IV. RECENT IMPORTANT RESULTS FROM THE MAPS PROGRAM

For the past nine years, IBP has been publishing monitoring results from MAPS (DeSante 1992, DeSante and Burton 1994, DeSante et al. 1993, 1996, 1998, 2000). These papers have documented pronounced annual variation in regional productivity indices as well as the pattern that increases or decreases in productivity in a given year are typically followed by respective increases or decreases in population size the following year (DeSante et al. 1996, 1998). More recently, MAPS data have yielded interesting research and management related results (Nott and DeSante in press, Nott et al. in press). Several of the more important of these are described below.

A. Patterns of productivity as a function of nest location and migration strategy - DeSante (2000) described patterns of productivity indices at two spatial scales: all of eastern North America and the Sierra Nevada physiographic stratum. Productivity indices for species groups at both spatial scales varied as a function of nest location (in descending order: cavity, ground, open-cup tree, and open-cup shrub nesters) and migration strategy (again in descending order: permanent residents, temperate-wintering migrants, and Neotropical-wintering migrants). These patterns agree with those found by direct nest monitoring and those predicted from theoretical considerations, are robust with respect to time and space, and thus apparently reflect real population processes at multiple spatial scales.

B. The development and utilization of transient models in MAPS mark-recapture analyses - Not all individual adult birds captured as part of MAPS protocol are resident in the study area during the breeding season. Some, such as floaters, failed breeders, and post-breeding dispersing individuals, may be merely passing through the study area and have essentially zero probability of being recaptured there at a later date. The inclusion of such transient individuals in standard mark-recapture analyses violates the basic assumption that all individuals have an equal probability of recapture and causes substantial underestimation of survival-rates. This problem can be overcome by use of a transient model (Pradel et al. 1997, Nott and DeSante in press) that utilizes both between- and within-year information to estimate the proportion of residents among newly captured adults and the survival rate of those resident adults.

Figure 1 shows that survival rate estimates in the range of 0.4 to 0.5 obtained for target species from the standard CJS non-transient model were increased by 12% to 20% through the use of the transient model. Moreover, the precision of the survival rate estimates from the transient model averaged 7.5% higher than the precision of the estimates obtained from the standard CJS non-transient model (Nott and DeSante in press). These transient models are now being employed in all mark-recapture analyses of MAPS data. Nevertheless, survival rate estimates from MAPS and virtually all mark-recapture experiments on landbirds, including estimates obtained from use of the transient model, are confounded by emigration of breeding individuals and, therefore, are actually estimates of apparent survival.

Figure 1. Relationship between 1992-1998 MAPS continent-wide, time-constant annual adult survival rates from use of the within- and between-year transient model (TMSURVIV) versus use of the standard Cormack-Jolly-Seber (CJS) non-transient model for 89 species. Adapted from Nott and DeSante in press.

C. Relationships between adult survival rate estimates from MAPS and body mass and migration strategy - Although previous researchers have made broad inferences about variation in avian survivorship, they generally have done so by comparing survival rates of two or more populations of a single species (e.g., Greenberg 1980) or by aggregating multi-species data from many disparate sources (e.g., Martin 1995). The latter studies have been hampered by the fact that the survivorship values from different studies were derived from many different field methods and analytical models, each of which has its own unique biases. In contrast, survival rate estimates from MAPS are derived from modified Cormack-Jolly-Seber mark-recapture analyses that include a between- and within-year transient model and are applied to continent-wide data generated by a standardized mark-recapture methodology. As a result, ecological and geographical correlates of adult survival rates can be examined with much greater rigor than ever before.
Figure 2 shows time-constant 1992-1998 annual adult survival rates plotted against the natural logarithm of mean body mass (Dunning 1992, Sibley 2000) for 89 target species and for three groupings of these species classified according to migration strategy (permanent residents; temperate-wintering migrants; Neotropical-wintering migrants).

Figure 2. Relationships between time-constant annual adult survival rates from 1992-98 continent-wide MAPS data and the logarithm of the mean body mass for each of three migratory-strategy species groups (permanent residents, temperate-wintering migrants, and Neotropical-wintering migrants) and for all species. IBP unpublished data.

