Catches per unit of effort of bigeye tuna: a new analysis with regression trees and simulated annealing

ENGLISH: We analyzed catches per unit of effort (CPUE) from the Japanese longline fishery for bigeye tuna (Thunnus obesus) in the central and eastern Pacific Ocean (EPO) with regression tree methods. Regression trees have not previously been used to estimate time series of abundance indices fronl CPUE data. The "optimally sized" tree had 139 parameters; year, month, latitude, and longitude interacted to affect bigeye CPUE. The trend in tree-based abundance indices for the EPO was similar to trends estimated from a generalized linear model and fronl an empirical model that combines oceanographic data with information on the distribution of fish relative to environmental conditions. The regression tree was more parsimonious and would be easier to implement than the other two nl0dels, but the tree provided no information about the nlechanisms that caused bigeye CPUEs to vary in time and space. Bigeye CPUEs increased sharply during the mid-1980's and were more variable at the northern and southern edges of the fishing grounds. Both of these results can be explained by changes in actual abundance and changes in catchability. Results from a regression tree that was fitted to a subset of the data indicated that, in the EPO, bigeye are about equally catchable with regular and deep longlines. This is not consistent with observations that bigeye are more abundant at depth and indicates that classification by gear type (regular or deep longline) may not provide a good measure of capture depth. Asimulated annealing algorithm was used to summarize the tree-based results by partitioning the fishing grounds into regions where trends in bigeye CPUE were similar. Simulated annealing can be useful for designing spatial strata in future sampling programs. SPANISH: Analizamos la captura por unidad de esfuerzo (CPUE) de la pesquería palangrera japonesa de atún patudo (Thunnus obesus) en el Océano Pacifico oriental (OPO) y central con métodos de árbol de regresión. Hasta ahora no se han usado árboles de regresión para estimar series de tiempo de índices de abundancia a partir de datos de CPUE. EI árbol de "tamaño optimo" tuvo 139 parámetros; ano, mes, latitud, y longitud interactuaron para afectar la CPUE de patudo. La tendencia en los índices de abundancia basados en árboles para el OPO fue similar a las tendencias estimadas con un modelo lineal generalizado y con un modelo empírico que combina datos oceanográficos con información sobre la distribución de los peces en relación con las condiciones ambientales. EI árbol de regresión fue mas parsimonioso y seria mas fácil de utilizar que los dos otros modelos, pero no proporciono información sobre los mecanismos que causaron que las CPUE de patudo valiaran en el tiempo y en el espacio. Las CPUE de patudo aumentaron notablemente a mediados de los anos 80 y fueron mas variables en los extremos norte y sur de la zona de pesca. Estos dos resultados pueden ser explicados por cambios en la abundancia real y cambios en la capturabilidad. Los resultados de un arbal de regresión ajustado a un subconjunto de los datos indican que, en el OPO, el patudo es igualmente capturable con palangres regulares y profundos. Esto no es consistente con observaciones de que el patudo abunda mas a profundidad e indica que clasificación por tipo de arte (palangre regular 0 profundo) podría no ser una buena medida de la profundidad de captura. Se uso un algoritmo de templado simulado para resumir los resultados basados en el árbol clasificando las zonas de pesca en zonas con tendencias similares en la CPUE de patudo. El templado simulado podría ser útil para diseñar estratos espaciales en programas futuros de muestreo. (PDF contains 45 pages.)

Eventos de vida produtores de estresse e acidentes de trabalho: Estudo Pró-Saúde, Rio de Janeiro - Brasil

