Scaling up local ecological knowledge to prioritise areas for protection: Determining Philippine pangolin distribution, status and threats

Abstract Local ecological knowledge (LEK) can provide cost-effective baseline ecological data across large geographical areas, and is increasingly seen as an important source of information for rare and cryptic species. However, to date, its use as a practical tool for prioritising conservation action is limited. Pangolins are the world’s most heavily trafficked wild mammals and all species are in decline. The Philippine pangolin (Manis culionensis) is Critically Endangered but conservation efforts are hindered by a lack of knowledge on where populations still exist and where in situ action should be prioritised. We conducted the first range-wide systematic survey for the species using household interviews (n = 1296) to provide new data on pangolin distribution, status and threats, and to assess the use of LEK for highlighting priority areas for conservation. LEK about pangolins was high (87% of respondents recognised pangolins and provided further information), with evidence of pangolin occurrence in 17 of the 18 municipalities surveyed. The majority (70%) of respondents had seen a pangolin, but most (72%) perceived pangolins to be ‘rare’ or ‘very rare’, and local use of pangolins was reported across the species’ range. Spatial differences in sighting frequencies, perceived abundance and reported population trends were observed, providing an important baseline to identify priority sites for targeted research and community-based pangolin conservation.


Introduction 27
The importance of robust data to inform conservation management is widely recognised by 28 conservation practitioners (Sutherland et al., 2020;Yoccoz et al., 2001). An evidence-based 29 approach allows changes in wildlife populations to be tracked and can identify key areas that are in 30 need of protection, helping to direct limited resources to where they are most needed (Collen et al.,31 respondent knowledge levels may differ or be influenced by socio-demographic parameters 72 (Beaudreau and Levin, 2014; Iniesta-Arandia et al., 2014; Papworth et al., 2009). This presents 73 challenges when working across large geographical areas, as random respondent selection is 74 needed to achieve adequate sample sizes for analysis, complicating efforts to ensure respondent 75 knowledge levels are comparable across study areas. Further, whereas LEK data can determine 76 species' presence or absence, they cannot determine absolute abundance, an important metric in 77 spatial prioritisation of conservation effort; LEK data might instead be restricted to providing broad-78 level insights and relative abundance patterns, and are limited spatially to areas subject to human 79 use that may coincide with anthropogenic threats but not necessarily with areas of high species 80 abundance. However, uncertainty and bias can be reduced through appropriate data collection and 81 critical analysis that accounts for socio-demographic variation within datasets. For example, 82 inclusion of additional "control species" within survey design permits comparison of between-83 species relative abundance patterns, and assessment of whether data variation is likely to reflect 84 underlying ecological patterns or instead variation in respondent knowledge or experience (Turvey 85 et al., 2015). The use of interspecies comparisons has been used elsewhere to provide insights into 86 species distributions and relative abundance patterns for other rare and cryptic species (Turvey et 87 al., 2015) and increases the likelihood of respondents reporting potentially sensitive information on 88 the target species, alongside reducing social desirability bias by removing the focus from the 89 species of interest (Newing, 2011). 90 Here, we explore the use of LEK to help identify priority areas for community-based conservation 91 using the Philippine pangolin (Manis culionensis) as a case study. Endemic to Palawan Province 92 (mainland Palawan and associated islands) in the Philippines, this species is classified as Critically 93 Endangered on the IUCN Red List (Schoppe et al., 2019), and is an example of a rare mammal 94 which, like other pangolin species, is rarely detected through general biodiversity surveys (Ichu et  In this study, we use a multi-species LEK survey in communities living close to natural areas to 105 Filipino, Cuyonon or Bisayan languages by interviewers local to Palawan Province to ensure 146 appropriate positionality and minimise social desirability bias (Newing, 2011). Interviewers received 147 a week's training, followed by two rounds of pilot surveys to trial and reformat question structure and 148 wording. 149 Questionnaires consisted of both closed and open-ended questions, took up to 35 minutes to 150 complete (Appendix 1), and were completed on android tablets using the software Open Data Kit 151 (Hartung et al., 2010). Data on respondent attributes and socio-demographics were collected, 152 followed by questions focussed on the respondent's LEK in relation to five species (Table 1). 153 154 J o u r n a l P r e -p r o o f