Positive linear relationships were found between adult survival rates and ln (body mass) for each species group and were significant (P<0.05) for all groups except permanent residents. An analysis of co-variance (ANCOVA), which took body mass into consideration, showed significant (P=0.01) variation in annual adult survival rates among the three migration-strategy species groups, with both permanent residents and Neotropical-wintering migrants having higher survivorship than temperate-wintering migrants. Interestingly, the species group with the lowest average survival rate, temperate-wintering migrants, also had the steepest slope for its survival rate versus body mass relationship, suggesting that the low survival rates for species in this group were especially pronounced among species with small body mass. This may suggest that species with small body mass are better off either by migrating to tropical latitudes where overwintering climates are predictably benign, or by adapting to predictably harsh climatic conditions and foregoing migration. The poorest strategy (at least as regards adult survivorship) may to be that of migrating to areas where overwintering climate may sometimes be unpredictably harsh, such that costs of migration are always incurred without always reaping the benefits.

D. Measures of productivity and survival from MAPS are consistent with observed population trends - DeSante (1995) showed that reproductive indices based on the ratio of young to adult captures can provide unbiased estimators of actual productivity if the capture probabilities of young and adult birds are equal. This is unlikely to be the case, however, because the young captured by the MAPS protocol are primarily juveniles dispersing from the surrounding landscape, while the numbers of dispersing adults are inflated by captures of the breeding adults that are resident at the station during much of the MAPS season (DeSante 1995). Thus we might expect MAPS reproductive indices to underestimate actual productivity.

Considerable evidence is accumulating, however, to indicate that measures of productivity and survival from MAPS are generally capable of producing modeled population growth rates for multiple species that correlate with observed population trends for those species (DeSante et al. 1999). Moreover, such relationships have been demonstrated at multiple spatial scales, ranging from the smaller scale of a single national forest, national park or military installation, through the larger scale of groups of national forests or military installations within different geographic areas, and finally to the largest scale of the entire continent. These demonstrations indicate that although MAPS productivity indices may indeed be biased low, the biases remain relatively consistent over time and space and among various species, including those with widely different nest locations and migration strategies.

An example of such a relationship for multiple species on a single national forest is shown in Figure 3. Here we see that trends in adult captures for eight target species were significantly positively related to modeled population changes obtained from data pooled from six MAPS stations operated from 1992 through 1995 on Wenatchee National Forest (DeSante et al. 1999). Similar relationships have been obtained for a number of other national forests and parks including Flathead, Umatilla, Willamette, and Siuslaw National Forests and Denali, Yosemite, and Shenandoah National Parks (DeSante et al. 1999).

Rsq=0.569, P<0.025
Figure 3. Relationship between trends in adult captures and modeled population changes calculated from reproductive indices and survival estimates from 1992-1995 MAPS data for eight species on Wenatchee National Forest. Trends in adult captures were weighted by the reciprocal of their standard errors and the size of each point reflects the relative weight of each species. From DeSante et al. 1999.

E. MAPS productivity indices and survival rate estimates can be used to identify the proximate demographic cause(s) of population decline - DeSante et al. (2001b) recently described and evaluated a technique for identifying the proximate demographic cause(s) of population change. The approach involves modeling spatial variation in vital rates (productivity and survivorship) both as a function and not as a function of spatial variation in population trends, and using Akaike's Information Criteria (AIC) to select the appropriate (area-dependent or area-independent) model (Burnham and Anderson 1992).

We conducted these analyses at two spatial scales. At the larger scale, we examined 1992-1998 BBS and MAPS data for Gray Catbird. We modeled productivity and survival rates from MAPS stations located in BBS physiographic strata where catbirds were significantly (P<0.01) increasing, as well as strata where they were significantly decreasing. We found that catbird productivity was best modeled as independent of area, while adult survival rates for catbirds were best modeled as area dependent. Moreover, differences in adult survival rates were of the magnitude needed to cause the observed differences in population trends. We concluded that low adult survival rate, rather than low productivity, was the proximate demographic cause of population decline for Gray Catbirds in the physiographic strata where they were declining.

