|Development of Management
Objectives for Breeding Birds in the Mississippi Alluvial Valley
Allan J. Mueller1, Daniel J. Twedt2, and Charles R. Loesch3
Bird Conservation Plans (BCPs) for each physiographic area will make critical contributions to the national Partners In Flight (PIF) conservation plan. To be most useful, these regional BCPs should promote on-the-ground conservation actions by developing quantified, site-specific habitat and population objectives. As a model for the PIF planning process, the North American Waterfowl Management Plan has had great success in putting conservation on the ground through the preparation of detailed regional plans with objectives that focus conservation efforts, provide funding justifications, and provide perspective on the "big picture."
Frequently we do not have firm scientific information to quantify conservation issues. However, if we wait for all of the information that we think we require, the time for effective conservation action may pass. We therefore must move forward and make conservation recommendations as soon as possible, based on the best information currently available. As new information becomes known, recommendations can be modified. This iterative method of operation, called adaptive management, is becoming widely accepted in the conservation/scientific community (Franklin 1995, Kirchhoff et al. 1995, Meffe and Viederman 1995, Petit et al. 1995). This paper presents a general model for setting detailed, regional bird conservation objectives, and describes the application of this model, using the best available information in the Mississippi Alluvial Valley (MAV).
Our generalized model for setting regional bird habitat and population objectives consists of a six-step process (Table 1). The issues addressed in this model should be covered in all bird conservation planning efforts, although the sequence of steps and the emphasis on each one will vary among local situations.
Table 1. A model process for setting bird conservation goals
This model shares the philosophy of and is compatible with the processes described by Petit et al. 1995
Step 1. Establish species priorities
In an ecosystem or landscape approach to planning we often are confronted with trying to meet the conservation needs of many bird (and other) species with widely varying ecological requirements. The conservation needs of some species, however, will be greater than others. The PIF prioritization process (Hunter et al. 1993, Carter et al. this volume) can be modified to fit any situation, and will help focus the jumble of apparently conflicting conservation needs.
Step 2. Establish habitat priorities
Species priorities should help to establish habitat priorities. Depending on the location and prioritization scheme, habitat priorities can emphasize breeding, wintering, or migration stopover habitat.
Step 3. Identify habitat requirements to maintain individual populations of priority species groups in priority habitat(s)
Habitat requirements of priority species must be identified explicitly to effectively direct the implementation of conservation actions. (This is the first point at which we face the inadequacy of our information base.) First, the habitat needs of each high-priority species should be defined and quantified. That is, the habitat area sufficient to support and maintain a population, however it is definedviable, source, etc.of a species must be quantified. Then, the needs of all priority species occurring in a habitat type can be considered collectively. Species requiring similar conditions can be grouped into suites; habitat requirements for each suite should be based on the needs of the single most demanding species in the suite.
Step 4. Determine the location and extent of existing habitat suitable for meeting the habitat requirements of individual populations of priority species groups
Knowledge of the current distribution, configuration, condition, and extent of key habitat types is required to set realistic habitat objectives. A Geographic Information System (GIS) or some comparable database is essential in this assessment. Although the expense of assembling a GIS specifically dedicated to PIF planning may be prohibitive, GIS is a widely used tool. For example, most states and major universities operate a GIS and probably have land use/cover data for at least part of any given planning area. If GIS is unavailable, other databases that are less site specific, such as river basin studies and forest inventories, can provide much useful information on the habitat composition of a given physiographic area. However, even when sophisticated spatial imagery is used, assessing the many habitat characteristics that determine the quality of an area for priority species usually requires on-the-ground bird inventory work to verify estimates of habitat extent and condition.
Step 5. Set site-specific habitat objectives
Having defined habitat requirements for priority species, and having identified the location and extent of existing habitat that is suitable for meeting those requirements, the next step is to determine whether the existing habitat is adequate to provide long-term support for secure bird populations. If the current situation is satisfactory, then habitat objectives should be framed in terms of maintaining existing conditions. If the situation is unsatisfactory, then objectives should recommend acquisition or restoration of habitat or changes in management of existing, non-suitable habitat. These recommendations can, at least at first, be opportunistic. That is, they can build on existing efforts that may not specifically be dedicated to birds, or they can build on cooperative arrangements that benefit birds but are not prohibitively expensive to partners. Objectives should be ambitious, but realistic. Site-specific objectives have a much better chance of being implemented than general recommendations for a region. Local knowledge of conservation opportunities should be used to help set site-specific objectives.
