Printer Friendly

The wetland vegetation of Caddo Lake.

Abstract. -- Caddo Lake, located along the boundary of northeastern Texas and northwestern Louisiana, contains extensive areas of relatively undisturbed forested wetlands and has been designated a wetland of international importance under the Ramsar Convention. Vegetation and water levels from 30 plots were sampled. Aerial photographs of the 3000 ha Caddo Lake State Park and Wildlife Management Area were used to map the park into 119 survey areas, each representing a distinct patch of vegetation. Both data sets were separately subject to Two Way Indicator Species Analysis (TWINSPAN) classification and Detrended Correspondence Analysis (DCA) ordination. Water levels were strongly correlated with the first axis of a DCA ordination of the 30 plots which arranged plots along a flooding gradient from mesic natural levees and terraces to cypress swamps. These results were corroborated by analysis of data from the 119 survey areas and were used to identify six community types: Rich Mesic Slopes, Mesic Bottomland Ridges, Bottomland Oak Flats, Cypress-Water elm Swamps, Closed-Canopy Cypress Swamps and Deep water Cypress Swamps. These communities are comparable to other wetland communities described for the region.


Forested wetlands in the southern USA have suffered extensive losses since the time of European settlement (Mitsch & Gosselink 1993). Most remaining areas exist as isolated fragments of less than 100 ha (Gosselink et al. 1990). Between 1940 and 1980, the national rate of loss of forested wetlands in the USA was 2.8 million ha per year with most of the loss occurring in the South (Abernathy & Turner 1987). In eastern Texas, only an estimated 37% of the presettlement riparian forests still existed in 1980 (Frye 1987). Although much of the loss is due to conversion to agriculture, water resource projects are also a threat; over 240,000 ha of Texas bottomland forest were destroyed by impoundments such as Lake of the Pines, Toledo Bend Reservoir and Sam Rayburn Reservoir (Frye 1987).

Caddo Lake, located in northeastern Texas and northwestern Louisiana, is associated with 3000 ha of contiguous forested wetlands on public lands (Taylor et al. 1996) and additional wetlands on private lands. It provides an intact forested wetland landscape on a scale seldom seen today in the South. Consequently, Caddo Lake is recognized as a wetland habitat of international importance by the Ramsar Convention, an international convention named after Ramsar, Iran, the place of its adoption (Navid 1989; Davis 1994). Caddo lake, Catahoula Lake in Louisiana, and the Okefenokee Swamp in Georgia are the only three Ramsar wetlands in the southern USA excluding peninsular Florida (US Fish and Wildlife Service 1996).

There have been many hydrologic, geomorphic and vegetation studies of bottomlands and swamps in the South including Chambless & Nixon (1975), Thompson (1980), Bedinger (1981), Mathies et al (1983), Miller (1990), Nixon et al. (1990), Shankman & Drake (1990), Shankman (1991), Barrett (1995) and Crouch & Golden (1997). However, in spite of Caddo Lake's importance as a major wetland landscape, only limited non-published information about its wetland vegetation (Hine & Nixon 1992; Sheffield 1995; MacRoberts 1979) is available. This paper provides a quantitative description and gradient analysis of the wetland plant communities of Caddo Lake, focusing on the public lands of the Caddo Lake State Park and Wildlife Management Area.


Caddo Lake lies between Mooringsport, Louisiana, and Karnack, Texas at approximately 32[degrees] 42' 38" N, 94[degrees] 8' 25" W along Big Cypress Bayou, a tributary of the Red River. It is in the Pineywoods vegetation area of Hatch et al. (1990), and the Southeastern Mixed Forest Province of Bailey et al. (1994). The climate is humid-subtropical with a mean annual low temperature of 11[degrees]C, an average high temperature of 22[degrees]C, an annual mean of 116.8 cm of rainfall, and negligible snowfall (Larkin & Bomar 1983).

Historically called Ferry Lake, Caddo Lake is a drowned floodplain, one of several lakes that formed around the year 1800 as a result of a 120 km-long series of log jams (the Great Raft) that blocked the main channel of the Red river (Barrett 1995). The Raft forced water into side channels of the Red River and caused water to back up into tributaries, forming Caddo Lake and the now extinct Sodo Lake. Steam boats entered Caddo Lake via side channels of the Red River, contributing to the commerce of the time (Dahmer 1988). Removal of the Great Raft in 1873 drained Sodo Lake and caused Caddo Lake to slowly recede until a dam built in 1914 near Mooringsport, Louisiana, stabilized the lake and preserved the cypress and bottomland wetlands of the formerly natural lake (Dahmer 1988; Barrett 1995). The extent of the alteration of the original hydrologic regime and consequent influence on vegetation is unknown, although sedimentation rates were higher prior to the break up of the Raft and dam construction as the result of inflow of sediment-rich water from the Red River (Barrett 1995).

Caddo Lake, including its wetlands, covers roughly 10,7200 ha at capacity of 51.36 m above mean sea level, but there is considerable seasonal variation in water levels and in the area actually flooded (USGS 1:2400 Topographic map; A.I.D. Associates 1993; US Army Corps of Engineers 1994b). The eastern portion of Caddo Lake, much of which lies in Louisiana, is the deepest and consists largely of open water (Barrett 1995). The western portion, in Texas, is the focus of this study. It is a "freshwater delta" (Barrett 1995) occurring on drowned stream channels, point bar deposits, natural levees and lacustrine deposits. These features provide for a diversity of communities including ponds, cypress swamps, hardwood flats and mesic communities.


Two transects, perpendicular to Cypress Bayou, were randomly located along a 1.5 km baseline in the Caddo Lake State Park and Wildlife Management Area. The transects, with a combined length of 1500 m followed a moisture and elevation gradient from a mesic natural levee (53.4 m above sea level) through a bottomland oak flat, to a cypress swamp (50.6 m above sea level). Thirty points, each defining the center of a series of nested plots, were located at 50 m intervals along the transects. The samples, a modified and simplified form of the "Whitaker method" (Shmida 1984), included a series of nested plots. Each point defined a 1000 [m.sup.2] circular plot, a 500 [m.sup.2] circular plot, a 100 [m.sup.2] circular plot, a 3.16 m by 3.16 m (10 [m.sup.2]) plot, and a 1 [m.sup.2] plot. Plot centers were permanently marked with tagged aluminum stakes. Inundated plots were marked by recording the distance and azimuth from a nearby tagged reference tree.

Diameter at 1.5 m (breast) height (dbh) was recorded for each tree greater than 10 cm dbh in the 500 [m.sup.2] plot and the number of stems of shrubs and saplings less than 10 cm dbh but greater than 1 m high was tallied for species in the 100 [m.sup.2] plot. Ground layer species (herbaceous species and woody plants less than 1 m tall) were assigned "occurrence ranks" based on their occurrence in different sized plots. Species occurring in 1 [m.sup.2] plots were assigned an occurrence rank of "five" while those found in the 10 [m.sup.2] plot but not in the 1 [m.sup.2] plot received a rank of "four", and those occurring in the 100 [m.sup.2] but not the smaller plots received a rank of "three". Species not found in the 100 [m.sup.2] plot but observed within a radius of approximately 18m (1000 [m.sup.2] area) of the plot center were given a rank of "two" if three or more individuals or colonies were encountered and a rank of "one" if only one or two individuals were encountered or if plants were found only in one or two small, localized colonies. Percentage cover was also estimated for all ground cover species in the 10 [m.sup.2] plot. Species not found in the 10 [m.sup.2] plot but in the larger sample area were given a "trace" coverage of 0.1%. The coverage, projected on the ground, and occurrence of the epiphyte Tillansia usneoides (L.) L. was included with the ground layer.

