|
nif:isString
|
-
We incorporated local knowledge and expertise into a fishery-independent sampling design that was adopted after a series of workshops with fishermen, harbor officials, and scientists from agency, academic, and conservation organizations [12], [13]. The protocols we developed for monitoring MPAs were based on a stratified random sampling design wherein we used seafloor maps and fishermen’s knowledge to stratify the sampling areas by habitat for nearshore rockfishes (Sebastes spp.) and then sampled those areas by conducting standardized hook-and-line fishing surveys (see [13] for more detail about methods). REF sites for each MPA were selected based on their proximity, similarity in depth, bathymetry, substrate characteristics, and oceanographic conditions found within MPAs. All REF sites were from 0.5–10 km away from the corresponding MPAs. We intentionally chose REF sites close to MPAs because we believed that environmental variation among different areas would be a greater source of difference than any spillover that might occur immediately following MPA designation. Sampling occurred in four different geographic areas along 350 km of coastline. Each area contained a MPA and REF site. Within each site multiple grid cells were established and served as sample units. Sampling areas included three State Marine Reserves that are closed to all fishing (Point Lobos, Piedras Blancas, and Point Buchon), one State Marine Conservation Area that is closed to all fishing (Año Nuevo) but allows harvest of giant kelp (Macrocystis pyrifera) by hand, and the REF sites associated with each of these MPAs (Fig. 1). The Año Nuevo, Point Lobos, Piedras Blancas, and Point Buchon MPAs encompass areas of 26.4 km2, 14.0 km2, 26.9 km2, and 17.4 km2, respectively. Within the boundaries of each MPA and REF site, 500 m by 500 m grid cells were delineated in rocky habitats in water < 40 m deep (to limit fishing mortality associated with barotrauma). A total of 22 grid cells in Año Nuevo, 17 cells in Point Lobos, 57 cells in Piedras Blancas, and 22 cells in Point Buchon were designated, numbered, and then chosen at random to be sampled on a given day. Our work in these MPAs was approved by the California Department of Fish and Wildlife (CDFW) as part of CDFW scientific collecting permits #2613 and #6681. Also, we obtained permission from the California Department of Parks and Recreation to sample in the marine reserve in the Point Lobos State Park.
Figure data removed from full text. Figure identifier and caption: 10.1371/journal.pone.0118502.g001 Location of the four study areas in central California.Año Nuevo State Marine Conservation Area (SMCA), and the Point Lobos, Piedras Blancas, and Point Buchon State Marine Reserves (SMRs). Sampling protocols and ethics statement: Surveys were conducted annually in the four areas from 2007–2013 (except that surveys in the Piedras Blancas area began in 2008). Surveys occurred in the late summer period from mid–July through September when ocean conditions in the region are most consistent. Each MPA or REF site was sampled two days a month (usually within a few days of each other) for two months per year. Before each day of fishing, four of the grid cells in a given MPA or REF site were randomly chosen for sampling. Volunteer anglers were recruited from various fishing clubs, online fishing websites, and from previous collaborative studies. We used a standardized mixture of fishing gear (metal jigs, feathered lures, and barbless baited hooks) in order to capture a variety of species and cover the spectrum of typical hook-and-line fishing gear used in this region [13]. Captured fishes were identified to species, measured, tagged with a T-bar anchor tag (unless the fish was in poor condition or was too small to tag), and released. Lengths reported are total length, defined as the distance from the tip of the snout to the most posterior part of the caudal fin without compressing the tail. We recorded the locations (latitude and longitude) and depths where fishes were released. The effects of barotrauma were reduced with venting needles and descending devices, and by minimizing the duration of time that the fishes were on board the vessel. We aimed to process and release fish in < 5 min in order to minimize effects of barotrauma and handling stress. The San Jose State University and the California Polytechnic State University Institutional Animal Care and Use Committee (IACUC) approved this study as San Jose State University IACUC Animal Protocol number 824 and Cal Poly IACUC Animal Protocol number 1205.
