For Peer Review 20 8 21 22 23 9 José F

Page 1 of 56 American Journal of Primatology

1 2 3 4 1 Title: Density and distribution of western chimpanzees around a bauxite deposit in 5 6 7 2 the Boé Sector, Guinea-Bissau. 8 9 3 10 11 4 Running title: Western chimpanzees of Boé Sector. 12 13 14 5 15 16 6 José F. C. Wenceslau1 and Filipe S. Dias2,3,4*, Tiago A. Marques5,6 and David L. 17 18 7 Miller6 19 For Peer Review 20 8 21 22 23 9 José F. C. Wenceslau and Filipe S. Dias should be considered joint first authors 24 25 10 26 27 11 1 Foundation Chimbo, Amsterdam, The Netherlands 28 29 12 2 Centre for Applied Ecology “Prof. Baeta Neves” (CEABN – InBIO), School of 30 31 32 13 Agriculture, University of Lisbon, Tapada da Ajuda, 1349-017 Lisbon, Portugal 33 34 14 3 CIBIO/InBio, Centro de Investigação em Biodiversidade e Recursos Genéticos, 35 36 15 Laboratório Associado, Universidade do Porto, Campus Agrário de Vairão, 4485- 37 38 661 Vairão, Portugal 39 16 40 41 17 4 CIBIO/InBio, Centro de Investigação em Biodiversidade e Recursos Genéticos, 42 43 18 Laboratório Associado, Instituto Superior de Agronomia, Universidade de Lisboa, 44 45 19 Tapada da Ajuda, 1349-017 Lisbon, Portugal 46 47 5 48 20 Centre for Research into Ecological & Environmental Modelling and School of 49 50 21 Mathematics & Statistics, University of St Andrews, St Andrews, Scotland, United 51 52 22 Kingdom 53 54 55 56 57 58 59 60 John Wiley & Sons American Journal of Primatology Page 2 of 56

1 2 3 4 23 6 Centro de Estatística e Aplicações da Universidade de Lisboa, Departamento de 5 6 7 24 Biologia Animal, Faculdade de Ciências da Universidade de Lisboa, Portugal 8 9 25 10 11 26 *Correspondence to: Filipe Dias ([email protected]), CIBIO - Instituto Superior 12 13 14 27 de Agronomia Tapada da Ajuda), Tapada da Ajuda, 1349-017 Lisboa 15 16 17 18 19 For Peer Review 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 John Wiley & Sons Page 3 of 56 American Journal of Primatology

1 2 3 4 28 Research Highlights 5 6 7 29 8 9 30 • Approximately 18 nest building western chimpanzees inhabit the 10 11 31 surroundings of a bauxite deposit in the SW of Guinea-Bissau; 12 13 14 32 15 16 33 • The construction of a mine can have adverse direct and indirect effects on 17 18 34 this population. 19 For Peer Review 20 21 35 22 23 36 24 25 37 26 27 28 38 29 30 39 31 32 40 33 34 41 35 36 37 42 38 39 43 40 41 44 42 43 44 45 45 46 46 47 48 47 49 50 51 48 52 53 49 54 55 56 57 58 59 60 John Wiley & Sons American Journal of Primatology Page 4 of 56

1 2 3 4 50 Abstract 5 6 7 51 The Boé sector in southeast Guinea-Bissau harbors a population of western 8 9 52 chimpanzees (Pan troglodytes verus) that inhabits a mosaic of forest and savanna. 10 11 53 The Boé sector contains a substantial bauxite deposit in a region called Ronde Hill, 12 13 14 54 and there are plans for the construction of a mine, which may endanger the 15 16 55 chimpanzee population. In a one-week survey in May 2013, we used the standing 17 18 56 crop nest counts method to obtain the number of chimpanzee nests and from that 19 For Peer Review 20 57 estimate the density and abundance of chimpanzees. We carried out five 1 km line 21 22 23 58 transects that covered the bauxite deposit and surrounding valleys. We used 24 25 59 density surface modeling to analyze habitat preferences, then predicted 26 27 60 chimpanzee nest density and distribution based on environmental variables. We 28 29 30 61 found the projected location of the mine partially coincides with an area of high 31 32 62 predicted abundances of chimpanzee nests and is surrounded by highly suitable 33 34 63 areas for chimpanzees (northeast and southwest). We conclude the mine could 35 36 37 64 have significant direct and indirect effects on this population of chimpanzees 38 39 65 whose impacts must be carefully considered and properly mitigated if the mine is 40 41 66 built. 42 43 67 44 45 46 68 Keywords: western chimpanzee, Boé, bauxite mining, Guinea-Bissau, Density 47 48 69 Surface Modelling 49 50 70 51 52 53 54 55 56 57 58 59 60 John Wiley & Sons Page 5 of 56 American Journal of Primatology

1 2 3 4 71 1. Introduction 5 6 7 72 Western chimpanzees (Pan troglodytes verus, Schwarz) are a subspecies of 8 9 73 chimpanzee whose distribution ranges from tropical lowland forests in Liberia, Côte 10 11 74 d'Ivoire, and Sierra Leone to savannas in Guinea, Guinea-Bissau, Senegal, and 12 13 14 75 Mali, that can also inhabit some highly humanized agro-forestry systems in these 15 16 76 regions (Kühl et al., 2017). Western chimpanzees are currently listed as Critically 17 18 77 Endangered in the International Union for the Conservation of Nature’s Red List 19 For Peer Review 20 78 (Humle et al., 2016). The population of western chimpanzees declined by 80% and 21 22 23 79 lost 20% of its range from 1990 to 2014 (Kühl et al., 2017). The most significant 24 25 80 losses occurred in Côte d'Ivoire, where the population declined by 90%, mostly due 26 27 81 to deforestation, poaching, and infectious diseases (Campbell, Kuehl, N’Goran 28 29 30 82 Kouamé & Boesch, 2008). In Senegal and Ghana, there are fewer than 1000 31 32 83 individuals (Kormos & Bakarr 2003; Danquah, Oppong, Akom & Sam, 2012) and in 33 34 84 Benin, Togo and Burkina-Faso western chimpanzees are probably extinct (Ginn, 35 36 37 85 Robison, Redmond & Nekaris, 2013; Khül et al., 2017). 38 39 86 In Guinea-Bissau, chimpanzees were declared extinct in 1988, but subsequent 40 41 87 surveys found populations in the Quinara and Tombali regions (in the southwest) 42 43 88 and in Medina do Boé (a sector south of the Gabu region; Gippoliti, Embalo & 44 45 46 89 Sousa, 2003; Brugiere, Badjinca, Silva & Serra, 2009). No country-wide 47 48 90 abundance estimates are available for Guinea-Bissau, but some surveys suggest 49 50 91 the population may range between 600 and 1000 individuals (Gippoliti et al. 2003). 51 52 53 92 A study in Lagoas de Cufada Natural Park, in Quinara region, estimated 137 54 55 93 individuals (95% CI: 51–390) (Carvalho, Marques & Vicente, 2013). In Southern 56 57 58 59 60 John Wiley & Sons American Journal of Primatology Page 6 of 56

1 2 3 4 94 Cantanhez National Park, in the Tombali region, a study reported fewer than 100 5 6 7 95 chimpanzees (Sousa, Barata, Sousa, Casanova, & Vicente, 2011). In the Boé 8 9 96 sector, Serra, Silva, & Lopes (2007) interviewed hunters and other knowledgeable 10 11 97 locals and came to an estimate of 710 individuals. The main threat to western 12 13 14 98 chimpanzees in Guinea-Bissau is habitat loss and fragmentation due to expanding 15 16 99 plantations of banana, cashew, and other fruits (Gippoliti et al., 2003). Expansion 17 18 100 of mining operations can also impact chimpanzees, as some studies conducted in 19 For Peer Review 20 101 other West African countries have suggested (Diallo, 2010; Humle et al., 2016). 21 22 23 102 Mining operations can have direct and indirect impacts on great apes (Arcus 24 25 103 Foundation, 2014). The construction of mines can cause habitat loss, and mining 26 27 104 operations can cause water contamination and habitat degradation (Kusin et al., 28 29 30 105 2017, Mensah et al., 2015). The noise from mineral extraction can disturb apes 31 32 106 and cause them to move to other areas, thus disrupting their behaviors and social 33 34 107 structure. The construction of roads for transporting minerals and workers can 35 36 37 108 cause habitat loss, fragmentation, and increase disturbance (Arcus Foundation, 38 39 109 2014, Carvalho et al., 2013, Gippoliti et al., 2003; Hockings & Humle, 2009). The 40 41 110 influx of new workers brought to work on mines can increase bushmeat hunting 42 43 111 (Laurence et al., 2005) and promote conversion of forest into agricultural areas to 44 45 46 112 cultivate crops. Frequent contact between humans and chimpanzees can also 47 48 113 increase the probability of transmission of diseases for which chimpanzees lack 49 50 114 immunity, such as bacterial respiratory diseases (Köndgen et al., 2008) and Ebola 51 52 53 115 (Arcus Foundation, 2014, Devos, Sanz, Morgan, Onononga & Laporte, 2008). 54 55 116 56 57 58 59 60 John Wiley & Sons Page 7 of 56 American Journal of Primatology

