Ricker Stock Recruitment Model Research Articles

Stochastic modelling and synthesis of dynamic fish recruitment productivity in the Celtic Seas ecoregion

Abstract The Celtic Seas ecoregion (CSE) is undergoing climatic and ecosystem changes, which can induce changes in fish productivity. Globally, the productivity of many stocks has shown evidence of change over decadal timescales. Varying factors might drive these dynamics in the CSE, but for many stocks, these mechanisms have not been fully understood to be included in management advice. We study dynamic productivity for 28 stocks in the Celtic Seas by tracking integrated stochastic signals in the relationship between stock size and recruitment using state-space modelling applying Peterman’s Productivity Method. Our research objectives were to (i) fit Ricker stock–recruitment models with time-varying parameters to all age- or length-based assessed stocks in the CSE, (ii) evaluate which parameters vary in time, (iii) examine temporal characteristics of historical recruitment productivity, and (iv) evaluate productivity correlation across stocks. For 22 out of 28 stocks, at least one of the three time-varying parameter models had a better fit than the time-invariant model. In the CSE, fish productivity has diverse temporal patterns, with some stocks displaying relevant long-term decreasing productivity trends. Getting insight into temporal changes in recruitment productivity is very valuable and has important implications for sustainable fisheries.

Assessment and management of the temperate stock of Pacific sardine (Sardinops sagax) in the south of California Current System

Knowing the population dynamics and the variations in the abundance of the Pacific sardine is essential for suitable fishing management. Due to the possibility that the different stocks of Sardinops sagax that can inhabit the south of California Current System (CCS) were present in the catches, the caught of the temperate stock were discriminated from the total landed and, subsequently was assessed the dynamics of temperate stock of Pacific sardine (TSPS) for the period from 1989 to 2021. For the above, we use a statistical analysis of catch-at-age (ACE), which is an integrated analysis model that includes three indices of relative abundance (catch rate, acoustic surveys and evaluations of eggs and larvae). Acoustic indices and capture rate (CPUE) denoted changes in population abundance better than the index of eggs and larvae. The total biomass has shown great interannual variability oscillating between 853,476 and 1,592,519 t; a similar trend was shown by the spawning biomass, oscillating between 404,189 t and 770,484 t. With the modified Ricker Stock-Recruitment model, the minimum biomass Bmin = 50,000 t and the exploitation rate in Emrsy = 0.251 year−1 were estimated. Considering behavior of proportion between the recorded catch and the estimated biologically acceptable catch to each year (Cobs/BAC), we can infer that the TSPS has been sustainably exploited throughout analyzed period, except for 2014 and 2017 fishing seasons, when the BAC was exceeded by around 20%. Harvest control rule, by depending on a biomass fraction, estimated with an integrated model ACE, is considered a management suitable strategy to TSPS.

Incorporating spatial heterogeneity and environmental impacts into stock-recruitment relationships for Gulf of Maine lobster

ABSTRACT Functional stock-recruitment relationships (SRRs) are often difficult to quantify and can differ over space. Additionally, climate change adds to the complexity of recruitment dynamics. This paper's aim was to incorporate spatial heterogeneity and environmental effects on productivity in SRRs with American lobster in the Gulf of Maine (GOM) as a case study. GOM lobster recruitment has substantially increased since the mid-2000s, due to improved survival rates of pre-recruits and increased spawning stock biomass (SSB). GOM bottom water temperatures have increased at a rate of 0.2ºC per decade, which caused lobster settlement area to expand and improved survival rates. We first estimated local SSB using bottom trawl survey data and a geostatistical model. Using estimated SSB, recruitment data from a ventless trap survey, and an interpolated bottom water temperature field, we developed modified Ricker stock-recruitment models accounting for spatial heterogeneity and temperature impacts with varying coefficient generalized additive models. Results showed that temperature significantly impacted recruitment. Changes in temperature mediated productivity differed between the eastern and western GOM. Our study demonstrated that the incorporation of spatial heterogeneity and environmental effects impacts our understanding of SRRs. These methods can be applied to other species to understand recruitment dynamics influenced by climate change.

