2006 SOUTHEASTERN NATURALIST 5(2):295–310 Assessment of Habitat Values for Indicator Species and Avian Communities in a Riparian Forest Dwight Barry1,5,*, Richard A. Fischer2, Karl W. Hoffman3, Tami Barry3, Earl G. Zimmerman4, and Kenneth L. Dickson3 Abstract - Habitat suitability index (HSI) models, which are meant to analyze the habitat value of an area for a particular species of interest, are commonly used to evaluate species, communities, and habitats for management or impact assessment. However, many assessments are conducted under strict time and money limitations, and this use of HSI may lead to inappropriate interpretations. The purpose of our work was to evaluate the relationships between three HSI models, the presence of their attendant species, and related avian communities at common assessment time scales. We compared the habitat values of a north Texas bottomland forest measured by HSI models for three species commonly used as indicators of this habitat type, Hairy Woodpecker, Pileated Woodpecker, and Barred Owl, with (1) the actual presence of each species, and (2) the presence of a larger forest-dependent avian community. HSI values did not correlate well with either occurrences of the indicator species or communities for any of the three models. Our results suggest that these models are not necessarily effective indicators of actual habitat use by indicator species, and may not be useful in helping managers make decisions that may affect an entire community related by habitat type. Positive assessments of habitat value through HSI values should be appraised carefully before management decisions are initiated, perhaps through outside review and/or a clear discussion of the context. The primary value of HSI use may be the ability to measure changes in the habitats themselves over time, and not exclusively as representations of species or community presence. Introduction Management strategies and assessments on conservation lands are often predicated on the assumption that indicator species and their habitat requirements can be used effectively to represent the current or potential distribution of a wider array of species. If this is the case, there should be a strong correlation between the results of habitat analyses for appropriate indicator species, and the presence of a larger suite of native species, guilds, and/or communities of management interest. We have observed that habitat assessments for indicator species are commonly used as a surrogate for extensive floral and faunal surveys in the applied environmental 1Lewisville Wildlife Management Area, PO Box 310559, Denton, TX 76203. 2US Army Engineer Research and Development Center, Environmental Laboratory, ATTN: CEERD-EE-E, 3909 Halls Ferry Road, Vicksburg, MS 39180. 3Institute of Applied Sciences, University of North Texas, PO Box 310559, Denton, TX 76203. 4Department of Biology, University of North Texas, PO Box 305220, Denton, TX 76203.5Current address - Peninsula College, 1502 East Lauridsen Boulevard, Port Angeles, WA 98362. Corresponding author - dwightb@pcadmin.ctc.edu. 296 Southeastern Naturalist Vol. 5, No. 2 assessment field. In our experience, these assumptions are often taken for granted and have not been extensively tested in contexts specific to management and assessment. Furthermore, due to funding, time, or other administrative constraints, many potential violations in the interpretation and application of these methods can occur, leading to poorly-grounded management decisions. One use of indicator species for management and impact-assessment work is the Habitat Evaluation Procedure (HEP), developed by the US Fish and Wildlife Service (USFWS) to quantify habitat variables important to a given target species or set of target species chosen to represent the larger faunal communities of which they are a part (USFWS 1980a, 1980b, 1981). The target species’ life-history requirements are quantified in a habitat suitability index (HSI) model that is specific to each species and its primary habitat(s), and sometimes specific to a particular region. Theoretical and empirical support for the HSI framework comes from ecological theories related to habitat selection, niche partitioning, and limiting factors. Studies abound in the literature that have evaluated, critiqued, and/or supported the validity of many of these assumptions (e.g., Hall 1988, Harris and Kangas 1988, Hobbs and Hanley 1990, O’Neil and Carey 1986, Roloff and Kernohan 1999). The HSI models use various habitat metrics (e.g., number of trees ≥ 51cm dbh/0.4 ha) to evaluate particular sites for habitat suitability for a given species. Each model provides a numerical index of habitat suitability on a 0.0 to 1.0 scale, based on the assumption that there is a positive relationship between the index and habitat “carrying capacity” (USFWS 1981). The HSI model is then used to predict the habitat suitability of the site for that particular animal. HSI values are combined with the total acreage of a given cover type to calculate habitat units (HUs) within the assessment areas. HSI values from several species’ models may also be averaged for a given cover type. The use of HEP and HSI models is common in habitat and impact assessment for management in the southeast (e.g., Institute of Applied Sciences 1995), including in politically contentious developments in bottomland hardwood (BLH) forests. Unfortunately, HSI models are often used in practice as implicit indicators for entire communities, such as the use of a forest-dependent bird species model to represent all forest-dependent birds in the area. The potential result of this misapplication is that, given a high habitat-value number for a particular habitat type, a manager with little-to-no background in wildlife management could assume that the habitat is good for most or even all of the animals in that habitat. While this is unfortunate, it is also quite common for decision makers to have little background in ecology and sampling methods and no practical contact with the data collectors. This is often the case when sampling and reporting is contracted out of the agency, and, as a result, there can be a significant disconnect between field work and its results and eventual description of management alternatives. 2006 D. Barry et al. 297 Much of the BLH ecosystem in Texas is becoming increasingly fragmented by urban development, and rapid habitat assessments for impact assessment, mitigation, and conservation are commonplace. Bottomland and riparian forests, especially in north Texas, provide food and shelter for an impressive array of bird species (Barry et al. 2000, Conner and Dickson 1997, Hoffman 2001, Kellison and Young 1997). Riparian forests are typically a small part of any landscape, but they are essential habitat for many species of birds. For example, riparian areas in the western United States make up less than one percent of the total landscape, but are used by more species of breeding birds than any other habitat type in North America (Fischer and Fischenich 2000, Knopf et al. 1988). Unfortunately, the many species of breeding birds that inhabit BLH forest interiors, such as Coccyzus americanus L. (Yellow-billed Cuckoo), Seiurus motacilla Vieillot (Louisiana Waterthrush), and Protonotaria citrea Boddaert (Prothonotary Warbler), have been declining over the past several decades as their habitats become increasingly fragmented (Conner and Dickson 1997, Sauer et al. 1999). There is a need to investigate the relationship between indicator habitat values and birds in riparian forest ecosystems. These investigations are critical to determining if indicator species (as represented by their habitat models) are as effective tools for the conservation and management of these forests as managers often assume. We explored the HSI index values of a BLH forest in north Texas and the relationships between these values and the avian community. Specifically, our purpose was to explore whether HSI models work for their intended targets when applied under typical management conditions in the southeast, and whether or not HSI values obtained from indicator-species’ models can reflect the presence of other forest species in a particular field situation. Field Site Description The Ray Roberts Greenbelt Wildlife Management Area comprises nearly 2000 ha near the center of Denton County, TX. Approximately 500 ha of the area contain remnant stands and connecting corridors of BLH along the Elm Fork of the Trinity River (Fig.1). The remaining 1500 ha are regenerating forests, old fields, and savannah scattered among the remnant forest patches. The vegetation of the riparian forest is dominated by Ulmus crassifolia Nutt. (cedar elm), Celtis laevigata Willd. and C. reticulata Torr. (hackberry), and Fraxinus pennsylvanica Marsh. (green ash), with occasional occurrences of Quercus macrocarpa Michx. (bur oak), Q. shumardii Buckl. (Shumard oak), Carya illinoensis (Wangenh.) K. Koch (pecan), and Populus deltoides Bartr. ex Marsh. (eastern cottonwood) (Barry and Kroll 1999, 2003; Barry et al. 2000). This elm-ash-hackberry type is recognized as a mature stage in many southeastern bottomland hardwood forests (Hodges 1997). The understory is a mixture of Smilax rotundifolia L. (common greenbrier), Toxicodendron radicans (L.) Kuntze (poison ivy), Symphoricarpos orbiculatus Moench (coralberry), and Elymus virginicus L. (Virginia wild rye). The surrounding 298 Southeastern Naturalist Vol. 5, No. 2 landscape matrix above the bottomland is predominantly rangeland and agriculture, with a few scattered areas of rural development. Denton County occupies approximately 2450 km2 in north-central Texas, just north of the Dallas/Ft. Worth metroplex. Throughout the county, soil type is the key factor explaining native vegetation distribution (Bailey 1995, USDA 1980). The climate is humid subtropical, with hot summers and mild winters (mean temperatures of 4–10 °C in winter and 27–32 °C in summer), with moderate rainfall of 890 mm per year and periodic drought (Bailey 1995, NWS 2005). Figure 1. The Ray Roberts Greenbelt WMA Forest and sampling locations. 