A proposta desta tese consiste em um recorte de uma abordagem mais ampla da determinação dos acidentes de trabalho, e tem como objetivo geral investigar o perfil de acidentes de trabalho que acometem funcionários técnico-administrativos do quadro efetivo de uma universidade do Rio de Janeiro. Para o alcance do objetivo, esta tese está estruturada em dois artigos, e como tal pretende explorar o perfil sócio-demográfico e ocupacional de funcionários públicos na ocorrência de acidentes de trabalho (Artigo 1); e investigar a associação entre os eventos de vida produtores de estresse (EVPE) e a ocorrência de acidentes de trabalho (Artigo 2). Dados seccionais da fase 2 de um estudo de natureza prospectiva (Estudo Pró-Saúde) foram coletados entre 3572 funcionários. A história de acidentes de trabalho foi captada por meio de perguntas dicotômicas (sim vs. não) para cada um dos seguintes tipos de acidentes: perfuração com agulha; perfuração com outro objeto; corte; queimadura; choque elétrico; contusão ou distensão muscular; fratura, entorse ou luxação; e envenenamento ou intoxicação. O período de referência para aferição tanto dos EVPE quanto da ocorrência de acidente de trabalho correspondeu aos 12 meses anteriores a aplicação de questionário autopreenchível. No artigo 1 utilizou-se a técnica de análise de correspondência múltipla para delimitar agrupamentos de funcionários quanto ao perfil sócio-demográfico e ocupacional associado à ocorrência de acidente de trabalho, de acordo com as seguintes Característica : sexo, idade, escolaridade, renda per capita, ocupação, setor e local de trabalho. No artigo 2, a associação entre EVPE e acidentes de trabalho foi avaliada através de análise multivariada por meio de modelo lineares generalizados (logpoisson), sendo os resultados expressos através de razões de prevalência (RP) ajustadas e seus respectivos intervalos de 95% de confiança (IC95%). A prevalência total de acidentes no período de 12 meses foi de 25,6%. Dos tipos de acidentes referidos, o mais frequente foi a contusão ou fratura, com cerca de (10,2%) de relatos. Em seguida, aparecem as perfurações com agulha (6,5%). Os resultados da análise de correspondência revelam três grupos, destacando-se aquele formado pelos que sofreram perfuração com agulha com um perfil que abrange os auxiliares de enfermagem, trabalham no Hospital Universitário e setores adjacentes, especificamente em setores de terapia intensiva, emergência, cirurgia geral, clinica geral e ambulatório. Em relação à associação com EVPE, ter sido testemunha de agressão foi o evento mais fortemente associado com acidentes de trabalho (RP= 1,98, IC95%= 1,67; 2,34). Este estudo trouxe informações acerca da importância das características sócio-demográficas e de aspectos psicossociais na ocorrência dos acidentes de trabalho que podem ser úteis na elaboração de medidas para a prevenção desse importante problema de saúde pública.

Growth of Skipjack tuna (Katsuwonus pelamis) in the eastern Pacific Ocean as estimated from tagging data

English: Data obtained from tagging experiments initiated during 1953-1958 and 1969-1981 for skipjack tuna from the coastal eastern Pacific Ocean (EPO) are reanalyzed, using the Schnute generalized growth model. The objective is to provide information that can be used to generate a growth transition matrix for use in a length-structured population dynamics model. The analysis includes statistical approaches to include individual variability in growth as a function of length at release and time at liberty, measurement error, and transcription error. The tagging data are divided into northern and southern regions, and the results suggest that growth rates differ between the two regions. The Schnute model provides a significantly better fit to the data than the von Bertalanffy model, a sub-model of the Schnute model, for the northern region, but not for the southern region. Individual variation in growth is best described as a function of time at liberty and as a function of growth increment for the northern and southern regions, respectively. Measurement error is a significant part of the total variation, but the results suggest that there is no bias caused by the measurement error. Additional information, particularly for small and large fish, is needed to produce an adequate growth transition matrix that can be used in a length-structured population dynamics model for skipjack tuna in the EPO. Spanish: Los datos obtenidos de los experimentos de marcado iniciados durante los períodos de 1953- 1958 y de 1969-1981 para el atún barrilete en las costas del Océano Pacífico Oriental (OPO) fueron analizados nuevamente, utilizando el modelo de crecimiento generalizado de Schnute. El objetivo es brindar información que sea útil para producir una matriz sobre la tran-sición de crecimiento que pueda utilizarse en un modelo de dinámica poblacional estructurado por talla. El análisis usa enfoques estadísticos para poder incluir la variabilidad individual del crecimiento como función de la talla de liberación y tiempo en libertad, el error de medición, y el error de transcripción. Los datos de marcado son divididos en regiones norte y sur, y los resultados sugieren que las tasas de crecimiento en las dos regiones son diferentes. En la región norte, pero no en la región sur, el modelo de Schnute se ajusta significativamente mejor a los datos que el modelo von Bertalanffy, un sub-modelo del modelo de Schnute. La mejor descripción de la variación individual en el crecimiento es como una función del tiempo en libertad y como una función del incremento de crecimiento para las regiones norte y sur, respectivamente. El error de medición es una parte significativa de la variación total, pero los resultados sugieren que no existe un sesgo causado por el error de medición. Se necesita información adicional, particularmente para peces pequeños y grandes, para poder producir una matriz de transición de crecimiento adecuada que pueda utilizarse en el modelo de dinámica poblacional estructurado por tallas para el atún barrilete en el OPO.

