Analyses/Maps

Fish

Habitat Suitability Modeling/Maps

Introduction

Habitat suitability modeling (HSM) is a tool for predicting the suitability of habitat for a given species based on known affinities with environmental parameters. This technique was chosen for this project to provide a synoptic view of habitat suitability for specific species as well as assess habitat suitability for species assemblages. One HSM technique is termed "habitat suitability index (HSI) modeling". HSI models are simple mathematical expressions for calculating a unitless index of habitat quality as a function of one or more environmental variables. Using GIS, these index values can be mapped and analyzed to portray areas of potential distribution for a species (Brown et al., 2000). High-quality habitat may provide high carrying capacity and support higher rates of growth, survival, or reproduction for a given species, whereas low-quality or unsuitable habitat may have little or no carrying capacity (Brown et al., 2000). The HSI methods were adapted from the U.S. Fish and Wildlife Service (USFWS) Habitat Evaluation Procedures program (USFWS, 1980a, 1980b, 1981) to provide spatially explicit estimates of suitability across the entire study area. It is important to note that the model results depict potentially suitable habitat for a given species and not actual distribution. This section provides the methodology, results, validation, and interpretation of HSI models developed for selected adult and juvenile stages of commercially and recreationally important groundfish and invertebrate species. The models are based on species' affinities to substrate types and bathymetric ranges (Monaco et al., 1998).

Summary of Habitat Suitability Model Maps

Blue Rockfish	Description	PDF Map
Boccacio	Description	PDF Map
California Market Squid	Description	PDF Map
Canary Rockfish	Description	PDF Map
Chilipepper Rockfish	Description	PDF Map
Darkblotched Rockfish	Description	PDF Map
Dover Sole	Description	PDF Map
Dungeness Crab	Description	PDF Map
English Sole	Description	PDF Map
Hake	Description	PDF Map
Lingcod	Description	PDF Map
Longspine Thornyhead	Description	PDF Map
Petrale Sole	Description	PDF Map
Pacific Sanddab	Description	PDF Map
Rex Sole	Description	PDF Map
Sablefish	Description	PDF Map
Shortspine Thornyhead	Description	PDF Map
Widow Rockfish	Description	PDF Map
Yelloweye Rockfish	Description	PDF Map
Yellowtail Rockfish	Description	PDF Map
All Flatfish	Description	PDF Map
Mean Species Overlap	Description	PDF Map
All Rockfish	Description	PDF Map
Shallow Shelf Assemblage	Description	PDF Map
Slope Assemblage	Description	PDF Map

Download Habitat Suitability Model ArcView Grids

Data and Analyses

Environmental Data: Initially bathymetry, benthic substrate type, and bottom temperature were chosen as the environmental data to be included in the models. Although water temperature is an influential factor that affects species distributions and movements (See ELR), several factors led to the exclusion of bottom temperature from final model development: 1) information regarding species associations with bottom temperature were too general or absent from scientific literature; 2) statistical analyses revealed collinearity between bottom temperatures collected with NMFS trawl samples and bathymetry; and 3) since most of the species modeled are benthic organisms where bottom temperature is not highly variable, numerous authors state that depth is the most significant factor regulating species distributions (Gabriel and Tyler, 1980; Matthews and Richards, 1991; Yoklavich et al., 2000; Williams and Ralston, 2002). As a result, water temperature was eliminated as a modeling variable. This does not preclude using water temperature as a variable for modeling pelagic species, however, more information will have to be collected to explore their affinities for this variable. Based on these considerations, bathymetry and substrate data were used to map HSI model results (See HSI Results Ð Mapping below). Numerous data sources were combined to produce a digital, high resolution map of bathymetry. Bathymetry was rasterized with 70 m cell size for most of the study area for depths to 4810 m (See Section 2.1.X). Benthic substrate was mapped from Pt. Arena in the north to Pt. Sal in the south to conform to the latitudinal limits of the study area. Substrates were characterized using 5 classifications: sand, mud, rock, pebble/cobble/gravel, and mud/rock mix. See Section 2.1.X for more information on benthic substrate map development.

Species selected for HSM: The primary criteria to select species for which HSMs were developed was their commericial and ecological importance. In addition, several species were included based on recommendations by staff members from the NMSP. Overall, 20 species were modeled, 14 of which included models for adult and juvenile distribution. Species with two life stage models include: bocaccio, canary rockfish, chilipepper rockfish, darkblotched rockfish, longspine thornyhead, shortspine thornyhead, lingcod, sablefish, Pacific whiting (hake), dover sole, english sole, petrale sole, rex sole, and Pacific sanddab. Potential adult distributions were modeled for Dungeness crab, California market squid, blue rockfish, widow rockfish, yelloweye rockfish, and yellowtail rockfish. Some of these species were chosen to represent species assemblages as determined in Section 2.1.1, and mapped to display the potential distribution of suitable habitats for the assemblage. For example, cluster analyses determined that Dover sole, sablefish, and shortspine thornyhead were commonly captured in NMFS trawl surveys that indicated a deep water shelf assemblage for these species.