2006 D. Barry et al. 299 Methods Avian sampling We established 62 permanent sampling stations along a transect placed within the middle of the riparian forest, roughly following the channel of the Elm Fork. Sampling stations were placed approximately 250 meters apart to avoid double counting individual birds (Hamel et al. 1996, Ralph et al. 1995). The stations were marked on a map, and then relocated in the field each season using a GPS unit and aerial photographs. To test the value of HSI model results, we conducted point counts of birds four times per year in 1999 and 2000 with sampling periods during winter, spring and fall migrations, and the summer breeding season. Within each sampling season, all points were surveyed within a 2-week period, and surveyors were trained to standardize experience level. We sampled each point once during each of the eight sampling seasons to simulate the routine practice in management and impact assessment of sampling a location once, based on when the funding becomes available or the directive to sample is given. The order of visitation to all points was determined randomly within each season. Surveys were conducted as 10-minute, 50-m radius point counts (Hamel et al. 1996, Ralph et al. 1995). Sampling began at first light and proceeded for approximately four hours, which allowed for potential detection of some nocturnal (such as owls) as well as most diurnal species. The 50-m radius was part of a nested sampling design that recorded approximate distance and direction of all birds seen or heard within three distance rings of 25-m, 50-m, and beyond 50-m (Hamel et al. 1996). Mirroring constraints associated with typical assessments, we did not collect data in a manner that would allow for adjustment of abundance estimates by detection probabilities. For purposes of data analysis, we excluded any birds that were recorded outside of the 50- m radius, as we wished to focus on birds likely to be using specific habitat at each point count station. We classified species detected based on whether or not they depend upon a large amount of forest for their life history requirements (Ehrlich et al. 1988, Robbins et al. 1989, Sauer et al. 1999, Stokes and Stokes 1996, Whitcomb et al. 1981, and the personal experience of the authors). Forest birds (Table 1) were chosen for this analysis due to their sensitivity to fragmentation, their general population declines across the range of BLH, their status as species of management and conservation concern for several governmental and nongovernmental organizations, and their potential to serve as charismatic or “flagship” conservation species in efforts to protect, manage, or restore riparian forests (Conner and Dickson 1997, NABCI 2000). Habitat sampling The main focus of the study was to match habitat values as measured by HSI models at particular locations with any birds that were encountered in these same locations, as these values are sometimes used in managerial 300 Southeastern Naturalist Vol. 5, No. 2 practice as surrogates for entire animal communities. Thus, we evaluated each permanent sampling station using the methods described in the HSI models for Picoides villosus L. (Hairy Woodpecker [HAWO]; Sousa 1987), Dryocopus pileatus L. (Pileated Woodpecker [PIWO]; Schroeder 1983), and Strix varia Barton (Barred Owl [BDOW]; Allen 1987) (Table 2). HSI values were then calculated for each plot using the formulas described in each model. These three species were chosen because they are regional residents that can serve as both indicator and flagship species, HSI models for each are widely available, these models have been used in the region, and because suitable spatial and physical habitat for each species, according to the qualitative descriptions in the models, exists within the Greenbelt. While each of our indicator species have large home ranges and could have been passing through a given sampling site, we assumed that because they were in the site, it was being used for some aspect of their life history. Although Table 1. Forest birds of the Ray Roberts Greenbelt detected during the study. Common name Scientific name Cooper’s Hawk Accipiter cooperii Bonaparte Red-shouldered Hawk Buteo lineatus Gmelin Yellow-billed Cuckoo Coccyzus americanus L. Barred Owl Strix varia Barton Ruby-throated Hummingbird Archilochus colubris L. Red-bellied Woodpecker Melanerpes carolinus L. Yellow-bellied Sapsucker Sphyrapicus varius L. Downy Woodpecker Picoides pubescens L. Hairy Woodpecker Picoides villosus L. Pileated Woodpecker Dryocopus pileatus L. Eastern Wood Pewee Contopus virens L. Great-crested Flycatcher Myiarchus crinitus L. Red-eyed Vireo Vireo olivaceus L. Warbling Vireo Vireo gilvus Vieillot Gray-cheeked Thrush Catharus minimus Lafresnaye Hermit Thrush Catharus guttatus Pallas Brown Creeper Certhia americana Bonaparte Winter Wren Troglodytes troglodytes L. Blue-gray Gnatcatcher Polioptila caerulea L. Ruby-crowned Kinglet Regulus calendula L. Golden-crowned Kinglet Regulus satrapa Lichtenstein Carolina Chickadee Poecile carolinensis Audubon Tufted Titmouse Baeolophus bicolor L. White-breasted Nuthatch Sitta carolinensis Latham Tennessee Warbler Vermivora peregrine Wilson, A. Northern Parula Parula americana L. Yellow-rumped Warbler Dendroica coronata L. Black-and-white Warbler Mniotilta varia L. Prothonotary Warbler Protonotaria citrea Boddaert Ovenbird Seiurus aurocapilla L. Canada Warbler Wilsonia Canadensis L. Summer Tanager Piranga rubra L. Rose-breasted Grosbeak Pheucticus ludovicianus L. 2006 D. Barry et al. 301 there are a wide variety of potential problems associated with this assumption (Garshelis 2000, Hall 1988), in our experience, this is a routine approach to rapid assessment in this region. We categorized the forest class of the sampling stations as forest patch or corridor by delineating the forested area of the Ray Roberts Greenbelt in a GIS through onscreen digitizing of 1-m resolution USGS Denton East and Green Valley digital orthophoto data. Once the