James R. Bence

TEMPORAL AND SPATIAL VARIATION IN ENVIRONMENTAL IMPACT ASSESSMENT

We compare two approaches to designing and analyzing monitoring studies to assess chronic, local environmental impacts. Intervention Analysis (IA) compares Before and After time series at an Impact site; a special case is Before–After, Control–Impact (BACI), using comparison sites as covariates to reduce extraneous variance and serial correlation. IVRS (impact vs. reference sites) compares Impact and Control sites with respect to Before–After change, treating the sites as experimental units. The IVRS estimate of an “effect” is the same as that of the simplest BACI (though not of others), but IVRS estimates error variance by variation among sites, while IA and BACI estimate it by variation over time. These approaches differ in goals, design, and models of the role of chance in determining the data. In IA and BACI, the goal is to determine change at the specific Impact site, so no Controls are needed. IA does not have controls and BACIs are not experimental controls, but covariates, deliberately chosen to ...

ANALYSIS OF SHORT TIME SERIES: CORRECTING FOR AUTOCORRELATION'

Short time series are common in environmental and ecological studies. For sample sizes of 10 to 50, I examined the performance of methods for adjusting confidence intervals of the mean and parameters of a linear regression for autocorrelation. Similar analyses are common in econometric studies, and serious concerns have been raised about the adequacy of the common adjustment approaches, especially for estimating the slope of a linear regression when the explanatory variable has a time trend. Use of a bias-corrected estimate of the autocorrelation, either in an adjusted t test or in a two-stage approach, outperformed other methods, including maximum likelihood and bootstrap estimators, in terms of confidence interval coverage. The bias correction was, however, sometimes awk- ward to apply. It was generally better to test for autocorrelation at the 0.5 level and use ordinary least squares if the test was not significant, although this pretesting mainly helped for weak autocorrelation and small sample sizes. For the best methods, the coverage was sometimes still substantially less than the stated 95% when autocorrelation was strong, even for sample sizes as large as 50. This was true for estimates of the mean, the regression intercept, and, when the explanatory variable had a time trend, the slope. Simulation results and an example show that different adjustment methods can produce substantially different estimates and confidence intervals. Cautious interpretation of confidence intervals and hy- pothesis tests is recommended.

Space‐Limited Recruitment in Open Systems: The Importance of Time Delays

We study the dynamic behavior of open systems where older or larger individuals can inhibit the recruitment of juveniles into the population. Our approach is to examine the predictions of simple dynamic models. In these models, settlement rate into open space is decoupled from local population density. Our more elaborate models are special cases of Roughgarden et al.s (1985) model for an open marine population with space—limited recruitment. We found, as did Roughgarden et al., that two qualitatively different dynamic outcomes were possible: a stable steady state, and cyclic fluctuation in population density and space occupied. The crucial factor needed to produce cyclic fluctuations is a time delay between settlement and recruitment into the adult population. The introduction of density—dependent mortality in adults serves to stabilize the limit cycle. The effect of growth or development rate on stability depends upon the relationship between these rates and individual size or age. This is because increased growth may act to increase the area occupied at equilibrium (a destabilizing factor), but can cause developmental time lags to either increase or decrease in length. As a result, an unstable system can be stabilized by either an increase or a decrease in juvenile growth rate when adults do not grow, or grow more slowly than juveniles.

DETECTING THE ECOLOGICAL EFFECTS OF ENVIRONMENTAL IMPACTS: A CASE STUDY OF KELP FOREST INVERTEBRATES'