Effects of Adult Biomass and Environmental Conditions on Bigheaded Carp Reproductive Output

AbstractBighead Carp Hypothalmichthys nobilis and Silver Carp Hypothalmichthys moltrix (hereafter collectively referred to as Bigheaded Carp) have spread throughout the majority of the Mississippi River since the 1970s. The current northern invasion edge of Bigheaded Carp in the Upper Mississippi River (UMR) spans between Pools 14 and 20 because of limited passage at Lock and Dam (LD) 19. Mechanisms limiting adult Bigheaded Carp abundance above LD19 are unknown but may be due in part to lack of reproductive success influenced by adult abundance and environmental factors. Our objective was to investigate how relative adult biomass and river temperature and discharge affect maximum annual Bigheaded Carp larval production in the UMR using a Ricker stock-recruitment model. Adult Bigheaded Carp relative biomass (kg/h) was estimated annually with boat electrofishing and larvae were collected every 10 d between May and August 2014–2017 in Pools 14–20 in the UMR. Adult relative biomass ranged from 0.0 to 880.9 kg/h, whereas maximum annual larval densities ranged from 0.0 to 2,869.4 larvae/m3. After accounting for variability among pools and years, the most supported linear Ricker stock-recruitment model indicated the number of recruits per spawner decreased with increasing adult relative biomass and increased with mean discharge. Our results highlight the importance of adult biomass and river discharge conditions for reproduction of Bigheaded Carp along leading edges of invasion. Management strategies that aim to maintain low adult abundance where reproduction is not yet occurring could help limit population increases via reproduction, whereas reducing high adult biomass (e.g., commercial harvest, barriers) may result in greater Bigheaded Carp reproductive output in the UMR.

Covariability of Fraser River sockeye salmon productivity and phytoplankton biomass in the Gulf of Alaska

AbstractUsing satellite ocean color measurements and sockeye salmon stock–recruit data from 1995 to 2016, our analysis reveals a significant, positive relationship between Fraser River sockeye productivity and summer chlorophyll a concentrations (Chl‐a) in the northwestern Gulf of Alaska. The correlation is the strongest for the area adjacent to the continental shelf near Kodiak Island (northern Gulf of Alaska), during midsummer of the juvenile sockeye salmon ocean entry year. The inclusion of (Chl‐a) data from this region significantly improved the performance of Ricker stock–recruitment models for ten out of eighteen Fraser River stocks. We discuss the potential oceanographic processes driving the correlation between (Chl‐a) and sockeye productivity. During midsummer, positive sea‐level anomalies in the offshore waters near Kodiak Island suggest the presence of mesoscale eddies, which can transport nutrients to this region and concentrate phytoplankton and zooplankton biomass, thus supporting enhanced sockeye salmon prey abundance. These results suggest a mechanism linking mesoscale oceanographic dynamics to sockeye salmon productivity, and identify a potentially important region for future studies of Fraser River sockeye salmon survival.

The effect of harvesting on the spatial synchrony of population fluctuations

Harvesting in space affects, in general, the spatial scale of the synchrony in the population fluctuations, which determines the size of the areas subjected to simultaneous quasi-extinction risk. Here we show that harvesting reduces the population synchrony scale if it depends more strongly on population fluctuations than the density dependence of the growth rate in the absence of harvesting. We show that constant and proportional harvesting always increases the spatial scale, using a theta-logistic model for density regulation. We also provide exact scaling results under harvesting for the Beverton–Holt and the Ricker stock-recruitment models that are commonly applied, e.g. in fisheries. Our results indicate that harvest in areas with large abundances should be encouraged to avoid increase of the spatial scale of synchrony in the population fluctuations that can lead to unexpected quasi-extinction of populations over large areas. Our results quantify this harvesting impact giving the resulting scales of spatial synchrony of population fluctuations. This emphasizes the importance of estimating the form of density dependence as well as the dependency of harvest upon population density of exploited populations, in order to get reliable predictions of the size of areas that can undergo simultaneous quasi-extinction.