forest extent within the study area was delineated, all forest that was within 100 m of non-forest habitat (i.e., edge forest) was removed from the GIS layer, leaving polygons of interior forest. The 100-m distance was chosen because edge effects that cause microclimatic variation or ecological edge effects are minimized or absent at this distance (Cadenasso and Pickett 2001, McGarigal and McComb 1995, Oliver and Larson 1990). Sampling stations that fell within these remaining interior forest polygons were considered to be patch forest, while stations that did not occur in forest interior were considered to be occurring in the corridors of forest connecting remnant patches. Data analysis The full data set from the 62 point count stations was subdivided by forest class, with 43 stations occurring in corridors and 19 stations in patch forests. We calculated the total, mean, and cumulative species richness and abundance for individual sampling periods, each season (years pooled), year (seasons pooled), and overall (all data pooled). Also, the HSI model species were subdivided out of the main data set and compared with HSI values for each year (all seasons pooled), both years (all data pooled), season (both years pooled), and each season individually. In our results, we report both means and medians for our data, as mean HSI values (“averages”) are most often reported in applied work, while medians are what some of our statistical analyses evaluate; the differences between the two are important and instructive. Table 2. Description of habitat metrics for the habitat suitability index models employed in this study. Species Plot size Metrics Hairy Woodpecker 0.4 ha Number of snags > 25-cm dbh/ha (HAWO) Mean dbh of overstory trees Percent canopy cover of overstory trees Percent canopy closure of pines (optional) Barred Owl 0.4 ha Number of trees ≥ 51-cm dbh (BDOW) Mean dbh of overstory trees Percent canopy cover of overstory trees Pileated Woodpecker 0.4 ha Number of trees ≥ 51-cm dbh (PIWO) Percent canopy cover of overstory trees Number of tree stumps > 0.3 m in height and > 18 cm in diameter Number of logs > 18 cm in diameter Number of snags > 38-cm dbh Mean dbh of snags > 38 cm 302 Southeastern Naturalist Vol. 5, No. 2 Using Statistica 5.5 software (Statsoft 1999), we analyzed our data at an alpha level of 0.1, as we were most interested in identifying patterns worthy of further study. The HSI values were not normally distributed, so differences in HSI values between the presence/absence of model species and patch/corridor locations were evaluated using the Mann-Whitney U Test. Chi-square analysis was used to determine if each indicator species occurred in either patch or corridor forests more often than would be expected by chance. Spearman’s Rank Order Correlation was used to explore relationships between the HSI values and avian metrics as outlined above. Any comparisons that were non-significant were dropped from further analysis. In addition, the HSI values for each model were stratified by 75th and 25th percentiles, and any station with an HSI value in the 75th percentile was considered the best habitat in the Greenbelt relative to overall HSI habitat values for each model. Bayesian analysis allows for an estimation of the confidence one might have in the use of the models in this particular situation; using Excel 2000 software (Microsoft 2000), stratified HSI data were analyzed with actual occurrence of model species using Bayes’ theorem and data-based prior distributions (see Anderson [1998] for an example of this approach). This was done to determine the probability that model species will actually occur at a given station, given the observation of higher quality habitat from HSI data. Results Habitat suitability index values The results of the habitat suitability index (HSI) analysis indicated that suitable habitat in varying quantities occurred for each species within the Greenbelt. The HSI values for the HAWO model ranged from 0 to 0.95, with a mean value of 0.65 (median = 0.75). Sites with moderate HSI values were distributed heterogeneously throughout the forest, with a near-optimal maximum of 0.95 occurring at 11 stations in both patch and corridor forests. The minimum of 0.0 occurred at six stations in both patch and corridor forests. There was no significant difference in habitat value between patch (mean = 0.67, median = 0.85) and corridor (mean = 0.65, median = 0.70) forests for the HAWO model (U = 352, p = 0.38). For the BDOW model, HSI values ranged from 0 to 1, with a mean value of 0.64 (median = 0.77). There was a significant difference (U = 273.5, p = 0.04) between BDOW corridor HSI values (mean = 0.69, median = 0.77) and patch values (mean = 0.59, median = 0.67). The mean dbh index value for the BDOW model was higher in corridors than patches, which influenced the overall result. However, the 0.1 difference in average (and median) habitat value (HSI) between patches and corridors may not be biologically meaningful; indeed, the Barred Owl did not seem to show a preference for either forest class (see below). For the PIWO model, HSI values ranged from 0 to 0.85, with a mean value of 0.1 (median = 0). There was no significant difference (U = 681, p = 2006 D. Barry et al. 303 0.21) between HSI values of the corridor (mean = 0.07, median = 0.0) and patch (mean = 0.18, median = 0.0) forests. Most