Picardi, Matthew E. Boone, Thomas A. Clay, Samantha C. Patrick, Vilma S. Romero-Romero and Mathieu Basille 2022-03-06 1 Introduction This is the companion website for the manuscript “Recent trends in movement ecology of animals and human mobility,” a quantitative review of animal and human movement literature in 2009-2018 from Joo et al., serving as the manuscript’s supplementary information page. The R codes for the analyses are available in this GitHub repository. A version of this manuscript is available as an arXiv pre-print. 1.1 Abstract of the manuscript Movement is fundamental to life, shaping population dynamics, biodiversity patterns and ecosystem structure. In 2008, the Movement Ecology Framework (MEF; Nathan et al. 2008) introduced an integrative theory of organism movement—linking internal state, motion capacity, and navigation capacity to external factors—which has been recognized as a milestone in the field. Since then, the study of movement experienced a technological boom, which provided massive quantities of tracking data of both animal and human movement globally and at ever finer spatio-temporal resolutions. In this work, we provide a quantitative assessment of the state of research within the MEF, focusing on animal movement, including humans and invertebrates, and excluding movement of plants and microorganisms. Using a text mining approach, we digitally scanned the content of >8000 papers from 2009-2018 available online, identified tools and methods, and assessed all components of the MEF. Over the past decade, the publication rate has increased considerably, along with major technological changes, such as an increased use of GPS devices and accelerometers and a majority of studies now using the R software environment for statistical computing. However, animal movement research still largely focuses on the effect of environmental factors on movement, with much less focus on motion and navigation. We discuss the potential for technological and methodological advances in the field to lead to more integrated and interdisciplinary research and increased exploration of key movement processes like navigation and the evolutionary, physiological, and life-history consequences of movement. "],["data-collection-and-processing.html", "2 Data collection and processing 2.1 Identification of movement ecology (mov-eco) papers 2.2 Downloading whole manuscripts 2.3 Extracting the Material and Methods (M&M) sections", " 2 Data collection and processing 2.1 Identification of movement ecology (mov-eco) papers Fig. 2.1. Workflow to identify movement ecology papers. 2.1.1 What is a movement ecology paper? We defined mov-eco papers as scientific peer-reviewed papers that studied the voluntary movement of one or more living individuals. This included humans. 2.1.2 Search keywords We used Web of Science (WoS) as a search engine for the papers. We defined mov-eco papers as scientific peer-reviewed papers that studied the voluntary movement of one or more living individuals. This included humans. Very few papers mention “movement ecology” in their abstracts, so we did not use “movement ecology” as a search phrase. After much testing, we came up with the following groups of words: Group 1 - Behavior: behavio Group 2 - Movement: movement, moving, motion, spatiotemporal, kinematics, spatio-temporal Group 3 - Biologging: telemetry, geolocat, biologg, accelerom, gps, geo-locat, bio-logg, reorient, vhf, argos, radar, sonar, gls, vms, animal-borne Group 4 - Individuals: animal, individual, human, person, people, player, wildlife, fishermen Paper abstracts had to have words from at least 3 of the groups above to be selected. An initial search revealed these keywords to pick up papers from a variety of fields such as biochemistry, medicine, physics and economics. As such, unless they had words from group 3, paper abstracts could not contain the following words: Group 5 - Missleading words: cell, DNA, enzyme, strain, neurons, atom, molecule, lymph, cortex, cortic, neurotransmi, patient prosthese, eye, particle, tectonic, counsel, cognit, market, spine, questionnaire, sendentary, insulin The search on WoS was made over a final selection of 273 journals; made in parallel with keyword tunning. 2.1.3 Cleaning and filtering results in R The grouping criteria were applied to the Topic field in WoS, which searched the Title, Abstract, and Keywords sections. We downloaded the search results from WoS, which contain information on title, keywords, abstracts and authors, among others. We downloaded all references in raw .txt format as it was the valid input for the refsplitr package available for R (R Core Team (2018)) on github (https://github.com/ropensci/refsplitr) (Fournier et al. (2019)). Refsplitr reads in multiple WoS files, parses addresses, and performs author matching. We used the references_read function to compile the .txt files into one data sheet. To be sure that the papers shown in our search in WoS were respecting our search criteria for the abstracts, we applied the same filters described above to the downloaded search results via R. In addition to the grouping criteria we filtered by Document Type to only allow ‘Articles,’‘Proceedings Papers,’ and ‘Reviews.’ This process yielded 8007 papers. The code can be downloaded here. 2.1.4 Quality control From the cleaned results, we took a random sample of 100 papers (i.e. with title, abstracts and other features taken from WoS). We then read the abstracts and classified them into “mov-eco” and “not mov-eco.” If the percentage of mov-eco papers (i.e. precision) was lower than 80%, the word criterion used for the search would be improved (e.g. adding more words, editing some, changing the rules for the groups). This is how we came up with the groups introduced above (section 2.1). We obtained 90% of precision. That means that, from the papers that we had, almost all of them were mov-eco papers. We also wanted to obtain a recall or sensitivity rate to quantify, from all mov-eco papers in the literature, how many we had in our search results. As it is impossible to obtain the “real” list of mov-eco papers in the literature, we instead looked at the list of papers published by the journal Movement Ecology. We found that 69% were in our list. An estimated sensitivity of 69% and precision of 90% implies that though we did not get the whole population of mov-eco papers in our set, there is a high certainty that those obtained are mov-eco papers. 2.1.5 Possible biases We have no reason to believe that our search criteria have introduced biases to our results. The relatively short list of words is due to the fact that other words we tried were actually reducing our precision, providing us many papers that were not about movement ecology. Of course, it is always possible that we forgot to try an important word. A possible bias could come from WoS: we were not able to get papers that were not in WoS, which depends on WoS agreements. 2.1.6 Differences with other approaches selecting and analyzing mov-eco papers Holyoak et al. (2008): Their goal was to find papers about movement of organisms or gametes, so their definition of movement ecology was somewhat broader than ours. We were inspired by their procedure, and tried the terms that they showed in the paper that would be consistent with our definition of mov-eco. Like us, they used WoS to build their literature dataset. They had a two-step criterion to select the papers. First, they screened the WoS for papers that contained their keywords. Then, they narrowed down the selection by excluding non-ecological journals from their initial results. Two of their coauthors decided on a list of 496 journals. We applied a similar procedure but using a modified set of keywords because we found their criteria to be too broad for our definition, and then two coauthors (R.J. and S.P.) decided on 273 journals. Among the remaining articles, they selected a random sample of 1000 papers for quality control, rating them as relevant or not. Their overall success rate (similar to our precision) was 77%. Fraser et al. (2018): They also used the WoS. In “Topics,” they searched for “ecology” and either “movement,” “migrat,” “home range,” “dispersal” or “track.” Their combination of words was too vague in our opinion, and they did not mention any quality control (e.g. precision, recall, specificity, sensitivity) statistic. 2.2 Downloading whole manuscripts Fig. 2.2. Downloading movement ecology papers. We used the fulltext package (Chamberlain (2019)) in R, using Elsevier, Springer, Scopus, Wiley, BMC and PLOS one API keys. We downloaded the articles we had access to, as xml or pdf documents. We downloaded a total of 4060 complete manuscripts, representing 51% of our list of mov-eco papers. The codes to download papers can be downloaded clicking on this link. 2.3 Extracting the Material and Methods (M&M) sections Fig. 2.3. Summary of procedure to extract Material and Methods section from each paper. For some analyses (section 3), we needed the Material and methods section of the manuscripts. We created codes for .xml (click to download) and .pdf (click to download) files. The former was built using the functions of the xml2 package (Wickham, Hester, and Ooms (2018)). The later calls the readPDF and Corpus functions from the tm (Feinerer and Hornik (2018)) package. In order to write the codes, we took account of the structure of the papers in either format, and aimed at finding section names related to “Methods,” “Data” or “Statistical Analysis.” Not all papers had an M&M section (e.g. reviews or perspective papers). We were able to extract 3674 M&M sections (46% of mov-eco paper results and 90% of fully downloaded papers). The possibility to download articles was conditioned by openness of data from the publishers and data & text mining agreements with the institutions we accessed them from. Fig. 2.4. Number of articles about movement ecology of animals and human mobility identified by our algorithm and per journal. In blue, the number of articles that were downloaded and a methods section was identified. In yellow, the number of articles that were not downloaded. Only the journals with more than 20 identified articles are shown in the graphs. References "],["analyses.html", "3 Data analysis 3.1 Topic analysis 3.2 Taxonomical identification 3.3 Movement ecology framework (MEF) 3.4 Tracking devices 3.5 Software 3.6 Statistical methods", " 3 Data analysis A total of 8007 results (papers) from 2009-2018 were obtained. The proportion of animal and human movement papers from the total number of scientific papers extracted from WoS was higher in the last years (Table and Figures below; download the code to reproduce them here). Year Movement articles All articles Proportion movement/all (\\(10^{-4}\\)) 2009 485 1139611 4.25 2010 479 1186928 4.03 2011 564 1262956 4.47 2012 666 1323677 5.03 2013 791 1398009 5.66 2014 878 1438134 6.11 2015 978 1709898 5.72 2016 937 1775745 5.28 2017 1073 1838351 5.84 2018 1156 1928507 5.99 Table 3.1. Number of articles in movement ecology of animals and human mobility, and articles in scientific literature in general, published from 2009 to 2018 according to the Web of Science. Proportion movement/all refers to the proportion of articles in that studied movement. Fig. 3.1. Proportion of scientific articles extracted from WoS that were about animal or human movement for each year between 2009 and 2018. The solid line and grey area represent the fitted local polynomial regression (estimates and error bounds, respectively). Several dimensions of the mov-eco literature were analyzed: research topics, taxonomical groups studied, components of the movement ecology framework studied, tracking devices used, software tools used, and statistical methods applied. Depending on the dimension, we either analyzed the title, keywords, abstract or material and methods (M&M). The sections used for each aspect of the analysis are detailed in the following table. Dimension Title Keywords Abstract M&M Topics X Taxonomy X X X Framework X X X Devices X X X X Software X X X X Methods X X X X Table 3.2. Paper sections used to analyze each dimension. 3.1 Topic analysis Fig. 3.2. Stages of topic analysis. The topics were not defined a priori. Instead, we fitted Latent Dirichlet Allocation (LDA) models to the abstracts (Blei, Ng, and Jordan (2003)). 3.1.1 The model LDAs are Bayesian mixture models that assume the existence of a fixed number \\(K\\) of topics behind the abstracts. Each topic can be characterized by a multinomial distribution of words with parameter \\(\\beta\\), drawn from a Dirichlet distribution with parameter \\(\\gamma\\). Each document \\(d \\in {1, ..., D}\\) is composed by a mixture of topics, drawn from a multinomial distribution with parameter \\(\\theta\\), which is drawn from a Dirichlet distribution with parameter \\(\\alpha\\). For each word \\(w\\) in document \\(d\\), first a hidden topic \\(z\\) is selected from the multinomial distribution with parameter \\(\\theta\\). From the selected topic \\(z\\), a word is selected based on the multinomial distribution with parameter \\(\\beta\\). The log-likelihood of a document \\(d = \\{w_1,...,w_N\\}\\) is \\(l(\\alpha,\\beta) = \\log(p(d|\\alpha,\\beta)) = \\log\\int\\sum_z\\left[\\prod_{n=1}^{N} p(w_i|z_i,\\beta)p(z_i|\\theta)\\right]p(\\theta|\\alpha)d\\theta\\) Here we used the LDA model with variational EM estimation (Wainwright and Jordan (2008), Blei, Ng, and Jordan (2003)) implemented in the topicmodels package. All the details of the model specification and estimation are in Grün and Hornik (2011). The model assumes exchangeability (i.e. the order of words is negligible), that topics are uncorrelated, and that the number of topics is known. The most commonly used criterion to choose a number of topics is the perplexity score or likelihood of a test dataset (De Waal and Barnard (2008)). Basically, this quantity measures the degree of uncertainty a language model has when predicting some new text (for this study, a new abstract of a paper). Lower values of the perplexity is good and it means the model is assigning higher probabilities. However, the perplexity score measures predictive capacities, rather than having actual humanly-interpretable latent topics (Chang et al. (2009)). In fact, using this score could result in there being too many topics; see Griffiths and Steyvers (2004) who analyzed PNAS abstracts and obtained 300 topics. Hence, we decided to fix the number of topics to 15, as a reasonable value that would not be too large than we could not interpret them, or too small that the topics were too general. Fig. 3.3. Schematic representation of the links between words, documents and topics. Each document is a mixture of topics. Each topic is modeled as a distribution of words. Each word comes out of one of these topics. Source of the image: Blei, D.M. 2012. Probabilistic topic models. Communications of the ACM, 55(4), 77-84. Fig. 3.4. Schematic representation of the Latent Dirichlet model described above. 3.1.2 Preprocessing To improve the quality of our LDA model outputs, we cleaned the data by 1) removing redundant words for identifying topics (e.g. prepositions and numbers), 2) converting all British English words to American English so they would not be seen as different words, 3) lemmatizing (i.e. extracting the lemma of a word based on its intended meaning, with the aim of grouping words under the same lemma) (Ingason et al. (2008)), 4) filtering out words that were only used once in the whole set of abstracts. R packages tidytext (Silge and Robinson (2016)), tm and textstem (Rinker (2018)) were used in this stage (click to download). 