Search USGS⌕ Search

Geology topics

W.A. Link

Publications and source records attributed to W.A. Link.

At least 55 records · Page 3Linked to original sources
USGS2002

Model-based estimation of individual fitness

Fitness is the currency of natural selection, a measure of the propagation rate of genotypes into future generations. Its various definitions have the common feature that they are functions of survival and fertility rates. At the individual level, the operative level for natural selection, these rates must be understood as latent features, genetically determined propensities existing at birth. This conception of rates requires that individual fitness be defined and estimated by consideration of the individual in a modelled relation to a group of similar individuals; the only alternative is to consider a sample of size one, unless a clone of identical individuals is available. We present hierarchical models describing individual heterogeneity in survival and fertility rates and allowing for associations between these rates at the individual level. We apply these models to an analysis of life histories of Kittiwakes (Rissa tridactyla ) observed at several colonies on the Brittany coast of France. We compare Bayesian estimation of the population distribution of individual fitness with estimation based on treating individual life histories in isolation, as samples of size one (e.g. McGraw & Caswell, 1996).

Book chapter↗
USGS2002

Random effects and shrinkage estimation in capture-recapture models

We discuss the analysis of random effects in capture-recapture models, and outline Bayesian and frequentists approaches to their analysis. Under a normal model, random effects estimators derived from Bayesian or frequentist considerations have a common form as shrinkage estimators. We discuss some of the difficulties of analysing random effects using traditional methods, and argue that a Bayesian formulation provides a rigorous framework for dealing with these difficulties. In capture-recapture models, random effects may provide a parsimonious compromise between constant and completely time-dependent models for the parameters (e.g. survival probability). We consider application of random effects to band-recovery models, although the principles apply to more general situations, such as Cormack-Jolly-Seber models. We illustrate these ideas using a commonly analysed band recovery data set.

Book chapter↗
USGS2002

Using Christmas Bird Count data in analysis of population change

The scientific credibility of Christmas Bird Count (CBC) results depend on the development and implementation of appropriate methods of statistical analysis. The key to any successful analysis of CBC data is to begin with a careful review of how the limitations of the data are likely to influence the results of the analysis, then to choose methods of analysis that accommodate as much as possible the limitations of the survey. For our analyses of CBC data, we develop a flexible model for effort adjustment and use information from the data to guide the selection of the best model. We include geographic structuring to accommodate the regional variation in number of samples, use a model that allows for overdispersed poisson data appropriate for counts, and employ empirical Bayes procedures to accommodate differences in quality of information in regional summaries. This generalized linear model approach is very flexible, and can be applied to a variety of studies focused on factors influencing wintering bird populations. In particular, the model can be easily modified to contain covariates, allowing for assessment of associations between CBC counts and winter weather, disturbance, and a variety of other environmental factors. These new survey analysis methods have added value in that they provide insights into changes in survey design that can enhance the value of the information. The CBC has been extremely successful as a tool for increasing public interest in birding and bird conservation. Use of the information for bird conservation creates new demands on quality of information, and it is important to maintain a dialogue between users of the information, information needs for the analyses, and survey coordinators and participants. Our work as survey analysts emphasizes the value and limitations of existing data, and provides some indications of what features of the survey could be modified to make the survey a more reliable source of bird population data. Surveys only remain useful if they adapt to current needs, while still maintaining consistency with historical goals.

American Birds↗
USGS1999

Controlling for varying effort in count surveys: An analysis of Christmas Bird Count data

The Christmas Bird Count (CBC) is a valuable source of information about midwinter populations of birds in the continental U.S. and Canada. Analysis of CBC data is complicated by substantial variation among sites and years in effort expended in counting; this feature of the CBC is common to many other wildlife surveys. Specification of a method for adjusting counts for effort is a matter of some controversy. Here, we present models for longitudinal count surveys with varying effort; these describe the effect of effort as proportional to exp(B effortp), where B and p are parameters. For any fixed p, our models are loglinear in the transformed explanatory variable (effort)p and other covariables. Hence, we fit a collection of loglinear models corresponding to a range of values of p and select the best effort adjustment from among these on the basis of fit statistics. We apply this procedure to data for six bird species in five regions, for the period 1959-1988.

Journal of Agricultural, Biological, and Environme↗
USGS1999

Modeling pattern in collections of parameters

Wildlife management is increasingly guided by analyses of large and complex datasets. The description of such datasets often requires a large number of parameters, among which certain patterns might be discernible. For example, one may consider a long-term study producing estimates of annual survival rates; of interest is the question whether these rates have declined through time. Several statistical methods exist for examining pattern in collections of parameters. Here, I argue for the superiority of 'random effects models' in which parameters are regarded as random variables, with distributions governed by 'hyperparameters' describing the patterns of interest. Unfortunately, implementation of random effects models is sometimes difficult. Ultrastructural models, in which the postulated pattern is built into the parameter structure of the original data analysis, are approximations to random effects models. However, this approximation is not completely satisfactory: failure to account for natural variation among parameters can lead to overstatement of the evidence for pattern among parameters. I describe quasi-likelihood methods that can be used to improve the approximation of random effects models by ultrastructural models.