PIWO HSI values were low; 43 of the 62 stations sampled had a value of 0.0. Comparisons of HSI habitat values to model species occurrence Over the two-year study, HAWO was detected 47 times (at 39 of the 62 sampling stations), with most detections occurring in the fall seasons of both years. The HAWO did not occur significantly more often in patch forests (n = 29) than in corridor forests (n = 10) based on availability (χ2 = 1.16, df = 1, p = 0.29). Median HSI value at stations at which HAWO occurred was 0.8 (n = 39, mean = 0.7, sd = 0.26), whereas stations where it did not occur had a median HSI value of 0.65 (n = 23, mean = 0.57, sd = 0.34). However, this difference was not significant (U = 341, p = 0.11). The BDOW was detected 36 times (at 29 stations), and it also did not show any significant differences in occurrence based upon availability of forest class (χ2 = 2.09, df = 1, p = 0.15). Stations at which BDOW was present or absent had median HSI values of 0.77 (n = 29, mean = 0.68, sd = 0.31) and 0.67 (n = 33, mean = 0.64, sd = 0.29), respectively, a difference that was not significant (U = 427.5, p = 0.47). The PIWO was detected 26 times (at 18 stations) in two years, with most detections occurring during the summer. PIWO detections were concentrated in and around the largest patches of forest in the Greenbelt. It is unknown if the birds at these locations represent breeding pairs or individuals. The occurrence of the PIWO in this forest is the first confirmation of their presence in the area since 1986 (K. Steigman, McKinney, TX, pers. comm.). Unlike the other two indicator species, the PIWO occurred significantly more often in patch forests than would be expected by chance (χ2 = 6.50, df = 1, p < 0.011). The median HSI value was 0.0 for sites with (n = 18, mean = 0.09, sd = 0.14) and without (n = 44, mean = 0.11, sd = 0.22) PIWO presence, a non-significant difference (U = 357, p = 0.46) (Table 3). Bayesian analysis revealed that none of the HSI models satisfactorily predicted actual occurrence of model species in the study area, given the prior distribution based on good habitat (i.e., top quartiles of HSI value by model). The HAWO model performed the best, showing a probability of 0.667 that the higher quality habitat in the Greenbelt (by its HSI value) would have at least one HAWO. Analysis of BDOW data exhibited a Table 3. Summary HSI values by model species presence/absence (n refers to number of stations). HSI value n Median (95% CI) Mean (± 95% CI) Range HAWO HSI Presence 39 0.80 (0.71–0.85) 0.70 (0.09) 0.00–0.95 Absence 23 0.65 (0.50–0.85) 0.57 (0.15) 0.00–0.95 BDOW HSI Presence 29 0.77 (0.67–0.88) 0.68 (0.12) 0.00–1.00 Absence 33 0.67 (0.59–0.84) 0.64 (0.10) 0.03–1.00 PIWO HSI Presence 18 0.00 (0.00–0.19) 0.09 (0.07) 0.00–0.42 Absence 44 0.00 (0.00–0.00) 0.11 (0.07) 0.00–0.85 304 Southeastern Naturalist Vol. 5, No. 2 probability of 0.5 that the BDOW actually occurred at a station considered by the HSI model to be the better habitat within the Greenbelt. Analysis of the PIWO model showed a probability of 0.375 that the better habitat as shown by the PIWO HSI value would have a PIWO present. Comparisons of HSI values to avian communities Of the 109 species of birds detected during the designated sampling period over the 2 years of this study, 33 species were considered forest-dependent birds (Table 1). For the whole sampling period, a cumulative average of 31 species occurred at each station, with a mean of 12 forest-dependent species per point. The presence or absence of the indicator species (HAWO, PIWO, BDOW) did not show a relationship with forest-bird species richness, either in patch or corridor forest (Fig. 2). Only the HAWO HSI values had any significant correlations with forestbird species richness (Table 4). No significant correlations occurred with species richness and either the PIWO or the BDOW HSI values, and no Table 4. Significant correlations between HAWO HSI values and forest avian community metrics. Avian community metric Spearman’s R p Species richness (fall 1998) 0.26 0.045 Species richness (spring 1999) 0.29 0.022 Species richness (summer 1999) 0.30 0.018 Species richness (winter 2000) 0.29 0.027 Species richness (year 1 average) 0.36 0.004 Species richness (overall average) 0.36 0.004 Species richness (year 1 cumulative) 0.36 0.004 Species richness (overall cumulative) 0.38 0.002 Species richness (year 1 total) 0.36 0.004 Species richness (overall total) 0.32 0.012 Figure 2. Presence of HSI model species compared with cumulative counts of forestbird species richness for the study duration. 