Background/Question/Methods Small mammals in desert environments are important links in ecosystem energy flow, with their densities often effecting the abundance of their primarily predators.  Granivorous small mammals also affect seed dispersal and can shape vegetation characteristics.  Consequently, factors affecting small mammal abundance are important in understanding community and ecosystem structure and dynamics.  Small mammal abundance has been attributed to levels of precipitation/primary productivity (bottom up) and/or removal by predation (top down).  However, small mammal abundance can vary greatly over the landscape in relatively short (< 5 km) distances.  Similarities in precipitation amounts and predator abundance on this small scale do not explain this within-landscape variation in abundance.  However, on this scale, habitat heterogeneity can be high. Under the landscape of fear model, such habitat heterogeneity can result in proportionally different amounts of safe vs risky habitat and consequently, predation risk among adjacent areas.  We predicted that density of small mammals in an area is inversely related to the proportion of risky habitat/level of predation risk of an area.  We tested this prediction with kangaroo rats (Dipodomys spp.) in two geographically separate areas of the Chihuahaun Desert:  Mapimi Biosphere reserve in Durango Mexico and the Jornada Experimental range in New Mexico.  We measured giving up densities (GUDs) in 8 distinct areas of similar shrub/grass habitat with different densities of kangaroo rats. GUD’s were measured with commercial 50 x30 x 10 cm aluminum buffet trays containing 3 kg of sand and 4 g of millet seed.  The specific prediction was that GUDs would be inversely related to kangaroo rat densities.
Results/Conclusions  In the Jornada Experimental range we found significantly lower GUDs in the high density grids (2.1 ± 0.1 g, SE vs 3.3 ± 0.1g, group t = 8.89, P ≤ 0.001).  In Mapimi we found a significant negative relationship between kangaroo rat density and GUDs (P = 0.026, r² = 0.924, n = 4).  We conclude that kangaroo rat density is inversely related to predation risk.  Further research will test if this relationship is because of lethal factors, i.e. predators can kill more animals in areas of higher risk, or non-lethal factors, i.e. predation risk per se reduces the use of an area by prey.