abstract
We constructed ecosystem models using the Ecopath with Ecosim software to evaluate whether predation by killer whales might explain the decline of Steller sea lions since the late 1970s in the western Aleutian Islands. We also sought to understand why sea lions increased in the presence of killer whales in Southeast Alaska. Modeling results reproduced the time series of abundances for exploited species and sea lions in both ecosystems. Simulation results suggest that killer whale predation contributed to the decline of sea lions in the western Aleutians, but that predation was not the primary cause of the population decline. Predation could however have become a significant source of mortality during the 1990s when sea lions numbers were much lower. In Southeast Alaska, predation was also found to be a significant source of mortality in the 1960s when sea lions were low, but ceased to control population growth through the 1980s and 1990s. Overall, the ecosystem models suggest that large populations of Steller sea lions can withstand predation, but that small populations are vulnerable to killer whales.

abstract
The hypothesis that commercial whaling caused a sequential megafaunal collapse in the North Pacific Ocean by forcing killer whales to eat progressively smaller species of marine mammals is not supported by what is known about the biology of large whales, the ecology of killer whales and the patterns of ecosystem change that took place in Alaska, British Columbia, and elsewhere in the world following whaling. A comparative analysis shows that populations of seals, sea lions and sea otters increased in British Columbia following commercial whaling, unlike the declines noted in the Gulf of Alaska and Aleutian Islands. The declines of seals and sea lions that began in western Alaska around 1977 were mirrored by increases in numbers of these species in British Columbia. A more likely explanation is the seal and sea lion declines and other ecosystem changes in Alaska stems from a major oceanic regime shift that occurred in 1977. Killer whales are unquestionably a significant predator of seals, sea lions and sea otters but not because of commercial whaling.

abstract
Springer et al. [Springer, A.M., Estes, J.A., van Vliet, G.B., Williams, T.M., Doak, D.F., Danner, E.M., Forney, K.A., Pfister, B., 2003. Sequential megafaunal collapse in the North Pacific Ocean: an ongoing legacy of industrial whaling? Proceedings of the National Academy of Sciences 100 (21), 12,223–12,228] hypothesized that great whales were an important prey resource for killer whales, and that the removal of fin and sperm whales by commercial whaling in the region of the Bering Sea/Aleutian Islands (BSAI) in the late 1960s and 1970s led to cascading trophic interactions that caused the sequential decline of populations of harbor seal, northern fur seal, Steller sea lion and northern sea otter. This hypothesis, referred to as the Sequential Megafaunal Collapse (SMC), has stirred considerable interest because of its implication for ecosystem-based management. The SMC has the following assumptions: (1) fin whales and sperm whales were important as prey species in the Bering Sea; (2) the biomass of all large whale species (i.e., North Pacific right, fin, humpback, gray, sperm, minke and bowhead whales) was in decline in the Bering Sea in the 1960s and early 1970s; and (3) pinniped declines in the 1970s and 1980s were sequential. We concluded that the available data are not consistent with the first two assumptions of the SMC. Statistical tests of the timing of the declines do not support the assumption that pinniped declines were sequential. We propose two alternative hypotheses for the declines that are more consistent with the available data. While it is plausible, from energetic arguments, for predation by killer whales to have been an important factor in the declines of one or more of the three populations of pinnipeds and the sea otter population in the BSAI region over the last 30 years, we hypothesize that the declines in pinniped populations in the BSAI can best be understood by invoking a multiple factor hypothesis that includes both bottom–up forcing (as indicated by evidence of nutritional stress in the western Steller sea lion population) and top–down forcing (e.g., predation by killer whales, mortality incidental to commercial fishing, directed harvests). Our second hypothesis is a modification of the top–down forcing mechanism (i.e., killer whale predation on one or more of the pinniped populations and the sea otter population is mediated via the recovery of the eastern North Pacific population of the gray whale). We remain skeptical about the proposed link between commercial whaling on fin and sperm whales, which ended in the mid-1960s, and the observed decline of populations of northern fur seal, harbor seal, and Steller sea lion some 15 years later.

