Center For Advanced Spatial Technologies (CAST)

THE ARKANSAS GAP ANALYSIS PROJECT

FINAL REPORT

ANALYSIS BASED ON STEWARDSHIP AND MANAGEMENT STATUS

TABLE OF CONTENTS

1. INTRODUCTION

2. LANDCOVER CLASSIFICATION AND MAPPING

3. PREDICTED ANIMAL DISTRIBUTIONS AND SPECIES RICHNESS

4. LAND STEWARDSHIP

5. ANALYSIS BASED ON STEWARDSHIP AND MANAGEMENT STATUS

5.1. Introduction

5.2. Landcover Analysis

5.2.1. Description of the Landcover Analysis Table

5.2.2. Analyzed by Ownership/Results

5.2.3. Analyzed by Management Status/Results

5.3. Limitations and Discussion

5.4. Predicted Species Distributions Analysis

5.4.1. Description of the Species Distributions Analysis Table

5.4.2. Analyzed by Ownership/Results

5.4.3. Analyzed by Management Status/Results

5.4.4. Analysis of Statewide Endemics

5.4.5. Prioritization of hexagons based on gap species:

5.4.6. Ecoregional Analysis

5.5. Limitations and Discussion

6. CONCLUSIONS AND MANAGEMENT IMPLICATIONS

7. DATA USE AND AVAILABILITY

8. LITERATURE CITED

9. GLOSSARY

10. GLOSSARY OF ACRONYMS

11. APPENDICES AND MAPS

5.1. Introduction

As described in the general introduction to this report (see chapter 1), the primary objective of gap analysis is to provide information on the distribution and status of several elements of biological diversity. This is accomplished by first producing: maps of landcover types (see chapter 2), predicted distributions for selected animal species (see chapter 3), and land stewardship and management status (see chapter 4). Intersecting the land stewardship and management map with the distribution of the elements results in tables that summarize the area and percent of total mapped distribution of each element in different land stewardship and management categories. The data are provided in a format that allows users to carry out inquires about the representation of each in different land stewardship and management categories as appropriate to their own management objectives. This forms the basis of gap analysis' mission to provide land owners and managers with the information

necessary to conduct informed policy development, planning, and management for biodiversity maintenance.

Although gap analysis "seeks to identify habitat types and species not adequately represented in the current network of biodiversity management areas," (GAP Handbook, Version 1, Preface, pg. i) it is unrealistic to create a standard definition of "adequate representation" for either landcover types or individual species (Noss et al. 1995). A practical solution to this problem is to report both percentages and absolute area of each cover type in biodiversity management areas (as described above) and allow the user to determine which types are adequately represented in natural areas. Clearly, opinions will differ among users, but this is an issue of policy, not scientific analysis.

The network of Conservation Data Centers (CDC) and Natural Heritage Programs (NHPs) established cooperatively by The Nature Conservancy and various state agencies maintain detailed data bases on the locations of rare elements of biodiversity. Gap analysis cooperatively uses these data to develop predicted distributions of potentially suitable habitat for these elements, which may be valuable for identifying research needs and preliminary considerations for restoration or reintroduction, but conservation of such elements is best accomplished through the fine-filter approach of the above organizations. It is not the role of gap analysis to duplicate or disseminate Heritage Program or CDC Element Occurrence Records. Users interested in more specific information about the location, status, and ecology of populations of such species are directed to their state Heritage Program or CDC.

Currently, landcover types and terrestrial vertebrates are the primary focus of gap analysis' mapping efforts, however, other components of biodiversity, such as aquatic organisms or selected groups of invertebrates may be incorporated into gap analysis distributional data sets. Where appropriate, gap analysis data may also be analyzed to identify the location of a set of areas in which most or all landcover types or species are predicted to be represented. The use of "complementarity" analysis, that is, an approach that additively identifies a selection of locations that may represent biodiversity rather than "hot spots of species richness" may prove most effective for guiding biodiversity maintenance efforts. Several quantitative techniques have been developed recently that facilitate this process (see Pressey et al. 1993, Williams et al. 1996, Csuti et al., 1997, for details). These areas become candidates for field validation and may be incorporated into a system of areas managed for the long-term maintenance of biological diversity.