The depth of water covering inundated plots or the depth to water in a shallow pit on non-flooded plots was measured for each plot on 6 December 1996, a time when the majority of the plots were inundated and the water level at a nearby gauging station (Tall Pines Lodge) was observed at 52.03 m above mean sea level. A level and sighting-stick were used to measure the elevation of two plots that were more than 1m above the water. Since the gauging station was relatively close (< 1000 m) to the plots and negligible current was observed, variation in the elevation of the water surface or "head" was considered to be negligible. The elevation of each plot was estimated by subtracting the depth of the plot below water from the water level elevation (52.03 m).

Since a goal of this study was to survey and map the plant communities of the entire Caddo Lake State Park and Wildlife Management Area, a survey methodology was devised that would facilitate a rapid survey of the area, but still obtain data suitable for quantitative analysis. Polygons, each representing a distinct patch of vegetation, were drawn on January 1993 color infrared aerial photographs covering the Caddo Lake State Park and Wildlife Management Area. During the summer and fall of 1994 and 1995, 119 of these areas were visited and a representative portion of each area roughly 1 ha in size was surveyed. All overstory species (> 10 cm diameter at 1.5 m (breast) height (dbh), mid story species (< 10 cm dbh but > 1 m high), and ground layer (herbaceous plants and woody plants < 1m high) were identified and ranked on a five-point abundance scale within their respective stratum. Notes on hydrology, soils, and wildlife also were recorded. In addition to providing a survey of most of the Park, these data provided an independent data set with which to corroborate the results of the transect plot data.

Voucher specimens were deposited in the herbarium of Stephen F. Austin State University (ASTC). Nomenclature follows Hatch et al. (1990) and Kartesz (1994) for species not found in Hatch et al. (1990).

Data analysis. -- Databases were compiled for both the transect and survey data sets. Overstory and ground layer data from both data sets were separately subjected to Detrended Correspondence Analysis (DCA) ordination (Hill 1979a; Hill & Gauch 1980) and Two Way Indicator Species Analysis (TWINSPAN, Hill 1979b): TWINSPAN is a hierarchical, divisive polythetic classification which classifies samples into groups with similar species composition. Input data for each species in each sample included log-transformed density (recommended by Harcombe et al. 1993) for the transect overstory, occurrence ranks for the transect ground layer, and abundance ranks for the survey data. Linear regression (Steele & Torrie 1980; Ludwig & Reynolds 1988) was used to relate DCA scores to water levels for the transect data set. Field notes were used to relate a DCA of the 119 survey sites to a flooding gradient and to other observed environmental factors.


DCA and other ordination techniques summarize complex, multivariate, samples-by-species data sets by arranging samples objectively along several axes on the basis of their species composition. Vegetation samples can be graphed as a scatter diagram based on ordination scores, in which points that are near each other represent samples with similar species composition while points distant from each other represent samples with dissimilar species composition (Hill & Gauch 1980; Gauch 1982; Jongman et al. 1995). Ordination axes are generally interpreted as being gradients in species composition reflecting an underlying environmental gradient such as one of flooding or nutrients, (Gauch 1982; Ludwig & Reynolds 1988).


TWINSPAN and DCA results for both the overstory and ground layer, field notes and observations of aerial photographs were used to classify the 119 survey samples into community types. Since plant communities generally change continuously along environmental gradients, (Gleason 1926; Gauch 1982) all classifications are somewhat artificial; arbitrary decisions occasionally had to be made among adjacent types when there was lack of agreement between results for overstory, ground flora, TWINSPAN and DCA. Ordination results were usually given preference over TWINSPAN results for these decisions. The classification was corroborated by comparison of the community descriptions with the results from the transect data set and was used along with aerial photographs to develop a map of the wetland vegetation of Caddo Lake State Park and Wildlife Management Area and adjacent areas.


DCA ordinates species simultaneously with samples (Hill & Gauch 1980; Hill 1979a). Species occurring nearby on an ordination tend to occur in the same type of samples, while distant species are generally found in different types of samples. Likewise TWINSPAN classifies species into groups on the basis of the sites they occur in (Hill 1979b; Gauch 1982). TWINSPAN and DCA results were used to classify Caddo Lake species into "ecological species groups": groups of species which respond similarly to environmental gradients and tend to occur together on similar types of sites (Muller-Dombois & Ellenberg 1974; Barnes et al. 1982). Presence of several members of a group is a strong indication of a site's ecological conditions and community type. For ground layer species, only those listed as community indicators or strong preferentials (occurred in at least 57% of the samples for the group) during one of the divisions of TWINSPAN were included in the ecological species groups.


Gradient analysis. -- The first axis of a DCA ordination of the 30 transect plots based on the occurrence rank of ground layer species (Figure 1) was strongly correlated with water levels on the plots ([R.sup.2] = 0.74, p < 0.01, Figure 2). The ordination reflected a gradient from well-drained mesic sites on a natural levee along Cypress Bayou (high first DCA axis ordination scores) to seasonally flooded oak-dominated flats (intermediate scores) to semi-permanently inundated water elm and cypress swamps (low scores). TWINSPAN classified the plots into five groups (the symbols in Figures 1 and 2) and showed close agreement with the DCA arrangement (Figure 1).

Mesic sites (open circles in Figure 2) were above 52.03 m above sea level and were not inundated at the time of measurement. Three subtypes of Bottomland Oak Flats were evident (Figure 1): somewhat open-canopied sites on slight mounds with a dense ground layer dominated by Erianthus strictus Baldw. (DCA score 87-161), intermediate sites with a closed willow oak (Quercus phellos L.) canopy and an abundance of Carex joorii Bailey (DCA score 162-206), and wetter sites with an overcup oak (Q. lyrata Walt.) dominated overstory and a sparse ground layer (DCA score > 207). The Erianthus oak sites (open squares, Figures 1 and 2) largely occurred between 51.9 and 52.1 m above sea level. The other Bottomland Oak Flats (triangles) occurred below 51.9 m with the overcup oak-dominated stands (solid triangles, Figure 2) tending to occupy the lowest sites at about 51.36-51.5 m. Two types of swamps were also evident along the first DCA axis: Cypress Water-elm Swamps (solid squares of Figures 1 and 2), had a dense sub-canopy of water elm (Planera aquatica (Walt.) J. F. Gmel. and some soil-rooted ground flora such as Brunnichia ovata (Walt.) Shinners and Saururus cernuus L., while cypress swamps (solid circles) were pure stands of bald cypress (Taxodium distichum (L.) L. Rich.) dominated by floating and submersed plants on the surface layer. Swamps occurred below the normal pool elevation of Caddo Lake (51.36m above sea level), with cypress swamps generally in deeper water than Cypress Water-elm Swamps (Figure 2). The second DCA axis was related to variation among cypress swamps: Open-canopied stands with a dense surface layer of Nuphar luteum (L.) Sibth. & Sm, Nelumbo lutea (Willd.) Pers., had high second axis scores while closed-canopy stands, their surface layer dominated by Lemnaceae species and Egeria densa Planch. had low scores (Figure 1).


