The scale and cause of spatial heterogeneity in strength of temporal density dependence

The importance of density dependence in natural communities continues to spark much debate because it is fundamental to population regulation. We used temporal manipulations of density to explore potentially stabilizing density dependence in early survivorship among six local populations of a tropical damselfish (Dascyllus flavicaudus). Specifically, we tested the premise that spatial heterogeneity in the strength of temporal density dependence would reflect variation in density of predators, the agent of mortality. Our field manipulations revealed that mortality among successive cohorts of young fishes was density dependent at each reef, but that its strength varied by approximately 1.5 orders of magnitude. This spatial heterogeneity was well predicted by variation among the six reefs in the density of predatory fishes that consume juvenile damselfishes. Because density dependence arose from competition for enemy-free space within a shelter coral, the mortality consequence of the competition depended on the neighborhood density of predators. Thus, the scale of heterogeneity in the density dependence largely reflected attributes of the environment that shaped the local abundance of predators. These results have important implications for how ecologists explore regulatory processes in nature. Failure to account for spatial variation could frequently yield misleading conclusions regarding density dependence as a stabilizing process, obscure underlying mechanisms influencing its strength, and provide no insight into the spatial scale of the heterogeneity. Further, models of population dynamics will be improved when experimental approaches better estimate the magnitude and causes of variation in strength of stabilizing density dependence.     

Spatial heterogeneity in distribution and ecology of Western Palearctic birds

Species vary in abundance and heterogeneity of spatial distribution, and the ecological and evolutionary consequences of such variability are poorly known. Evolutionary adaptation to heterogeneously distributed resources may arise from local adaptation with individuals of such locally adapted populations rarely dispersing long distances and hence having small populations and small overall ranges. We quantified mean population density and spatial heterogeneity in population density of 197 bird species across 12 similarly sized regions in the Western Palearctic. Variance in population density among regions differed significantly from a Poisson distribution, suggesting that random processes cannot explain the observed patterns. National estimates of means and variances in population density were positively correlated with continental estimates, suggesting that means and variances were maintained across spatial scales. We used Morisita's index of population abundance as an estimate of heterogeneity in distribution among regions to test a number of predictions. Heterogeneously distributed passerine bird species as reflected by Morisita's index had small populations, low population densities, and small breeding ranges. Their breeding populations had been consistently maintained at low levels for considerable periods of time, because the degree of genetic variation in a subsample of non-passerines and passerines was significantly negatively related to heterogeneity in distribution. Heterogeneously distributed passerine species were not more often habitat specialists than homogeneously distributed species. Furthermore, heterogeneously distributed passerine species had high annual adult survival rates but did not differ in annual fecundity from homogeneously distributed species. Heterogeneously distributed passerine species rarely colonized urban habitats. Finally, homogeneously distributed bird species were hosts to a greater diversity of blood parasite species than heterogeneously distributed species. In conclusion, small breeding ranges, population sizes, and population densities of heterogeneously distributed passerine bird species, combined with their low degree of genetic variability, and their inability to colonize urban areas may render such species particularly susceptible to human-influenced global climatic changes.     

Density dependence in marine fish populations revealed at small and large spatial scales

Experimental manipulation of population density has frequently been used to demonstrate demographic density dependence. However, such studies are usually small scale and typically provide evidence of spatial (within-generation) density dependence. It is often unclear whether small-scale, experimental tests of spatial density dependence will accurately describe temporal (between-generation) density dependence required for population regulation. Understanding the mechanisms generating density dependence may provide a link between spatial experiments and temporal regulation of populations. In this study, I manipulated the density of recently settled kelp rockfish (Sebastes atrovirens) in both the presence and absence of predators to test for density-dependent mortality and whether predation was the mechanism responsible. I also examined mortality of rockfish cohorts within kelp beds throughout central California to evaluate temporal (between-generation) density dependence in mortality. Experiments suggested that short-term behavioral responses of predators and/or a shortage of prey refuges caused spatial density dependence. Temporal density dependence in mortality was also detected at larger spatial scales for several species of rockfish. It is likely that short-term responses of predators generated both spatial and temporal density dependence in mortality. Spatial experiments that describe the causal mechanisms generating density dependence may therefore be valuable in describing temporal density dependence and population regulation.     

Empirical modeling of atmospheric deposition in mountainous landscapes.

