Low resilience at the early stages of recovery of the semi-arid Chaco forest—Evidence from a field experiment

Abstract

Resilience "the capacity of an ecosystem to recover from disturbance" is a popular concept but quantitative empirical studies are still uncommon. This lack of empirical evidence is especially true for semi-arid ecosystems in the face of the combined and often confounding impacts of land use and climate changes. We designed a methodology to disentangle vegetation responses to land-use exclusion and weather variability, and piloted it at the southern extreme of the Gran Chaco forest, the most extensive seasonally dry forest in South America. We established 16 pairs of neighbouring fenced and unfenced plots in four ecosystem types resulting from different long-term land-use regimes under the same climate and on highly similar soil parental material. From lower to higher land-use intensity, related with logging and livestock grazing and trampling, these types were: primary forest (no land use in the last 50 years), secondary forest, closed species-rich shrubland and open shrubland. In each plot we monitored plant species composition during the first 5 years following land-use exclusion, and evaluated the resilience as the rate of change of vegetation towards the primary forest, considered as the reference ecosystem. We found that during the first 5 years of exclusion and despite the high rainfall, only grass cover in the secondary forest showed positive resilience (recovery towards the reference ecosystem). The rest of the variables in the other ecosystem types showed either no significant change (null resilience) or even transitioned away from the reference state (negative resilience). Synthesis. The lack of detectable recovery after 5 years of exclusion suggests that (a) long-term land use, even at lower intensities, has affected the sources of resilience of this ecosystem; (b) rainy periods do not necessarily speed up recovery as suggested in the literature; and (c) study designs should incorporate the variation of the reference ecosystem in order to differentiate the effect of land use from other factors in a context of climate change. Although still confined to the early post-disturbance stages, our findings suggest that recovery of these systems may be slower and more complicated than predicted in the literature on the basis of space-for-time substitutions.