Gaussian process framework for temporal dependence and discrepancy functions in Ricker-type population growth models

Density dependent population growth functions are of central importance to population dynamics modelling because they describe the theoretical rate of recruitment of new individuals to a natural population. Traditionally, these functions are described with a fixed functional form with temporally constant parameters and without species interactions. The Ricker stock-recruitment model is one such function that is commonly used in fisheries stock assessment. In recent years, there has been increasing interest in semiparametric and temporally varying population growth models. The former are related to the general statistical approach of using semiparametric discrepancy functions, such as Gaussian processes (GP), to model deviations around the expected parametric function. In the latter, the reproductive rate, which is a key parameter describing the population growth rate, is assumed to vary in time. In this work, we introduce how these existing Ricker population growth models can be formulated under the same statistical approach of hierarchical GP models. We also show how the time invariant semiparametric approach can be extended and combined with the time varying reproductive rate using a GP model. Then we extend these models to the multispecies setting by incorporating cross-covariances among species with a continuous time covariance structure using the linear model of coregionalization. As a case study, we examine the productivity of three Pacific salmon populations. We compare the alternative Ricker population growth functions using model posterior probabilities and leave-one-out cross validation predictive densities. Our results show substantial temporal variation in maximum reproductive rates and reveal temporal dependence among the species, which have direct management implications. However, our results do not support inclusion of semiparametric discrepancy function and they suggest that the semiparametric discrepancy functions may lead to challenges in parameter identifiability more generally.

Empirically based models of oceanographic and biological influences on Pacific Herring recruitment in Prince William Sound

Abundances of small pelagic fish can change dramatically over time and are difficult to forecast, partially due to variable numbers of fish that annually mature and recruit to the spawning population. Recruitment strength of age-3 Pacific Herring (Clupea pallasii) in Prince William Sound, Alaska, is estimated in an age-structured model framework as a function of spawning stock biomass via a Ricker stock-recruitment model, and forecasted using the 10-year median recruitment estimates. However, stock size has little influence on subsequent numbers of recruits. This study evaluated the usefulness of herring recruitment models that incorporate oceanographic and biological variables. Results indicated herring recruitment estimates were significantly improved by modifying the standard Ricker model to include an index of young-of-the-year (YOY) Walleye Pollock (Gadus chalcogrammus) abundance. The positive relationship between herring recruits-per-spawner and YOY pollock abundance has persisted through three decades, including the herring stock crash of the early 1990s. Including sea surface temperature, primary productivity, and additional predator or competitor abundances singly or in combination did not improve model performance. We suggest that synchrony of juvenile herring and pollock survival may be caused by increased abundance of their zooplankton prey, or high juvenile pollock abundance may promote prey switching and satiation of predators. Regardless of the mechanism, the relationship has practical application to herring recruitment forecasting, and serves as an example of incorporating ecosystem components into a stock assessment model.

Application of a Delay-difference model for the stock assessment of southern Atlantic albacore (Thunnus alalunga)

Delay-difference models are intermediate between simple surplus-production models and complicated age-structured models. Such intermediate models are more efficient and require less data than age-structured models. In this study, a delay-difference model was applied to fit catch and catch per unit effort (CPUE) data (1975–2011) of the southern Atlantic albacore (Thunnus alalunga) stock. The proposed delay-difference model captures annual fluctuations in predicted CPUE data better than Fox model. In a Monte Carlo simulation, white noises (CVs) were superimposed on the observed CPUE data at four levels. Relative estimate error was then calculated to compare the estimated results with the true values of parameters α and β in Ricker stock-recruitment model and the catchability coefficient q. a is more sensitive to CV than β and q. We also calculated an 80% percentile confidence interval of the maximum sustainable yield (MSY, 21756 t to 23408 t; median 22490 t) with the delay-difference model. The yield of the southern Atlantic albacore stock in 2011 was 24122 t, and the estimated ratios of catch against MSY for the past seven years were approximately 1.0. We suggest that care should be taken to protect the albacore fishery in the southern Atlantic Ocean. The proposed delay-difference model provides a good fit to the data of southern Atlantic albacore stock and may be a useful choice for the assessment of regional albacore stock.