1 2 3 4 117 The Boé sector is located in the southeast of Guinea-Bissau and presents the 5 6 7 118 highest altitudes in the country. The region contains lateritic plateaus, mostly close 8 9 119 to the border with Guinea, with considerable amounts of bauxite (Diallo, 2010). 10 11 120 Ronde Hill is where bauxite prospecting first began in the 1970s by Russian 12 13 14 121 investors. In 2008, Bauxite Angola S.A. continued prospecting in association with 15 16 122 Compagnie Bauxite de Guinée and built a road in the region. This road connects 17 18 123 the deposit with the Republic of Guinea and is meant to facilitate the transportation 19 For Peer Review 20 124 of machinery for bauxite exploitation (Wit, 2011). Mining has not started and is 21 22 23 125 contingent on agreements between Bauxite Angola S.A. and the Guinea-Bissau 24 25 126 government that include the improvement of transportation infrastructure. Mining 26 27 127 would take place at the crest of the hill, an area important for maintaining water 28 29 30 128 quantity and quality in the Jabere and Paramaka rivers and adjacent valleys (Wit, 31 32 129 2011). Since these valleys host a population of western chimpanzees (Wit, 2011), 33 34 130 it is crucial to assess the distribution of chimpanzees to understand the possible 35 36 37 131 effects of mining and to develop mitigation strategies. 38 39 132 Here we estimate the abundance and distribution of chimpanzee populations in 40 41 133 Ronde Hill and adjacent valleys to assess the potential impacts of a bauxite mine. 42 43 134 We 1) determined the density and abundance of nest building chimpanzees based 44 45 46 135 on the distribution of nests and 2) analyzed the overlap between chimpanzee nests 47 48 136 and the mining area to assess potential impacts. 49 50 137 51 52 53 138 54 55 139 56 57 58 59 60 John Wiley & Sons American Journal of Primatology Page 8 of 56

1 2 3 4 140 2. Methods 5 6 7 141 Study area 8 9 142 The survey was conducted over approximately 47 km2, comprising Ronde Hill, 10 11 143 which includes the prospected bauxite deposit, and the basins of the rivers 12 13 14 144 Paramaka and Jabere rivers and its tributaries, Barquere, Gra, Jabeje, Mussa and 15 16 145 Tuncotanca creeks (Fig. 1). This site is in the southern limit of the Boé sector, 17 18 146 which is close to the border with the Republic of Guinea (11° 41' N, 13° 54' W). The 19 For Peer Review 20 147 nearest human settlements are the villages of Capebonde in Guinea-Bissau and 21 22 23 148 Paramakadow and Paramakaley on the Guinean side of the border. Soils in Ronde 24 25 149 Hill are shallow and mostly in the early stages of laterization. As a consequence, 26 27 150 savanna is predominant, and forests occur only where the topsoil layer is deeper 28 29 30 151 than one meter and does not flood for prolonged periods (Wit & Reintjes, 1989). 31 32 152 Ethics statement 33 34 153 The present study complies with the Principles for the Ethical Treatment of 35 36 37 154 Non-Human Primates of the American Society of Primatologists. This research was 38 39 155 also approved by Guinea-Bissau’s Instituto da Biodiversidade e das Areas 40 41 156 Protegidas (IBAP). Since the sampling methods we used did not require direct 42 43 157 contact between researchers and chimpanzees, disturbance and health threats to 44 45 46 158 chimpanzees were minimal. 47 48 159 49 50 160 51 52 53 161 54 55 162 56 57 58 59 60 John Wiley & Sons Page 9 of 56 American Journal of Primatology

1 2 3 4 163 Estimating the abundance of chimpanzees 5 6 7 164 Since directly counting chimpanzees is often impractical, surveyors usually use 8 9 165 indirect methods. In our case this involved counting nests, which chimpanzees 10 11 166 build using branches and leaves. Nests are relatively easy to detect, remain visible 12 13 14 167 for weeks, months, or even years and can be counted with distance sampling 15 16 168 techniques (Buckland et al. 2001, Thomas et al. 2010). Chimpanzee abundances 17 18 169 can then be estimated by combining the density of nests with nest construction 19 For Peer Review 20 170 rates, nest decay rates, and the proportion of the population that builds the nests 21 22 23 171 (see below). 24 25 172 We established five parallel transects (each 1 km, North-South orientation) that 26 27 173 were spaced one kilometer apart and encompassed Ronde Hill and adjacent 28 29 30 174 valleys. During the first week of May 2013, three people followed the Standing 31 32 175 Crop Nest Count (SCNC) protocol (Spehar et al., 2010): they walked along each 33 34 176 transect carrying a GPS device (Garmin eTrex 10) and recorded the coordinates of 35 36 37 177 chimpanzee nests and the perpendicular distance between each nest and the 38 39 178 transect with a measuring tape. The decay stage of each nest was recorded 40 41 179 following the scale used by Plumptre & Reynolds (1997): 1- if the nest is still fresh 42 43 180 and stable, with green leaves and feces or feeding signs underneath, 2- if it is still 44 45 46 181 solid, but the leaves have signs of drying, 3- if the nest presents only dried leaves 47 48 182 and/or is starting to lose its structure, and 4- if it lost every leaf but is still 49 50 183 recognizable as a nest due to the presence of broken branches and twigs. The 51 52 53 184 surrounding environment around each nest was also classified according to four 54 55 185 categories: 1) "primary forest" for pristine forested habitats or forests in later 56 57 58 59 60 John Wiley & Sons American Journal of Primatology Page 10 of 56

1 2 3 4 186 successional stages, 2) "secondary forest" for agricultural land abandoned for 5 6 7 187 longer than five years that present dense mid-story and is starting to regain canopy 8 9 188 closure, 3) "fallow" for agricultural fields abandoned for less than four years or still 10 11 189 active, and 4) "savanna" for open or sparsely arborized grasslands. Contrary to the 12 13 14 190 work of Bryson-Morrison, Tzanopoulos, Matsuzawa & Humle (2017) in Bossou, 15 16 191 Republic of Guinea, our classification of "primary forest" encompasses mature and 17 18 192 riverine forests, our "secondary forest" category includes young secondary forests 19 For Peer Review 20 193 and our "fallow" class corresponds to all types of highly disturbed habitats they 21 22 23 194 identified in their study. 24 25 195 Chimpanzees tend to build nests in groups (Ogawa, Idani, Moore, Pintea & 26 27 196 Hernandez-Aguilar, 2007). As recommended by Buckland et al. (2001), we 28 29 30 197 considered clusters of nests as our observation unit instead of individual nests. To 31 32 198 create clusters, we grouped nests with the same age class that were within 20 33 34 199 meters of each other post hoc. Some studies have used thresholds of 50 meters 35 36 37 200 (e.g., Morgan & Sanz, 2006; Sousa et al., 2011), but based on our observations in 38 39 201 the field we decided to choose 20 meters to reduce the risk of grouping different 40 41 202 clusters together (see Marchesi, Marchesi, Fruth & Boesch 1995, Ogawa et al. 42 43 203 2007, Kouakou, Boesch, & Kuehl 2009). 44 45 46 204 Since chimpanzees show marked preferences for nesting sites (Carvalho, 47 48 205 Meyer, Vicente & Marques, 2015; Bryson-Morrison et al., 2017), we used Density 49 50 206 Surface Modelling (DSM) to model the abundance of clusters of nests (Hedley & 51 52 53 207 Buckland, 2004; Miller, Burt, Rexstad & Thomas, 2013) as a function of 54 55 208 environmental covariates that include topographic variables, distance to rivers, 56 57 58 59 60 John Wiley & Sons Page 11 of 56 American Journal of Primatology

1 2 3 4 209 roads and villages, percentage of cover of different land uses and Shannon-Wiener 5 6 7 210 land-use diversity (Table 1). Each of the transects was split into five 200 meter 8 9 211 segments for modelling. This is a two-stage approach that involves 1) fitting a 10 11 212 detection function to the clusters of nests and using it to estimate abundances in 12 13 14 213 transect segments with a Horvitz–Thompson-like estimator (Borchers, Buckland, 15 16 214 Goedhart, Clarke, & Hedley, 1998) and 2) building a generalized additive model 17 18 215 (Wood, 2017) to model estimated cluster abundances per transect segment as a 19 For Peer Review 20 216 function of environmental covariates. 21 22 23 217 We fitted uniform, half-normal and hazard-rate detection functions and included 24 25 218 observation-level covariates that may have affected nest detection, such as nest 26 27 219 cluster size, mean nest age class and land use cover (savanna, primary forest, 28 29 30 220 secondary forest or fallows). In dense forests and areas with dense understory, 31 32 221 nest detection can be lower. Observed distances were truncated at 50 meters 33 34 222 based on the visual inspection of the detection function superimposed on a 35 36 37 223 histogram of distances (Buckland et al., 2001) (Appendix 1). The goodness of fit of 38 39 224 each detection function was assessed with the Cramer-von Mises test and the 40 41 225 Kolmogorov-Smirnov test (Buckland et al., 2004). The best detection function was 42 43 226 selected using the Akaike's Information Criteria (AIC). All calculations were 44 45 46 227 performed in R 3.6 (R Core Team, 2019) using the package "Distance" version 47 48 228 0.9.8 (Miller, Rexstad, Thomas, Marshall & Laake, 2016). 49 50 229 We used Generalized Additive Models (GAMs) to model the abundance of 51 52 53 230 clusters of nests. The expected abundance in each segment was modeled with 54 55 231 Tweedie or negative binomial distribution as a function of several covariates. 56 57 58 59 60 John Wiley & Sons American Journal of Primatology Page 12 of 56

1 2 3 4 232 GAMs were fitted with the R package "dsm" version 2.2.17 (Miller et al., 2013). 5 6 7 233 Thin plate regression splines (Wood, 2003) were used as the basis for the model's 8 9 234 smooth terms. The model is initiated by considering that the fit is extremely wiggly. 10 11 235 Then the fitting procedure induces a penalization that essentially means the final 12 13 14 236 wigglyness is driven by the data. (Wood, 2017). To minimize the effects of 15 16 237 correlation among covariates, we considered only those variables with an 17 18 238 individually significant association (p<0.05) with nest cluster abundance. 19 For Peer Review 20 239 Furthermore, we calculated variance inflation factors (VIF; Fox & Weisberg, 2010) 21 22 23 240 and eliminated covariates with a VIF > 3. After fitting the model with all variables, 24 25 241 we removed non-significant terms to reduce concurvity. Smoothness selection was 26 27 242 performed via restricted maximum likelihood (REML). Smooth terms were selected 28 29 30 243 using approximate p-values (p<0.05) and by adding an additional penalty that 31 32 244 allowed each smooth term to be removed during model fitting (Marra and Wood, 33 34 245 2011). Spatial autocorrelation was assessed by examining a correlogram of 35 36 37 246 deviance residuals. To validate the final models, we analyzed deviance residuals 38 39 247 and checked for normal distribution and constant variance (Wood, 2017). To 40 41 248 calculate the density of chimpanzees we divided the estimated nest density by the 42 43 249 nest production rate and nest decay rate (Plumptre, 2003), following a formula 44 45 46 250 modified after Kühl, Maisels, Ancrenaz & Williamson (2008): 47 D�� 48 251 D = 49 ��ℎ�� × t 50 51 252 Where r is the estimated rate of nest production per individual per day and t is the 52 53 253 estimated mean life of a nest. Both values can be calculated only by performing 54 55 56 57 58 59 60 John Wiley & Sons Page 13 of 56 American Journal of Primatology

