2012 NORTHEASTERN NATURALIST 19(2):279–296 Parsimony Analysis of East Coast Salt Marsh Plant Distributions Joseph W. Rachlin1,*, Richard Stalter2, Dwight Kincaid3, and Barbara E. Warkentine1,4 Abstract - A parsimony algorithm was used to evaluate the distribution and co-occurrence of 46 vascular salt marsh-associated species in 20 coastal salt marshes from Biscayne Bay National Park, FL, to Sable Island Marine Protected Sanctuary, NS, Canada. The method considers each salt marsh as if it were a taxon, and the presence or absence of a particular vascular plant species as a “character state” of that taxon. Using this information, a 20 x 46 data matrix was created and examined by multivariate ordination techniques and by parsimony analysis using the program WinClada running over NONA. A hierarchical clustering showed that the salt marsh sites on the eastern seaboard of North America formed two main clusters, one including all of the Florida sites and South Carolina, and the second including all of the more northern sites: North Carolina, Virginia, New Jersey, New York, and Sable Island, NS, Canada. Within the large southern cluster, we find two major sub-clusters separating the Florida marshes from those of South Carolina. Likewise, within the large northern cluster, we find two major sub-clusters separating North Carolina and Virginia from the other northern marshes. An essentially similar pattern of site grouping was also observed using the ordination technique of non-metric multidimensional scaling, in which the southern marshes all aligned to the left of the origin, while the more northern marshes align to its right. Parsimony analysis yielded twelve equally parsimonious trees from which a strict consensus tree was constructed. The topology of the consensus tree clearly shows two major clades, a southern one and a northern one, with the division occurring between South and North Carolina. The main southern clade is supported by the presence of Sporobolus virginicus (Seashore Dropseed), while the main northern clade is supported by the presence of Ruppia maritima (Widgeongrass). Spartina alterniflora (Smooth or Atlantic Cordgrass), which we take as the single species that defines the salt marsh on the eastern coast of North America, was present in all of the sampled sites except Biscayne Bay National Park, FL, a mangrove swamp. Introduction The techniques of hierarchical clustering and ordination has been shown to have great value as an explorative tool for finding and evaluating patterns in ecological data (Gauch 1982; Legendre and Legendre 1998; Manly 2005, 2007; 1Laboratory for Marine and Estuarine Research (La MER), Department of Biological Sciences, Lehman College, City University of New York, 250 Bedford Park Boulevard West, Bronx, NY 10468-1589. 2Department of Biological Sciences, St. John’s University, 8000 Utopia Parkway, Queens, NY 11439. 3Department of Biological Sciences, Lehman College, City University of New York, 250 Bedford Park Boulevard West, Bronx, NY 10468- 1589. 4Science Department, SUNY Maritime College, 6 Pennyfield Avenue, Bronx, NY 10465-4198. *Corresponding author - joseph.rachlin@lehman.cuny.edu. 280 Northeastern Naturalist Vol. 19, No. 2 McCune and Grace 2002; Pielou 1984). That parsimony analysis also has a place in recognizing ecological patterns was initially demonstrated by Lambshead and Paterson (1986) in their seminal paper on the use of numerical cladistics in analyzing ecological data, and more recently by Rachlin et al. (2008) in their evaluation of ichthyofaunal distribution and co-occurrence in an urban stream; by Herrera-Vasquez et al. (2008) to demonstrate that the freshwater ichthyofauna of Costa Rica formed two panbiogeographic tracks, an Atlantic and a Pacific along the length of the country; and by Wenzel and Luque (2008) in their study of ecological succession and changes in community structure. Following from these studies, the goal of this current study is to further demonstrate the efficacy of using parsimony analysis as a tool for ecological exploratory analysis, along with more conventional multivariate analyses such as hierarchical cluster analysis and non-metric multidimensional scaling. In all of these studies, sampling sites are treated as “taxa” and the species, either as abundance or simply the presence or absence of the species of interest at the sampled sites, are treated as “character states” of those “taxa”; it is our hypothesis that parsimony analysis would partition the sampled sites, based on their vegetation, in an analogous manner to that obtained by the other multivariate techniques used. Further, the graphic display of the parsimony analysis would explicitly show the co-occurrence of the salt marsh vegetation on a site-by-site basis, which is not available in hierarchical cluster analysis, non-metric multidimensional scaling, nor other multivariate or grouping procedures. The value of such a graphic display of the distribution of these plants among the various sites and their co-occurrence or community structure as elucidated by parsimony analysis provides the desired detail to allow for future studies of the various factors which constrain these distributions and community structures. In the papers by Rachlin et al. (2008) and Wenzel and Luque (2008), parsimony analysis was performed using the program WinClada (Nixon 2002) running over NONA (Goloboff 1993). Both programs are