Atmospheric deposition has long been recognized as an important source of pollutants and nutrients to ecosystems. The need for reliable, spatially explicit estimates of total atmospheric deposition (wet + dry + cloud) is central, not only to air pollution effects researchers, but also for calculation of input-output budgets, and to decision makers faced with the challenge of assessing the efficacy of policy initiatives related to deposition. Although atmospheric deposition continues to represent a critical environmental and scientific issue, current estimates of total deposition have large uncertainties, particularly across heterogeneous landscapes such as montane regions. We developed an empirical modeling approach that predicts total deposition as a function of landscape features. We measured indices of total deposition to the landscapes of Acadia (121 km2) and Great Smoky Mountains (2074 km2) National Parks (USA). Using approximately 300-400 point measurements and corresponding landscape variables at each park, we constructed a statistical (general linear) model relating the deposition index to landscape variables measured in the field. The deposition indices ranged over an order of magnitude, and in response to vegetation type and elevation, which together explained approximately 40% of the variation in deposition. Then, using the independent landscape variables available in GIS data layers, we created a GIS-relevant statistical nitrogen (N) and sulfur (S) deposition model (LandMod). We applied this model to create park-wide maps of total deposition that were scaled to wet and dry deposition data from the closest national network monitoring stations. The resultant deposition maps showed high spatial heterogeneity and a four- to sixfold variation in "hot spots" and "cold spots" of N and S deposition ranging from 3 to 31 kg N x ha(-1) x yr(-1) and from 5 to 42 kg S x ha(-1) x yr(-1) across these park landscapes. Area-weighted deposition was found to be up to 70% greater than NADP plus CASTNET monitoring-station estimates together. Model-validation results suggest that the model slightly overestimates deposition for deciduous and coniferous forests at low elevation and underestimates deposition for high-elevation coniferous forests. The spatially explicit deposition estimates derived from LandMod are an improvement over what is currently available. Future research should test LandMod in other mountainous environments and refine it to account for (currently) unexplained variation in deposition.     

The triple helix of Plantago lanceolata: genetics and the environment interact to determine population dynamics

The theory of evolution via natural selection predicts that the genetic composition of wild populations changes over time in response to the environment. Different genotypes should exhibit different demographic patterns, but genetic variation in demography is often impossible to separate from environmental variation. Here, we asked if genetic variation is important in determining demographic patterns. We answer this question using a long-term field experiment combined with general linear modeling of deterministic population growth rates (lambda), deterministic life table response experiment (LTRE) analysis, and stochastic simulation of demography by paternal lineage in a short-lived perennial plant, Plantago lanceolata, in which we replicated genotypes across four cohorts using a standard breeding design. General linear modeling showed that growth rate varied significantly with year, spatial block, and sire. In LTRE analysis of all cohorts, the strongest influences on growth rate were from year x spatial block, and cohort x year x spatial block interactions. In analysis of genetics vs. temporal environmental variation, the strongest impacts on growth rate were from year and year x sire. Finally, stochastic simulation suggested different genetic composition among cohorts after 100 years, and different population growth rates when genetic differences were accounted for than when they were not. We argue that genetic variation, genotype x environment interactions, natural selection, and cohort effects should be better integrated into population ecological studies, as these processes should result in deviations from projected deterministic and stochastic population parameters.     

The roles of competition and environmental heterogeneity in the maintenance of behavioral variation and covariation

Many models of selection predict that populations will lose variation in traits that affect fitness. Nonetheless, phenotypic variation is commonly observed in natural populations. We tested the influences of competition and spatial heterogeneity on behavioral variation within and among populations of Merriam's kangaroo rats (Dipodomys merriami) and tested for the differential expression of trait correlations. We found that populations of D. merriami exhibited more aggression at sites with more competition. Contrary to theoretical predictions and empirical results in other systems, the sites with the greatest spatial heterogeneity and highest levels of competition did not exhibit the most behavioral variation among individuals. However, the greatest within-individual behavioral variability in boldness (response to cues of predator presence) was exhibited where spatial heterogeneity was highest. Aggression and boldness of D. merriami were highly repeatable, that is, individuals behaved in a consistent manner over time, and the two behaviors were also highly correlated. Interestingly, the strength of this correlation was greatest where the competitive community was least diverse. These findings add to increasing evidence that natural populations of animals exhibit patterns of behavioral covariance, or personality structure, and suggest that competitive variation may act to erode personality structure.     