5.2. Landcover Analysis

Analysis of landcover to stewardship data was conducted using traditional GIS polygon overlay technique. Prior to analysis, stewardship polygons were dissolved by code to reduce redundant polygons that for example, distinguished between privately owned land within and outside National Forest boundaries. Using ARC/INFO INTERSECT, AR-GAP landcover was compared to AR-GAP stewardship (ownership and management). INTERSECT creates an output coverage containing new polygons that describe the coincidence between the two input coverages (e.g., landcover and management). Ownership and management were independently processed against landcover. Attribution from the source data was joined to form the output coverage polygon attribute table (PAT). That PAT was exported to Microsoft Excel to produce the summary tables (Tables 5.1., 5.2.) describing the relationship between landcover and stewardship. Stewardship water was removed from the analysis.

5.2.1. Description of the Landcover Analysis Tables:

Tables 5.1. and 5.2. encapsulate the mapped distribution of landcover by ownership and management status respectively. The tables are organized by landcover class in the first column. Each class occupies two rows (described by the second column as 1. Area in hectares and 2. Percent of row total). Data columns follow for each land ownership and management status class. The last column

Table 5.1. Area (ha) and percentage of AR-GAP landcover classes mapped in ownership categories.

Table 5.1. continued.

corresponds to the row totals for each landcover class.

5.2.2. Analyzed by Ownership/Results:

Table 5.1. prominently displayed the dominance of private/unknown ownership in the proportional distribution of AR-GAP landcover classes. A majority (17 of 31, 55%) of AR-GAP natural landcover classes had more than 90% of their distribution falling in private/unknown ownership. One class, T.1.B.3.a.I (Fagus grandifolia; 54%), had more than 50% of its area mapped in another land ownership category (US National Park Service) besides private ownership.

Only USDA Forest Service ownership contained more than 30% of the mapped distribution of any landcover class. Three classes, T.1.A.9.b.I (Pinus echinata; 36%), T.1.B.2.b.II (Quercus spp. - Pinus

Table 5.1. continued.

echinata - Carya spp.; 31%), and T.2.B.3.a.I (Pinus echinata - Quercus spp.; 31%), met that standard. Another notable distribution was P.1.B.3.c.III (Quercus falcata var. pagodifolia) with 21% of its mapped area in AGFC ownership. Military Reservations captured 14% of the mapped area of T.1.A.9.c.I (Juniperus virginiana). Only two other classes even accrued more than one percent of their area in Military ownership.

AGFC, second largest landowner (in area administered), also expressed the highest diversity of AR-

Table 5.1. continued.

GAP landcover classes with two thirds (66%) of mapped classes occurring under its administration. Intuition would suggest that the smallest landowner would most likely demonstrate the least diversity when compared against the AR-GAP landcover map. In fact, all 52 ha of Nature Conservancy ownership consisted of P.1.B.3.c.V (Quercus nuttallii).

5.2.3. Analyzed by Management Status/Results:

Six landcover classes, T.1.B.3.a.I (Fagus grandifolia; 54%), P.1.B.3.c.I (Quercus lyrata; 10%), P.1.B.3.c.III (Quercus falcata var. pagodifolia; 21%), P.1.B.3.c.IV (Celtis laevigata; 26%), P.1.B.3.d.I (Taxodium distichum; 20%), and P.1.B.3.d.II (Nyssa; 18%), had more than 10% of their mapped distribution is status 1 and 2 combined. Of those classes, only P.1.B.3.d.II (Nyssa; 1.16%) had more than 1% of its distribution in status 1. Overall, three other classes, T.1.B.2.b.IV (Juniperus virginiana; 2.53%), T.1.B.2.b.II (Quercus spp. - Pinus echinata - Carya spp.; 1.98%), and T.1.B.3.a.II (Quercus alba - mixed hardwoods; 1.31%) accumulated more than one percent of their

Table 5.2. Area (ha) and percentage of AR-GAP landcover classes mapped in management categories.

distributions in status 1. The latter two were also the most abundant (Table 2.4.) natural landcover

Table 5.2. continued.

classes in the state. Overall, 25 of 31 AR-GAP natural landcover classes (81%) maintained less than 10% of their mapped distributions in status 1 or 2. Twenty seven AR-GAP natural landcover classes (87%) retained less than 20% of their mapped area in status 1 or 2.

