Is Community Structure a Reliable Proxy for Habitat When Constructing Indices of Relative Abundance?

One of the biggest hurdles with statistical modeling of fisheries survey data is the high proportion of zero observations.  A zero arises from one of two possibilities, the species was present at the location sampled but not observed (“observational” zero), or the survey location was in an area of unsuitable habitat for the species and the species is not expected to be present (“structural” zero).  Although large-scale generalizations of defining suitable habitat are typically made based on depth and/or latitude, it can be very difficult to separate the structural zeros from observational zeros in most fisheries-dependent data.  We compare indices of abundance developed from two methods.  The first is the Stephens-MacCall method, which identifies species commonly caught with the target species.  Indices of abundance from this method are calculated at the port or county level.  The second method is based on high-resolution baythmetric data, from which we overlay fishing stops on individual rocky reefs, and subset the data based on the presence of the target species on a reef.  Indices of abundance are reef-based from the second method.  We illustrate the comparison for China rockfish.