Estimation of population parameters from aerial counts of North American mallards: a cautionary tale.

The accuracy of population estimates strongly interferes with our ability to obtain unbiased estimates of population parameters based on analyses of time series of population fluctuations. Here we use long-term data on fluctuations in the size of Mallard populations collected as part of the May Breeding Waterfowl Survey covering a large section of North America. We assume a log-linear model of density dependence and use a hierarchical Bayesian state-space approach in which all parameters are assumed to be realizations from a common underlying distribution. Thus, parameters for different populations are not allowed to vary independently of each other. We then simulated independent time series of aerial counts, using the estimated parameters and adding various levels of observation error. These simulations showed that the estimates of stochastic population growth rate and strength of density dependence were biased even when moderate sampling errors were present. In contrast, the estimates of the environmental stochasticity and the carrying capacity were unbiased even for short time series and large observation error. Our results underline the importance of reducing the magnitude of sampling error in the design of large-scale monitoring programs of population fluctuations.     

Competition among eucalyptus trees depends on genetic variation and resource supply

Genetic variation and environmental heterogeneity fundamentally shape the interactions between plants of the same species. According to the resource partitioning hypothesis, competition between neighbors intensifies as their similarity increases. Such competition may change in response to increasing supplies of limiting resources. We tested the resource partitioning hypothesis in stands of genetically identical (clone-origin) and genetically diverse (seed-origin) Eucalyptus trees with different water and nutrient supplies, using individual-based tree growth models. We found that genetic variation greatly reduced competitive interactions between neighboring trees, supporting the resource partitioning hypothesis. The importance of genetic variation for Eucalyptus growth patterns depended strongly on local stand structure and focal tree size. This suggests that spatial and temporal variation in the strength of species interactions leads to reversals in the growth rank of seed-origin and clone-origin trees. This study is one of the first to experimentally test the resource partitioning hypothesis for intergenotypic vs. intragenotypic interactions in trees. We provide evidence that variation at the level of genes, and not just species, is functionally important for driving individual and community-level processes in forested ecosystems.     

Incorporating known sources of uncertainty to determine precautionary harvests of saltwater crocodiles.

It has been demonstrated repeatedly that the degree to which regulation operates and the magnitude of environmental variation in an exploited population will together dictate the type of sustainable harvest achievable. Yet typically, harvest models fail to incorporate uncertainty in the underlying dynamics of the target population by assuming a particular (unknown) form of endogenous control. We use a novel approach to estimate the sustainable yield of saltwater crocodile (Crocodylus porosus) populations from major river systems in the Northern Territory, Australia, as an example of a system with high uncertainty. We used multimodel inference to incorporate three levels of uncertainty in yield estimation: (1) uncertainty in the choice of the underlying model(s) used to describe population dynamics, (2) the error associated with the precision and bias of model parameter estimation, and (3) environmental fluctuation (process error). We demonstrate varying strength of evidence for density regulation (1.3-96.7%) for crocodiles among 19 river systems by applying a continuum of five dynamical models (density-independent with and without drift and three alternative density-dependent models) to time series of density estimates. Evidence for density dependence increased with the number of yearly transitions over which each river system was monitored. Deterministic proportional maximum sustainable yield (PMSY) models varied widely among river systems (0.042-0.611), and there was strong evidence for an increasing PMSY as support for density dependence rose. However, there was also a large discrepancy between PMSY values and those produced by the full stochastic simulation projection incorporating all forms of uncertainty, which can be explained by the contribution of process error to estimates of sustainable harvest. We also determined that a fixed-quota harvest strategy (up to 0.2K, where K is the carrying capacity) reduces population size much more rapidly than proportional harvest (the latter strategy requiring temporal monitoring of population size to adjust harvest quotas) and greatly inflates the risk of resource depletion. Using an iconic species recovering from recent extreme overexploitation to examine the potential for renewed sustainable harvest, we have demonstrated that incorporating major forms of uncertainty into a single quantitative framework provides a robust approach to modeling the dynamics of exploited populations.     

