Breeding distribution (BD)

Generally, species with the widest breeding distributions are the least vulnerable to deleterious environmental changes and catastrophic events. BD is based on the proportion of North America that is covered by a species' breeding range. We have defined North America to include the main body of the continent (excluding Greenland) through Panama and the islands of the Caribbean, comprising an area of 22,059,680 km² (National Geographic Society 1993). A species inhabiting a breeding area that represents 2.5% or less of North America is assigned a BD score of 5 because species with such small ranges could be significantly and negatively affected by a catastrophic event, such as a hurricane (Table 1). At the other extreme, species inhabiting breeding ranges covering 20% or more of North America should be at relatively low risk from stochastic and other negative environmental events and are assigned a score of 1. Species with ranges between 2.5% and 20% were ranked on a graded scale (Table 1). The primary sources of information for bird distribution for BD and ND (see below) were maps from two well-known field guides (National Geographic Society 1987, Howell and Webb 1995).

Nonbreeding distribution (ND)

Nonbreeding Distribution reflects the proportion of North America or equivalent area that is covered by a species' nonbreeding range. It is scored in the same manner and using the same scale as BD (Table 1). This parameter is based on the smallest nonbreeding range that a species occupies. Therefore, species that move through severe migratory bottlenecks, such as Red Knots (Calidris canutus) staging at Delaware Bay, receive high scores to reflect the risks associated with such concentration.

Relative abundance (RA)

Relative Abundance measures the abundance of a species in appropriate habitat within its range relative to other bird species. Species that are uncommon are assumed to be at greater risk and receive higher RA scores. Scores for RA are determined using one of two methods. Where BBS data are available, RA values are assigned on the basis of the mean number of birds on the 10 BBS routes on which a species is most abundant (Table 2) using an analysis derived from Price et al. (1995). A less-quantitative approach is necessary for species for which BBS data are unavailable or unreliable, including many arctic, boreal, and tropical species whose breeding ranges largely fall outside of the area sampled by the BBS. In these cases, RA scores are assigned on a qualitative scale of abundance (Table 2) on the basis of published notes and expert opinions, as described in earlier versions of the PIF prioritization process (Carter and Barker 1993, Hunter et al. 1993). All departures from BBS-derived criteria and justifications for departures are documented in the database.

Threats to breeding (TB) and threats to nonbreeding (TN)

These parameters reflect recent and predicted threats that may put a species at risk of decline or extirpation from an area. TB and TN are scored using a standardized scale for evaluating conditions, including what has happened in the past and what is likely to happen in the future, that affect an area's ability to support healthy populations of a bird (Table 3). These conditions include the amount of suitable habitat, other factors that affect the species' survival or reproductive success, including cowbird parasitism, pesticides, and predation, and the capacity of the species to withstand or recover from negative conditions. Present conditions are based on the percent of conditions that existed around 1945 that still exist today. Future conditions are the percent of current conditions anticipated to exist in subsequent decades.

We illustrate this process for two species that inhabit the Central Shortgrass physiographic area. Much of the shortgrass prairie, the preferred habitat of McCown's Longspur (Calcarius mccownii), has been lost or degraded over the last two centuries (Knopf 1994). Perhaps 51 to 75% of the habitat that existed in 1945 has been lost (with 25 to 49% remaining). Using the 25 to 49% remaining column, McCown's Longspur could only score a 3, 4, or 5 (Table 3). Using similar reasoning, we estimate that loss rates of shortgrass habitat will not exceed 50% in the next few decades, but that not all shortgrass prairie that exists today will exist in the future. Therefore, McCown's Longspur in this physiographic area scores a TB of 4. The Horned Lark (Eremophila alpestris) also occurs in the same shortgrass prairie habitat but would be scored differently. The conversion of shortgrass to agricultural lands has not been as detrimental to Horned Larks as to many other species, so perhaps 100% of past conditions remain for this species, leaving possible scores of 1 to 5. Because future conditions are likely to favor Horned Larks (i.e. no future loss), the species scores a 1 for TB. Although the scoring process for TB is heavily based on amount of habitat, it also may incorporate factors related to the ability of an area to support a species. For example, if we concluded for Horned Larks that new threats (e.g. pesticides, plowing of nests) existed that could limit breeding success to less than 25% of what would occur under current conditions, then Horned Larks would receive a TB score of 5.

TN is scored using the same criteria as TB (Table 3). As with Nonbreeding Distribution discussed earlier, TN is named and scored to reflect not only wintering conditions but also threats faced during migration. Species that pass through severe ecological bottlenecks during migration may score higher than they would based on winter conditions alone. Species that reside within an area throughout the year are often assigned the same TB and TN scores, but those that change habitats within that area, or are otherwise more or less vulnerable in winter, could receive different scores for the two parameters.