A DCA ordination of the 30 transect plots on the basis of log-transformed density of overstory trees provided an arrangement of samples much like that of the ground layer ordination (Figure 3). As with the ground layer, first axis DCA scores were strongly correlated with water levels ([R.sup.2] = 0.84, p < 0.01, Figure 2). Wet Bottomland Oak Flats with their overcup-oak dominated canopy were more distant from the other bottomland hardwood samples than for the ground layer ordination. The overstory ordination arrangement also corresponded closely with the ground-layer TWINSPAN classification, represented by the symbols displayed in Figures 1, 2 and 3).

Mean overstory density (Table 1) for mesic and Bottomland Oak Flat communities was generally in the range of the 300-600 stems/ha that was reported for the Big thicket (Marks & Harcombe 1981). Maximum density was observed among the Cypress-Water elm Swamps, while Open-canopied Cypress Swamps in deeper water showed the lowest density (Table 1). Excluding the open Erianthus sites, mean basal area for the bottomland Oak Flats (28 [m.sup.2]/ha, Table 1) was similar to the 29 [m.sup.2]/ha reported on equivalent sites from the Big Thicket (Marks & Harcombe 1981), but was lower than that reported from other southeastern bottomland forests (Robertson et al. 1978). Basal area was higher for the swamps (Table 1), in part because of buttressing of the stems. However, only one Closed-canopy Cypress Swamp sample (90 [m.sup.2]/ha) approached the 130 [m.sup.2]/ha reported for the cypress/tupelo stand described in Marks & Harcombe (1981).


The first axis of a DCA ordination of the 119 survey areas based on ground layer species arranged samples in a pattern similar to that of the transect data set (Figure 4). As with the transect data, the ordering of sites reflected a gradient from semi-permanently flooded swamps to seasonally-flooded oak flats to temporarily-flooded natural levees and terraces. However, it was not possible to identify subgroups of Bottomland Oak Flats with this less precise and coarser-scale data set. The first axis of a DCA ordination based of the abundance ranks of overstory trees from the 119 survey sites was highly correlated ([r.sup.2] = 0.95, p < 0.01) with the first axis of the survey ground layer ordination and reflected a gradient similar to that of the ground layer. The second axis of the ground layer ordination was largely related to variation among mesic sites (Figure 4). Sites with high axis 2 scores were associated with steep slopes and narrow ravines near the Caddo Lake State Park campground and headquarters area while terraces and natural levees from higher portions of the Cypress Bayou floodplain had low scores. Ravines and slopes did not occur on the transects which were wholly within the floodplain.

Ecological species groups. -- DCA orders species simultaneously with samples, displaying near one another in ordination space those species that occur in similar sites and are presumably ecologically similar (Hill & Gauch 1980). Since species ordinations are based on same information that samples ordinations are (Gauch 1982), the ordering of the 136 ground layer species encountered in the 30 transect plots along the first DCA axis was strongly related to water depth as was the first axis of the samples ordination. Species characteristic of well drained mesic sites (Chasmanthium sessiliflorum (Poir) Yates) had low first axis scores, species found on seasonally flooded oak flats (Carex joorii) had intermediate scores, and swamp species (Ceratophyllum demersum L.) had high scores (Figure 5). TWINSPAN classified ground layer species into five ecological species groups; groups of species that respond similarly to environmental factors and tend to occur together on similar types of sites (Mueller-Dombois & Ellenberg 1974; Barnes et al. 1982). The species groups are listed in Table 2 and their species plotted in ordination space in Figure 5. Only 44 species listed as indicator species or observed to be strong preferential species (present on more than 57% of the samples for a given TWINSPAN division) were included in the species group list. Twenty-eight of the 136 species encountered had a mean coverage of [greater than or equal to]0.75% for at least one community type (Table 3). Most of these dominant species also had indicator value as a comparison of Tables 2 and 3 reveals. Species groups followed the flooding gradient from the mesic Chasmanthium group to the aquatic Ceratophyllum group (Table 1, Figure 5). Although 363 species were encountered in the much larger survey data set, the classification and list of indicator species from the 119 survey areas was similar to that from the plots. Overstory species for the transect plots were likewise classified by TWINSPAN (Table 1).


It is also informative to observe changes in abundance for individual species along an environmental gradient (Figure 6). Generally the distributions species along the flooding gradient in this study (expressed by the first DCA axis) are consistent with the bell-shaped Gausian curves characteristic of the distributions of many species along an environmental gradient (Gauch 1982). For example, cypress (Taxodium distichum) seedlings plotted along the first ground layer DCA axis for the survey data reach their peak abundance in shallow cypress-water elm swamps and decline on drier sites and in deeper water where germination is limited by lack of soil exposure (Figure 6). Likewise, Ceratophyllum demersum is abundant on swamp sites but becomes progressively rarer on drier sites. Carex joorii shows peak abundance near the middle of the gradient among the bottomland oak flats, while Chasmanthium sessiliflorum is most abundant on mesic sites. Distributions for other species can be inferred from observing rows in Tables 1-3.

Community types. -- The 119 survey samples were classified into six community types (Rich-Mesic Slopes and Creek Bottoms, Mesic Bottomland Ridges, Bottomland Oak Flats, Cypress Water-elm Swamps, Closed Canopy Cypress Swamps and Deep-Water Open Cypress Swamps) on the basis of multivariate analysis of both overstory and ground flora (Figure 4). This classification forms the basis for a map of the wetland vegetation of Caddo Lake State Park and Wildlife Management Area (Figure 7). Fifty-two percent of the samples (Figure 4) and 56% of the area mapped in Figure 7 consisted of one of the three Taxodium dominated swamp types (Table 4); a much higher proportion of cypress swamps than that of any other Texas wetland landscape the authors observed.

(1) Rich Mesic Slopes and Creek Bottoms: These hardwood-dominated communities occurred on steep sheltered slopes, creek bottoms, and ravines along the edge of the Caddo Lake basin. The community was mainly observed in the campground and headquarters portion of Caddo Lake State Park, which according to USGS topographic maps, has some of the greatest topographic relief on the Caddo watershed. Ground flora included members of the Chasmanthium group as well as a rich assemblage of mesophytic plants including Florida maple (Acer barbatum Michx.), cross vine (Bignonia capreolata L.), Christmas fern (Polystichum acrostichoides (Michx.) Shott.) and Canada snakeroot (Sanicula canadensis L.) that were not found elsewhere. Unlike the remaining five types, these are not wetlands and they did not occur within the Caddo Lake floodplain (<54m elevation). However, these sites have better moisture relations than the adjacent uplands and were possibly better protected from presettlement fires by their topographic location.