Biotic and Abiotic Factors Influencing Cisco Recruitment Dynamics in Lake Superior during 1978–2007

AbstractHistorically, the Cisco Coregonus artedi was abundant throughout the Great Lakes basin, but anthropogenic influences caused the collapse of stocks during the mid‐1900s, and fishery managers are currently exploring options for restoration. To increase understanding of biotic and abiotic factors influencing Cisco recruitment dynamics throughout the Great Lakes, we used Ricker stock–recruitment models with rearing‐habitat‐weighted indices of recruitment and adult spawning stock size in Lake Superior to identify and quantify (1) the appropriate spatial scale for modeling age‐1 recruitment dynamics and (2) the effects of biotic and abiotic factors on age‐1 recruitment dynamics within the regions identified for modeling. Cisco recruitment variation in Lake Superior was best described by a regional model with separate stock–recruitment relationships for western, southern, eastern, and northern regions. The spatial scale for modeling was approximately 260 km. Age‐1 recruitment was negatively correlated with adult spawning stock size in all four regions. Multifactor models suggested (1) a positive correlation between age‐1 recruitment and the interaction between wind speed and air temperature on a lakewide scale and (2) a negative correlation between age‐1 recruitment and the biomass of Rainbow Smelt Osmerus mordax on a regional scale. Large‐scale abiotic factors are beyond the control of fishery managers, so harvest of adult Ciscoes and potential predators and competitors should be carefully managed to achieve desired goals in Lake Superior. Although our study was limited to Cisco stocks in Lake Superior, we believe that our general findings can be more broadly applied (albeit with caution) to the restoration and management of remnant stocks throughout the Great Lakes basin.Received June 4, 2013; accepted August 16, 2013

Density‐Dependence and Environmental Conditions Regulate Recruitment and First‐Year Growth of Common Carp in Shallow Lakes

AbstractRecruitment dynamics of fishes can determine their abundance and subsequent effects of adult populations on ecosystem properties. Common Carp Cyprinus carpio is a widespread invasive species that can exert negative density‐dependent effects on aquatic food webs. However, little is known concerning processes that regulate their recruitment. We used summer‐through‐fall electrofishing catch rates of age‐0 Common Carp to investigate the importance of biotic (stock size, prey availability, egg and juvenile predation, intraspecific competition) and abiotic (water level fluctuation, temperature, wind events) factors on recruitment and growth in 13 lakes across eastern South Dakota from 2008 to 2010. Mean relative abundance of age‐0 Common Carp was highly variable spatially and temporally, ranging from 0 to 152.2 fish per hour of electrofishing. Ricker stock–recruitment models indicated that peak production of age‐0 Common Carp occurred when adult Common Carp relative abundance was low. Recruitment also increased with spring–summer temperatures, decreased with wind events, and was affected by annual water level fluctuations. Age‐0 Common Carp growth, estimated by mean size in August, was highly variable (mean TL = 42–165 mm) and influenced by the abundance of conspecifics and wind events, indicating that density‐dependent interactions may have occurred during early life stages. Combined, our results suggest that the early life history of Common Carp is regulated by density‐dependent processes and abiotic environmental conditions, which provide new insights into mechanisms regulating recruitment of this widespread invasive species.

Life Form and Life History Explain Variation in Population Processes in a Grassland Community Invaded by Exotic Plants and Mammals

The existence of general characteristics of plant invasiveness is still debated. One reason we may not have found these characteristics is because we do not yet understand how processes underlying population dynamics contribute to community composition in invaded communities. Here I modify Ricker stock-recruitment models to parameterize processes important to community dynamics in an invaded grassland community: immigration, maximum intrinsic growth rate, self-regulation, and limitation by other species. I then used the parameterized models in a multi-species stochastic simulation to determine how processes affected long-term community dynamics. By parameterizing the models using the frequency of the 18 most common species in the grassland, I determined that life history and life form are stronger predictors of underlying processes than is native status. Immigration maintains exotic annual grasses and the dominant native perennial grass in the community. Growth rate maintains other perennial species. While the model mirrors the frequency of native species well, exotic species have lower observed than parameterized frequencies, suggesting that they are not reaching their potential frequency. These results, combined with results from past research, suggest that disturbance may be key to maintaining exotic species in the community. Here I showed that a continuous modified Ricker model fit discrete grassland frequency data well. This allowed me to model the dominant species in the community simultaneously and gain insight into the processes that determine community composition.

