Modeling the distribution of the dwarf palmetto (Sabal minor; arecaceae) in McCurtain County, Oklahoma.
McCurtain County is in southeastern Oklahoma, bordering Texas to the south and Arkansas to the east. Elevations are 86-582 m above sea level (National Elevation Dataset, http://ned.usgs.gov/). Average annual precipitation is 132 cm, average annual temperature is 17[degrees]C, and there is an average of 15 days/year when temperature is below 26.7[degrees]C. The growing season averages 215 days with the average first freeze occurring on 3 November and the last on 30 March (Oklahoma Climatological Survey, http://climate.ok.gov/). Vegetation associations in McCurtain County (B. W. Hoagland, http://geo.ou.edu/botanical) include bottomland forest, cypress-bottoms (Cupressaceae) forest, loblolly pine (Pinus taeda), and oak-pine (Quercus-Pinus) forest (Reinking, 2004). All S. minor collected were in the southeastern portion of McCurtain County (B. W. Hoagland et al., http://geo.ou.edu/botanical).
Despite S. minor being listed as critically imperiled in Oklahoma, little has been published about its geographic range. Other studies that have examined the distribution of palms relied upon aerial photographs (Everitt et al., 1996) or satellite imagery (McMorrow, 2001). However, because S. minor is an understory species, detection by aerial photographs or satellite imagery is problematic due to vegetation that obscures these plants from above.
One approach to mapping the range of S. minor in Oklahoma is to use modeling of its ecological niche. Traditional models use presence-absence data in conjunction with various ecogeographical variables (e.g., temperature, precipitation) to identify factors that affect distribution (Anderson et al., 2003). This information can be analyzed using Genetic Algorithm for Rule-set Prediction, which uses a heuristic approach to identify factors correlated with presence or absence of a species (Anderson et al., 2003). However, data indicating absence at a site can be misleading because the species may have been either truly absent or not detected (Hirzel et al., 2002). As a result, false absences may potentially skew predictions of the model. An alternative approach is to use only data indicating presence to identify the geographic range of an organism (Hirzel et al., 2002). An Ecological Niche Factor Analysis identifies ecogeographical variables associated with specimens and generates a habitat-suitability map (A. H. Hirzel et al., http://gi.leica-geosystems.com/ default.aspx) using the program BioMapper (A. H. Hirzel et al., http://www.unil.ch/biomapper). This analysis summarizes multiple ecogeographical variables into a few independent factors similar to a principal-components analysis. The first factor is the marginality factor, which maximizes the difference in environmental conditions between the niche and the study area. Subsequent factors are specialization factors and are created by computing the direction that maximizes the remaining variance between the study area and locations where the species occurs. We used Ecological Niche Factor Analysis to identify ecogeographical variables that affect the distribution of S. minor and to model its distribution in McCurtain County, Oklahoma.
MATERIALS AND METHODS--To determine where S. minor was documented to occur in McCurtain County, Oklahoma, we obtained data from the Oklahoma Biological Survey Oklahoma Vascular Plants Database (B. W. Hoagland et al., http://geo.ou.edu/botanical) and from the United States Department of Agriculture Forest Service. We also traveled to McCurtain County in February 2008 to confirm sites where S. minor had been reported and to record locations where S. minor was encountered incidentally. We then created a preliminary model of the ecological niche using ArcGIS 9.2 (Crosier et al., 2004) that was based on elevation, use of land, and vegetational cover to identify areas that potentially were suitable for S. minor. In March 2008, we collected data at 46 additional sites in the county.
We used ArcGIS 9.2 (Crosier et al., 2004) to create a grid of 100 by 100 1-[m.sup.2] cells (10,000 [m.sup.2] or 1 ha) for McCurtain County. As McCurtain County consists of 494,743 ha, the grid consisted of 494,743 cells. Each cell then had variables assigned to it, including landuse, and vegetational cover from both the United States Geological Survey and aerial imagery (downloaded from United States Department of Agriculture Geospatial Data Gateway, http://datagateway.nrcs.usda.gov/; classified using ERDAS IMAGINE 9.1; ERDAS, Atlanta, Georgia), distance to surface water, amount of water flowing into each cell (calculated using ArcGIS 9.2; Crosier et al., 2004), elevation, temperature, precipitation, slope, aspect, type of soil, and whether S. minor was discovered at each cell on the grid. We then used BioMapper 4 to perform an Ecological Niche Factor Analysis (A. H. Hirzel, http://www2.unil.ch/biomapper/). Based on observations in the field, populations of S. minor appeared to be spatially clumped, so we added a spatial-autoregressive component (i.e., eastings and northings in the Universal Transverse Mercator coordinate system) to the model.
