Search

The invasive grass cheatgrass (Bromus tectorum L.) presents major challenges for land management and habitat conservation in the western United States. Feral horses (Equus ferus caballus) have become overabundant in some areas of the West and can impact fragile semiarid ecosystems. Amid ongoing efforts to control cheatgrass in the Great Basin, we conducted a study to determine if feral horses contribute to the spread of cheatgrass through distribution via their feces. We collected feral horse fecal samples from Little Book Cliffs Herd Management Area in western Colorado in 2014. Fecal samples were dried, and 20 from each of 3 collection sessions were cultivated to examine germination success. Six species germinated from 18 samples (30%; mostly one plant per sample where germination occurred), including cheatgrass from 8% of samples. In a separate study we examined the diet of this same horse population using fecal plant DNA barcoding. Plant species that germinated were rare in the diet and germinated from fewer samples than expected relative to their detection in the diet. Our results suggest that feral horses could be contributing to cheatgrass propagation. Native ungulates and domestic cattle also have this potential. Although management of all large ungulates is necessary to mitigate cheatgrass spread, control of feral horse numbers is particularly necessary.

Evaluating the conditions where a species can persist is an important question in ecology both to understand tolerances of organisms and to predict distributions across landscapes. Presence data combined with background or pseudo-absence locations are commonly used with species distribution modeling to develop these relationships. However, there is not a standard method to generate background or pseudo-absence locations, and method choice affects model outcomes. We evaluated combinations of both model algorithms (simple and complex generalized linear models, multivariate adaptive regression splines, Maxent, boosted regression trees, and random forest) and background methods (random, minimum convex polygon, and continuous and binary kernel density estimator (KDE)) to assess the sensitivity of model outcomes to choices made. We evaluated six questions related to model results, including five beyond the common comparison of model accuracy assessment metrics (biological interpretability of response curves, cross-validation robustness, independent data accuracy and robustness, and prediction consistency). For our case study with cheatgrass in the western US, random forest was least sensitive to background choice and the binary KDE method was least sensitive to model algorithm choice. While this outcome may not hold for other locations or species, the methods we used can be implemented to help determine appropriate methodologies for particular research questions.

Using localized studies to understand how ecosystems recover can create uncertainty in recovery predictions across landscapes. Large archives of remote sensing data offer opportunities for quantifying the spatial and temporal factors influencing recovery at broad scales and predicting recovery. For example, energy production is a widespread and expanding land use among many semi-arid ecosystems of the Western United States dominated by sagebrush (Artemisia spp.), a keystone species providing a variety of ecological services. With remotely-sensed (Landsat) estimates of vegetation cover collected every 2–5 years from southwestern Wyoming, USA, over nearly three decades (1985–2015), we modeled changes in sagebrush cover on 375 former oil and gas well pads in response to weather and site-level conditions. We then used modeled relationships to predict recovery time across the landscape as an indicator of resilience for vegetation after well pad disturbances, where faster recovery indicates a greater capacity to recover when similarly disturbed. We found the rate of change in sagebrush cover generally increased with moisture and temperature, particularly at higher elevations. Rate of change in sagebrush cover also increased and decreased with greater percent sand and larger well pads, respectively. We predicted 21% of the landscape would recover to pre-disturbance conditions within 60 years, whereas other areas may require >100 years for recovery. These predictions and maps could inform future restoration efforts as they reflect resilience. This approach also is applicable to other disturbance types (e.g., fires and vegetation removal treatments) across landscapes, which can further improve conservation efforts by characterizing past conditions and monitoring trends in subsequent years.