Estimation of the growth curve parameters in Macrobrachium rosenbergii

Growth is one of the most important characteristics of cultured species. The objective of this study was to determine the fitness of linear, log linear, polynomial, exponential and Logistic functions to the growth curves of Macrobrachium rosenbergii obtained by using weekly records of live weight, total length, head length, claw length, and last segment length from 20 to 192 days of age. The models were evaluated according to the coefficient of determination (R2), and error sum off square (ESS) and helps in formulating breeders in selective breeding programs. Twenty full-sib families consisting 400 PLs each were stocked in 20 different hapas and reared till 8 weeks after which a total of 1200 animals were transferred to earthen ponds and reared up to 192 days. The R2 values of the models ranged from 56 – 96 in case of overall body weight with logistic model being the highest. The R2 value for total length ranged from 62 to 90 with logistic model being the highest. In case of head length, the R2 value ranged between 55 and 95 with logistic model being the highest. The R2 value for claw length ranged from 44 to 94 with logistic model being the highest. For last segment length, R2 value ranged from 55 – 80 with polynomial model being the highest. However, the log linear model registered low ESS value followed by linear model for overall body weight while exponential model showed low ESS value followed by log linear model in case of head length. For total length the low ESS value was given by log linear model followed by logistic model and for claw length exponential model showed low ESS value followed by log linear model. In case of last segment length, linear model showed lowest ESS value followed by log linear model. Since, the model that shows highest R2 value with low ESS value is generally considered as the best fit model. Among the five models tested, logistic model, log linear model and linear models were found to be the best models for overall body weight, total length and head length respectively. For claw length and last segment length, log linear model was found to be the best model. These models can be used to predict growth rates in M. rosenbergii. However, further studies need to be conducted with more growth traits taken into consideration

Catch rates and demographics of loggerhead sea turtles (Caretta caretta) captured from the Charleston, South Carolina, shipping channel during the period of mandatory use of turtle excluder devices (TEDs)

Trawling was conducted in the Charleston, South Carolina, shipping channel between May and August during 2004–07 to evaluate loggerhead sea turtle (Caretta caretta) catch rates and demographic distributions. Two hundred and twenty individual loggerheads were captured in 432 trawling events during eight sampling periods lasting 2–10 days each. Catch was analyzed by using a generalized linear model. Data were fitted to a negative binomial distribution with the log of standardized sampling effort (i.e., an hour of sampling with a net head rope length standardized to 30.5 m) for each event treated as an offset term. Among 21 variables, factors, and interactions, five terms were significant in the final model, which accounted for 45% of model deviance. Highly significant differences in catch were noted among sampling periods and sampling locations within the channel, with greatest catch furthest seaward consistent with historical observations. Loggerhead sea turtle catch rates in 2004–07 were greater than in 1991–92 when mandatory use of turtle excluder devices was beginning to be phased in. Concurrent with increased catch rates, loggerheads captured in 2004–07 were larger than in 1991–92. Eighty-five percent of loggerheads captured were ≤75.0 cm straight-line carapace length (nuchal notch to tip of carapace) and there was a 3.9:1 female-to-male bias, consistent with limited data for this location two decades earlier. Only juvenile loggerheads ≤75.0 cm possessed haplotypes other than CC-A01 or CC-A02 that dominate in the region. Six rare and one un-described haplotype were predominantly found in June 2004.