(2) Mesic Bottomland Ridges and Flats: These rarely-flooded communities, found within the Cypress Bayou floodplain (<54 m elevation), occurred on the crests of natural levees and meander scrolls that developed along stream channels and along the gradually sloping Pleistocene low terraces adjacent to the lower wetlands. Stands were commonly dominated by loblolly pine (Pinus taeda L.), sweetgum (Liquidambar styraciflua L.), southern red oak (Quercus falcata Michx.) and other hardwoods. Important shrubs included Ilex decidua Walt and Vaccinium arboreum Marsh. Ground vegetation was dominated by members of the Chasmanthium and Carex joorii groups especially Chasmanthium sessilflorum, sedge (Carex joorii) and greenbriar (Smilax spp.).

(3) Bottomland Oak Flats: These communities occurred on the lower portions of islands and levees that gently slope down into the wetter swamps. Sites were seasonally flooded and soils poorly drained, but were above water for most of the year. Most stands were dominated by willow oak (Q. phellos), with lesser amounts of overcup oak (Q. lyrata), blackgum (Nyssa sylvatica Marsh.) and sweetgum (Liquidambar styraciflua). Wetter stands transitional to swamps were generally dominated by Q. lyrata. Important shrubs included Crataegus opaca Hook & Arn., Diospyros virginiana L., Styrax americana Lam., Ilex decidua and on lower sites, Foresteria acuminata (Michx.) Poir. Ground vegetation included Carex joorii, and other members of the C. joorii group. Members of the Brunnichia group were also common, especially on wetter sites. Bottomland Oak Flats can be divided into three community subtypes: (a) willow oak flats, dominated by willow oak and C. Joorii, (b) sites on slight mounds with a somewhat open canopy of willow oak and a dense ground cover of Erianthus strictus and (c) overcup oak-dominated flats on lower, wetter sites. The environmental and/or historic factors maintaining the canopy openings of the Erianthus subtype are not known, but the groundflora appears to be transitional between Mesic Bottomland Ridges and the remaining Bottomland Oak flats (Figure 1, Table 1).

(4) Cypress Water-elm Swamps: Cypress-water elm communities were transitional between oak flats and swamps. Occurring slightly below the current normal pool elevation of the lake (Figure 2), they were flooded for much of the year, but had significant periods of exposure. Cypress dominated the overstory but there was an abundant sub-canopy of water elm (Planera aquatica). Shrubs were sparse, but included Foresteria acuminata, and Cephalanthus occidentalis L. The Ground layer was sparse during low water periods, largely represented by the Brunnichia group. When flooded, species from the Ceratophyllum group were found.

(5) Closed-Canopy Cypress Swamps: These swamps, often associated with abandoned stream channels, occurred lower on the landscape than the cypress water-elm swamps, and were usually under water. The high density of cypress indicated that historically there were periods in which the sites were exposed enabling significant regeneration. The tree layer was pure baldcypress (T. distichum). Shrubs were limited to scattered Cephalanthus occidentalis growing from stumps and logs. By mid summer, the surface was covered by members of the Ceratophyllum group including Egeria densa, fanwort (Cabomba caroliniana) duckmeat (Spirodela punctata (Meyer) Thomps.), water meal (Wolffia columbiana Karst) and water fern (Azolla caroliniana Willd.).

(6) Open (Deep Water) Cypress Swamps: In deeper water, cypress stands became less dense. The deepest swamps were reduced to scattered trees growing in open water covered by floating and submersed aquatic plants. As with the previous community, members of the Ceratophyllum group dominated the surface. Dense colonies of Nuphar luteum were also common.


Caddo Lake wetland vegetation corresponds to a flooding and elevation gradient from mesic, rarely-flooded terraces and levees to nearly continuously flooded cypress swamps. While removal of the Great Raft and dam construction may have modified hydrology, modern vegetation appears to be closely adjusted to the current flooding regime as evidenced by the correspondence between the normal pool elevation of 51.36 m and the boundary between semi-permanently flooded swamps and seasonally flooded Bottomland Oak Flats (Figure 2).

Other bottomland and swamp vegetation studies. -- Other authors have described hydrologic gradients and plant communities comparable to those at Caddo Lake from other locations in the South. Crouch & Golden (1997) surveyed the flora of an Alabama bottomland forest where a seasonally flooded floodplain was dominated by bottomland hardwoods including Quercus phellos, Q. pagoda Raf., Q. nigra L and L. styraciflua, with Q. lyrata and Carya aquatica (Michx f.) Nutt. dominating a somewhat lower slough. An area of slopes and ravines adjacent to the floodplain contained a rich mixture of hardwoods similar to the Rich Mesic Steep Slopes of Caddo Lake.

Thompson (1980) described mesic terrace bottoms, mixed softwood levees, oak hardwood bottoms and shallow Taxodium distichum swamps from a floodplain forest in Missouri. The recently-formed mixed softwood levees, characterized by pioneer trees such as Salix nigra Marsh. were not observed at Caddo Lake, possibly because the high-energy stream flows responsible for creating such landforms have not been present on Big Cypress Bayou during the last 200 yr as a result of its partial impoundment. Although many southern bottomland species were present, Thompson (1980) also reported a number of species such as Acer saccharinum L. and Quercus palustris Muenchh. from this more northerly site that were not found at Caddo Lake.

Texas and Louisiana studies. -- Marks & Harcombe (1981) identified Floodplain Hardwood Forest and Swamp Cypress-Tupelo Forest from the Big Thicket of southeastern Texas. Their two highest Floodplain Hardwood stands are roughly equivalent to Mesic Bottomland Ridges, except that Fagus grandifolia, Ehrh., abundant in the Big Thicket sites, was absent at Caddo. The remaining Big Thicket Floodplain Hardwood stands correspond to Caddo Lakes's Bottomland Oak Flats except that Quercus lyrata is less important for the Big Thicket and Carpinus caroliniana Walt., abundant in the Big Thicket understory, is absent at Caddo. Nyssa aquatica L., dominant in the Big Thicket swamp, was absent from the corresponding swamps at Caddo Lake.

Mathies et al (1983) described a series of wetland communities from Cunningham Brake in Nachitoches Parish, Louisiana. Taxodium distichum and N. aquatica, were associated with areas flooded for most of the year and lowland hardwoods including Q. phellos, Q. nigra, Q. laurifolia Michx., Q. falcata (=Q. pagoda ?), and L. styraciflua dominated seasonally flooded bottomlands. Open sand bars and stream edges along Kisatchie Bayou, characterized by pioneer species such as Betula nigra L. and Salix nigra, are equivalent to the softwood levees of Thompson (1980), but with the exception of a few sites observed on the spoils of artificial ditches, were not observed at Caddo Lake.

Delcourt (1976) described cane ridges, levee slope hardwoods (equivalent to Caddo's Bottomland Oak Flats), and cypress/tupelo swamps in her reconstruction of presettlement northern Louisiana vegetation. It is notable that although presettlement accounts described giant cane Arundinaria gigantea (Walt.) Muhl. as dominating natural levees, (equivalent to Mesic Bottomland Ridges), cane was uncommon on such sites at Caddo Lake.