We evaluated four methods of constructing maps of habitat suitability (median, geometric mean, harmonic mean, and minimal distance) using a continuous Boyce index (Hirzel et al., 2006). The median algorithm for calculating habitat suitability assumes that the median distribution of the species for each ecogeographical March 2011 Butler et al.--Dwarf palmetto in Oklahoma 67 variable is the preferred habitat and it assumes that distributions are symmetrical. The geometric-mean algorithm is similar to the median algorithm but does not assume symmetry. The harmonic mean is similar to the geometric mean but it gives a high weight to all observations (rather than mean values for distribution of the species in each ecogeographical variable). Minimal distance gives each observation the same weight, rather than accounting for density. A further description of each method is in A. Hirzel et al. (http://www.unil.ch/biomapper). We used a continuous Boyce Curve (Hirzel et al., 2006) to classify habitat as either suitable (suitability >50%) or unsuitable (suitability <50%). Robustness of the model was assessed using a continuous Boyce index (Hirzel et al., 2006), which can range from 0 to 1, with models closer to 1 being more robust.
[FIGURE 1 OMITTED]
RESULTS--Sabal minor was at 41 locations representing 29 1-ha cells on grids (Fig. 1a). Of the locations, 28 were new and 13 were from herbarium collections. Although all records in the Oklahoma Vascular Plants Database were from southeastern McCurtain County, S. minor occurred as far north as Beavers Bend State Park (34[degrees]08.34'N, 94[degrees]41.16'W) ca. 40 km north of the Red River. There were [greater than or equal to]20 individuals in this population, primarily clustered along the Mountain Fork River (a specimen from here was deposited in the Herbarium Division of the Museum of Natural History at the University of Central Oklahoma). Although we sampled points north of Beavers Bend State Park prior to performing the Ecological Niche Factor Analysis, we did not find S. minor further north. The greatest density of S. minor was in southeastern McCurtain County, particularly near Red Slough Wildlife Management Area (33[degrees]44.31'N, 94[degrees]40.15'W) where in some areas it was the dominant understory plant with [less than or equal to]74 mature plants/100 [m.sup.2]. Most S. minor were in low-lying deciduous forests (primarily dominated by Celtis laevigata, Nyssa sylvatica, Quercus lyrata, and Q. phellos), but some were in open fields and, at two locations, in upland stands of loblolly pine.
The best model for calculating habitat suitability was the harmonic-mean model as it had the highest continuous Boyce-index value of 0.519. The distribution of S. minor was associated positively with average annual minimum temperatures and was associated negatively with elevation. Populations of S. minor also were strongly autocorrelated spatially and were most common on silty clay loam. The habitat-suitability map (Fig. 1b) shows that the greatest concentration of S. minor should occur in southeastern McCurtain County, but that suitable habitat for this species occurs in a wider area of the county than records in herbarium collections suggest.
DISCUSSION--Although S. minor is most common in extreme southeastern McCurtain County, its distribution is wider than previously reported, extending ca. 40 km north of the Red River and ca. 25 km north of any record from herbarium specimens. It is unclear whether S. minor was present historically this far north or whether this is a recent expansion of its geographic range. However, the relatively large population (ca. 20 plants) coupled with the observation of two individuals with fruit suggests that this population has been present for several years.
The harmonic-mean model provided the best model of habitat suitability as it had the highest continuous Boyce-index value. In general, harmonic-mean models are more appropriate for relatively small samples (A. H. Hirzel et al., http://www2.unil.ch/biomapper/). This model indicated that presence of S. minor was associated positively with average annual minimum temperatures and negatively associated with elevation. McCurtain County is at the northwestern edge of the range of S. minor, so it was not surprising to find that S. minor was restricted to the lowest elevations having the warmest temperatures. Sabal minor in McCurtain County usually was clustered in areas with silty clay; previously, it was reported to tolerate a wide range of soils from sandy to loamy limestone-derived soils (Henderson, 1986). Given that S. minor was primarily in swampy conditions, it is possible that the distribution in McCurtain County may be influenced by the superior water-retaining capacity of silty clay soils as opposed to sandy and loamy soils, which have greater porosity (Rowell, 1994) and better drainage.
The greatest density of potentially suitable habitat was in southeastern McCurtain County (Fig. 1b). Immediately to the north of this area, the elevation rises slightly, decreases, then rises again resulting in a narrow belt of suitable habitat in lower elevations ca. 5-10 km north of the largest area of suitable habitat. North of this belt the elevation increases and potentially suitable habitat becomes sparser.