At the smaller scale, we examined six years (1994-1999) of MAPS data from stations on military installations in both the western and eastern Midwest. We conducted analyses on five target species that showed significant negative or positive trends in adult captures on installations in either the western or eastern Midwest, and trends with the opposite sign on installations in the other area. For all five species, we found that low productivity on the installations where the species was declining was a cause of population decline. Low adult survival was an additional cause of decline for Gray Catbird and Yellow-breasted Chat. These are important results because they confirm that MAPS data can be used to identify the vital rate(s) responsible for population declines and, thus, the vital rate(s) toward which management actions should be directed.

F. MAPS productivity indices, coupled with landscape-level habitat data, can be used to identify management strategies for reversing population declines - A critical management goal of MAPS is to identify management actions and conservation strategies to reverse population declines by quantifying relationships between reproductive indices and landscape-level habitat characteristics (Askins and Philbrick 1987). Ideally, habitat variables should be measured in the landscape surrounding the station that includes the area from within which the dispersing juveniles captured by MAPS protocol have originated. The size of this area undoubtedly varies from species to species, and possibly varies geographically and among habitats for a given species. Although the size of this area is unknown for virtually all species, radio telemetry data demonstrate that dispersing juvenile and post-breeding adult Wood Thrushes generally disperse less than four km from their nests and often to edge locations that have dense shrub cover and an abundance of fruit (Anders 1996, Anders et al. 1997).

Using funding supplied by the DoD Legacy Resources Management Program, we have begun to investigate relationships between bird captures and landscape characteristics within four-km-radius areas surrounding MAPS stations on military installations. For example, for each of the nine most common target species on Jefferson Proving Ground, Indiana, we established logarithmic relationships between bird captures and various landscape metrics based upon 30-m resolution Multi-Resolution Land Characterization (MRLC) Consortium remote-sensed data (Bara 1994). Then, from these fitted logarithmic curves, we calculated the relationships between reproductive indices (young/adult) and landscape metrics (Fig. 4).

Figure 4. (A) Numbers of individual adult (o) and young (x) birds of four forest interior species captured per 3600 net-hours at six MAPS stations operated during 1994-1999 on Jefferson Proving Ground, Indiana, as a function of mean forest patch size in the 4-km-radius area surrounding each station. (B) Relationship between reproductive index (young/adult) and mean forest patch size at Jefferson Proving Ground for these four species (obtained from the fitted curves in A). IBP unpublished data.

Figure 4a shows these results for four target species (Ovenbird, Acadian Flycatcher, Wood Thrush, Kentucky Warbler) as a function of mean forest patch size, the single landscape metric that showed the strongest correlation with number of adults captured for each of the four species. These four species are generally considered to be forest interior species and, for each of them, numbers of both adults and young were significantly (P<0.05) positively correlated with mean forest patch size at the six stations. Even more interesting were the relationships between reproductive index and mean forest patch size (Fig. 4b). For each species, a threshold patch size (the patch size associated with the 45 degree inflection point of the relationship) was found, below which reproductive indices increased rapidly with increasing forest patch size and above which increases in forest patch size produced relatively small increases in reproductive indices.

Both the threshold patch size and the sharpness of the threshold varied among species. Of the four, the reproductive index for Ovenbird was the most sensitive to mean forest patch size; that is, its threshold patch size was highest (about 30 ha) and its threshold was least sharp of the four species. This is in accordance with recent literature on Ovenbirds (Porneluzi et al. 1993, Burke and Nol 1998). Acadian Flycatcher showed the least sensitive response of reproductive index to mean forest patch size; its threshold patch size was lowest and its threshold was sharpest with very little increase above 20 ha. Reproductive indices for Wood Thrush and Kentucky Warbler showed intermediate sensitivity to mean forest patch size. These tolerances to forest fragmentation are also similar to those previously reported (Gibbs and Faaborg 1990, Robinson et al. 1995), but here, for the first time, we are able to relate the vital rate actually causing the area sensitivity to habitat conditions in the local landscape.

These results have profound management implications. When these types of analyses become fully developed, it should be possible to calculate, from MAPS survivorship and population trend data, the critical values of productivity needed to reverse population declines and produce positive population trends. It should then be possible to predict the values of various landscape metrics that would be needed to produce such reproductive indices. The development of such landscape-level management strategies is one of the ultimate goals of the MAPS Program.

III: Goals and Objectives   V:5-year Plan and 3-year Objectives