available at http://www.cladistics. com. The advantage of using these programs is that in the graphical output the sites are shown as terminal “taxa” and the species as “character states” of those “taxa”, which are plotted explicitly on the topology of the tree or trees generated by the analysis. Further, those species uniquely supporting a clade or being unique to a site are differentially marked and identified from those species not uniquely supporting a clade and/or not unique to a site—that is they occur in more than one position in the generated topology of sites. In the present study, to further demonstrate the efficacy of using parsimony analysis to uncover patterns of species distribution and co-occurrence in ecological data, we turned our attention to the salt marsh communities of the east coast of North America. Using the published data generated by one of us (R. Stalter), who has been studying the floristics of these salt marsh communities for over thirty-six years, we extracted a data set of 46 vascular salt marsh species distributed in 20 sites from Florida north to Sable Island, Canada. Nineteen of these sites are salt marshes as defined by the presence of Spartina alterniflora Loisel 2012 J.W. Rachlin, R. Stalter, D. Kincaid, and B.E. Warkentine 281 (Atlantic Cordgrass) and one, Biscayne Bay National Park, FL, that is considered a mangal because of the presence of all four species of mangrove plants: Avicennia germinans L. (Black Mangrove), Conocarpus erectus L. (Button Mangrove), Laguncularia racemosa (L.) Gaertn (White Mangrove), and Rhizophora mangle L. (Red Mangrove) (Reimold 1977). This southernmost site was included because it also contains 19 of the 46 vascular salt marsh plants included in this study and likely represents a transition zone between more southern mangrove swamps and the more northern salt marshes. The specific objectives of this study were: 1) to visualize if the salt marsh plant community changed along a latitudinal gradient; 2) to trace cooccurrence of species and the groupings of sites; 3) to ascertain if any of the vascular salt marsh plants were unique to specific sites, and for those that were not, the extent of their range; and 4) to compare the results of parsimony analysis with those of multivariate methods more commonly used to evaluate and ordinate communities (e.g., hierarchical cluster analysis and non-metric multidimensional scaling). Methods The data for this study came from a series of floristic survey studies, based on multiple site visits during the growing seasons of salt marsh vegetation, and published between 1972 and 2006. These studies do not represent quantitative ecological inventory analyses. Each published paper represents an exhaustive floristic survey based on the results of at least three years of coverage of a site over all regional growing seasons, in which the entire habitat was walked and the presence of species recorded. At least one voucher specimen of each species found was taken for herbarium deposition in the A.C. Moore Herbarium, University of South Carolina, Columbia, SC. Table 1 lists the 20 sites included in this study, their latitudes, and their specific reference source. We carefully examined each one of these published papers to vet the site-bysite presence of the vascular salt marsh plants included in this study, leaving out Zostera marina L. (Common Eelgrass) and other coastal submerged vegetation that were sometimes included in the floristic data for these sites. Table 1 also lists the site codes used throughout the analyses. Figure 1, a map of the east coast of the United States from Florida north to Connecticut, shows the approximate locations of 19 of the 20 marshes evaluated in this study. The 20th salt marsh, Sable Island, NS, Canada, at a latitude of 43°56'N, is approximately 168 nautical miles north of the northernmost US site (Orient Beach State Park, NY) and is shown as an insert in Figure 1. Because plant nomenclature has changed over the 34 years encompassed by this study, the problem of synonyms was dealt with in the following manner. Species in South Carolina originally classified after Radford et al. (1971), species from North Carolina to New York originally classified after Gleason and Cronquist (1991), and species from Florida originally classified after Wunderlin 282 Northeastern Naturalist Vol. 19, No. 2 (1998), are now classified for consistency according to Kartesz (1994). Genus and species names were then checked with the large Integrated Taxonomic Information System Database (ITIS 2011). All taxa were identified to species and recorded by salt marsh study site. A 20- site by 46-species data matrix was constructed (Table 2a) in which the presence of a species at a particular site was coded as “1” and its absence coded as “0”. In this coding, we treat each site as if it were a “taxon” and the presence/absence of a plant species as a character state of that “taxon” (Lambshead and Paterson 1986, Rachlin et al. 2008, Wenzel and Luque 2008). Table 2b lists the codes for each of the 46 plant species examined in this study along with their naming authorities. This data matrix was analyzed by two multivariate techniques, hierarchical cluster analysis, and the ordination technique of non-metric multidimensional scaling. In addition to the many other possible methods of analysis, we also did not include indicator species analysis (De Caceres and