157
Questions on the pangolin, porcupine and hornbill were randomised to remove any potential order 158 bias. Photographs (sourced locally or from www.arkive.org) were used to present each animal and 159 engage respondents in the interview process (Nash et al., 2016). Follow-up questions asked 160 respondents if they recognised each species, and if so, whether they had seen it, the calendar year 161 of their last sighting, last-sighting location (within or outside village boundaries, habitat types, and 162 specific habitat characteristics), frequency of sightings, and perceptions on the conservation status 163 and population trends (covering the past ten years) for each species. Open-ended questions on 164 cultural values and local uses of wildlife were also included, providing respondents with the 165 opportunity to discuss personal or local beliefs and uses of pangolins. Respondents were also 166 asked their opinions on conservation importance and willingness to be involved in conservation 167 respondents recognised or had seen each species were investigated using generalised linear mixed 174 models (GLMMs) using a binomial error structure, as the response variables are binary (yes/no). 175 The R package "glmmTMB" was used for analysis. Variables influencing how a respondent: i) 176 perceived species population changes, ii) perceived species abundance, iii) perceived the 177 importance of conservation, and iv) reported their willingness to help monitor wildlife were 178 investigated using ordinal logistic regression models using the R package "ordinal". Models were 179 fitted using the "clmm" function to allow for the inclusion of random effects. Ordinal logistic 180 regression models were also used to investigate factors influencing perceived abundance and trend 181 scores across all species. Variables for inclusion were selected a priori (Appendix 2, Table 1). Post-182 hoc tests using the R package 'emmeans' were conducted to compare between groups. Chi-183 squared tests were used to test for associations between respondent recognition, sightings and 184 perceptions of pangolins and respondent recognition, sightings, and perceptions of other Palawan 185 endemic species, using the subset of respondents who could recognise all four species. 186 J o u r n a l P r e -p r o o f 8

Perceived pangolin abundance and trends 266
Across the subset of respondents who could recognise a pangolin, 72% (n=811) perceived the 267 pangolin population in their village to be either 'rare' or 'very rare', 22% (n=248) perceived it to be 268 'common' or 'very common', and 6% (n=64) were unsure. Municipality had a significant effect on 269 perceived pangolin abundance (CLMM, X²=43.405, df=16, p<0.001; Appendix 2: Figure 1), as did 270 recent pangolin sightings (CLMM, X²=150.220, df=1, p<0.001), with respondents who reported 271 seeing pangolins in 2018-2019 more likely to give a higher abundance score. Age and gender did 272 not have a significant effect, but occupation did (CLMM, X²=9.881, df=4, p<0.042), with farmers 273 more likely to report higher abundance scores compared to respondents in private employment. and occupation had no significant effect, but respondents who had seen a pangolin recently were 280 more likely to report either no population changes or increasing population trends (CLMM, 281 X²=32.146, df=1, p<0.001). 282 Across focal species, pangolins were most frequently perceived to be rare or very rare by 283 respondents (Figure 5b), and were most frequently reported to have declined over the past ten 284 years ( Figure 5a). Species abundance scores were significantly different (CLMM, X²=1450.69, df=3, 285 p<0.001), with respondents significantly more likely to report lower abundance scores for pangolins 286 compared to all other species. Perceived abundance was also influenced significantly by 287 municipality (CLMM, X²=65.44, df=14, p<0.001; Appendix 2: Figure

Local uses of pangolins 316
Local use of pangolins was reported by 49% (n=553) of respondents, with many respondents 317 describing multiple uses: 75% (n=492) of descriptions related to pangolin consumption, 20% 318 (n=131) related to pangolin trade, and 5% (n=30) related to medicinal use. Pangolin scales, blood 319 and internal organs were all reported to have medicinal properties and used to treat conditions such 320 as asthma, tuberculosis, stomach aches, lung conditions and back pain (Appendix 2: Table 2). At 321 the village level, 99% of villages (n=71 in 17/18 municipalities) reported pangolin consumption, 71% 322 (n=51 in 15/18 municipalities) reported pangolin trade, and 28% (n=20 in 9/18 municipalities) 323 reported use of pangolins for medicinal purposes. Other cultural uses were reported at low 324 frequencies (n=48), including the burning of scales to protect against bad spirits or to ward off 325 insects in rice fields (Appendix 2: Table 3). 326

