The identification of efficient management strategies that reduce protected species bycatch while also minimizing impacts on fishing livelihoods is a global conservation challenge. Identifying such strategies requires understanding levels of bycatch relative to management targets as well as the relationship between bycatch risk and potential management actions. In this study, we use ratio estimation to reconstruct bycatch of select marine mammal and seabird species in the California >= 3.5" set gillnet fishery from 1981 to 2022 and random forest models to identify potential drivers and hotspots of bycatch risk. We find that bycatch has dropped precipitously since the 1980s as a result of management-induced reductions in fishing effort, but at significant costs to fisheries participation and revenues. Recent marine mammal bycatch ranges from 0.1 % to 4.0 % of the potential biological removal and marine mammal populations are recovering. Spatial-temporal correlates of bycatch risk were more important than fishing-related correlates of risk, suggesting that spatial-temporal closures would reduce bycatch more reliably than mesh size or soak time restrictions. For each species, we identified 1-3 hotspots of bycatch risk as candidates for temporary seasonal closures. Bycatch risk for harbor seal (Phoca vitulina) and California sea lion (Zalophus californianus), the species with the greatest bycatch risk, is especially high from April 1st to June 15th, suggesting that hotspot closures during this 2.5-month time period could be particularly efficient. Our study also highlights the value of competing multiple sample balancing approaches to identify methods that best predict rare bycatch events.

Entanglements and vessel strikes impact large whales worldwide. Post-event health status is often unknown because whales are seen once or over short spans that conceal long-term health declines. Well-studied populations with high site fidelity verified by photo-ID offer opportunity to confirm deaths, health declines and recoveries. We used known outcome entanglements and vessel strikes of right whales ( Eubalaena glacialis ) and humpback whales ( Megaptera novaeangliae ) to model probabilities of deaths, health declines and recoveries with Random Forest (RF) classification trees. Variables included presence or absence of phrases from case narratives (‘deep laceration’, ‘cyamid’, ‘healing’, ‘superficial’) and a categorical variable for vessel size. Health status post-entanglement was correctly classified in 95.7% of right whale and 93.6% of humpback whale cases (expected by chance=50%). Health status post-vessel strike was correctly classified in 91.4% of right whale and 88.6% of humpback whale cases. Important variables included cyamid presence, emaciation, discolored skin, constricting entanglements, gear-free resightings, superficial or healing lacerations, and vessel size. Cross-validated RF models were applied to unknown outcome cases to estimate the probability of deaths, health declines and recoveries. Total serious injuries (probability of death or health decline > 0.50) assigned by RF were nearly equal to current injury assessment methods applied by biologists for known outcomes. However, RF consistently predicted higher serious injury totals for unknown outcomes, suggesting that current assessment methods may underestimate risk for cases lacking details or long-term observations. Advantages of the RF method include: 1) risk models are based on known outcomes; 2) unknown outcomes are assigned post-event health status probabilities; and 3) identification of important predictor variables improves data collection standards.

Harbor porpoises, Phocoena phocoena, off California, comprise four recognized population stocks: Morro Bay (MOR), Monterey Bay (MRY), San Francisco-Russian River (SFRR), and Northern California-Southern Oregon (NCSO). The three southernmost stocks experienced substantial bycatch in gill net fisheries during the 1970s and 1980s. While the SFRR stock received full protection from gill nets in 1989, the MOR and MRY stocks continued to experience at least some bycatch through 2001-2002. We examined long-term population trends for these four harbor porpoise stocks, based on two sets of systematic, aerial line-transect surveys conducted off California during summer/fall of 1986-2017. We applied a Bayesian hierarchical framework to specify a process model of population density and an observation model of porpoise counts during line-transect surveys. Growth rates were estimated for periods with and without bycatch. Posterior distributions indicate the MOR, MRY, and SFRR stocks, respectively, grew at 9.6%, 5.8%, and 6.1% per year after gill nets were largely or fully eliminated for each stock. Abundance off northern California appears stable or slightly increasing. This study provides a first empirical estimate of maximum net reproductive rate for harbor porpoise (at least 9.6%), and demonstrates that porpoise populations can recover from substantial gill net impacts if bycatch is eliminated.