1 2 3 4 254 detailed field studies and may vary between populations and geographic areas. 5 6 7 255 Because of time constraints, we could not estimate these parameters in our study 8 9 256 area, so we used estimates from other studies. For r we used 1.09 nests/day per 10 11 257 individual from Plumptre & Reynolds (1997) in Budongo Forest Reserve, Uganda. 12 13 14 258 For t we chose 194 days from Fleury-Brugiere & Brugiere (2010) in the Haut Niger 15 16 259 National Park, Republic of Guinea. This estimate was considered the most suitable 17 18 260 given the proximity to our study area and similarities in climate and vegetation. 19 For Peer Review 20 261 Unfortunately, these studies did not provide the variances for these parameters. 21 22 23 262 Therefore the variances of chimpanzee densities will be underestimated. 24 25 263 To assess the potential impacts of the construction of the mine on 26 27 264 chimpanzees, we used the density surface model to calculate the predicted 28 29 30 265 abundance of nests in the study area. We combined uncertainty from the spatial 31 32 266 model (GAM) with that of detectability (detection function) using the delta method 33 34 267 (assuming independence between these two components) using "dsm.var.gam" 35 36 37 268 from the R package "dsm" (Miller et al 2013). Finally, we analyzed the overlap 38 39 269 between the bauxite deposit and the areas where the model predicts higher 40 41 270 abundances of nests. 42 43 271 44 45 46 272 3. Results 47 48 273 We counted 608 nests during the surveys, which we grouped in 116 clusters. 49 50 274 The number of nests per cluster averaged 5.2 ± SD 6.7. 51 52 53 275 54 55 276 56 57 58 59 60 John Wiley & Sons American Journal of Primatology Page 14 of 56

1 2 3 4 277 Detection function 5 6 7 278 We selected a hazard-rate key function with cluster size as a covariate by AIC. 8 9 279 The truncation distance for the detection function was 50 m and selected by 10 11 280 comparing test statistics from the Cramer–von Mises and Kolmogorov–Smirnov 12 13 14 281 goodness of fit tests. The average detection probability was 0.534, and the 15 16 282 coefficient of variation was 0.068 (Fig. 2). A complete comparison of the detection 17 18 283 functions can be found in the Supplementary Information (Table S1), along with all 19 For Peer Review 20 284 the R code required to reproduce our results. Figure 2 shows relatively few 21 22 23 285 detections close to the transect, which was caused by lower detectability of nests 24 25 286 in areas with dense forest or dense understorey. This did not have important 26 27 287 effects on the fit of the detection function. 28 29 30 288 Density surface models 31 32 289 The density surface model with a Tweedie distribution provided the best fit for 33 34 290 the data (see quantile-quantile plot, Fig. 3). The abundance of clusters of nests 35 36 37 291 was higher in areas with a northwest exposure, closer to seasonal rivers, in areas 38 39 292 with a low cover of savanna and with a high Shannon-Wiener diversity of land uses 40 41 293 (Fig. 4). 42 43 294 Estimated abundance of nests and chimpanzees 44 45 46 295 The model predicted the occurrence of 3878 nests in the study area. The 47 48 296 coefficient of variation from the GAM was 0.2481, and the coefficient of variation of 49 50 297 the detection function 0.1271. The total coefficient of variation for the estimate was 51 52 53 298 0.2788 (calculated using the delta method). Following Equation 1, the estimated 54 55 299 abundance of nest building chimpanzees in Ronde hill is N = 18 (95% CI: 11-31). 56 57 58 59 60 John Wiley & Sons Page 15 of 56 American Journal of Primatology

1 2 3 4 300 This estimate corresponds to a density of 0.3898 individuals/km2 (95% CI: 0.2280– 5 6 7 301 0.6664). 8 9 302 The overlap between chimpanzees' nests and the proposed mine 10 11 303 Predicted abundances of nests are not very high (< 20 nests) at the top of 12 13 14 304 Ronde hill, where the mine is going to be built (there is some overlap in the 15 16 305 northwestern part) (Fig. 5). The overlap between areas with a high predicted 17 18 306 abundance of nests (>40 nests/km2) and the future area of the mine is 0.2 km2. 19 For Peer Review 20 307 21 22 23 308 4. Discussion 24 25 309 In this study, we estimated the distribution and abundance of chimpanzees 26 27 310 with the standing crop nest counts method and compared it with the future location 28 29 30 311 of a bauxite mine. Overall, the predicted abundances of nests in location of the 31 32 312 mine were relatively low, which can probably be explained by the fact that the top 33 34 313 of Ronde Hill is covered by savanna and devoid of suitable trees for building nests. 35 36 37 314 Still, the northeastern part of the mine coincides with an area of high observed and 38 39 315 predicted nest density (>40 nests/km2), that also contains the only accessible year- 40 41 316 round source of water in a 2 kilometer radius. This area is probably an essential 42 43 317 refuge for western chimpanzees, which are already suffering from habitat loss due 44 45 46 318 to agricultural pressure from the neighboring village of Capebonde. 47 48 319 We estimated the total abundance of nest building chimpanzees in the study 49 50 320 areas was 18 (95% CI: 11-31), corresponding to 0.3898 individuals/km2 (95% CI: 51 52 53 321 0.2280–0.6664). Camera traps active during fieldwork placed in the valley of the 54 55 322 Jabere river during identified at least 18 weaned chimpanzees (JFCW et al. 56 57 58 59 60 John Wiley & Sons American Journal of Primatology Page 16 of 56

1 2 3 4 323 unpublished data). Our estimate is within the range of estimates obtained in other 5 6 7 324 studies that also used the standing crop nest counts method. In Senegal, Pruetz et 8 9 325 al. (2002) estimated 0.13 individuals/km2, in the Republic of Guinea Fleury- 10 11 326 Brugiere & Brugiere (2010) estimated 0.87 individuals/km2 (95% CI: 0.73 – 1.04) 12 13 14 327 and in Lagoas de Cufada Natural Park in Guinea-Bissau Carvalho et al. (2013) 15 16 328 found 0.22 individuals/km2 (95% CI: 0.08 – 0.62). 17 18 329 The density surface model suggests that chimpanzees prefer to build nests in 19 For Peer Review 20 330 areas facing northeast, with higher Shannon-Wiener land use diversity, with low 21 22 23 331 cover by savanna, and close to seasonal rivers. These results are in line with the 24 25 332 findings from other studies, which suggest that western chimpanzees can tolerate 26 27 333 some human disturbance (Brugiere et al., 2009; Bryson-Morrison et al., 2017) and 28 29 30 334 inhabit mosaics containing savanna, riparian forests, dense forests and more open 31 32 335 habitats (Carvalho et al., 2013, 2015). In Lagoas de Cufada Natural Park (Guinea- 33 34 336 Bissau), Carvalho et al. (2015) found that chimpanzees prefer to build nests in 35 36 37 337 dense forests, contrary to our findings. Dense forests in Ronde hill are often close 38 39 338 to frequently used agricultural areas which are avoided by chimpanzees. This type 40 41 339 of avoidance behavior has also been observed in the Republic of Guinea (Bryson- 42 43 340 Morrison et al., 2017). 44 45 46 341 Because of logistical constraints, we could conduct only one survey. We 47 48 342 suggest that future research in the study area should focus on analyzing how 49 50 343 chimpanzees use habitats throughout the year. It would also be useful to determine 51 52 53 344 whether the chimpanzees that occur in Ronde Hill are part of one or several 54 55 345 communities, and whether these communities are connected to those in the 56 57 58 59 60 John Wiley & Sons Page 17 of 56 American Journal of Primatology

1 2 3 4 346 Republic of Guinea. This information would allow us to better understand and 5 6 7 347 prevent the possible impacts of the construction of the mine on this population of 8 9 348 western chimpanzees. 10 11 349 12 13 14 350 5. Conclusion 15 16 351 The results of the study show that only a small part of the proposed mine 17 18 352 coincides with areas of high chimpanzee's nests abundance. This small area of 19 For Peer Review 20 353 overlap presents one of the highest abundances of nests in the whole study area 21 22 2 23 354 (>40 nests/km ). In the remaining area around the mine, predicted nest densities 24 25 355 are low, which probably reflects the fact that it is currently covered by grassland 26 27 356 savanna and does not contain trees suitable for building nests. The projected 28 29 30 357 location of the mine borders two areas of high abundance of chimpanzee's nests 31 32 358 (northeast and southwest), therefore it is likely to be used by chimpanzees. The 33 34 359 data we gathered, combined with the existing knowledge on impacts of mining on 35 36 37 360 great ape populations, suggests the construction of the mine is likely to have 38 39 361 significant direct and indirect effects on this population of chimpanzees. We 40 41 362 recommend that if the mine is approved, authorities should carefully consider direct 42 43 363 and indirect impacts on this population of chimpanzees and implement appropriate 44 45 46 364 mitigation and compensation measures. 47 48 365 49 50 366 51 52 53 367 54 55 368 56 57 58 59 60 John Wiley & Sons American Journal of Primatology Page 18 of 56