Detecting the environmental impacts of human activities on natural communities is a central problem in applied ecology. It is a difficult problem because one must separate human perturbations from the considerable natural temporal variability displayed by most populations. In addition, most human perturbations are generally unique and thus unreplicated. This raises the problem of deciding whether observed local effects are due to human intervention or to the natural differences in temporal patterns that often occur among different sites. These problems can be successfully addressed with the Before-After/Control-Impact (BACI) sampling design, in which Impact and Control sites are sampled contemporaneously and repeatedly in periods Before and After the human perturbation of interest. In the present case, we use this design to examine the ecological effects of the cooling water discharge from a coastal nuclear power plant in southern California. The results suggest some general lessons about the process of impact assessments that are applicable in many ecological contexts. In systems where plants and animals are long-lived and recruit sporadically, the rates of change in density are often so low that sampling more than a few times per year will introduce serial correlations in the data. As a result, for studies of few years duration, few samples will be taken. A small sample size means that the tests of the assumptions underlying the statistical analyses, e.g., independence and additivity, will have low power. This injects uncertainty into the conclusions. Small sample size also means that the power to detect any but very large effects will be low. In our study, sampling periods of 2- yr both Before and After the impact were not long enough to detect a halving or doubling of populations at the impact site. We concluded that there were significant environmental impacts because: (1) the effect size was generally very large (°-75%); (2) there was a consistent pattern among species; (3) there were two Impact sites, and effects were larger at the site nearest the discharge; (4) the observed effects accorded with physical changes that could be linked with the source of impact; and (5) a number of alternative mechanisms, unrelated to the source of impact, were examined and rejected. Relative to control populations, there were statistically significant reductions in density of snails, sea urchins, and sea stars, all of which occurred primarily on rocky substrates. All of the reductions were larger at the Impact station about 0.4 km from the discharge than at a second Impact station 1.4 km away. The most plausible mechanisms for the declines seem to be linked to the turbidity plume created by the power plant and the resultant increase in suspended inorganic and organic materials (+46% at the Impact site nearest the discharge). Any associated flux of fine particles on rocks would have deleterious effects on many of the hard benthos. Populations of two filter-feeding species, a gorgonian coral and a sponge, showed relative increases in density. Although the increase in populations of filter feeders could be related to the ingestion, killing, and discharge of tons of plankton by the cooling system, an alternative natural mechanism was also considered reasonable. Monitoring studies or relatively long-lived organisms will often have low power to detect ecologically significant changes in density. The present study of kelpforest organisms extended over nearly 6 yr, yet the resulting statistical tests generally had power of <30% to detect a doubling or halving in density at a significance level of .05. In such a community it would be a mistake to conclude that there were no significant ecological effects based on conventional hypothesis tests. Unless there is a willingness to accept the fact that changes in natural populations on the order of 50% will often go undetected, the standards and types of evidence used to demonstrate environmental impacts must be changed.

Recruitment Variability of Alewives in Lake Michigan

We used a long-term series of observations on alewife Alosa pseudoharengusabun- dance that was based on fall bottom-trawl catches to assess the importance of various abiotic and biotic factors on alewife recruitment in Lake Michigan during 1962-2002. We first fit a basic Ricker spawner-recruit model to the lakewide biomass estimates of age-3 recruits and the cor- responding spawning stock size; we then fit models for all possible combinations of the following four external variables added to the basic model: an index of salmonine predation on an alewife year-class, an index for the spring-summer water temperatures experienced by alewives during their first year in the lake, an index of the severity of the first winter experienced by alewives in the lake, and an index of lake productivity during an alewife year-classs second year in the lake. Based on an information criterion, the best model for alewife recruitment included indices of salmonine predation and spring-summer water temperatures as external variables. Our analysis corroborated the contention that a decline in alewife abundance during the 1970s and early 1980s in Lake Michigan was driven by salmonine predation. Furthermore, our findings indicated that the extraordinarily warm water temperatures during the spring and summer of 1998 probably led to a moderately high recruitment of age-3 alewives in 2001, despite abundant salmonines. A key problem in fisheries research is predicting recruitment from a given level of spawning stock size (Sissenwine et al. 1988; Myers et al. 2001; Kehler et al. 2002). Fish recruitment can be strong- ly influenced by many abiotic and biotic factors, including water temperature, water movements, predation, and spawning stock size (Sissenwine 1984; Hilborn and Walters 1992). Although im- portant factors affecting recruitment may vary across ecosystems (Madenjian et al. 1996), inter- esting patterns may emerge by comparing recruit- ment analyses for populations of a species across ecosystems (Myers 1998). An invasion of alewives Alosa pseudoharengus during the 1940s proved to be an important stressor to the Lake Michigan ecosystem (Wells and

INDIRECT EFFECTS AND BIOLOGICAL CONTROL OF MOSQUITOES BY MOSQUITOFISH

SUMMARY (1) The mosquitofish, Gambusia affinis (Baird & Girard), reduced the abundance of the pest mosquito Culex tarsalis Coquillet in a rice field in San Joaquin Valley, California. (2) The mosquitofish also reduced the abundance of other aquatic invertebrate taxa, including predatory insects and zooplankton. (3) The reduction in predatory insect abundance, due to the fish, reduced mosquito mortality due to insects. (4) The reduction in the abundance of alternative prey (mainly zooplankton) increased predation rates on mosquitoes by individual fish. A low abundance of zooplankton increased the predation rates of individual predatory insects. (5) Successful control of mosquitoes results from a direct negative effect by mosquitofish that more than outweighs indirect positive effects of the mosquitofish on mosquitoes. In other studies, the introduction of mosquitofish increased the abundance of mosquito larvae; indirect positive effects of the type demonstrated in this study could account for these results.