3.1.3 Model fitting The parameter estimates of the LDA model were obtained by running 20 replicates of the models (with the VEM estimation method), and keeping the one with the highest likelihood; click here to download the code. 3.1.4 Model outputs From the fitted LDA model, we can obtained: \\(E(\\beta | z,w)\\), as the posterior expected values of word distribution per topic, denoted by \\(\\hat{\\beta}\\), and \\(E(\\theta\\_d | z)\\), the posterior topic distribution per abstract, denoted by \\(\\gamma\\) in the package. The \\(\\hat{\\beta}\\) values were thus a proxy of the importance of a word in a topic. They were used to interpret and label each topic, and to create wordclouds for each topic, where the area occupied by each word was proportional to its \\(\\hat{\\beta}\\) value. Fig. 3.5. Wordclouds of each topic based on \\(\\hat{\\beta}\\) values. Download the codes for the plot here. Since \\(\\gamma\\) indicated the degree of association between an abstract and a topic, we obtained a sample (click to download) of the 5 most associated abstracts to each topic, to aid the interpretation of the topics. Based on these outputs, the topics were interpreted as: 1) Social interactions and dispersal, 2) Movement models, 3) Habitat selection, 4) Detection and data, 5) Home ranges, 6) Aquatic systems, 7) Foraging in marine megafauna, 8) Biomechanics, 9) Acoustic telemetry, 10) Experimental designs, 11) Activity budgets, 12) Avian migration, 13) Sports, 14) Human activity patterns, 15) Breeding ecology. For an extended description of these topics, see the main text of the manuscript. The sum of \\(\\gamma\\) values for each topic (\\(\\sum_d E(\\theta_d | z_k)\\) for each \\(k\\)) served as proxies of the “prevalence” of the topic relative to all other topics and were used to rank them. Fig. 3.6. Measure of prevalence of each topic. Download the codes for the plot here. Fig. 3.7. Time series of the relative prevalence of each topic every year. The code can be downloaded from here. A heatmap of the \\(\\gamma\\) values also showed that most papers were evidently more associated to one topic and few were split into several topics. Fig. 3.8. Heatmap of \\(\\gamma\\) values per abstract and topic. The corresponding code can be downloaded clicking here. 3.1.5 Model assessment We assess the consistency and interpretation of the LDA results via internal and external expert judgement, respectively. 3.1.5.1 Consistency In the heatmap, we showed that most papers were more strongly associated to one topic. To check for consistency, for each topic, we compared its wordcloud with one obtained from abstracts that were highly associated with the topic (\\(\\gamma > 0.75\\)). The two wordclouds should be telling a very similar story, thus visually resemble, with very small differences due to the abstracts being composed –in a small proportion –by other topics as well. We selected the papers with \\(\\gamma > 0.75\\), and computed the number of times each unique word occurred in the abstracts related to the topic. We divided those values by the total number of words in the topic to get a relative frequency \\(\\eta\\). We then created wordclouds for each topic, where the area occupied by each word was proportional to its \\(\\eta\\) value. These wordclouds were overall consistent with the topic wordclouds, i.e. most words were the same, and the words that differ provided complementary information to interpret the topics and understand to which types of abstracts they were strongly associated to. Fig. 3.9. Wordclouds of each topic based on most strongly associated abstracts. The code for this plot can be downloaded clicking here. 3.1.5.2 Interpretation To assess the interpretability of the topics, we performed a word intrusion analysis, i.e. people are given the task to identify a word injected into the top‐terms of each topic. For each topic, we identified the 4 highest-probability words (click to download), i.e. with the highest \\(\\hat{\\beta}\\) values. We did not use a higher number of words in the word intrusion tests as we considered that it could have required more effort from the researchers who kindly and voluntarily participated in this process. Then, we took a high-probability word from another topic –that was not ‘movement’ or ‘behavior’ as they were highly related to most topics –and added it to the group; download file with intruder here. We asked 10 researchers in the field to identify the intruder in each group of words without any more explanations, and suggested them to answer fast so that they would not have to overthink; their answers are here. We then computed the number of correct answers for each topic. A high score for a topic would indicate that the topic was easy to interpret using the 4 highest associated words to it. The scores are shown in the table below. The codes written for this analysis are downloadable clicking here. Topic number Topic label Score (from 0 to 10) 1 Social interactions and dispersal 7 2 Movement models 10 3 Habitat selection 9 4 Detection and data 5 5 Home ranges 8 6 Aquatic systems 8 7 Foraging in marine megafauna 10 8 Biomechanics 0 9 Acoustic telemetry 1 10 Experimental designs 3 11 Activity budgets 2 12 Avian migration 9 13 Sports 0 14 Human activity patterns 5 15 Breeding ecology 4 Table 3.3. Word intrusion score for each topic. Topics 1, 2, 3, 5, 6, 7 and 12 got high scores (>7). A few of the low-scored topics (\\(\\leq 5\\)) had a word intrusion that could be considered relatively general for the researchers thus confusing, such as “tag” (topic 4) and “data” (topics 9 and 15). Moreover, the two topics highly associated with humans, 13 and 14, also had low scores that could be a consequence of the researchers (all ecologists) not expecting any human related group. Overall, this analysis shows that only half of the topics are easily interpretable using the 4 highest-probability words. Even for us, the researchers involved in this study, it was necessary to look at the abstracts with the highest association to each topic in order to be sure of the interpretation of some of the topics. An alternative approach to model assessment in topics is through topic prediction on an independent data set, but this should be performed when the goal of the study is to predict over new data sets, which is not the case here. The word intrusion approach is not an exhaustive assessment of topic interpretability, but it allows putting our results into perspective: some topics have a clear and easy interpretation and some others are really hard to interpret. 3.2 Taxonomical identification To identify the taxonomy of the organisms studied in the papers, the ITIS (Integrated Taxonomic Information System) database (USGS Core Science Analytics and Synthesis) was used to detect names of any animal species (kingdom Animalia) that were mentioned in the abstracts, titles and keywords. We screened these sections for latin and common (i.e., vernacular) names of species (both singular and plural), as well as common names of higher taxonomic levels such as orders and families. We excluded ambiguous terms that are used as common names for taxa but also have a current language meaning; for example: “Here,” “Scales,” “Costa,” “Ray,” etc. Because we wanted to consider humans as a separate category, we excluded “Homo sapiens” from the search terms, but used the following non-ambiguous terms to identify papers that focus on movement ecology of humans: "player", "players", "patient", "patients", "child", "children", "teenager", "teenagers", "people", "student", "students", "fishermen", "person", "tourist", "tourists", "visitor", "visitors", "hunter", "hunters", "customer", "customers", "runner", "runners", "participant", "participants", "cycler", "cyclers", "employee", "employees", "hiker", "hikers", "athlete", "athletes", "boy", "boys", "girl", "girls", "woman", "women", "man", "men", "adolescent", "adolescents". In cases where words may be suffixes of larger words, we used regular expression notation to exact match words, e.g ‘man’ must match only the word ‘man’ and not ‘manually.’ We excluded words that could have an ambiguous meaning: “passenger” may appear in papers that mention passenger pigeons; “driver” may be used to refer to a causing factor. After having identified any taxon mentioned in a paper, we summarized taxa at the Class level (except for superclasses Osteichthyes and Chondrichthyes which we merged into a single group labeled Fish, and for classes within the phylum Mollusca and the subphylum Crustacea which we considered collectively). Thus, each paper was classified as focusing on one or more class-like groups, as in Holyoak et al. (2008): Fish, Mammals, Birds, Reptiles, Amphibians, Insects, Crustaceans, Mollusks, and others. For the purpose of our analysis, we kept humans as a separate category and did not count them within Class Mammalia. The quality control procedure consisted in selecting a random sample of 100 abstracts and verifying that the common taxonomical group was correctly identified. The accuracy was 93%. The code for taxonomical identification can be downloaded here. 3.2.1 Outputs Fig. 3.10. Number of species studied in each year for the five classes with most studied species. More species have been studied in the last years. The code for plot can be downloaded here. Fig. 3.11. Proportion of papers in each year studying each of the five most commonly studied taxonomic groups. A paper can study several taxonomic groups, hence the proportions for each year can sum up to more than one. The code for plot can be downloaded here. Fig. 3.12. For each topic, relative frequencies of papers studying each taxonomical group. Only papers with more than 50% of association to each topic are used for this graph. The code for plot can be downloaded here. 3.3 Movement ecology framework (MEF) A unifying conceptual framework for movement ecology was proposed in Nathan et al. (2008). It consisted of four components: external factors (i.e. the set of environmental factors that affect movement), internal state (i.e. the inner state affecting motivation and readiness to move), navigation capacity (i.e. the set of traits enabling the individual to orient), and motion capacity (i.e. the set of traits enabling the individual to execute movement). The outcome of the interactions between these four components would be the observed movement path (plus observation errors). To assess the study of the different components of the movement ecology framework, we built what we call here a “dictionary.” A dictionary is composed of concepts and associated words. Here, the concepts of interest were the components of the framework (i.e. internal state, external factor, motion and navigation), and their associated words were the terms potentially used in the abstracts to refer to the study of each component. For example, terms like “memory,” “sensory information,” “path integration” or “orientation” were used to identify the study of navigation. The framework dictionary is downloadable here. To assessed how well the dictionary identified the components in the papers, a quality control procedure was established. For each aspect, a random sample of 100 papers was selected, and a coauthor who did not lead the construction of the dictionary was randomly selected to check if in those papers the categories of the dictionary were correctly identified (i.e. accuracy). The accuracy was 91%. We replicated this analysis for the 1999-2008 period for comparison purposes. 3.3.1 Outputs Component 2009-2018 1999-2008 External factors 77.3% 76.7% Internal state 49.0% 45.7% Motion capacity 26.2% 27.3% Navigation capacity 9.0% 11.7% Table 3.4. Framework components. The values are the percentages of abstracts (where information on the framework was gathered) that use terms related to each component. Fig. 3.13. Representation of the components of the movement ecology framework and how much they were studied in the last decade: external factors, internal state, motion and navigation capacities, whose interactions result in the observed movement path. The size of each component box is proportional to the percentage of papers (in parentheses) tackling them irrespectful of whether they are only about this component or in combination with another one. The latter is specified through the segments that join the components to the observed movement path. One fill color corresponds to papers that only studied one component, while two or more colors correspond to papers that tackled two or three components, respectively (the ones from those colors). The width of the segment is proportional to the percentage of papers that studied that combination (or single component). Only combinations corresponding to \\(>5\\%\\) of papers are shown; e.g. combinations involving navigation and papers studying navigation on its own had \\(<5\\%\\) of papers each therefore they are not shown in the graph. This graph was made using Inkscape. The percentages in it and an alternative graphical representation of them can be found in this downloadable code. Fig. 3.14. Representation of the components of the movement ecology framework and how much they were studied between 1999 and 2008: external factors, internal state, motion and navigation capacities, whose interactions result in the observed movement path. The size of each component box is proportional to the percentage of papers (in parentheses) tackling them irrespectful of whether they are only about this component or in combination with another one. The latter is specified through the segments that join the components to the observed movement path. One fill color corresponds to papers that only studied one component, while two or more colors correspond to papers that tackled two or three components, respectively (the ones from those colors). The width of the segment is proportional to the percentage of papers that studied that combination (or single component). Only combinations corresponding to \\(>5\\%\\) of papers are shown; e.g. combinations involving navigation and papers studying navigation on its own had \\(<5\\%\\) of papers each therefore they are not shown in the graph. This graph was made using Inkscape. The percentages in it and an alternative graphical representation of them can be found in this downloadable code. Fig. 3.15. Proportion of papers in each year focusing on each component of the MEF. A study can focus on more than one component, hence the proportions for each year can sum up to more than one. The code for plot can be downloaded here. Fig. 3.16. Proportion of papers studying each taxonomic class for each component of the MEF. Papers associated to several components were accounted for in several frames. The code for plot can be downloaded here. 3.4 Tracking devices We grouped tracking devices in 11 categories. These categories were meant to be as monophyletic as possible, and so broad categories had to be defined. Also, use of one technology does not rule out the use of another technology. E.g. radio + GPS is frequently used, and marine studies frequently have an array of sensors on them that may combine multiple technologies. These categories are: Light loggers: Any technology that records light levels and derives locations based on the timing of twilight events. Satellite: Any tag that collects location via, and sends data to, satellites, so that data can be accessed remotely. Frequently, the ARGOS system. Radio telemetry: Any technology that infers location based on radio telemetry (VHF/UHF frequency). Sometimes it is used in addition to either light logger or GPS technology, though this distinction cannot be readily inferred using our methodology. Camera: They include any device that records location/presence via photos or video; mainly camera traps with known locations where the capture of the individual implies the location. Video: They include any device that records movement via video. Acoustic: Any technology that uses sound to infer location, either in a similar way to radio telemetry or in an acoustic array where animal vocalizations are recorded and the location of sensors in the array are used to obtain an animal location. Pressure: Any technology that records pressure readings (frequently in the water), and these changes infer vertical movement, such as through a water column. Accelerometer: Any technology that is placed on a subject and measures the acceleration of the tag, and this delta infers movement. Body conditions: Any technology that uses body condition sensors to collect data on the subject that may be associated with a movement or lack thereof, such as temperature and heart rate. GPS: Any technology that uses Global Position System satellites to calculate the location of an object. Can be handheld GPS devices, or animal-borne tags. Radar: Any technology that uses “radio detection and ranging” devices to track objects. Can be large weather arrays or tracking radars. Encounter: Any analog tracking method where the user must capture the subject and place a marker on the subject. The recapturing/resighting of the subject infers the movement. This category was difficult to capture in our study due to a lack of specific phrases that can be used. Their use was assessed with a dictionary approach. The dictionary can be downloaded here. To assess how well the dictionary identified the types of devices in the papers, a quality control procedure was established. For each aspect, a random sample of 100 papers was selected, and a coauthor who did not lead the construction of the dictionary was randomly selected to check if in those papers the categories of the dictionary were correctly identified (i.e. accuracy). The accuracy was 84%. To assess if we were identifying animal-borne GPS rather than hand-held GPS, we took a sample of 100 papers already associated to GPS according to our dictionary. 