Step 6. Set Meta-population Goals
Ideally we would set overall population goals before we establish habitat objectives. We would know how many individuals (i.e., populations) of a species are needed for a secure population (i.e., meta-population) to assure the long-term stability of the species. Unfortunately, this information does not exist for most species addressed here. In addition, unlike conservation models that start with defined population goals (e.g., waterfowl), this model is being applied to bird species that do not have adequate population estimates. Consequently, meta-population goals should be set based on a pragmatic evaluation of what is possible, tempered by the best available scientific evaluation of what is needed for long-term species stability. Population goals may be established in terms of the total number or overall density of birds, the number or distribution of populations constituting the meta-population, source-sink or meta-population dynamics, population trends, or security of existing habitat.
THE MISSISSIPPI ALLUVIAL VALLEY EXAMPLE
Despite the radical habitat changes that have occurred in the past two centuries in the 9.7 million ha MAV (Brown et al. this volume), this physiographic area still retains significant habitat values for wintering waterfowl, breeding forest birds, and other transient and resident wildlife. This example focuses on retaining, restoring, and enhancing those values specifically for forest breeding birds. Our long-term, overall goal is to establish and maintain source populations of all breeding bird species in the MAV.
Step 1. Establish species priorities
We used the PIF prioritization process (Hunter et al. 1993, Carter et al. this volume) to set breeding bird species priorities in the MAV (Table 2). Although we focused on breeding birds, we recognize that the MAV is important winter habitat for vast numbers of temperate migrants as well as in-transit habitat for long distance migrants. We tentatively assume that conditions sufficient for breeding birds also will be sufficient for these other species; this assumption needs to be tested rigorously. Additionally, some areas not suitable to high-priority breeding birds can be very important for wintering and transient birds. Ultimately, these factors need to be incorporated into the overall BCP for the MAV but are beyond the scope of this paper.
Table 2. Breeding bird species priorities in the Mississippi Alluvial Valley
86 additional species have priority scores of 19 or less
aBLH = Breeds in or requires bottomland hardwood forest as a component of breeding habitat.
Step 2. Establish habitat priorities
Six of the seven MAV species that have breeding season prioritization scores of 24 or more nest in bottomland hardwood forest (Table 2). Based on this and the historical ecosystem structure of the MAV, we selected bottomland hardwood forest as the highest-priority habitat type for breeding bird conservation in this region.
Step 3. Identify habitat requirements to maintain individual populations of priority species groups in priority habitat
Habitat requirements conceptually can be separated into issues of quality and quantity. Qualitative factors such as vegetative structure, plant species composition, successional stage, flood regime, and other microhabitat features affect the ability of bottomland hardwood habitat to support a diversity of breeding bird species (Pashley and Barrow 1992). Given time and even a marginally natural flood regime, we assume that most sites of sufficient size will achieve the internal diversity to support the needs of most birds in this system.
Much of the topography of the lower Mississippi Valley floodplain consists of ridges and swales, with high, dry sites interwoven with low, wet sites. Over recent history, however, agriculture has claimed almost all of the high sites, leaving only the wettest places for forest and wildlife. These wet sites, regardless of the time that has passed since major disturbance, may not provide conditions necessary for some of the highest priority birds in this system, such as Cerulean Warbler and Swainson's Warbler. Therefore, we must ensure that a sufficient number of forest patches are of average wetness or drier. Habitat quantity must be considered with an awareness that the current landscape of the MAV is at least 75 percent deforested (MacDonald et al. 1979), and most remaining forested patches are small and isolated (Rudis 1995). Because the vast majority of this system is unlikely to be reforested, planners must determine the necessary size, configuration, number, distribution, and interconnectivity of remaining forest patches.
To maintain bird populations, a forest patch should be of sufficient size to support source populations of all priority bird species, with little likelihood of extirpation or genetic degradation. Smaller patches will provide adequate habitat for only a subset of priority species. To determine necessary patch sizes, we used two sources of information: (1) empirical studies and (2) mathematically derived theoretical genetically viable populations.
