) who 34 made our research possible. We also thank data proofers/entry (J. Whitaker, A. Blackert, S. 35Ramakrishnan). We are grateful to anonymous reviewers and to D. Westneat and P. Porneluzi for their 36 reviews that greatly improved our manuscript. 37 AbstractThe

The factors that affect resource selection by a foraging herbivore can vary according to the resources or conditions associated with particular levels of organization in the environment, and to the scales over which the herbivore perceives and responds to those resources and conditions. To investigate the role of forage in this hierarchical process, we studied resource selection by a mixed‐feeding large herbivore, the impala (Aepyceros melampus). We focussed on three spatial scales: plant species, feeding station and feeding patch. In paired sites where impala were and were not observed, we identified the plant species from which animals fed, the attributes of the plants, and the characteristics of the broader site. Across all three scales, plant species available as forage was central in determining resource selection by impala. At the species level, that effect was modified by the nutritional quality (greenness) and whether it was during a period of forage abundance or scarcity (season). At the feeding‐station level, overall greenness and biomass of the station were important, but their effects were modified by the season. At the feeding‐patch level, broader‐scale factors such as the type of vegetation cover had an important influence on resource selection. The grass Panicum maximum was a preferred forage species and a key resource determining the locations of feeding impala. Our findings support the idea that selection by a foraging herbivore at fine scales (i.e. diet selection) can have consequences for broader‐scale selection that result in observed patterns of habitat use and animal distribution.

Key factors affecting metapopulation dynamics of animals include patch size, isolation, and patch quality. For wetland‐associated species, hydrology can affect patch availability, connectivity, and potentially habitat quality; and therefore drive metapopulation dynamics. Wetlands occurring on natural river floodplains typically have more dynamic hydrology than anthropogenic wetlands. Our overall objective was to assess the multiyear spatial and temporal variation in occupancy and turnover rates of a semi‐aquatic small mammal at two hydrologically distinct wetland complexes. We live‐trapped marsh rice rats (Oryzomys palustris) for 3 yr and >50 000 trap nights at nine wetland patches on the Mississippi River floodplain and 14 patches at a reclaimed surface mine in southern Illinois. We used dynamic occupancy modeling to estimate initial occupancy, detection, colonization, and extinction rates at each complex. Catch per unit effort (rice rats captured/1000 trap nights) was markedly higher at the floodplain site (28.1) than the mining site (8.1). We found no evidence that temperature, rainfall, or trapping effort affected detection probability. Probability of initial occupancy was similar between sites and positively related to patch size. Patch colonization probability at both sites was related negatively to total rainfall 3 weeks prior to trapping, and varied across years differently at each site. We found interacting effects of site and rainfall on extinction probability: extinction increased with total rainfall 3 months prior to trapping but markedly more at the floodplain site than at the mining site. These site‐specific patterns of colonization and extinction are consistent with the rice rat metapopulation in the floodplain exhibiting a habitat‐tracking dynamic (occupancy dynamics driven by fluctuating quality), whereas the mineland complex behaved more as a classic metapopulation (stochastic colonization & extinction). Our study supports previous work demonstrating metapopulation dynamics in wetland systems being driven by changes in patch quality (via hydrology) rather than solely area and isolation.

Citation: van der Merwe, J., and E. C. Hellgren. 2016. Spatial variation in trophic ecology of small mammals in wetlands: support for hydrological drivers. Ecosphere 7(11):e01567. 10.1002/ecs2.1567Abstract. Food webs composed of similar consumers can vary based on nutrient input, habitat structure, and other factors. For wetland-associated species, fluctuating water levels can potentially affect habitat quality, which in turn can affect trophic diversity. Our objective was to determine spatiotemporal variation in the trophic structure of small mammals at two wetland complexes (floodplain and mineland) in southern Illinois. We live-trapped small mammals during 2011-2013 at nine wetland patches on the Mississippi River floodplain and 14 patches at a reclaimed mineland. We collected hair samples from six species of small mammals (n = 428) at these wetland complexes. We analyzed C and N stable isotopes for three mammal taxa (Oryzomys palustris, Peromyscus spp., Microtus ochrogaster) to compare diet between species, sites, and times. Food sources (vegetation and invertebrates) were collected to form the isotopic baseline. We found no seasonal difference in diet composition, but isotopic values varied between sites and species. At the floodplain site, both δ 15 N and δ 13 C isotopic values for Oryzomys were more variable and completely enclosed that of Peromyscus. At the mining site, Peromyscus were at a lower trophic level (δ 15 N) and had a separate and less variable δ 13 C values from Oryzomys. Microtus was at a lower trophic position than the other two species at both sites. These results point to reduced niche overlap between Oryzomys and Peromyscus at the mining site, perhaps due to lower habitat quality and limited suitable resources. At the floodplain site, we conclude that more dynamic hydrology gave rise to higher biodiversity and more resources allowing small mammals to use similar food items.