83 referred to GPS tags, 12 were hand held, and 5 were neither tag nor hand-held (e.g. GPS coordinates in general, or General Practitioners). 3.4.1 Output Fig. 3.17. Proportion of papers in each year using the five most commonly used tracking devices. A study can use more than one device, hence the proportions for each year can sum up to more than one. The code for plot can be downloaded here. 3.5 Software Here we also used a dictionary approach. We used expert opinion to compile all known software used in movement ecology. The 33 software in our list were 1) R, 2) Python, 3) SPSS, 4) Matlab, 5) SAS, 6) MARK (program Mark and not R package unmarked), 7) Java, 8) C (if researchers wrote C code themselves; i.e some tracking Radars for pre-processing), 9) Fortran, 10) WinBUGS, 11) Agent-Analyst, 12) BASTrack, 13) QGIS, 14) GRASS, 15) Microsoft Excel, 16) Noldus observer, 17) fragstats, 18) postgis (we separated postGIS from the database category because its high spatial analytical capabilities), 19) databases (any relational database, likely for data management and summarizing necessarily for analyitical use), 20) e-surge, 21) m-surge, 22) u-care, 23) Genstat, 24) Biotas, 25) Statview, 26) Primer-e, 27) PAST, 28) STATA, 29) Statistica, 30) UCINET, 31) Mathcad, 32) Vicon, and 33) GME (geospatial modeling environment). The dictionary with the terms used can be downloaded here. For quality control, we examined a random sample of 50 papers. The accuracy was 88%. 3.5.1 Output Fig. 3.18. Proportion of papers in each year using the five most commonly used software. A study can use more than one software, hence the proportions for each year can sum up to more than one. The code for plot can be downloaded here. 3.6 Statistical methods Within our dictionary approach, we first used expert opinion to compile all known statistical methods (and terms used for them) that could have been used in movement ecology, resulting in 188 terms; click here to download the corresponding file. We classified statistical methods into Spatial, Time-series, Movement, Spatiotemporal, Social, and Generic. Their definitions are below: Spatial: spatial statistical methods (e.g. geostatistics) Time-series: time series methods (e.g. functional data analysis) Movement: statistical method used for the study of movement (e.g. behavioral change point analysis) Spatiotemporal: spatiotemporal but not movement methods (e.g. spatiotemporal geostatistics) Social: statistical methods that are not exclusively for movement, but that characterize or model social processes (e.g. social networks) Generic: generic statistical methods that could be used in any type of study, that are not inherently spatial, temporal or social (e.g. a regression analysis) Terms related to hypothesis tests were first considered but ultimately removed; we considered that the tendency of papers to present p-values could be biasing researchers towards the use of hypothesis tests, thus creating a bias towards general methods. For quality control, we examined a random sample of 50 papers. The accuracy was 84%. 3.6.1 Outputs Fig. 3.19. Proportion of papers in each year mentioning each type of statistical method. A study can use more than one type of method, hence the proportions for each year can sum up to more than one. The code for the plot can be downloaded here. generic Movement spatial time-series social spatiotemporal 67.8% 32.6% 18.9% 16.5% 3.3% 0.3% Table 3.5. Percentage of papers using each type of statistical method. The code for this table can be downloaded here. trigram n linear mixed models 231 linear mixed effects 229 generalized linear mixed 202 mixed effects models 202 linear mixed model 188 markov chain monte 180 chain monte carlo 178 akaike’s information criterion 174 akaike information criterion 162 minimum convex polygon 158 information criterion aic 146 monte carlo mcmc 133 correlated random walk 129 mixed effects model 117 hidden markov model 116 Table 3.6. Most common statistical trigrams in M&M sections of papers (with more than 100 mentions in papers). To reconstruct this table, readers can refer to this downloadable code. *The code to process the dictionaries of the framework, tracking devices, software, statistical methods, and taxonomy is downloadable here. References "],["survey-about-movement-ecology.html", "4 Survey about movement ecology 4.1 Description of the survey 4.2 Participation in the survey 4.3 Movement ecology framework 4.4 Taxa 4.5 Tracking devices 4.6 Software 4.7 Methods 4.8 Big changes in the field 4.9 Summary", " 4 Survey about movement ecology 4.1 Description of the survey As a complementary source of information for our review, we elaborated a survey to get the perspectives of movement ecologists about the field and how it is evolving. The exact formulation of the questions in the survey can be downloaded here. 4.2 Participation in the survey The survey was designed to be completely anonymous. There was no previous selection of the participants and no probabilistic sampling was involved. The survey was advertised by mailing lists (r-sig-geo and r-sig-ecology), individual emails to researchers, the lab’s website https://mablab.org, the Gordon Research Conference and Seminar on Movement Ecology of Animals, and Twitter. It took place during the Winter/Spring of 2019. The survey got exemption from the Institutional Review Board at University of Florida (IRB02 Office, Box 112250, University of Florida, Gainesville, FL 32611-2250). A total of 82 people participated in the survey, and 61 answered at least one question concerning the field (and not just their years of experience in research or the field). No question was mandatory, so participants could opt to not answer some questions. We first asked the participants how many years they have been doing research, and how many years they have been working on animal movement. They have spent a median of 10 years in research, and a median of 6 years in animal movement. 4.3 Movement ecology framework In a 2008 article, “A movement ecology paradigm for unifying organismal movement research,” Nathan et al. defined a movement ecology framework where the movement propagation process is produced by the motion and the navigation processes, with internal and external factors affecting movement. We asked the participants: Would you say that most research articles in movement ecology analyze these components of the movement ecology framework? Would you say that the these components of the movement ecology framework are currently being more, less or equally studied compared to 10 years ago? The results are shown in the graphs below. Most participants perceived that external factors are being addressed in most movement ecology papers, that motion is being addressed in at least half of the literature, and that navigation and external factors are addressed in less than half. Interestingly, when asked which components were most studied now than 10 years ago, most participants agreed on internal factors. 4.4 Taxa We asked the participants which taxa they considered to be studied the most in movement ecology, and to select up to 3 taxa. The number of votes per taxon are shown in the graph below. Birds and then mammals were indicated as the most studied. 4.5 Tracking devices We asked the participants the following questions: Which tracking device do you consider to be used the most in movement ecology? (up to 3) Which tracking devices do you think are used more often now compared to 10 years ago? (up to 3) Which tracking devices do you think are used less often now compared to 10 years ago? (up to 3) The results are shown in the graphs above. GPS and satellite tags (e.g. PSAT, PTT) are the most used devices, according to the participants. When compared to 10 years ago, they expressed that accelerometers and GPS devices are more used than before, while radio tags and encounter techniques (e.g. capture mark recapture, banding, direct observation) are less used. 4.6 Software We asked the participants the following questions: Which software do you think is used the most for movement analysis? (up to 3) For movement analysis, which software do you think are used more often now compared to 10 years ago? (up to 3) For movement analysis, which software do you think are used less often now compared to 10 years ago? (up to 3) The results are shown in the graphs above. R, and in the a lesser degree, ArcGIS, are the most used software, according to the participants. When compared to 10 years ago, they expressed that R is more used than before, while Matlab, ArcGIS, SPSS and SAS are less used. 4.7 Methods We asked the participants the following questions: Which statistical/mathematical methods do you consider to be used the most for movement analysis? Which methods do you think are used more often now compared to 10 years ago? (up to 3) Which methods do you think are used less often now compared to 10 years ago? (up to 3) There are many analytical tools applied/developed in movement ecology. To keep it simple, we provided the participants with 9 (arbitrarily selected) options: generalized linear models (GLMs) and generalized additive models (GAMs), machine learning, model selection criteria, multivariate exploratory methods, net squared displacement (NSD), spatial point processes, state-space and Hidden Markov models (SSMs and HMMs, respectively), step and resource selection functions (SSFs and RSFs, respectively), and test statistics and p-values. We provided an “Other” option for researchers to indicate methods that would not be within the 9 other options. In the end, the “Other” option was poorly used. The results are shown in the graphs above. The most used methods in movement ecology, according to the participants, are SSFs and RSFs, SSMs and HMMs, GLMs and GAMs. When compared to 10 years ago, they expressed that SSMs and HMMs, as well as machine learning, and in a lesser degree, SSFs and RSFs are more used, while hypothesis tests are less used. The “Other” option was used for speed threshold (1) and mininum convex polygon and kernel density utilization (1), for the first question; visual analytics (1), and continuous time movement modeling (1) for the second question; descriptive metrics (1) and “none” (1) for the third. 4.8 Big changes in the field We asked the participants three final open questions. In your opinion, what has revolutionized the field in the last 10 years? (Please keep it to three topics) 33 participants answered this first question. 31 answers were related to tagging devices (smaller, long-lasting, cheaper, battery-saving devices) that gave rise to longer or higher resolution data on many species. 15 answers were related to software development and computational power, 6 to statistical methods, 2 to the availability of remote sensing data, 2 to availability of tracking data online (in open data portals like movebank), 1 to molecular markers, 1 to citizen science, and there were other references to “mechanistic thinking and approaches” (1), the increase in the number of studies (1) and Nathan et al. 2008 paper in PNAS (1). Compared to 10 years ago, what would you be able to work on now that you could not do 10 years ago? (Please keep it to three topics; your answer may involve taxa, devices, methods or others) 29 participants answered this question. 28 answers were related to tagging devices that allowed working with smaller species, obtaining longer datasets, tracking more individuals and having high resolution data on their movement; among them, 4 mentioned the opportunity to work with accelerometry data. 5 answers were related to working with environmental data to link to the movement data (one mentioned it to do habitat selection analysis), 4 people mentioned studies in physiology, 3 of behavior, 3 of movement models and analysis in general, 1 studies on the mechanisms underlying movement, 1 multitaxa comparison, 1 citizen science thanks to apps on the phone. 2 people mentioned computational simulation and 2 processing data. In your opinion, what will revolutionize the field in the next 10 years? (Please keep it to three topics) 32 participants answered this question. 21 gave answers related to tagging devices (including use of ICARUS technology, 3D monitoring, development of more reliable and less invasive devices, improvement of accelerometers, and combined use of devices). 5 people referred to the development of statistical and mathematical methods to study movement processes (including machine learning techniques); 3 mentioned the availability of high resolution remote sensing data; 2 answers were related to software and computing power, 2 to connecting movement ecology to physiology, 2 to evolution, 1 to ontogeny, 1 to large-scale navigation experiments, 1 to integrating navigation, genetics, environment, physiology, life history and population dynamic studies; 1 to the development of data platforms, and 1 to “improved conceptualization of the field.” 