UC Berkeley

Little is known about how metrics of biodiversity and abundance scale in ecologically disturbed and disrupted systems. Natural disturbances have a fundamental role in structuring ecological communities, and the study of these processes and extension to novel ecological disruptions is of increasing importance due to global change and mounting human impacts. Numerous studies have demonstrated the importance of natural disturbance in determining basic ecological properties of an ecosystem, including species diversity, membership, and relative abundances of those species, as well as overall productivity. Although estimating ecological metrics at both the species and community level is of critical importance to conservation goals, predicting the impacts of disturbance and disruption, including anthropogenic changes, on ecosystems is a major problem for ecological theory for several reasons. Disturbances are diverse in type, create patches that are internally heterogeneous, interact with site-specific disturbance legacies, and have different effects over multiple spatial and temporal scales. In contrast, empirical studies providing the basis for development of models tend to focus on short time-scales and relatively homogeneous systems with steady-state dynamics. Sites that experience single disturbances or are part of disturbance regimes also pose a challenge to ecological theory because they represent open, non-equilibrium systems that are not tractable with equilibrium mathematics. Additionally, the spatial scale at which a disturbance is studied will affect the conclusions that are drawn about communities or their component species. Nevertheless, the ubiquity and importance of disturbance to ecosystems continues to motivate a search for generality in disturbance and landscape ecology. In this dissertation, I apply an information entropy based theory of macroecology to ecosystems in transition, or have otherwise experienced ecological disruption. This leads to comparable results between systems, and forms a basis for cross-system comparisons of ecosystems in transition.

The maximum information entropy inference procedure (MaxEnt) has been proven to produce the least-biased estimates of a probability distribution, given prior knowledge of a system. Empirical values make up the prior knowledge of the system, and constrain the mean, variance, or higher moments of a given distribution. An extension of the MaxEnt procedure, the Maximum Entropy Theory of Ecology (METE) takes a macroecological approach to estimating plot- to landscape- to biome-scale species diversity, abundance, and energetics metrics, using only the relationships between four non-adjustable state variables S0 (total species), A0 (area under consideration), N0 (total abundance), and E0 (total metabolic energy), and no adjustable parameters to characterize the scaling of diversity and abundances of species in a system. Until this work, METE has mainly been tested in steady-state and minimally disturbed systems.

In Chapter 1, “Disturbance macroecology: a comparative study of plant species’ abundances and distributions in different-age post-fire stands of Bishop pine (Pinus muricata),” I investigate how metrics of biodiversity and abundance are scale in a plant community that is largely structured by a dominant, disturbance-dependent species. We target two different-aged stands in a region of high wildfire activity, one a characteristically mature stand with a diverse understory, and one more recently disturbed by a stand-replacing fire 17 years previously. We compare the stands using various macroecological metrics of species richness, abundance and spatial distributions that are predicted by METE, which does not rely on steady-state or equilibrium assumptions, and is therefore well-suited to be a null model for ecosystems in transitional states. Ecological patterns in the mature stand more closely match METE predictions than do data from the more recently disturbed stand. This suggests METE’s predictions are more robust in late successional, slowly changing, or steady-state systems than those in rapid flux with respect to species composition, abundances, and body sizes. These findings highlight the need for a macroecological theory that incorporates natural disturbance and other ecological perturbations into its predictive capabilities, because most natural systems are not in a steady state.

In Chapter 2, “Macroecology for management: Testing an information-entropy-based theory of macroecology against anthropogenic disruption of high-Sierra meadows, I investigate the extent to which anthropogenic changes to an ecosystem, in the form of grazing by large, introduced herbivores, are detectable using METE, and small (<1 ha) replicate census plots. Anthropogenically-induced ecological disruptions (anthropogenic disruptions) have been overlooked by macroecological theory because they represent ecosystems in various states of transition that result from non-natural selection on the community. While critically important to understand for conservation reasons, anthropogenic disruptions are, in general, not comparable to each other, nor to other ecological disturbances that are natural in origin. Here, we use METE to examine the effects of an anthropogenically-induced novel disturbance regime of grazing by horses in high Sierra Nevada meadows on the species-abundance distributions (SAD), number of singleton species, and the species-level spatial abundance distributions (SSADs) (a measure of spatial aggregation) for all species in three pairs of grazed and ungrazed meadows, each meadow containing a system of plots set up across a moisture gradient. We find that number of singleton species may be a better indicator of ecological disruption than the shape of the SAD in systems where the differences in community structure are subtle. We also find that the METE SSAD performs better than all other models tested for both grazed and ungrazed plots. We suggest ways of augmenting tests of the METE SSAD to refine theory for management relevance.

In Chapter 3, “Empirical tests of within- and across-species energetics in a diverse plant community,” I (with my coauthors) test the metabolic predictions of METE for herbaceous plants in a subalpine meadow. METE is an extremely general macroecological theory that predicts spatial, abundance and metabolic rate distributions of species, and the interrelationships of these metrics for any system defined by a set of basic community state variables. It therefore also predicts body-size distributions if a metabolic scaling relationship between metabolism and body size is assumed. Many fundamental properties of ecological systems and interactions are tied to body size, and a related metric, the metabolic rate distribution, both within and across species. Extensive tests of METE’s macroecological predictions in multiple ecosystems and with multiple taxa generally support its species diversity and abundance predictions, but two related predictions had not been evaluated against full community census data until this study: the distribution of metabolic rates of individuals within species as a function of the abundance of the species, and the distribution of average individual metabolic rates across species. We show that while METE realistically predicts the distribution of individual metabolic rates across the entire community, the within and across species predictions generally fail. We also test the energy-equivalence type prediction that arises as a consequence of the prediction for the distribution of average individual metabolic rates across species. We suggest several possible explanations for the empirical deviations from theory, and distinguish between the expected deviations caused by ecological disturbance and those deviations that might be corrected within the theory.

Taken together, these results indicate that it is possible to extend macroecological theory to ecosystems that experience natural disturbances and other ecological disruptions. Because we find that there are regular empirical deviations from theory in ecosystems that have experienced some sort of disturbance, we can conclude that the values and ratios of the four state variables (A0, S0, N0 and E0) used by METE are not sufficient to describe the dynamics of real ecosystems. These regular deviations, however, are interesting in their own right, because they suggest where ecological processes may influence the shape of empirical macroecological distributions. This will provide a framework for comparing and eventually predicting the various effects of disturbance on biodiversity, in the contexts of disturbance regimes, anthropogenic change, and mixtures of both.