Population trend (PT) and population trend data quality (PTDQ)

Population trend scores, reflecting the magnitude and direction of population change, are assigned globally as well as locally for each physiographic area. When possible, population trend scores are assigned on the basis of BBS data (Table 4) as analyzed by the BBS laboratory of the United States Geological Survey's Biological Resources Division (J. Sauer pers. comm.). Each PT score is linked with a PTDQ score that assesses the quality of BBS data based on sample size and statistical significance.

A population trend must meet the thresholds of magnitude and reliability associated with PTDQ scores of A, B, or C to warrant either a very high (4 or 5, declining) or a very low (1, increasing) score (Table 4). Species not meeting these minimum requirements are assigned a PT of 3 (trend unknown) and receive PTDQ scores of D, E, or F. A PTDQ score of D is assigned to species for which a large sample size indicates positive or negative nonsignificant trends of at least 1%. This often applies to birds whose local populations fluctuate greatly from year to year, possibly reflecting fluctuating habitat conditions and/or possibly obscuring long-term trends. Species assigned a PTDQ score of E1 or E2 have insufficient sample sizes. Species not detected on the BBS in the area are given a PTDQ of F and usually are locally rare or difficult to detect.

A PT score of 2 reflects two possibilities that are differentiated from each other by the PTDQ score and the magnitude of the trend. A PT of 2 with a PTDQ of A2 or B2 and a trend between ±1% indicates a well-sampled species whose population is essentially stable. A PT of 2 with a PTDQ of C1 or C2 and an increasing trend (at least 1% per year) reflects a species that is probably increasing, but without statistical certainty.

In cases where BBS data are unsatisfactory, other population trend data or the opinions of local experts may be substituted with the scoring mimicking the criteria in Table 4. PT scores not derived from the BBS are assigned a PTDQ score of X, and the origin of the information is noted in the database.

Area importance (AI)

This variable is intended to evaluate how important a particular physiographic area is to the conservation of a given species. AI scores compare the abundance of a species within a physiographic area relative to other areas throughout its range (Table 5). Because these scores are based on an index of relative abundance, they are not influenced by the size of the geographic unit in question. Data for this criterion are derived from the BBS, when they are adequate. The first step is to identify the physiographic area within the bird's range with the maximum mean number of individuals per BBS route. Mean numbers of birds per route for other physiographic areas are compared with the physiographic area maximum. For cases in which BBS data are not available or seem to be misleading, AI can be assigned on the basis of local review by experts knowledgeable about the distribution and abundance of the species, following as closely as possible the rationale behind the quantitative method.

Other data quality and supplemental scores

Although PTDQ is the only supplemental value currently assigned, it may be beneficial to document the quality of information used in developing scores for other variables. In the past, PIF has assigned some data quality scores for TB and TN, but these have been rarely used or have caused confusion. Hunter et al. (1993), following Millsap et al. (1990), described a potentially useful supplemental score termed “Management Needs” that documents the amount of management attention that a species is receiving. This could assist in differentiating among species that are receiving an adequate amount of attention from those with similar prioritization scores that may require additional management effort.

Weighting and sorting

At present, the seven variables in the prioritization process are equally weighted in the database. When the database was initially developed, suggestions were made to weight certain parameters in situations where one factor might be considered more important than another. A particular score may be emphasized or de-emphasized by being multiplied by a factor greater than or less than 1 and used to generate a new total score. We caution against such weighting, however, because the relative contribution of the seven parameters to the overall vulnerability of a species usually is unclear. Rather, we suggest that using parameters as sorting factors may result in conservation priorities that are easier to interpret.

As a first step in sorting priority scores, an AI threshold can be chosen that eliminates a species from the peripheral parts of its range, where conservation efforts would be ineffective. Efforts in a physiographic area generally should be limited to species with an AI score of 2 or higher. A three-tiered sorting of species with an AI score of 4 or 5 in the top rank, 3 or 2 in the second rank, and 1 in the third rank would increase emphasis on species for which an area has the greatest conservation responsibility. When reliable BBS data exist, PT scores provide a useful second sorting, after AI, in that PT ultimately is the parameter that is the object and measure of conservation action. Birds with high scores for PT may be more in need of conservation action than are birds with low scores. An additional sorting on the basis of TB in breeding habitat and TN in wintering or migration habitat will also focus conservation actions where they potentially can do the most good, particularly where local PT is uncertain. In addition, sorting by PTDQ may be useful because species that receive both a high total score and a PTDQ that indicates poor data generally are those that are most in need of attention in developing monitoring programs.