Spatial patterns reveal negative density dependence and habitat associations in tropical trees

Understanding how plant species coexist in tropical rainforests is one of the biggest challenges in community ecology. One prominent hypothesis suggests that rare species are at an advantage because trees have lower survival in areas of high conspecific density due to increased attack by natural enemies, a process known as negative density dependence (NDD). A consensus is emerging that NDD is important for plant-species coexistence in tropical forests. Most evidence comes from short-term studies, but testing the prediction that NDD decreases the spatial aggregation of tree populations provides a long-term perspective. While spatial distributions have provided only weak evidence for NDD so far, the opposing effects of environmental heterogeneity might have confounded previous analyses. Here we use a novel statistical technique to control for environmental heterogeneity while testing whether spatial aggregation decreases with tree size in four tropical forests. We provide evidence for NDD in 22% of the 139 tree species analyzed and show that environmental heterogeneity can obscure the spatial signal of NDD. Environmental heterogeneity contributed to aggregation in 84% of species. We conclude that both biotic interactions and environmental heterogeneity play crucial roles in shaping tree dynamics in tropical forests.     

Assessing tiger population dynamics using photographic capture-recapture sampling

Although wide-ranging, elusive, large carnivore species, such as the tiger, are of scientific and conservation interest, rigorous inferences about their population dynamics are scarce because of methodological problems of sampling populations at the required spatial and temporal scales. We report the application of a rigorous, noninvasive method for assessing tiger population dynamics to test model-based predictions about population viability. We obtained photographic capture histories for 74 individual tigers during a nine-year study involving 5725 trap-nights of effort. These data were modeled under a likelihood-based, "robust design" capture-recapture analytic framework. We explicitly modeled and estimated ecological parameters such as time-specific abundance, density, survival, recruitment, temporary emigration, and transience, using models that incorporated effects of factors such as individual heterogeneity, trap-response, and time on probabilities of photo-capturing tigers. The model estimated a random temporary emigration parameter of gamma" = gamma' = 0.10 +/- 0.069 (values are estimated mean +/- SE). When scaled to an annual basis, tiger survival rates were estimated at S = 0.77 +/- 0.051, and the estimated probability that a newly caught animal was a transient was tau = 0.18 +/- 0.11. During the period when the sampled area was of constant size, the estimated population size N(t) varied from 17 +/- 1.7 to 31 +/- 2.1 tigers, with a geometric mean rate of annual population change estimated as lambda = 1.03 +/- 0.020, representing a 3% annual increase. The estimated recruitment of new animals, B(t), varied from 0 +/- 3.0 to 14 +/- 2.9 tigers. Population density estimates, D, ranged from 7.33 +/- 0.8 tigers/100 km2 to 21.73 +/- 1.7 tigers/100 km2 during the study. Thus, despite substantial annual losses and temporal variation in recruitment, the tiger density remained at relatively high levels in Nagarahole. Our results are consistent with the hypothesis that protected wild tiger populations can remain healthy despite heavy mortalities because of their inherently high reproductive potential. The ability to model the entire photographic capture history data set and incorporate reduced-parameter models led to estimates of mean annual population change that were sufficiently precise to be useful. This efficient, noninvasive sampling approach can be used to rigorously investigate the population dynamics of tigers and other elusive, rare, wide-ranging animal species in which individuals can be identified from photographs or other means.     

Linking patch-use behavior, resource density, and growth expectations in fish

Optimality theory rests on the assumptions that short-term foraging decisions are driven by variation in environmental quality, and that these decisions have important implications for long-term fitness. These assumptions, however, are rarely tested in a field setting. We linked behavioral foraging decisions in food patches with measures of environmental quality covering larger spatial (resource density) or temporal (growth parameters) scales. In 10 lakes, we measured the food density at which benthic fish give up foraging in experimental food patches (giving-up density, GUD), quantified the biomass of benthic invertebrates, and calculated the maximum individual size (L(infinity)) of bream (Abramis brama L.), a typical benthivore in these lakes. We found positive relationships between resource density and both GUD and L(infinity), and a positive relationship between L(infinity) and GUD. Prey characterized as vulnerable to predation contributed most to the relationships between resource density and either GUD or L(infinity). A path analysis showed that resource density and L(infinity) directly explained 54% and 28%, respectively, of the variation in GUD, whereas 86% of the variation in L(infinity) was explained by resource density, with mostly indirect contribution from GUD. We conclude that the short-term foraging behavior of benthivores matched our expectations based on optimality theory by being positively linked to variables on environmental quality operating at both a larger spatial scale and a longer temporal scale.     