Response variables considered in this study include catch-per-angler-hour (CPUE), biomass caught-per-angler-hour (BPUE), and the mean lengths of fishes caught during a sampling cell visit. Catch rates are reported as the annual mean CPUE of fishing with two hooks per fishing rod and were calculated by dividing the total number of fishes caught by total angler hours of fishing in a sampling cell in a day. BPUE was calculated for each fish caught by converting lengths to weights for each fish using published length-weight relationships for each species. Weights were then used to calculate the total BPUE during each visit to a sampling cell. CPUE and BPUE for each grid cell in each day were averaged over each year to estimate trends in CPUE and BPUE over time in MPA and REF sites. CPUE and BPUE in the 2007 and 2008 sample years were deemed to be starting conditions, except for the pre-existing reserve portion of Point Lobos MPA that was treated as distinct from the newly designated MPA area because fishing has been prohibited there since 1973 (Fig. 2, labeled “1973 MPA Area”). Estimates of mean lengths, CPUE, and BPUE were generated for all species caught, but analyses presented here focus on the eleven most abundant species, those that each represent > 1% of the total catch for all areas combined.
Figure data removed from full text. Figure identifier and caption: 10.1371/journal.pone.0118502.g002 Locations of the sampling in the Point Lobos area.Grid cells, 500 m by 500 m in size, were established as sampling locations in both the Point Lobos marine protected area (MPA) and reference (REF) site. The area designated by diagonal lines around Point Lobos has been closed to all fishing since 1973.
Geographic comparison of species composition: Unpublished maps of the rock lithology and associated habitats in central California (http://walrus.wr.usgs.gov/mapping/csmp/, Accessed 5 January 2015) led us to expect that species compositions should be more similar between a given MPA and REF than among other areas along the coast. To test this, we conducted a multivariate comparison of species composition among areas and between MPA and REF sites. We did this by comparing the similarities between the relative species compositions among areas and sites using cluster analysis and non-metric multi-dimensional scaling (MDS) plots based on Bray-Curtis similarity indices (PRIMER v.6). In addition to the graphical analyses, we compared differences in species compositions among areas with a permutation MANOVA, using the “Vegan” package in the “R” statistical program.
We used a mixed-model, repeated-measures analysis of variance (ANOVA) to test for differences in species abundance and mean sizes of fishes inside and outside of MPAs, both at the time of implementation, and as change over time using data from the entire seven-year sampling period (SAS v9.4). Main effects in the model included area (the four MPA areas surveyed), site (MPA or REF) and sample year as a continuous covariate. A day of sampling in a grid cell was used as the sample unit. Models were run individually for each of the 11 most abundant species recorded, and with all of those species combined. Response variables were evaluated for normality using probability distribution plots prior to analysis, leading us to apply a square root transformation of CPUE and BPUE. Analyses were conducted for each area individually and for all areas together as an analysis of network-wide MPA effects. Also, we conducted a factor analysis in SPSS to identify groups of species with highly correlated distributions, then used those groups to determine how well CPUE obtained from one species could predict relative abundance of co-occurring species.
Multiple-regression analysis using environmental variables: To assess the amount of variance in the BPUE response variable that could be accounted for by spatial (sample cell, site, area) or temporal (sample year) factors relative to environmental conditions at the time of sampling, we conducted a multiple regression analysis using forward model selection and calculated variation accounted for by each variable (semi-partial eta-square). Environmental variables used in this study as potential correlates to individual species’ catch rates included in situ recordings of sea-surface temperature and wind-speed, depth and bottom relief measurements recorded from ship-board echo sounders, observed numbers of harbor seals (Phoca vitulina) and California sea lions (Zalophus californianus) in the area at the time of fishing, buoy-derived estimates of wave energy, winds, and temperature calculated using 3, 24, 48, and 72 hour moving averages, and estimates of depth and relief derived from multibeam sonar data collected and processed by the California Seafloor Mapping Project [14]. Seafloor digital elevation models of the MPA and REF sites generated at 2 m resolution were used to calculate the mean, range, and standard deviation of three types of seafloor characteristics: depth, slope, and rugosity. Mean estimates of each of these bathymetric variables were calculated from the area within each sampling cell. All environmental variables used in model selection were first tested for colinearity (variance inflation factor < 3).
|
![[This page as RDF]](http://data.gesis.org/somesci/static/rdf-icon.gif){kind=icon}