Date of Award
Ph.D. Oceanography/Marine Biology
Bernhard M. Riegl
Richard S. Spieler
Richard E. Dodge
Linking small-scale measurements of species distributions to broad-scale seascapes is necessary to understanding and predicting organismal distributions and their dynamics. This applies to reef fish populations as well. Reef fish studies are often limited to small spatial scales because of logistical and economic constraints; however, viewing the data at larger spatial scales might elucidate unforeseen relationships and patterns and facilitate regional management and conservation efforts. To address this growing need, an empirical model was created to predict reef fish abundance and species richness for the entire seascape using the relationship between the fish, benthic habitats, and GIS-derived topographic complexity metrics from a subset of in situ survey data.
The essential inputs for this model included a large-scale, high-resolution bathymetric survey of the seascape; accurate, spatially defined and characterized benthic habitats of the seascape; and a spatially defined, in situ survey of the reef fish population with a statistically powerful number of samples within many of the defined habitats. Two studies were performed to obtain the model inputs.
The first study (Part II) integrated laser bathymetry, acoustic ground discrimination, subbottom profiling, and aerial photography to create a habitat map for the nearshore benthic habitats of Broward County, Florida, USA from 0 to 35m depth. A mosaic of interpolated, sun-shaded, high-resolution laser bathymetry data served as the foundation upon which acoustic ground discrimination, subbottom profiling, aerial photography, and groundtruthing data aided in resolving habitats. Mapping protocols similar to NOAA’s Biogeography Branch Caribbean coral reef ecosystem maps were used to allow for a comparable output. Expert-driven visual interpretation outlined geomorphological features at a scale of 1:6000 with a minimum mapping unit of 1 acre, pre-defined by the NOAA protocol. Acoustic data were then used to differentiate areas of similar morphology by their acoustic diversity and look at within feature variation. Of the approximately 112 km2 mapped, 56.62 km2 were coral reef and colonized hardbottom (50.42%), 54.78 km2 were unconsolidated sediments (46.80%), and 0.43 km2 were other categories (2.78%). Three linear reef complexes were depicted. The outermost linear reef has a mature windward reef morphology including a spur and groove system, which was absent on the other two reef lines. Different benthic habitats were found on the outer versus middle and inner reefs. A considerable amount of colonized pavement was found inshore. The Broward map yielded a high overall accuracy of 89.6%, only slightly less than the photo interpreted NOAA Caribbean maps (overall accuracy of 91.1%). User and producer accuracies within each category were also comparable. Similar methodology can be used in other areas where photo interpretation is not feasible.
The second study (Part III) analyzed reef fish assemblage relationships to in situ and GIS topographic measurements across the seascape to evaluate the possibilities of using GIS metrics as a proxy for prediction models. In situ topographic complexity was measured for 370 point count fish surveys spanning the reef seascape. GIS topographic measurements were taken from a high-resolution bathymetric dataset of each survey’s footprint. The sites were characterized for seascape analysis by the independent benthic habitat map from Part II. Reef fish abundance and species richness increased with increasing topographic complexity, but the data were weakly correlated due to high variability suggesting that it is not the only controlling factor on the assemblage. Seascape characterization elucidated two distinct assemblages; one shallow and one deep. Topographic complexity better correlated to species richness in the shallow habitats than in deeper ones, whereas, it correlated to abundance the strongest in the deeper habitats. In situ measurement yielded the highest correlations, but the GIS metrics followed the same trends therefore they can be used as proxies for reef fish distribution models.
The results from the previous two studies were assembled into a model framework to project the relationship of reef fish abundance and richness to topographic metrics in the different habitats across the entire seascape. A squared polygonal grid of the entire seascape was created at the same resolution as the fish surveys and topographic statistics were calculated for every square I the grid. Grid polygons which fell outside modeled habitats (e.g. sand) were filtered and discarded. The linear regression equations of the reef fish/GIS topography relationship (Part III) were used to predict the abundance and richness of fish for the prediction grid in each modeled habitat. The topographic statistic from each grid polygon was entered as the x value (GIS metric) in the regression equation which was then solved for y (abundance of richness). The output was rounded to the nearest whole number and populated in the GIS for the appropriate grid polygon. A similarity percentage analysis (SIMPER) between habitats calculated the dominant percentage (top 70%) of each species in each habitat (Part IV). These percentages were used to estimate the abundance of the dominant species in each grid cell from the predicted total abundance. This resulted in a seascape of polygons (15.24m by 15.24m grid cells) with predicted abundance and richness values for three GIS topographic metrics, elevation, volume, and surface rugosity. These were then displayed as maps for viewing, querying, and statistical analyses.
Prediction model output analysis evinced similar relationships as the input data for both abundance and species richness, thus this model enabled viewing of the relationship between reef fishes and their habitats over the entire seascape. Comparison between predicted and empirical data showed significant, but low agreement for all of the topographic metrics. The elevation model performed best in this comparison with both abundance (r2=0.27) and richness (r2=0.39). The fact that the prediction data was not strongly correlated to the input data, but the statistical relationships were evident between datasets, means that the model is best used for comparative analyses instead of gross estimates.
This model has many scientific and management applications like the estimation of fish stocks, the designation of marine protected areas, and baseline comparisons to future surveys. It also gives statistical support to management and conservation decisions, giving resource managers a powerful tool to support their actions. This framework design is a simple approach that lends itself to adaptation and could easily be modified to look at different ecological processes (other than fish) and their relationships to many types of seascape variables. To increase model accuracy, better understandings of the appropriate measurement scale and fish operational scales are needed as well as more research on the dynamics of how reef fish relate to topographic complexity and the other ecological factors influencing their distributions across the seascape.
Brian K. Walker. 2008. A Seascape Approach to Predicting Reef Fish Distribution. Doctoral dissertation. Nova Southeastern University. Retrieved from NSUWorks, Oceanographic Center. (3)