Legendre 2009, Dufrene and Legendre 1997) as this technique did not seem suitable for our data and for our objectives. Further, it was our desire to conduct all of the analyses using proven non-commercial software packages available online, and the current version of PAST used for the multivariate analysis part of this paper does not support indicator species analysis. It is supported as an R language package, but using it as such requires familiarity and experience with the R language, not yet a common achievement among investigators. Table 1. Salt marsh sites, listed from southernmost to northernmost, for parsimony analysis of species co-occurrence and references to publications from which the site specific species data was obtained. Site # Site name Code Latitude Reference 1 Biscayne Bay National Park, flBNP,fl25°39'N Stalter et al. 1999 2 Turtle Mound, flTM,fl28°55'N Stalter and Kincaid 2004 3 Green Mound, flGM,fl29°06'N Stalter and Kincaid 2004 4 Tomoka State Park, flTSP,fl29°20'N Stalter and Kincaid 2004 5 Turtle Island, SC TI,SC 32°03'N Stalter 1973 6 Hunting Island State Park, SC HISP,SC 32°23'N Stalter 1985 7 Otter Island, SC OI,SC 32°29'N Stalter 1972 8 Isle of Palms, SC IoP,SC 32°48'N Stalter 1975 9 Bull Island, SC BI,SC 32°54'N Stalter 1984 10 Huntington Beach State Park, SC HBSP,SC 33°30'N Stalter 1971, 1978 11 Outer Banks, NC OB,NC 35°46'N Stalter and Lamont 1997 12 Fisherman Island, VA FI,VA 37°05'N Stalter and Lamont 2000a 13 Assateague Island, VA AI,VA 37°55'N Stalter and Lamont 1990 14 Cape May Point State Park, NJ CMSP,NJ 38°56'N Sutton et al. 1990 15 Little Beach Island, NJ LBI,NJ 39°28'N Stalter 1994 16 Sandy Hook, NJ SH,NJ 40°25'N Stalter and Lamont 2000b 17 Jamaica Bay Wildlife Refuge, NY JBWR,NY 40°37'N Stalter and Lamont 2002 18 Fire Island National Sea Shore, NY FINSS,NY 40°39'N Stalter et al. 1986 19 Orient Beach State Park, NY OBSP,NY 41°08'N Lamont and Stalter 1991 20 Sable Island, NS, Canada SI,CAN 43°56'N Stalter and Lamont 2006 2012 J.W. Rachlin, R. Stalter, D. Kincaid, and B.E. Warkentine 283 Non-metric multidimensional scaling is selected as the ordination method of choice as it is a “free-ordination” technique that attempts to find patterns in the data in the absence of predetermined guiding principles as to what might cause these patterns; it also provides a synthesis of the information about the Figure 1. Map showing approximate locations of the sampled US east coast salt marsh sites from Florida north to New York, with insert showing Sable Island, NS, Canada. See Table 1 for site codes. 284 Northeastern Naturalist Vol. 19, No. 2 Table 2a. Data matrix of 20 stations and 46 salt marsh vascular plant species. 0 = species absence at a particular site, and 1 = species presence at a particular site. Station codes as per Table 1. Species codes listed in Table 2b. Species Stations 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 BNP,fl0 0 1 0 0 1 1 0 1 1 1 0 0 1 0 0 0 1 1 0 0 0 1 1 0 0 0 1 0 0 1 0 1 0 0 1 1 1 0 1 0 1 0 1 0 0 TM,fl0 0 0 0 0 1 0 0 0 0 0 1 0 0 0 0 0 1 0 0 0 0 1 0 0 0 0 1 0 0 0 0 0 0 0 0 0 0 1 0 0 1 0 0 0 0 GM,fl0 0 0 0 0 0 0 0 1 1 1 1 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 1 0 0 1 0 1 1 1 0 1 0 0 0 0 TSP,fl0 0 0 0 0 0 1 0 1 1 0 1 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 0 1 0 0 0 0 TI,SC 0 0 1 0 0 0 1 0 1 1 0 0 0 1 0 0 0 0 0 1 0 1 0 1 0 0 0 0 0 1 0 1 1 0 0 0 0 0 1 1 0 1 0 1 0 0 HISP,SC 0 0 1 1 1 0 1 0 1 1 0 0 1 1 0 0 0 0 1 1 0 1 0 1 1 0 0 0 0 1 1 0 1 0 1 0 0 1 1 1 0 1 0 1 0 0 OI,SC 0 0 1 1 0 0 1 0 1 1 0 0 1 1 0 0 0 0 0 1 0 1 0 1 0 0 0 0 0 1 0 1 1 0 0 0 0 1 1 1 0 1 0 1 0 0 IoP,SC 0 1 1 1 1 0 1 0 0 1 0 0 0 1 0 0 0 0 1 1 0 1 0 1 1 0 0 0 0 1 1 0 1 1 0 0 1 1 1 1 1 1 0 1 0 0 BI,SC 0 1 1 0 1 0 1 0 1 1 0 0 1 1 0 0 0 0 1 1 0 1 0 1 1 0 0 0 0 0 1 0 1 0 1 0 0 1 1 1 0 1 0 1 0 0 HBSP,SC 0 0 1 1 1 0 1 0 0 0 0 0 0 0 0 0 0 0 0 1 0 1 0 1 0 0 0 0 0 1 0 0 1 0 0 0 1 1 1 1 0 1 0 1 0 0 OB,NC 1 1 1 1 1 0 0 0 0 1 0 1 0 1 1 1 1 0 1 1 0 1 0 1 0 0 0 0 1 1 1 1 1 1 1 1 0 1 1 1 1 0 0 1 0 0 FI,VA 1 1 1 1 1 0 0 0 0 1 0 0 0 1 0 1 1 0 0 1 1 1 0 1 0 0 0 0 0 1 1 1 1 1 1 1 0 1 1 1 0 0 1 1 0 0 AI,VA 1 1 1 1 1 0 0 1 0 1 0 0 0 1 1 1 1 0 0 1 1 1 0 1 0 0 0 0 1 1 1 1 1 1 0 1 0 1 1 1 1 0 0 1 0 0 CMPSP,NJ 0 1 0 0 1 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 1 0 0 0 0 0 1 0 1 1 0 0 0 1 1 0 0 1 1 1 1 0 0 0 0 0 LBI,NJ 0 0 1 1 1 0 0 1 0 0 0 0 0 1 0 0 0 0 0 1 1 0 0 1 0 1 0 0 1 1 0 1 1 0 1 1 0 1 1 1 1 0 0 1 0 0 SH,NJ 0 0 1 1 1 0 0 1 0 0 0 0 0 1 0 0 0 0 0 1 1 0 0 1 0 1 0 0 1 0 0 1 1 0 0 0 0 1 1 1 1 0 1 1 0 0 JBWR,NY 0 1 1 1 1 0 0 1 0 0 0 0 0 1 1 1 0 0 0 1 1 0 0 1 0 0 0 0 0 1 0 1 0 1 0 0 0 1 1 1 1 0 1 0 1 0 FINSS,NY 0 0 0 1 1 0 0 1 0 0 0 0 0 1 0 0 0 0 0 1 1 0 0 1 0 0 1 0 0 1 0 1 0 0 1 0 0 1 1 1 0 0 0 0 1 0 OBSP,NY 1 1 1 1 1 0 0 1 0 0 0 0 0 1 0 1 1 0 0 1 1 0 0 1 0 1 0 0 1 1 0 1 1 1 1 0 0 1 1 1 1 0 0 1 1 1 SI,CAN 0 0 0 0 1 0 0 0 0 0 0 0 0 0 1 1 0 0 0 0 0 0 0 0 0 1 0 0 1 0 0 0 0 0 0 0 0 1 1 0 1 0 0 0 0 0 2012 J.W. Rachlin, R. Stalter, D. Kincaid, and B.E. Warkentine 285 Table 2b. Species codes for Table 2a. Code Species Code Species 1 Agalinis maritima (Raf.) Raf. (Saltmarsh False Foxglove) 28 Rhizophora mangle L. (Red Mangrove) 2 Aster subulatus Michx. (Annual Saltmarsh Aster) 29 Ruppia maritima L. (Widgeongrass) 3 Aster tenuifolius L. (Saline Aster) 30 Sabatia stellaris Pursh (Rose of Plymouth) 4 Atriplex arenaria Nutt. (Quelite) 31 Salicornia bigelovii Torr. (Dwarf Saltwort) 5 Atriplex prostrata Boucher ex