Get access

Contingent effects of water balance variation on tree cover density in semiarid woodlands


*Correspondence: Andrew J. Kerkhoff, Department of Ecology and Evolutionary Biology, University of Arizona, 1041 E. Lowell St., BSW 310, Tucson, AZ 85721. Tel.: +1 520 626 3336. Fax. +1 520 621 9190. E-mail:


Aim  The local distribution of woody vegetation affects most functional aspects of semiarid landscapes, from soil erosion to nutrient cycling. With growing concern about anthropogenic climate change, it has become critical to understand the ecological determinants of woody plant distribution in semiarid landscapes. However, relatively little work has examined the determinants of local variation in woody cover. Here we examine water balance controls associated with patterns of tree cover density in a topographically complex semiarid woodland.

Location  Los Pinos Mountains, Sevilleta National Wildlife Refuge LTER, New Mexico, USA.

Methods  To explore the relationship between local water balance variation and tree cover density, we used a combination of high-resolution (1 × 1 m), remotely sensed imagery and quantitative estimates of water balance variation in space and time. Regression tree analysis (RTA) was used to identify the environmental parameters that best predict variation in tree cover density.

Results  Using six predictor variables, the RTA explains 39% of the deviance in tree cover density over the landscape. The relationship between water balance conditions and tree cover density is highly contingent; that is, similar tree cover densities occur under very different combinations of water balance parameters. Thus, the effect of one environmental parameter on tree cover density depends on the values of other parameters. After tree cover density is adjusted for water balance conditions, residual variation is related to tree cover density in the neighbourhood of a particular location.

Conclusion  In semiarid landscapes, vegetation structure is largely controlled by water supply and demand. Results presented here indicate that localized feedbacks and site-specific historical processes are critical for understanding the responses of semiarid vegetation to climate change.