Desertification involves a shift in dominant vegetative composition from an ecosysytem dominated by perennial grasses to one dominated by shrubs and bare soil [Daily 1995; Van Auken 2000; Scheffer et al. 2001; Jackson et al. 2002]. Thus, desertification is the process in which grasslands are converted into shrublands. Desertification is widespread in arid systems throughout the world and thought to be largely irreversible over timescales relevant to management because few studies have reported natural vegetation recovery in desertified grasslands [Laycock 1991; Van Auken 2000; Rasmussen et al. 2001; Valone et al. 2002]. Mechanisms to explain desertification include changes in climate, herbivory and fire regime, but most workers agree that overgrazing by livestock has played a major role [Schlesinger et al. 1990; Laycock 1991; Fleischner 1994; Archer et al. 1995; Daily 1995; Van Auken 2000].
In attempting to explain the apparent irreversibility of desertification, theoretical models typically focus on changes in soil properties [e.g., Walker et al. 1981; Schlesinger et al. 1990; Rosenfeld et al. 2001; van de Koppel and Rietkerk 2004]. Many models are based on a positive feedback between grass cover and water infiltration rate [e.g., Kelly and Walker 1976; Walker et al. 1981; Breman and de Wit 1983; Rietkerk and van de Koppel 1997; Rietkerk et al. 1997; van de Koppel et al. 1997]. These models demonstrate that once grass cover is reduced by livestock grazing or drought beyond a threshold that results in desertification, grass recovery is inhibited because water infiltration rates are insufficient for grass re-establishment [e.g., Walker et al. 1981; van de Koppel and Rietkerk 2000; Scheffer et al. 2001; Rietkerk et al. 2002; van de Koppel et al. 2002; van de Koppel and Rietkerk 2004]. Such alternate stable state models have come to dominate the desertification literature (e.g., Briske et al. 2005; Peters et al. 2006).
However, grass cover is only one mechanism known to affect water infiltration rate. A large literature has demonstrated that soil compaction can significantly reduce water infiltration rate [e.g., Rhoades et al. 1964; Rauzi and Hanson 1966; Wood and Blackburn 1981; Warren et al. 1986; Abdel-Magid et al. 1987; Gamougoun 1984; Schlesinger 1990; Fleischner 1994]. Although theoreticians do acknowledge the effect of livestock trampling on soil compaction, it is typically assumed to be negligible in relation to the effects of grass canopy cover on water infiltration rate.
Recently, four studies have reported significant increases in perennial grass abundance at formerly desertified sites: i.e., reversals of desertification. Fuhlendorf et al.  reported significant increases in perennial grasses in a 25-year livestock exclosure in Texas, U.S.A. while Valone et al.  reported a 300% increase in perennial grass cover at a 39-year livestock exclosure in Arizona U.S.A. Similarly, a 20-year reduction in human activities, including livestock grazing, was associated with significant vegetation recovery at a desertified site in the African Sahel [Rasmussen et al. 2001, Herrmann et al. 2005] while 30 years of protection from livestock resulted in perennial grass recovery at a desertified site in China [Zhang et al. 2005]. These observations suggest that desertification is potentially reversible, at least in some circumstances. Existing desertification models cannot account for these changes in vegetation or their timescale.
Castellano and Valone  recently suggested a novel mechanism to explain the above reversals of desertification. They postulate that long-term livestock removal permits recovery of soil compaction which leads to increased water infiltration rates and subsequent increases in perennial grass abundance. They evaluated this hypothesis by comparing soil compaction, water infiltration rates and vegetation at three nearby grazing exclosures that differed in time since livestock removal [Castellano and Valone 2007]. The three sites, all within 6 km of one another and on similar soils, had been free of livestock grazing for 10, 26 and 45 years. At each site, they measured water infiltration, soil compaction and vegetation inside and outside the long-term grazing exclusion fence. At all three sites, water infiltration was higher and soil compaction lower inside compared to outside the grazing fence. These differences were significant for the 26 and 45 year exclusion sites but not the 10 year exclusion site. Furthermore, a comparison across sites revealed that relative water infiltration (inside [ungrazed] versus outside [grazed] measurements) and relative soil compaction (outside versus inside measurements) increased significantly with time since livestock exclosure. In other words, relative water infiltration rates were highest and relative soil compaction was lowest inside the oldest livestock exclosure. In addition, while all three sites were desertified at the beginning of the 20th century, only the 45 year exclusion site had experienced a recovery in perennial grasses. Of note, that site had experienced no perennial grass recovery during the first 20 years of livestock exclosure [Valone et al. 2002]. Thus, these data suggest that in these sites, more than two decades of livestock removal are required for sufficient reductions in soil compaction and concomitant increases in water infiltration to support perennial grass recovery.
The Castellano and Valone  mechanism suggests that perennial grasses can recover on desertified sites following long-term livestock removal via increased water infiltration rates that occur due to subsequent recovery from soil compaction. Additionally, this mechanism can also explain the multi-decadal time lags observed between the reduction of livestock grazing and perennial grass recovery [e.g. Fuhlendorf et al. 2001; Rasmussen et al. 2001; Valone et al. 2002; Zhang et al. 2005] because soils in arid systems recover slowly from compaction. Release from compaction requires freeze-thaw or wet-dry cycles and the frequency of such cycles decreases with increasing aridity and increasing temperature [Seybold et al. 1999].
While the results of Castellano and Valone  are insightful, they come from three nearby sites in one valley in Arizona (the San Simon), a valley that experienced dramatic desertification at the end of the 19th century [Bahre 1991]. As such, one possibility is that the patterns observed are an artifact of desertification and may not be general. Clearly, their proposed mechanism requires further evaluation and testing to assess its generality. Here, I further evaluate the effect of long-term livestock removal on water infiltration rates by reporting data from five additional sites in southern Arizona. The sites differ in time since livestock removal and degree of historic desertification but are similar in soil type and elevation.
Was this article helpful?