Journal of Wildlife Management↗
USGS1999

Factors influencing counts in an annual survey of Snail Kites in Florida

Snail Kites ( Rostrhamus sociabilis ) in Florida were monitored between 1969 and 1994 using a quasi-systematic annual survey. We analyzed data from the annual Snail Kite survey using a generalized linear model where counts were regarded as overdispersed Poisson random variables. This approach allowed us to investigate covariates that might have obscured temporal patterns of population change or induced spurious patterns in count data by influencing detection rates. We selected a model that distinguished effects related to these covariates from other temporal effects, allowing us to identify patterns of population change in count data. Snail Kite counts were influenced by observer differences, site effects, effort, and water levels. Because there was no temporal overlap of the primary observers who collected count data, patterns of change could be estimated within time intervals covered by an observer, but not for the intervals among observers. Modeled population change was quite different from the change in counts, suggesting that analyses based on unadjusted counts do not accurately model Snail Kite population change. Results from this analysis were consistent with previous reports of an association between water levels and counts, although further work is needed to determine whether water levels affect actual population size as well as detection rates of Snail Kites. Although the effects of variation in detection rates can sometimes be mitigated by including controls for factors related to detection rates, it is often difficult to distinguish factors wholly related to detection rates from factors related to population size. For factors related to both, count survey data cannot be adequately analyzed without explicit estimation of detection rates, using procedures such as capture-recapture.

The Auk↗
USGS1999

On the importance of controlling for effort in analysis of count survey data: Modeling population change from Christmas Bird Count data

Count survey data are commonly used for estimating temporal and spatial patterns of population change. Since count surveys are not censuses, counts can be influenced by 'nuisance factors' related to the probability of detecting animals but unrelated to the actual population size. The effects of systematic changes in these factors can be confounded with patterns of population change. Thus, valid analysis of count survey data requires the identification of nuisance factors and flexible models for their effects. We illustrate using data from the Christmas Bird Count (CBC), a midwinter survey of bird populations in North America. CBC survey effort has substantially increased in recent years, suggesting that unadjusted counts may overstate population growth (or understate declines). We describe a flexible family of models for the effect of effort, that includes models in which increasing effort leads to diminishing returns in terms of the number of birds counted.

Book chapter↗
USGS1999

Regional analysis of population trajectories from the North American Breeding Bird Survey

The North American Breeding Bird Survey (BBS) was started in 1966, and provides information on population change and distribution for most of the birds in North America. The geographic extent of the survey, and the logistical compromises needed to survey such a large area, present many challenges for estimation from BBS data. In this paper, we describe the survey and discuss some of the limitations of the survey design and implementation. Analysis of the survey has evolved over time as new statistical methods and insights into the analysis of count data are developed. Survey results and analysis tools for the BBS are now available over intemet; we present new methods that use generalized linear models for estimation of population change and empirical Bayes procedures for regional summaries.

Book chapter↗
USGS1999

Predicting chick survival and productivity of Roseate Terns from data on early growth

Early growth of Roseate Tern (Sterna dougallii) chicks is a strong predictor of chick survival and hence of productivity. We developed discriminant functions to predict chick survival from body-masses measured during the first 3 days of life. Productivity is estimated by assuming that almost all A-chicks (first-hatched in each brood) survive to fledging, and using the discriminant functions to predict survival of B-chicks (second-hatched in each brood). A relation between survival rates and classification rates is derived, allowing discriminant function results to be used in predicting survival rates. In the absence of predation, the resulting estimates of chick survival and productivity are almost as good as those obtained by more intensive methods, but require much less effort and much less disturbance. This approach might be useful for other seabird species in which chick survival is determined primarily by parental performance.

Waterbirds↗
USGS1998

Unbiasedness

Unbiasedness is probably the best known criterion for evaluating the performance of estimators. This note describes unbiasedness, demonstrating various failings of the criterion. It is shown that unbiased estimators might not exist, or might not be unique; an example of a unique but clearly unacceptable unbiased estimator is given. It is shown that unbiased estimators are not translation invariant. Various alternative criteria are described, and are illustrated through examples.

Book chapter↗
USGS1997

Estimation of population trajectories from count data

Monitoring of changes in animal population size is rarely possible through complete censuses; frequently, the only feasible means of monitoring changes in population size is to use counts of animals obtained by skilled observers as indices to abundance. Analysis of changes in population size can be severely biased if factors related to the acquisition of data are not adequately controlled for. In particular we identify two types of observer effects: these correspond to baseline differences in observer competence, and to changes through time in the ability of individual observers. We present a family of models for count data in which the first of these observer effects is treated as a nuisance parameter. Conditioning on totals of negative binomial counts yields a Dirichlet compound multinomial vector for each observer. Quasi-likelihood is used to estimate parameters related to population trajectory and other parameters of interest; model selection is carried out on the basis of Akaike's information criterion. An example is presented using data on Wood thrush from the North American Breeding Bird Survey.