2006 D. Barry et al. 305 significant correlations occurred between the abundance of any forest bird and HSI habitat values for any of the three model species. An analysis of potential correlations between HSI model values and measures of the forestbird community showed 10 significant but weak correlations between the HAWO model and forest-bird species richness; the highest correlation among these was for the HAWO model and cumulative forest-bird species richness (R = 0.38, p = 0.002). Most correlations occurred with data from the first year of sampling (Fall 1998–Summer 1999); only one correlation occurred with second year (Fall 1999–Summer 2000) data: HAWO HSI and forest-bird species richness in Winter 2000 (R = 0.29, p = 0.027). Since we found a total of only 10 correlations out of 102 data-set comparisons made that were significant when analyzed at an α = 0.1 level, and since when using an α = 0.1 threshold, 10 of every 100 correlations found could be due to chance alone, the degree of statistical correlation is weak and there is a strong likelihood that it does not represent biological significance. Discussion Our evaluation suggests that these HSI models, as routinely used in the region, may not provide a basis to justify the assumed link between habitat value and the presence of indicator species and other habitat-related fauna. Thus, they may not be helpful as indicators of actual habitat use or presence/ absence by the species. The few, weak correlations between HSI values and the forest-bird community did not occur any more frequently than might be expected by chance alone, nor were the correlation values high enough to suggest biological relationships among habitat values and the forest-bird community. The results summarized in Tables 3 and 4 suggest that correlations could arise at any point in time regardless of what relationships might actually exist between birds and their habitat. Taken together, the models did not distinguish between areas with and without detections any better than at random. Bayesian analysis provided the same implications, that we can assign little confidence to the use of these models in this forest. Thus, if persons conducting an impact-assessment project had chosen to use these models in a particular season or year, they might detect a set of relationships between the birds and the forest habitat that would not necessarily be the same in any other season or year. As a whole, these results suggest that the use of HSI models as a surrogate for indicator species and/or avian community presence is inappropriate given the sampling conditions. The results of our habitat sampling suggested that little differences exist between designated corridors and patches, and that it may be possible to maintain much of the habitat value present in corridors that connect patches. There is empirical evidence that these corridors are important for a variety of animals that use them to migrate, disperse, or otherwise move between and among habitats throughout the landscape (e.g., Beier and Noss 1998, Machtans et al. 1996). While our studies have shown that similar habitat values can be present in larger patches as well as within the corridors that 306 Southeastern Naturalist Vol. 5, No. 2 connect them (Barry et al. 2000), the relationships among animals and their habitats and landscape contexts must be more fully understood and evaluated before managers should feel encouraged by a positive assessment of habitat value. Because corridors are often seen as one of the most important conservation tools in fragmented landscapes (Beier and Noss 1998), this could be seen as an important finding. However, additional research on actual patterns of animal use and movement through any conservation corridor is needed to determine whether corridor habitat, especially as defined by HSI models, is truly functional. Indeed, the results of this study indicate that habitat evaluation for indicator species using the HSI format may not provide the support these evaluations have often been assumed to have. Although habitat factors, such as structural diversity, are often presented in the literature and in “common knowledge” as strongly related to bird diversity (e.g., Flather et al. 1992, MacArthur and MacArthur 1961), recent research suggests that spatial heterogeneity is a central causal factor of faunal diversity in ecosystems, as the landscape mosaic itself often explains more of the variation in diversity of fauna than within-patch factors such as site-specific habitat characteristics (Franklin 1993, Haas 1995, Henderson et al. 1985, Keitt et al. 1997, Machtans et al. 1996, Pearson 1993, Pickett and Cadenasso 1995, Robbins et al. 1989, St. Clair et al. 1998, Storch 2002, Whitcomb et al. 1981). Other studies of the avian and mammal communities occurring on the Ray Roberts Greenbelt support these views (Barry et al. 2000, Hoffman 2001). Taken as a whole, these studies strongly suggest that the landscape may impose important topdown constraints on avian response to habitat-level factors. Because many animal species react to landscape factors and other spatial considerations, the improvement of HSI models will have to include explicit analyses of landscape context (e.g., Larson et al. 2003). The use of HSI models has one extremely useful feature: they provide numerical values that represent habitat conditions in a given place. These values can be measured from year to year or longer, and can provide a useful analysis of trends and changes in habitat features. So, while they may not necessarily have any direct relationship with their target species or a larger community in the form which most of them take (i.e., habitat-feature based), they can provide important, more or less objective standards by which changes (e.g., those wrought by impacts) can be measured directly. Because the use of numbers is often a considerable part of decision making, this is a particularly valuable use of the models without having to make claims regarding the relationships between habitat and wildlife. Our study was not designed to determine whether HSI models are inherently flawed or if a broader selection of models is