Assessment of habitat quality for juvenile California halibut (Paralichthys californicus) in a seasonally arid estuary

We evaluated habitat quality for juvenile California halibut (Paralichthys californicus) in a Pacific Coast estuary lacking in strong salinity gradients by examining density, recent otolith growth rates, and gut fullness levels of wild-caught and caged juveniles for one year. Juveniles <200 mm standard length were caught consistently in the inner, central, and outer sections of the estuary. The density of juveniles was two times higher in the inner estuary during most of the year, consistent with active habitat selection by part of the population. A generalized linear model indicated temperature, sampling time, and the interaction between salinity and temperature were significantly related to density. However, the model explained only 21% of the variance. Gut fullness levels of wild-caught juveniles were highest during the summer, but recent otolith growth rates were not related to temperature. The proportion of individuals feeding successfully indicated that seasonal differences in food availability are more important than spatial variation in prey abundance in driving feeding success. Feeding success of caged fishes was limited, precluding the use of growth rates as indicators of local habitat quality. However, marginal increment widths were reliable indicators of somatic growth at low growth rates over two-week periods. The relatively high growth rates and abundance of small wild-caught juveniles found throughout the estuary indicates that the entire estuary system has the potential for serving as nursery habitat.

Ungewohnliche Gewichsabnahme von Dorsch im 2. Jahresquartal in Bornholm Becken (ICES-Gebeit 25) und der Arkonasee (ICES-Gebiet 24)

Cod captured by commercial fi shery in the Bornholm Basin in quarter 2 of 2001 were not suitable for the mechanical processing due to low condition and weight. The absolute mean weight of cod captured by the commercial fishery in the Arkona Sea and Bornholm Basin in the same quarter during the last fi ve years was studied to describe its development. The results of a GLM (Generalized Linear Model) analysis showed similar development of body weight in the Bornholm Basin and in the Arkona Sea between 2007 and 2011. The mean weight of cod in the Bornholm Basin increased from 2007 to 2008 in both areas followed by a relative stable weight until 2009 and a decrease until 2011. In the Arkona Sea the mean weight of cod 2009 has decreased in comparison to 2008, then have increased 2010 slightly and last have decreased in 2011. The analyses showed that the weight of cod is signifi cantly infl uenced by length, age and maturity of individuals.

Growth dynamics of the spinner shark (Carcharhinus brevipinna) off the United States southeast and Gulf of Mexico coasts: a comparison of methods

The age and growth dynamics of the spinner shark (Carcharhinus brevipinna) in the northwest Atlantic Ocean off the southeast United States and in the Gulf of Mexico were examined and four growth models were used to examine variation in the ability to fit size-at-age data. The von Bertalanffy growth model, an alternative equation of the von Bertalanffy growth model with a size-at-birth intercept, the Gompertz growth model, and a logistic model were fitted to sex-specific observed size-at-age data. Considering the statistical criteria (e.g., lowest mean square error [MSE], high coefficient-of-determination, and greatest level of significance) we desired for this study, the logistic model provided the best overall fit to the size-at-age data, whereas the von Bertalanffy growth model gave the worst. For “biological validity,” the von Bertalanffy model for female sharks provided estimates similar to those reported in other studies. However, the von Bertalanffy model was deemed inappropriate for describing the growth of male spinner sharks because estimates of theoretical maximum size (L∞) indicated a size much larger than that observed in the field. However, the growth coefficient (k= 0.14/yr) from the Gompertz model provided an estimate most similar to that reported for other large coastal species. The analysis of growth for spinner shark in the present study demonstrates the importance of fitting alternative models when standard models fit the data poorly or when growth estimates do not appear to be realistic.

Escapement of the Cape rock lobster (Jasus lalandii ) through the mesh and entrance of commercial traps

Metal-framed traps covered with polyethylene mesh used in the fishery for the South African Cape rock lobster (Jasus lalandii) incidentally capture large numbers of undersize (<75 mm CL) specimens. Air-exposure, handling, and release procedures affect captured rock lobsters and reduce the productivity of the stock, which is heavily fished. Optimally, traps should retain legalsize rock lobsters and allow sublegal animals to escape before traps are hauled. Escapement, based on lobster morphometric measurements, through meshes of 62 mm, 75 mm, and 100 mm was investigated theoretically under controlled conditions in an aquarium, and during field trials. SELECT models were used to model escapement, wherever appropriate. Size-selectivity curves based on the logistic model fitted the aquarium and field data better than asymmetrical Richards curves. The lobster length at 50% retention (L50) on the escapement curve for 100-mm mesh in the aquarium (75.5 mm CL) approximated the minimum legal size (75 mm CL); however estimates of L50 increased to 77.4 mm in field trials where trapentrances were sealed, and to 82.2 mm where trap-entrances were open. Therfore, rock lobsters that cannot escape through the mesh of sealed field traps do so through the trap entrance of open traps. By contrast, the wider selection range and lower L25 of field, compared to aquarium, trials (SR = 8.2 mm vs. 2.6 mm; L25 =73.4 mm vs. 74.1 mm), indicate that small lobsters that should be able to escape from 100-mm mesh traps do not always do so. Escapement from 62-mm mesh traps with open entrance funnels increased by 40−60% over sealed traps. The findings of this study with a known size distribution, are related to those of a recent indirect (comparative) study for the same species, and implications for trap surveys, commercial catch rates, and ghost fishing are discussed.