Total scores and independence among parameters

Beissinger et al. (2000) caution against summing the seven rank scores to derive a single value because of the potential for lack of independence among the parameters. Although scores for some parameters can be shown to be correlated, this does not necessarily result from lack of biological independence. For example, a species may have a small breeding distribution (high BD score) but occur in high relative abundance (low RA score), and its populations may or may not be threatened (high or low TB). The seven parameters in the species prioritization process were selected partly because of our ability to score them independently. That in practice some species exhibit correlations among scores suggests to us patterns of conservation need, rather than lack of independence. Species that score high on multiple parameters, and therefore have high total scores, exhibit compounding evidence of vulnerability. These species may have a small total distribution, occur in low relative abundance, show threats to breeding and wintering populations, and exhibit a significantly declining population trend. Other species that show vulnerability only in one or two of these categories would have moderate total scores, the meaning of which is unclear without inspection of the component scores. Very low and very high total scores, however, have easily interpretable meanings in terms of whether some sort of conservation attention is warranted.

Beissinger et al. (2000) suggested a categorical approach that sorts species primarily according to thresholds of scores for Population Trend, Relative Abundance, and threats. In their analysis of the breeding bird species of New York state, categorical rank (e.g. high vs. moderate concern) was highly correlated (r_s = 0.76) with the sum of the seven parameter scores. More important, the same species of greatest conservation concern were clearly identified by both approaches. A rigid categorical approach has a drawback, however, in that the greatest emphasis (weighting) may inadvertently be placed on parameters for which scientific data are unreliable. The combination of approaches used by PIF in its planning process (see below) places species into priority categories based on their total scores as well as on combinations of scores for the component parameters, especially AI and PT. This is essentially a categorical approach in which a species can enter the priority pool by exceeding any one of several thresholds.

Responsibility versus concern

Not all species identified through the prioritization process as high priority are of immediate conservation concern or require immediate management actions in all areas. Rosenberg and Wells (1995, 2000) introduced the concept of “area responsibility” to highlight an area's share in the responsibility for long-term conservation of species, including those that are not currently declining or threatened in that area. Such species may have relatively high total scores, including high AI, BD, or RA, but low to moderate locally derived PT and TB (or TN for wintering birds) scores. These may be species for which the physiographic area should assume conservation responsibility and for which close monitoring is warranted. Area responsibility also is a prominent component of species prioritization efforts in Canada (Dunn et al. 1999). In contrast, species that are most in need of on-the-ground conservation actions in a region receive high scores for local concern (PT and TB) and global priority (very high total scores) but may score high or low for area responsibility (AI). Focusing on high-responsibility species or global-priority species that are of immediate concern is an improvement over many state threatened and endangered species lists, which often contain peripheral populations that are locally rare and declining and that may have little consequence for long-term conservation of the listed species.

PIF bird conservation plans

We believe that the most important application of priority scores to date is their use in the drafting of Bird Conservation Plans for all of the physiographic areas and/or states of the continental United States. From among the breeding avifauna of each physiographic area, a priority species pool is derived using a combination of criteria from the prioritization database. A species may be included in this pool if it is in immediate need of conservation action, has a global high-priority status, or has a large population for which there is high local responsibility. All species in this priority pool are then grouped into habitat-species suites, each consisting of one or more co-occurring species. Within habitat-species suites, specific conservation needs or actions may be suggested by patterns of priority scores. If, for example, there are consistently poor PTDQs for species in a suite, increased monitoring effort may be required in that habitat. Habitats used by one or more species with very high total scores, high PT and TB, and high AI scores may be in need of rapid remedial management or other conservation actions. Conditions in habitats for suites that consist of many birds with high total scores and high AI scores but low PT and TB scores may be adequate at present but in need of long-term maintenance.

The plans will also include quantitative population and habitat objectives for habitat-species suites, suggestions for implementation actions to achieve conservation objectives, and an adequate program to evaluate the success or failure of those actions. The overall intent of the PIF planning process is to guide the efficient expenditure of conservation resources to ensure the long-term health of populations of landbirds throughout the United States, with an understanding that comparable efforts may be required in other countries to provide sufficient wintering and migration habitat for many migratory species.

Watch lists

PIF prioritization scores have been used to generate a national Watch List for the United States (Carter et al. 1996, Pashley 1996) that is based on the sum of global scores for six parameters (AI is omitted because it never receives a global score). The value of a national Watch List is to highlight species that may be in need of conservation attention throughout their range and to attract interest in bird conservation. Such a national list based on summed global scores is less useful for conservation planning at regional or local scales than are physiographic area scores in which AI, PT, and TB are scored locally. Therefore, we urge caution in the use of the national Watch List, or any other compilation of species at an inappropriate geographic scale, as a guide for conservation priorities.