Recovery of Subtidal Benthic Macroinvertebrate Communities Following Natural and Experimental Disturbances

Abstract

The recovery processes of subtidal benthic macroinvertebrate communities following large-scale natural and meso-scale experimental disturbances were studied in Wellington Harbour, New Zealand, a temperate semi-enclosed embayment. This is the first time that long-term effects (>1 year post-disturbance) of a naturally occurring toxic plankton bloom have been investigated in the Southern hemisphere. For 2 years macroinvertebrate communities were studied at three sites of differing hydrodynamic regime. Samples were taken with a Van Veen grab and washed through a 500 μm mesh. Community recovery following the bloom was site-specific. Multivariate analyses revealed that at two sites community recovery was not completed >3 years post-bloom, whereas at the third site the community composition oscillated from year to year, but did not show any signs of a sequential recovery process. The hydrodynamic regime was identified as a major factor influencing the observed recovery processes. Communities exposed to an active hydrodynamic regime were less affected by the bloom and recovered faster, as they were naturally in a perpetual state of recovery as indicated by a dominance of r-selected species. The community at the hydrodynamically less active site was more affected by the bloom. Complete recovery to the pre-disturbance climax community dominated by K-selected species was estimated to take 4-5 years, if not interrupted by other disturbances. For the first time a defaunation experiment was conducted in a hydrodynamically active site to mimic the effects of a plankton bloom on the benthic macroinvertebrate community. Three sediment plots of 25 m2 were covered by plastic tarpaulins, thereby creating a benthic die-off caused by oxygen depletion. This method of defaunation had not been used in the subtidal before. Community recovery was studied for 1 year and compared with community composition in undisturbed control plots. Macroinvertebrate samples were taken by diver-operated cores and washed through a 500 μm mesh. Recovery was slow until after 70 days when abundance and number of species increased synchronously in disturbed and control plots. Multivariate analyses showed that community composition fluctuated strongly in the first 100 days. After 1 year, although disturbed and control communities were converging, differences in community composition were still significant. Time for complete recovery was estimated to be approximately 2 years. Predictions of current succession models were generally fulfilled in both studies. Recovered communities were similar in their composition to either pre-disturbance or surrounding communities. The major deviation from model predictions was that no abundance peak of opportunistic species occurred in either study. Timing of the disturbance, in both studies past the major macroinvertebrate recruitment peak, and the hydrodynamic regime were identified as major factors influencing recovery processes of the communities studied. Such deviation from model predictions indicates that the general models cannot take into account the multiplicity and complexity of factors influencing recovery processes. Thus, their applicability in predicting recovery times and endpoints for specific disturbances at specific locations is limited. Location-specific models might be a useful alternative. Recommendations are made to combine uni- and multivariate techniques to assess recovery processes due to their different sensibilities to changes in community composition.