Ecological resilience in a primate community affected by gold mining in Suriname

Tropical habitats in South America and Africa are being transformed by artisanal gold mining (ASM) but few studies have addressed how mining impacts animals at the community level.  We assessed long-term ecological resilience to mining disturbance for seven primate species (Allouatta macconnelli, Ateles paniscus, Cebus olivaceus, Chiropotes sagulatus, Pithecia pithecia, Saguinus midas, and Sapajus apella) in Brownsberg Nature Park, Suriname over a 20-year period. Using eleven trails and unpaved roads to calculate “encounter rates” (species encountered/km walked), we compared encounter frequency, encounter location, and group size across four community-wide surveys in 2003, 2013, 2014, and 2023. We hypothesized that primate response to gold mining would 1) affect species encounter rates; 2) shift the location of encounters relative to mining activity, and 3) impact group sizes. Intraspecific variation in encounter rates from 2003 to 2023 did not vary significantl,y but minimum group sizes declined for all species (four species showing significant declines). The three more recent surveys also showed that two species were encountered in areas close to the top of the mountain.  We suggest that in the context of intensified mining, the Brownsberg primate community maintained stable encounter rates for all species and some species shifted their ranges as evidenced by greater encounter rates farther from the periphery of the study area. While this suggests a capacity for resilience in the face of mining-related disturbances, the decline in group sizes may be an early sign of an insidious community-wide effect.  

Methods
Decadal surveys of seven species of primates at Brownsberg Nature Park, Suriname, South America, allowed us to compare variation in species encounter rates, encounter locations, and minimum group size from 2013-2023. We also compared habitat loss due to artisanal gold mining from 2003 to 2020 using satellite imaging files.

We collected encounter rates by walking trails and recording the observation of primate species, their location and minimum group size in four surveys (2003, 2013, 2014, 2023). These data were analyzed using multivariate techniques:

We used general linear modeling (GLM) to assess factors influencing primate encounter rates (ERs) at BNP. We first conducted a single GLM across the full data set to test the effect of species differences and survey year on ERs. For this test, ER was the dependent variable, with species and survey year as the independent variables. Species were designated as a repeated effect (due to repeat measures each survey year) and ERs were pooled across all surveyed trails (i.e., each species had a cumulative ER for each survey year). We conducted Bonferroni post-hoc tests to compare pairwise differences in ERs between species.
Jonckheere-Terpstra tests for ordered alternatives to compare minimum group size across survey years for each species.
We created a "plateau index" to compare encounters in two habitat types. Data were analyzed using a repeated measures ANOVA to test for differences in location (plateau or slope) between and among the 2003, 2013, 2014, and 2023 surveys, with pairwise Bonferroni post hoc tests.

Habitat loss was compared using satellite images of the survey area from 2003, 2013 and 2020. To quantify forest cover, we overlaid a 100 x 100 m grid on the survey area. We then visually classified each grid cell into one of four categories based on forest cover: 100% forested, >50% forested, <50% forested, 0% forested. From these estimates, we calculated total forested area for each survey year, and using 2003 as a baseline, calculated the percent loss between 2003, 2013, and 2020.