Validation and improvement of species distribution models for structure-forming invertebrates in the eastern Bering Sea with an independent survey

Abstract

Supplement 1. Stereo drop camera descriptionThe two stereo drop camera systems were comprised of two machine-vision cameras spaced approximately 30 cm apart in underwater housings that were connected via ethernet cables to a computer also in an underwater housing. On the first drop-camera system, one of the paired cameras recorded monochromatic still images sized at 1.45 megapixels (JAI, CM-140GE), while the other camera collected 1.73 megapixel color still images (JAI, AB-201GE). On the second drop-camera system, one of the paired cameras recorded monochromatic still images sized at 1.45 megapixels (JAI, CM-140GE), while the other camera collected 2.82 megapixel color still images (Prosilica GX 1920C). Lighting was provided by four strobe lights constructed of four Bridgelux® BXRA LED arrays capable of producing 1,300 lumens at 10.4 W. The computer, cameras, and lights were powered by a 28 V NiMH battery pack. Synchronous images were collected and recorded from each of the cameras at a frequency of one image per second. Each of the systems was enclosed in an aluminum cage to protect the components from damage. Additionally, a 1/4 inch diameter coaxial cable provided a connection from the drop-camera system to the winch at the surface, allowing images from the monochrome camera were viewed in real time at a rate of four images per second. This allowed the height of the camera to be actively controlled to keep it just above the seafloor using a quick response electric winch. Supplement 2. Distribution modeling of structure forming invertebrates and cross-validation using bottom trawl survey and camera survey data Bottom trawl survey modelsThe initial distribution modeling was carried out using bottom trawl survey data collected on the NOAA Fisheries, Alaska Fisheries Science Center, eastern Bering Sea outer shelf and slope surveys from 2002 to 2012 and was reported in Sigler et al. (2015). Briefly, the invertebrate distributions were predicted using generalized additive models (GAM) to determine the relationships between environmental variables (latitude*longitude, depth, slope, long-term average bottom temperature, ocean color, mean current speed, maximum tidal current speed, sediment grain size and sediment sorting which is the standard deviation of grain size) and observations of presence in bottom trawl survey catches for each structure-forming invertebrate group. All modeling was carried out in R software using the mgcv package (Wood 2006) and diagnostics were performed using the PresenceAbsence package. A binomial distribution was used to model presence or absence data and backwards term selection was employed so that the full model including all variables was fit first and

Corals, Canyons, and Conservation: Science Based Fisheries Management Decisions in the Eastern Bering Sea

“…Other published density estimates for coral are available for Pribilof and Zhemchug canyons from a study conducted in 2007 (Miller et al, 2012), where densities of gorgonian corals in Pribilof Canyon were estimated at 0.73 individuals * m −2 , and estimates in Zhemchug Canyon were 0.13 individuals * m −2 . These density estimates are higher than those reported by Rooper C. N. et al (2016), likely due to differences in sampling design. While Rooper C. N. et al (2016) randomized transects, Miller et al (2012 placed transects systematically where coral was likely to be seen based on prior knowledge.…”

“…Results from Rooper C. N. et al (2016) show that deepwater corals are present along the Bering Sea slope, with average densities of 0.005 individuals * m −2 . The densities and heights of corals were highest in and adjacent to Pribilof Canyon.…”

“…Coral densities likely differ between the Aleutian Islands/Bowers Bank, Gulf of Alaska, and Eastern Bering Sea because of differences in the amount of appropriate substrates available for colonization. In the studies above that performed random camera surveys in the Aleutian Islands, Bowers Bank, and Gulf of Alaska, the percentage of transects with rocky habitat (cobble, boulder, exposed bedrock) present somewhere along the transect was 63, 42, and 58%, respectively (C. Rooper, AFSC-RACE Division, unpublished data), compared to 3.2% for the Eastern Bering Sea slope (Rooper C. N. et al, 2016). Deep-sea corals require rocky habitats on which to settle, and results from the Eastern Bering Sea show a distinct lack of appropriate habitat (Rooper C. N. et al, 2016), which explains the low density of deep-sea corals in Pribilof and Zhemchug canyons.…”

“…Corals occurred on 32 of 250 survey transects, and did not occur in the transects sampled in Bering or Pervenets Canyon, or the regions between Bering and Pribilof canyons or between Zhemchug and Pervenets canyons (Figure 1). Corals were identified by Rooper C. N. et al (2016) to three families (Primnoidae, Plexauridae, and Isididae), and two genera (Plumarella, and Swiftia). The densities of corals found were very low, and ranged from 0 to 0.28 individuals * m −2 (mean 0.005, SE 0.002).…”

High-Resolution Habitat Suitability Models for the Conservation and Management of Vulnerable Marine Ecosystems on the Louisville Seamount Chain, South Pacific Ocean

“…The quantity, quality, and distribution of species presence records, the availability of true absence records, and the environmental predictor variables used can all influence the reliability of the models (Araújo and Guisan, 2006;Guisan et al, 2006). Recent efforts to improve the accuracy of habitat suitability models, and thus their usefulness for the management of the impact of fishing on VMEs, have included use of abundance data (as opposed to relying only on presence-absence data), ground-truth model validation, ensemble modeling, and estimates of model uncertainty (Rooper et al, 2014(Rooper et al, , 2016Anderson et al, 2016a,b;Robert et al, 2016).…”