Empirical studies were used primarily for Swallow-tailed Kite (Cely and Sorrow 1990, Meyer and Collopy 1990) and Cerulean Warbler (Hamel 1992a).
To determine forest patch-size requirements for theoretical genetically viable populations of other species we used the formula:
A = (N D) + B, where
A = area of forest patch required to support a source population, N = number of reproductive units (usually breeding pairs) required for a source population, D = breeding density (usually expressed as ha/breeding pair), and B = the area of a 1 km forested buffer around the forest core (forest core = N D).
To determine N, we first considered the work of Soule (1987), who guessed that a population size "in the low thousands" should represent an adequate minimum viable population for vertebrates, although he strongly cautioned that the size should be independently calculated for each species. Thomas (1990) generally concurred with this estimate. We assumed that individuals of a species in one block of habitat in the MAV are not genetically isolated from individuals in other patches. Furthermore, with the exception of the Ivory-billed Woodpecker (which undoubtedly is extinct in the United States), virtually all of the high-priority birds in this system are Neotropical migrants, which show very low natal site fidelity (Sherry and Holmes 1989, Roth and Johnson 1993). This suggests a high likelihood of gene flow among patches. Therefore, retaining populations above the "low thousands" in the entire physiographic area should ensure viability from a genetic perspective. But even though genetic deterioration within blocks does not seem to be a threat if populations in the physiographic area (or whatever planning area is under consideration) are high enough, a target number of birds for each patch is required to ensure a source population. A proposed minimum effective population of 500 breeding adults (Franklin 1980) was adopted by the U. S. Fish and Wildlife Service (1985) as the minimum size for each of several populations in the recovery plan for the Red-cockaded Woodpecker (Picoides borealis). For monogamous species this N constitutes 250 breeding pairs. However, establishing conservation goals at the minimum threshold, based upon a series of unverified assumptions, seems fraught with peril. Therefore, to provide adequate population levels in the face of these uncertainties, we set N at 500 breeding pairs per forest patch.
For the value of D, we used average breeding densities from Breeding Bird Censuses, as summarized for the southeastern United States by Hamel (1992b). We realize, however, that because of differences in habitat quality, birds might not occur in the MAV at densities as high as those reported in the literature. Even under optimal conditions, bird density in bottomland hardwoods is determined by the frequency of occurrence of necessary patchily distributed microhabitat features, e.g., thickets for Swainson's Warblers, cypress brakes for Yellow-throated Warblers (Dendroica dominica), etc. This is another reason for adopting a target of 500 breeding pairs per forest patch; this number both increases the number above a theoretically determined minimum and reflects our assumption that birds may occur at only one-half the densities reported in ideal conditions.
Finally, because the agricultural matrix that dominates the MAV generally is considered hostile to birds breeding within forest patches, we used an adjustment factor (B) to account for this degradation in suitability. Robinson et al. (1995) found that nest predation and parasitism were high even in large forest patches (2,200 ha) in landscapes with a low percentage of forest cover. Working in Illinois and Missouri, Thompson (1994) found that female Brown-headed Cowbirds (Molothrus ater) traveled an average of 1.2 km between feeding and breeding sites. Undesirable edge effects also can extend to mating patterns. Van Horn et al. (1995) found that male Ovenbirds (Seiurus aurocapillus) singing on territories less than 300 meters from the edge of the forest were much more likely to be unpaired than males from the interior of the forest. For planning purposes, we assumed that a 1.0 km forest buffer surrounding an interior forest core will reduce these negative impacts. Only those pairs within the forest core (N D) are assumed to reproduce at a rate sufficient to serve as a source population.
Large forest patches also are required to maintain the density of breeding individuals that facilitates extra-pair mating systems found in many Neotropical migrants (Morton 1989, Wagner 1993, Stutchbury and Morton 1995). We assumed that patches designed to include a core large enough to support a source population within a 1 km buffer also will mitigate for these other issues of area sensitivity. Clearly, all of the assumptions in this process need to be tested. Because the area of a 1 km buffer will vary with the geometric configuration of each forest patch, the area requirements of each forest patch will differ. For planning purposes, until the actual areas of interior forest within each forest patch are determined, doubling the core forest area [(N D) 2] generally will result in forest patch requirements that approximate or exceed a 1 km buffer around the desired interior forest area.