3.5.
Respondent willingness to engage in conservation and perceived importance of 327 conservation 328 Across respondents, few reported low or no willingness to help monitor wildlife and protecting 329 wildlife was largely perceived to be important or very important (Appendix 2: Figures 6 and 7). 330 However, willingness scores and importance scores were both significantly influenced by 331 municipality (willingness: CLMM, X²=49.268, df=17, p<0.001, Appendix 2: Figure 6; importance: 332 CLMM, X²=40.140, df=17, p<0.001, Appendix 2: Figure 7); in particular, respondents in Bataraza 333 and Brooke's Point were more likely to give lower willingness and perceived importance scores 334 compared to respondents in Aborlan, Busuanga, Culion, Narra, Puerto Princesa, Quezon and 335 Roxas. Gender significantly influenced scores, with male respondents more likely to give higher 336 willingness scores (CLMM, X²=9.717, df=1, p<0.002) and higher importance scores (CLMM, 337 X²=5.905, df=1, p<0.015). Education also significantly influenced both willingness scores (CLMM, 338 X²=15.433, df=4, p<0.004) and importance scores (CLMM, X²=16.546, df=4, p<0.002), with 339 respondents with college-level education more likely to give higher willingness scores than 340 respondents with no, elementary or high school-level education, and respondents with high school 341 or college-level education more likely to give higher importance scores than respondents with no or 342 elementary-level education. Occupation and ethnicity influenced willingness to help monitor wildlife, 343 with those of 'other' occupations less likely than all other occupations to give high willingness 344 scores, and respondents of Tagbanua ethnicity more likely to give higher willingness scores 345 compared to all other ethnicities. Occupation and ethnicity did not influence perceived importance of 346 wildlife protection and age did not significantly influence either model. All model results are 347 presented in Appendix 2, Table 4. However, although recent sightings indicate the species' continued persistence across the 381 province, most participants considered it to be either rare or very rare, and declines were 382 reported in every municipality. Further, abundance ratings for pangolins were significantly 383 lower than results for other Palawan species, most of which are also threatened and 384 declining, with the majority of respondents perceiving the pangolin population in their local 385 area to be rare or very rare. Pangolins were also the species most regularly reported to be 386 declining, with the majority of respondents reporting declines, and pangolin trends 387 significantly worse than trends reported for all other species. Indeed, despite high levels of 388 respondent recognition and overall sighting frequencies for pangolins, mean sighting dates 389 and recent pangolin sightings were the lowest for all our target species. High overall sighting 390 frequencies coupled with relatively few recent sightings and high probability of reporting 391 declines are indicative of substantial recent declines in pangolin populations. These results 392 suggest that in absolute terms, pangolins are probably now relatively rare across the 393 province, and despite a wide distribution, populations are likely to be small and declining. 394 There is therefore an urgent need to establish conservation efforts before it is too late to help 395 the species. 396 Though in overall terms, multiple metrics of pangolin status indicate that the species is now 397 relatively rare across the province, geographical differences in sighting frequencies, trends 398 and perceived status were seen, suggesting that occurrence and threats may not be equally 399 This could be a concern due to pangolins' myrmecophagous diets (Chao et  and Klenk, 2020). By combining LEK and local attitudes, studies such as this can provide an 504 initial baseline to better understand local considerations and demographic influences at an 505 earlier stage in the conservation planning process. This can help to prioritise conservation 506 efforts to areas with higher potential success and move beyond the use of LEK solely for 507 data collection, towards a more integrated approach that views local knowledge as a 508 legitimate and central part of the management process and provides a starting point for 509 collaborative and inclusive conservation (Latulippe and Klenk, 2020). 510

Conclusion 511
Our data indicate that conservation initiatives for Philippine pangolins need to be scaled up 512 and developed as a priority. Sighting frequencies, perceived pangolin abundance, and 513 willingness to help monitor wildlife are higher in northern municipalities, and we suggest that 514 these areas could be focused on initially. However, although we found geographical variation 515 across multiple metrics of pangolin population status, recent pangolin reports are 516 documented across the province, and high levels of local support for conservation offer hope 517 that it is not too late to develop range-wide conservation initiatives. With limited data on 518 pangolin status and threats available for some municipalities prior to this research, we hope 519 this study will provide the evidence needed to encourage municipal government bodies 520 across Palawan to engage in pangolin conservation efforts. 521 Our findings provide evidence that LEK data can offer valuable insights to confirm species' 522 presence, assess their status, and understand local use and values. LEK therefore not only 523 Combining local knowledge and occupancy analysis for a rapid assessment of the 578 forest elephant Loxodonta cyclotis in Cameroon's timber production forests. Oryx 1-11.   Interviews can be subject to interviewer bias, with the potential for questions to be answered differently depending on who is asking the questions. This was therefore included in the models to check for bias. Trade General trade reports 50 "people sell the scales" "people catch it and sell it" "there was trading before" "the meat is for food, and scales for sale -₱10,000 per kilo" "trading 2017 price was ₱12,000 per kilo of scales, people eat the meat also" "2014 buyer from Puerto asking to buy the whole pangolin alive for ₱5000" "in community scales is for sale"

Random
Recent trade reports 11 "trading is still ongoing" "even now there is trading but just hiding, catching by chance" "this year 2019 -₱700-1,200 per kilo and it is common food for some people here" "last year there was people catching it to sell" "trading here even now and ₱5,000 per kilo of scales, meat is ₱300 per kilo" "before people hunt it to sale but not now anymore"

Medicinal
Medicinal -general uses 19 "traditional medicine" "scales are medicine" "blood is medicine" "eat the meat, it can heal sick people" "internal organ is medicine" "only the scales for medicine -burn and drink the charcoal of scales" J o u r n a l P r e -p r o o f Medicine to treat asthma 10 "blood and liver cure asthma" "scales and blood as treatment for asthma" "blood as cure for asthma" "scales and blood medicine for asthma" Medicinal value for women who have given birth 2 "the poop is good to increase the health of woman who have given birth. Dry the poop and grill it and drink the ash" "scales are medicine for women who have given birth" Medicinal value for back and body pain 1 "scales medicine for body pain and backbone"       J o u r n a l P r e -p r o o f