Lake Trout Mortality and Abundance in Southern Lake Huron

Abstract Populations of lake trout Salvelinus namaycush in the main basin of Lake Huron collapsed in the 1940s because of predation by sea lampreys Petromyzon marinus and commercial fishing. Efforts to rehabilitate lake trout have emphasized reduction of mortality and the stocking of hatchery-reared lake trout to reestablish populations. We fit a statistical catch-at-age model for lake trout in the southern main basin of Lake Huron using a maximum likelihood approach to estimate mortality rates and abundance during 1984–1993. This represents the first such analysis for lake trout in the Great Lakes, and a flexible application of the approach proved useful for integrating diverse information and assessing population and mortality trends. Sea lamprey-induced mortality and recruitment of lake trout to age 1 were calculated external to model fitting. Recruitment was based on numbers of lake trout stocked because natural recruitment is negligible. Sea lamprey-induced mortality rates were based on observed woun...

Yellow Perch Dynamics in Southwestern Lake Michigan during 1986–2002

Abstract We examined the role of harvest in the collapse of the population of yellow perch Perca flavescens in southwestern Lake Michigan during the mid to late 1990s. After the great decrease in this population at that time, commercial fisheries in Illinois and Wisconsin were closed during 1996–1997 (and have remained closed), and stricter regulations were placed on recreational fisheries. Reproductive failure has been implicated as the primary cause of the population collapse, but the role of fishing in the collapse was not rigorously investigated in previous studies. We conducted an age-, size-, and sex-structured stock assessment of yellow perch to estimate population size and examine historical trends in fishing mortality in Illinois and Wisconsin waters of southwestern Lake Michigan. Model estimates indicated that yellow perch abundance in 2002 was less than 10% of the 1986 abundance in Wisconsin and about 20% of the respective population in Illinois. Annual mortality rates for females age 4 and old...

Sea Lamprey (Petromyzon marinus) Parasite-host Interactions in the Great Lakes

Prediction of how host mortality responds to efforts to control sea lampreys (Petromyzon marinus) is central to the integrated management strategy for sea lamprey (IMSL) in the Great Lakes. A parasite-host submodel is used as part of this strategy, and this includes a type-2 multi-species functional response, a developmental response, but no numerical response. General patterns of host species and size selection are consistent with the model assumptions, but some observations appear to diverge. For example, some patterns in sea lamprey marking on hosts suggest increases in selectivity for less preferred hosts and lower host survival when preferred hosts are scarce. Nevertheless, many of the IMSL assumptions may be adequate under conditions targeted by fish community objectives. Of great concern is the possibility that the survival of young parasites (parasitic-phase sea lampreys) varies substantially among lakes or over time. Joint analysis of abundance estimates for parasites being produced in streams and returning spawners could address this. Data on sea lamprey marks is a critical source of information on sea lamprey activity and potential effects. Theory connecting observed marks to sea lamprey feeding activity and host mortality is reviewed. Uncertainties regarding healing and attachment times, the probability of hosts surviving attacks, and problems in consistent classification of marks have led to widely divergent estimates of damages caused by sea lamprey. Laboratory and field studies are recommended to provide a firmer linkage between host blood loss, host mortality, and observed marks on surviving hosts, so as to improve estimates of damage.

Population Regulation, Convergence, and Cannibalism in Notonecta (Hemiptera)

In a population convergence experiment, the initial densities of adults of the predatory backswimming bug Notonecta hoflmanni were set above and below a putative equilibrium density in stock tanks. The experiment was done at two constant rates of food supply (wingless Drosophila) for the larger instars (in natural pools Notonecta feed mainly on terrestrial arthropods that fall on the water surface). It was predicted that the densities of the resulting populations would converge on an equilibrium set by the rate of food supply for the larger instars. The tanks also contained zooplankton (mainly Daphnia), which were the main food supply of the smaller instars of Notonecta. The resulting over- wintering populations converged towards the appropriate equilibrium densities, via density- dependent and food-dependent fecundity and then cannibalism. However, the populations overshot their equilibria, producing overconvergence. In natural populations such over- convergence might tend to produce 2-yr cycles in abundance. Overconvergence resulted from the insensitivity of the survivorship of the original adults (at least over the short term) to differences in food supply between treatments, allowing them to continue to affect (via cannibalism and reproduction) the eventual density of the new overwintering popu- lation. Because Notonecta population density was determined by the externally supplied, locally uncoupled food supply, even though the early instars depended for food largely upon dynamic populations of zooplankton, the dynamics of the Notonecta population were simpler than a description of the food web might suggest.