4.9 Summary Most participants in the survey had less than a decade of experience in the field, so their farthest point of comparison in the past is from less than 10 years. Participants identified external factors as the dominant component in movement ecology papers, though recognizing that there is an increasing but still weaker interest in the link between internal factors and movement. Birds and mammals were identified as the most studied taxa in the field. GPS tags were identified as the most used tags, while the use of accelerometers is increasing and the use of radio tags, decreasing. The R software is the most used software according to the participants, and it is far more used than 10 years ago. On the other hand, Matlab, ArcGIS, SPSS and SAS use have decreased in the last decade. RSFs, SSFs, SSMs, HMMs, as well as GLMs and GAMs, are the most used statistical methods in movement ecology. SSM, HMM and machine learning techniques have experienced a notable increase in their use in the last decade, while the use of statistical tests, mainly, have decreased. When asked for what has revolutionized the field and what will play a major role in the next 10 years, most of the answers were related to tracking devices (miniaturization, reliability, battery saving, combined use of them). In a lesser degree, software, data and methods to analyze movement were mentioned. Most participants identified tracking technology as the game changer in the field. "],["r-session-information.html", "5 R session information", " 5 R session information devtools::session_info() ─ Session info ──────────────────────────────────────────── setting value version R version 3.6.3 (2020-02-29) os Ubuntu 18.04.4 LTS system x86_64, linux-gnu ui RStudio language (EN) collate en_US.UTF-8 ctype en_US.UTF-8 tz America/Nassau date 2020-05-22 ─ Packages ─────────────────────────────────────────────── package * version date lib source aRxiv 0.5.19 2019-08-08 [1] CRAN (R 3.6.1) assertthat 0.2.1 2019-03-21 [1] CRAN (R 3.6.1) audio 0.1-6 2019-03-19 [1] CRAN (R 3.6.1) backports 1.1.5 2019-10-02 [1] CRAN (R 3.6.1) beepr * 1.3 2018-06-04 [1] CRAN (R 3.6.1) bibtex 0.4.2 2017-06-30 [1] CRAN (R 3.6.1) bit 1.1-14 2018-05-29 [1] CRAN (R 3.6.1) bit64 0.9-7 2017-05-08 [1] CRAN (R 3.6.1) bitops 1.0-6 2013-08-17 [1] CRAN (R 3.6.1) blob 1.2.0 2019-07-09 [1] CRAN (R 3.6.1) bookdown 0.16 2019-11-22 [1] CRAN (R 3.6.1) broom 0.5.2 2019-04-07 [1] CRAN (R 3.6.1) callr 3.3.2 2019-09-22 [1] CRAN (R 3.6.1) caTools 1.17.1.2 2019-03-06 [1] CRAN (R 3.6.1) cellranger 1.1.0 2016-07-27 [1] CRAN (R 3.6.1) cli 2.0.2 2020-02-28 [1] CRAN (R 3.6.3) clipr 0.7.0 2019-07-23 [1] CRAN (R 3.6.1) cluster 2.1.0 2019-06-19 [1] CRAN (R 3.6.3) codetools 0.2-16 2018-12-24 [4] CRAN (R 3.6.3) colorspace 1.4-1 2019-03-18 [1] CRAN (R 3.6.1) cowplot * 1.0.0 2019-07-11 [1] CRAN (R 3.6.1) crayon 1.3.4 2017-09-16 [1] CRAN (R 3.6.1) crul 0.9.0 2019-11-06 [1] CRAN (R 3.6.1) curl 4.3 2019-12-02 [1] CRAN (R 3.6.3) data.table 1.12.6 2019-10-18 [1] CRAN (R 3.6.1) DBI 1.1.0 2019-12-15 [1] CRAN (R 3.6.1) dbplyr * 1.4.2 2019-06-17 [1] CRAN (R 3.6.1) dendextend 1.12.0 2019-05-11 [1] CRAN (R 3.6.1) desc 1.2.0 2018-05-01 [1] CRAN (R 3.6.1) details * 0.2.1 2020-01-12 [1] CRAN (R 3.6.3) devtools 2.2.1 2019-09-24 [1] CRAN (R 3.6.1) digest 0.6.25 2020-02-23 [1] CRAN (R 3.6.3) dotCall64 1.0-0 2018-07-30 [1] CRAN (R 3.6.1) dplyr * 0.8.3 2019-07-04 [1] CRAN (R 3.6.1) DT 0.10 2019-11-12 [1] CRAN (R 3.6.1) ellipsis 0.3.0 2019-09-20 [1] CRAN (R 3.6.1) evaluate 0.14 2019-05-28 [1] CRAN (R 3.6.1) fansi 0.4.1 2020-01-08 [1] CRAN (R 3.6.3) farver 2.0.3 2020-01-16 [1] CRAN (R 3.6.3) fastmap 1.0.1 2019-10-08 [1] CRAN (R 3.6.1) fields 10.0 2019-11-11 [1] CRAN (R 3.6.1) forcats * 0.4.0 2019-02-17 [1] CRAN (R 3.6.1) foreach 1.4.7 2019-07-27 [1] CRAN (R 3.6.1) foreign 0.8-76 2020-03-03 [4] CRAN (R 3.6.3) fs 1.3.1 2019-05-06 [1] CRAN (R 3.6.1) fulltext * 1.3.0 2019-06-27 [1] CRAN (R 3.6.1) gclus 1.3.2 2019-01-07 [1] CRAN (R 3.6.1) gdata 2.18.0 2017-06-06 [1] CRAN (R 3.6.1) generator * 0.1.0 2015-08-26 [1] CRAN (R 3.6.1) generics 0.0.2 2018-11-29 [1] CRAN (R 3.6.1) ggmap 3.0.0 2019-02-05 [1] CRAN (R 3.6.1) ggplot2 * 3.3.0 2020-03-05 [1] CRAN (R 3.6.3) ggwordcloud * 0.5.0 2019-06-02 [1] CRAN (R 3.6.1) glue 1.4.0 2020-04-03 [1] CRAN (R 3.6.3) gplots 3.0.1.1 2019-01-27 [1] CRAN (R 3.6.1) gridExtra * 2.3 2017-09-09 [1] CRAN (R 3.6.1) gtable 0.3.0 2019-03-25 [1] CRAN (R 3.6.1) gtools 3.8.1 2018-06-26 [1] CRAN (R 3.6.1) haven 2.2.0 2019-11-08 [1] CRAN (R 3.6.1) heatmaply * 0.16.0 2019-05-11 [1] CRAN (R 3.6.1) hms 0.5.2 2019-10-30 [1] CRAN (R 3.6.1) hoardr 0.5.2 2018-12-02 [1] CRAN (R 3.6.1) htmltools 0.4.0 2019-10-04 [1] CRAN (R 3.6.1) htmlwidgets 1.5.1 2019-10-08 [1] CRAN (R 3.6.1) httpcode 0.2.0 2016-11-14 [1] CRAN (R 3.6.1) httpuv 1.5.2 2019-09-11 [1] CRAN (R 3.6.1) httr 1.4.1 2019-08-05 [1] CRAN (R 3.6.1) iterators 1.0.12 2019-07-26 [1] CRAN (R 3.6.1) janeaustenr 0.1.5 2017-06-10 [1] CRAN (R 3.6.1) jpeg 0.1-8.1 2019-10-24 [1] CRAN (R 3.6.1) jsonlite 1.6.1 2020-02-02 [1] CRAN (R 3.6.3) KernSmooth 2.23-17 2020-04-26 [4] CRAN (R 3.6.3) knitr 1.28 2020-02-06 [1] CRAN (R 3.6.3) koRpus * 0.11-5 2018-10-28 [1] CRAN (R 3.6.1) koRpus.lang.en * 0.1-3 2019-05-13 [1] CRAN (R 3.6.1) labeling 0.3 2014-08-23 [1] CRAN (R 3.6.1) later 1.0.0 2019-10-04 [1] CRAN (R 3.6.1) lattice 0.20-41 2020-04-02 [4] CRAN (R 3.6.3) lazyeval 0.2.2 2019-03-15 [1] CRAN (R 3.6.1) lifecycle 0.2.0 2020-03-06 [1] CRAN (R 3.6.3) lubridate 1.7.4 2018-04-11 [1] CRAN (R 3.6.1) magrittr * 1.5 2014-11-22 [1] CRAN (R 3.6.1) maps 3.3.0 2018-04-03 [1] CRAN (R 3.6.1) maptools 0.9-8 2019-10-05 [1] CRAN (R 3.6.1) MASS 7.3-51.6 2020-04-26 [4] CRAN (R 3.6.3) Matrix 1.2-18 2019-11-27 [4] CRAN (R 3.6.1) memoise 1.1.0 2017-04-21 [1] CRAN (R 3.6.1) microdemic 0.4.0 2018-10-25 [1] CRAN (R 3.6.1) mime 0.9 2020-02-04 [1] CRAN (R 3.6.3) miniUI 0.1.1.1 2018-05-18 [1] CRAN (R 3.6.1) modelr 0.1.5 2019-08-08 [1] CRAN (R 3.6.1) modeltools 0.2-22 2018-07-16 [1] CRAN (R 3.6.1) munsell 0.5.0 2018-06-12 [1] CRAN (R 3.6.1) network 1.15 2019-04-02 [1] CRAN (R 3.6.1) nlme 3.1-147 2020-04-13 [4] CRAN (R 3.6.3) NLP * 0.2-0 2018-10-18 [1] CRAN (R 3.6.1) pillar 1.4.2 2019-06-29 [1] CRAN (R 3.6.1) pkgbuild 1.0.6 2019-10-09 [1] CRAN (R 3.6.1) pkgconfig 2.0.3 2019-09-22 [1] CRAN (R 3.6.1) pkgload 1.0.2 2018-10-29 [1] CRAN (R 3.6.1) plotly * 4.9.1 2019-11-07 [1] CRAN (R 3.6.1) plyr * 1.8.4 2016-06-08 [1] CRAN (R 3.6.1) png 0.1-7 2013-12-03 [1] CRAN (R 3.6.1) prettyunits 1.0.2 2015-07-13 [1] CRAN (R 3.6.1) processx 3.4.1 2019-07-18 [1] CRAN (R 3.6.1) promises 1.1.0 2019-10-04 [1] CRAN (R 3.6.1) ps 1.3.0 2018-12-21 [1] CRAN (R 3.6.1) purrr * 0.3.3 2019-10-18 [1] CRAN (R 3.6.1) R6 2.4.1 2019-11-12 [1] CRAN (R 3.6.1) rappdirs 0.3.1 2016-03-28 [1] CRAN (R 3.6.1) RColorBrewer * 1.1-2 2014-12-07 [1] CRAN (R 3.6.1) Rcpp 1.0.4.6 2020-04-09 [1] CRAN (R 3.6.3) rcrossref 0.9.2 2019-05-04 [1] CRAN (R 3.6.1) readr * 1.3.1 2018-12-21 [1] CRAN (R 3.6.1) readxl 1.3.1 2019-03-13 [1] CRAN (R 3.6.1) refsplitr * 1.0 2020-05-22 [1] Github (ropensci/refsplitr@ad6e199) registry 0.5-1 2019-03-05 [1] CRAN (R 3.6.1) remotes 2.1.0 2019-06-24 [1] CRAN (R 3.6.1) rentrez 1.2.2 2019-05-02 [1] CRAN (R 3.6.1) reprex 0.3.0 2019-05-16 [1] CRAN (R 3.6.1) reshape2 1.4.3 2017-12-11 [1] CRAN (R 3.6.1) RgoogleMaps 1.4.4 2019-08-20 [1] CRAN (R 3.6.1) rjson 0.2.20 2018-06-08 [1] CRAN (R 3.6.1) rlang 0.4.5 2020-03-01 [1] CRAN (R 3.6.3) rmarkdown 2.1 2020-01-20 [1] CRAN (R 3.6.3) rplos 0.8.6 2019-05-03 [1] CRAN (R 3.6.1) rprojroot 1.3-2 2018-01-03 [1] CRAN (R 3.6.1) RSQLite * 2.2.0 2020-01-07 [1] CRAN (R 3.6.1) rstudioapi 0.11 2020-02-07 [1] CRAN (R 3.6.3) rvest 0.3.5 2019-11-08 [1] CRAN (R 3.6.1) rworldmap 1.3-6 2016-02-03 [1] CRAN (R 3.6.1) scales 1.1.0 2019-11-18 [1] CRAN (R 3.6.1) seriation 1.2-8 2019-08-27 [1] CRAN (R 3.6.1) sessioninfo 1.1.1 2018-11-05 [1] CRAN (R 3.6.1) shiny 1.4.0.2 2020-03-13 [1] CRAN (R 3.6.3) slam 0.1-46 2019-10-26 [1] CRAN (R 3.6.1) SnowballC 0.6.0 2019-01-15 [1] CRAN (R 3.6.1) solrium 1.1.4 2019-11-02 [1] CRAN (R 3.6.1) sp 1.3-2 2019-11-07 [1] CRAN (R 3.6.1) spam 2.4-0 2019-11-02 [1] CRAN (R 3.6.1) storr 1.2.1 2018-10-18 [1] CRAN (R 3.6.1) stringdist 0.9.5.5 2019-10-21 [1] CRAN (R 3.6.3) stringi 1.4.6 2020-02-17 [1] CRAN (R 3.6.3) stringr * 1.4.0 2019-02-10 [1] CRAN (R 3.6.1) sylly * 0.1-5 2018-07-29 [1] CRAN (R 3.6.1) sylly.en 0.1-3 2018-03-19 [1] CRAN (R 3.6.1) testthat 2.3.0 2019-11-05 [1] CRAN (R 3.6.1) textstem * 0.1.4 2018-04-09 [1] CRAN (R 3.6.1) tibble * 2.1.3 2019-06-06 [1] CRAN (R 3.6.1) tidyr * 1.0.0 2019-09-11 [1] CRAN (R 3.6.1) tidyselect 0.2.5 2018-10-11 [1] CRAN (R 3.6.1) tidytext * 0.2.2 2019-07-29 [1] CRAN (R 3.6.1) tidyverse * 1.3.0 2019-11-21 [1] CRAN (R 3.6.1) tm * 0.7-6 2018-12-21 [1] CRAN (R 3.6.1) tokenizers 0.2.1 2018-03-29 [1] CRAN (R 3.6.1) topicmodels * 0.2-8 2018-12-21 [1] CRAN (R 3.6.1) TSP 1.1-7 2019-05-22 [1] CRAN (R 3.6.1) usethis 1.5.1 2019-07-04 [1] CRAN (R 3.6.1) utf8 1.1.4 2018-05-24 [1] CRAN (R 3.6.1) vctrs 0.2.0 2019-07-05 [1] CRAN (R 3.6.1) viridis * 0.5.1 2018-03-29 [1] CRAN (R 3.6.1) viridisLite * 0.3.0 2018-02-01 [1] CRAN (R 3.6.1) webshot 0.5.2 2019-11-22 [1] CRAN (R 3.6.1) whisker 0.4 2019-08-28 [1] CRAN (R 3.6.1) withr 2.1.2 2018-03-15 [1] CRAN (R 3.6.1) xfun * 0.12 2020-01-13 [1] CRAN (R 3.6.3) XML 3.98-1.20 2019-06-06 [1] CRAN (R 3.6.1) xml2 * 1.3.1 2020-04-09 [1] CRAN (R 3.6.3) xtable 1.8-4 2019-04-21 [1] CRAN (R 3.6.1) yaml 2.2.1 2020-02-01 [1] CRAN (R 3.6.3) zeallot 0.1.0 2018-01-28 [1] CRAN (R 3.6.1) [1] /home/rjoo/R/x86_64-pc-linux-gnu-library/3.6 [2] /usr/local/lib/R/site-library [3] /usr/lib/R/site-library [4] /usr/lib/R/library "],["references.html", "References", " References "]] +[["index.html", "Recent trends in movement ecology of animals and human mobility 1 Introduction 1.1 Abstract of the manuscript", " Recent trends in movement ecology of animals and human mobility Rocío Joo, Simona Picardi, Matthew E. Boone, Thomas A. Clay, Samantha C. Patrick, Vilma S. Romero-Romero and Mathieu Basille 2022-05-10 1 Introduction This is the companion website for the manuscript “Recent trends in movement ecology of animals and human mobility,” a quantitative review of animal and human movement literature in 2009-2018 from Joo et al., serving as the manuscript’s supplementary information page. The R codes for the analyses are available in this GitHub repository. A version of this manuscript is available as an arXiv pre-print. 1.1 Abstract of the manuscript Movement is fundamental to life, shaping population dynamics, biodiversity patterns and ecosystem structure. In 2008, the Movement Ecology Framework (MEF; Nathan et al. 2008) introduced an integrative theory of organism movement—linking internal state, motion capacity, and navigation capacity to external factors—which has been recognized as a milestone in the field. Since then, the study of movement experienced a technological boom, which provided massive quantities of tracking data of both animal and human movement globally and at ever finer spatio-temporal resolutions. In this work, we provide a quantitative assessment of the state of research within the MEF, focusing on animal movement, including humans and invertebrates, and excluding movement of plants and microorganisms. Using a text mining approach, we digitally scanned the content of >8000 papers from 2009-2018 available online, identified tools and methods, and assessed all components of the MEF. Over the past decade, the publication rate has increased considerably, along with major technological changes, such as an increased use of GPS devices and accelerometers and a majority of studies now using the R software environment for statistical computing. However, animal movement research still largely focuses on the effect of environmental factors on movement, with much less focus on motion and navigation. We discuss the potential for technological and methodological advances in the field to lead to more integrated and interdisciplinary research and increased exploration of key movement processes like navigation and the evolutionary, physiological, and life-history consequences of movement. "],["data-collection-and-processing.html", "2 Data collection and processing 2.1 Identification of movement ecology (mov-eco) papers 2.2 Downloading whole manuscripts 2.3 Extracting the Material and Methods (M&M) sections", " 2 Data collection and processing 2.1 Identification of movement ecology (mov-eco) papers Fig. 2.1. Workflow to identify movement papers. 2.1.1 What is a movement ecology paper? We defined mov-eco papers as scientific peer-reviewed papers that studied the voluntary movement of one or more living individuals. This included humans. 2.1.2 Search keywords We used Web of Science (WoS) as a search engine for the papers. We defined mov-eco papers as scientific peer-reviewed papers that studied the voluntary movement of one or more living individuals. This included humans. Very few papers mention “movement ecology” in their abstracts, so we did not use “movement ecology” as a search phrase. After much testing, we came up with the following groups of words: Group 1 - Behavior: behavio Group 2 - Movement: movement, moving, motion, spatiotemporal, kinematics, spatio-temporal Group 3 - Biologging: telemetry, geolocat, biologg, accelerom, gps, geo-locat, bio-logg, reorient, vhf, argos, radar, sonar, gls, vms, animal-borne Group 4 - Individuals: animal, individual, human, person, people, player, wildlife, fishermen Paper abstracts had to have words from at least 3 of the groups above to be selected. An initial search revealed these keywords to pick up papers from a variety of fields such as biochemistry, medicine, physics and economics. As such, unless they had words from group 3, paper abstracts could not contain the following words: Group 5 - Missleading words: cell, DNA, enzyme, strain, neurons, atom, molecule, lymph, cortex, cortic, neurotransmi, patient prosthese, eye, particle, tectonic, counsel, cognit, market, spine, questionnaire, sendentary, insulin The search on WoS was made over a final selection of 273 journals; made in parallel with keyword tunning. 2.1.3 Cleaning and filtering results in R The grouping criteria were applied to the Topic field in WoS, which searched the Title, Abstract, and Keywords sections. We downloaded the search results from WoS, which contain information on title, keywords, abstracts and authors, among others. We downloaded all references in raw .txt format as it was the valid input for the refsplitr package available for R (R Core Team (2018)) on github (https://github.com/ropensci/refsplitr) (Fournier et al. (2019)). Refsplitr reads in multiple WoS files, parses addresses, and performs author matching. We used the references_read function to compile the .txt files into one data sheet. To be sure that the papers shown in our search in WoS were respecting our search criteria for the abstracts, we applied the same filters described above to the downloaded search results via R. In addition to the grouping criteria we filtered by Document Type to only allow ‘Articles,’‘Proceedings Papers,’ and ‘Reviews.’ This process yielded 8007 papers. The code can be downloaded here. 2.1.4 Quality control From the cleaned results, we took a random sample of 100 papers (i.e. with title, abstracts and other features taken from WoS). We then read the abstracts and classified them into “mov-eco” and “not mov-eco.” If the percentage of mov-eco papers (i.e. precision) was lower than 80%, the word criterion used for the search would be improved (e.g. adding more words, editing some, changing the rules for the groups). This is how we came up with the groups introduced above (section 2.1). We obtained 90% of precision. That means that, from the papers that we had, almost all of them were mov-eco papers. We also wanted to obtain a recall or sensitivity rate to quantify, from all mov-eco papers in the literature, how many we had in our search results. As it is impossible to obtain the “real” list of mov-eco papers in the literature, we instead looked at the list of papers published by the journal Movement Ecology. We found that 69% were in our list. An estimated sensitivity of 69% and precision of 90% implies that though we did not get the whole population of mov-eco papers in our set, there is a high certainty that those obtained are mov-eco papers. 2.1.5 Possible biases We have no reason to believe that our search criteria have introduced biases to our results. The relatively short list of words is due to the fact that other words we tried were actually reducing our precision, providing us many papers that were not about movement ecology. Of course, it is always possible that we forgot to try an important word. A possible bias could come from WoS: we were not able to get papers that were not in WoS, which depends on WoS agreements. 2.1.6 Differences with other approaches selecting and analyzing mov-eco papers Holyoak et al. (2008): Their goal was to find papers about movement of organisms or gametes, so their definition of movement ecology was somewhat broader than ours. We were inspired by their procedure, and tried the terms that they showed in the paper that would be consistent with our definition of mov-eco. Like us, they used WoS to build their literature dataset. They had a two-step criterion to select the papers. First, they screened the WoS for papers that contained their keywords. Then, they narrowed down the selection by excluding non-ecological journals from their initial results. Two of their coauthors decided on a list of 496 journals. We applied a similar procedure but using a modified set of keywords because we found their criteria to be too broad for our definition, and then two coauthors (R.J. and S.P.) decided on 273 journals. Among the remaining articles, they selected a random sample of 1000 papers for quality control, rating them as relevant or not. Their overall success rate (similar to our precision) was 77%. Fraser et al. (2018): They also used the WoS. In “Topics,” they searched for “ecology” and either “movement,” “migrat,” “home range,” “dispersal” or “track.” Their combination of words was too vague in our opinion, and they did not mention any quality control (e.g. precision, recall, specificity, sensitivity) statistic. 