1 2 3 4 369 6. Acknowledgments 5 6 7 370 We thank CHIMBO Foundation, especially the director Annemarie 8 9 371 Goedmakers and the board advisor Piet Wit for providing logistical and financial 10 11 372 support for this survey and for their suggestions that significantly improved this 12 13 14 373 work. We further thank the five CHIMBO's village committee members of 15 16 374 Capebonde (Amadou Camará, Mangabói Culubali, Mari Cante, Boibalo Bangura, 17 18 375 and Ali Camará) for their help in data collection and colleagues Jitske Willemsen 19 For Peer Review 20 376 and Menno de Boer for their company and support during the field survey. We are 21 22 23 377 also extremely thankful to Bauxite Angola S.A. for providing us with their 24 25 378 prospecting data and for the meeting with JFCW to discuss possible outcomes of 26 27 379 this research. This work was approved by the Instituto da Biodiversidade e das 28 29 30 380 Áreas Protegidas (IBAP) of Guinea-Bissau and complied with the Principles for the 31 32 381 Ethical Treatment of Non-Human Primates of the American Society of 33 34 382 Primatologists (ASP). JFCW was supported by Ciências sem Fronteiras 35 36 37 383 scholarship from Conselho Nacional de Desenvolvimento Científico e Tecnológico 38 39 384 (CNPq), Brazil (grant number 221350/2012-8). FSD was funded by FEDER funds 40 41 385 through the Operational Programme for Competitiveness Factors - COMPETE and 42 43 386 by National Funds through FCT - Foundation for Science and Technology under 44 45 46 387 the UID/BIA/50027/2013 and POCI-01-0145-FEDER-006821. TAM thanks partial 47 48 388 support by CEAUL (funded by FCT - Fundação para a Ciência e a Tecnologia, 49 50 389 Portugal, through the project UID/MAT/00006/2019). We also wish to thank the 51 52 53 390 editor and the anonymous reviewers for their comments that significantly improved 54 55 391 this manuscript. 56 57 58 59 60 John Wiley & Sons Page 19 of 56 American Journal of Primatology

1 2 3 4 392 7. References 5 6 7 393 Arcus Foundation. (2014). Extractive industries and ape conservation (state of the 8 9 394 Apes). https://doi.org/10.1017/CBO9781107590274 10 11 395 Borchers, D. L., Buckland, S. T., Goedhart, P. W., Clarke, E. D., & Hedley, S. L. 12 13 14 396 (1998). Horvitz-Thompson estimators for double-platform line transect 15 16 397 surveys. Biometrics, 54(4), 1221–1237. 17 18 398 Brugiere, D., Badjinca, I., Silva, C., & Serra, A. (2009). Distribution of chimpanzees 19 For Peer Review 20 399 and interactions with humans in Guinea-Bissau and Western Guinea, West 21 22 23 400 Africa. Folia Primatologica, 80, 353–358. 24 25 401 https://doi.org/10.1159/000259335 26 27 402 Bryson-Morrison, N., Tzanopoulos, J., Matsuzawa, T., & Humle, T. (2017). Activity 28 29 30 403 and habitat use of chimpanzees (Pan troglodytes verus) in the 31 32 404 anthropogenic landscape of Bossou, Guinea, West Africa. International 33 34 405 Journal of Primatology, 38(2), 282–302. http://doi.org/10.1007/s10764-016- 35 36 37 406 9947-4 38 39 407 Buckland, S. T., Anderson, D. R., Burnham, K. P., Laake, J. L., Borchers, D. L., & 40 41 408 Thomas, L. (2001). Introduction to Distance Sampling: Estimating 42 43 409 Abundance of Biological Populations. New York: Oxford University Press. 44 45 46 47 48 410 Buckland, S. T., Anderson, D. R., Burnham, K. P., Laake, J. L., Borchers, D. L., & 49 50 411 Thomas, L. (2004). Advanced Distance Sampling: Estimating Abundance of 51 52 412 Biological Population. New York: Oxford University Press. 53 413 54 55 56 57 58 59 60 John Wiley & Sons American Journal of Primatology Page 20 of 56

1 2 3 4 414 Campbell, G., Kuehl, H., N’Goran Kouamé, P., & Boesch, C. (2008). Alarming 5 6 7 415 decline of West African chimpanzees in Côte d’Ivoire. Current Biology, 8 9 416 18(19), R903–R904. 10 11 417 http://doi.org/https://doi.org/10.1016/j.cub.2008.08.015 12 13 14 418 Carvalho, J. S., Marques, T. A., & Vicente, L. (2013). Population status of Pan 15 16 419 troglodytes verus in Lagoas de Cufada Natural Park, Guinea-Bissau. PloS 17 18 420 One, 8(8), e71527. http://doi.org/10.1371/journal.pone.0071527 19 For Peer Review 20 421 Carvalho, J. S., Meyer, C. F. J., Vicente, L., & Marques, T. A. (2015). Where to 21 22 23 422 nest? Ecological determinants of chimpanzee nest abundance and 24 25 423 distribution at the habitat and tree species scale. American Journal of 26 27 424 Primatology, 77(2), 186–199. http://doi.org/10.1002/ajp.22321 28 29 30 425 Danquah, E., Oppong, S., Akom, E., & Sam, M. (2012). Preliminary survey of 31 32 426 chimpanzees and threatened monkeys in the Bia-Goaso Forest Block in 33 34 427 Southwestern Ghana. African Primates. African Primates 7(2), 163-174. 35 36 37 428 https://doi.org/10.1017/S0030605313001191 38 39 429 Devos, L., Sanz, C., Morgan, D., Onononga, J., & Laporte, N. (2008). Comparing 40 41 430 ape densities and habitats in Northern Congo: surveys of sympatric gorillas 42 43 431 and chimpanzees in the Odzala and Ndoki regions, American Journal of 44 45 46 432 Primatology 13, 1–13. http://doi.org/10.1002/ajp.20514 47 48 433 Diallo, M.S. (2010). Praticas das indústrias extractivas na Africa ocidental, Síntese 49 50 434 comparativa de quatro estudos de casos (Guiné-Bissau, Guinée, Senegal, 51 52 53 435 Serra Leoa). Gland, Switzerland: IUCN, Dakar, Senegal: ASAN, Bissau, 54 55 56 57 58 59 60 John Wiley & Sons Page 21 of 56 American Journal of Primatology

1 2 3 4 436 Guinea-Bissau: AD, Conakry, Guinea: Guinée ecologie, Freetown, Sierra 5 6 7 437 Leone: EFA. 8 9 438 ESRI 2011. ArcGIS Desktop: Release 10. Redlands, CA: Environmental Systems 10 11 439 Research Institute. 12 13 14 440 Fleury-Brugiere, M.-C., & Brugiere, D. (2010). High population density of Pan 15 16 441 troglodytes verus in the Haut Niger National Park, Republic of Guinea: 17 18 442 Implications for local and regional conservation. International Journal of 19 For Peer Review 20 443 Primatology, 31(3), 383–392. http://doi.org/10.1007/s10764-010-9391-9 21 22 23 444 Fox, J., & Weisberg, H. S. (2010). An R companion to applied regression (2nd 24 25 445 ed.). Thousand Oaks, CA: SAGE Publications. 26 27 446 Ginn, L., Robison, J., Redmond, I., & Nekaris, K. (2013). Strong evidence that the 28 29 30 447 West African chimpanzee is extirpated from Burkina Faso. Oryx, 47(3), 31 32 448 325-326. doi:10.1017/S0030605313000434 33 34 449 Gippoliti, S., Embalo, D. S., Sousa, C. (2003). Guinea-Bissau. In: R. Kormos, C. 35 36 37 450 Boesch, M. I. Bakarr, & T. M. Butynski (Eds.), West african chimpanzees. 38 39 451 status survey and conservation action plan (pp. 55–61). Gland, Switzerland 40 41 452 and Cambridge, UK: IUCN/SSC Primate Specialist Group (PSG). 42 43 453 Hedley, S. L., & Buckland, S. T. (2004). Spatial models for line transect sampling. 44 45 46 454 Journal of Agricultural, Biological, and Environmental Statistics, 9(2), 181. 47 48 455 http://doi.org/10.1198/1085711043578 49 50 456 Hockings, K., & Humle, T. (2009). Best practice guidelines for the prevention and 51 52 53 457 mitigation of conflict between humans and great apes. Gland, Switzerland: 54 55 458 IUCN/SSC Primate Specialist Group (PSG). 56 57 58 59 60 John Wiley & Sons American Journal of Primatology Page 22 of 56

1 2 3 4 459 Humle, T., Boesch, C., Campbell, G., Junker, J., Koops, K., Kuehl, H. & Sop, T. 5 6 7 460 2016. Pan troglodytes ssp. verus (errata version published in 2016). The 8 9 461 IUCN Red List of Threatened Species 2016: 10 11 462 e.T15935A102327574. http://dx.doi.org/10.2305/IUCN.UK.2016- 12 13 14 463 2.RLTS.T15935A17989872.en. 15 16 464 Köndgen, S., Kühl, H., N’Goran, P. K., Walsh, P. D., Schenk, S., Ernst, N., … 17 18 465 Leendertz, F. H. (2008). Pandemic human viruses cause decline of 19 For Peer Review 20 466 endangered great apes. Current Biology, 18(4), 260–264. 21 22 23 467 https://doi.org/10.1016/j.cub.2008.01.012 24 25 468 Kormos, R., & Bakarr, M.I. (2003). Regional Summary. In: R. Kormos, C. Boesch, 26 27 469 M. I. Bakarr, & T. M. Butynski (Eds.), West african chimpanzees. status 28 29 30 470 survey and conservation action plan (pp. 27–29). Gland, Switzerland and 31 32 471 Cambridge, UK: IUCN/SSC Primate Specialist Group (PSG). 33 34 472 Kouakou, C. Y., Boesch, C., & Kuehl, H. (2009). Estimating chimpanzee population 35 36 37 473 size with nest counts: validating methods in Taï National Park. American 38 39 474 Journal of Primatology, 71(6), 447–57. http://doi.org/10.1002/ajp.20673 40 41 475 Kühl, H. S., Maisels, F., Ancrenaz, M., & Williamson, E. A. (2008). Best practice 42 43 476 guidelines for surveys and monitoring of great ape populations. Gland, 44 45 46 477 Switzerland: IUCN/SSC Primate Specialist Group (PSG). 47 48 478 Kühl, H. S., Kalan, A. K., Arandjelovic, M., Aubert, F., D’Auvergne, L., 49 50 479 Goedmakers, A., … Boesch, C. (2016). Chimpanzee accumulative stone 51 52 53 480 throwing. Scientific Reports, 6, 22219. http://doi.org/10.1038/srep22219 54 55 56 57 58 59 60 John Wiley & Sons Page 23 of 56 American Journal of Primatology