Relation to regional classification systems. -- Van Lear & Jones (1987) developed an ecological site classification system for terraces and floodplains associated with the Savannah River in South Carolina. Resembling the Caddo Lake communities, ecological land units are arranged along a flooding gradient from deep water cypress swamps, shallow cypress and tolerant hardwood swamps, laurel oak-water oak floodplains characterized by prolonged flooding, and loblolly pine-sweetgum-oak communities on terraces above the active floodplain.

Larson et al. (1981) and Christenson (1988) described a "zonal classification" of southeastern alluvial wetlands which divides them into five zones based on the frequency and duration of flooding. Mesic Bottomland Ridges essentially correspond to zone V (infrequently flooded transition to uplands), Bottomland Oak Flats correspond to zone IV (seasonally flooded forests of backwaters and flats), Overcup oak-dominated lower Bottomland Oak Flats correspond to Zone III (Lower hardwood swamp forest), and the three swamp types described for Caddo lake correspond to zone II (intermittently exposed river swamp forest). Zone I represents the permanent water of river channels and lakes.

The SAF Forest Cover Type classification (SAF 1980), widely used among forestry professionals in the US, places Mesic Bottomland Ridges in the Loblolly Pine-Hardwood cover type (SAF type 82). Higher bottomland Oak Flats (Erianthus openings and Willow Oak Flats best fit the Sweetgum-Willow Oak cover type (SAF type 92), but also resemble the Willow Oak- Water Oak- Diamondleaf Oak type (SAF type 88). Overcup oak-dominated Bottomland Oak Flats correspond to the Overcup Oak- Water Hickory cover type (SAF type 96). The three cypress swamp community types of Caddo lake belong to the Baldcypress cover type (SAF type 101), although Planera aquatica, common at Caddo Lake, is described as an understory associate for the Baldcypress/Tupelo cover type (SAF type 102), but not for the pure Baldcypress type.

In 1998, The Nature Conservancy released a draft of a comprehensive classification of terrestrial vegetation for the southeastern United States. The system classifies vegetation into Alliances which are in turn made up of one or more Associations (Weakley et al. 1998). While the Caddo Lake communities can generally be placed into an appropriate Alliance, in many cases an Association has yet to be described that closely fits them. Additional quantitative case studies such as this one will be invaluable in refining the classification and in increasing our understanding of regional patterns of vegetation.

Mesic bottomland Ridges appear to best fit in the Pinus taeda-Quercus (phellos/nigra/laurifolia) Temporarily Flooded Forest Alliance (I.C.3.N.b.070), but they also have affinity to the Pinus taeda-Quercus (pagoda/shumardii/michauxii) Temporarily Flooded Forest Alliance (I.C.3.N.b.060). The willow oak-dominated Bottomland Oak Flats best fit into the Quercus phellos Seasonally Flooded Alliance (I.B.2.N.e.130), but the Alliance usually occurs in upland depressions, rather than bottomlands. Willow Oak Flats also resemble the Quercus (phellos/nigra/laurifolia) Temporarily Flooded Forest Alliance (I.B.2.N.d.250) although there are understory differences and Big Cypress Bayou is not a "blackwater or low sediment/nutrient river" as is associated with this Alliance (Weakley et al. 1998). The lower, overcup oak-dominated Bottomland Oak Flats belong to the Quercus lyrata (Carya aquatica) Seasonally Flooded Forest Alliance (I.B.2.N.e.100) and possibly to the Quercus lyrata/Liquidambar styraciflua/Foresteria acuminata Forest Association. Cypress-Water elm Swamps best fit the Taxodium distichum/Nyssa Seasonally Flooded Forest Alliance (I.B.2.N.e.190) although Nyssa spp. were absent from Caddo swamps. Weakley et al. (1998) also described a Planera aquatica Seasonally Flooded Forest Alliance (I.B.2.N.e.090) with an understory similar to that observed in Cypress-Water elm Swamps at Caddo Lake. However, the emergent cypress canopy at Caddo is too dense to fit the description for this Alliance. Closed Canopy and Open Cypress Swamps belong to the Taxodium distichum Semi-permanently Flooded Forest Alliance (I.B.2.N.f.060) and to the Taxodium distichum/Lemna minor. Association, although at Caddo Spirodela spp. and Wolfia spp. dominate the surface rather than Lemna spp. Some of the larger openings in the Open Cypress Swamps (such as portions of Clinton Lake, Figure 7) belong to the Nuphar lutea Permanently Flooded Herbaceous Alliance (V.C.2.N.a.040) where Nuphar dominates, or to the southern variant of the Potomogeton spp./Ceratophyllum spp./Elodea spp. Permanently Flooded Herbaceous Alliance (V.C.2.N.a.065).
Table 1: Mean basal area ([m.sup.2]/ha) and absolute density (stems/ha,
in parentheses) for each tree species > 10 cm dhh encountered in 29
plots along a transect from the Caddo Lake State Park and Wildlife
Management Area. The 30th plot, a heterogeneous plot spanning a mesic
slope and a cypress swamp (See Figure 4) was omitted from the
tabulation. Samples are classified into community types on the basis of
ground layer TWINSPAN and DCA while species are classified into groups
on the basis of overstory TWINSPAN. MB = Mesic Bottomland Ridges, BFE =
Bottomland Oak Flats: Erianthus openings, BFW = Bottomland Oak Flats:
Willow oak-dominated, BFO = Bottomland Oak Flats: Overcup oak-dominated,
CW = Cypress-Water elm Swamps, CS = Closed-canopy Cypress Swamps, and
OC = Open Cypress Swamps.


Acer rubrum L. 0.38 (43) 0 (0) 0 (0) 0 (0)
Celtis laevigata
 Willd. 0.05 (3) 0 (0) 0 (0) 0 (0)
Nyssa sylvatica
 Marsh. 1.52 (43) 0 (0) 0 (0) 0 (0)
Pinus taeda L. 11.80 (47) 0 (0) 0 (0) 0 (0)
Quercus nigra L. 7.01 (80) 1.20 (4) 0 (0) 0 (0)
Quercus pagoda Raf. 1.48 (13) 0 (0) 0 (0) 0 (0)
Sassafras albidum
 (Nutt.) Nees 0.18 (7) 0 (0) 0 (0) 0 (0)
Vaccinium arboreum
 Marsh. 0.04 (3) 0 (0) 0 (0) 0 (0)
Betula nigra L. 0.69 (17) 0.92 (28) 0.40 (6) 0 (0)
Carya aquatica
 (Michx.f.) Nutt. 0.08 (7) 0.48 (12) 0.64 (9) 0 (0)
Crataegus opaca
 Hook. & Arn. 0 (0) 0.12 (8) 0.20 (11) 0 (0)
 styraciflua L. 3.93 (110) 3.44 (92) 3.32 (66) 0 (0)
Quercus phellos L. 1.59 (20) 13.50 (116) 11.51 (183 0 (0)
Diospyros virginiana
 L. 0.11 (3) 0 (0) 0.05 (3) 0.21 (20)
Gleditsia aquatica
 Marsh. 0 (0) 0 (0) 0 (0) 0.72 (27)
Quercus lyrata Walt. 1.32 (17) 1.61 (36) 9.10 (51) 17.46 (153)
Salix nigra Marsh. 0 (0) 0 (0) 0 (0) 1.55 (7)
Planera aquatica
 Marsh. 0 (0) 0 (0) 0.11 (9) 5.74 (173)
Taxodium distichum
 (L.) L. Rich 0 (0) 0 (0) 2.27 (23) 2.79 (60)
TOTAL 30.18 (413) 21.27 (296) 27.59 (361) 28.47 (440)


