In the preceding section, emphasis was placed on monitoring birds over large areas, and the monitoring method emphasized was the point count. Indeed, point counts are useful when a large area must be surveyed, or if all that is required is a simple list of the species present. However, it is becoming increasingly clear that to meet many objectives pertinent to research and management, we must know something about the demographic parameters of bird populations. Demographic parameters include vital rates such as productivity and survival, and estimates of population increase or decrease obtained from vital rates. Point counts do not provide these data. In fact, presence/absence, or even abundance, can be a misleading indicator of habitat quality (Van Horne 1983). For example, forest fragments occupied by a species may not be suitable breeding habitat if the individuals detected are mostly unmated males (Gibbs and Faaborg 1990), or if pairs present consistently produce no young (i.e., a sink; Pulliam 1988). Any attempt to relate population viability to landscape characteristics or to alternative management practices is also meaningless without some estimate of the vital rates of the populations in question. That sentiment is echoed by many of the papers in this section.

Several general approaches are used in demographic studies. Three such methods—intensive nest monitoring and netting, nest monitoring alone, and constant-effort mist netting—are represented in this section. A fourth method, using behavioral cues as an index to reproductive success, is not covered here but is useful when working with species whose nests are extremely difficult to find (Vickery et al. 1992).

For someone initiating a project that includes estimation of productivity, BBIRD protocols provide several advantages. First, a standardized set of variables, including methods for point counts and habitat measurement, already has been established. Many aspects of logistics (e.g., number and size of plots, crew size and makeup) also have been developed, although methods often have to be fine-tuned for each specific project. Therefore, the investigator does not have to start from scratch when developing protocols. Also, much of the BBIRD program revolves around nest monitoring and measurement. Thus, the program follows the spirit of intensive studies by directly assessing the productivity of individual nests. A drawback is that BBIRD protocols do not require marking individual birds, so that while nesting success and productivity of individual nests are assessed, seasonal productivity is not.

Another interesting problem that rests not so much with BBIRD as with the individual investigator using BBIRD is that the investigator is encouraged to find, monitor, and measure as many nests of as many species as possible, usually on several plots. This can present a problem in that most species will be represented only by a few nests that, while they contribute to the overall database, are too few to make any sort of population inference in that particular study. Effort expended on those species will detract from effort perhaps better spent on abundant species (Hejl and Holmes, this volume). From an experimental design standpoint, if treatments (fragmentation, habitat type, management practice) are associated with or assigned to individual plots, then sample sizes (i.e., number of nests per species) within plots are often not adequate to estimate parameters of interest such as nesting success (Hensler and Nichols 1981). In such cases, one is often forced to combine several plots from the same treatment and commit sacrificial pseudoreplication (Hurlbert 1984) to examine differences among treatments. This problem can sometimes be overcome by focusing effort on the nests of a few focal bird species and ignoring the rest (Hejl and Holmes, this volume).

Last, using reproductive success alone can be misleading because it is only one component of annual productivity. However, if additional information is known, such as the length of the breeding period and the time between nesting attempts, then models such as those developed by Pease and Gryzbowski (1995) can estimate annual productivity. These models also take into account the fact that brood parasitism precludes renesting as long as the nest is active, which in some instances can significantly modify estimates of annual productivity.

Constant-effort mist netting takes an approach that is fundamentally different from the two above. It entails capturing birds in mist nets using standardized procedures and drawing inferences about productivity and survival from the numbers and ages of birds captured both among and within years. In the USA, it is best exemplified by the Monitoring Avian Productivity and Survivorship (MAPS) program, discussed in this section by DeSante, Rosenberg et al., and Nur et al. Nests are generally not monitored (although doing so as a check on productivity estimates is encouraged). Rather, assessment of productivity is based on captures of hatch-year (HY) birds in mist nets. Standardized protocols call for > 10, 12-m nets to be operated throughout the breeding season for a total of 10-12 days. It assumes that the intraspecific ratio of HY birds to adult birds is an index to the productivity for that species in that area. It therefore bypasses outcomes of individual nests, instead assessing seasonal output. Its chief advantage is that, in addition to assessing productivity, it also allows annual survivorship (the probability that a bird will survive one year) to be estimated when performed in one location over a number of years.

There are other advantages. First, a smaller crew is needed. In some locations, a single experienced bander can operate all the nets. Also, the system lends itself to training new banders by giving them experience in removing birds from nets and banding them while the supervisor is nearby. More than a few ornithologists determine their career while under the careful tutelage of an experienced and patient bander. Also, the system lends itself to public demonstrations of field work with migratory birds, meeting outreach and education needs along with monitoring and research. Last, protocols call for point counts and additional habitat measurements, although these are not as extensive as BBIRD habitat measurements.

It is perhaps human nature to consider one approach superior and to reject the other approaches based on the above criteria. However, if one considers the intensive approach to be overall the superior one --  but unrealistic because of expertise, cost, or other practical considerations -- then it is perhaps useful to consider the two standardized methods, BBIRD and MAPS, to be components of the intensive approach. That is, intensive studies include both nest monitoring and banding, and therefore the BBIRD and MAPS approaches complement each other to some extent. One might start out by using either BBIRD or MAPS protocols, depending on preference, and endeavor to integrate the other approach into the study at a later date.

A cautionary note also is in order. Potential investigators need to be aware of problems associated with estimation of demographic parameters. Survival estimation is problematic because of the tendency of long-distance migrants to disperse far from the site where they were first captured. Although still alive, they are indistinguishable from dead birds if not resighted. This problem, discussed here by Marshall et al., is so severe for hatch-year birds dispersing from their natal site that first year survival is hardly ever estimated. Estimation of productivity also is problematic because it can vary greatly both within and among years (e.g., Wilson and Cooper 1998). Reproductive success estimated from nests monitored during only a small part of the breeding season, or from only one year, can therefore be very misleading.

Using standardized approaches is valuable for several reasons. Large data sets from nationwide sources promote collaboration. Indeed, some demographic questions are answerable only over large scales by the joint analysis of several data sets. For example, variation in population parameters across a species’ range, regional patterns of productivity or survival, or other regional comparisons are possible only with large data sets such as these. Robinson et al. (1995) provide an excellent example of what can be accomplished by combining data sets from several locations (Robinson and Morse discuss linking local and regional demographic studies in this section). In this case, the authors verified what had long been suspected, that reproductive success of forest interior bird species is inversely related to the degree of fragmentation in the surrounding landscape. However, this result was based on data from the midwestern USA only. It remains to be seen if the rest of us can work collaboratively in other regions of North America to obtain the information we so desperately need about these birds.