Integrating field and satellite data for spatially explicit inference on the density of threatened arboreal primates

Abstract

Spatially explicit models of animal abundance are a critical tool to inform conservation planning and management. However, they require the availability of spatially diffuse environmental predictors of abundance, which may be challenging, especially in complex and heterogeneous habitats. This is particularly the case for tropical mammals, such as nonhuman primates, that depend on multi-layered and species-rich tree canopy coverage, which is usually measured through a limited sample of ground plots. We developed an approach that calibrates remote-sensing imagery to ground measurements of tree density to derive basal area, in turn used as a predictor of primate density based on published models. We applied generalized linear models (GLM) to relate 9.8-ha ground samples of tree basal area to various metrics extracted from Landsat 8 imagery. We tested the potential of this approach for spatial inference of animal density by comparing the density predictions for an endangered colobus monkey, to previous estimates from field transect counts, measured basal area, and other predictors of abundance. The best GLM had high accuracy and showed no significant difference between predicted and observed values of basal area. Our species distribution model yielded predicted primate densities that matched those based on field measurements. Results show the potential of using open-access and global remote-sensing data to derive an important predictor of animal abundance in tropical forests and in turn to make spatially explicit inference on animal density. This approach has important, inherent applications as it greatly magnifies the relevance of abundance modeling for informing conservation. This is especially true for threatened species living in heterogeneous habitats where spatial patterns of abundance, in relation to habitat and/or human disturbance factors, are often complex and, management decisions, such as improving forest protection, may need to be focused on priority areas.

Effects of anthropogenic disturbance on primate density at the landscape scale

“…() and September to November 2015 by Cavada et al. () (number of primate social groups detected in Supporting Information). Data collected followed the distance sampling approach (Buckland et al.…”

“…Primate data were collected July 2011 to February 2012 and August to November 2012 by Araldi et al (2014) and Barelli et al (2015) and September to November 2015 by Cavada et al (2017b) (number of primate social groups detected in Supporting Information). Data collected followed the distance sampling approach (Buckland et al 2001); 2-km-long transects were uniformly distributed in each forest block according to a randomly placed grid ( Fig.…”

Model selection using information criteria, but is the “best” model any good?

“…One study found that the null model was the best-fitting model in two of four cases (Major, Buxton, Schacter, Conners, & Jones, 2017), suggesting little support for the ecological hypotheses of interest. Several papers examined the match between observed and fitted values for an averaged model (Edwards, Massam, Haugaasen, & Gilroy, 2017;Mitchell, Bakker, Vincent, & Davies, 2017), while another undertook cross-validation on a model-averaged result (Cavada et al, 2017). The other 64 papers did not evaluate model goodness-of-fit in an absolute sense.…”

Aligning quantitative vegetation classification and landscape scale mapping: updating the classification approach of the Regional Ecosystem classification system used in Queensland

“…Species abundance can be measured in several ways useful for classifying vegetation communities (Kent 2012) and in Queensland the recommended data collection methodology focuses on three; basal area, stem counts and percentage cover . All of these can be used to estimate relative dominance and abundance (for example (Lehmann et al 2014;Memiaghe et al 2016;Cavada et al 2017;Eldridge et al 2018); however, the arguments supporting the use of percentage cover as the abundance measure in quantitative class definition procedures for the RE system are compelling. The most common abundance measure used globally in vegetation classification systems (De Cáceres et al 2018) is percentage cover and within Australia it is used to determine national vegetation types (Executive Steering Committee for Australian Vegetation Information and Department of the Environment and Heritage 2003; Hnatiuk et al 2009).…”