Three landcover classes, T.1.A.9.b.I (Pinus echinata; 36%), T.1.B.2.b.II (Quercus spp. - Pinus echinata - Carya spp.; 31%), and T.2.B.3.a.I (Pinus echinata - Quercus spp.; 31%), contained over 30% of their mapped area in management status 3. Six landcover classes had 100% of their mapped distribution occurring in status 4. Those six classes ranked 28, 30, 31, 32, 34, and 36 in relative land area (Table 2.4.).

5.3. Limitations and Discussion

The low area percentages of classes mapped in status 1 and 2 were not surprising given the distribution of publicly managed lands in the state (Table 4.2.). In fact, with 90% of the state falling into private/unknown management (status 4), it seemed intuitive that 30 of 31 (97%) AR-GAP

natural landcover classes would have less than 50% of their mapped distribution in status 1 or 2. More surprising was that the one AR-GAP natural landcover class (T.1.B.3.a.I (Fagus grandifolia; 54%)) with over 50% of its mapped area in status 1 or 2 was dispersed in only 2 polygons (361 ha) (Table 2.4.). Since much of the ground truth (CISC and TMI, see chapter 2) for classification of AR-GAP natural landcover came from publicly managed area, the distribution of those landcover classes may be potentially biased toward those management areas.

5.4. Predicted Species Distributions Analysis

5.4.1. Description of the Species Distributions Analysis Table:

Summary tables were generated for species treated by AR-GAP, a small section of which is reproduced in Tables 5.3-5.10. For most of these tables, the first three fields are identical.

Table 5.3. Species Statewide Distribution (Appendix 11.13.).

The first field is a unique number assigned to each species; sorting by this number will place species in taxonomic order (birds, amphibians, reptiles and mammals). The second field pertains to minimum area requirement (see Table 5.12 for explanation of values), and the third field is common name of species. For many species, the predictive model is a combination of terrestrial vegetation types and water features. Total statewide habitat predicted areas are shown in Table 5.3, the fourth field is hectares of terrestrial predicted habitat, fifth is hectares of water predicted habitat, and sixth field is summed predicted habitat. For example, Table 5.3 shows Ovenbirds are Type 2, have been predicted to occur in 4,492,140 ha of terrestrial habitat, zero hectares of water habitat with a sum total of 4,492,140 ha of habitat in Arkansas.

5.4.2. Analyzed by Ownership/Results:

Each completed terrestrial vertebrate predictive model was compared to 16 ownership categories (Table 5.4 for terrestrial habitats, Table 5.5 for water habitats). Fields three through 18 identify total hectares of predicted terrestrial habitat that each particular agency managed. For instance, USDA Forest Service owned 929,907 ha of terrestrial Ovenbird habitat and 3,356,713 ha of terrestrial Ovenbird habitat are privately held (Table 5.4). Fields 20 through 37 identify total hectares of predicted water habitat that each particular agency managed. For instance, USDA Forest Service owned 50,104 ha of Louisiana Waterthrush water habitat, while an additional 320,788 ha of Louisiana Waterthrush water habitat were privately held (5.5). Ovenbirds do not use water habitats and therefore their amount of water habitat is zero.

Table 5.4. Species/Ownership Status _ Terrestrial (Appendix 11.14.)

Table 5.5. Species/Ownership Status - Water. (Appendix 11.14.)