Acer rubrum L. 0 (0) 0 (0) 0 (0)
Celtis laevigata
 Willd. 0 (0) 0 (0) 0 (0)
Nyssa sylvatica
 Marsh. 0 (0) 0 (0) 0 (0)
Pinus taeda L. 0 (0) 0 (0) 0 (0)
Quercus nigra L. 0 (0) 0 (0) 0 (0)
Quercus pagoda Raf. 0 (0) 0 (0) 0 (0)
Sassafras albidum
 (Nutt.) Nees 0 (0) 0 (0) 0 (0)
Vaccinium arboreum
 Marsh. 0 (0) 0 (0) 0 (0)
Betula nigra L. 0 (0) 0 (0) 0 (0)
Carya aquatica
 (Michx.f.) Nutt. 0 (0) 0 (0) 0 (0)
Crataegus opaca
 Hook. & Arn. 0 (0) 0 (0) 0 (0)
 styraciflua L. 0 (0) 0 (0) 0 (0)
Quercus phellos L. 0 (0) 0 (0) 0 (0)
Diospyros virginiana
 L. 0 (0) 0 (0) 0 (0)
Gleditsia aquatica
 Marsh. 0.88 (40) 0 (0) 0 (0)
Quercus lyrata Walt. 1.45 (10) 0 (0) 0 (0)
Salix nigra Marsh. 0 (0) 0 (0) 0 (0)
Planera aquatica
 Marsh. 7.41 (360) 0 (0) 0 (0)
Taxodium distichum
 (L.) L. Rich 43.41 (480) 58.01 (450) 19.08 (170)
TOTAL 53.15 (890) 58.01 (450) 19.08 (170)

Table 2. Ecological species groups for Caddo Lake wetland communities.
Groups are based on a TWINSPAN classification of 136 ground flora
species from 30 plots along a transect from the Caddo Lake State Park
and Wildlife Management Area, except that the TWINSPAN did not separate
the overcup oak-dominated subtype from the remaining Bottomland Oak
Flats. The division is recognized on the basis of DCA. Only 44 species
with strong indicator preference are listed. Groups are named after a
representative species that has both a high frequency of occurrence in
the community types characterized by the group and a high fidelity to
those communities. Symbols after the group name (listed in order of
importance) refer to the communities in which the species are most often
found: MB = Mesic Bottomland Ridges, BFE = Bottomland Oak Flats:
Erianthus openings, BFW = Bottomland Oak Flats: Willow oak-dominated,
BFO = Bottomland Oak Flats: Overcup oak-dominated, CW = Cypress-Water
elm Swamps, CS = Closed-canopy Cypress Swamps, and OC = Open Cypress
Swamp. Species codes, used in Figure 5, are formed from the scientific
name of each species.

Species Frequency of occurrence (% of sites)
1) Chasmanthium group: MB, BFE Code MB BFE BFW BFP CW CS OC

Acalypha virginica L. ACVI 71 40 13 0 0 0 0
Acer rubrum L. ACRU 71 0 0 0 0 0 0
Ascyrum hypericoides L. HYHY 100 60 0 0 0 0 0
Berchemia scandens (Hill) K.
 Koch BESC 86 40 0 0 0 0 0
Botrychium biternatum
 (Savig.) Underw. BOBI 71 0 0 0 0 0 0
Chasmanthium sessiliflorum
 (Poir) Yates CHSE 71 0 0 0 0 0 0
Dichanthelium dichotomum (L.)
 Gould DIDI 71 60 0 0 0 0 0
Ilex opaca Soland. in Ait. ILOP 57 0 0 0 0 0 0
Oxalis dillenii Jacq. OXDI 86 0 0 0 0 0 0
Pinus taeda L. PITA 100 20 0 0 0 0 0
Quercus nigra L. QUNI 57 0 0 0 0 0 0
Smilax bona-nox L. SMBN 100 20 0 0 0 0 0
Smilax rotundifolia L. SMRO 100 60 13 0 0 0 0
Trachelospermum difforme
 (Walt.) Gray TRDI 86 60 38 0 0 0 0
Toxicodendron radicans (L.) O.
 Ktze. TORA 86 20 0 0 0 0 0
Ulmus alata Michx. ULAL 71 0 0 0 0 0 0

2) Carex joorii Group: BFE, BFW, BFO, MB.

Carex joorii Bailey CAJO 57 100 100 100 0 0 0
Diospyros virginiana L. DIVR 57 60 100 100 0 0 0
Erianthus strictus Baldw. ERST 71 100 62 66 0 0 0
Hibiscus moscheutos L. HIMO 14 80 62 66 0 0 0
Liquidambar styraciflua L. LIST 86 80 75 33 0 0 0
Styrax americanus Lam. STAM 43 100 62 66 0 0 0

3) Brunnichia group: CW, BF, CS, BFO, BFW, BFE.

Boehmeria cylindrica (L.) Sw. BOCY 29 60 50 33 100 100 50
Brunnichia ovata (Walt.)
 Shinners BROV 29 100 100 100 100 100 0
Cephalanthus occidentalis L. CEOC 0 40 25 33 50 100 0
Quercus lyrata Walt. QULY 29 20 55 66 50 0 0
Planera aquatica (Walt.) J. F.
 Gmel. PLAQ 14 100 87 100 100 50 0
Saururus cernuus L. SACE 0 0 13 66 100 0 0
Taxodium distichum (L.) L.
 Rich. TADI 29 80 87 100 100 67 0
Vitis aestivalis Michx. VIAE 14 100 75 33 100 0 0

4) Nuphar group: OS, CS, CW. MB BFE BFW BFO CW CS OC

Ludwigia palustris (L.) Ell. LUPA 0 0 0 33 50 67 100
Ludwigia peploides
 (Kunth in H.B.K.) Raven LUPE 0 0 0 0 0 0 100
Myriophyllum aquaticum (Vell.)
 Verd. MYAQ 0 0 0 0 50 67 100
Nelumbo lutea (Willd.) Pers. NELU 0 0 0 0 0 34 100
Nuphar luteum (L.) Sibth. & Sm. NULU 0 0 0 0 0 34 100

5) Ceratophyllum group: CS, OS, CW

Azolla caroliniana Willd. AZCA 0 0 0 0 0 67 50
Cabomba caroliniana Gray CACA 0 0 0 0 0 67 50
Ceratophyllum demersum L. CEDE 0 0 0 0 0 67 50
Egeria densa Planch. EGDE 0 0 0 13 100 100 50
Hydrolea uniflora Raf. HYUN 0 0 0 33 100 100 100
Limnobium spongia
 (Bosc.) L.C. Rich ex Steud. LISP 0 0 0 0 100 100 50
Potomogeton nodosus Poir. PONO 0 0 0 0 100 67 0
Spirodela punctata (Meyer.)
 Thomps. SPPU 0 0 0 0 0 100 100
Wolffia columbiana Karst. WOCO 0 0 0 0 0 100 100