abstract
The timing and extent of the negative population trend in the abundance of the western stock of Steller sea lions has not been geographically uniform. A stochastic metapopulation dynamics model is developed for Steller sea lions. This model allows for geographical differences in factors affecting population processes, and can be parameterized to represent a wide range of hypotheses for the decline in Steller sea lion abundance. Bayesian and maximum likelihood methods are used to fit this model to pup and non-pup count data, age structure samples, and survival estimates. Inferences from model selection criteria highlight the spatial variability in the types of impact deemed to provide most parsimonious representation of the data. Bayesian posteriors for the estimated model parameters show that many combinations of parameter values are able to provide similar fits to the data, even given a specific hypothesis for the decline. This highlights the uncertainty in the precise nature of the impact of these hypotheses. Indeed, while pup production is generally estimated consistently among models, estimates of the size of other components of the Steller sea lion population (such as total population size) depend greatly on the assumptions regarding the cause of the decline. The results demonstrate that future simulation modeling approaches will require more formal, spatial, and mechanistic descriptions of the manner in which specific hypotheses for the decline affect the population.

abstract
Steller sea lions (Eumetopias jubatus) increased in the eastern portion of their range while declining in the Gulf of Alaska and Aleutian Islands from the late 1970s to late 1990s. We constructed ecosystem models of the central and western Aleutians and of Southeast Alaska to simultaneously evaluate four hypotheses explaining sea lion dynamics: killer whale (Orcinus orca) predation, ocean productivity, fisheries, and competition with other species. Comparisons of model predictions to historical time series data indicate that all four factors likely contributed to the trends observed in sea lion numbers in both ecosystems. Changes in ocean productivity conveyed by the Pacific Decadal Oscillation influenced the abundance trajectory of several species. Fishing could have affected the ecosystem structure by influencing the abundance of Atka mackerel (Pleurogrammus monopterygius) in the Aleutians, and herring (Clupea pallasii) in Southeast Alaska. Halibut (Hypoglossus stenolepis) in the Aleutians and arrowtooth flounder (Reinhardtius stomias) in Southeast Alaska appear to impede sea lion population growth through competitive interactions. Predation by killer whales was important when sea lions were less abundant in the 1990s in the Aleutians and in the 1960s in Southeast Alaska, but appear to have little effect when sea lion numbers were high.

abstract
A variety of reasons have been postulated for the decline of Steller sea lions (Otariidae: Eumetopias jubatus) in the Northeast Pacific. To date, however, it has proved impossible to distinguish among these reasons given the available data. In principle, experimental management based on spatial replication of treatments could be used to discriminate among some of these hypotheses. A simulation protocol was developed and applied to evaluate the power of a set of potential experiments to distinguish between whether the cause of the decline was fishing-induced or due to other factors. The simulations are based on an operating model that is individual-based and spatially explicit, and can be parameterized to represent the implications of a range of possible causes for the decline. This model can be used to generate the types of data typically available for the western stock of Steller sea lions. Experiments based on splitting four of the regions identified for past ana! lyses of population dynamics information into sectors that are either open to some fishing or completely closed are considered. The performance of these experiments is, however, poor, owing to the impact of movement, different historical trends in different areas, demographic stochasticity, and the likely size of the effect that the experiments are attempting to detect. These results suggest that the currently available information imply that large-scale experimental manipulation by means of additional spatial closures, where the results are analyzed by examining trends in pup counts, is unlikely to provide an effective means of discriminating among alternative hypotheses for the declining Steller sea lions in Alaska.