The Spatial Scale for Cisco Recruitment Dynamics in Lake Superior during 1978–2007

AbstractThe cisco Coregonus artedi was once the most abundant fish species in the Great Lakes, but currently cisco populations are greatly reduced and management agencies are attempting to restore the species throughout the basin. To increase understanding of the spatial scale at which density‐independent and density‐dependent factors influence cisco recruitment dynamics in the Great Lakes, we used a Ricker stock–recruitment model to identify and quantify the appropriate spatial scale for modeling age‐1 cisco recruitment dynamics in Lake Superior. We found that the recruitment variation of ciscoes in Lake Superior was best described by a five‐parameter regional model with separate stock–recruitment relationships for the western, southern, eastern, and northern regions. The spatial scale for modeling was about 260 km (range = 230–290 km). We also found that the density‐independent recruitment rate and the rate of compensatory density dependence varied among regions at different rates. The density‐independent recruitment rate was constant among regions (3.6 age‐1 recruits/spawner), whereas the rate of compensatory density dependence varied 16‐fold among regions (range = −0.2 to −2.9/spawner). Finally, we found that peak recruitment and the spawning stock size that produced peak recruitment varied among regions. Both peak recruitment (0.5–7.1 age‐1 recruits/ha) and the spawning stock size that produced peak recruitment (0.3–5.3 spawners/ha) varied 16‐fold among regions. Our findings support the hypothesis that the factors driving cisco recruitment operate within four different regions of Lake Superior, suggest that large‐scale abiotic factors are more important than small‐scale biotic factors in influencing cisco recruitment, and suggest that fishery managers throughout Lake Superior and the entire Great Lakes basin should address cisco restoration and management efforts on a regional scale in each lake.Received March 5, 2011; accepted January 5, 2012

A recruitment-mortality model in the precautionary management toolkit of African tropical inland, single-species fisheries

A recruitment-mortality model and a Beverton–Holt or Ricker stock-recruitment model are combined to derive stock performance metrics. A composite (age-aggregated) yield-per-recruit (CYPR) emerges from these derivations. Upon a balanced equilibrium (e.g., recruitment>yield), the CYPR maximizes at an asymptote (Asy) beyond which CYPR cannot drop whatever the age-at-harvest. F0.5Asy, the fishing mortality corresponding to 0.5Asy, may form the basis of a conservative feedback-policy rule. The approach aims to complement the management tools for inland single-species fisheries in tropical Africa, but can be applied for any single-species fishery. It is illustrated using data for the Zambian Bukabuka (Lates stappersii) fishery of Lake Tanganyika.

Old Tales in Recent Context: Current Perspective on Yellow Perch Recruitment in Lake Michigan

AbstractTo evaluate factors affecting recruitment of yellow perch Perca flavescens in southwestern Lake Michigan, a Ricker stock–recruitment model was fitted to age‐0 abundance estimated from a fall bottom trawl survey (1989–2007) and spawning stock biomass estimates. Models were built for all possible combinations of four external variables: spring–summer water temperature, June nearshore zooplankton density, spring abundance of alewives Alosa pseudoharengus, and an index of wind‐induced transport of yellow perch larvae. Four relatively parsimonious models were deemed robust, and three external variables were represented (spring–summer water temperature, spring alewife abundance, and June zooplankton density). The Ricker stock–recruitment model with the lowest value of Akaike's information criterion (i.e., the best‐supported model) included spring–summer water temperature and spring alewife abundance. The basic Ricker stock–recruitment model ranked second, followed by the Ricker model that included June zooplankton density. The fourth Ricker model that was deemed robust included spring–summer water temperature. The four best models provided a good fit to age‐0 yellow perch abundance estimates based on fall seining catches along the Illinois shoreline of Lake Michigan during 1989–2007. Our findings indicate that over the last two decades, yellow perch recruitment benefited during years associated with relatively warm water, low alewife abundance, and high zooplankton density.