needed; rather, we suggest that the untailored use of HSI models for the generation of management alternatives (e.g., the NEPA process) is a serious problem, one that is too often overlooked by managers who may not have the knowledge to judge whether or not a study using these models was done appropriately. Our study 2006 D. Barry et al. 307 also suggests that management and impact assessments evaluated under typical applied-data collection conditions (esp. short and/or inappropriate time frames) should be carefully appraised to determine their true worth. This can be accomplished most beneficially through a peer review assessment of the intended application of HSI values within both initial scope of work proposals and draft/final reports. This latter type of review should be conducted before offering the results and (potentially flawed) alternative management strategies to the public for scoping. If the use of HSI models cannot accurately reflect wildlife/habitat relationships at a given site, managers must be honest that their use is only for analysis of changes in habitat features, and does not necessarily reflect potential wildlife impacts. Acknowledgments Thanks to K. Steigman and M. Guilfoyle for many hours of field work, and to S. Mabey and four anonymous reviewers who substantially increased the quality of this manuscript. Literature Cited Allen, A.W. 1987. Habitat suitability index models: Barred Owl. Biological Report 82(10.143). National Ecology Center, US Fish and Wildlife Service, US Department of the Interior, Washington, DC. Anderson, J.L. 1998. Embracing uncertainty: The interface of Bayesian statistics and cognitive psychology. Conservation Ecology [online] 2(1):2. Available at: http:/ /www.consecol.org/vol2/iss1/art2. Bailey, R.G. 1995. Description of the ecoregions of the United States. US Forest Service, US Department of Agriculture, Washington, DC. Barry, D., and A.J. Kroll. 1999. A phytosociological description of a remnant bottomland hardwood forest in Denton, Texas. Texas Journal of Science 51(4):309–316. Barry, D., and A.J. Kroll. 2003. A phytosociological description of a remnant bottomland hardwood forest in Denton, Texas. LLELA Research Note 5 [online]. Lewisville Lake Environmental Learning Area / Lewisville Wildlife Management Area, Lewisville, TX. Available at: http://www.ias.unt.edu/llela/documents/ llelanote5.pdf. Barry, D., K. Hoffman, S.S. Holcomb, K.L. Dickson, and E.G. Zimmerman. 2000. Ray Roberts Greenbelt corridor study final report. Institute of Applied Sciences, University of North Texas, Denton, TX. Beier, P., and R.F. Noss. 1998. Do habitat corridors provide connectivity? Conservation Biology 12(6):1241–1252. Cadenasso, M.L., and S.T.A. Pickett. 2001. Effect of edge structure on the flux of species into forest interiors. Conservation Biology 15(1):91–97. Conner, R.N., and J.G. Dickson. 1997. Relationships between bird communities and forest age, structure, species composition, and fragmentation in the West Gulf Coastal Plain. Texas Journal of Science 49(3) Supplement:123–138. Ehrlich, P.R., D.S. Dobkin, and D. Wheye. 1988. The Birder’s Handbook: A Field Guide to the Natural History of North American Birds. Simon and Schuster, New York, NY. 720 pp. 308 Southeastern Naturalist Vol. 5, No. 2 Fischer, R.A., and J.C. Fischenich. 2000. Design recommendations for riparian corridors and vegetated buffer strips. Report ERDC TN-EMRRP-SR-24. US Army Corps of Engineers Research and Development Center, Vicksburg, MS. Flather, C.H., S.J. Brady, and D.B. Inkley. 1992. Regional habitat appraisals of wildlife communities: A landscape-level evaluation of a resource planning model using avian distribution data. Landscape Ecology 7(2):137–147. Franklin, J.F. 1993. Preserving biodiversity: Species, ecosystems, or landscapes? Ecological Applications 3(2):202–205. Garshelis, D.L. 2000. Delusions in habitat evaluation: Measuring use, selection, and importance. Pp. 111–165, In L. Boitani and T. Fuller (Eds.). Research Techniques in Animal Ecology: Controversies and Consequences. Columbia University Press, New York, NY. 464 pp. Hall, C.A.S. 1988. An assessment of several of the historically most influential theoretical models used in ecology and of the data provided in their support. Ecological Modelling 43:5–31. Hamel, P.B., W.P. Smith, D.J. Twedt, J.R. Woehr, E. Morris, R.B. Hamilton, and R.J. Cooper. 1996. A land manager’s guide to point counts of birds in the Southeast. General Technical Report GTR-SO-120. US Forest Service, US Department of Agriculture, Southern Experiment Station, Asheville, NC. Harris, L.D., and P. Kangas. 1988. Reconsideration of the habitat concept. Transactions of the 53rd North American Wildlife and Natural Resources Conference 53:137–144. Haas, C.A. 1995. Dispersal and use of corridors by birds in wooded patches on an agricultural landscape. Conservation Biology 9(4):845–854. Henderson, M.T., G. Merriam, and J. Wegner. 1985. Patchy environments and species survival: Chipmunks in an agricultural mosaic. Biological Conservation 31:95–105. Hobbs, N.T., and T.A. Hanley. 1990. Habitat evaluation: Do use/availability data reflect carrying capacity? Journal of Wildlife Management 54(4):515–522. Hodges, J.D. 1997. Development and ecology of bottomland hardwood sites. Forest Ecology and Management 90:117–125. Hoffman, K.W. 2001. Riparian forest width and the avian community in a greenbelt corridor setting. M.Sc. Thesis. University of North Texas, Denton, TX. Institute of Applied Sciences. 