Biodiversity influences the effects of oil disturbance on coastal ecosystems

Abstract

Biodiversity can enhance the response of ecosystems to disturbance. However, whether diversity can reduce the ecological effect of human-induced novel and extreme disturbances is unclear. In April 2010, the Deepwater Horizon (DwH) platform exploded, allowing an uncontrolled release of crude oil into the northern Gulf of Mexico. Initial surveys following the spill found that ecological impacts on coastal ecosystems varied greatly across habitat-type and trophic group; however, to date, few studies have tested the influence of local biodiversity on these responses. We used a meta-analytic approach to synthesize the results of 5 mesocosm studies that included 10 independent oil experiments and 5 independent oil + dispersant experiments. We tested whether biodiversity increased the resistance and/or resilience of coastal ecosystems to oil disturbance and whether a biodiversity effect depended on the type of diversity present (taxonomic or genetic) and/or the response type measured (population, community, or ecosystem level). We found that diversity can influence the effects of oiling, but the direction and magnitude of this diversity effect varied. Diversity reduced the negative impact of oiling for within-trophic-level responses and tended to be stronger for taxonomic than genetic diversity. Further, diversity effects were largely driven by the presence of highly resistant or quick to recover taxa and genotypes, consistent with the insurance hypothesis. However, we found no effect of diversity on the response to the combination of oil and dispersant exposure. We conclude that areas of low biodiversity may be particularly vulnerable to future oil disturbances and provide insight into the benefit of incorporating multiple measures of diversity in restoration projects and management decisions.

Keywords: Deepwater Horizon; diversity‐stability; genetic diversity; oiling; species diversity.

FIGURE 1

Estimated mean Hedge's d effect size of oiling ±95% confidence intervals in monoculture (light grey) and polycultures (black) across (a) response level (population, community, or ecosystem) and (b) diversity type (genetic or taxonomic). The numbers in parentheses (n, k) represent the number of effect sizes used in the models for both monoculture and polycultures within each response level or diversity type (k) and the number of independent studies from which those effect sizes were sourced from (n). A positive d indicates that oiling increased performance, while a negative d indicates that oiling reduced performance. 95% confidence intervals encompassing zero indicate no effect of oiling. * denotes significant different among monocultures and polycultures in oil effect within that response level (from post hoc linear contrast analyses) at level of p = .07

FIGURE 2

Estimated mean Hedge's d effect size of oil + dispersant ±95% confidence intervals between polyculture (black) and (a) average monoculture (grey, closed square) and (b) best monocultures (grey, open diamond). The number of effect sizes used in the models for both monoculture and polycultures was k = 5 for n = 2 independent studies from which those effect sizes were sourced from. A positive d indicates that oil + dispersant increased performance, while a negative d indicates that oil + dispersant reduced performance. 95% confidence intervals encompassing zero indicate no effect of oil + dispersant. Best monoculture was determined by the monoculture that was least impacted by oiling treatment

FIGURE 3

Estimated mean Hedge's d effect size of oiling ±95% confidence intervals in best monoculture (light gray) and polycultures (black) across (a) response level (population, community, or ecosystem) and (b) diversity type (genetic or taxonomic). The numbers in parentheses (n, k) represent the number of effect sizes used in the models for both monoculture and polycultures within each in response level or diversity (k) and the number of independent studies from which those effect sizes were sourced from (n). A positive d indicates that oiling increased performance, while a negative d indicates that oiling reduced performance. 95% confidence intervals encompassing zero indicate no effect of oiling. Best monoculture was determined by the monoculture that was least impacted by oiling treatment. Because we removed any experiment or response that did not replicate individual monocultures in this dataset, the average polyculture values are different from Figure 1