abstract
We estimated the risk that the Steller sea lion will be extirpated in western Alaska using a population viability analysis (PVA) that combined simulations with statistically fitted models of historical population dynamics. Our analysis considered the roles that density-dependent and density-independent factors may have played in the past, and how they might influence future population dynamics. It also established functional relationships between population size, population growth rate and the risk of extinction under alternative hypotheses about population regulation and environmental variability. These functional relationships can be used to develop recovery criteria and guide research and management decisions. Life table parameters (e.g., birth and survival rates) operating during the population decline (1978?2002) were estimated by fitting simple age-structured models to time-series of pup and non-pup counts from 33 rookeries (subpopulations). The PVA was carried out by projecting all 33 subpopulations into the future using these estimated site-specific life tables (with associated uncertainties) and different assumptions about carrying capacities and the presence or absence of density-dependent population regulation. Results suggest that the overall predicted risk of extirpation of Stelsler sea lions as a species in western Alaska was low in the next 100 yr under all scenarios explored. However, most subpopulations of Steller sea lions had high probabilities of going extinct within the next 100 yr if trends observed during the 1990s were to continue. Two clusters of contiguous subpopulations occurring in the Unimak Pass area in the western Gulf of Alaska/eastern Aleutian Islands and the Seguam?Adak region in the central Aleutian Islands had relatively lower risks of extinction. Risks of extinction for a number of subpopulations in the Gulf of Alaska were reduced if the increases observed since the late 1990s continue into the fu ture. The risks of subpopulations going extinct were small whe n densit ydependent compensation in birth and survival rates were assumed, even when random stochasticity in these vital rates was introduced.

abstract
Pup mortality and the timing of birth of South American sea lions Otaria flavescens were investigated to determine the possible relationship between fluctuations in prey availability in the Peruvian upwelling ecosystem and current and future reproductive success of sea lions during six consecutive breeding seasons. Our study from 1997 to 2002 encompassed the strongest El Nino on record and one La Nina event. Pup mortality ranged from 13% before El Nino to 100% during El Nino, and was negatively correlated with prey availability. Abortions were also more frequent when prey availability was low. However, pup mortality remained high following El Ni~no due to the punctuated short-term effects it had on population dynamics and subsequent maternal behaviour. Births occurred later in the season after years of low food availability and earlier following years of high food availability. The peak of pupping coincided with the peak of mortality in all years, and may have ! been the product of intensive competition between bulls at the peak of the breeding season. The stronger and more frequent El Ninos that appear to be occurring along the Peruvian coast may produce significant stochastic changes in future births and pup mortality, which may place the vulnerable South American sea lion population in Peru at greater risk.

abstract
Ecosystem models were constructed for the Antarctic and the Bering Sea that incorporate current understanding of biological interactions of species within the ecosystem (i.e., who eats whom and how much). Within the limitations that are inherent to simulations, both models suggest that removal of large whales had little measurable effect on lower trophic levels or on the dynamics of other species in their polar ecosystems. Trophic interactions failed to explain the magnitude of changes in the biomass of the major species groups in the Antarctic and Bering Sea. Nor did fin-fisheries appear to have had a significant effect on the abundance of non-targeted species. This may mean that environmental effects (which were not modeled) play an important role in influencing the dynamics of marine ecosystems. Oceanographic factors such as changes in water temperature or ocean currents likely result in variations in ecosystem production and species recruitment patterns which are not captured by our Ecopath models. The Ecopath modeling approach is a powerful means of synthesizing knowledge about ecosystems and the factors that influence ecosystem dynamics. They provide a straightforward means for estimating trophic levels and niche overlaps with other species to assess the potential for resource competition. While the models failed to support the hypotheses that large whales play a significant structural role in the Antarctic and Bering Sea ecosystems, they do support what most already know ?- i.e., that populations of large whales are easily reduced to low numbers, but take a long, long time to recover. They also help in recognizing the need to consider factors other than food web interactions when assessing the status of cetaceans, as well as highlighting the potential tradeoffs that can result when other species are removed from ecosystems.

