Introduction

To monitor a group of animals implies that there is a need for information on the population status or health that can only be met by collecting data. As every species is inherently countable and inherently of interest, some process for choosing which species or species groups to survey must take place. Below is a list of factors to consider that may influence which species you choose and which life stage or parameter about that species you monitor. Such choices will influence your sampling frame, the numbers of plots or samples you need, costs, number and types of personnel needed; and your choice of monitoring technique. Consequently, thoughtful consideration at this time in formulating your choices will circumvent future problems.

Any long-term monitoring program must be thought of as a long-term investment, more so in many ways than the "permanent " buildings on your site or the management plans you create. The reason: you cannot go back in time. Data not collected or poorly collected cannot be recollected at some later date. That lost year is gone, lost; irretrievable by any means. The monitoring program you create now will become the reference point for those species for that time forward. Your decisions have long-term consequences!

After considering the factors below, you should have all the information you need to justify your species choices. Listing and articulating those justifications will strengthen your proposals and arguments to create, fund, and continue your monitoring program.

Political and Philosophical Factors:

Long-term monitoring requires a long-term commitment of resources. Because those resources could be invested elsewhere, it makes sense to take some time in the beginning to outline your rationale and justification for choosing which species to monitor. Some important questions to address in that justification are:

• Are these species tied to ongoing management on your land?
• Are these species of local, regional, or national importance?
• Is your land of special importance to these species?
• Will the data you plan to collect become part of a larger regional network?
• Are these species already monitored sufficiently by existing programs?

Biological Factors:

Abundance

An abundant species (given that it is reasonably detectable) will require less effort to count on the whole than a rare one. Uncommon species, despite often being the species you are the most interested in protecting or monitoring, have the problem of being (surprise!) absent from many of your sampling stations, causing you to increase your sampling effort in order to accommodate all those zeros. When an uncommon species is detected, its numbers are usually low. Because resulting datasets are largely composed of zeros and ones, different analysis approaches specifically designed to handle these situations are required. Sampling distributions will almost always be non-normal and statistical transformations unsatisfying. You may need to pool your data across sites and use presence-absence or non-parametric analyses, limiting the types of statements you can make about your populations. Other things being equal, monitoring abundant species or employing a technique that permits large numbers of animals will be easier.

Detectability

Detectability can be defined as the probability of locating, identifying, and counting each individual of a species if it is present at your counting station while you are there to sample it. It makes common sense, all else being equal, to sample species that are easier to find and more consistently detectable. Some examples of highly detectable species during portions of the year are: Spring Peepers, California Treefrogs, Tiger Swallowtails, Red-eyed Vireos. Examples of species with low detectability rates include most snakes, and carnivores.

For most species, detectability varies from place to place, season to season and, more problematically, over time. This variation in detectability may bias estimates of change and should be accounted for in your monitoring program. A more detailed discussion of the problems with detectability in creating monitoring programs can be found here.

Variation in counts

Any monitoring program's ability to estimate change is influenced by the amount of year-to-year variability in that program's counts. High variation in those counts will diminish your ability to estimate trends. This count variability also influences the sampling effort (e.g. the number of plots, the amount of time or effort spent at each plot, and the number of replicates of the plot within a season) needed to detect trends of the magnitude established by the organizers of the survey. Intuitively, more sampling effort must be expended when there is high variability.

We have collected counts for many species from the literature and from published and unpublished databases, calculated their variability, and put them into a database that you can access at our CV (coefficient of variation) database site (links at top of this page). In the sample size section of this web site, we help step you through determining how to estimate sample sizes based on either your own pilot data (only rarely is this sufficient) or the CV database.

Unbiased Sampling Techniques Available

While variation in counts has the effect of masking long-term trends, bias in population indices and estimators can cause monitoring programs to reach false conclusions regarding the trajectory of those population changes. For most species groups there are usually sampling approaches (e.g., mark-recapture estimators) that are unbiased if the proper model is chosen or have indices that correct for the common biasing factors such as weather and observer differences, using co-variables.