Factors Affecting Recruitment of Age‐0 Muskellunge in Escanaba Lake, Wisconsin, 1987–2006

AbstractWe modeled variation in recruitment (R) of age‐0 muskellunge Esox masquinongy to identify factors influencing their abundance in Escanaba Lake, Wisconsin. Muskellunge R declined over the study period and ranged from 0.00 to 1.85 age‐0 fish/km of shoreline (mean = 0.42 age‐0 fish/km of shoreline). A Ricker stock–recruitment model determined that the following factors explained 88% of the variation in annual R of age‐0 muskellunge between 1987 and 2006: abundance and age structure of the adult muskellunge population, abundance of bluntnose minnow Pimephales notatus, abundance of age‐3 and older (age‐3+) walleyes Sander vitreus, abundance of age‐0 white suckers Catostomus commersonii, and coefficient of variation (CV) of May water temperatures. Abundance of adult muskellunge (≥76.2 cm total length) accounted for only 1% of the variation in R and showed no significant relationship with R. Abundance of bluntnose minnow improved the model fit to 40% of the variation in R and indicated that higher R was achieved with greater numbers of bluntnose minnow. The average age of adult muskellunge further improved the model fit to 59% of the variation in R, suggesting that R increased when more young adults were present in the population. The abundance of age‐3+ walleyes enhanced the model fit to 69% of the variation in R and indicated that greater R occurred with high numbers of walleyes. The abundance of age‐0 white suckers improved the model to explain 77% of the variation in R and indicated that more recruits occurred when numbers of age‐0 white suckers were low. Finally, the CV of May water temperatures further improved the model to explain 88% of the variation in R and signified that lower variability in May water temperatures was beneficial to recruitment success. We interpret the model results to mean that muskellunge R in Escanaba Lake is regulated by the reproductive potential of the adult muskellunge population, forage availability, variation in May water temperatures, and other community dynamics.

Factors affecting recruitment dynamics of Great Lakes sea lamprey ( Petromyzon marinus) populations

Knowledge of stock–recruitment dynamics is as important for control of pest species such as the sea lamprey ( Petromyzon marinus) as it is for sustainable harvest management of exploited fish species. A better understanding of spatial and temporal variation in recruitment of pest populations may inform managers on where and when to effectively apply different control methods. Sea lamprey stock–recruitment data combined from streams across the Great Lakes basin into a Ricker stock–recruitment model indicated both compensation (density-dependent survival) and a large amount of density-independent recruitment variation. We evaluated the use of a mixed-effects model to look at common year effects and stream-level variables that could affect productivity and growing season length, with the objective of identifying factors that may explain this density-independent variation in recruitment. After selecting the “best model”, we tested factors that might affect recruitment variation, using a Great Lakes dataset comprising 97 stream–years. Lake Superior tributaries, streams with larger numbers of lamprey competitors, and streams regularly requiring lampricide treatment showed significantly higher recruitment. Alkalinity and thermal regulation did not affect the observed recruitment pattern among streams. In four long-term study streams we observed significant variation among streams, tested as a fixed effect, but no evidence suggested a common pattern of variation among years. Differences in recruitment among streams were consistent with evidence of quality of spawning and larval habitat among streams. Our findings suggest that management models should account for differences in recruitment dynamics among sea lamprey-producing streams, but not common year effects.