HSI Data/SI Development: Initially, suitability index (SI) values for bathymetry and substrate type were developed through literature review and modeled in GIS. During October 2002, the methodological approach and results were peer-reviewed by NMSP and NMFS staff (see review panel below) who suggested that, where sufficient data were available, bathymetry SI values should be developed using NMFS trawl data. In addition, the panel requested separate models for adults and juveniles. As a result, a subset of NMFS trawl data on the shelf (1977-1995) and slope (1984-1999) for the entire west coast were used to develop SI values for adults and juveniles for most species. SI values for bathymetry were developed from NMFS trawl data by fitting a polynomial regression to bathymetric classes and mean species abundance (log transformed) (Figure 1). Since trawl samples were not collected in waters less than 50 m, the bathymetric classes begin at 50 m with a range of 20 m between classes. The fitted curve was weighted by sampling effort to account for disproportionate sample sizes within bathymetric classes. Predicted mean abundance along the curve was then used to calculate SI's for each bathymetric class by dividing each mean abundance value by the maximum observed across the bathymetric gradient (Table 1) (Rubec et al., 1999). Resultant values were multiplied by 10 to scale SI's by whole integers (0-10) as reclassification of environmental grids is done using ArcView which does not recognize decimals. For species that had limited or no trawl data, SI values were developed from bathymetric ranges reported in the literature (Christensen et al., 1997, Brown et al., 2000). Table 2 displays a sample data matrix generated from literature sources, where presence (1) or absence (0) is coded within the bathymetric classes for each particular species and life stage. In this technique, the total number of references that denote presence of the species are summed within each depth class and then divided by the total number of references examined to obtain the final SI value. Literature review provided only general ranges of species occurrence in relation to bathymetry, therefore, classes of 50 m were chosen to confidently develop SI values rather than the 20 m classes used above. Differentiating depth ranges for adults and juveniles from literature sources was difficult due to lack of data, therefore, only adult SI values could be developed using this technique. SI's for affinities with substrate were also created using this technique. SI's for juveniles based on bathymetry were developed using NMFS trawl data, when available, or were simply not modeled where trawl data was limited or absent. Contrasting evidence exists within the literature that bathymetric preferences can shift for many groundfish species based on latitude (PFMC, 1999; Williams and Ralston, 2002). For the present study, it was assumed that depth preference was similar regardless of latitude, although further exploration into this reported trend is currently underway. Similarly, preference for substrate was assumed to be the same throughout each species range.

The NMFS trawls were conducted within waters of depths between 50 Ð 1300 m; therefore bathymetric SI's outside this range could not be calculated. Depth information within the literature exists for most species outside of this range, but was omitted from modeling and mapping to match the depth range associated with the NMFS trawls. Trawls were conducted during June to November, thus models for different "seasons" could not be created. Thus, modeled map surfaces represent species potential distributions for water depths from 50-1300 meters and for the summer and late fall time period. Also, many of the species modeled exhibit inshore/offshore migrations based on habitat shifts associated with life history requirements and/or spawning activity. Additional data will have to be collected to reflect these shifts in abundance and distribution; therefore, no attempt was made to model them here.

HSI Results-Mapping: Once SI values were determined for bathymetry and substrate type, these values were inserted into the environmental grids (see Environmental Variables page xx). Once each species' suitability indices were derived (either through regression or through the literature), the values were combined with the bathymetry and substrate map layers (see Environmental Variables) to calculate an index of habitat suitability. The habitat suitability was calculated as the geometric mean of suitability indices (Si) for the two (n) environmental factors (Rubec et al. (1999):

The resulting maps display the potential suitability of cells for each species based on the strength of their affinities to depth and substrate type in each cell. The map displays habitat suitability in a unitless index from 0 (unsuitable) to 10 (highly suitable).

Validation of HSI Model: The remaining subsets (i.e. independent data) of NMFS trawl data from the shelf (1998-2001) and slope (2000-2001) were used to assess model performance. Mean abundance was calculated for each species from these data and superimposed over the predicted HSI values and compared by regressing observed catch data on the predicted HSI values. The statistical results are not intended to be definitive tests of the model, but provide supporting evidence for the existence and strength of the relationship between the model predictions and the catch data. It is important to note that these models are based on two independent parameters and are not the definitive predictors of habitat utilization for these species. Fishery-dependent data from CDF&G recreational surveys were also used to validate models for species that had limited trawl information, i.e. species that display affinities for rocky substrates (rockfishes) and had poor representation in trawl data. If the model performs correctly, this validation procedure should demonstrate increasing mean abundance with increasing habitat suitability.

Integrative Maps: Management plans are often developed for a group of species that exhibit similar life history strategies. Selected species assemblages, as defined in Section 2.1.1, were analyzed and mapped to identify the spatial distribution of their important habitats. In addition, two analyses were conducted which examine the overlap of highly suitable habitat based on all species for which HSI maps were developed. Areas with the most overlap of high suitability could be considered important habitats for selected groundfish.

Analytical Products

As part of the biogeographic assessment, digital data were developed as products from the study. Digital bathymetry and substrate maps were created as ArcView shape and raster files. Three representative HSI models are presented: bocaccio (adult and juvenile), Dover sole (adult and juvenile), and Dungeness crab (adult). The remaining 31 species HSI maps are on the CD-ROM. Representative maps displaying habitat importance based on all HSI models and select species assemblages are also included. Additional integrative maps for shelf assemblages, all rockfish, and all flatfish, are also included on the CD-ROM.