DC. (Hastate Orache) 32 Salicornia europaea sensu Wolff & Jefferies, non L. (Slender Glasswort) 6 Avicennia germinans (L.) L. (Black Mangrove) 33 Salicornia virginica L. (Virginia Glasswort) 7 Baccharis angustifolia Michx. (Saltwater False Willow) 34 Scirpus americanus auct. non Pers. (American Bulrush) 8 Bassia hirsuta (L.) Asch. (Hairy Smotherweed) 35 Scirpus robustus Pursh (Sturdy Bulrush) 9 Batis maritima L. (Saltwort) 36 Sesuvium maritimum (Walt.) B.S.P. (Slender Seapurslane) 10 Borrichia frutescens (L.) DC (Bushy Seaoxeye) 37 Sesuvium portulacastrum (L.) L. (Shoreline Seapurslane) 11 Conocarpus erectus L. (Button Mangrove) 38 Solidago sempervirens L. (Seaside Goldenrod) 12 Cynanchum angustifolium Pers. (Gulf Coast Swallowwort) 39 Spartina alterniflora Loisel. (Atlantic Cordgrass) 13 Cynanchum palustre (Pursh) Heller (Gulf Coast Swallowwort) 40 Spartina patens (Aiton) Muhl. (Marshhay Cordgrass) 14 Distichlis spicata (L.) Greene (Desert Saltgrass) 41 Spergularia maritima (All.) Chiov. (Media Sandspurry) 15 Eleocharis halophila (Fernald & Brackett) 42 Sporobolus virginicus (L.) Kunth (Seashore Dropseed) Fernald & Brackett (Saltmarsh Spikerush) 43 Suaeda calceoliformis (Hook.) Moq. (Paiuteweed) 16 Eleocharis parvula (Roem. & Schult.) Link ex Bluff, 44 Suaeda linearis (Elliott) Moq. (Annual Seepweed) Nees & Schauer (Dwarf Spikerush); 45 Suaeda maritima (L.) Dumort. (Herbaceous Seepweed) 17 Fimbristylis castanea (Michx.) Vahl (Marsh Fimbry) 46 Triglochin maritima L. (Arrowgrass) 18 Heliotropium angiospermum Murr. (Scorpion’s-tail) 19 Heliotropium curassavicum L. (Quail Plant) 20 Iva frutescens L. (Bigleaf Sumpweed) 21 Juncus gerardii Loisel. (Saltmarsh Rush) 22 Juncus roemerianus Scheele (Needlegrass Rush) 23 Laguncularia racemosa (L.) Gaertn. f. (White Mangrove) 24 Limonium carolinianum (Walt.) Britt. (Carolina Sea-lavender) 25 Limonium nashii Small (Carolina Sea-lavender) 26 Plantago maritima L. (Goose Tongue) 27 Puccinellia fasciculata (Torr.) E.P. Bicknell (Saltmarsh Alkaligrass) 286 Northeastern Naturalist Vol. 19, No. 2 similarity of sample units to one another (Gauch 1982, Gerhard et al. 2004, Hammer and Harper 2006, Manly 2005, McCune and Grace 2002, Peck 2010). As such it is ideally suited for our data, which comes from a data set based on floristic collections made over a 34-year time span. Hierarchical cluster analysis was performed using Ward’s method (Hammer and Harper 2006, McCune and Grace 2002), and non-metric multidimensional scaling used the Bray-Curtis measure of similarity (Hammer and Harper 2006). The technique of non-metric multidimensional scaling is performed by the program such that each run is actually a sequence of eleven trials from which the one with the smallest stress is chosen; ten of the trials have random starting configurations, and the eleventh uses principal coordinate analysis to compute the starting configuration. The algorithm implemented in PAST to accomplish this is based on a relatively new approach developed by Taguchi and Oono (2005). Stress values below 0.1 are considered “good” (Hammer and Harper 2006). In addition, the Shepard plot of the obtained versus the observed (target) ranks indicates the quality of the result. Under ideal conditions, all points in the Shepard plot should fall on a straight ascending line (x = y), the number of points in the plot for n original data points will be a total of (n2 - n) / 2 (Hammer and Harper 2006). These multivariate analyses were carried out using the program PAST version 2.12 (Hammer et al. 2001). This statistical program in its most recent version can be downloaded at no cost from http://folk.uio. no/ohammer/past. The parsimony analysis was performed using WinClada (Nixon 2002) running over NONA (Goloboff 1993) with “slow or Deltran optimization”, and a null station consisting of all zeros was added to the data set of Table 2a to root the tree or trees generated in the analysis. Results The results of the multivariate techniques are shown in Figures 2, 3, and 4. The hierarchical cluster analysis (Fig. 2) essentially forms two main clusters representing a southern section from Florida north to and including South Carolina, and a northern section from North Carolina to Sable Island Canada. The southern section is further subdivided into two subsections essentially separating the Florida marshes from those of South Carolina. The northern section is likewise essentially divided into two main subdivisions separating the North Carolina and Virginia marshes from those of New York, New Jersey, and Canada. Figure 3 presents the results of non-metric multidimensional scaling (NMDS) analysis, the multivariate ordination technique used in this study. NMDS provides a spatial representation of the data suitable for ecological community analysis (Gauch 1982, Gerhard et al. 2004, Hammer and Harper 2006, Harrison and Whitfield 2006, McCune and Grace 2002, Peck 2010), and the figure shows that all of the southern marshes lie to the left of the origin, while the northern marshes from North Carolina to Canada lie to the right of the origin; axis 1, 2012 J.W. Rachlin, R. Stalter, D. Kincaid, and B.E. Warkentine 287 which is the axis accounting for this separation, had an r2 value of 0.8516. This separation echoes the major divisions observed in the hierarchical cluster analysis (Fig. 2). The separation of marshes above and below the origin is weakly accounted for by axis 2, which