2.2 Downloading whole manuscripts Fig. 2.2. Downloading movement ecology papers. We used the fulltext package (Chamberlain (2019)) in R, using Elsevier, Springer, Scopus, Wiley, BMC and PLOS one API keys. We downloaded the articles we had access to, as xml or pdf documents. We downloaded a total of 4060 complete manuscripts, representing 51% of our list of mov-eco papers. The codes to download papers can be downloaded clicking on this link. 2.3 Extracting the Material and Methods (M&M) sections Fig. 2.3. Summary of procedure to extract Material and Methods section from each paper. For some analyses (section 3), we needed the Material and methods section of the manuscripts. We created codes for .xml (click to download) and .pdf (click to download) files. The former was built using the functions of the xml2 package (Wickham, Hester, and Ooms (2018)). The later calls the readPDF and Corpus functions from the tm (Feinerer and Hornik (2018)) package. In order to write the codes, we took account of the structure of the papers in either format, and aimed at finding section names related to “Methods,” “Data” or “Statistical Analysis.” Not all papers had an M&M section (e.g. reviews or perspective papers). We were able to extract 3674 M&M sections (46% of mov-eco paper results and 90% of fully downloaded papers). The possibility to download articles was conditioned by openness of data from the publishers and data & text mining agreements with the institutions we accessed them from. Fig. 2.4. Number of articles about movement ecology of animals and human mobility identified by our algorithm and per journal. In blue, the number of articles that were downloaded and a methods section was identified. In yellow, the number of articles that were not downloaded. Only the journals with more than 20 identified articles are shown in the graphs. References "],["analyses.html", "3 Data analysis 3.1 Topic analysis 3.2 Taxonomical identification 3.3 Movement ecology framework (MEF) 3.4 Tracking devices 3.5 Software 3.6 Statistical methods", " 3 Data analysis A total of 8007 results (papers) from 2009-2018 were obtained. The proportion of animal and human movement papers from the total number of scientific papers extracted from WoS was higher in the last years (Table and Figures below; download the code to reproduce them here). Year Movement articles All articles Proportion movement/all (\\(10^{-4}\\)) 2009 485 1139611 4.25 2010 479 1186928 4.03 2011 564 1262956 4.47 2012 666 1323677 5.03 2013 791 1398009 5.66 2014 878 1438134 6.11 2015 978 1709898 5.72 2016 937 1775745 5.28 2017 1073 1838351 5.84 2018 1156 1928507 5.99 Table 3.1. Number of articles in movement ecology of animals and human mobility, and articles in scientific literature in general, published from 2009 to 2018 according to the Web of Science. Proportion movement/all refers to the proportion of articles in that studied movement. Several dimensions of the mov-eco literature were analyzed: research topics, taxonomical groups studied, components of the movement ecology framework studied, tracking devices used, software tools used, and statistical methods applied. Depending on the dimension, we either analyzed the title, keywords, abstract or material and methods (M&M). The sections used for each aspect of the analysis are detailed in the following table. Dimension Title Keywords Abstract M&M Topics X Taxonomy X X X Framework X X X Devices X X X X Software X X X X Methods X X X X Table 3.2. Paper sections used to analyze each dimension. 3.1 Topic analysis Fig. 3.2. Stages of topic analysis. The topics were not defined a priori. Instead, we fitted Latent Dirichlet Allocation (LDA) models to the abstracts (Blei, Ng, and Jordan (2003)). 3.1.1 The model LDAs are Bayesian mixture models that assume the existence of a fixed number \\(K\\) of topics behind the abstracts. Each topic can be characterized by a multinomial distribution of words with parameter \\(\\beta\\), drawn from a Dirichlet distribution with parameter \\(\\gamma\\). Each document \\(d \\in {1, ..., D}\\) is composed by a mixture of topics, drawn from a multinomial distribution with parameter \\(\\theta\\), which is drawn from a Dirichlet distribution with parameter \\(\\alpha\\). For each word \\(w\\) in document \\(d\\), first a hidden topic \\(z\\) is selected from the multinomial distribution with parameter \\(\\theta\\). From the selected topic \\(z\\), a word is selected based on the multinomial distribution with parameter \\(\\beta\\). The log-likelihood of a document \\(d = \\{w_1,...,w_N\\}\\) is \\(l(\\alpha,\\beta) = \\log(p(d|\\alpha,\\beta)) = \\log\\int\\sum_z\\left[\\prod_{n=1}^{N} p(w_i|z_i,\\beta)p(z_i|\\theta)\\right]p(\\theta|\\alpha)d\\theta\\) Here we used the LDA model with variational EM estimation (Wainwright and Jordan (2008), Blei, Ng, and Jordan (2003)) implemented in the topicmodels package. All the details of the model specification and estimation are in Grün and Hornik (2011). The model assumes exchangeability (i.e. the order of words is negligible), that topics are uncorrelated, and that the number of topics is known. The most commonly used criterion to choose a number of topics is the perplexity score or likelihood of a test dataset (De Waal and Barnard (2008)). Basically, this quantity measures the degree of uncertainty a language model has when predicting some new text (for this study, a new abstract of a paper). Lower values of the perplexity is good and it means the model is assigning higher probabilities. However, the perplexity score measures predictive capacities, rather than having actual humanly-interpretable latent topics (Chang et al. (2009)). In fact, using this score could result in there being too many topics; see Griffiths and Steyvers (2004) who analyzed PNAS abstracts and obtained 300 topics. Hence, we decided to fix the number of topics to 15, as a reasonable value that would not be too large than we could not interpret them, or too small that the topics were too general. Fig. 3.3. Schematic representation of the links between words, documents and topics. Each document is a mixture of topics. Each topic is modeled as a distribution of words. Each word comes out of one of these topics. Source of the image: Blei, D.M. 2012. Probabilistic topic models. Communications of the ACM, 55(4), 77-84. Fig. 3.4. Schematic representation of the Latent Dirichlet model described above. 3.1.2 Preprocessing To improve the quality of our LDA model outputs, we cleaned the data by 1) removing redundant words for identifying topics (e.g. prepositions and numbers), 2) converting all British English words to American English so they would not be seen as different words, 3) lemmatizing (i.e. extracting the lemma of a word based on its intended meaning, with the aim of grouping words under the same lemma) (Ingason et al. (2008)), 4) filtering out words that were only used once in the whole set of abstracts. R packages tidytext (Silge and Robinson (2016)), tm and textstem (Rinker (2018)) were used in this stage (click to download). 3.1.3 Model fitting The parameter estimates of the LDA model were obtained by running 20 replicates of the models (with the VEM estimation method), and keeping the one with the highest likelihood; click here to download the code. 3.1.4 Model outputs From the fitted LDA model, we can obtained: \\(E(\\beta | z,w)\\), as the posterior expected values of word distribution per topic, denoted by \\(\\hat{\\beta}\\), and \\(E(\\theta\\_d | z)\\), the posterior topic distribution per abstract, denoted by \\(\\gamma\\) in the package. The \\(\\hat{\\beta}\\) values were thus a proxy of the importance of a word in a topic. They were used to interpret and label each topic, and to create wordclouds for each topic, where the area occupied by each word was proportional to its \\(\\hat{\\beta}\\) value. Fig. 3.5. Wordclouds of each topic based on \\(\\hat{\\beta}\\) values. Download the codes for the plot here. Since \\(\\gamma\\) indicated the degree of association between an abstract and a topic, we obtained a sample (click to download) of the 5 most associated abstracts to each topic, to aid the interpretation of the topics. Based on these outputs, the topics were interpreted as: 1) Dispersal, 2) Movement models, 3) Habitat selection, 4) Detection and data, 5) Home ranges, 6) Aquatic systems, 7) Foraging in marine megafauna, 8) Biomechanics, 9) Acoustic telemetry, 10) Experimental designs, 11) Activity budgets, 12) Migration, 13) Sports, 14) Human activity patterns, 15) Breeding ecology. For an extended description of these topics, see the main text of the manuscript. The sum of \\(\\gamma\\) values for each topic (\\(\\sum_d E(\\theta_d | z_k)\\) for each \\(k\\)) served as proxies of the “prevalence” of the topic relative to all other topics and were used to rank them. A heatmap of the \\(\\gamma\\) values also showed that most papers were evidently more associated to one topic and few were split into several topics. Fig. 3.6. Heatmap of \\(\\gamma\\) values per abstract and topic. The corresponding code can be downloaded clicking here. 3.1.5 Model assessment We assess the consistency and interpretation of the LDA results via internal and external expert judgement, respectively. 3.1.5.1 Consistency In the heatmap, we showed that most papers were more strongly associated to one topic. To check for consistency, for each topic, we compared its wordcloud with one obtained from abstracts that were highly associated with the topic (\\(\\gamma > 0.75\\)). The two wordclouds should be telling a very similar story, thus visually resemble, with very small differences due to the abstracts being composed –in a small proportion –by other topics as well. We selected the papers with \\(\\gamma > 0.75\\), and computed the number of times each unique word occurred in the abstracts related to the topic. We divided those values by the total number of words in the topic to get a relative frequency \\(\\eta\\). We then created wordclouds for each topic, where the area occupied by each word was proportional to its \\(\\eta\\) value. These wordclouds were overall consistent with the topic wordclouds, i.e. most words were the same, and the words that differ provided complementary information to interpret the topics and understand to which types of abstracts they were strongly associated to. Fig. 3.7. Wordclouds of each topic based on most strongly associated abstracts. The code for this plot can be downloaded clicking here. 3.1.5.2 Interpretation To assess the interpretability of the topics, we performed a word intrusion analysis, i.e. people are given the task to identify a word injected into the top‐terms of each topic. For each topic, we identified the 4 highest-probability words (click to download), i.e. with the highest \\(\\hat{\\beta}\\) values. We did not use a higher number of words in the word intrusion tests as we considered that it could have required more effort from the researchers who kindly and voluntarily participated in this process. Then, we took a high-probability word from another topic –that was not ‘movement’ or ‘behavior’ as they were highly related to most topics –and added it to the group; download file with intruder here. We asked 10 researchers in the field to identify the intruder in each group of words without any more explanations, and suggested them to answer fast so that they would not have to overthink; their answers are here. We then computed the number of correct answers for each topic. A high score for a topic would indicate that the topic was easy to interpret using the 4 highest associated words to it. The scores are shown in the table below. The codes written for this analysis are downloadable clicking here. Topic number Topic label Score (from 0 to 10) 1 Social interactions and dispersal 7 2 Movement models 10 3 Habitat selection 9 4 Detection and data 5 5 Home ranges 8 6 Aquatic systems 8 7 Foraging in marine megafauna 10 8 Biomechanics 0 9 Acoustic telemetry 1 10 Experimental designs 3 11 Activity budgets 2 12 Avian migration 9 13 Sports 0 14 Human activity patterns 5 15 Breeding ecology 4 Table 3.3. Word intrusion score for each topic. Topics 1, 2, 3, 5, 6, 7 and 12 got high scores (>7). A few of the low-scored topics (\\(\\leq 5\\)) had a word intrusion that could be considered relatively general for the researchers thus confusing, such as “tag” (topic 4) and “data” (topics 9 and 15). Moreover, the two topics highly associated with humans, 13 and 14, also had low scores that could be a consequence of the researchers (all ecologists) not expecting any human related group. Overall, this analysis shows that only half of the topics are easily interpretable using the 4 highest-probability words. Even for us, the researchers involved in this study, it was necessary to look at the abstracts with the highest association to each topic in order to be sure of the interpretation of some of the topics. An alternative approach to model assessment in topics is through topic prediction on an independent data set, but this should be performed when the goal of the study is to predict over new data sets, which is not the case here. The word intrusion approach is not an exhaustive assessment of topic interpretability, but it allows putting our results into perspective: some topics have a clear and easy interpretation and some others are really hard to interpret. 3.2 Taxonomical identification To identify the taxonomy of the organisms studied in the papers, the ITIS (Integrated Taxonomic Information System) database (USGS Core Science Analytics and Synthesis) was used to detect names of any animal species (kingdom Animalia) that were mentioned in the abstracts, titles and keywords. We screened these sections for latin and common (i.e., vernacular) names of species (both singular and plural), as well as common names of higher taxonomic levels such as orders and families. We excluded ambiguous terms that are used as common names for taxa but also have a current language meaning; for example: “Here,” “Scales,” “Costa,” “Ray,” etc. Because we wanted to consider humans as a separate category, we excluded “Homo sapiens” from the search terms, but used the following non-ambiguous terms to identify papers that focus on movement ecology of humans: "player", "players", "patient", "patients", "child", "children", "teenager", "teenagers", "people", "student", "students", "fishermen", "person", "tourist", "tourists", "visitor", "visitors", "hunter", "hunters", "customer", "customers", "runner", "runners", "participant", "participants", "cycler", "cyclers", "employee", "employees", "hiker", "hikers", "athlete", "athletes", "boy", "boys", "girl", "girls", "woman", "women", "man", "men", "adolescent", "adolescents". In cases where words may be suffixes of larger words, we used regular expression notation to exact match words, e.g ‘man’ must match only the word ‘man’ and not ‘manually.’ We excluded words that could have an ambiguous meaning: “passenger” may appear in papers that mention passenger pigeons; “driver” may be used to refer to a causing factor. After having identified any taxon mentioned in a paper, we summarized taxa at the Class level (except for superclasses Osteichthyes and Chondrichthyes which we merged into a single group labeled Fish, and for classes within the phylum Mollusca and the subphylum Crustacea which we considered collectively). Thus, each paper was classified as focusing on one or more class-like groups, as in Holyoak et al. (2008): Fish, Mammals, Birds, Reptiles, Amphibians, Insects, Crustaceans, Mollusks, and others. For the purpose of our analysis, we kept humans as a separate category and did not count them within Class Mammalia. The quality control procedure consisted in selecting a random sample of 100 abstracts and verifying that the common taxonomical group was correctly identified. The accuracy was 93%. The code for taxonomical identification can be downloaded here. 3.2.1 Outputs Fig. 3.8. Number of species studied in each year for the five classes with most studied species. More species have been studied in the last years. The code for plot can be downloaded here. Fig. 3.9. Proportion of papers in each year studying each of the five most commonly studied taxonomic groups. A paper can study several taxonomic groups, hence the proportions for each year can sum up to more than one. The code for plot can be downloaded here. Fig. 3.10. For each topic, relative frequencies of papers studying each taxonomical group. Only papers with more than 50% of association to each topic are used for this graph. The code for plot can be downloaded here. 3.3 Movement ecology framework (MEF) A unifying conceptual framework for movement ecology was proposed in Nathan et al. (2008). It consisted of four components: external factors (i.e. the set of environmental factors that affect movement), internal state (i.e. the inner state affecting motivation and readiness to move), navigation capacity (i.e. the set of traits enabling the individual to orient), and motion capacity (i.e. the set of traits enabling the individual to execute movement). The outcome of the interactions between these four components would be the observed movement path (plus observation errors). To assess the study of the different components of the movement ecology framework, we built what we call here a “dictionary.” A dictionary is composed of concepts and associated words. Here, the concepts of interest were the components of the framework (i.e. internal state, external factor, motion and navigation), and their associated words were the terms potentially used in the abstracts to refer to the study of each component. For example, terms like “memory,” “sensory information,” “path integration” or “orientation” were used to identify the study of navigation. The framework dictionary is downloadable here. To assessed how well the dictionary identified the components in the papers, a quality control procedure was established. For each aspect, a random sample of 100 papers was selected, and a coauthor who did not lead the construction of the dictionary was randomly selected to check if in those papers the categories of the dictionary were correctly identified (i.e. accuracy). The accuracy was 91%. We replicated this analysis for the 1999-2008 period for comparison purposes. 3.3.1 Outputs Component 2009-2018 1999-2008 External factors 77.3% 76.7% Internal state 49.0% 45.7% Motion capacity 26.2% 27.3% Navigation capacity 9.0% 11.7% Table 3.4. Framework components. The values are the percentages of abstracts (where information on the framework was gathered) that use terms related to each component. Fig. 3.11. Representation of the components of the movement ecology framework and how much they were studied in the last decade: external factors, internal state, motion and navigation capacities, whose interactions result in the observed movement path. The size of each component box is proportional to the percentage of papers (in parentheses) tackling them irrespectful of whether they are only about this component or in combination with another one. The latter is specified through the segments that join the components to the observed movement path. One fill color corresponds to papers that only studied one component, while two or more colors correspond to papers that tackled two or three components, respectively (the ones from those colors). The width of the segment is proportional to the percentage of papers that studied that combination (or single component). Only combinations corresponding to \\(>5\\%\\) of papers are shown; e.g. combinations involving navigation and papers studying navigation on its own had \\(<5\\%\\) of papers each therefore they are not shown in the graph. This graph was made using Inkscape. The percentages in it and an alternative graphical representation of them can be found in this downloadable code. External Internal Motion Navigation 2009-2018 Count 2009-2018 Percentage 1999-2008 Count 1999-2008 Percentage X - - - 2371 33.3 418 33.8 X X - - 1768 24.8 287 23.2 - X - - 663 9.3 96 7.8 X - X - 485 6.8 104 8.4 X X X - 424 6.0 69 5.6 - - X - 383 5.4 56 4.5 - X X - 373 5.2 63 5.1 X - - X 176 2.5 44 3.6 X X - X 136 1.9 15 1.2 - - - X 83 1.2 23 1.9 X - X X 73 1.0 12 1.0 - X - X 50 0.7 16 1.3 - - X X 49 0.7 15 1.2 X X X X 45 0.6 11 0.9 - X X X 41 0.6 8 0.6 Table 3.5. Number and percentage of articles in movement ecology of animals and human mobility studying each combination of components of the Movement Ecology Framework for the decades 2009-2018 and 1999-2008. In each row, an X in the column indicates the studied component. Fig. 3.12. Representation of the components of the movement ecology framework and how much they were studied between 1999 and 2008: external factors, internal state, motion and navigation capacities, whose interactions result in the observed movement path. The size of each component box is proportional to the percentage of papers (in parentheses) tackling them irrespectful of whether they are only about this component or in combination with another one. The latter is specified through the segments that join the components to the observed movement path. One fill color corresponds to papers that only studied one component, while two or more colors correspond to papers that tackled two or three components, respectively (the ones from those colors). The width of the segment is proportional to the percentage of papers that studied that combination (or single component). Only combinations corresponding to \\(>5\\%\\) of papers are shown; e.g. combinations involving navigation and papers studying navigation on its own had \\(<5\\%\\) of papers each therefore they are not shown in the graph. This graph was made using Inkscape. The percentages in it and an alternative graphical representation of them can be found in this downloadable code. Fig. 3.13. Proportion of papers in each year focusing on each component of the MEF. A study can focus on more than one component, hence the proportions for each year can sum up to more than one. The code for plot can be downloaded here. Fig. 3.14. Proportion of papers studying each taxonomic class for each component of the MEF. Papers associated to several components were accounted for in several frames. The code for plot can be downloaded here. 3.4 Tracking devices We grouped tracking devices in 11 categories. These categories were meant to be as monophyletic as possible, and so broad categories had to be defined. Also, use of one technology does not rule out the use of another technology. E.g. radio + GPS is frequently used, and marine studies frequently have an array of sensors on them that may combine multiple technologies. These categories are: Light loggers: Any technology that records light levels and derives locations based on the timing of twilight events. Satellite: Any tag that collects location via, and sends data to, satellites, so that data can be accessed remotely. Frequently, the ARGOS system. Radio telemetry: Any technology that infers location based on radio telemetry (VHF/UHF frequency). Sometimes it is used in addition to either light logger or GPS technology, though this distinction cannot be readily inferred using our methodology. Camera: They include any device that records location/presence via photos or video; mainly camera traps with known locations where the capture of the individual implies the location. Video: They include any device that records movement via video. Acoustic: Any technology that uses sound to infer location, either in a similar way to radio telemetry or in an acoustic array where animal vocalizations are recorded and the location of sensors in the array are used to obtain an animal location. Pressure: Any technology that records pressure readings (frequently in the water), and these changes infer vertical movement, such as through a water column. Accelerometer: Any technology that is placed on a subject and measures the acceleration of the tag, and this delta infers movement. Body conditions: Any technology that uses body condition sensors to collect data on the subject that may be associated with a movement or lack thereof, such as temperature and heart rate. GPS: Any technology that uses Global Position System satellites to calculate the location of an object. Can be handheld GPS devices, or animal-borne tags. Radar: Any technology that uses “radio detection and ranging” devices to track objects. Can be large weather arrays or tracking radars. Encounter: Any analog tracking method where the user must capture the subject and place a marker on the subject. The recapturing/resighting of the subject infers the movement. This category was difficult to capture in our study due to a lack of specific phrases that can be used. Their use was assessed with a dictionary approach. The dictionary can be downloaded here. To assess how well the dictionary identified the types of devices in the papers, a quality control procedure was established. For each aspect, a random sample of 100 papers was selected, and a coauthor who did not lead the construction of the dictionary was randomly selected to check if in those papers the categories of the dictionary were correctly identified (i.e. accuracy). The accuracy was 84%. To assess if we were identifying animal-borne GPS rather than hand-held GPS, we took a sample of 100 papers already associated to GPS according to our dictionary. 83 referred to GPS tags, 12 were hand held, and 5 were neither tag nor hand-held (e.g. GPS coordinates in general, or General Practitioners). 3.4.1 Output Fig. 3.15. Proportion of papers in each year using the five most commonly used tracking devices. A study can use more than one device, hence the proportions for each year can sum up to more than one. The code for plot can be downloaded here. 3.5 Software Here we also used a dictionary approach. We used expert opinion to compile all known software used in movement ecology. The 33 software in our list were 1) R, 2) Python, 3) SPSS, 4) Matlab, 5) SAS, 6) MARK (program Mark and not R package unmarked), 7) Java, 8) C (if researchers wrote C code themselves; i.e some tracking Radars for pre-processing), 9) Fortran, 10) WinBUGS, 11) Agent-Analyst, 12) BASTrack, 13) QGIS, 14) GRASS, 15) Microsoft Excel, 16) Noldus observer, 17) fragstats, 18) postgis (we separated postGIS from the database category because its high spatial analytical capabilities), 19) databases (any relational database, likely for data management and summarizing necessarily for analyitical use), 20) e-surge, 21) m-surge, 22) u-care, 23) Genstat, 24) Biotas, 25) Statview, 26) Primer-e, 27) PAST, 28) STATA, 29) Statistica, 30) UCINET, 31) Mathcad, 32) Vicon, and 33) GME (geospatial modeling environment). The dictionary with the terms used can be downloaded here. For quality control, we examined a random sample of 50 papers. The accuracy was 88%. 3.5.1 Output Fig. 3.16. Proportion of papers in each year using the five most commonly used software. A study can use more than one software, hence the proportions for each year can sum up to more than one. The code for plot can be downloaded here. 3.6 Statistical methods Within our dictionary approach, we first used expert opinion to compile all known statistical methods (and terms used for them) that could have been used in movement ecology, resulting in 188 terms; click here to download the corresponding file. We classified statistical methods into Spatial, Time-series, Movement, Spatiotemporal, Social, and Generic. Their definitions are below: Spatial: spatial statistical methods (e.g. geostatistics) Time-series: time series methods (e.g. functional data analysis) Movement: statistical method used for the study of movement (e.g. behavioral change point analysis) Spatiotemporal: spatiotemporal but not movement methods (e.g. spatiotemporal geostatistics) Social: statistical methods that are not exclusively for movement, but that characterize or model social processes (e.g. social networks) Generic: generic statistical methods that could be used in any type of study, that are not inherently spatial, temporal or social (e.g. a regression analysis) Terms related to hypothesis tests were first considered but ultimately removed; we considered that the tendency of papers to present p-values could be biasing researchers towards the use of hypothesis tests, thus creating a bias towards general methods. For quality control, we examined a random sample of 50 papers. The accuracy was 84%. 3.6.1 Outputs Fig. 3.17. Proportion of papers in each year mentioning each type of statistical method. A study can use more than one type of method, hence the proportions for each year can sum up to more than one. The code for the plot can be downloaded here. generic Movement spatial time-series social spatiotemporal 67.8% 32.6% 18.9% 16.5% 3.3% 0.3% Table 3.6. Percentage of papers using each type of statistical method. The code for this table can be downloaded here. trigram n linear mixed models 231 linear mixed effects 229 generalized linear mixed 202 mixed effects models 202 linear mixed model 188 markov chain monte 180 chain monte carlo 178 akaike’s information criterion 174 akaike information criterion 162 minimum convex polygon 158 information criterion aic 146 monte carlo mcmc 133 correlated random walk 129 mixed effects model 117 hidden markov model 116 Table 3.7. Most common statistical trigrams in M&M sections of papers (with more than 100 mentions in papers). To reconstruct this table, readers can refer to this downloadable code. *The code to process the dictionaries of the framework, tracking devices, software, statistical methods, and taxonomy is downloadable here. References "],["survey-about-movement-ecology.html", "4 Survey about movement ecology 4.1 Description of the survey 4.2 Participation in the survey 4.3 Movement ecology framework 4.4 Taxa 4.5 Tracking devices 4.6 Software 4.7 Methods 4.8 Big changes in the field 4.9 Summary", " 4 Survey about movement ecology 4.1 Description of the survey As a complementary source of information for our review, we elaborated a survey to get the perspectives of movement ecologists about the field and how it is evolving. The exact formulation of the questions in the survey can be downloaded here. 4.2 Participation in the survey The survey was designed to be completely anonymous. There was no previous selection of the participants and no probabilistic sampling was involved. The survey was advertised by mailing lists (r-sig-geo and r-sig-ecology), individual emails to researchers, the lab’s website https://mablab.org, the Gordon Research Conference and Seminar on Movement Ecology of Animals, and Twitter. It took place during the Winter/Spring of 2019. The survey got exemption from the Institutional Review Board at University of Florida (IRB02 Office, Box 112250, University of Florida, Gainesville, FL 32611-2250). A total of 82 people participated in the survey, and 61 answered at least one question concerning the field (and not just their years of experience in research or the field). No question was mandatory, so participants could opt to not answer some questions. We first asked the participants how many years they have been doing research, and how many years they have been working on animal movement. They have spent a median of 10 years in research, and a median of 6 years in animal movement. 4.3 Movement ecology framework In a 2008 article, “A movement ecology paradigm for unifying organismal movement research,” Nathan et al. defined a movement ecology framework where the movement propagation process is produced by the motion and the navigation processes, with internal and external factors affecting movement. We asked the participants: Would you say that most research articles in movement ecology analyze these components of the movement ecology framework? Would you say that the these components of the movement ecology framework are currently being more, less or equally studied compared to 10 years ago? The results are shown in the graphs below. Most participants perceived that external factors are being addressed in most movement ecology papers, that motion is being addressed in at least half of the literature, and that navigation and external factors are addressed in less than half. Interestingly, when asked which components were most studied now than 10 years ago, most participants agreed on internal factors. 