Evaluating relative abundance indices for terrestrial herbivores from large‐scale camera trap surveys

Abstract

The ability to directly monitor animal populations across time and space is a key element of wildlife conservation and management, but logistically difficult to achieve. Photographic capture rates from camera trap surveys can provide relative abundance indices (RAIs) for a wide variety of medium‐ to large‐bodied wildlife species. RAIs are less complex than other estimation methods and are commonly used when true abundance is difficult or costly to measure. However, this method is controversial as it does not account for potential bias arising from imperfect detection. Here, we evaluate the reliability and precision of RAI estimates drawn from a large‐scale camera trap survey for ten African herbivore species by comparing them against preexisting aerial survey data. RAIs correlated strongly with independent estimates, particularly when indices were derived from counts of all photographed animals. RAIs were most reliable for species that were nonmigratory, occurred in open habitats and had high rates of daily movement. Increasing survey coverage and duration both had strong but comparable effects on improving RAI precision. Our results suggest that RAIs from camera traps hold substantial promise as a tool for monitoring herbivore relative abundances, and we provide guidelines on the utility of this approach for ecological inference.

An empirical evaluation of camera trap study design: How many, how long and when?

“…Although detection rate can be a misleading measure of relative abundance if field protocols are not standardized (Sollmann, Mohamed, Samejima, & Wilting, ), it has been tied mechanistically to abundance (Rowcliffe, Field, Turvey, & Carbone, ), and has been shown to reflect animal density in a number of studies using grids or stratified random sampling (e.g. Palmer et al, ; Parsons et al, ), and we therefore refer to it here as an index of relative abundance.…”

Using lorelograms to measure and model correlation in binary data: Applications to ecological studies

“…Defining a time‐lag to independence is crucial when estimating capture rate (i.e. the number of independent events within a certain time frame), an index of abundance often used in studies of unmarked populations (Carbone et al, ; O'Brien, Kinnaird, & Wibisono, ; Palmer, Swanson, Kosmala, Arnold, & Packer, ). The lorelogram could also help guide data aggregation decisions in occupancy studies .…”

The African Lion: A Long History of Interdisciplinary Research

“…The huge archive of labeled images also allowed for the development of new methods for computer-based image recognition (Norouzzadeh et al, 2018). The success of SnapshotSerengeti has so far resulted in over 40 different camera-trap projects on Zooniverse, and we have recently expanded the original project to become SnapshotSafari (see www.SnapshotSafari.org), which provides species classifications for dozens of camera-trap grids in multiple African countries with the ultimate goal of helping wildlife managers assess the population status of the lions' primary prey species in their respective reserves (see Palmer et al, 2018).…”