A stable home: Autocorrelated Kernel Density Estimated home ranges of the critically endangered Elongated tortoise

Abstract

Home range analysis is a standard and fundamental concept in ecology used to describe animal space use over their lifetimes. Connecting home range sizes with animal characteristics, location, and habitat can be used to inform conservation decisions. Reptiles are frequently lacking robust estimates of space use, particularly reptiles in tropical regions. Here we analyse a publicly available dataset, collected by the authors of this study, describing the movements of Critically Endangered Elongated tortoises Indotestudo elongata. The tortoise data included the locations of 17 tortoises (12 females, 5 males) collected on average once every three days for an average duration of 353.76 SE ± 33.10 days. We use these data to estimate the home range of Elongated tortoise, and explore how tortoise size and sex influences home range size. To mitigate issues resulting from low effective sample sizes and low temporal resolution of the data, we used a modern home range estimation method – Autocorrelated Kernel Density Estimators (AKDE). We found 14 of 17 individuals appear to be occupying a stable home range (using variograms to determine range residency). The average AKDE home range for all 14 individuals with range residency was 44.81 ± 10.44 ha. Bayesian Regression Models suggest comparable size estimates between male and female home ranges, despite males being physically larger than females in both mass and carapace length. These AKDE home range estimates have the added utility of being more comparable with other studies, less susceptible to errors from a suboptimal tracking regime, and are well positioned for inclusion in future meta-analyses. Keywords: testudine, autocorrelated kernel density estimator, spatial ecology, space use, Thailand, Indotestudo elongata

The Importance of Representative Sampling for Home Range Estimation in Field Primatology

“…A similar sampling duration (45-136 days) was required in a study on giant anteaters (Myrmecophaga tridactyla), where the average home range crossing time was approximately 2 days (Giroux et al, 2021). In contrast, a study on elongated tortoises (Indotestudo elongata) estimated a home range crossing time of 17 days on average (although sometimes much longer for particular individuals), and even with 1 year of consistent sampling, researchers were not able to achieve adequate effective sample sizes for several individuals (Montano et al, 2021). Studies on Mongolian gazelles have shown estimated range crossing times of around 5-6 months (Fleming et al, 2014a).…”

“…A thorough evaluation of the AKDE compared against commonly used approaches, such as the MCP, KDE, and Local Convex Hull (LocoH), shows that the AKDE is superior in capturing space use out of sample and is the only estimator capable of producing unbiased estimates with low effective sample sizes . While AKDE is increasingly adopted in studies of animal ranging behavior (Crabb et al, 2022;Desbiez et al, 2020;Lenske & Nocera, 2018;McEvoy et al, 2019;Montano et al, 2021;Naveda-Rodríguez et al, 2022;Poessel et al, 2022), the overwhelming majority of studies still use methods, such as the MCP and KDE (Börger et al, 2020;Fleming & Calabrese, 2017;Hemson et al, 2005;Laver & Kelly, 2010;Powell, 2000). The potential presence of significant, systematic biases from applying these methods to estimate home-range area is concerning, because it can lead to erroneous conservation decision-making (Gaston et al, 2008) or inaccurate meta-analyses that influence public opinion and theoretical frameworks (Noonan et al, 2020).…”

Spatial ecology to strengthen invasive snake management on islands

“…To check for AKDE's assumptions on range residency 47 , we first built up individual variograms using the ggplot2 v.3.35 package 48 and visually decided which ones we considered were stable. Secondly, we explored the effective sample size of all individuals (DOF area resulting from the 'akde' function) and retained for home range calculations only those with a value greater than 10 (following Montano et al 49 ). We also calculated home range using the Minimum Convex Polygon (MCP) 50 (100%, 95%, and 50%) and Kernel Density Estimators (KDE) 51 (95% and 50%) to allow comparisons with previous studies.…”

Home range ecology of Indian rock pythons (Python molurus) in Sathyamangalam and Mudumalai Tiger Reserves, Tamil Nadu, Southern India

“…We generated R package citations with the aid of grateful v.0.0.3 61 . The code we used was based upon that used by 62 that is available at https:// osf. io/ rxu6f/.…”