Protected species bycatch can be rare, making it difficult for fishery managers to develop unbiased estimates of fishing-induced mortality. To address this problem, we use Bayesian time-series models to estimate the bycatch of humpback whales ( Megaptera novaeangliae ), which have been documented only twice since 2002 by fishery observers in the United States West Coast sablefish pot fishery, once in 2014 and once in 2016. This model-based approach minimizes under- and over-estimation associated with using ratio estimators based only on intra-annual data. Other opportunistic observations of humpback whale entanglements have been reported in United States waters, but, because of spatio-temporal biases in these observations, they cannot be directly incorporated into the models. Notably, the Bayesian framework generates posterior predictive distributions for unobserved entanglements in addition to estimates and associated uncertainty for observed entanglements. The United States National Marine Fisheries Service began using Bayesian time-series to estimate humpback whale bycatch in the United States West Coast sablefish pot fishery in 2019. That analysis resulted in estimates of humpback whale bycatch in the fishery that exceeded the previously anticipated bycatch limits. Those results, in part, contributed to a review of humpback whale entanglements in this fishery under the United States Endangered Species Act. Building on the humpback whale example, we illustrate how the Bayesian framework allows for a wide range of commonly used distributions for generalized linear models, making it applicable to a variety of data and problems. We present sensitivity analyses to test model assumptions, and we report on covariate approaches that could be used when sample sizes are larger. Fishery managers anywhere can use these models to analyze potential outcomes for management actions, develop bycatch estimates in data-limited contexts, and guide mitigation strategies.

AbstractSpecies distribution models (SDMs) are important management tools for highly mobile marine species because they provide spatially and temporally explicit information on animal distribution. Two prevalent modeling frameworks used to develop SDMs for marine species are generalized additive models (GAMs) and boosted regression trees (BRTs), but comparative studies have rarely been conducted; most rely on presence‐only data; and few have explored how features such as species distribution characteristics affect model performance. Since the majority of marine species BRTs have been used to predict habitat suitability, we first compared BRTs to GAMs that used presence/absence as the response variable. We then compared results from these habitat suitability models to GAMs that predict species density (animals per km2) because density models built with a subset of the data used here have previously received extensive validation. We compared both the explanatory power (i.e., model goodness of fit) and predictive power (i.e., performance on a novel dataset) of the GAMs and BRTs for a taxonomically diverse suite of cetacean species using a robust set of systematic survey data (1991–2014) within the California Current Ecosystem. Both BRTs and GAMs were successful at describing overall distribution patterns throughout the study area for the majority of species considered, but when predicting on novel data, the density GAMs exhibited substantially greater predictive power than both the presence/absence GAMs and BRTs, likely due to both the different response variables and fitting algorithms. Our results provide an improved understanding of some of the strengths and limitations of models developed using these two methods. These results can be used by modelers developing SDMs and resource managers tasked with the spatial management of marine species to determine the best modeling technique for their question of interest.

Observer program design and evaluation often overlook the challenges of documenting rare-event bycatch. To support and facilitate consideration of threatened, endangered, and protected species bycatch in evaluating observer programs and assessing fisheries impacts, we developed a software tool to assess observer coverage with respect to several objectives for documenting or estimating rare-event bycatch. The ObsCovgTools package for the R programming language, also available as an online application, predicts observer coverage performance for a given total fishery effort in relation to three metrics: (1) the conditional probability of observing any bycatch given that bycatch occurred in the fishery and the probability of any bycatch in the total fishery effort, (2) the upper confidence limit for total bycatch when none is observed, and (3) precision (coefficient of variation) of the bycatch estimate. We describe the tool; explore how specific observer coverage targets for these metrics vary with total effort, BPUE, and dispersion index; and apply it to evaluate observer coverage in the California drift gillnet fishery. Our results underscore the importance of considering effort as well as percentage in assessing how well an observer program documents bycatch. We caution that rare species interactions may not be documented in many observer programs, and should be anticipated through a complementary risk assessment approach. The tool's modular design and open source programming approach encourage adaptation and augmentation to address additional objectives or complexities in sampling design or estimation.

Pacific leatherback turtles (Dermochelys coriacea) are critically endangered, and declines have been documented at multiple nesting sites throughout the Pacific. The western Pacific leatherback forages in temperate and tropical waters of the Indo-Pacific region, and about 38–57% of summer-nesting females from the largest remaining nesting population in Papua Barat (Indonesia) migrate to distant foraging grounds off the U.S. West Coast, including neritic waters off central California. In this study, we examined the trend in leatherback abundance off central California from 28 years of aerial survey data from coast-wide and adaptive fine-scale surveys. We used a Bayesian hierarchical analysis framework, including a process model of leatherback population density and an observation model relating leatherback observations to distance sampling methods. We also used time-depth data from biologgers deployed on 21 foraging leatherback turtles in the study area to account for detection biases associated with diving animals. Our results indicate that leatherback abundance has declined at an annual rate of −5.6% (95% credible interval −9.8% to −1.5%), without any marked changes in ocean conditions or prey availability. These results are similar to the nesting population trends of −5.9% and −6.1% per year estimated at Indonesian index beaches, which comprise 75% of western Pacific nesting activity. Combined, the declining trends underscore the need for coordinated international conservation efforts and long-term population monitoring to avoid extirpation of western Pacific leatherback turtles.