Sabal minor currently is listed by the Oklahoma Natural Heritage Inventory as S1, critically imperiled, within the state. Given that 28 new locations were discovered during this study, it may be worthwhile to re-evaluate the status of this species in Oklahoma. However, although S. minor was locally common near Red Slough Wildlife Management Area, it was scarce and local (i.e., only one plant or a small group of plants were present) at most locations elsewhere in McCurtain County. At least two areas with large numbers of S. minor have been cleared in McCurtain County in the past 10 years. Continued monitoring of the population of this plant is recommended, both for conservation purposes and to determine if an expansion of the geographic range may be occurring. Fruits of this species are dispersed by birds and water (Zona, 1997), which may facilitate expansion of the range. In North Carolina, a population of S. minor was reported recently from Martin County, ca. 45 km north of previously known populations, suggesting that S. minor may be expanding its range northward in North Carolina (Tripp and Dexter, 2006). We suggest further monitoring of S. minor in Oklahoma to investigate whether a similar expansion of the range may be occurring.
We thank R. Bastarache and personnel of the United States Forest Service for assistance in locating populations of S. minor in McCurtain County, the Oklahoma Vascular Plants Database for herbarium data, and the Oklahoma State University Forest Resources Center for providing accommodations in McCurtain County. Thanks to L. Stabler for comments on an early draft. This research was supported by the Office of Research and Grants at the University of Central Oklahoma.
Associate Editor was Janis K. Bush.
Submitted 30 May 2009. Accepted 26 July 2010.
ANDERSON, R. P., D. LEW, AND A. T. PETERSON. 2003. Evaluating predictive models for species' distributions: for selecting optimal models. Ecological Modelling 162:211-232.
CROSIER, S., B. BOOTH, K. DALTON, A. MITCHELL, AND K. CLARK. 2004. ArcGIS 9--getting started with ArcGIS. March 2011 Butler et al.--Dwarf palmetto in Oklahoma 69
Environmental Systems Resarch Institute Press, Redlands, California.
EVERITT, J. H., F. W. JUDD, D. E. ESCOBAR, M. A. ALANIZ, M. R. DAVIS, AND W. MACWHORTER. 1996. Using remote sensing and spatial information technologies to map Sabal palm in the Lower Rio Grande Valley of Texas. Southwestern Naturalist 41: 218-226.
GOLDMAN, D. H. 1999. Distribution update: Sabal minor in Mexico. Palms 43:40-44.
HENDERSON, A. 1986. A review of pollination studies in the Palmae. Botanical Review 52:221-259.
HIRZEL, A. H., J. HAUSSER, D. CHESSEL, AND N. PERRIN. 2002. Ecological-niche factor analysis: how to compute habitat-suitability maps without absence data? Ecology 83:2027-2036.
HIRZEL, A., G. LE LAY, V. HELFER, C. RANDIN, AND A. GUISAN. 2006. Evaluating the ability of habitat suitability models to predict species presences. Ecological Modeling 199:142-152.
HOAGLAND, B. W. 2000. The vegetation of Oklahoma: a classification for landscape mapping and conservation planning. Southwestern Naturalist 45:385-420.
MCMORROW, J. 2001. Linear regression modeling for the estimation of oil palm age from Landsat TM. International Journal of Remote Sensing 22: 2243-2264.
REINKING, D. L., EDITOR. 2004. Oklahoma breeding bird atlas. University of Oklahoma Press, Norman.
RIFFLE, R. L., AND P. CRAFT. 2003. An encyclopedia of cultivated palms. Timber Press, Portland, Oregon.
ROWELL, D. L. 1994. Soil science: methods and applications. Longman Scientific and Technical, Essex, United Kingdom.
TRIPP, E. A., AND K. G. DEXTER. 2006. Sabal minor (Arecaceae): a new northern record of palms in eastern North America. Castanea 71:172-177.
ZONA, S. 1990. A monograph of Sabal (Arecaceae: Coryphoideae). Aliso 12:583-666.
ZONA, S. 1997. The genera of Palmae (Arecaceae) in the southeastern United States. Harvard Papers in Botany 11:71-107.
* Correspondent: firstname.lastname@example.org
CHRISTOPHER J. BUTLER,* JENNIFER L. CURTIS, KIMBERLY MCBRIDE, DAVID ARBOUR, AND BERLIN HECK
Department of Biology, University of Central Oklahoma, 100 North University Drive, Edmond, OK 73034 (CJB, JC, KM)
Oklahoma Department of Wildlife Conservation, Route 4, Box 2900, Broken Bow, OK 74728 (DA) 109 Kaye Drive, Broken Bow, OK 74728 (BH)
|Printer friendly Cite/link Email Feedback|
|Author:||Butler, Christopher J.; Curtis, Jennifer L.; McBride, Kimberly; Arbour, David; Heck, Berlin|
|Date:||Mar 1, 2011|
|Previous Article:||Annual frequency of clutches of Pseudemys texana and Trachemys scripta at the headwaters of the San Marcos River in Texas.|
|Next Article:||Demography of a semelparous, high-elevation population of Sceloporus bicanthalis (Lacertilia: Phrynosomatidae) from the Nevado de Toluca Volcano,...|