Temporal trends (2000–2011) and influences on fishery-independent catch rates for loggerhead sea turtles (Caretta caretta) at an important coastal foraging region in the southeastern United States

Seasonal trawling was conducted randomly in coastal (depths of 4.6–17 m) waters from St. Augustine, Florida, (29.9°N) to Winyah Bay, South Carolina (33.1°N), during 2000–03, 2008–09, and 2011 to assess annual trends in the relative abundance of sea turtles. A total of 1262 loggerhead sea turtles (Caretta caretta) were captured in 23% (951) of 4207 sampling events. Capture rates (overall and among prevalent 5-cm size classes) were analyzed through the use of a generalized linear model with log link function for the 4097 events that had complete observations for all 25 model parameters. Final models explained 6.6% (70.1–75.0 cm minimum straight-line carapace length [SCLmin]) to 14.9% (75.1–80.0 cm SCLmin) of deviance in the data set. Sampling year, geographic subregion, and distance from shore were retained as significant terms in all final models, and these terms collectively accounted for 6.2% of overall model deviance (range: 4.5–11.7% of variance among 5-cm size classes). We retained 18 parameters only in a subset of final models: 4 as exclusively significant terms, 5 as a mixture of significant or nonsignificant terms, and 9 as exclusively nonsignificant terms. Four parameters also were dropped completely from all final models. The generalized linear model proved appropriate for monitoring trends for this data set that was laden with zero values for catches and was compiled for a globally protected species. Because we could not account for much model deviance, metrics other than those examined in our study may better explain catch variability and, once elucidated, their inclusion in the generalized linear model should improve model fits.

Distribution of ommastrephid paralarvae in the eastern tropical Pacific

Jumbo squid (Dosidicus gigas) and purpleback squid (Sthenoteuthis oualaniensis) (Teuthida: Ommastrephidae) are thought to spawn in the eastern tropical Pacific. We used 10 years of plankton tow and oceanographic data collected in this region to examine the reproductive habits of these 2 ecologically important squid. Paralarvae of jumbo squid and purpleback squid were found in 781 of 1438 plankton samples from surface and oblique tows conducted by the Southwest Fisheries Science Center (NOAA) in the eastern tropical Pacific over the 8-year period of 1998–2006. Paralarvae were far more abundant in surface tows (maximum: 1588 individuals) than in oblique tows (maximum: 64 individuals). A generalized linear model analysis revealed sea-surface temperature as the strongest environmental predictor of paralarval presence in both surface and oblique tows; the likelihood of paralarval presence increases with increasing temperature. We used molecular techniques to identify paralarvae from 37 oblique tows to species level and found that the purpleback squid was more abundant than the jumbo squid (81 versus 16 individuals).

Effects of El Niño–Southern Oscillation events on catches of Bigeye Tuna (Thunnus obesus) in the eastern Indian Ocean off Java

The effects of El Niño–Southern Oscillation events on catches of Bigeye Tuna (Thunnus obesus) in the eastern Indian Ocean (EIO) off Java were evaluated through the use of remotely sensed environmental data (sea-surface-height anomaly [SSHA], sea-surface temperature [SST], and chlorophyll a concentration), and Bigeye Tuna catch data. Analyses were conducted for the period of 1997–2000, which included the 1997–98 El Niño and 1999–2000 La Niña events. The empirical orthogonal function (EOF) was applied to examine oceanographic parameters quantitatively. The relationship of those parameters to variations in catch distribution of Bigeye Tuna was explored with a generalized additive model (GAM). The mean hook rate was 0.67 during El Niño and 0.44 during La Niña, and catches were high where SSHA ranged from –21 to 5 cm, SST ranged from 24°C to 27.5°C, and chlorophyll-a concentrations ranged from 0.04 to 0.16 mg m–3. The EOF analysis confirmed that the 1997–98 El Niño affected oceanographic conditions in the EIO off Java. The GAM results indicated that SST was better than the other environmental factors (SSHA and chlorophyll-a concentration) as an oceanographic predictor of Bigeye Tuna catches in the region. According to the GAM predictions, the highest probabilities (70–80%) for Bigeye Tuna catch in 1997–2000 occurred during oceanographic conditions during the 1997–98 El Niño event.