had an r2 of only 0.0699. It can also be seen that all of the Florida sites lie to the left of the South Carolina sites, and the North Carolina and Virginia sites form a tight cluster on the right side of the origin in the upper quadrant, indicating similarity in species space. This NMDS ordination technique transforms the distances between the stations, as defined by their inclusive species, into ranks, and compares these ranked distances with the ranks of the distances in the ordination plot. The quality of the result may be determined by assessing the resulting stress, which is a measure of how much the ranked distance in the ordination deviate from the original ranked distances (Hammer and Harper 2006). A stress value of 0.1 is considered good, and our value of 0.09737 Figure 2. Hierarchical cluster analysis, using Ward’s method, of the sampled sites derived from the data matrix (Table 2a). See Table 1 for site codes. 288 Northeastern Naturalist Vol. 19, No. 2 is thus quite acceptable. Another way of assessing the quality of the NMDS ordination is to generate a Sheppard plot as shown in Figure 4. This procedure plots the ranks obtained from the ordination against the original ranks. Ideally, these values should be the same and produce a linear plot. The number of plotted points in a Sheppard Plot is (n2 - n) / 2, where “n” equals the number of original data points (Hammer and Harper 2006). In our case, the number of plotted points is 190 since we are considering the ordination of 20 stations, and the plotted points do produce an essentially linear plot. Parsimony analysis yielded 12 equally parsimonious trees, which were collapsed on the ambiguous nodes to yield the strict consensus tree presented in Figure 5. We used Deltran or slow optimization, as this tends to push the character states, in this case salt marsh species, as far up towards the terminal taxa (sampled sites, coded as per Table 1) as possible. The display of the character states (species) is represented as either a closed black circle or an open white circle. A black circle indicates that a particular species is on a branch leading Figure 3. Non-metric multidimensional scaling analysis of sampled sites derived from the data matrix (Table 2a). See Table 1 for site codes. 2012 J.W. Rachlin, R. Stalter, D. Kincaid, and B.E. Warkentine 289 to a clade, and that it uniquely supports that clade; further, it is usually found in all members of that clade. If the black circle is found only on the terminal branch of the clade, it indicates uniqueness of that salt marsh plant to that site. Open white circles indicate that the salt marsh species is neither unique to a particular clade nor uniquely supports a clade. The numbers above the circles identify the species, and the number below the circle, either a “1” or a “0”, indicates the presence or absence of that species at that location. For example, in Figure 5, black circle 39 representing the species Spartina alterniflora, with a “1” below it, is found on the base branch leading to all of the sites examined, and is therefore considered the one species common to and defining all salt marshes studied from Turtle Mound, Florida (TM,FL) to Sable Island, Canada (SI,CAN). The unique absence of Spartina alterniflora from Biscayne Bay National Park, Florida (BNP,FL) indicated by a “0” below the open circle numbered 39 on the terminal branch leading directly to this site would indicate that this site is not a true salt marsh; in fact, it is a mangal. Reading back from the terminal site, to the base of the tree and noting all of the species represented with a “1” below the circles, either black or white, indicates the entire assemblage of salt marsh species at that site. The first thing that becomes obvious in Figure 5 is that once again there is a separation into two major clades representing the southern salt marshes Figure 4. Shepard plot derived from non-metric multidimensional scaling analysis. 290 Northeastern Naturalist Vol. 19, No. 2 and the northern ones, and again the separation occurs in the vicinity of North Carolina, which aggregates with the northern marshes. The northern clade is supported by species #29 (Ruppia maritima), which is present at all but three of Figure 5. Strict consensus tree derived from parsimony analysis of the data matrix (Table 2a). See Table 1 for site codes. Closed black circles represent species either unique to a site or uniquely supporting a clade of sites. Open circles represent species not unique to a site or clade. The number “1” below any circle represents the presence of that species, while a “0” represents its absence. Code numbers above circles represent species that are identified below the tree. 2012 J.W. Rachlin, R. Stalter, D. Kincaid, and B.E. Warkentine 291 the northern marshes from the Outer Banks, NC (35°46'N) to Sable Island, NS, Canada (43°56'N); it was absent from Fisherman Island, VA (37°05'N), Jamaica Bay Wildlife Refuge, NY (40°37'N), and Fire Island National Sea Shore, NY (40°39'N). Among the northern clades, # 46 (Triglochin maritima [Arrowgrass]), is unique to Orient Beach State Park, NY. There are no other salt marsh plants in either the northern or the southern clade that were found in only one of our study sites. The southern clade is supported by species #42 (Sporobolus virginicus), which is present in all of the southern marshes including the Biscayne Bay National Park, FL, a