Table 3. Mean percentage cover of major species (all species encountered
with mean cover > 0.75% in at least 1 community type) for species from
30 plots along a transect from the Caddo Lake State Park and Wildlife
Management Area. Communities and species are classified on the basis of
TWINSPAN except that the overcup oak-dominated subtype is separated from
the remaining Bottomland Oak Flats on the basis of DCA. MB = Mesic
Bottomland Ridges, BFE = Bottomland Oak Flats: Erianthus openings, BFW =
Bottomland Oak Flats: Willow oak-dominated, BFO = Bottomland Oak Flats:
Overcup oak-dominated, CW = Cypress-Water elm Swamps, CS = Closed-canopy
Cypress Swamps, and OC = Open Cypress Swamp.

 Mean percentage cover

Acer rubrum 0.8 0 0 0 0 0 00
Berchemia scandens 1.0 0 0 0 0 0 0
Botrychium biternatum 1.0 0 0 0 0 0 0
Chasmanthium sessiliflorum 2.1 0 0 0 0 0 0
Dichanthelium dichotomum 1.7 0.1 0 0 0 0 0
Smilax bona-nox 0.8 0 0 0 0 0 0
Smilax rotundifolia 2.4 0.3 0.1 0 0 0 0
Ulmus alata 1.7 0 0 0 0 0 0
Carex joorii 7.7 16.0 16.1 0.4 0 0 0
Erianthus strictus 2.9 19.3 0.2 0.1 0 0 0
Ilex decidua Walt. 0.3 1.3 0 0.2 0 0 0
Liquidambar styraciflua 0.5 0.8 0.2 0 0 0 0
Brunnichia ovata 0.5 0.4 1.7 3.7 0.5 0.1 0.1
Tillandsia usneoides (L.) L. 15.1 2.3 2.4 8.5 25.0 25.0 0.3
Cyperus erythrorhyzos Muhl. 0 0 0 0 0 5.0 0
Cyperus odoratus L. 0 0 0 0 0 1.0 0.1
Ludwigia palustris 0 0 0 0 0.1 1.1 0.1
Ludwigia peploides 0 0 0 0 0 1.7 0.3
Lindernia dubia (L.) Penn. 0 0 0 0 0.1 6.7 0
Nelumbo lutea 0 0 0 0 0 16.7 37.5
Nuphar lutea 0 0 0 0 0 10.0 30.0
Cabomba caroliniana 0 0 0 0 0 1.0 35.0
Ceratophyllum demersum 0 0 0 0 0 10.0 20.0
Egeria densa 0 0 0 0 0.1 16.7 42.5
Hydrolea uniflora 0 0 0 0 0.5 28.5 0.3
Limnobium spongia 0 0 0 0 0.8 21.7 20.5
Spirodela punctata 0 0 0 0 0 14.0 20.5
Wolffia columbiana 0 0 0 0 0 20.0 48.0

Table 4. Approximate aerial extent of wetland community types from the
Caddo Lake State Park and Wildlife Management Area (estimated from the
aerial photograph that forms the basis for the map in Figure 7). RM =
Rich-Mesic Slopes and Ravines, MB = Mesic Bottomland Ridges, BO =
Bottomland Oak Flats, CW = Cypress-Water elm Swamps, CS = Closed-canopy
Cypress Swamps, OC = Open Cypress Swamp, and DI = Disturbed Uplands.

Community Approx. hectares Percentage of area

RM 48 1.4
MB 767 21.8
BO 596 17.0
CW 906 25.8
CS 319 9.1
OC 752 21.4
DI 121 3.5
Total 3509 100.0


The authors acknowledge the National Biological Service for their support of this study. We also thank Jim Neal, Jim Johnston, William Sheffield, Tommy Pritchard, Edward Hughs and Virginia Burkett for their contributions. Dr. P. A. Harcombe and three anonymous reviewers provided valuable comments on the manuscript.


Abernathy, V., & R. E. Turner. 1987. Forested wetlands: 1940-1980. Bioscience 37:721-727.

A.I.D. Associates. 1993. Caddo Lake: contoured depth map. A.I.D. Associates, Dallas Texas.

Bailey, R. G., P. E. Avers, T. King & W.H. McNab, eds. 1994. Ecoregions and subregions of the United States (map). U.S. Geological Survey, Washington D.C. Scale 1:7,500,000; colored.

Barnes, B. V, K. S. Pregitzer, T. A. Spies & V. H. Spooner. 1982. Ecological forest site classification. Journal of Forestry, 80:493-498.

Barrett, M. L. 1995. Sedimentary record of a 19th century Red River Raft lake: Caddo Lake, Louisiana. The Compass, 72:3-11.

Bedinger, M. S. 1981. Hydrology of bottomland hardwood forests of the Mississippi embayment. In Wetlands of bottomland hardwood forests (J. R. Clark & J. Benforado eds.), Elsevier, Amsterdam, pp. 161-176.

Chambless, L. F. & E. S. Nixon. 1975. Woody vegetation--soil relations in a bottomland forest in east Texas. Texas Journal of Science, 26:407-416.

Christenson N. L. 1988. Vegetation of the southeastern coastal plain. In North American Terrestrial Vegetation (M. G. Barbour & W. D. Billings eds.). Cambridge University Press, Cambridge. Pp. 317-355.

Crouch, V. E. & M. S. Golden. 1997. Floristics of a bottomland forest and adjacent uplands near the Tombigbee River, Choctaw County, Alabama. Castanea, 62:219-238.

Dahmer, F. 1988. Caddo Was ... a short history of Caddo Lake. Everett, Bossier City, Louisiana, 82 pp.

Davis, T. J. (ed) 1994. The Ramsar Convention Manual: A guide to the Convention on Wetlands of International Importance especially as waterfowl habitat. Ramsar Convention Bureau, Gland, Switzerland, 194 pp.

Delcourt, H. R. 1976. Presettlement vegetation of the north of Red River Land District, Louisana. Castanea, 41:122-139.

Frye, R. G. 1987. Bottomland Hardwoods: Current supply, status, habitat quality and future impact from reservoirs. Pp. 24-28 in Bottomland Hardwoods in east Texas. Proceedings of an interagency workshop on the status & ecology of bottomland hardwoods in Texas (C. A. Mc Mahan & R. G. Frye, eds.). Texas Parks & Wild. Dept., Austin. PWD-RP-133-3/87, 387 pp.

Gleason, H. A. 1926. The individualistic concept of the plant association. Bulletin of the Torrey Botanical Club 53:1-20.

Gosselink J. G, L. C. Lee & T. A. Muir eds. 1990. Ecological Process and Cumulative Impacts: illustrated by bottomland hardwood wetland Ecosystems. Lewis Publishers, Chelsea, MI. 708pp.

