Western Indian Ocean coral and fish normalized site richness collected between 1991 to 2000

Aim: Strong social-ecological trade-offs between resource extraction and protection have created challenges for large, protected area management in natural-resource-dependent countries. Therefore, local governments and community conservation activities are becoming common and need information about low environmental exposure and high biodiversity for planning localized conservation activities.

Location: the western Indian Ocean

Methods: Coral reef sites were evaluated for local scale environmental and species richness to elucidate local patterns in spatial heterogeneity. Local coral and fish taxonomic richness were normalized to partially account for common and heterogeneous disturbances to coral cover and fish biomass. Residuals were evaluated for patterns of local diversity with geography, environmental stress, and by machine learning to evaluate the relationship with 21 specific environmental variables.

Results: High variability in richness was found at similar latitudes where richness was high. Relationships with specific environmental and human influences variables were complex and spatially heterogeneous. Expected large-scale biogeographic variables influenced richness but variability and environmental influences were highly specific and localized. Among the environmental and human influence variables examined, ~ 8 variables contributed 8 to 25% of the variance to the richness of both coral and fishes.

Main conclusions: Decisions to focus small-scale conservation on locally biodiverse locations could contribute to species persistence by planning for local heterogeneity in richness and stress. From this specific data set, sites in the Pemba Channel between the Tanzanian mainland and Pemba Island, and northern Mozambique and Madagascar fit these characteristics.

Methods
Field study sites were undertaken in 4 ecoregions and 7 countries that ranged in latitude from 2.04°S (Kenya) to 26.08°S (Mozambique) and longitudes of 32.96°E (Mozambique) to 57.71°E in (Mauritius). Sites were located on the windward and leeward sides of coral reefs in depths from 1.5 to 20 meters depth at low tide (the region’s tidal range is ~1 to 4 meters). Sites were all located on calcium carbonate coral bottoms colonized by hard and soft corals and various algae, with sand and seagrass being a smaller portion of the benthic cover (McClanahan & Muthiga 2016). Sites were distributed among four fisheries management categories, namely, high compliance reserves (no-take closures), low compliance reserves, and restricted and unrestricted fishing, as previously described (McClanahan et al. 2015). Low compliance closures were areas legally gazetted as marine reserves but where fishing was evident by personal observations or reports in the literature. Restricted fishing locations had restrictions on the usage of small-meshed nets or spearguns. Sites were not randomly selected but biased towards sampling marine protected areas and comparable reference sites.

At each location, coral surveys were conducted in 7 countries (Comoros, Kenya, Madagascar, Mauritius, Mayotte, Mozambique, and Tanzania) using a roving observer method. Two experienced observers (N.A. Muthiga and T.R. McClanahan) sampled ~40 m2 over a broad range of ~1000 meters or 40 minutes of sampling. For each survey, an observer assessed the coral community in a series of haphazardly selected replicate quadrats (~2 m2), such that richness was the number of taxa encountered per ~40 m2 (McClanahan et al. 2007). Within each quadrat, hard coral colonies (>5 cm) were identified to genus and the coverage of hard coral was estimated to the nearest 5%. One identification exception was Porites, which was classified as either branching, massive, or Porites rus. A second was Galaxea, which was classified as G. astreata or G. fasciscularis. These divisions were made because of the different life histories and possible functions of these two common genera.

Fish families were sampled by snorkel and scuba diving using two separate belt transect methods. The first method estimated the number of species and biomass for the same 500 m2 belt (5-m x 100-m) transect (McClanahan 2019). One to 9 belt transect replicates were completed per study site between 1991 and 2020. Individual fish were counted in 9 preselected families that included the Acanthuridae, Balistidae, Chaetodontidae, Diodontidae, Labridae, Monacanthidae, Pomacanthidae, Pomacentridae, and Scaridae. These 9 families contain a mix of life histories and fished and unfished species and were chosen for their high number of species and as indicators of the total species richness in a location or region (Allen and Werner 2002). Individuals from the different families were counted during subsequent passes of the same belt transect and combined to create a metric of richness as the number of species per 500 m2 in these 9 families.

The number of coral and fish species were counted in 657 site x transect x time replications. Subsequently, replicated transects within sites were pooled into unique 346 sites of which 2 were outliers. Following outlier and pooling procedure, the final dataset of 344 unique sites had 286 coral and 320 fish replicates. Taxonomic richness was normalized to z-scores (i.e., -3 to +3 SDs) using the residuals of the best-fit species-coral cover and fish species-biomass relationships. The residuals were derived from a best-fit to a logistic model where residuals were extracted and normalized. Residuals are hereafter referred to as residual richness. Standardization was also applied to the GSM and site susceptibility metrics. The normalization reduced sampling bias that might influence testing for associations with environmental factors.