Local spatial modeling of white-tailed deer distribution

Complex spatial heterogeneity of ecological systems is difficult to capture and interpret using global models alone. For this reason, recent attention has been paid to local spatial modeling techniques. We used one local modeling approach, geographically weighted regression (GWR), to investigate the effects of local spatial heterogeneity on multivariate relationships of white-tailed deer distribution using land cover patch metrics and climate factors. The results of these analyses quantify differences in the contributions of model parameters to estimates of deer density over space. A GWR model with local kernel bandwidth was compared to a GWR model with global kernel bandwidth and an ordinary least-squares regression (OLS) model with the same parameters to evaluate their relative abilities in modeling deer distributions. The results indicated that the GWR models predicted deer density better than the traditional ordinary least-squares model and also provided useful information regarding local environmental processes affecting deer distribution. GWR model comparisons showed that the local kernel bandwidth GWR model was more realistic than the global kernel bandwidth GWR model, as the latter exaggerated local spatial variation. The parameter estimates and model statistics (e.g., model R²) of the GWR models were mapped using geographic information systems (GIS) to illustrate local spatial variation in the regression relationship and to identify causes of large-scale model misspecifications and low estimation efficiencies.     

Population dynamics of the barnacle <Emphasis Type="Italic">Chthamalus montagui</Emphasis> at two spatial and temporal scales in northern Spain

This study describes spatial and temporal patterns of variability in population parameters in the barnacle Chthamalus montagui Southward in three localities of northern Spain and evaluates whether density-dependent settlement may regulate population dynamics. The sampling design considered two spatial scales, localities and sites within localities, and two temporal scales, years and six month intervals. Density, amount of free space, mortality, growth rate and magnitude of settlement (both absolute and per unit of free space) were obtained from photographs of permanent quadrats and from direct counts in the field. The number of settlers in scraped and untouched quadrats was used to estimate the importance of the presence of conspecifics in settlement. Significant variation at the two spatial and temporal scales was found for most parameters. Large spatial and temporal variations in adult mortality rate, density, and settlement were observed. Patterns of mortality were not consistent with differences in density among localities. Differences in settlement among localities were maintained through time. We suggest that magnitude of settlement is regulated by persistent features such as topography or local water circulation. We assume that early post-settlement mortality does not differ among localities. In the absence of differential mortality, settlement determines average population density within localities. Within localities, settlement was independent of density and free space. No consistent evidence was found on preferential settlement at the vicinity of conspecifics. The main conclusion is that density-dependent settlement is not relevant for the regulation of the populations of C. montagui in the northern Spain. Regulation might occur by density-dependent processes within the adult fraction of the population and/or the larval phase before settlement.     

Incorporating diverse data and realistic complexity into demographic estimation procedures for sea otters.

Reliable information on historical and current population dynamics is central to understanding patterns of growth and decline in animal populations. We developed a maximum likelihood-based analysis to estimate spatial and temporal trends in age/sex-specific survival rates for the threatened southern sea otter (Enhydra lutris nereis), using annual population censuses and the age structure of salvaged carcass collections. We evaluated a wide range of possible spatial and temporal effects and used model averaging to incorporate model uncertainty into the resulting estimates of key vital rates and their variances. We compared these results to current demographic parameters estimated in a telemetry-based study conducted between 2001 and 2004. These results show that survival has decreased substantially from the early 1990s to the present and is generally lowest in the north-central portion of the population's range. The greatest temporal decrease in survival was for adult females, and variation in the survival of this age/sex class is primarily responsible for regulating population growth and driving population trends. Our results can be used to focus future research on southern sea otters by highlighting the life history stages and mortality factors most relevant to conservation. More broadly, we have illustrated how the powerful and relatively straightforward tools of information-theoretic-based model fitting can be used to sort through and parameterize quite complex demographic modeling frameworks.     

Genetic differentiation of Texas Gulf Coast populations of the blue crab Callinectes sapidus

Allelic frequencies at three polymorphic, enzyme-encoding gene loci (GOT-2, EST-1, EST-2) were determined for Callinectes sapidus (Rathbun) megalopae and adults sampled along the Texas coast of the Gulf of Mexico. Significant temporal and spatial variation was observed at all three loci. Primary findings included: (1) megalopal allelic frequencies often differed significantly from those observed among neighboring adult populations; (2) larval allelic frequencies appeared to vary seasonally, with populations showing sharp differences in the summer months but tending to be more homogeneous in winter; (3) allelic frequencies among adult populations were significantly heterogeneous, but only one locus (EST-2) showed significant temporal variation; (4) juvenile and adult crabs sampled within one bay showed no size-specific differences in allelic frequencies. The spatial heterogeneity in allelic frequencies suggests that interpopulation gene flow is not sufficient to overcome population differentiation resulting from genetic drift and/or natural selection. Temporal variation in larval allelic frequencies suggests seasonal changes in larval source populations which may result from population differences in spawning season or developmental times or from seasonal changes in coastal current patterns.     