Research vessel and aerial platforms were used between 1997 and 2000 to collect genetic and photographic data from a small populationof right whales that summers in the southeastern Bering Sea. Totals of 11 and six unique individuals were identified using photographicand genetic methods, respectively. Single matches between years occurred using both methods, and all genetic samples turned out to befrom male whales. Long-term research is needed to estimate the size of this population and to determine what threats the whales may befacing.

ABSTRACTThe California sea lion (Zalophus californianus) population in the United States has increased steadily since the early 1970s. The Marine Mammal Protection Act of 1972 (MMPA) established criteria for management of marine mammals based on the concept of managing populations within the optimal sustainable population (OSP), defined as a range of abundance from the maximum net productivity level (MNPL) to carrying capacity (K). Recent declines in California sea lion pup production and survival suggest that the population may have stopped growing, but the status of the population relative to OSP and MNPL is unknown. We used a time series of pup counts from 1975 to 2014 and a time series of mark‐release‐resight‐recovery data from 1987 to 2015 for survival estimates to numerically reconstruct the population and evaluate the current population status relative to OSP using a generalized logistic model. We demonstrated that the population size in 2014 was above MNPL and within its OSP range. However, we also showed that population growth can be dramatically decreased by increasing sea surface temperature associated with El Niño events or similar regional ocean temperature anomalies. In this analysis we developed a critical tool for management of California sea lions that provides a better understanding of the population dynamics and a scientific foundation upon which to base management decisions related to complex resource issues involving this species. Published 2018. This article is a U.S. Government work and is in the public domain in the USA.

Whale entanglements in US west coast fishing gear are largely represented by opportunistic sightings, and some reports lack species identifications due to rough seas, distance from whales, or a lack of cetacean identification expertise. Unidentified entanglements are often ignored in species risk assessments and thus, entanglement risk is underestimated. To address this negative bias, a species identification model was built from random forest (RF) classification trees using 199 identified entanglements (‘model data’). Humpback Megaptera novaeangliae and gray whales Eschrichtius robustus represented 92% of identified entanglements; the remaining 8% were minke whales Balaenoptera acutorostrata, fin whales B. physalus, blue whales B. musculus, and sperm whales Physeter macrocephalus. Predictor variables included year, gear type, location, season, sea surface temperature, water depth, and a multivariate El Niño index. Crossvalidated species classifications were correct in 78% (155/199) of cases, significantly higher (p < 0.001, permutation test) than the 49% correct classification rate expected by chance. The RF model correctly classified 91% of humpback whale cases, 64% of gray whale cases, and 100% of sperm whale cases, but misclassified all minke, blue, and fin whale cases. The cross-validated RF classification-tree species model was used to classify 35 entanglements without species identifications (‘novel data’) and each case was assigned a probability of belonging to each of 6 model data species. This approach eliminates the negative bias associated with ignoring unidentified entanglements in species risk assessments. Applications to other wildlife studies where some detections are unidentified include fisheries bycatch, line-transect surveys, and large-whale vessel strikes.

Recovery of cetacean carcasses provides data on levels of human-caused mortality, but represents only a minimum count of impacts. Counts of stranded carcasses are negatively biased by factors that include at-sea scavenging, sinking, drift away from land, stranding in locations where detection is unlikely, and natural removal from beaches due to wave and tidal action prior to detection. We estimate the fraction of carcasses recovered for a population of coastal bottlenose dolphins (Tursiops truncatus), using abundance and survival rate data to estimate annual deaths in the population. Observed stranding numbers are compared to expected deaths to estimate the fraction of carcasses recovered. For the California coastal population of bottlenose dolphins, we estimate the fraction of carcasses recovered to be 0.25 (95% CI = 0.20– 0.33). During a 12 yr period, 327 animals (95% CI = 253–413) were expected to have died and been available for recovery, but only 83 carcasses attributed to this population were documented. Given the coastal habits of California coastal bottlenose dolphins, it is likely that carcass recovery rates of this population greatly exceed recovery rates of more pelagic dolphin species in the region.

1Marine Mammal and Sea Turtle Division, Southwest Fisheries Science Center, NOAA, La Jolla, CA 92037. 2Marine Mammal Laboratory, Alaska Fisheries Science Center, NOAA, Seattle, WA 98115. 3Protected Resources Division, NMFS West Coast Region, NOAA, Long Beach, CA, 90802. 4Protected Resources Division, NMFS West Coast Region, NOAA, Seattle, WA 98115. 5Fishery Resource and Monitoring Division, Northwest Fisheries Science Center, NOAA, Seattle, WA 98112.