1 2 3 4 481 Kühl, H. S., Sop, T., Williamson, E. A., Mundry, R., Brugière, D., Campbell, G., … 5 6 7 482 Boesch, C. (2017). The Critically Endangered western chimpanzee 8 9 483 declines by 80%. American Journal of Primatology, 79(9). 10 11 484 https://doi.org/10.1002/ajp.22681 12 13 14 485 Kusin, F. M., Rahman, M. S. A., Madzin, Z., Jusop, S., Mohamat-Yusuff, F., Ariffin, 15 16 486 M., & Z, M. S. M. (2017). The occurrence and potential ecological risk 17 18 487 assessment of bauxite mine-impacted water and sediments in Kuantan, 19 For Peer Review 20 488 Pahang,Malaysia. Environmental Science and Pollution Research, 24(2), 21 22 23 489 1306–1321. http://doi.org/10.1007/s11356-016-7814-7 24 25 490 Laurance, W. F., Croes, B. M., Tchignoumba, L., Lahm, S. A., Alonso, A. Lee, M. 26 27 491 E. & Ondzeano, C. (2006). Impacts of roads and hunting on central African 28 29 30 492 rainforest mammals. Conservation Biology, 20, 1251–1261. 31 32 493 https://doi.org/10.1111/j.1523-1739.2006.00420.x 33 34 494 Marchesi, P., Marchesi, N., Fruth, B., & Boesch, C. (1995). Census and distribution 35 36 37 495 of chimpanzees in Côte d’Ivoire, Primates, 36(4), 591 – 607. 38 39 496 https://doi.org/10.1007/BF02382880 40 41 497 Marra, G., & Wood, S. N. (2011). Practical variable selection for generalized 42 43 498 additive models. Computational Statistics & Data Analysis, 55(7), 2372– 44 45 46 499 2387. http://doi.org/https://doi.org/10.1016/j.csda.2011.02.004 47 48 500 Mensah, A. K., Mahiri, I. O., Owusu, O., Mireku, O. D., Wireko, I., & Kissi, E. A. 49 50 501 (2015). Environmental impacts of mining: a study of mining communities in 51 52 53 502 Ghana. Applied Ecology and Environmental Sciences, 3(3), 81–94. 54 55 503 http://doi.org/10.12691/aees-3-3-3 56 57 58 59 60 John Wiley & Sons American Journal of Primatology Page 24 of 56

1 2 3 4 504 Miller, D. L., Rexstad, E., Burt, L., Bravington, M. V., & Hedley, S. (2019). dsm: 5 6 7 505 Density surface modelling of distance sampling data (Version 2.2.17). 8 9 506 Retrieved from https://cran.r-project.org/web/packages/dsm/index.html 10 11 507 Miller, D. L., Burt, M. L., Rexstad, E. A., & Thomas, L. (2013). Spatial models for 12 13 14 508 distance sampling data: recent developments and future directions. 15 16 509 Methods in Ecology and Evolution, 4, 1001–1010. 17 18 510 https://doi.org/10.1111/2041-210X.12105 19 For Peer Review 20 511 Miller, D. L., Rexstad, E., Thomas, L., Marshall, L., & Laake, J. L. (2016). Distance 21 22 23 512 sampling in R. Journal of Statistical Software. 24 25 513 http://dx.doi.org/10.1101/063891. 26 27 514 Morgan, D., & Sanz, C. (2007). Best practice guidelines for reducing the impact of 28 29 30 515 commercial logging on great apes in Western Equatorial Africa. Occasional 31 32 516 paper of the IUCN Species Survival Commission, No 34, IUCN 33 34 517 Ogawa, H., Idani, G., Moore, J., Pintea, L., & Hernandez-Aguilar, A. (2007). 35 36 37 518 Sleeping parties and nest distribution of chimpanzees in the savanna 38 39 519 woodland, Ugalla, Tanzania. International Journal of Primatology, 28(6), 40 41 520 1397–1412. http://doi.org/10.1007/s10764-007-9210-0 42 43 521 Plumptre, A. J. (2003). Censuing chimpanzees. In: A. J. Plumptre, D. Cox, & S. 44 45 46 522 Mugume, (Eds.), The status of chimpanzees in Uganda. Albertine Rift 47 48 523 Technical Report Series No. 2. Wildlife Conservation Society, New York. 49 50 524 R Core Team (2017). R: a language and environment for statistical computing. 51 52 53 525 Vienna, Austria: R Foundation for Statistical Computing. 54 55 56 57 58 59 60 John Wiley & Sons Page 25 of 56 American Journal of Primatology

1 2 3 4 526 Serra, A., Silva, C., & Lopes, E. (2007). Étude de faisabilité du projet 5 6 7 527 “Développement touristique de la Boé au profit de la conservation des 8 9 528 Chimpanzés et des populations locales”. Bissau, Guinea-Bissau: WWF, 10 11 529 Chimbo Foundation, IUCN and IBAP. 12 13 14 530 Sousa, J., Barata, A. V, Sousa, C., Casanova, C. C. N., & Vicente, L. (2011). 15 16 531 Chimpanzee oil-palm use in southern Cantanhez National Park, Guinea- 17 18 532 Bissau. American Journal of Primatology, 73(5), 485–97. 19 For Peer Review 20 533 http://doi.org/10.1002/ajp.20926 21 22 23 534 Spehar, S. N., Mathewson, P. D., Wich, S. A., Marshall, A. J., Kühl, H., & Meijaard, 24 25 535 E. (2010). Estimating orangutan densities using the standing srop and 26 27 536 marked nest count methods: lessons learned for conservation. Biotropica, 28 29 30 537 42(6), 748–757. https://doi.org/10.1111/j.1744-7429.2010.00651.x 31 32 538 Wit, P. (2011). Démarrage des activités d’exploitation de bauxite dans la Boé par 33 34 539 l’entreprise “Bauxite Angola”, Daridibó-Chimbo, Bissau. 35 36 37 540 Wit, P., & Reintjes, H. C. (1989). An agro-ecological survey of the Boé province, 38 39 541 Guinea Bissau. Agriculture, Ecosystems & Environment, 27(1), 609–620. 40 41 542 http://doi.org/https://doi.org/10.1016/0167-8809(89)90121-7 42 43 543 Wood, S. N. (2003). Thin plate regression splines. Journal of the Royal Statistical 44 45 46 544 Society: Series B, 65(1), 95-114. https://doi.org/10.1111/1467-9868.00374 47 48 545 Wood, S. N. (2017). Generalized Additive Models: An Introduction with R (2nd ed.). 49 50 546 Boca Raton, FL: Chapman and Hall/CRC. 51 52 53 547 54 55 548 56 57 58 59 60 John Wiley & Sons American Journal of Primatology Page 26 of 56

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 For Peer Review 20 21 22 23 24 25 26 27 28 29 30 549 31 32 33 550 34 35 551 Figure 1 - Map showing the study area including the location of Ronde hill, the 36 37 552 future location of the mine, transects, nest clusters, roads, rivers and closest 38 39 40 553 villages. The top inset shows the location of Guinea-Bissau and the bottom inset 41 42 554 the location of the study area in this country. 43 44 555 45 46 47 556 48 49 557 50 51 52 53 54 55 56 57 58 59 60 John Wiley & Sons Page 27 of 56 American Journal of Primatology

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 For Peer Review 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 558 35 36 37 559 Figure 2 - Selected detection function (hazard-rate with cluster size as covariate) 38 39 560 for clusters of nests overlaid onto a histogram of observed distances. 40 41 561 42 43 562 44 45 46 563 47 48 564 49 50 565 51 52 53 566 54 55 567 56 57 58 59 60 John Wiley & Sons American Journal of Primatology Page 28 of 56

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 For Peer Review 20 21 22 23 24 25 568 26 27 569 Figure 3 - Comparison of models with Tweedie (left) and negative binomial (right) 28 29 30 570 response distributions by quantile-quantile plots. Good fit is indicated by agreement 31 32 571 between observed and fitted (residual) quantiles (i.e., points being close to the red 33 34 572 line). 90% reference bands are shown in grey allowing judgment of the deviation 35 36 37 573 from the line. The negative binomial points fall further away from the red line than 38 39 574 those for the Tweedie, indicating model misspecification. 40 41 575 42 43 576 44 45 46 577 47 48 578 49 50 579 51 52 53 54 55 56 57 58 59 60 John Wiley & Sons Page 29 of 56 American Journal of Primatology

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 For Peer Review 20 21 22 23 24 25 26 27 28 29 30 31 32 33 580 34 35 581 Figure 4 – Smooth functions for “aspect”, “distance to closest seasonal river”, 36 37 38 582 “savanna and “Shannon-Wiener” land use diversity. Grey shading corresponds to 39 40 583 95% confidence bands, numbers in brackets on the vertical axis labels give the 41 42 584 effective degrees of freedom of the term (1 corresponds to a linear term). 43 44 45 585 46 47 586 48 49 587 50 51 52 53 54 55 56 57 58 59 60 John Wiley & Sons American Journal of Primatology Page 30 of 56

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 For Peer Review 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 588 36 37 38 589 Figure 5 – Predicted abundance of nests overlaid with the location of the transects 39 40 590 (black lines), location of clusters of nests (red dots) and future location of the mine 41 42 591 (pink line). Ronde hill is shown by the green line and rivers are shown by blue 43 44 592 lines. 45 46 47 593 48 49 594 50 51 595 52 53 54 596 55 56 57 58 59 60 John Wiley & Sons Page 31 of 56 American Journal of Primatology

