Recovery Dynamics in Benthic Communities: Balancing Detail with Simplification
299
14.2 Searching for Generality Part I
In attempting to make comparisons between studies conducted in different
locations it is important to identify some common and meaningful processes
or measures. Meta-analysis provides some statistical tools for the development of quantitative syntheses of separate studies (e.g., Gurevitch and Hedges
1993, 1999; Arnqvist and Wooster 1995; Englund et al. 1999; Osenberg et al.
1999). However, sophisticated meta-analysis will be meaningless unless the
variables used to compare studies are ecologically meaningful.
Even when only a qualitative synthesis is possible, we must be aware of
potential confounding variables when making comparisons. Disturbance is a
very nebulous term (see Pickett and White 1985 for definitions) and disturbance events can occur at various intensities, frequencies and extents, all of
which are likely to have important consequences for the subsequent ecological recovery (Zajac et al. 199~). Potential problems can also arise with
operational definitions of succession/recovery end points for a defaunated
patch. Birth and mortality rates of the benthos within the patch plus their flux
across the patch boundary (i. e., immigration and emigration) determine the
path of recovery. Also important are the broader scale temporal changes in the
benthic community of the surrounding undisturbed sediments, because these
provide the pool of potential colonists for patch recovery. We must also assess
recovery by comparison with community structure in the adjacent sediments
rather than against some theoretical baseline. Confidence intervals for
density estimates in and out of experimental plots may overlap extensively
during recovery (indicating no significant difference between experimental
plots and ambient sediments) but subsequently once again separate. Assuming an adequate sample size is used, this difference can result from recruitment events; changes in spatial variance, (particularly when densities are
low); or the merging of experimental patterns with broader scale temporal
patterns (such as a seasonal decline in population density). Thus, it is
important that similarities in density and assemblage structure in
experimental plots and ambient sediments persist over time before recovery
is assumed. Unfortunately, this is frequently not the case given the limited
time over which many experiments are sampled.
The "classical" succession models for marine soft-sediment macrobenthic
communities were developed by Pearson and Rosenberg (1978) and Rhoads
et al. (1978). These conceptual models are very similar, although the former
was focused on recovery processes in space and the latter on time. Both conceptual models were developed using a combination of broad-scale survey
work and detailed smaller scale experimentation and observation. These
models have become the cornerstones of the assessment of environmental
impacts on macrobenthic communities and the framework against which
299
14.2 Searching for Generality Part I
In attempting to make comparisons between studies conducted in different
locations it is important to identify some common and meaningful processes
or measures. Meta-analysis provides some statistical tools for the development of quantitative syntheses of separate studies (e.g., Gurevitch and Hedges
1993, 1999; Arnqvist and Wooster 1995; Englund et al. 1999; Osenberg et al.
1999). However, sophisticated meta-analysis will be meaningless unless the
variables used to compare studies are ecologically meaningful.
Even when only a qualitative synthesis is possible, we must be aware of
potential confounding variables when making comparisons. Disturbance is a
very nebulous term (see Pickett and White 1985 for definitions) and disturbance events can occur at various intensities, frequencies and extents, all of
which are likely to have important consequences for the subsequent ecological recovery (Zajac et al. 199~). Potential problems can also arise with
operational definitions of succession/recovery end points for a defaunated
patch. Birth and mortality rates of the benthos within the patch plus their flux
across the patch boundary (i. e., immigration and emigration) determine the
path of recovery. Also important are the broader scale temporal changes in the
benthic community of the surrounding undisturbed sediments, because these
provide the pool of potential colonists for patch recovery. We must also assess
recovery by comparison with community structure in the adjacent sediments
rather than against some theoretical baseline. Confidence intervals for
density estimates in and out of experimental plots may overlap extensively
during recovery (indicating no significant difference between experimental
plots and ambient sediments) but subsequently once again separate. Assuming an adequate sample size is used, this difference can result from recruitment events; changes in spatial variance, (particularly when densities are
low); or the merging of experimental patterns with broader scale temporal
patterns (such as a seasonal decline in population density). Thus, it is
important that similarities in density and assemblage structure in
experimental plots and ambient sediments persist over time before recovery
is assumed. Unfortunately, this is frequently not the case given the limited
time over which many experiments are sampled.
The "classical" succession models for marine soft-sediment macrobenthic
communities were developed by Pearson and Rosenberg (1978) and Rhoads
et al. (1978). These conceptual models are very similar, although the former
was focused on recovery processes in space and the latter on time. Both conceptual models were developed using a combination of broad-scale survey
work and detailed smaller scale experimentation and observation. These
models have become the cornerstones of the assessment of environmental
impacts on macrobenthic communities and the framework against which