Making and breaking a new ecological theory: Does maximum information entropy predict community structure in evolving ecosystems?

Background/Question/Methods

The maximum information entropy theory of ecology (METE) predicts species abundance and individual metabolic rates by maximizing statistical randomness while constraining the observed number of species, individuals and total metabolic rate. Thus METE is a neutral, conditional on incorporation of specific information (i.e. species, individuals and energy). Thus like other neutral theories, it provides a powerful null model. Unlike other neutral theories, new information related to biological mechanism can be easily incorporated, allowing one to test the importance of those mechanisms in structuring communities. METE very accurately predicts distributions of abundance and metabolic rate in many real ecosystems. However, we ask, does the unique geologic history and evolution of the Hawaiian archipelago produce ecosystems that deviate from the predictions of METE? Using published data on the diversity and biomass of terrestrial arthropods across a chronosequence of substrate ages we evaluate the predictions of METE, testing for systematic deviations caused by evolutionary history, specifically island and substrate age, recent invasion, and ancient colonization events.

Results/Conclusions

METE under-predicts the most dominate species and over-predicts medium-rare species in both youngest (order 100 years) and oldest (order 4 million years) communities. This signal is distributed differently across trophic guilds indicating that the reasons for METE's off-prediction could be varied depending on the unique eco-evolutionary histories of these groups. The fit for detritivores monotonically improves for older sites, while that for herbivores worsens for older sites; predators are universally well-predicted. Predator communities are also more heavily invaded, implying shared evolutionary histories may drive communities out of statistical steady-states. Genus to species ratios of Hawaiian arthropods have previously been used to estimate colonization and diversification. We find that these ratios do account for deviations in detritivores, but not in other guilds. To better account for higher taxonomic groups we present a new extension of METE that further constrains community structure based on the partitioning of species to genera. If historical contingencies relating to colonization influence community structure, including explicit constraints on genera should improve METE's predictiveness. The unique identities of species and the compositional similarity of sites did not influence the predictiveness of METE, implying observed patterns and not taxon-specific, but a property of the system at large.