Home range and shelter use by Sonoran Desert Tortoises at 2 populations in Arizona

At Sugarloaf we monitored female tortoises (range 184–289 mm midline carapace length [CL]) weekly using radio telemetry in 1991–1993 and 1997–2005 as part of a reproductive ecology study (Averill-Murray 2002; Averill-Murray et al. 2018). During the 11 years of radio-tracking, all burrows used by both telemetered and opportunistically encountered tortoises were marked with numbered aluminum tags (n = 522). We use the term “burrow” specifically to refer to a subsurface cavity &gt;1/2 the tortoise’s length either formed by erosion or excavated by a tortoise or another animal (Burge 1978). We categorized burrow types as “rock” (cover provided by rocks or boulders, including large boulder piles in which the tortoise could not be visualized), “soil” (cover provided by soil or vegetation with no substantial rock component), and “packrat” (freestanding white-throated woodrat Neotomoa albigula nest independent of other shelter types; packrat nests inside rock shelters were categorized as “rock”). We did not individually mark “pallets” (shallow, scraped out areas &lt;1/2 tortoise length) or other temporary shelters unmodified by the tortoise (e.g., under trees, shrubs, or rock overhangs).At FMR, as part of a habitat-use study we attached radio transmitters (ATS, AVM, Telonics, or Wildlife Materials) to the anterior carapace of adult tortoises as above (Riedle et al. 2008). During the winter months (November through February) when tortoises were inactive, we located tortoises once a week. During the activity season (March through October) we located tortoises 2–3 times weekly, obtaining both morning and evening locations. From 2000 to 2003, we searched areas in which tortoises might occur within FMR, concentrating on sites suitable for burrow excavation, especially including incised washes with caliche caves and the volcanic hill. We also searched all washes within the study area, whether incised or not, and spent considerable time on the alluvial fans. We marked burrows in which we observed tortoises with individually numbered aluminum tags (n = 124), and we mapped locations of all caliche caves large enough to shelter a tortoise ≥180 mm CL (n = 463). Burrow types included caliche, rock, soil, and packrat.In order to directly compare patterns of space and burrow use between sites and to minimize potential variation due to inter-annual environmental variation, we estimated home ranges at both sites for the years of common study, 2000–2003. We only included adult tortoises (all &gt;220 mm CL) in the analysis, and we excluded observations between the first and last dates of hibernation each year, estimating the date that each tortoise terminated hibernation as the last day the tortoise was observed inside or &lt; 10 m from its hibernaculum. We calculated variograms, fit movement models, and estimated home ranges with area-corrected, optimally weighted, autocorrelated kernel density estimation (wAKDEC) using the <i>ctmm</i> package in R. We estimated total home ranges across the entire study period using perturbative Hybrid REML (pHREML) wAKDE_C conditional on the selected underlying movement model for each tortoise. We estimated core areas as the area encompassed by the 50% AKDE isopleth, the proportion of the total (95%) home range area contained by the 50% core area (PA), and the intensity of core area use (ICU = core area isopleth/[50% core area/95% AKDE area]). We used the ‘meta’ function in <i>ctmm</i> to calculate population-level home range and core area estimates that account for estimation uncertainty, as well as overall effect sizes between sites and sex-specific effect sizes within and between sites. We calculated the number of burrows available to each tortoise and estimated burrow densities by clipping vectors of the mapped burrows to the 50% AKDE core areas, the 95% AKDE home ranges, and the 50–95% non-core areas in QGIS 3.34.1. <b>Location data are not included because </b><b><i>Gopherus morafkai</i></b><b> is a protected species. Inquiries for these data should be made to the Arizona Game and Fish Department.</b>We compared home ranges estimated in this study with 1) MCP estimates for the same individuals for up to four years at FMR reported by Riedle et al. (2008) and 2) AKDE estimates based on up to 10 years of data for the same individuals at Sugarloaf reported by Averill-Murray et al. (2020). To compare MCP and AKDE estimates at FMR, we fit a logistic regression of the ratio of MCP to AKDE estimates against effective sample size (N ̂area) using the <i>quantreg</i> package in R. A good fit of the data to a logistic curve would indicate that MCP estimates underestimate AKDE estimates at low N ̂area. We conducted a repeated-measures ANOVA of the paired estimates from each tortoise using the <i>rstatix</i> package in R. A significant test result would indicate that home-range estimates from the 4-year subset of data are not equivalent to estimates from the full dataset.We hypothesized that adult AKDE home ranges were equivalent between sexes and sites and that ranges were not affected by the number of available burrows (loge transformed). We used the R package <i>metafor</i> in mixed-effects, within-study meta-analyses to explicitly account for the variation (heterogeneity) among the true effects arising from estimated uncertainty in home-range size within individual tortoises. We conducted a permutation test to validate the robustness of the final model and to adjust inferences accordingly. We compared density of the total number of burrows within AKDE home ranges between sites and sexes with generalized least squares regression (GLS) using R package <i>nlme</i>. We repeated the meta-analysis for 50% core areas and 50–95% non-core areas and followed by assessing burrow density versus sex, site, and home range portion with repeated-measures ANOVA. We also compared the relationship of ICU to the number of burrows within core areas and between sites and sexes with a generalized linear model and Gamma distribution with a logarithmic link function in R package <i>glmmTMB</i>.We used GLS in <i>nlme</i> to evaluate the hypotheses that the number of burrows used by individual tortoises did not differ between sites, sexes, or relative to the number of burrows available within their 95% AKDE home ranges, 50% core areas, and 50–95% non-core areas.