5.4.3. Analyzed by Management Status/Results:

Each completed terrestrial vertebrate predictive model was compared to four management categories (Table 5.6 for terrestrial habitats, Table 5.7 for water habitats). The fourth field is hectares of terrestrial habitat predicted to occur in the highest level of protection (Status 1), fifth is hectares of terrestrial habitat predicted to occur in the second highest level of protection (Status 2), sixth is hectares of terrestrial habitat predicted to occur in the third highest level of protection (Status 3), and seventh is hectares of terrestrial habitat predicted to occur in water areas. For instance, 57,265 ha of terrestrial Ovenbird habitat were Status 1, 100,325 ha were Status 2, 945,199 ha were Status 3. Status 4 lands can be obtained by subtraction and 7,877 ha were water. This last category was composed of areas in which AR-GAP vegetation map identifies was terrestrial habitat but which AR-GAP ownership map labeled as water. Two different sources of data were used to define water in these map layers. The result was that some areas labeled as terrestrial vegetation in the vegetation map were labeled water in the ownership map (and visa versa). Table 5.7 is identical in format to Table 5.6.

Combining above data, it was possible to show percentage distribution of each species in Status 1 and 2 (Table 5.8 for terrestrial habitats, Table 5.9 for water habitats, Table 5.10 for total percentage of habitat). The fourth field of Table 5.8 is hectares of terrestrial habitat in Status 1 and 2, fifth is hectares of terrestrial habitat, and sixth is percentage of terrestrial habitat in Status 1 and 2. For example, 157,591 ha of terrestrial Ovenbird habitat were in Status 1 and 2 and 4,492,140 ha of terrestrial Ovenbird habitat were predicted to occur in Arkansas; therefore 4% of Ovenbird habitat was included Status 1 and 2. The format for Table 5.9 is identical to Table 5.8. The fourth field of Table 5.10 shows total percentage of habitat (terrestrial and water) that is in Status 1 and 2. The fifth, sixth, and seventh fields are markers for species that have less than 10%, 20% and 50% of their habitats protected (respectively). For instance, while only 4% of available Ovenbird habitat was found within Status 1 and 2, 15% of Swainson's Warbler habitat was in Status 1 and 2. Under 10% of Arkansas's area was owned by federal or state agencies. Given this value, it was understandable that only 49 species (15% of AR-GAP species) had 10% or more of their predicted distribution in

Status 1 and 2 and only Rufus-crowned Sparrows had more than 20% of it's predicted distribution in Status 1 and 2 (Table 5.11). Contrastingly, 68 species had less than 1% of their predicted habitat within Status 1 and 2. Superficially, these statistics may seem to point towards the conclusion that much of Arkansas's biodiversity is underprotected however, information found in Tables 5.3-5.10 must be put into proper ecological context. Arkansas's state boundary is a political, not natural boundary and many of species found in Arkansas range widely over many ecosystems. Only once proper regional or ecosystem analyses have been performed can concrete conclusions on biodiversity status be formulated (see ecoregional analysis).

Table 5.6. Species/Management Status _ Terrestrial. (Appendix 11.15.)

Table 5.7. Species/Management Status _ Water. (Appendix 11.15.)

Table 5.8. Percentage of terrestrial habitat in category 1 or 2. (Appendix 11.16.)

Table 5.9. Percentage of water habitat in category 1 or 2. (Appendix 11.16.)

Table 5.10. Percentage of total habitat in category 1 or 2. (Appendix 11.16.)

5.4.4. Analysis of Statewide Endemics:

Only two terrestrial vertebrate species treated by AR-GAP are endemic to Arkansas, Caddo Mountain salamander (Plethodon caddoensis) and Fourch Mountain salamander (Plethodon fourchensis). Caddo Mountain salamanders are restricted to Caddo mountains of western Arkansas and Fourch Mountain salamanders to Fourche and Irons Fork mountains of western Arkansas. Caddo Mountain salamanders prefer cool, wet habitats (such as north facing slopes) and Fourch Mountain salamanders inhabit a wide variety of forested habitats. While neither species is endangered or threatened, their status is currently undergoing review (Stan Trauth, pers. com). Both species have sizable amounts of predicted habitat within Arkansas (726,295 ha for Caddo Mountain salamander, 661,261 ha for Fourch Mountain salamander), however both have only 2 % of their predicted habitat within Status 1 and 2. Due to difficulties in spatially predicting where these species may be present (as noted in introduction), specific management recommendations should await review of each of these species status.