The topic of bias, sampling approach, and the correction for changes in detectability are some of the most contentious in the monitoring field. Be sure to read the sampling approaches section of this web site and the associated discussions on detectability and choice of approach.

For detailed information about the biases associated with individual monitoring techniques we have listed known biases within each technique's account on this web site. All else being equal, species that have well-developed, unbiased sampling approaches and techniques should be favored.

Multiple Species

Many counting techniques permit data from more than one species to be collected at one time. A technique that collects useful information on an entire group of species yields a more complete portrait of an area's population health and status than does information for but a single species. Often these data also permit the estimation of species richness (total number of species present) and there is usually a decrease in the cost of data collection per species. Furthermore, collecting information on a entire community of species allows you to compare trends among species, search for patterns of change within an ecological guild, and will give you insights into as many aspects of your landscape as you have species being monitored.

Species Identification Problems

The ability of an employee or volunteer to identify the species being monitored needs to be considered. Techniques that require an observer to find, catch, or locate animals are all prone to bias as observers change and grow in their identification abilities. Uncorrected, these observer differences can greatly alter the results of your monitoring program, masking changes you are trying to detect in animal populations.

Under each of our technique's write-ups we provide detailed information about each techniques' vulnerability to observer differences.

Species Surrogates

Traditionally, wildlife monitoring implies population monitoring of adult vertebrates. However, at times monitoring the targeted adult animals may not be a simple choice. This is especially true if:

• The species is uncommon
• The species is shy or otherwise difficult to find
• The available techniques or approaches are costly
• There are safety factors associated with counting them directly

In these cases surrogates such as habitat, food plants, or associated (more common) species could be chosen. However, a great deal of caution has to be taken with this approach, because in these cases you are NOT measuring the species populations. Without periodic checks of the real population, your monitoring of surrogates could easily mislead those that use the data into mistaken notions about the population status of the species of interest.

Habitat monitoring becomes a focus when monitoring is incorporated into management impacts and actions. For more information about integrating monitoring with management activities see the adaptive management sections of this site.

Features to measure

There are 3 population attributes that can be measured in a monitoring program:

• Population (numbers of animals alive)
• Fecundity (birth rate)
• Survivorship (death rate)

Counts of adults are most commonly used in monitoring programs. However, information about the juvenile age classes, birth rates, and survival often reveal more information about the health of the population than do simple counts of adults, though the costs of collecting such information are substantially higher than collecting count data.

Keystone species and ecological dominants

Within any unit of land, some species have a disproportionate effect on the ecology of the site either through sheer numbers and biomass or through their ability to control the diversity and pattern of other species via predation or habitat modification. A species of such consequence would often be worthy of monitoring as it, in effect, is managing and controlling population numbers on the site.

Environmental indicators, indicator species, bioindicators, and management indicator species

Every species has a range of conditions under which it thrives. Remove any component of those conditions and the species disappears or no longer successfully reproduces. Thus the continued presence of a species is an indication that the environmental conditions which it requires remain. By choosing to monitor a set of species that require high quality environments, specialized habitats, or conditions that a manager may want to promote a sense of the region's environmental health can be made. Since environmental or habitat health is often difficult for us to measure directly, due to the many factors (often unknown or ephemeral) that contribute to the conditions, it is often easier to measure the status of the species that require them to develop an assessment.

EPA has a comprehensive web site about the selection, use, and analysis of bioindicators and Indices of Biotic Integrity for aquatic systems: BioIndicators of Watershed Health

The U. S. Forest Service uses indicator species in their management plans, usually calling them Management Indicator Species. Their Management Indicator Species report for Daniel Boone National Forest is online.

Jim Karr has developed a number of assessment techniques that combine the results individual species abundance indices to develop a multispecies index that aims to summarize the biological well-being of a location, often now termed its "biological integrity." References to his work are available at his web site.

Authors' note: we could really use some additional links or summary references on this topic, but were unable to locate any. If you know of such a thing, please drop us a line at sam_droege@usgs.gov. Thank you.

Next step: What types of information you need to collect about the species you have chosen.