Gauch, H. G., Jr. 1982. Multivariate analysis in community ecology. Cambridge University Press, New York. 298 pp.

Harcombe, P. A., J. S. Glitzenstein, R. G. Knox, S. L. Orzell & E. L. Bridges. 1993. Vegetation of the longleaf pine region of the west Gulf coastal plain. In Proceedings of the 18th Tall Timbers Fire Ecology Conference, the longleaf pine ecosystem: ecology, restoration, and management (S. M. Hermann, ed.). Tall Timbers Research Station, Tallahassee, Florida, 418 pp.

Hatch S. L, K. N. Gandhi & L. E. Brown. 1990. Checklist of the vascular plants of Texas. Texas Agricultural Experiment Station, College Station, Texas, 158pp.

Hine D. N. & E. S. Nixon. 1992. Preliminary checklist of the ferns and herbaceous flowering plants of Caddo Lake State Park and vicinity. In East Texas and its many ecosystems, Proceedings, Native Plant Society of Texas, Georgetown, Texas, 118 pp.

Hill, M. O. 1979a. DECORANA--a FORTRAN program for detrended correspondence analysis and reciprocal averaging. Ithaca, New York. Ecology and systematics, Cornell University.

Hill M. O. 1979b. TWINSPAN--a FORTRAN program for arranging multivariate data in an ordered two-way table by classification of the individuals and attributes. Ithaca, New York. Ecology and systematics, Cornell University.

Hill, M. O. & H. G. Gauch, Jr. 1980. Detrended correspondence analysis: an improved ordination technique. Vegetation, 42:47-50.

Jongman, R. H. G., C. F. ter Braak & O. F. R. van Tongren. eds. 1995. Data analysis in community and landscape ecology, Cambridge University Press, Cambridge, 299 pp.

Kartesz, J. T. 1994. A synonymized checklist of the vascular flora of the United States, Canada and Greenland, 2nd ed. Timber Press, Portland, 622pp.

Larkin, E. J. & G. W. Bomar. 1983. Climatic Atlas of Texas. Texas Dept. of Water Resources, Austin, Texas, 151 pp.

Larson, J. S., M. S. Bedinger, C. F. Bryan, S. Brown, R. T. Huffman, E. L. Miller, D. G. Rhodes & B. A. Touchet. 1981. Transition from wetlands to uplands in southeastern bottomland hardwood forests in Wetlands of bottomland hardwood forests, (J. R. Clark and J. Benforado, eds.), Elsvier, Amsterdam, pp. 225-273.

Ludwig J. A. & J. F. Reynolds. 1988. Statistical ecology. Wiley, New York, 337 pp.

MacRoberts, M. H. 1979. Checklist of the plants of Caddo Parish, Louisiana. Bull. 1, Museum of Life Sciences. Louisiana State Univ., Shreveport, Louisiana, 54 pp.

Marks, P. L & P. A. Harcombe 1981. Forest vegetation of the Big Thicket, Southeast Texas. Ecological Monographs, 52:287-305.

Mathies, P. S., W. C. Holmes & A. S. Allen. 1983. The vascular flora of Cunningham Brake, a cypress-gum swamp in Natchitoches Parish, Louisiana. Castanea, 48:24-31.

Miller, N. A. 1990. Effects of permanent flooding on bottomland hardwoods and implications for water management in the Forked Deer River floodplain. Castanea, 55:106-112.

Misch W. J. & J. G. Gosselink. 1993. Wetlands. 2nd edition. Van Nostrand Reinhold, New York, 722 pp.

Mueller-Dombois, D. & H. Ellenberg. 1974. Aims and methods of vegetation ecology. New York, Wiley, 547pp.

Navid, D. 1989. The international law of migratory species: the Ramsar Convention. Natural Resources Journal, 29:1001-10016.

Nixon, E. S., G. A. Sullivan, S. D. Jones, G. D. Jones & J. K. Sullivan. 1990. Species diversity of woody vegetation in the Trinity River basin, Texas. Castanea, 55:97-105.

SAF (Society of America Foresters). 1980. Forest Cover Types of the United States and Canada (F. H. Erye, ed.). Society of American foresters, Washington D.C., 148 pp.

Shankman D. 1991. Forest regeneration on abandoned meanders of a coastal plain river in western Tennessee. Castanea, 56:157-167.

Shankman D. & L. G. Drake. 1990. Channel migration and regeneration in baldcypress in western Tennessee. Physical Geography, 11:343-352.

Sheffield, W. J. 1995. A summer-fall ecological reconnaissance of the Big Cypress Bayou watershed, Texas and Louisiana. Texas Parks & Wildlife Dept. Austin, 73 pp.

Shmida, A. 1984. Whittaker's plant diversity sampling method. Israel Journal of Botany, 33:41-46.

Steele, R. G. D. & J. A. Torrie. 1980. Principals and procedures of statistics: a biometrical approach. 2nd edition. Mc.Graw-Hill, New York, 633 pp.

Taylor, P. D., J. R. Harwood, V. Silvey, L. Dickinson & T. Duguid. 1996. Caddo Lake State Park and Wildlife Management Area. A conceptual master plan. Center for Environmental Design Research, University of Texas-Arlington, 46 pp.

Thompson, R. L. 1980. Woody vegetation and floristic affinities of Mingo Wilderness Area, a northern terminus of southern floodplain forest, Missouri. Castanea, 45:194-212.

US Army Corps of Engineers. 1994b. Geomorphic Investigations: Red River waterway project, Shreveport, Louisiana to Dangerfield, Texas reach re evaluation study. UASCOE, Vicksburg, Mississippi, 142 pp.

US Fish and Wildlife Service. 1996. Fact Sheet: Ramsar Convention on wetlands of international importance especially as waterfowl habitat. US Fish and Wildlife Service, Office of International Affairs, Arlington, Virginia, 2 pp.

Van Lear, D. H. & S. M. Jones. 1987. An example of site classification in the southeastern coastal plain based on vegetation and land type. Southern Journal of Applied Forestry, 11:23-28.

Weakley, A. S., K. D. Patterson, S. Landaal, M. Pyne and others (compilers). 1998. International classification of ecological communities: Terrestrial vegetation of the southeastern United States. Working draft of August 1998. The Nature Conservancy, Southeast Regional Office, Southern Conservation Science Dept., Community Ecology Group, Chapel hill, North Carolina, 680 pp.

James E. Van Kley and Douglas N. Hine

Department of Biology, PO. Box 13003

Stephen F. Austin State University

Nacogdoches, Texas 75962

JEVK at:
COPYRIGHT 1998 Texas Academy of Science
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1998 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Van Kley, James E.; Hine, Douglas N.
Publication:The Texas Journal of Science
Geographic Code:1U7LA
Date:Nov 1, 1998
Previous Article:Preliminary observations on breeding by the Llano pocket gopher, Geomys texensis.
Next Article:Importance of arbuscular mycorrhizae to drymass production of a native Texas [C.sub.3] and [C.sub.4] grass.

Terms of use | Privacy policy | Copyright © 2018 Farlex, Inc. | Feedback | For webmasters