Landscape Effects on Carnivore Community Dynamics in an Agro-prairie Ecosystem PDF Download

Author: Ty Jordan Werdel
Publisher:
ISBN:
Category :
Languages : en
Pages : 0

Get Book Here

Book Description
The Great Plains region has undergone extensive conversion of native prairies to agriculture production and energy development since European colonization. Temperate prairies, including remaining prairies within the Great Plains, are considered among Earth's most imperiled ecosystems. Prairie patches now exist as components of a landscape mosaic proportionately dominated by cultivated agriculture. These contemporary human-modified landscapes may structure species' distributions, influence community dynamics, and supplant established abiotic range-limiting processes. Understanding the direction and scale of these processes, and how they are affected by landscape composition and configuration, is necessary to enhance conservation efforts. Carnivore communities may be most affected by landscape changes due to negative interactions with humans and their inherent biological traits; however, information regarding landscape-scale effects on the existing suite of carnivores in the Great Plains is lacking. I examined how landscape composition and characteristics influenced site occupancy probabilities and turnover rates by swift foxes (Vulpes velox), the spatial and temporal interactions between swift foxes and coyotes (Canis latrans), and carnivore richness in agro-prairie ecosystems. Additionally, I strategically identified native prairie areas to focus conservation and management of remaining swift fox habitat. During 2018-2020, I used detection/non-detection data from camera traps at 381 randomly selected sites distributed throughout a landscape mosaic comprising the westernmost 31 counties (7.16 million ha) of Kansas, USA. I subsequently used presence/absence data from these sites across three years to infer species-specific responses to landscape change and carnivore community dynamics. To evaluate effects of landscape composition and configuration on site occupancy probabilities and turnover rates by swift fox, I used a distance-weighted scale of effect of landscape metrics within multi-season occupancy models. Swift foxes were more likely to occur at sites with moderate landcover diversity within 254.47 ha, greater proportion of shortgrass prairie (7.07 ha) and loamy soil types (0.79 ha), and lower proportions of Conservation Reserve Program (CRP) landcover (78.54 ha). Swift foxes were more likely to colonize sites with less diverse landcover, a greater proportion of loamy soil types, and lower proportions of CRP landcover. Swift foxes were insensitive to the proportion of row-crop agriculture surrounding sites (3.14 ha). To evaluate landscape composition effects on swift foxes and coyote (the apex predator in the region) spatiotemporal interactions, I used a Bayesian hierarchical multi-season occupancy model to evaluate spatial interactions, and a coefficient of overlap of temporal activity to assess factors affecting temporal interactions. Mean persistence of swift foxes differed across sites where coyotes were not detected (0.66; SE = 0.001) and where coyotes were detected (0.39; SE=0.001). The coefficient of overlap at sites surrounded by lower proportions of CRP (≥0.10) differed (95% CIs did not overlap) from coefficient of overlap of all other landscape effects. The spatial distribution of swift foxes was positively influenced (Species Interaction Factor [SIF] > 1) by coyote presence through space and time at low proportions of CRP (≤0.04). SIF decreased as proportion of CRP increased; however, Bayesian confidence intervals overlapped SIF = 1, suggesting that swift foxes were spatially distributed independent of coyotes through space and time at greater proportions of CRP (>0.04). I used a structural equation model to test hypotheses of multiple direct and indirect relationships between landscape composition and configuration and prey availability on carnivore richness. My hypothesized model (X2 = 23.92, df = 24, P = 0.47) explained 27% of the variance of carnivore richness. Agriculture, native prairie, landcover diversity, CRP, water availability, prey occurrence, and sampling effort all had direct positive effects on my measure of carnivore richness, while loamy tableland soil had only an indirect effect. To strategically identify native prairie areas for conservation of swift fox habitat, I created a predicted swift fox occupancy map based on my most-supported, stacked single-season occupancy model. I identified predicted occupancy rate (range = 0.01-0.46) where sensitivity equaled specificity (0.09) within a receiver operating characteristic curve, and reclassified the predicted occupancy map to include only predicted occupancy rates >0.09, and again for a more targeted approach with predicted occupancy rates >0.18. These two maps were intersected with a map of grassland proportions >0.60 to identify areas that were expected to have relatively high occupancy and survival rates by swift fox. Swift foxes were more likely to occur at sites with low levels of landscape diversity ([Beta] = -0.411 ± 0.140), greater proportions of native grassland ([Beta] = 0.375 ± 0.154) and loamy tableland soils ([Beta] = 0.944 ± 0.188), and lower proportions of CRP landcover ([Beta] = -1.081 ± 0.360). Identified native grassland conservation areas totaled 84,420.24 ha (mean patch size = 162.66 ha [SE = 29.67]). Conservation areas located on privately owned working lands included 82,703.86 ha, while conservation areas located within the boundaries of federal, state, and non-governmental organizations (NGO) parcels included 1,716.38 ha. My results provide a unique understanding of how landscape composition and configuration, intraguild competition, and prey availability drive carnivore community dynamics in agro-prairie ecosystems. Additionally, my research elucidated constraints to range expansions for an iconic prairie-obligate carnivore (swift fox) at the edge of their range, while also identifying areas for strategic conservation for their populations.