abstract
Two distinct viability models are developed for Steller sea lions (Eumetopias jubatus )to evaluate the sensitivity of extinction risk to various levels of stochasticity,spatial scale and density dependence.These models include a metapopulation model,Analysis of the Likelihood of Extinction (ALEX;Possingham et al.,1992; Possingham,H., Davies,I.A.,Noble,I.1992.ALEX 2.2 Operation Manual.Department of Applied Mathematics,University of Adelaide,Adelaide,SA 5005;Australia.),and a model that incorpo- rates both sampling and process error in estimating population parameters from timeseries data (Gerber and DeMaster,1999; Gerber,L.R.,DeMaster,D.P.1999.An approach to endangered species act classification of long-lived vertebrates:a case study of north Pacific humpback whales.Conservation Biology 13 (5);1203 –1214.).Results are compared with a third model that encompasses three different geographic scales (York et al.,1996;York,A.E.,Merrick,R.L.,Loughlin,T.R.1996.An analysis of the Steller Sea lion metapopulation in Alaska.In:McCullough,D.R.(Ed.),Metapopulations and Wildlife Conservation.Island Press, Covelo,CA pp.259 –292).The combination of modeling approaches provides a basis for considering how model parameterization and the selection of classification criteria affect both model results and potential status determinations.Results from the models generally agree with regard to central tendency,25th and 75th percentile times to extinction.For Steller sea lions,the distributions of time to extinction for each model were narrower than the range of extinction distributions between models.If this finding applies generally to listed species,it would suggest that more than one viability model should be considered when listing decisions are made.On a more applied basis,the results of our analysis provide a quantitative assessment of extinction risk of Steller sea lions in the context of its status pursuant to the US Endangered Species Act.

abstract
Over the past 10 years there has been increasing criticism of management decisions that are based on single species approaches and a call for the implementation of ecosystem approaches. The major criticism of single species models is that they cannot predict changes in community struc ture. Unfortunately, our experience in modeling the Bering Sea shows that these same criticisms can also be leveled against ecosystem models. We employed trophic mass balance models (Ecopath and Ecosim) to examine some possible explanations for the changes that occurred in the Bering Sea between the 1950s and 1980s. We removed fish and mammals from the modeled system and tracked how other components of the eco system responded. Our mass balance models indicate that neither whal ing nor commercial fisheries were sufficient to explain the 400% increase in pollock biomass and other changes that may have occurred between the two time periods. The simulations further suggest that environmental factors, affecting recruitment or primary production, may be more impor tant in determining the dynamics of the Bering Sea ecosystem than preda tor prey interactions alone. These findings illustrate that mass balance models that do not account for the impact of climate variability on year class strength cannot provide reliable estimates of trends in marine fish production. However, our models can show how predation and fishing can affect trophic interactions among species. As such, ecosystem models are a useful scientific tool to identify gaps in understanding and data needs, but are unlikely to ever replace single species models. They may instead complement and provide parameters to single species models. Ecosystem models such as ours are still in the early stages of develop ment and will become increasingly more important as a management tool as they begin to incorporate spatial and oceanographic/climatic information.

abstract
The size of Steller sea lion populations in the Gulf of Alaska and Aleutian Islands was estimated by applying life table statistics to counts of pups and adults (non-pups) at rookery sites. Total population size was 5.10 times the number of pups counted or 3.43 times the number of adults counted. Only 55% of the adult population return to rookeries during the summer. Data compiled from published and unpublished sources for all 39 major rookeries in Alaska suggest that the total number of Steller sea lions (including pups) rose from 250 000 to 282 000 between the mid 1950s and the mid 1970s. Since 1980 it has decreased by over 70% (roughly 5% per year) to about 76 000 animals in 1992. Most of the decline appears to have slowed or stopped within two subareas of these large regions- the eastern Aleutians and western Gulf of Alaska. Increases have been occurring in the smaller populations of southeast Alaska. It is not known why these changes have occurred.

keywords     abundance, Steller sea lions, Alaska, decline, #2

abstract
A coalition of North Pacific fishing industry groups has been working to address pending restrictions on the Gulf of Alaska and Eastern Bering Sea commercial fishery as a result of possible decline of the Steller sea lion population. A component of the resea.rch into this problem is the investigataioonf techniques to automate the counting of sea lions from aerial video of the Alaska coast. Currently, sea-lion counts are completed mmually from 35mm slides, and there is some concern regarding t,he accuracy and repeatabilityof such an approach. It is proposed that computer-assisted counting from video tape or digitised slides (CDROM) could provide a better alternative to manual counting in terms of improved spatial coverage, improved reliability/consistency and reduced labour costs.