Biometrics↗
USGS1996

A resource conservative procedure for comparison of dose-response relationships

The evaluation of effects of toxicants on a wildlife community can be complicated by varying responses among the community's constituent populations. Even within populations, considerable variability in dose-response relations may result from different avenues of exposure to the toxicant. Full-scale investigations of the dose-response relations among a variety of species and avenues of exposure can therefore be prohibitively expensive, whether this expense is measured by the number of experimental animals needed, by the human resources committed to the study, or by laboratory expenses. We propose an abbreviated protocol for investigations of multiple dose-response relations that is designed to limit these expenses. The protocol begins with the judicious choice of a baseline dose-response relation to be estimated by a full-scale study involving a minimum of five doses levels, with 10 subjects per dose level. This relation is then used as the basis for rapid screening of subsequent dose-response relations, which are compared to the baseline relation by testing for differences in the median effective dosages. These secondary studies can consist of as few as 14 animals exposed to the estimated median lethal concentration from the baseline study. We describe MS-DOS-compatible software available from the authors that can be used to analyze these data.

Environmental Toxicology and Chemistry↗
USGS1996

The 1994 and 1995 summary of the North American Breeding Bird Survey

Data from the North American Breeding Bird Survey were used to estimate continental and regional changes in bird populations for the 2-year periods of 1993-1994 and 1994-1995. These 2-year changes were placed in the context of population trends estimated over the 1966-1995 interval. The 2-year changes were more positive during the 1993-1994 period, when 54.2% of all species exhibited positive continental trend estimates. This percentage was reduced to 47.7% during 1994-1995, as compared with 50.5% of all species having positive continental trend estimates over then entire survey period. In general, the percentage of increasing species in the Central and Western BBS regions was highest during 1993-1994, with a very marked decline in the Western BBS Region during 1994-1995. The percentage was highest in the Eastern BBS Region during 1994-1995. The continental and regional percentages of species with positive trend estimates were also analyzed for 12 groups of North American birds having shared life-history traits. Over the entire survey period, grassland birds remain the species group with the smallest percentage of increasing species. Trends during these 2-year intervals do not indicate any consistent improvement in the overall declines experienced by grassland birds since the mid-1960s.

Bird Populations↗
USGS1995

Estimation and confidence intervals for empirical mixing distributions

Questions regarding collections of parameter estimates can frequently be expressed in terms of an empirical mixing distribution (EMD). This report discusses empirical Bayes estimation of an EMD, with emphasis on the construction of interval estimates. Estimation of the EMD is accomplished by substitution of estimates of prior parameters in the posterior mean of the EMD. This procedure is examined in a parametric model (the normal-normal mixture) and in a semi-parametric model. In both cases, the empirical Bayes bootstrap of Laird and Louis (1987, Journal of the American Statistical Association 82, 739-757) is used to assess the variability of the estimated EMD arising from the estimation of prior parameters. The proposed methods are applied to a meta-analysis of population trend estimates for groups of birds.

Biometrics↗
USGS1994

Density estimation using the trapping web design: A geometric analysis

Population densities for small mammal and arthropod populations can be estimated using capture frequencies for a web of traps. A conceptually simple geometric analysis that avoid the need to estimate a point on a density function is proposed. This analysis incorporates data from the outermost rings of traps, explaining large capture frequencies in these rings rather than truncating them from the analysis.

Biometrics↗
USGS1994

On the importance of sampling variance to investigations of temporal variation in animal population size

Our purpose here is to emphasize the need to properly deal with sampling variance when studying population variability and to present a means of doing so. We present an estimator for temporal variance of population size for the general case in which there are both sampling variances and covariances associated with estimates of population size. We illustrate the estimation approach with a series of population size estimates for black-capped chickadees (Parus atricapillus) wintering in a Connecticut study area and with a series of population size estimates for breeding populations of ducks in southwestern Manitoba.

Oikos↗
USGS1994

Observer differences in the North American Breeding Bird Survey

Because count data collected in many bird surveys are only an index to population size, factors that can influence the counts must be identified and incorporated into analyses. Observer quality is often ignored in analyses of population changes from survey data, but observers differ in methods and capabilities and, hence, tend to count different numbers of birds. We assess the consequences of between-observer differences in counts for estimation of population trends in the North American Breeding Bird Survey. Observer differences in numbers of birds counted were found in 50% of the 369 species we examined. For many species, observers in later years tended to count more birds than observers in earlier years, suggesting an increase in observer quality over time. Analysis of population trends from 1966 through 1991 indicates that failure to include observers as covariables in the analysis results in an overly optimistic view of population trends.

The Auk↗