mangal. In addition, species #9 (Batis maritima [Saltwort]) uniquely supports all of the clade after Turtle Mound, FL, but is absent from Huntington Beach, SC and Isle of Palms, SC, and species number 14 (Distichlis spicata [Desert Saltgrass]), also supporting the same clade, but not uniquely, is only absent from Huntington Beach State Park, SC. In fact, from a latitudinal perspective, Distichlis spicata is present in all of the sampled southern salt marshes between latitudes 25°39'N (Biscayne Bay NP, FL) and 32°54'N (Bull Island, SC) except for the Turtle Mound, flsite (28°55'N). It is, however, also found in all of the northern marshes except for the Sable Island, NS site. Discussion In order to understand the distribution and co-occurrence of salt marsh vegetation along the east coast of North America, we first obtained a relatively large data set consisting of 20 sites, comprising a total of 46 salt marsh vascular species. This data set was first examined using the techniques of multivariate analysis as a means of exploration to determine the existence of any structure in the data set (Gauch 1982, Legendre and Legendre 1998, Manly 2005, McCune and Grace 2002, Pielou 1984). The techniques used consisted of hierarchical cluster analysis and the ordination technique of non-metric multidimensional scaling. Hierarchical cluster analysis actually groups sites in an evident hierarchy; however, since it is known that most clustering algorithms result in a hierarchy even if the elements are not hierarchically related (Legendre and Legendre 1998), several investigators have used non-metric multidimensional scaling to validate their results (Gerhard et al. 2004, Harrison and Whitfield 2006, Rachlin et al. 2008). In the current study, the use of PAST version 2.12 (Hammer et al. 2001) provided modules for both the hierarchical cluster analysis and non-metric multidimensional scaling with the ability to choose one of several distance measures to provide output with minimum stress, and the ability to generate a Shepard plot to further validate the results. Used in conjunction, hierarchical classification and ordination, because of their complementary natures, provide a powerful tool in discerning patterns in the structure of community data (Adam 1978, Brazner and Beals 1997). Figures 2 and 3 clearly show a fairly consistent structure in the aggregation and clustering of the sites examined in this study. There also is a general north– south trending which would be anticipated based on the mean temperatures and 292 Northeastern Naturalist Vol. 19, No. 2 tidal and climatic conditions normally experienced on a latitudinal gradient along the east coast of North America (Adam 1978, Hartig et al. 2002, Pennings et al. 2005, Pethick 1981). What is of particular interest was the clear break between southern and northern marshes occurring between Huntington Beach State Park, SC and the Outer Banks, NC, approximately 426 km distant. This observation first clearly seen in the hierarchical cluster analysis (Fig. 2) was also confirmed in the parsimony analysis (Fig. 5), and a closer examination of Figure 5 clearly shows that there are five species (Agalinis maritima [Saltmarsh Foxglove], Fimbristylis castanea [Marsh Frimby], Juncus gerardii [Saltmarsh Rush], Ruppia maritima, and Triglochin maritima) uniquely associated with the northern marshes, and four species (Baccharis angustifolia (Saltwater False Willow), Batis maritima, Limonium nashii (Carolina Sea-lavender), and Sporobolus virginicus) uniquely associated with the southern marshes. Of further interest is that some species, such as Juncus roemerianus (Needlegrass Rush), are found in both northern and southern marshes, being present in all of the South Carolina marshes and also in the North Carolina and Virginia sites, but the congeneric Juncus gerardii is exclusive to the northern marshes (Fig. 5). This observation is consistent with the earlier distributional report of Eleuterius (1976). One can also see that Spartina alterniflora is associated with all of the salt marshes but is absent from Biscayne Bay National Park, FL. This site, which is included in this study because of the presence of 20 of the 46 other salt marsh vascular plant species, is actually classified as a mangal rather than a salt marsh and likely represents a transition zone between the more southern mangrove marshes and the more northern salt marshes (Reimold 1977). The parsimony analysis yielded 12 equally parsimonious trees, which indicated that the data did not have enough resolution to produce a single unambiguous tree showing a unique hierarchical grouping of all of the sites. When this occurs, one selects the strict consensus tree option to produce a single tree collapsed on the ambiguous nodes. An examination of Figure 5 shows that the strict consensus tree is in essential agreement with the results of the multivariate analyses. While the multivariate analyses showed that there was indeed structure in the data, the advantage of the parsimony analysis is that in addition to showing that the data has an ordered structure, one can graphically visualize the precise nature of that structure by treating the sites as if they were taxa and the presence of salt marsh plant species as character states of those taxa. This visual representation permits a direct reading, from the terminal branch back to the base, of all species present at the terminal