abstract
From June 1985 through June 1992 regular aerial surveys over Southeast Alaska, the Gulf of Alaska, and Aleutian Islands were conducted to monitor the distribution and abundance of Steller sea lions (Eumetopias jubatus). Furthermore, during 1992 the use of repetitive aerial surveys was introduced to study survey variability. The purposes of our present study were two-fold: (1) To investigate alternative statistical procedures for estimating population trend rates from aerial survey counts, and (2) to apply these procedures to the 1989-92 counts. We focused on 1989-92 because of the more abundant data for that period and to determine the most recent direction of population abundance relative to the well-documented steep decline prior to 1989. We concluded that parametric bootstrapping was the most appropriate interval estimation procedure. For this procedure the 90% confidence interval for the rate of change between the 1989 and 1992 Kenai-Kiska Recovery Plan trend site counts was (-10.19%, 2.62%). For glJ rookery and haul-out sites within the Kenai-Kiska region for which counts were made for both 1989 and 1992, the 90% confidence interval was (-6.20%, 6.81%). For sites outside the Kenai-Kiska region, the 90% confidence interval for the rate of change between 1989 and 1992 was (-37.92%, -6.93%). The parametric bootstrap procedure had the weakness of including data from only pairs of years; thus, 1990-91 information was excluded from the above interval estimates. Consequently, we investigated the use of both generalized linear modeling (GLM) and generalized estimating equations (GEE) as tools for analyzing all four years' data together. GLM seemed less appropriate than GEE, and the latter procedure yielded conclusions similar to those of parametric bootstrapping: 1989-92 stability for sea lion abundance in the Kenai-Kiska region, and some likelihood of decline from 1989-1992 considering all sites together.

abstract
The goal of our study was to assess the status of Steller sea lions in Alaska, review their population biology, and develop a simulation model to explore the role that harvesting and incidental kills by fisheries may have played in the sea lion decline. We also attempted to relate the population declines to the amount of fish caught in the Gulf of Alaska and Aleutian Islands, and to the number of vessels fishing from 1950 to 1990. Using life tables to estimate population size, the numbers of Steller sea lions were estimated for all rookeries for which information was available in each of six areas in the Gulf of Alaska and Aleutian Islands. The total population appears to have risen from 150,000 to 210,000 from the mid 1950s to 1967. The population was then stable for roughly ten years, then increased to 225,000 by 1979. Since then it has decreased to about 85,000. Most of the decline took place in Area 3 (Kodiak region) but there were also significant declines in Areas 4 to 6 (westward of Kodiak). Increases have occurred in the smaller populations of Areas 1 and 2 (southeast Alaska and Prince William Sound). A major growth in domestic fisheries occurred after the declaration of 200 mile zones. The traditional fisheries for salmon, herring and halibut were augmented by major groundfish fisheries. The decline in the numbers of Steller sea lions has been coincidental with the growth in the numbers and size of vessels and the increase in catch. The stabilization in the numbers of Steller sea lions in the Gulf of Alaska from 1956 to 1980 can be attributed to the direct effect of incidental capture in fishing gear, the shooting of sea lions and the harvesting of adults and pups. However, these factors explain but a small portion of the recent population decline, from 1980 to the present. Some sea lions are missing in the arithmetic of population dynamics which cannot be accounted for by movements of animals from one area to another. Whether these losses are caused by the removal of food resources is a circumstantial possibility, but evidence of local abundance of food resources at particular times of the year for particular segments of the population is needed to build a convincing case. Other causes, such as diseases and parasites must also be kept in mind as possible contributing factors. Research on Steller sea lions should focus on the decline in abundance since 1980, changes in body size, the diet at various seasons of the year, bioenergetics and nutritional requirements, and assessment of local abundance of various food items. Long term research on the ecosystem dynamics of the region will be necessary for long term management of all living resources but how best to focus that research is a matter of current scientific debate that will not be resolved quickly.