Abstract:

This dataset contains Oyster Reef Restoration Habitat Suitability Index (ORRHSI) of Texas Bays and Estuaries created from Maximum Entropy (MaxEnt) and an oyster reef layer that contains area-specific metrics of successful restoration projects. It includes input and output data for a MaxEnt Habitat Suitability model for the Eastern oyster (Crassostrea virginica) using Texas Parks and Wildlife (TPWD) hydrographic water quality data and catch data of oysters collected in TPWD oysters sampling, bag seines, gill nets, and trawl data spanning January 1986 through December 2016. Water quality and oyster data were extracted from the TPWD long-term Fisheries Independent Monitoring Program for all estuaries in Texas. Spatial representations of water quality were integrated with oyster presence data to create the ORRHSI, which characterizes conditions based on suitability for oyster reef restoration. The oyster reef dataset includes polygons that represent the spatial extent of oyster reefs and contains information about each reef.

Suggested Citation:

Reisinger, Anthony, Jennifer Beseres Pollack, and James Gibeaut. 2020. Oyster Reef Restoration Habitat Suitability Index of Texas Bays & Estuaries. Distributed by: GRIIDC, Harte Research Institute, Texas A&M University–Corpus Christi. doi:10.7266/n7-htxh-3g51

Purpose:

The aim of this project was to create a standardized Oyster Reef Restoration Habitat Suitability Index framework to guide oyster reef restoration in all Texas bay systems. By integrating bay-wide, long-term data on oyster populations and environmental variables, we have created maps that allow users to prioritize restoration actions based on their potential for success. In addition, we created maps that illustrate area-specific metrics of success (spat, oyster density), in order to more effectively evaluate accomplishment of project goals.

Data Parameters and Units:

Water quality rasters inputs: These are single file that represent interpolated values of the quartiles of water quality data collected at TPWD stations. The prefix of the file name represents the temporal quartile of the TPWD station data. File name prefixes for the quartiles are as follows: 0.025, q2p5*.asc, 0.05 q5*.asc, 0.25, q25*.asc, 0.50, q50*.asc, 0.75 q75*.asc, 0.95 q95*.asc, and 0.975, q97.5*.asc. The suffix of the file name represents the TPWD water quality measurements at each station for salinity [practical salinity units (PSU)]; turbidity [nephelometric turbidity units (NTU)]; dissolved oxygen [percent dissolved oxygen]; and temperature [degrees Celsius]. Reef Polygons: RQI.shp feature classes: indnew; Reef Quality Index (RQI); Avg_spat [Average number of spat on oyster shells (double)]; Avg_nDead [average number of dead shells collected in dredge (double)]; Avg_nLive [average number of Live shells collected in dredge (double)]. Habitat Suitability Polygons: OHSIv6.shp grid code; Numeric Habitat Suitability VaHSI rank [rank 1-7 (integer)]; Habitat Suitability Rank [Text description of rank, text].

Methods:

Environmental measurements collected by Texas Parks and Wildlife (TPWD) were imported as point data into a Geographic Information System (GIS; ArcGIS 10.4.1, ESRI). These data were then aggregated based on TPWD sampling stations for bag seines, gill nets, and trawl data. Median latitude and longitude were calculated for each sampling station and these values were used to create a new point dataset for each sampling station in the TPWD dataset. At these sampling stations, temporal percentiles (0.025, 0.05, 0.25, 0.50, 0.75, 0.95, 0.975) were calculated for salinity, turbidity, dissolved oxygen, and temperature measurements to represent the range of temporal variability at each sampling station. The TPWD sampling stations median location were spaced 1 minute of longitude on average from each other (approximately 1.8 km), representing the spatial resolution of the TPWD environmental sampling scheme. Although higher resolution data would potentially reveal more localized patterns and processes, finer-scale data were not available. The temporal percentiles of salinity, turbidity, temperature, and dissolved oxygen were then spatially interpolated across the all the estuaries of the Texas coast using a local polynomial interpolation weighted by the temporal frequency of sampling at each station and gridded using a 250-meter cell size. Oyster dredge sampling data from TPWD were also imported as point data into the GIS. These data were then converted into live reef polygons using our previous methodology employed in Pollack et al 2012. For each contiguous live reef polygon, temporal aggregates of oyster sampling data were calculated which included a mean abundance of live oysters (>25 mm shell length), dead shell (>25 mm shell length), and spat (5–25 mm shell length) on live oysters and dead shell were calculated for each continuous reef polygon. A Reef suitability Index was calculated for each contiguous reef using the methodology in Pollack et al 2012. To create a presence dataset for oysters we extracted all latitude and longitude values of any oysters collected in TPWD oysters sampling, bag seines, gill nets, and trawl data spanning January 1986 through December 2016. If a live oyster was caught in any of the sampling we extracted from the TPWD dataset and imported them into the GIS. To develop a more robust oyster presence dataset we then filtered out spurious occurrences of oysters by only including samplings that caught five or more oysters per event. This resulted in 25,674 separate occurrences of oysters across all Texas estuaries for the time period. Oyster presence data were then resampled to match the TPWD temporal percentiles 250-meter grid cells. If more than one oyster occurrence was present in a single grid cell it was included as a single presence value for that grid cell. This was done so duplicates would not overfit the model to these sites with multiple occurrences in a single grid cell. Habitat Suitability Modeling: The habitat suitability index for oysters was created using the MaxEnt modeling software using a presence-only model for oysters. The model was randomly divided into training (70%) and test (30%) data sets to evaluate model performance; testing data were withheld from the model and used for evaluation (Phillips et al. 2006). The inputs for the model were the aforementioned temporal percentiles for salinity, turbidity, temperature, and dissolved oxygen and the presence data for oysters. Background samples (n = 10,000) were used to generate pseudo-absences for the model. Rasters of oyster habitat suitability index were then classified into 7 groups using the Natural Jenks with the highest values receiving a ranking of 1 to lowest receiving a 7 and converted into polygons. A polygon of navigation channels was buffered by 250 meters and overlaid on the HSI and assigned a 7.

Provenance and Historical References:

Beseres Pollack J, Cleveland A, Palmer TA, Reisinger AS, Montagna PA (2012) A Restoration Suitability Index Model for the Eastern Oyster (Crassostrea virginica) in the Mission-Aransas Estuary, TX, USA. PLoS ONE 7(7): e40839. https://doi.org/10.1371/journal.pone.0040839 Phillips, S. J., Anderson, R. P., & Schapire, R. E. (2006). Maximum entropy modeling of species geographic distributions. Ecological Modelling, 190(3-4), 231–259. https://doi.org/10.1016/j.ecolmodel.2005.03.026

Individual Files: