2.1. Study Area
The Mississippi Alluvial Valley Bird Conservation Region (
http://nabci-us.org/resources/bird-conservation-regions-map/#bcr26) is a relatively flat, weakly dissected alluvial plain of >10 million ha within 7 states: Illinois, Missouri, Arkansas, Kentucky, Tennessee, Mississippi, and Louisiana (
Figure 1). Topographic and hydrologic differences subdivide this region into 14 physiographic provinces [
9]. In this ecoregion, forest-dwelling birds are of great conservation concern because over two-thirds of the area that was formerly forested has been converted to other land uses.
Average annual precipitation is 114–165 cm. Natural vegetation has been cleared from most of this ecoregion [
1,
2,
10], being primarily converted to agriculture. Historically, extensive flooding dictated vegetative conditions, but levees, dikes, and dams have markedly altered the hydrology of the Mississippi Alluvial Valley [
11]. These hydrological changes have influenced the composition and structure of the remaining forested wetlands [
12,
13,
14].
Forest cover currently comprises approximately 30% of area within the Mississippi Alluvial Valley [
3]. Remaining floodplain forests are dominated by oak-gum-cypress and elm-ash-cottonwood cover types. Co-dominant species within these forest types include oaks [overcup (
Quercus lyrata), willow (
Quercus phellos), Nuttall (
Quercus nuttallii), water (
Quercus nigra), and cherrybark (
Quercus pagodaefolia)] as well as sweetgum (
Liquidambar styraciflua), water hickory (
Carya aquatica), sugarberry (
Celtis laevigata), American elm (
Ulmus americana), bald cypress (
Taxodium distichum), green ash (
Fraxinus pennsylvanica), and others [
1]. Oak-hickory forests occur on isolated upland inclusions (e.g., Crowley’s Ridge) within this floodplain. Co-dominant upland tree species include post (
Quercus stellate), southern red (
Quercus falcata), black (
Quercus velutina), chinkapin (
Quercus muehlenbergii), and white (
Quercus alba) oaks along with mockernut hickory (
Carya tomentosa) and others [
10].
2.2. Data Sources
Boundary: For our analyses, we used the Lower Mississippi Valley Joint Venture’s conservation planning boundary for the Mississippi Alluvial Valley Bird Conservation Region because it well delineates the transition from alluvial floodplain and deltaic lands to upland habitat. We included all upland inclusions that were wholly contained within this boundary (
Figure 1;
http://www.arcgis.com/home/item.html?id=c72185797b564b5995f44e9bc367163e).
Reforestation priority: Reforestation (i.e., afforestation) priorities for bird conservation have been established for restorable lands within the Mississippi Alluvial Valley. These priorities are intended to effectively increase the number of forest patches that harbor >2000 ha of core forest, while concurrently targeting more than 60% forest cover within local (320 km
2) landscapes and restoration of higher elevation bottomland hardwood forests [
6]. We extracted and used the highest (upper 10%) priority restoration zone from this reforestation decision support model (
Figure 1; LMVJV/FBBDSM_2011;
https://gisweb.ducks.org/arcgis/rest/services/LMVJV/FBBDSM_2011/MapServer).
Flood frequency: We used the inundation frequency of lands in the Gulf Coastal Plain and Ouachita Mountains (GCPO), including the Mississippi Alluvial Valley, that was developed from 50 Landsat scenes and 1334 total images depicting inundation extent under varying hydrologic conditions [
15]. Inundation frequency ranged from 0% to 100% (in
Supplementary Materials: GCPO Inundation Frequency Mosaic;
https://www.sciencebase.gov/catalog/item/5617e3c3e4b0cdb063e3fc35).
Conservation estate: We identified lands owned or managed by conservation-oriented entities, either public or private, and lands subjected to perpetual conservation-oriented easements or servitudes from 7 geographic information system (GIS) source files:
Protected Areas Database of the United States 2.0, 2018: From the U.S. Geological Survey Gap Analysis Project, this database included public and non-profit lands and waters. Most were public lands owned in fee title, but the database also contained long-term easements, leases, agreements, and congressional (e.g., Wilderness Area), executive (e.g., National Monument), and administrative (e.g., Area of Critical Environmental Concern) designations as documented in agency management plans (
https://doi.org/10.5066/P955KPLE).
National Conservation Easement Data, 2018: A public–private partnership database of locations for more than 150,000 conservation easements and land trusts throughout the United States (
https://www.conservationeasement.us/).
Wetlands Reserve Program, 2016: Location information for lands under federal conservation easements with the U.S. Department of Agriculture. These conservation easements included the Wetland Reserve Program, Wetland Reserve Enhancement Program, and Wetland Reserve Enhancement Partnership. These data are not publicly accessible.
The Nature Conservancy, Louisiana Lands, 2018: Locations of lands owned or managed by the non-governmental conservation organization, The Nature Conservancy in Louisiana. These data are not publicly accessible.
All GIS raster data were obtained or converted to 30 m (900 m
2) pixel resolution for analyses. Unless otherwise stated, GIS manipulations were accomplished within ArcMap (Version 10.5.1; Environmental Systems Research Institute, Redlands, CA, USA). The above files were merged to create a unified depiction of the current conservation estate within the Mississippi Alluvial Valley (
Figure 2; in
Supplementary Materials:
https://doi.org/10.5066/P90V76SY).
2.3. Forest Patches
Previous planning efforts for the conservation of forest-breeding birds in the Mississippi Alluvial Valley have made the biological assumption that birds occur at higher density, have increased probability of survival, and have greater reproductive success within forest interiors (i.e., core forest) [
6,
16,
17]. To mitigate the presumed detrimental influences associated with forest edges [
18], initial conservation planning in this region used a conservative 1000 m buffer from ‘hostile’ edges [
19]. More recent conservation plans have assumed that a buffer distance of 250 m is enough to mitigate the detrimental effects of hostile edges [
16].
We identified all extant forest patches (core forest plus the buffer), including reforested areas, within the Mississippi Alluvial Valley [
3]. After identifying and including non-hostile habitats, we extracted core-forest areas that were >250 m from a hostile forest edge [
16]. We considered cropland, pasture, grassland, aquaculture, urban, and suburban habitats to be hostile edges because these ecotones with forest tend to promote predator incursions [
20] and greater abundance of the nest parasite, brown-headed cowbird (
Molothrus ater) [
21]. Conversely, we considered shrublands, emergent wetlands, and natural water bodies to be non-hostile habitats, such that forest core habitats extended to the boundary of these non-hostile edges.
Once core forest was identified, we used the ERDAS Imagine (Hexagon Geospatial, Madison, Alabama) raster processing software to clump (i.e., group) and uniquely identify all contiguous areas of core forest, hereafter referred to as ‘core clumps’. Core clumps were separated from other clumps by at least one pixel (900 m2) around the entirety of the clump, such that corner connections (i.e., diagonally connected pixels) retained continuity of the clump. The area (ha) of each forest core clump was then calculated.
In addition to reliance on forest interior habitat, previous conservation planners also assumed that a large area of core forest is needed to ensure occupancy by enough breeding individuals to diminish the likelihood of extirpation of a species from the forest patch and to provide habitat diversity consistent with the needs of priority bird species [
5]. The minimum area of core forest previously recommended was 2000 ha [
6]. Ongoing evaluation of habitat needs for breeding birds in this ecoregion suggests that a 2000 ha area of core forest would support populations with less than 1% likelihood of extirpation over 100 years for 46 out of 56 (82%) breeding species. Therefore, our goal was to emphasize core forest of >2000 ha. We recognized that additional forest restoration adjacent to core clumps <2000 ha could result in core clumps that exceed this threshold area. Therefore, we retained all core clumps ≥1600 ha (80% of 2000 ha). In addition, because reforestation efforts continue to focus restoration within higher priority forest restoration zones, we retained all core clumps (regardless of ha area) that were adjacent to the highest (upper 10%) reforestation priority zones (
Figure 1).
We reestablished the entirety of forest patches for this set of core clumps that were ≥1600 ha or adjacent to high restoration priority zones (
Figure 3), by returning the 250 m non-core forest buffer. Concurrently, we retained only forested habitat by removing water and herbaceous wetland habitat from these forest patches.
2.4. Conservation–Protection
For each forest patch meriting consideration for conservation–protection, as described above, we determined the percentage of the patch that was outside the conservation estate. Thus, forest patches with a value of 100 had no existing conservation–protection and were in greatest need of forest protection. Conversely, those patches with a value of 0 were fully protected and no additional forest protection was warranted. We adjusted the perceived need for conservation–protection of each forest patch, which was initially based solely on percent area not protected, to account for location and hydrology.
2.6. Hydrology
To benefit priority forest-breeding birds, conservation plans previously placed increased emphasis on retention and restoration of bottomland forest sites that are less prone to prolonged flooding [
6,
16]. This emphasis on drier bottomland sites was because these forests had been disproportionately converted to agriculture [
2] and continue to be more suitable for conversion to non-forest use than flood-prone forests. Moreover, bottomland forests with limited flooding tend to support more understory vegetation and are therefore important for ground-nesting silvicolous bird species [
6]. The excessive loss of bottomland forests that are less prone to prolonged flooding may be exacerbated within the conservation estate by the bias of protected areas to be located on less-threatened land that is not easily converted to other uses [
22].
For each of the forest patches deemed to have merit for bird conservation by virtue of having a core forest habitat area >1600 ha or being located adjacent to high priority restoration zones, we calculated their mean flood probability from inundation frequency data [
15]. The resultant mean percent flood frequency was inverted and scaled (0–100) as a coefficient of dryness, such that 100 represented the least flood-prone forest patches and 0 represented the most flood-prone patches. Because we perceived drier forest patches to be of greater conservation value, we granted an increase in need for conservation–protection proportional to forest patch dryness (dryness coefficient x 0.2). As such, the least flood-prone forest patches received up to 20% increase in need for protection, whereas the most flood-prone patches received a negligible increase.