As an example of the completed calculation for one species, breeding Swainson's Warblers occur at a density of one pair per 4.7 ha (Hamel 1992b). If Swainson's Warblers occur over a large area at this density, then 500 pairs would require 2,350 ha. Applying the doubling factor as a surrogate for the 1 km buffer produces a desired forest patch size of 4,700 ha for one source population of this species.
To determine the minimum forest patch size required to support 500 breeding pairs for all MAV forest breeding species, we performed the above calculations for each species (Table 3). Next, we grouped the species into species suites based on their minimum area requirements. We used three forest patch sizes designed to meet the area requirements of three area-sensitive species groups: 4,000 to <8,000 ha, 8,000 to 40,000 ha, and >40,000 ha. A similar technique was used to determine the areal habitat needs of raptors in French Guiana (Thiollay 1989), Golden-cheeked Warblers (Dendroica chrysoparia) in Texas (Pease and Gingerich nd), and grizzly bears (Ursus arctos) in the Yellowstone ecosystem (Shaffer and Samson 1985). Wenny et al. (1993) discussed this process as one technique for determining areal habitat needs. A good deal of uncertainty is inherent in these assumptions and extrapolations. However, Robinson (this volume), working in the hardwood forests of Illinois, recommended greater-than 8,000 ha "macrosites" to maintain regional metapopulations, and Hamel (1992a) recommended 8,000 ha mature forest patches to secure Cerulean Warbler populations. The agreement of these independently derived figures adds confidence to our forest patch objectives.
Table 3. Forest patch size requirements to support 500 breeding pairs within the Mississippi Alluvial Valley.
aBased on Cely and Sorrow's (1990) work, a 40,000 hectare patch of bottomland hardwood forest would support only about 80 pairs of Swallow-tailed Kites. A secure (source) population would realistically have to be based on a regional (southeast US) population.
Step 4. Determine the extent and location of existing habitat suitable for meeting the requirements of individual populations of priority species groups
A GIS allowed an analysis of the current status of forested habitat in the MAV. Using 1992 Landsat thematic mapper images, we located and measured more than 35,000 forest patches 1 ha or larger (Figure 1). The average patch size is less than 40 ha. Fewer than one percent of the forest patches are larger than 4,000 ha, but they account for more than 52% of the total forest area. The GIS helped to identify opportunities in which relatively minor improvements of size or configuration through reforestation could create patches at or above threshold sizes. Maps produced through this process have been invaluable tools in all subsequent phases of bird conservation planning in the MAV.
Figure 1. Existing forest patches in the Mississippi Alluvial Valley circa 1992.
Step 5. Set site-specific habitat objectives
Having determined the areal habitat requirements for source populations of the high-priority species and having measured the amount of existing habitat that can support these populations, we had enough information to identify the specific locations desired for habitat protection/restoration. In addition to habitat requirements and existing forest locations, several other factors, such as flooding frequency and current land use, were used to identify proposed habitat protection/restoration sites. Where possible, restoration sites were centered on existing public land. Where linkages could logically be created, existing forest patches were combined to reach target sizes. For this reason, several existing 4,000 or 8,000 ha patches sometimes were combined into a proposed 40,000 ha patch.
Land use adjacent to existing or proposed forest patches was an important consideration in identifying and locating conservation areas. Adjacent land use can be beneficial, neutral, or hostile to bird survival and reproduction in forest patches. The Mississippi River and other large bodies of water are considered neutral, and the forested uplands on the periphery of the MAV are considered neutral or beneficial. Land uses that support large numbers of Brown-headed Cowbirds and predators are clearly hostile. Grazed levees, which support large populations of cowbirds, are one of the most hostile land uses. Crop lands are generally hostile, but this likely varies with the type of crop.
We identified 101 target forest patches (Table 4, Figure 2), but the number of these sites and their location is not final, and probably never will be. A massive reforestation effort will be necessary to create these patches, and developing them will be opportunity driven. As new opportunities arise and old patch objectives become unattainable, locations of target patches will change.