1995. Ray Roberts Lake post-impoundment environmental study: Year ten. US Army Corps of Engineers, Fort Worth District, Ft. Worth, TX. Kellison, R.C., and M.J. Young. 1997. The bottomland hardwood forest of the southern United States. Forest Ecology and Management 90:101–15. Keitt, T.H., D.L. Urban, and B.T. Milne. 1997. Detecting critical scales in fragmented landscapes. Conservation Ecology [online] 1(1):4. Available at: http:// www.consecol.org/vol1/iss1/art4. Knopf, F.L., R.R. Johnson, T. Rich, F.B. Samson, and R.C. Szaro. 1988. Conservation of riparian ecosystems in the United States. Wilson Bulletin 100:272–284. Larson, M.A., W.D. Dijak, F.R. Thompson, and J.J. Millspaugh. 2003. Landscapelevel habitat suitability models for twelve species in southern Missouri. General Technical Report NC-233. US Forest Service, US Department of Agriculture, North Central Experiment Station, St. Paul, MN. 2006 D. Barry et al. 309 MacArthur, R.H., and J.W. MacArthur. 1961. On bird species diversity. Ecology 42:594–598. Machtans, C.S., M.-A. Villard, and S.J. Hannon. 1996. Use of riparian buffer strips as movement corridors by forest birds. Conservation Biology 10(5):1366–1379. McGarigal, K., and W.C. McComb. 1995. Relationships between landscape structure and breeding birds in the Oregon Coast Range. Ecological Monographs 65(3):235–260. Microsoft. 2000. Microsoft Excel 2000. Microsoft, Inc. Redmond, WA. North American Bird Conservation Initiative (NABCI). 2000. The North American bird conservation initiative in the United States: A vision of American bird conservation. US Fish and Wildlife Service, Division of North American Waterfowl and Wetlands, Arlington, VA. 29 pp. National Weather Service (NWS). 2005. Dallas/Fort Worth Climatology. National Weather Service Fort Worth Weather Forecast Office. [online] Available at: http://www.srh.noaa.gov/fwd/CLIMO/dfw/dfwclimo.html. O’Neil, L., and A. Carey. 1986. Introduction: When habitats fail as predictors. Pp. 207–208, In J. Verner, C.J. Ralph, and M. Morrison (Eds.). Wildlife 2000: Modeling Habitat Relationships of Terrestrial Vertebrates. University of Wisconsin Press, Madison, WI. 470 pp. Oliver, C., and B. Larson. 1990. Forest stand dynamics. McGraw-Hill, New York, NY. 467 pp. Pearson, S.M. 1993. The spatial extent and relative influence of landscape-level factors on wintering bird populations. Landscape Ecology 8:3–18. Pickett, S.T.A., and M.L. Cadenasso. 1995. Landscape ecology: Spatial heterogeneity in ecological systems. Science 269:331–334. Ralph, C.J., J.R. Sauer, and S. Droege (Eds.). 1995. Monitoring bird populations by point counts. General Technical Report PSW-GTR-144. US Forest Service, US Department of Agriculture, Albany, CA. Robbins, C., D. Dawson, and B. Dowell. 1989. Habitat-area requirements of breeding forest birds of the middle Atlantic states. Wildlife Monographs 103:1–34. Roloff, G.J., and B.J. Kernohan. 1999. Evaluating reliability of habitat-suitability index models. Wildlife Society Bulletin 27(4):973–985. Sauer, J.R., J.E. Hines, I. Thomas, J. Fallon, and G. Gough. 1999. The North American breeding-bird survey: Results and analysis 1966–1998. Version 98.1 [online], USGS Patuxent Wildlife Research Center, Laurel, MD. Available at: http://www.mbr-pwrc.usgs.gov/bbs/bbs98.html. Schroeder, R.L. 1983. Habitat suitability index models: Pileated Woodpecker. FWS/ OBS-82/10.39. National Ecology Center, US Fish and Wildlife Service, US Department of the Interior, Washington, DC. Sousa, P.J. 1987. Habitat suitability index models: Hairy Woodpecker. Biological Report 82(10.146). National Ecology Center, US Fish and Wildlife Service, US Department of the Interior, Washington, DC. St. Clair, C.C., M. Belisle, A. Desrochers, and S. Hannon. 1998. Winter responses of forest birds to habitat corridors and gaps. Conservation Ecology [online] 2(2):13. Available at: http://www.consecol.org/vol2/iss2/art13/. Statsoft. 1999. Statistica 5.5. Statsoft, Inc. Tulsa, OK. Stokes, D., and L. Stokes. 1996. Stokes Field Guide to Birds: Eastern Region. Little, Brown, and Company, New York, NY. 496 pp. 310 Southeastern Naturalist Vol. 5, No. 2 Storch, I. 2002. On spatial resolution in habitat models: Can small-scale forest structure explain Capercaillie numbers? Conservation Ecology [online] 6(1):6. Available at: http://www.consecol.org/vol6/iss1/art6/. USDA. 1980. Soil survey of Denton County, Texas. Natural Resources Conservation Service, US Department of Agriculture,Washington, DC. USFWS. 1980a. Habitat as a basis for environmental assessment. ESM 101. US Fish and Wildlife Service, Division of Ecological Services, Washington, DC. USFWS. 1980b. Habitat evaluation procedures. ESM 102. US Fish and Wildlife Service, Division of Ecological Services, Washington, DC. USFWS. 1981. Standards for the development of habitat suitability index models for use in the Habitat Evaluation Procedures. ESM 103. US Fish and Wildlife Service, Division of Ecological Services, Washington, DC. Whitcomb, R.F., C.S. Robbins, J.F. Lynch, B.L. Whitcomb, M.K. Klimkiewicz, and D. Bystrak. 1981. Effects of forest fragmentation on avifauna of the eastern deciduous forest. Pp. 125–205, In R.L. Burgess and D.M. Sharpe (Eds.). Forest Island Dynamics in Man-dominated Landscapes. Springer-Verlag, New York, NY. 310 pp.