1 2 3 4 597 Table 1 - Covariates used in the spatial model (GAM). 5 6 Variables Description and Units Source 7 8 Slope Mean slope (degrees) ASTER GDEM 2.0 9 10 Aspect Mean aspect (radians) ASTER GDEM 2.0 11 Altitude Mean altitude (m) ASTER GDEM 2.0 12 13 Distance to closest permanent river Distance (m) JFCW 14 Distance to closest seasonal river Distance (m) JFCW 15 16 Distance to closest village Distance to the centroid of JFCW 17 the closest village (m) 18 19 Distance to closest roadFor PeerDistance Review (m) JFCW 20 Agriculture Area (ha) JFCW 21 22 Urban Area (ha) JFCW 23 24 Primary Forest Area (ha) JFCW 25 Secondary Forest Area (ha) JFCW 26 27 Savanna Area (ha) JFCW 28 29 Land use diversity Shannon-Wiener diversity - 30 Index 31 598 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 John Wiley & Sons American Journal of Primatology Page 32 of 56

1 2 Research Highlights 3 4 5 6 7 • Approximately 18 nest building western chimpanzees inhabit the surroundings of a 8 9 bauxite deposit in the SW of Guinea-Bissau; 10 11 12 13 14 • The construction of a mine can have adverse direct and indirect effects on this 15 16 population. 17 18 19 For Peer Review 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 John Wiley & Sons Page 33 of 56 American Journal of Primatology

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 For Peer Review 20 21 22 23 24 25 26 27 28 29 Figure 1 30 31 296x209mm (300 x 300 DPI) 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 John Wiley & Sons American Journal of Primatology Page 34 of 56

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 For Peer Review 20 21 22 23 24 25 26 27 28 29 30 Figure 2 31 32 396x285mm (72 x 72 DPI) 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 John Wiley & Sons Page 35 of 56 American Journal of Primatology

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 For Peer Review 20 21 22 23 24 25 Figure 3 26 27 381x213mm (96 x 96 DPI) 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 John Wiley & Sons American Journal of Primatology Page 36 of 56

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 For Peer Review 20 21 22 23 24 25 26 27 28 29 Figure 4 30 31 425x293mm (72 x 72 DPI) 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 John Wiley & Sons Page 37 of 56 American Journal of Primatology

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 For Peer Review 20 21 22 23 24 25 26 27 28 29 30 31 Figure 5 32 423x313mm (72 x 72 DPI) 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 John Wiley & Sons American Journal of Primatology Page 38 of 56

1 2 3 4 Supplementary information from ‘Density and 5 6 distribution of western chimpanzees around a bauxite 7 8 9 deposit in the Boé Sector, Guinea-Bissau’ 10 José F. C. Wenceslau 11 Filipe S. Dias 12 Tiago A. Marques 13 David L. Miller 14 15 16 17 1. Introduction 18 In this document we present the R code we used to generate the results we present and discuss in “Density 19 and distribution of western chimpanzeesFor Peer around a bauxite Review deposit in the Boé Sector, Guinea-Bissau.” 20 21 22 2. Load required packages 23 24 library(ggplot2) 25 library(gridExtra) 26 library(knitr) 27 library(mrds) 28 library(Distance) 29 library(dsm) 30 library(tweedie) 31 library(vegan) 32 library(viridis) 33 library(usdm) 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 1 60 John Wiley & Sons Page 39 of 56 American Journal of Primatology

1 2 3 2. Exploratory data analysis 4 5 2.1 Histogram of observed distances 6 7 8 9 Histogram of observed distances 10 11 12 13 14 15 16 17 18 19 For Peer Review 20 21 Frequency 22 23 24 25 26 27 0 2 4 6 8 10 12 28 29 0 20 40 60 80 30 31 distances 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 2 60 John Wiley & Sons American Journal of Primatology Page 40 of 56

1 2 3 2.2 Do observed distances change as function of covariates? 4 5 Observed distances vs Decay Observed distances vs Stratum 6 7 8 75 75 9 10 50 50 11 12

Distances 25 Distances 25 13 14 15 0 0 16 1234 Fallow Forest Savannah Secondary 17 Decay Stratum 18 19 Observed distancesFor vs ClusterPeer size Review 20 21 75 22 23 50 24 25 25 26 Distances 27 28 0 29 0 10 20 30 40 50 30 Cluster size 31 32 33 34 4. Fit detection functions 35 36 4.1 Conventional distance sampling (CDS) 37 38 df1c<-ds(data_scnc_clus, truncation=50, key = "unif", adjustment="cos",order=c(2)) 39 df2c<-ds(data_scnc_clus, truncation=50, key = "hn", adjustment="cos",order=c(2)) 40 df3c<-ds(data_scnc_clus, truncation=50, key = "hn", adjustment="herm") 41 df4c<-ds(data_scnc_clus, truncation=50, key = "hr", adjustment="poly") 42 43 4.2 Multiple-covariate distance sampling (MCDS) 44 45 #Hazard-rate function 46 df5c<-ds(data_scnc_clus, truncation=50, key="hr",formula=~size) 47 df6c<-ds(data_scnc_clus, truncation=50, key="hr",formula=~stratum) 48 df7c<-ds(data_scnc_clus, truncation=50, key="hr",formula=~decay) 49 df8c<-ds(data_scnc_clus, truncation=50, key="hr",formula=~size+stratum) 50 df9c<-ds(data_scnc_clus, truncation=50, key="hr",formula=~decay+stratum) 51 df10c<-ds(data_scnc_clus, truncation=50, key="hr",formula=~size+decay) 52 df11c<-ds(data_scnc_clus, truncation=50, key="hr",formula=~size+decay+stratum) 53 54 #Half-normal function 55 df12c<-ds(data_scnc_clus, truncation=50, key="hn",formula=~size) 56 df13c<-ds(data_scnc_clus, truncation=50, key="hn",formula=~stratum) 57 58 59 3 60 John Wiley & Sons Page 41 of 56 American Journal of Primatology

1 2 3 4 df14c<-ds(data_scnc_clus, truncation=50, key="hn",formula=~decay) 5 df15c<-ds(data_scnc_clus, truncation=50, key="hn",formula=~size+stratum) 6 df16c<-ds(data_scnc_clus, truncation=50, key="hn",formula=~decay+stratum) 7 df17c<-ds(data_scnc_clus, truncation=50, key="hn",formula=~size+decay) 8 df18c<-ds(data_scnc_clus, truncation=50, key="hn",formula=~size+decay+stratum) 9 10 4.3 Compare candidate detection functions based on AIC and goodness of ﬁt 11 test (Cramer von Mises) 12 13 df_table<-summarize_ds_models(df1c,df2c,df3c,df4c,df5c,df6c,df7c,df8c,df9c, 14 df10c,df11c,df13c,df14c,df15c,df16c,df18c,sort="AIC") 15 row.names(df_table)<-c() 16 kable(df_table[,c("Key function","Formula","C-vM p-value","$\\Delta$AIC")],digits=3, 17 caption="Table S1 - Candidate detection functions") 18 19 ForTable 1: TablePeer S1 - Candidate Review detection functions 20 21 22 Key function Formula C-vM p-value AIC 23 Hazard-rate ~size 0.391 0.000 24 Hazard-rate ~size + decay 0.414 0.360 25 Hazard-rate ~decay 0.440 1.245 26 Hazard-rate ~1 0.445 1.319 27 Half-normal ~decay 0.325 1.415 28 Half-normal ~1 0.346 2.344 29 Uniform with cosine adjustment terms of order 1,2 NA 0.353 3.020 30 Half-normal with cosine adjustment term of order 2 ~1 0.305 3.477 31 Hazard-rate ~size + stratum 0.404 3.764 32 Hazard-rate ~size + decay + stratum 0.397 4.087 33 Half-normal ~size + decay + stratum 0.306 4.448 34 Hazard-rate ~stratum 0.465 5.093 35 Hazard-rate ~decay + stratum 0.422 5.101 36 Half-normal ~decay + stratum 0.311 5.357 37 Half-normal ~size + stratum 0.316 5.510 38 Half-normal ~stratum 0.368 6.241 39 40 41 4.4 Summary and plot of the selected detection function 42 43 summary(df5c) 44 45 ## 46 ## Summary for distance analysis 47 ## Number of observations : 105 48 ## Distance range : 0 - 50 49 ## 50 ## Model : Hazard-rate key function 51 ## AIC : 788.5798 52 ## 53 ## Detection function parameters 54 ## Scale coefficient(s): 55 ## estimate se 56 ## (Intercept) 2.74574721 0.29804898 57 58 59 4 60 John Wiley & Sons American Journal of Primatology Page 42 of 56

1 2 3 ## size 0.05122658 0.04597067 4 ## 5 ## Shape coefficient(s): 6 ## estimate se 7 ## (Intercept) 0.6938415 0.2843393 8 ## 9 ## Estimate SE CV 10 ## Average p 0.534482 0.06794455 0.1271222 11 ## N in covered region 196.451903 28.28535362 0.1439811 12 13 plot(df5c,breaks=seq(0,50,by=5),showpoints=F, xlab=Distance (m), cex=1.5) 14 15 16 17 18 19 For Peer Review 20 21 22 23 24 25 26 27 28 29 30

31 Detection probability 32 33 34 35 0.0 0.2 0.4 0.6 0.8 1.0 36 37 0 10 20 30 40 50 38 39 Distance (m) 40 41 42 43 5. Density surface models 44 45 5.1 Covariates 46 47 1. altitude - mean altitude (m) 48 2. slope - mean slope (%) 49 3. zone_type - conservation (cz) or non-conservation zone (ncz) 50 4. aspect - mean aspect (radians) 51 5. dis_priv - distance to closest permanent river (m) 52 6. dis_sriv - distance to closest seasonal river (m) 53 7. dis_road - distance to closest road (m) 54 8. dis_city - distance to closest city (m) 55 9. agriculture - area of agriculture (ha) 56 10. urban - area of urban areas (ha) 57 58 59 5 60 John Wiley & Sons Page 43 of 56 American Journal of Primatology