Influence of soak time and fish accumulation on catches of reef fishes in a multispecies trap survey

Catch rates from fishery-independent surveys often are assumed to vary in proportion to the actual abundance of a population, but this approach assumes that the catchability coefficient (q) is constant. When fish accumulate in a gear, the rate at which the gear catches fish can decline, and, as a result, catch asymptotes and q declines with longer fishing times. We used data from long-term trap surveys (1990–2011) in the southeastern U.S. Atlantic to determine whether traps saturated for 8 reef fish species because of the amount of time traps soaked or the level of fish accumulation (the total number of individuals of all fish species caught in a trap). We used a delta-generalized-additive model to relate the catch of each species to a variety of predictor variables to determine how catch was influenced by soak time and fish accumulation after accounting for variability in catch due to the other predictor variables in the model. We found evidence of trap saturation for all 8 reef fish species examined. Traps became saturated for most species across the range of soak times examined, but trap saturation occurred for 3 fish species because of fish accumulation levels in the trap. Our results indicate that, to infer relative abundance levels from catch data, future studies should standardize catch or catch rates with nonlinear regression models that incorporate soak time, fish accumulation, and any other predictor variable that may ultimately influence catch. Determination of the exact mechanisms that cause trap saturation is a critical need for accurate stock assessment, and our results indicate that these mechanisms may vary considerably among species.

Magnitude and distribution of sea turtle bycatch in the sea scallop (Placopecten magellanicus) dredge fishery in two areas of the northwestern Atlantic Ocean, 2001–2002

In May 2001, the National Marine Fisheries Service (NMFS) opened two areas in the northwestern Atlantic Ocean that had been previously closed to the U.S. sea scallop (Placopecten magellanicus) dredge fishery. Upon reopening these areas, termed the “Hudson Canyon Controlled Access Area” and the “Virginia Beach Controlled Access Area,” NMFS observers found that marine turtles were being caught incidentally in scallop dredges. This study uses the generalized linear model and the generalized additive model fitting techniques to identify environmental factors and gear characteristics that influence bycatch rates, and to predict total bycatch in these two areas during May-December 2001 and 2002 by incorporating environmental factors into the models. Significant factors affecting sea turtle bycatch were season, time-of-day, sea surface temperature, and depth zone. In estimating total bycatch, rates were stratified according to a combination of all these factors except time-of-day which was not available in fishing logbooks. Highest bycatch rates occurred during the summer season, in temperatures greater than 19°C, and in water depths from 49 to 57 m. Total estimated bycatch of sea turtles during May–December in 2001 and 2002 in both areas combined was 169 animals (CV=55.3), of which 164 (97%) animals were caught in the Hudson Canyon area. From these findings, it may be possible to predict hot spots for sea turtle bycatch in future years in the controlled access areas.

N-best error simulation for training spoken dialogue systems

A recent trend in spoken dialogue research is the use of reinforcement learning to train dialogue systems in a simulated environment. Past researchers have shown that the types of errors that are simulated can have a significant effect on simulated dialogue performance. Since modern systems typically receive an N-best list of possible user utterances, it is important to be able to simulate a full N-best list of hypotheses. This paper presents a new method for simulating such errors based on logistic regression, as well as a new method for simulating the structure of N-best lists of semantics and their probabilities, based on the Dirichlet distribution. Off-line evaluations show that the new Dirichlet model results in a much closer match to the receiver operating characteristics (ROC) of the live data. Experiments also show that the logistic model gives confusions that are closer to the type of confusions observed in live situations. The hope is that these new error models will be able to improve the resulting performance of trained dialogue systems. © 2012 IEEE.