A Markov random field spatio-temporal analysis of ocean temperature

The National Oceanic Data Center (NODC) contains historical records from approximately 144,000 hydrographic stations in the North Atlantic. This data has been used by oceanographers to construct maps of point estimates of pressure, temperature, salinity and oxygen in the North Atlantic (Levitus (1994); Lozier et al. (1995)). Because data from any particular year are scarce, the previous maps have been for time-averaged values only. In addition, the maps have been reported without uncertainty estimates. This paper presents a Markov random field (MRF) analysis that can generate maps for specific time periods along with associated uncertainties. To estimate changes in oceanic properties over time previous oceanographic work has focused on differences between a few time periods each having many observations. Due to data scarcity this poses a severe restriction for both spatial and temporal coverage of climatic change. The MRF analysis provides a means for temporal modeling that does not require high data density at each time period. To demonstrate the usefulness of a MRF analysis of oceanic data we investigate the temporal variability along 24.5°N in the North Atlantic. Our results are compared to an earlier analysis (Parrilla et al. (1994)) where data from only three time periods was used. We obtain a more thorough understanding of the temperature change found by this previous study.     

Inferring evolutionary signals from ecological data in a plant-pathogen metapopulation

We followed the dynamics of local epidemics in three populations of a natural plant-pathogen system for four sequential years. We characterize the overwintering process with spatial statistics and use a stochastic, spatially explicit, modeling approach with Bayesian parameter estimation to study the spread of the infection during the growing season. Our modeling approach allows us to infer coevolutionary signals from spatiotemporal data on pathogen prevalence. Most importantly, we are able to assess the distribution of resistant hosts within the distribution of all host plants. We show that resistant hosts occur in areas with high pathogen encounter rates, and that the occurrence of resistance correlates with overwintering probability of the pathogen. The estimates for essentially all model parameters are characterized by a large amount of variation over the years and the populations. While the variation in the fraction of resistant hosts and in the force of infection is to a large extent explained by the population, the other model parameters (two parameters describing the shape of the dispersal kernel) vary essentially in an unpredictable manner, suggesting that much of the variation may occur at very fine spatial and temporal scales.     

Vegetation structure constrains primary production response to water availability in the Patagonian steppe

Grassland aboveground net primary production (ANPP) increases linearly with precipitation in space and time, but temporal models relating time series of ANPP and annual precipitation for single sites show lower slopes and regression coefficients than are shown by spatial models. The analysis of several ANPP time series showed lags in the ecosystem response to increased water availability, which may explain the difference between spatial and temporal models. The lags may result from constraints that ecosystems experience after drought. Our objective was to explore the structural constraints of the ANPP response to rainfall variability in a semiarid ecosystem, the Patagonian steppe, in southern Argentina. We designed a 3-yr rainfall manipulation experiment where we decreased water input with rainout shelters during two consecutive years, which included three levels of rainfall interception (30%, 55%, and 80%) and a control. In the third year, we irrigated one-half of the plots of each rainfall-interception treatment. We evaluated the immediate effects of drought on current-year ANPP and the effects of previous-year drought on vegetation recovery after water supplementation. ANPP (g x m(-2) x yr(-1)) was linearly related to annual precipitation input (APPT; mm/yr) along the experimental precipitation gradient (ANPP = 0.13 x APPT + 58.3; r2 = 0.34, P < 0.01), and this relationship was mostly accounted for by changes in the ANPP of grasses. Plant density (D; no. individuals/mm2) was related to the precipitation received during the drought period (D = 0.11 x APPT + 18; r2 = 0.39, P < 0.05). The recovery of plants after irrigation was lower for those plots that had experienced experimental drought the previous years relative to controls, and the lags were proportional to the intensity of drought. Therefore, our results suggest that the density of plants may constrain the recovery of vegetation after drought, and these constraints may determine lags that limit the capacity of the ecosystem to take advantage of wet years after dry years.