1 2 3 11. prim_forest - area of primary forest (ha) 4 12. savanna - area of savanna (ha) 5 13. sec_forest - area of secondary forest (ha) 6 14. diversity - Shannon-Wiener diversity of landuses 7 8 9 5.2 Calculate Shannon-Wiener landuse diversity 10 library(vegan) 11 segment_data$diversity<-diversity(segment_data[,11:15]) 12 13 14 15 16 17 18 19 For Peer Review 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 6 60 John Wiley & Sons American Journal of Primatology Page 44 of 56

1 2 3 5.3 Explore covariates 4 5 5.3.1 Histograms with the covariates 6 7 altitude slope aspect 8 10.0 10.0 10.0 9 7.5 7.5 7.5 10 5.0 5.0 5.0

count 2.5 count 2.5 count 2.5 11 0.0 0.0 0.0 12 100 150 200 250 5 10 15 20 25 100 200 300 13 altitude slope aspect 14 15 dis_priv dis_sriv dis_road 16 10.0 15 17 7.5 7.5 10 18 5.0 5.0 5 19 count 2.5 For Peercount Review count 2.5 20 0.0 0 0.0 21 0 500 1000 1500 2000 0 500 1000 0 500 1000 1500 2000 22 dis_priv dis_sriv dis_road 23 24 dis_city agriculture urban 25 8 100 125 6 75 100 26 75 4 50 27 50 count 2 count 25 count 25 28 0 0 0 29 0 1000 2000 3000 01234 0.0 0.5 1.0 1.5 2.0 30 dis_city agriculture urban 31 32 prim_forest savanna sec_forest 33 20 34 20 40 15 35 10 20 10 36 count count count 5 37 0 0 0 01234 01234 01234 38 prim_forest savanna sec_forest 39 40 diversity 41 42 10 43 5

44 count 45 0 46 0.0 0.2 0.4 0.6 47 diversity 48 49 50 5.3.2 Assess correlations between variables 51 52 vifstep(subset(segment_data,select=c(6:17,22)), th=3) 53 54 ## 2 variables from the 13 input variables have collinearity problem: 55 ## 56 ## sec_forest dis_road 57 58 59 7 60 John Wiley & Sons Page 45 of 56 American Journal of Primatology

1 2 3 ## 4 ## After excluding the collinear variables, the linear correlation coefficients ranges between: 5 ## min correlation ( urban ~ aspect ): -0.006550346 6 ## max correlation ( dis_priv ~ altitude ): 0.6224637 7 ## 8 ## ------VIFs of the remained variables ------9 ## Variables VIF 10 ## 1 altitude 2.679982 11 ## 2 slope 1.744131 12 ## 3 aspect 1.295983 13 ## 4 dis_priv 2.884064 14 ## 5 dis_sriv 1.666188 15 ## 6 dis_city 1.930337 16 ## 7 agriculture 1.674478 17 ## 8 urban 1.438926 18 ## 9 prim_forest 1.440064 19 ## 10 savanna 2.263407For Peer Review 20 ## 11 diversity 2.368425 21 22 23 5.4 Tweedie model 24 5.4.1 Fit the ﬁnal model 25 26 model_tw_c<- dsm(Nhat ~ s(aspect)+s(dis_sriv)+s(savanna)+s(diversity), 27 df5c, observation.data=data_scnc_clus, 28 segment.data=segment_data,engine="gam",family=tw(), 29 select=TRUE,method="REML") 30 summary(model_tw_c) 31 32 ## 33 ## Family: Tweedie(p=1.273) 34 ## Link function: log 35 ## 36 ## Formula: 37 ## Nhat ~ s(aspect) + s(dis_sriv) + s(savanna) + s(diversity) + 38 ## offset(off.set) 39 ## 40 ## Parametric coefficients: 41 ## Estimate Std. Error t value Pr(>|t|) 42 ## (Intercept) -9.0430 0.2508 -36.06 <2e-16 *** 43 ## --- 44 ## Signif. codes: 0 *** 0.001 ** 0.01 * 0.05 . 0.1 1 45 ## 46 ## Approximate significance of smooth terms: 47 ## edf Ref.df F p-value 48 ## s(aspect) 0.9159 9 1.190 0.000761 *** 49 ## s(dis_sriv) 1.2431 9 0.477 0.034616 * 50 ## s(savanna) 1.8387 9 1.953 5.24e-05 *** 51 ## s(diversity) 2.0221 9 1.259 0.002151 ** 52 ## --- 53 ## Signif. codes: 0 *** 0.001 ** 0.01 * 0.05 . 0.1 1 54 ## 55 ## R-sq.(adj) = 0.355 Deviance explained = 46.9% 56 ## -REML = 226.71 Scale est. = 8.8207 n = 125 57 58 59 8 60 John Wiley & Sons American Journal of Primatology Page 46 of 56

1 2 3 5.4.2 Model validation 4 5 par(mfrow=c(2,2)) 6 gam.check(model_tw_c) 7 8 9 Resids vs. linear pred. 10 11 12 13 14

15 residuals 16 − 4 0 4 − 4 0 4 17 residuals deviance −4−2 0 2 4 6 8 −4−2 0 2 18 19 theoreticalFor quantiles Peer Review linear predictor 20 21 22 23 Histogram of residuals Response vs. Fitted Values 24 25 26 27 28 Response Frequency

29 0 15 30 0 30 60 30 31 −4−2 0 2 4 6 0 10 20 30 40 32 33 Residuals Fitted Values 34 35 36 ## 37 ## Method: REML Optimizer: outer newton 38 ## full convergence after 13 iterations. 39 ## Gradient range [-0.000533106,0.0006441383] 40 ## (score 226.7065 & scale 8.820699). 41 ## Hessian positive definite, eigenvalue range [1.435505e-05,82.05384]. 42 ## Model rank = 37 / 37 43 ## 44 ## Basis dimension (k) checking results. Low p-value (k-index<1) may 45 ## indicate that k is too low, especially if edf is close to k. 46 ## 47 ## k edf k-index p-value 48 ## s(aspect) 9.000 0.916 0.94 0.81 49 ## s(dis_sriv) 9.000 1.243 0.79 0.14 50 ## s(savanna) 9.000 1.839 0.80 0.14 51 ## s(diversity) 9.000 2.022 0.91 0.67 52 par(mfrow=c(1,1)) 53 dsm.cor(model_tw_c, max.lag = 10,main="Assess autocorrelation") 54 55 56 57 58 59 9 60 John Wiley & Sons Page 47 of 56 American Journal of Primatology

1 2 3 4 5 Assess autocorrelation 6 7 8 9 10 11 12 13 14 15 16 17

18 Correlation 19 For Peer Review 20 21 22 23 24 0.0 0.2 0.4 0.6 0.8 1.0 25 0 1 2 3 4 5 6 7 8 9 10 26 27 28 Lag 29 30 31 concurvity(model_tw_c) 32 33 ## para s(aspect) s(dis_sriv) s(savanna) s(diversity) 34 ## worst 1.777762e-24 0.6080265 0.5696487 0.6269396 0.6599517 35 ## observed 1.777762e-24 0.2487394 0.4249006 0.4398928 0.5684356 36 ## estimate 1.777762e-24 0.2387085 0.4267187 0.4632259 0.5252859 37 38 5.4.3 Plot smoothers 39 40 par(mfrow=c(2,2)) 41 plot(model_tw_c, shade=TRUE, ylim=c(-5,2),fig.height=7,select=1,xlab="Aspect") 42 plot(model_tw_c, shade=TRUE, ylim=c(-5,2),fig.height=7,select=2,xlab="Distance to closest seasonal river") 43 plot(model_tw_c, shade=TRUE, ylim=c(-5,2),fig.height=7,select=3,xlab="Savanna") 44 plot(model_tw_c, shade=TRUE, ylim=c(-5,2),fig.height=7,select=4,xlab="Shannon-Wiener landuse diversity") 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 10 60 John Wiley & Sons American Journal of Primatology Page 48 of 56

1 2 3 4 5 6 7 8 9 10 11 12 13 14

15 s(aspect,0.92) s(dis_sriv,1.24) 16 17 18 19 − 5 − 3 − 1 0 1For 2 Peer Review− 5 − 3 − 1 0 1 2 20 50 100 150 200 250 300 0 200 600 1000 21 22 Aspect Distance to closest seasonal river 23 24 25 26 27 28 29 30 31 32 33 34 35 36 s(diversity,2.02) 37 s(savanna,1.84) 38 39 40 − 5 − 3 − 1 0 1 2 − 5 − 3 − 1 0 1 2 41 01234 0.1 0.2 0.3 0.4 0.5 0.6 0.7 42 43 Savanna Shannon−Wiener landuse diversity 44 45 46 par(mfrow=c(1,1)) 47 48 5.5 Negative binomial model 49 50 5.5.1 Fit the ﬁnal model 51 52 model_nb_c<- dsm(Nhat ~ s(slope)+s(aspect)+s(savanna), 53 df5c, observation.data=data_scnc_clus, 54 segment.data=segment_data,engine="gam",family=nb(), 55 select=TRUE,method="REML") 56 57 58 59 11 60 John Wiley & Sons Page 49 of 56 American Journal of Primatology

1 2 3 ## Warning in make.data(response, ddf.obj, segment.data, observation.data, : 4 ## Some observations are outside of detection function truncation! 5 6 summary(model_nb_c) 7 ## 8 ## Family: Negative Binomial(0.164) 9 ## Link function: log 10 ## 11 ## Formula: 12 ## Nhat ~ s(slope) + s(aspect) + s(savanna) + offset(off.set) 13 ## 14 ## Parametric coefficients: 15 ## Estimate Std. Error z value Pr(>|z|) 16 ## (Intercept) -8.8602 0.2391 -37.06 <2e-16 *** 17 ## --- 18 ## Signif. codes: 0 *** 0.001 ** 0.01 * 0.05 . 0.1 1 19 ## For Peer Review 20 ## Approximate significance of smooth terms: 21 ## edf Ref.df Chi.sq p-value 22 ## s(slope) 1.1315956 9 2.232 0.128 23 ## s(aspect) 0.0002205 9 0.000 0.597 24 ## s(savanna) 1.9722462 9 24.181 3.99e-07 *** 25 ## --- 26 ## Signif. codes: 0 *** 0.001 ** 0.01 * 0.05 . 0.1 1 27 ## 28 ## R-sq.(adj) = 0.158 Deviance explained = 26.6% 29 ## -REML = 252.32 Scale est. = 1 n = 125 30 31 32 5.5.2 Model validation 33 par(mfrow=c(2,2)) 34 gam.check(model_nb_c) 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 12 60 John Wiley & Sons American Journal of Primatology Page 50 of 56