site. Further, one can readily see which salt marsh plant species are unique to a particular site and which are not, and which grouping of sites share common salt marsh plants. In other words, the salt marsh vegetation community structure becomes readily discernable and comparisons among the sites are easily made. For example, looking at Assateague Island, VA (AI,VA) and reading from the terminal branch back to the base of the tree we see that this salt marsh contains a community of twenty-seven species represented by #15 Eleocharis halophila 2012 J.W. Rachlin, R. Stalter, D. Kincaid, and B.E. Warkentine 293 (Saltmarsh Spikerush), #36 Sesuvium maritimum (Slender Seapurslane), #31 Salicornia bigelovii (Dwarf Saltwort), #22 Juncus roemerianus, #10 Borrichia frutescens (Bushy Seaoxeye), #44 Suaeda linearis (Annual Seepweed), #34 Scirpus americanus (American Bulrush) , #33 Salicornia virginica (Virginia Glasswort), #17 Fimbristylis castanea, #16 Eleocharis parvula (Dwarf Spicebush), #2 Aster subulatus (Annual Saltmarsh Aster), #1 Agalinis maritima, #32 Salicornia europaea (Slender Glasswort), #24 Limonium carolinianum (Carolina Sea-lavender), #20 Iva frutescens (Bigleaf Sumpweed), #8 Bassia hirsuta (Hairy Smotherweed), #4 Atriplex arenaria, #3 Aster tenuifolius (Saline Aster), #40 Spartina patens (Marshhay Cordgrass), #30 Sabatia stellaris (Rose of Plymouth), #21 Juncus gerardii, #14 Distichlis spicata, #41 Spergularia maritima (Media Sandspurry), #38 Solidago sempervirens (Seaside Goldenrod), #29 Ruppia maritima, #5 Atriplex prostrata (Hastate Orache), and #39 Spartina alterniflora, but not #35 Scirpus robustus (Sturdy Bulrush), which has a “0” below it on the topology. Thus, we believe that the application of parsimony analysis to ecological data, as previously demonstrated by Lambshead and Paterson (1986), Rachlin et al. (2008), and Wenzel and Luque (2008), and as shown in this study, is an appropriate, timely, and valid approach to understanding, in fine detail, the community structure and species distributions and co-occurrences underlying the broad relationships implied by multivariate analyses. This parsimony analysis, which clearly shows the wide or narrow longitudinal distributions of individual plant species as well as their co-occurrence with other plant species, generally follows their greater or narrower tolerance to such environmental parameters as annual rainfall, temperature, and soil types and other edaphic conditions, which vary on a longitudinal basis along the east coast of the North American continent. The results of this parsimony analysis sets the stage for future studies to determine the specific environmental and spatial factors which constrain vascular salt marsh vegetation community structure and species distributions as elucidated in this study. It can be seen that the four specific objectives enumerated in the introduction of this paper have been successfully met. Literature Cited Adam, P. 1978. Geographical variation in British Saltmarsh vegetation. Journal of Ecology 66:339–366. Brazner, J.C., and E.W. Beals. 1997. Patterns in fish assemblages from wetland and beach habitats in Green Bay, Lake Michigan: A multivariate analysis of abiotic and biotic forcing factors. Canadian Journal of Fisheries and Aquatic Sciences 54:1743–1761. De Caceres, M., and P. Legendre. 2009. Associations between species and groups of sites: Indices and statistical inference. Ecology 90:3566–3574. Dufrene, M., and P. Legendre. 1997. Species assemblages and indicator species: The need for a flexible asymmetrical approach. Ecological Monographs 67:345–366. Eleuterius, L.N. 1976. The distribution of Juncus roemerianus in the salt marshes of North America. Chesapeake Science 17:289–292 294 Northeastern Naturalist Vol. 19, No. 2 Gauch, H.G., Jr. 1982. Multivariate Analysis in Community Ecology. Cambridge University Press, New York, NY. 298 pp. Gerhard, P., R. Moraes, and S. Molander. 2004. Stream fish communities and their associations to habitat variables in a rain forest reserve in southeastern Brazil. Environmental Biology of Fishes 71:321–340. Gleason, H.A., and A. Cronquist. 1991. Manual of Vascular Plants of the Northeastern United States and Adjacent Canada, 2nd Edition. New York Botanical Garden, New York, NY. 910 pp. Goloboff, P.A. 1993. NONA, Version 2, Published by the author, Tucuman, Argentina. Available online at http://www.cladistics.com. Accessed 14 November 2011. Hammer, Ø., and D.A.T. Harper. 2006. Paleontological Data Analysis. Blackwell Publishing, Malden, MA. 351 pp. Hammer, Ø., D.A.T. Harper, and P.D. Ryan. 2001. PAST: Paleontological Statistics Software Package for Education and Data Analysis. Palaeontoligica Electronica 4:1–9. Available online at http://folk.uio.no/ohammer/past. Accessed 14 November 2011. Harrison, T.D., and A.K. Whitfield. 2006. Estuarine typology and structuring of fish communities in South Africa. Environmental Biology of Fishes 75:269–293. Hartig, E.K., V. Gornitz, A. Kolker, F. Mushacke, and D. Fallon. 2002. Anthropogenic and climate-change impacts on salt marshes of Jamaica Bay, New York City. Wetlands 22:71–89. Herrera-Vasquez, J., W. Bussing, and F. Villalobos. 2008. Panbiogeographic analysis of Costa Rican freshwater fishes. Revista de Biologia Tropical (International of Journal Tropical Biology and Conservation ISSN-0034-7744) 56:165–170. Integrated Taxonomic Information System (ITIS). 