Recruitment variability of resident brown trout in peripheral populations from southern Europe

Summary1. Population regulation was studied for seven consecutive years (1992–98) in five rivers at the periphery of the distribution ofSalmo trutta, where the fish were living under environmental constraints quite different from those of the main distribution area.2. Recruitment is naturally highly variable and the populations had been earlier classified as overexploited. Thus we expected that densities of young trout in most populations would be too low for density‐dependent mortality to operate. We tested this by fitting the abundance of recruits to egg densities over seven consecutive years (stock–recruitment relationship), and used the results to judge whether exploitation should be restricted in the interests of conserving the populations.3. The density of 0+ trout in early September, as well as the initial density of eggs and parents, varied greatly among localities and years. The data for all populations fitted the Ricker stock–recruitment model. The proportion of variance explained by the population curves varied between 32% and 51%. However, in most cases the observations were in the density‐independent part of the stock–recruitment curve, where densities of the recruits increased proportionally with egg densities.4. Our findings suggest that recruitment densities in most rivers and years were below the carrying capacity of the habitats. Although density‐dependent mechanisms seemed to regulate fish abundance in some cases, environmental factors and harvesting appeared generally to preclude populations from reaching densities high enough for negative feedbacks to operate. The findings thus lend support toHaldane’s (1956)second hypothesis that changes in population density are primarily due to density‐independent factors in unfavourable areas and areas with low density due to exploitation. Exploitation should be reduced to allow natural selection to operate more effectively.

Environmental and Ecological Conditions Surrounding the Production of Large Year Classes of Walleye ( Sander vitreus) in Saginaw Bay, Lake Huron

The Saginaw Bay walleye population ( Sander vitreus) has not fully recovered from a collapse that began in the 1940s and has been dependent on stocking with only limited natural reproduction. Beginning in 2003, and through at least 2005, reproductive success of walleye surged to unprecedented levels. The increase was concurrent with ecological changes in Lake Huron and we sought to quantitatively model which factors most influenced this new dynamic. We developed Ricker stock-recruitment models for both wild and stock fish and evaluated them with second-order Akaike's information criterion to find the best model. Independent variables included adult alewife ( Alosa pseudoharengus) abundance, spring water temperatures, chlorophyll a levels and total phosphorus levels. In all, 14 models were evaluated for production of wild age-0 walleyes and eight models for stocked age-0 walleyes. For wild walleyes, adult alewife abundance was the dominant factor, accounting for 58% of the variability in age-0 abundance. Production of wild age-0 fish increased when adult alewives were scarce. The only other plausible factor was spring water temperature. Predictably, alewife abundance was not important to stocked fish; instead temperature and adult walleye abundance were more significant variables. The surge in reproductive success for walleyes during 2003–2005 was most likely due to large declines in adult alewives in Lake Huron. While relatively strong year classes (age-1 and up) have been produced as a result of increased age-0 production during 2003–2005, the overall magnitude has not been as great as the initial age-0 abundance originally suggested. It appears that over-winter mortality is higher than in the past and may stem from higher predation or slower growth (lower condition for enduring winter thermal stress). From this it appears that low alewife abundance does not assure strong walleye year classes in Saginaw Bay but may be a prerequisite for them.

Neural network and fuzzy logic models for pacific halibut recruitment analysis

Time series based neural network and fuzzy logic models were developed for fish recruitment analysis using fish stock–recruitment data incorporating environmental information. Data from the Pacific halibut stock with the Pacific Decadal Oscillation (PDO) index as the environmental variable was used for the model development. The neural network model was developed based on the multi-layer feed-forward neural network model with back-propagation learning algorithm. The fuzzy logic model was developed based on the fuzzification of the PDO index into two fuzzy sets denoted as “positive” and “negative” regimes. We compared the performance of these models to the traditional Ricker stock–recruitment model and the Ricker model with PDO as covariate. All four models showed significant residual time series structure in the form of first order autocorrelation. To account for this time series structure, an autoregressive component was added to each model. It was demonstrated that the fuzzy logic model outperformed the traditional Ricker stock–recruitment model, the Ricker model with PDO as covariate and the neural network model as measured by several diagnostic criteria. In addition, the simple Ricker stock–recruitment model corrected with the autoregressive time series residuals outperformed the Ricker model with PDO as covariate, neural network model and fuzzy logic model ignoring the time series residuals. The importance of the PDO to halibut recruitment is discussed in terms of developing a harvest policy.