1 2 3 4 5 Resids vs. linear pred. 6 7 8 9 10 11 12 13

14 residuals 15

16 residuals deviance 17

18 − 1.0 0.0 0.5 1.0 − 1.0 0.0 0.5 1.0 19 For Peer Review 20 −1.0 0.0 0.5 1.0 1.5 2.0 −1 0 1 2 21 22 theoretical quantiles linear predictor 23 24 25 26 Histogram of residuals Response vs. Fitted Values 27 28 29 30 31 32 33 34 35 Response 36 Frequency 37 38

39 0 10 20 30 40 0 10 30 50 70 40 41 −1.5 −0.5 0.5 1.0 1.5 0 2 4 6 8 10 12 42 43 Residuals Fitted Values 44 45 46 ## 47 ## Method: REML Optimizer: outer newton 48 ## full convergence after 11 iterations. 49 ## Gradient range [-8.869582e-05,9.496902e-05] 50 ## (score 252.3186 & scale 1). 51 ## Hessian positive definite, eigenvalue range [6.644452e-06,24.35469]. 52 ## Model rank = 28 / 28 53 ## 54 ## Basis dimension (k) checking results. Low p-value (k-index<1) may 55 ## indicate that k is too low, especially if edf is close to k. 56 ## 57 58 59 13 60 John Wiley & Sons Page 51 of 56 American Journal of Primatology

1 2 3 ## k edf k-index p-value 4 ## s(slope) 9.00000 1.13160 0.69 0.435 5 ## s(aspect) 9.00000 0.00022 0.74 0.770 6 ## s(savanna) 9.00000 1.97225 0.54 0.005 ** 7 ## --- 8 ## Signif. codes: 0 *** 0.001 ** 0.01 * 0.05 . 0.1 1 9 10 par(mfrow=c(1,1)) 11 dsm.cor(model_nb_c, max.lag = 10,main="Assess autocorrelation") 12 13 14 Assess autocorrelation 15 16 17 18 19 For Peer Review 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 Correlation 36 37 38 39 40 41 42 43 44 45 46

47 0.0 0.2 0.4 0.6 0.8 1.0 48 49 50 0 1 2 3 4 5 6 7 8 9 10 51 52 Lag 53 54 55 56 57 58 59 14 60 John Wiley & Sons American Journal of Primatology Page 52 of 56

1 2 3 4 concurvity(model_nb_c) 5 ## para s(slope) s(aspect) s(savanna) 6 ## worst 9.027777e-25 0.5553578 0.3767250 0.5460532 7 ## observed 9.027777e-25 0.4459200 0.1783605 0.4180740 8 ## estimate 9.027777e-25 0.3620254 0.1609830 0.4809394 9 10 11 5.5.3 Plot smoothers 12 par(mfrow=c(2,2)) 13 plot(model_nb_c, shade=TRUE, ylim=c(-5,2),fig.height=7,select=1,xlab="Slope") 14 plot(model_nb_c, shade=TRUE, ylim=c(-5,2),fig.height=7,select=2,xlab="Aspect") 15 plot(model_nb_c, shade=TRUE, ylim=c(-5,2),fig.height=7,select=3,xlab="Savanna") 16 par(mfrow=c(1,1)) 17 18 19 For Peer Review 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 15 60 John Wiley & Sons Page 53 of 56 American Journal of Primatology

1 2 3 4 5 6 7 8 9 10 11 12 13 14 s(aspect,0) 15 s(slope,1.13) 16 17 18 19 − 5 − 3 − 1 0 1For 2 Peer Review− 5 − 3 − 1 0 1 2 20 5 10 15 20 25 50 100 150 200 250 300 21 22 Slope Aspect 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 s(savanna,1.97) 38 39 40 − 5 − 3 − 1 0 1 2 41 01234 42 43 Savanna 44 45 46 47 5.6 Which model should we select? 48 par(mfrow=c(1,2)) 49 qq.gam(model_tw_c,rep=100,main="Tweedie") 50 qq.gam(model_nb_c,rep=100,main="Negative binomial") 51 52 53 54 55 56 57 58 59 16 60 John Wiley & Sons American Journal of Primatology Page 54 of 56

1 2 3 4 5 Tweedie Negative binomial 6 7 8 9 10 11 12 13 14 15 16 17 18 19

deviance residuals deviance For Peer Review residuals deviance 20 21

22 − 5 0 5 23 − 1 0 1 2 3 24 25 −4 0 2 4 6 −1.0 0.0 1.0 2.0 26 27 28 theoretical quantiles theoretical quantiles 29 30 31 par(mfrow=c(1,1)) 32 33 This plot shows a comparison of models with Tweedie (left) and negative binomial (right) response distributions 34 by quantile-quantile plots. Good fit is indicated by agreement between observed and fitted (residual) quantiles 35 (i.e., points being close to the red line). 90% reference bands are shown in grey allowing judgement of the 36 deviation from the line. The negative binomial points fall further away from the red line than those for the 37 Tweedie, indicating model misspecification. 38 39 6. Model predictions 40 41 6.1 Calculate oset 42 43 off.set <- (200 * 200) #grid is 200 m x 200 m 44 45 46 6.2 Predictions from the Tweedie model 47 48 6.2.1 Calculate predicted abundances 49 model_tw.pred_c <- predict(model_tw_c, preddata, off.set) 50 preddata$TW_ab_c<-unname(model_tw.pred_c) 51 52 53 6.2.2 Plot predicted abundances alongside transects and clusters of nests 54 55 p<-ggplot(preddata, aes(x, y))+theme_minimal() 56 p<-p + geom_raster(aes(fill = TW_ab_c)) 57 58 59 17 60 John Wiley & Sons Page 55 of 56 American Journal of Primatology

1 2 3 4 p<-p + scale_fill_viridis(name="Nests") 5 p<-p + geom_path(data=mine, aes(long, lat, colour="black")) 6 p<-p + geom_path(aes(x=POINT_X, y=POINT_Y, colour = "brown4"), data = Points_areas[26:185,]) 7 p<-p + geom_path(aes(x=POINT_X, y=POINT_Y, group = ORIG_FID, colour = "darkcyan"), 8 data = Points_river_roads[1:121,]) 9 p<-p + geom_line(aes(x, y, group = Transect.Label), data = segment_data) 10 p<-p + geom_point(aes(x=x, y=y), data = data_scnc_clus, colour = "red", 11 alpha = I(0.7)) 12 p<-p+scale_color_hue(labels = c("Mine", "Ronde Hill","Rivers","Transect"))+ 13 labs(color=) 14 p<-p+theme(panel.grid.major = element_blank(),panel.grid.minor = element_blank(), 15 axis.line = element_line(colour = "black")) 16 p<-p+theme(axis.text.x=element_blank(),axis.text.y=element_blank())+labs(x="",y="")+ 17 coord_cartesian(expand=FALSE) 18 p 19 For Peer Review 20 21 22 23 24 Nests 25 26 60 27 28 40 29 30 20 31 32 33 34 35 36 Mine 37 Ronde Hill 38 39 Rivers 40 41 42 43 44 45 46 47 48 6.2.3 Calculate prediction variances 49 dsm.var.gam 50 model_tw_var_c<- (model_tw_c, pred.data = preddata, off.set = off.set) summary 51 (model_tw_var_c) 52 ## Summary of uncertainty in a density surface model calculated 53 ## analytically for GAM, with delta method 54 ## 55 ## Approximate asymptotic confidence interval: 56 ## 2.5% Mean 97.5% 57 58 59 18 60 John Wiley & Sons American Journal of Primatology Page 56 of 56

1 2 3 ## 2268.237 3877.604 6628.855 4 ## (Using log-Normal approximation) 5 ## 6 ## Point estimate : 3877.604 7 ## CV of detection function : 0.1271222 8 ## CV from GAM : 0.2481 9 ## Total standard error : 1081.014 10 ## Total coefficient of variation : 0.2788 11 12 13 7. Calculate density and abundance of nest building chimpanzees 14 15 with the Tweedie model 16 To calculate the density of chimpanzees we use the following formula: 17 18 D_weaned_chimpanzee=D_nests/(r*t) 19 For Peer Review 20 21 22 where “r” is the estimated rate of nest production per individual per day estimated to be 1.09 23 nests/individual/day by Plumptre & Reynolds (1997) and “t” is the mean life of a nest estimated to be 194 24 days by Fleury-Brugiere & Brugiere (2010). 25 26 27 28 Following this formula, the estimated number of weaned chimpanzees in the study area is: 29 30 weaned_chimps<- as.numeric(model_tw_var_c$pred)/(1.09*194) 31 print(weaned_chimps) 32 ## [1] 18.33729 33 34 Now, we calculate the 95% conﬁdence intervals for the number of weaned chimpanzees in the study area 35 using the upper and lower bounds of the estimated number of nests (see above): 36 weaned_chimps_upper <- 6628.829/(1.09*194) 37 print(weaned_chimps_upper) 38 39 ## [1] 31.34791 40 41 weaned_chimps_lower <- 2268.236/(1.09*194) 42 print(weaned_chimps_lower) 43 ## [1] 10.72655 44 45 Considering the study area covers 47.04 squared kilometers, the number of chimpanzees per squared kilometer 46 and the corresponding 95% conﬁdence interval is: 47 estimate<- c(weaned_chimps, weaned_chimps_lower, weaned_chimps_upper) 48 final_value<- estimate/47.04 49 print(final_value) 50 51 ## [1] 0.3898234 0.2280304 0.6664096 52 53 54 55 56 57 58 59 19 60 John Wiley & Sons