2011. ITIS database. Available online at http://www.itis.gov. Accessed 14 November 2011. Kartesz, J.T. 1994. A Synonymized Checklist of the Vascular Flora of the United States, Canada, and Greenland, 2nd Edition. Vol. 2: Checklist. Timber Press, Inc., Portland OR. 816 pp. Lambshead, P.J.D., and G.L.J. Paterson. 1986. Ecological cladistics: An investigation of numerical cladistics as a method for analyzing ecological data. Journal of Natural History 20:895–909. Lamont, E.E., and R. Stalter. 1991. The vascular flora of Orient Beach State Park, Long Island, New York. Bulletin of the Torrey Botanical Club 118:459–468. Legendre, P., and L. Legendre. 1998. Numerical Ecology, 2nd (English) Edition. Elsevier Science, Amsterdam, The Netherlands. 853 pp. Manly, B.F.J. 2005. Multivariate Statistical Methods: A Primer, 3rd Edition. Chapman and Hall/CRC, Boca Raton, FL. 214 pp. Manly, B.F.J. 2007. Randomization, Bootstrap and Monte Carlo Methods in Biology, 3rd Edition. Chapman and Hall/CRC, Boca Raton, FL. 455 pp. McCune, B., and J.B. Grace. 2002. Analysis of Ecological Communities. MJM Software Design, Gleneden Beach, OR. 300 pp. Nixon, K.C. 2002. WinClada, Version 1.00.08. Published by the author, Ithaca, NY. Available online at http://www.cladistics.com. Accessed 14 November 2010. Peck, J.E. 2010. Mutivariate Analysis for Community Ecologists: Step-by-Step using PC-ORD. MJM Software Design, Gleneden Beach, OR. 162 pp. Pennings, S.C., M-B. Grant, and M.D. Bertness. 2005. Plant zonation in low-latitude salt marshes: Disentangling the roles of flooding, salinity, and competition. Journal of Ecology 93:159–167. 2012 J.W. Rachlin, R. Stalter, D. Kincaid, and B.E. Warkentine 295 Pethick, J.S. 1981. Long-term accretion rates on tidal salt marshes. Journal of Sedimentary Petrology 51:571–577. Pielou, E.C. 1984. The Interpretation of Ecological Data: A Primer on Classification and Ordination. John Wiley and Sons, New York, NY. 263 pp. Rachlin, J.W., B.E. Warkentine, and A. Pappantoniou. 2008. Use of a parsimony algorithm as a method for evaluating ichthyofaunal distribution and co-occurrence in an urban stream. Journal of Freshwater Ecology 23:1–11. Radford, A.E., H.E. Ahles, and C.R. Bell. 1971. Manual of the Vascular Flora of the Carolinas. 2nd Edition. University of North Carolina Press, Chapel Hill, NC. 1183 pp. Reimold, R.J. 1977. Mangals and salt marshes of the eastern United States. Pp. 157–166, In V.J. Chapman (Ed.). Ecosystems of the World 1: Wet Coastal Ecosystems. Elsevier Scientific Publishing Co., New York, NY. 428 pp. Stalter, R. 1971. The summer and fall flora of Huntington Beach State Park Georgetown County, South Carolina. Castanea 36:167–174. Stalter, R. 1972. The flora of outer Otter Island, Colleton County, South Carolina. Castanea 37:298–300. Stalter, R. 1973. The flora of Turtle Island, Jasper Co., South Carolina. Castanea 38:35–37. Stalter, R. 1975. The flora of the Isle of Palms, South Carolina. Castanea 40:4–13. Stalter, R. 1978. The plant communities of South Carolina’s barrier islands. Pp. 227–244, In R.R. Lewis and D.P. Cole (Eds.). Proceedings of the Fifth Annual Conference on the Restoration of Coastal Vegetation in Florida. Hillsboro County Community College, Tampa, FL. Stalter, R. 1984. The flora of Bull Island, Charleston County, South Carolina. Bartonia 50:27–30. Stalter, R. 1985. The flora of Hunting Island, Beaufort County, South Carolina. Bartonia 51:99–104. Stalter, R. 1994. The vegetation of Little Beach Island, New Jersey. Bartonia 58:97–100. Stalter, R., and D. Kincaid. 2004. The vascular flora of five Florida shell middens. Journal of the Torrey Botanical Society 131:93–103. Stalter, R., and E.E. Lamont. 1990. The vascular flora of Assateague Island, Virginia. Bulletin of the Torrey Botanical Club 117:48–56. Stalter, R., and E.E. Lamont. 1997. Flora of North Carolina’s Outer Banks, Ocracoke Island to Virginia. Journal of the Torrey Botanical Society 124:71–88. Stalter, R., and E.E. Lamont. 2000a. Vascular flora of Fisherman Island, Virginia. Journal of the Torrey Botanical Society 127:324–332. Stalter, R., and E.E. Lamont. 2000b. Vascular flora of Sandy Hook, New Jersey. Bartonia 60:105–116. Stalter, R., and E.E. Lamont. 2002. Vascular flora of Jamaica Bay Wildlife Refuge, Long Island, New York. Journal of the Torrey Botanical Society 129:346–358. Stalter, R., and E.E. Lamont. 2006. The historical and extant flora of Sable Island, Nova Scotia, Canada. Journal of the Torrey Botanical Society 133:362–374. Stalter, R., E.E. Lamont, and J. Northrup. 1986. Vegetation of Fire Island, New York. Bulletin of the Torrey Boanical Club 113:298–306. Stalter, R., J. Tamory, P. Lynch, and B. Lockwood. 1999. The vascular flora of Biscayne National Park, Florida. Sida, Contribtions to Botany 18:1207–1226. Sutton, P., R. Meyer, and R. Stalter. 1990. The vascular plants of Cape May Point State Park Cape May County, New Jersey. Bulletin of the Torrey Botanical Club 117:294–300. 296 Northeastern Naturalist Vol. 19, No. 2 Taguchi, Y.-h., and Y. Oono. 2005. Relational patterns of gene expression via non-metric multidimensional scaling analysis. Bioinformatics 21:730–740. Wenzel, J.W., and G.M. Luque. 2008. Parsimony analysis of ecological succession, a powerful tool for interpreting changes in community structure. Cladistics 24:746– 757. Wunderlin, R. 1998. Guide to the Vascular Plants of Florida. University Press of Florida, Gainesville, FL. 806 pp.