Boundary detection in an arid grazing landscape from the perspective of plants and small mammals in the Monte central Desert, Argentina

Background/Question/Methods

Land degradation in drylands modifies landscape structure, generating both contrasting patches and patches characterized by gradual change. At boundaries, such changes have consequences for ecosystem functioning because boundary dynamics are key drivers of landscape change. The aims of this study were 1) to detect the presence of boundaries created by grazing management in order to spatially delimit states of degradation in a temperate desert of Argentina; and 2) to determine the influence of grazing boundaries on small mammal assemblages. We considered three combinations of boundaries between habitats under passive restoration and continuous grazing: 1) in restored conditions (shrubland / woodland; SWr), 2) under continuous grazing conditions (shrubland-woodland; SWg) and 3) between passive restoration and grazing management (woodland; Wrg). For each particular type of habitat we also selected sampling sites over 200 m away from the boundaries (interior habitats). On each site, rodents were live trapped with Sherman-type traps and, in each trapping season, vegetation was measured using Braun Blanquet method. The rate of change of vegetation cover for regularly spaced data was analyzed with boundary detection algorithm. A generalized linear mixed model was used to identify the factors determining the presence of small mammals. 

Results/Conclusions

The moving split windows analysis indicated occurrence of sharp discontinuities in all boundaries. SWr and SWg showed higher variability among vegetation variables, and no clearly marked boundary was found for all variables at the same time. This reflects their complexity and natural heterogeneity. In Wrg, shrubs and brushwood represented the greatest change, comparing both management situations. In all, 230 individuals from 4 species (Cricetidae) were captured (25,200 trap-nights). Within SWr, shrubs, grasses and brushwood were related to presence of rodents, while bare soil, herbs and trees were negatively related to it. In SWg, rodent presence was higher within the boundary. In Wrg, this presence was associated with a lesser amount of bare soil and occurred in the restored part of the woodland. In the interior woodland, both restored and under management, rodent presence was related to grasses and herbs. In the grazed woodland, bare soil affected the assemblage negatively as well. In the interior shrubland, the explanatory variable was herbs. The net effect of grazing management on boundary habitats was to reduce their functional size by creating an edge effect. Conversely, in SWr, the similarity of vegetation structure among areas adjacent to the boundary could provide favorable microhabitats for species.