5.4.5. Prioritization of hexagons based on gap species:

The ultimate goal of AR-GAP is to identify species and areas that are unprotected or inadequately protected by existing networks of nature reserves. In order to accomplish this, every species included in AR-GAP was assigned a categorical numeral that represents minimum area requirement (MAR) for a viable population (Appendix 11.10.). While this concept of MAR is controversial, guidelines included within BRD-GAP are quite flexible (Table 5.12). A species is considered adequately protected if amount of predicted habitat within Status 1 exceeds its MAR. A preliminary analysis of spatial distributions of inadequately protected species (Appendix 11.17.) used Environmental Monitoring Assessment Program (EMAP) 635 km² United States hexagon tessellation (White et

Table 5.11. Species that have 10% or more of their habitat in category 1 or 2.

al. 1992) as units of analysis. A list of "gap species" for 260 Arkansas hexagons was generated and subjected to a greedy heuristic algorithm. This algorithm selects a hexagon(s) with the most "gap species" first, then a hexagon(s) with the most species not already represented by the first choice, and so on until all "gap species" are included. There are four families of hexagons that all equally protect Arkansas "gap species " (Figure 5.1.).

Table 5.12. Estimated Area Requirements for Viable Populations (from Gap Handbook).

5.4.6. Ecoregional Analysis:

In order to place data summarized by Tables 5.3-5.11 into proper ecological context, state analyses should ultimately be replaced by ecoregional analyses. Because Arkansas is the first Midwestern state to complete BRD-GAP, only a limited ecoregional analysis can be performed. Ecoregional boundaries defined by USDA Forest Service (Bailey 1980) (Figure 5.2.) were used and results are summarized in Table 5.13. It is hypothesized that gap species that do not have significant proportions of their area within Arkansas will be better managed and served by other states that contain more habitat. Ecoregions with a significant proportion of their area within Arkansas include, Boston Mountains, Arkansas Valley, and Ouachita Mountains. Boston Mountain section had 306 species (95% of AR-GAP species) of which 103 were "gap species" (90%). Arkansas Valley section had 305 species (95% of AR-GAP species) of which 97 were "gap species" (85%) and Ouachita Mountains had 295 species (91%) of which 95 were "gap species" (83%). Based on these summary statistics and previous analysis it would seem conservation efforts in the Boston Mountain would be most beneficial to under-protected biodiversity of Arkansas.

Table 5.13. Ecoregions (defined by Bailey, 1980) of Arkansas, their total extent and percentage found in Arkansas..

First choice

Third choice

Fifth choice

Second choice

Sixth choice

Fourth choice

Figure 5.1. The four unique families of hexagons needed to represent the 114 "gap species" in Arkansas. Hexagons of the same color are equivalent in species composition.

5.5. Limitations and Discussion

Results of this analysis must be viewed with caution for several reasons. A close examination of "gap species" revealed two problems with AR-GAP methodology. Many "gap species" inhabit water areas and some can be classified as common. AR-GAP identified these species as underprotected because water was a separate category in management maps. If a species was predicted to occur in

Figure 5.2. Ecoregional boundaries (Bailey 1980)

water areas, those habitats will never fall within the highest protection status (i.e. Status 1) and therefore species that exclusively use water habitats will by definition be "gap species". Interestingly, a similar problem occurred with species that inhabit urban areas. Urban areas were not protected, thus by definition exclusively urban species are "gap species"! Species primarily using either water or urban habitats are classed underprotected because their habitats are outside the network of natural reserves. A second problem with AR-GAP methodology was that water layers used in management status maps do not exactly coincide with water categories found in vegetation maps. This became evident in the analysis of management of individual species. Certain species known not to occur in water habitats have numerous hectares within the water management category.

Despite limitations outlined here, it is clear that research and associated products produced by AR-GAP will be valuable tools in aiding conservationists, public land managers, and industry representatives operating in Arkansas. While much research remains, especially in identifying and correcting weaknesses in methodologies, AR-GAP products can form a new framework discussion by all parties interested in biodiversity.

Central Till Plains, Beech-Maple Section

Mississippi

Alluvial

Basin

Section

Boston Mountains Section

Arkansas Valley Section

Ouachita Mountains Section

Mid Coastal Plains,

Western Section