The current distribution of target patches within the MAV is not even, largely reflecting existing opportunities. For example, more and larger patches exist in southern Louisiana than in northern Mississippi. As a result, the planning team tended to include marginal patches in northern Mississippi more frequently than in areas with adequate numbers of apparently higher quality sites. The most disturbing bias in patch distribution is that a majority of patches are in wetter parts of the MAV, either within the mainline levee systems, or in other areas where permanent or frequent flooding precludes consistent agricultural productivity. A concerted effort is needed to ensure that the range of conditions within the forest patches adequately represents the range of naturally occurring soil and community conditions in the MAV. This ultimately may require more or different forest restoration efforts than currently are contemplated.
Table 4. Distribution of 101 target forest patches in the Mississippi Alluvial Valley.
Figure 2. Proposed bird conservation areas in the Mississippi Alluvial Valley (some areas contain >1 targeted forest patch.
Step 6. Set meta-population goals
Assuming that each target patch truly will support a source population of the target species, does the number of patches in the three size classes represent an acceptable meta-population goal for the high-priority species? We feel cautiously optimistic that it does, with the possible exception of the Swallow-tailed Kite. Eventually a population and habitat viability analysis on the range of this species throughout the Southeast may be necessary to generate more reliable conservation goals. For all other species, we feel that the patch goals we have recommended in the MAV, if achieved, should preclude any local extinctions, and should allow population trends to stabilize (inasmuch as breeding ground conditions affect the survival and success of long distance migrants).
In some ways, however, the issue of sufficiency of population goals at the physiographic area level is not biological in nature, but instead depends upon the future demands of society for populations of birds and other elements of biological diversity. From this perspective, evaluating the sufficiency of these ambitious but realistic goals is difficult. The next phase of planning will involve establishing specific objectives for each of these target forest patches. These objectives will be based upon current size and configuration of forested habitat, ownership, intent of the landowners, flood regimes, and the avifauna. In general, forest habitat on public land, private industrial forests, and in limited partnership hunting clubs is considered secure. However, private landowner involvement also will be essential to achieve conservation objectives, because land acquisition by public and private conservation agencies never will be adequate. Indeed, this planning process is not intended to result in a land acquisition plan, but to serve as a guide to focus reforestation efforts of all kinds.
The model planning process for the MAV provides site-specific habitat objectives within the context of landscape level conservation needs. The process also gives land managers a perspective on how their management decisions blend with the overall conservation needs of the MAV, at least with regard to forest breeding birds. The process should aid local planning and help to direct, justify, and fund conservation projects. Many assumptions were made in setting these objectives, often based on little existing information. Research to test these assumptions is critical. Monitoring and evaluating the implementation of these recommendations also is essential (Twedt et al. this volume). Through adaptive management, objectives will change as research refines our assumptions, or if monitoring indicates that the intended results are not being achieved.
This process was a cooperative planning effort from the very beginning. Gary Myers, Executive Director of the Tennessee Wildlife Resources Agency, was the catalyst that started this effort, and he made many calls to free staff time to work on this project. The outline and structure of the plan was produced by a core team consisting of (in addition to the authors) Charles K. Baxter, U. S. Fish and Wildlife Service (USFWS); Cindy Brown, The Nature Conservancy (TNC), Louisiana; Robert J. Cooper, University of Georgia; Robert P. Ford, Tennessee Conservation League; Paul B. Hamel, USDA Forest Service; Robert Hatcher, Tennessee Wildlife Resources Agency; W. C. (Chuck) Hunter, USFWS; Seth E. Mott, USFWS; David N. Pashley, American Bird Conservancy; Lance Peacock, TNC, Arkansas; Karen Rowe, Arkansas Game and Fish Commission (AGFC); Cathy Shropshire, Mississippi Department of Wildlife, Fisheries and Parks; Mike Staten, Anderson-Tully Company; Mark Swan, TNC, Louisiana; Mark S. Woodrey, Mississippi Department of Wildlife, Fisheries and Parks; and Scott C. Yaich, AGFC. A large group of wildlife professionals contributed to the selection of restoration/protection sites at a meeting in Memphis, Tennessee, and in subsequent reviews. Their input was critical to making the habitat objectives as realistic as possible. As this plan evolves in response to better information and conservation opportunities, involvement by a wide range of biologists, business people, farmers, politicians, and the public will be essential for its successful implementation.
1 U. S. Fish and Wildlife Service
2 USGS Patuxent Wildlife Research Center
3 U. S. Fish and Wildlife Service