4.4 Taxa We asked the participants which taxa they considered to be studied the most in movement ecology, and to select up to 3 taxa. The number of votes per taxon are shown in the graph below. Birds and then mammals were indicated as the most studied. 4.5 Tracking devices We asked the participants the following questions: Which tracking device do you consider to be used the most in movement ecology? (up to 3) Which tracking devices do you think are used more often now compared to 10 years ago? (up to 3) Which tracking devices do you think are used less often now compared to 10 years ago? (up to 3) The results are shown in the graphs above. GPS and satellite tags (e.g. PSAT, PTT) are the most used devices, according to the participants. When compared to 10 years ago, they expressed that accelerometers and GPS devices are more used than before, while radio tags and encounter techniques (e.g. capture mark recapture, banding, direct observation) are less used. 4.6 Software We asked the participants the following questions: Which software do you think is used the most for movement analysis? (up to 3) For movement analysis, which software do you think are used more often now compared to 10 years ago? (up to 3) For movement analysis, which software do you think are used less often now compared to 10 years ago? (up to 3) The results are shown in the graphs above. R, and in the a lesser degree, ArcGIS, are the most used software, according to the participants. When compared to 10 years ago, they expressed that R is more used than before, while Matlab, ArcGIS, SPSS and SAS are less used. 4.7 Methods We asked the participants the following questions: Which statistical/mathematical methods do you consider to be used the most for movement analysis? Which methods do you think are used more often now compared to 10 years ago? (up to 3) Which methods do you think are used less often now compared to 10 years ago? (up to 3) There are many analytical tools applied/developed in movement ecology. To keep it simple, we provided the participants with 9 (arbitrarily selected) options: generalized linear models (GLMs) and generalized additive models (GAMs), machine learning, model selection criteria, multivariate exploratory methods, net squared displacement (NSD), spatial point processes, state-space and Hidden Markov models (SSMs and HMMs, respectively), step and resource selection functions (SSFs and RSFs, respectively), and test statistics and p-values. We provided an “Other” option for researchers to indicate methods that would not be within the 9 other options. In the end, the “Other” option was poorly used. The results are shown in the graphs above. The most used methods in movement ecology, according to the participants, are SSFs and RSFs, SSMs and HMMs, GLMs and GAMs. When compared to 10 years ago, they expressed that SSMs and HMMs, as well as machine learning, and in a lesser degree, SSFs and RSFs are more used, while hypothesis tests are less used. The “Other” option was used for speed threshold (1) and mininum convex polygon and kernel density utilization (1), for the first question; visual analytics (1), and continuous time movement modeling (1) for the second question; descriptive metrics (1) and “none” (1) for the third. 4.8 Big changes in the field We asked the participants three final open questions. In your opinion, what has revolutionized the field in the last 10 years? (Please keep it to three topics) 33 participants answered this first question. 31 answers were related to tagging devices (smaller, long-lasting, cheaper, battery-saving devices) that gave rise to longer or higher resolution data on many species. 15 answers were related to software development and computational power, 6 to statistical methods, 2 to the availability of remote sensing data, 2 to availability of tracking data online (in open data portals like movebank), 1 to molecular markers, 1 to citizen science, and there were other references to “mechanistic thinking and approaches” (1), the increase in the number of studies (1) and Nathan et al. 2008 paper in PNAS (1). Compared to 10 years ago, what would you be able to work on now that you could not do 10 years ago? (Please keep it to three topics; your answer may involve taxa, devices, methods or others) 29 participants answered this question. 28 answers were related to tagging devices that allowed working with smaller species, obtaining longer datasets, tracking more individuals and having high resolution data on their movement; among them, 4 mentioned the opportunity to work with accelerometry data. 5 answers were related to working with environmental data to link to the movement data (one mentioned it to do habitat selection analysis), 4 people mentioned studies in physiology, 3 of behavior, 3 of movement models and analysis in general, 1 studies on the mechanisms underlying movement, 1 multitaxa comparison, 1 citizen science thanks to apps on the phone. 2 people mentioned computational simulation and 2 processing data. In your opinion, what will revolutionize the field in the next 10 years? (Please keep it to three topics) 32 participants answered this question. 21 gave answers related to tagging devices (including use of ICARUS technology, 3D monitoring, development of more reliable and less invasive devices, improvement of accelerometers, and combined use of devices). 5 people referred to the development of statistical and mathematical methods to study movement processes (including machine learning techniques); 3 mentioned the availability of high resolution remote sensing data; 2 answers were related to software and computing power, 2 to connecting movement ecology to physiology, 2 to evolution, 1 to ontogeny, 1 to large-scale navigation experiments, 1 to integrating navigation, genetics, environment, physiology, life history and population dynamic studies; 1 to the development of data platforms, and 1 to “improved conceptualization of the field.” 4.9 Summary Most participants in the survey had less than a decade of experience in the field, so their farthest point of comparison in the past is from less than 10 years. Participants identified external factors as the dominant component in movement ecology papers, though recognizing that there is an increasing but still weaker interest in the link between internal factors and movement. Birds and mammals were identified as the most studied taxa in the field. GPS tags were identified as the most used tags, while the use of accelerometers is increasing and the use of radio tags, decreasing. The R software is the most used software according to the participants, and it is far more used than 10 years ago. On the other hand, Matlab, ArcGIS, SPSS and SAS use have decreased in the last decade. RSFs, SSFs, SSMs, HMMs, as well as GLMs and GAMs, are the most used statistical methods in movement ecology. SSM, HMM and machine learning techniques have experienced a notable increase in their use in the last decade, while the use of statistical tests, mainly, have decreased. When asked for what has revolutionized the field and what will play a major role in the next 10 years, most of the answers were related to tracking devices (miniaturization, reliability, battery saving, combined use of them). In a lesser degree, software, data and methods to analyze movement were mentioned. Most participants identified tracking technology as the game changer in the field. "],["r-session-information.html", "5 R session information", " 5 R session information devtools::session_info() ─ Session info ──────────────────────────────────────────── setting value version R version 3.6.3 (2020-02-29) os Ubuntu 18.04.4 LTS system x86_64, linux-gnu ui RStudio language (EN) collate en_US.UTF-8 ctype en_US.UTF-8 tz America/Nassau date 2020-05-22 ─ Packages ─────────────────────────────────────────────── package * version date lib source aRxiv 0.5.19 2019-08-08 [1] CRAN (R 3.6.1) assertthat 0.2.1 2019-03-21 [1] CRAN (R 3.6.1) audio 0.1-6 2019-03-19 [1] CRAN (R 3.6.1) backports 1.1.5 2019-10-02 [1] CRAN (R 3.6.1) beepr * 1.3 2018-06-04 [1] CRAN (R 3.6.1) bibtex 0.4.2 2017-06-30 [1] CRAN (R 3.6.1) bit 1.1-14 2018-05-29 [1] CRAN (R 3.6.1) bit64 0.9-7 2017-05-08 [1] CRAN (R 3.6.1) bitops 1.0-6 2013-08-17 [1] CRAN (R 3.6.1) blob 1.2.0 2019-07-09 [1] CRAN (R 3.6.1) bookdown 0.16 2019-11-22 [1] CRAN (R 3.6.1) broom 0.5.2 2019-04-07 [1] CRAN (R 3.6.1) callr 3.3.2 2019-09-22 [1] CRAN (R 3.6.1) caTools 1.17.1.2 2019-03-06 [1] CRAN (R 3.6.1) cellranger 1.1.0 2016-07-27 [1] CRAN (R 3.6.1) cli 2.0.2 2020-02-28 [1] CRAN (R 3.6.3) clipr 0.7.0 2019-07-23 [1] CRAN (R 3.6.1) cluster 2.1.0 2019-06-19 [1] CRAN (R 3.6.3) codetools 0.2-16 2018-12-24 [4] CRAN (R 3.6.3) colorspace 1.4-1 2019-03-18 [1] CRAN (R 3.6.1) cowplot * 1.0.0 2019-07-11 [1] CRAN (R 3.6.1) crayon 1.3.4 2017-09-16 [1] CRAN (R 3.6.1) crul 0.9.0 2019-11-06 [1] CRAN (R 3.6.1) curl 4.3 2019-12-02 [1] CRAN (R 3.6.3) data.table 1.12.6 2019-10-18 [1] CRAN (R 3.6.1) DBI 1.1.0 2019-12-15 [1] CRAN (R 3.6.1) dbplyr * 1.4.2 2019-06-17 [1] CRAN (R 3.6.1) dendextend 1.12.0 2019-05-11 [1] CRAN (R 3.6.1) desc 1.2.0 2018-05-01 [1] CRAN (R 3.6.1) details * 0.2.1 2020-01-12 [1] CRAN (R 3.6.3) devtools 2.2.1 2019-09-24 [1] CRAN (R 3.6.1) digest 0.6.25 2020-02-23 [1] CRAN (R 3.6.3) dotCall64 1.0-0 2018-07-30 [1] CRAN (R 3.6.1) dplyr * 0.8.3 2019-07-04 [1] CRAN (R 3.6.1) DT 0.10 2019-11-12 [1] CRAN (R 3.6.1) ellipsis 0.3.0 2019-09-20 [1] CRAN (R 3.6.1) evaluate 0.14 2019-05-28 [1] CRAN (R 3.6.1) fansi 0.4.1 2020-01-08 [1] CRAN (R 3.6.3) farver 2.0.3 2020-01-16 [1] CRAN (R 3.6.3) fastmap 1.0.1 2019-10-08 [1] CRAN (R 3.6.1) fields 10.0 2019-11-11 [1] CRAN (R 3.6.1) forcats * 0.4.0 2019-02-17 [1] CRAN (R 3.6.1) foreach 1.4.7 2019-07-27 [1] CRAN (R 3.6.1) foreign 0.8-76 2020-03-03 [4] CRAN (R 3.6.3) fs 1.3.1 2019-05-06 [1] CRAN (R 3.6.1) fulltext * 1.3.0 2019-06-27 [1] CRAN (R 3.6.1) gclus 1.3.2 2019-01-07 [1] CRAN (R 3.6.1) gdata 2.18.0 2017-06-06 [1] CRAN (R 3.6.1) generator * 0.1.0 2015-08-26 [1] CRAN (R 3.6.1) generics 0.0.2 2018-11-29 [1] CRAN (R 3.6.1) ggmap 3.0.0 2019-02-05 [1] CRAN (R 3.6.1) ggplot2 * 3.3.0 2020-03-05 [1] CRAN (R 3.6.3) ggwordcloud * 0.5.0 2019-06-02 [1] CRAN (R 3.6.1) glue 1.4.0 2020-04-03 [1] CRAN (R 3.6.3) gplots 3.0.1.1 2019-01-27 [1] CRAN (R 3.6.1) gridExtra * 2.3 2017-09-09 [1] CRAN (R 3.6.1) gtable 0.3.0 2019-03-25 [1] CRAN (R 3.6.1) gtools 3.8.1 2018-06-26 [1] CRAN (R 3.6.1) haven 2.2.0 2019-11-08 [1] CRAN (R 3.6.1) heatmaply * 0.16.0 2019-05-11 [1] CRAN (R 3.6.1) hms 0.5.2 2019-10-30 [1] CRAN (R 3.6.1) hoardr 0.5.2 2018-12-02 [1] CRAN (R 3.6.1) htmltools 0.4.0 2019-10-04 [1] CRAN (R 3.6.1) htmlwidgets 1.5.1 2019-10-08 [1] CRAN (R 3.6.1) httpcode 0.2.0 2016-11-14 [1] CRAN (R 3.6.1) httpuv 1.5.2 2019-09-11 [1] CRAN (R 3.6.1) httr 1.4.1 2019-08-05 [1] CRAN (R 3.6.1) iterators 1.0.12 2019-07-26 [1] CRAN (R 3.6.1) janeaustenr 0.1.5 2017-06-10 [1] CRAN (R 3.6.1) jpeg 0.1-8.1 2019-10-24 [1] CRAN (R 3.6.1) jsonlite 1.6.1 2020-02-02 [1] CRAN (R 3.6.3) KernSmooth 2.23-17 2020-04-26 [4] CRAN (R 3.6.3) knitr 1.28 2020-02-06 [1] CRAN (R 3.6.3) koRpus * 0.11-5 2018-10-28 [1] CRAN (R 3.6.1) koRpus.lang.en * 0.1-3 2019-05-13 [1] CRAN (R 3.6.1) labeling 0.3 2014-08-23 [1] CRAN (R 3.6.1) later 1.0.0 2019-10-04 [1] CRAN (R 3.6.1) lattice 0.20-41 2020-04-02 [4] CRAN (R 3.6.3) lazyeval 0.2.2 2019-03-15 [1] CRAN (R 3.6.1) lifecycle 0.2.0 2020-03-06 [1] CRAN (R 3.6.3) lubridate 1.7.4 2018-04-11 [1] CRAN (R 3.6.1) magrittr * 1.5 2014-11-22 [1] CRAN (R 3.6.1) maps 3.3.0 2018-04-03 [1] CRAN (R 3.6.1) maptools 0.9-8 2019-10-05 [1] CRAN (R 3.6.1) MASS 7.3-51.6 2020-04-26 [4] CRAN (R 3.6.3) Matrix 1.2-18 2019-11-27 [4] CRAN (R 3.6.1) memoise 1.1.0 2017-04-21 [1] CRAN (R 3.6.1) microdemic 0.4.0 2018-10-25 [1] CRAN (R 3.6.1) mime 0.9 2020-02-04 [1] CRAN (R 3.6.3) miniUI 0.1.1.1 2018-05-18 [1] CRAN (R 3.6.1) modelr 0.1.5 2019-08-08 [1] CRAN (R 3.6.1) modeltools 0.2-22 2018-07-16 [1] CRAN (R 3.6.1) munsell 0.5.0 2018-06-12 [1] CRAN (R 3.6.1) network 1.15 2019-04-02 [1] CRAN (R 3.6.1) nlme 3.1-147 2020-04-13 [4] CRAN (R 3.6.3) NLP * 0.2-0 2018-10-18 [1] CRAN (R 3.6.1) pillar 1.4.2 2019-06-29 [1] CRAN (R 3.6.1) pkgbuild 1.0.6 2019-10-09 [1] CRAN (R 3.6.1) pkgconfig 2.0.3 2019-09-22 [1] CRAN (R 3.6.1) pkgload 1.0.2 2018-10-29 [1] CRAN (R 3.6.1) plotly * 4.9.1 2019-11-07 [1] CRAN (R 3.6.1) plyr * 1.8.4 2016-06-08 [1] CRAN (R 3.6.1) png 0.1-7 2013-12-03 [1] CRAN (R 3.6.1) prettyunits 1.0.2 2015-07-13 [1] CRAN (R 3.6.1) processx 3.4.1 2019-07-18 [1] CRAN (R 3.6.1) promises 1.1.0 2019-10-04 [1] CRAN (R 3.6.1) ps 1.3.0 2018-12-21 [1] CRAN (R 3.6.1) purrr * 0.3.3 2019-10-18 [1] CRAN (R 3.6.1) R6 2.4.1 2019-11-12 [1] CRAN (R 3.6.1) rappdirs 0.3.1 2016-03-28 [1] CRAN (R 3.6.1) RColorBrewer * 1.1-2 2014-12-07 [1] CRAN (R 3.6.1) Rcpp 1.0.4.6 2020-04-09 [1] CRAN (R 3.6.3) rcrossref 0.9.2 2019-05-04 [1] CRAN (R 3.6.1) readr * 1.3.1 2018-12-21 [1] CRAN (R 3.6.1) readxl 1.3.1 2019-03-13 [1] CRAN (R 3.6.1) refsplitr * 1.0 2020-05-22 [1] Github (ropensci/refsplitr@ad6e199) registry 0.5-1 2019-03-05 [1] CRAN (R 3.6.1) remotes 2.1.0 2019-06-24 [1] CRAN (R 3.6.1) rentrez 1.2.2 2019-05-02 [1] CRAN (R 3.6.1) reprex 0.3.0 2019-05-16 [1] CRAN (R 3.6.1) reshape2 1.4.3 2017-12-11 [1] CRAN (R 3.6.1) RgoogleMaps 1.4.4 2019-08-20 [1] CRAN (R 3.6.1) rjson 0.2.20 2018-06-08 [1] CRAN (R 3.6.1) rlang 0.4.5 2020-03-01 [1] CRAN (R 3.6.3) rmarkdown 2.1 2020-01-20 [1] CRAN (R 3.6.3) rplos 0.8.6 2019-05-03 [1] CRAN (R 3.6.1) rprojroot 1.3-2 2018-01-03 [1] CRAN (R 3.6.1) RSQLite * 2.2.0 2020-01-07 [1] CRAN (R 3.6.1) rstudioapi 0.11 2020-02-07 [1] CRAN (R 3.6.3) rvest 0.3.5 2019-11-08 [1] CRAN (R 3.6.1) rworldmap 1.3-6 2016-02-03 [1] CRAN (R 3.6.1) scales 1.1.0 2019-11-18 [1] CRAN (R 3.6.1) seriation 1.2-8 2019-08-27 [1] CRAN (R 3.6.1) sessioninfo 1.1.1 2018-11-05 [1] CRAN (R 3.6.1) shiny 1.4.0.2 2020-03-13 [1] CRAN (R 3.6.3) slam 0.1-46 2019-10-26 [1] CRAN (R 3.6.1) SnowballC 0.6.0 2019-01-15 [1] CRAN (R 3.6.1) solrium 1.1.4 2019-11-02 [1] CRAN (R 3.6.1) sp 1.3-2 2019-11-07 [1] CRAN (R 3.6.1) spam 2.4-0 2019-11-02 [1] CRAN (R 3.6.1) storr 1.2.1 2018-10-18 [1] CRAN (R 3.6.1) stringdist 0.9.5.5 2019-10-21 [1] CRAN (R 3.6.3) stringi 1.4.6 2020-02-17 [1] CRAN (R 3.6.3) stringr * 1.4.0 2019-02-10 [1] CRAN (R 3.6.1) sylly * 0.1-5 2018-07-29 [1] CRAN (R 3.6.1) sylly.en 0.1-3 2018-03-19 [1] CRAN (R 3.6.1) testthat 2.3.0 2019-11-05 [1] CRAN (R 3.6.1) textstem * 0.1.4 2018-04-09 [1] CRAN (R 3.6.1) tibble * 2.1.3 2019-06-06 [1] CRAN (R 3.6.1) tidyr * 1.0.0 2019-09-11 [1] CRAN (R 3.6.1) tidyselect 0.2.5 2018-10-11 [1] CRAN (R 3.6.1) tidytext * 0.2.2 2019-07-29 [1] CRAN (R 3.6.1) tidyverse * 1.3.0 2019-11-21 [1] CRAN (R 3.6.1) tm * 0.7-6 2018-12-21 [1] CRAN (R 3.6.1) tokenizers 0.2.1 2018-03-29 [1] CRAN (R 3.6.1) topicmodels * 0.2-8 2018-12-21 [1] CRAN (R 3.6.1) TSP 1.1-7 2019-05-22 [1] CRAN (R 3.6.1) usethis 1.5.1 2019-07-04 [1] CRAN (R 3.6.1) utf8 1.1.4 2018-05-24 [1] CRAN (R 3.6.1) vctrs 0.2.0 2019-07-05 [1] CRAN (R 3.6.1) viridis * 0.5.1 2018-03-29 [1] CRAN (R 3.6.1) viridisLite * 0.3.0 2018-02-01 [1] CRAN (R 3.6.1) webshot 0.5.2 2019-11-22 [1] CRAN (R 3.6.1) whisker 0.4 2019-08-28 [1] CRAN (R 3.6.1) withr 2.1.2 2018-03-15 [1] CRAN (R 3.6.1) xfun * 0.12 2020-01-13 [1] CRAN (R 3.6.3) XML 3.98-1.20 2019-06-06 [1] CRAN (R 3.6.1) xml2 * 1.3.1 2020-04-09 [1] CRAN (R 3.6.3) xtable 1.8-4 2019-04-21 [1] CRAN (R 3.6.1) yaml 2.2.1 2020-02-01 [1] CRAN (R 3.6.3) zeallot 0.1.0 2018-01-28 [1] CRAN (R 3.6.1) [1] /home/rjoo/R/x86_64-pc-linux-gnu-library/3.6 [2] /usr/local/lib/R/site-library [3] /usr/lib/R/site-library [4] /usr/lib/R/library "],["references.html", "References", " References "]] diff --git a/docs/survey-about-movement-ecology.html b/docs/survey-about-movement-ecology.html index ab54197..9d0a262 100644 --- a/docs/survey-about-movement-ecology.html +++ b/docs/survey-about-movement-ecology.html @@ -24,7 +24,7 @@ - + @@ -247,7 +247,11 @@

4.2 Participation in the surveyThe survey was designed to be completely anonymous. There was no previous selection of the participants and no probabilistic sampling was involved. The survey was advertised by mailing lists (r-sig-geo and r-sig-ecology), individual emails to researchers, the lab’s website https://mablab.org, the Gordon Research Conference and Seminar on Movement Ecology of Animals, and Twitter. It took place during the Winter/Spring of 2019.

The survey got exemption from the Institutional Review Board at University of Florida (IRB02 Office, Box 112250, University of Florida, Gainesville, FL 32611-2250).

A total of 82 people participated in the survey, and 61 answered at least one question concerning the field (and not just their years of experience in research or the field). No question was mandatory, so participants could opt to not answer some questions.

-

+
+Histogram of the years in research of all participants (left), and histogram of the years working on movement (right). Both histograms are skewed to the right. +

+

+

We first asked the participants how many years they have been doing research, and how many years they have been working on animal movement. They have spent a median of 10 years in research, and a median of 6 years in animal movement.

@@ -258,12 +262,12 @@

4.3 Movement ecology framework

Would you say that the these components of the movement ecology framework are currently being more, less or equally studied compared to 10 years ago?

The results are shown in the graphs below. Most participants perceived that external factors are being addressed in most movement ecology papers, that motion is being addressed in at least half of the literature, and that navigation and external factors are addressed in less than half. Interestingly, when asked which components were most studied now than 10 years ago, most participants agreed on internal factors.

-

+

Horizontal stacked bar plots of the participants opinion on how much the community is focusing on each MEF component (left), and on how much focus there is compared to ten years ago (right). Results described in the text above.

4.4 Taxa

We asked the participants which taxa they considered to be studied the most in movement ecology, and to select up to 3 taxa. The number of votes per taxon are shown in the graph below. Birds and then mammals were indicated as the most studied.

-

+

Horizontal bar plot of the participants vote on which taxonomical group they considered to be the most studied in movement ecology. In decreasing order of votes: Aves, Mammalia, Sharks and Rays, Bony-fish, Homo Sapiens, Arthropoda, Reptilia, and Amphibia.

4.5 Tracking devices

@@ -273,8 +277,8 @@

4.5 Tracking devices

  • Which tracking devices do you think are used more often now compared to 10 years ago? (up to 3)

  • Which tracking devices do you think are used less often now compared to 10 years ago? (up to 3)

    • -

    -

    +

    Horizontal bar plots indicating participants' votes on which devices are the most used. In decreasing order of votes: GPS, satellite tags, radio tags, encounter, light loggers, acoustic telemetry, accelerometer, data storage tags, and images.

    +

    Left: horizontal bar plots indicating participants' votes on which devices are more used now compared to ten years ago. In decreasing order of votes: accelerometer, gps, acoustic telemetry, satellite tags, light loggers, images, data storage tags, encounter, radio tags. Right: horizontal bar plots indicating participants' votes on which devices are less used now compared to ten years ago. In decreasing order of votes: radio tags, encounter, satellite tags, light loggers, acoustic telemetry, images, none, and gps.

    The results are shown in the graphs above. GPS and satellite tags (e.g. PSAT, PTT) are the most used devices, according to the participants. When compared to 10 years ago, they expressed that accelerometers and GPS devices are more used than before, while radio tags and encounter techniques (e.g. capture mark recapture, banding, direct observation) are less used.

    @@ -285,8 +289,8 @@

    4.6 Software

  • For movement analysis, which software do you think are used more often now compared to 10 years ago? (up to 3)

  • For movement analysis, which software do you think are used less often now compared to 10 years ago? (up to 3)

    • -

    -

    +

    Horizontal bar plots indicating participants' votes on which software are the most used. In decreasing order of votes: R, ArcGIS, QGIS, Python, Matlab, RDBMs.

    +

    Left: horizontal bar plots indicating participants' votes on which software are more used now compared to ten years ago. In decreasing order of votes: R, QGIS, Python, Matlab, ArcGIS, RDBMs. Right: horizontal bar plots indicating participants' votes on which software are less used now compared to ten years ago. In decreasing order of votes: Matlab, ArcGIS, SPSS, SAS, RDBMs, Python.

    The results are shown in the graphs above. R, and in the a lesser degree, ArcGIS, are the most used software, according to the participants. When compared to 10 years ago, they expressed that R is more used than before, while Matlab, ArcGIS, SPSS and SAS are less used.

    @@ -298,8 +302,8 @@

    4.7 Methods

  • Which methods do you think are used less often now compared to 10 years ago? (up to 3)

    • There are many analytical tools applied/developed in movement ecology. To keep it simple, we provided the participants with 9 (arbitrarily selected) options: generalized linear models (GLMs) and generalized additive models (GAMs), machine learning, model selection criteria, multivariate exploratory methods, net squared displacement (NSD), spatial point processes, state-space and Hidden Markov models (SSMs and HMMs, respectively), step and resource selection functions (SSFs and RSFs, respectively), and test statistics and p-values. We provided an “Other” option for researchers to indicate methods that would not be within the 9 other options. In the end, the “Other” option was poorly used.

    -

    -

    +

    Horizontal bar plots indicating participants' votes on which statistical/mathematical methods are the most used. In decreasing order of votes: SSF and RSF, SSM and HMM, GLMs and GAMs, model selection, tests, NSD, spatial point, multivariate, other, and machine learning.

    +

    Left: horizontal bar plots indicating participants' votes on which statistical/mathematical methods are more used now compared to ten years ago. In decreasing order of votes: SSM and HMM, machine learning, SSF and RSF, model selection, multivariate, spatial point, GLMs and GAMs, NSD, and other. Right: horizontal bar plots indicating participants' votes on which statistical/mathematical methods are less used now compared to ten years ago. In decreasing order of votes: tests, NSD, multivariate, spatial point, SSF and RSF, model selection, other, and GLMs and GAMs.

    The results are shown in the graphs above. The most used methods in movement ecology, according to the participants, are SSFs and RSFs, SSMs and HMMs, GLMs and GAMs. When compared to 10 years ago, they expressed that SSMs and HMMs, as well as machine learning, and in a lesser degree, SSFs and RSFs are more used, while hypothesis tests are less used.

    The “Other” option was used for speed threshold (1) and mininum convex polygon and kernel density utilization (1), for the first question; visual analytics (1), and continuous time movement modeling (1) for the second question; descriptive metrics (1) and “none” (1) for the third.