Journal of the Torrey Botanical Society 144(2): 125–138, 2017.

Plant diversity and community composition in eastern North American serpentine barrens1 Kathryn M. Flinn2, Jennifer L. Mikes, and Hannah A. D. Kuhns Biology Department, Franklin and Marshall College P.O. Box 3003, Lancaster, PA 17604

Abstract. Serpentine barrens are globally rare, savanna habitats that support distinctive vegetation. In eastern North America, two major threats to serpentine barrens are the spread of invasive and the encroachment of trees into open habitats. We assessed these threats by comparing serpentine savannas and woodlands in the State Line serpentine barrens, southeast Pennsylvania. We asked which invasive species were most frequent and abundant in each habitat and how they may affect diversity and community composition and how succession of savannas to woodlands may affect plant diversity and community composition. We compared species richness, evenness, diversity, and beta diversity between savannas and adjacent woodlands and quantified differences in community composition. Invasive species comprised 16% of total plant cover, and the most frequent invasive species were Microstegium vimineum (Trin.) A. Camus, Lonicera japonica Thunb., and Elaeagnus umbellata Thunb. Savannas had substantially greater species richness, diversity, and beta diversity than woodlands. Of all species we found, 31% were unique to savannas. The cover of invasive species and the native greenbrier Smilax rotundifolia L. were negatively associated with the cover of characteristic native grasses and forbs. Although invasive species threaten diversity in serpentine barrens, tree encroachment may pose the greater threat. Savannas that become woodlands lose diversity and change composition. Fighting tree encroachment and thickets of Smilax spp. is evidently necessary to maintain the distinctive vegetation of serpentine barrens. These habitats are worth protecting because they harbor many rare species, increase landscape-level beta diversity, and make a unique contribution to regional biological diversity. Key words: beta diversity, indicator species, rank-abundance, savanna, succession

In forested landscapes, places that are too wet, naturally occurring, persistently open habitats too dry, or otherwise inimical to closed forests contribute disproportionately to regional diversity provide habitat for a flora that contrasts strongly (e.g., Flinn et al. 2008). Serpentine barrens in with surrounding vegetation, and they thus in- eastern North America are perhaps the best example of persistently open habitats in contrast- crease the landscape-level biological diversity ing matrices that support exceptional flora (Raja- more than their area might suggest (Anderson et karuna et al. 2009). al. 1999, Wright et al. 2002, Marks et al. 2008). Serpentine barrens are globally rare habitats that Because many parts of eastern North America occur on the unusual soils derived from ultramafic were largely forested for most of the Holocene, bedrock; these soils tend to be thin and nutrient poor, with low Ca:Mg ratios and high heavy metal 1 The authors thank Franklin and Marshall College for levels, especially nickel (Whittaker 1954, Walker supporting this project. We thank Tim Draude for help 1954, Harrison and Rajakaruna 2011). So called with plant identification; Jay Gregg for unpublished data; Roger Latham for advice and unpublished data; Eric because of their lack of utility for timber or Lonsdorf for help with beta diversity analyses; Kirk Miller agriculture, barrens consist of mosaics of plant for statistical discussions; and Janet Fischer, Sybil Gotsch, communities, ranging from savannas, in which Afshan Ismat, Roger Latham, Peter Marks, Mark Olson, scattered trees make up 10–25% of total cover, Tim Sipe, and two reviewers for comments on the manuscript. Thanks to Lancaster County Conservancy, through woodlands, with 25–60% tree cover, to The Nature Conservancy, Nottingham County Park forests, with . 60% tree cover, as well as several (Chester County, PA), Pennsylvania Department of wetland types (Latham and McGeehin 2012, Conservation and Natural Resources, and the Friends of the State Line Serpentine Barrens for their stewardship of Burgess et al. 2015a). Although all these commu- our sites. nities are distinct from surrounding nonserpentine 2 Author for correspondence: kfl[email protected]. Present vegetation, savannas have the greatest conserva- address: Biology Department, Baldwin Wallace tion value because they harbor nearly all the rare University, 275 Eastland Road, Berea, OH 44017. species characteristic of serpentine barrens (Lath- doi: 10.3159/TORREY-D-16-00030 am and McGeehin 2012). Eastern North American ÓCopyright 2017 by The Torrey Botanical Society Received for publication May 17, 2016, and in serpentine barrens have many rare , includ- revised form August 30, 2016; first published April 3, ing at least one endemic variety (Cerastium 2017. velutinum Raf. var. villosissimum (Pennell) J.K. 125 126 JOURNAL OF THE TORREY BOTANICAL SOCIETY [VOL. 144

Morton; Gustafson et al. 2003), one near-endemic the most detailed vegetation history. Using aerial species (Symphyotrichum depauperatum (Fernald) photographs, historical accounts, and ground Nesom) that occurs only on serpentine barrens in truthing, MacLachlan (1993) and Arabas (2000) Pennsylvania and Maryland and three mafic documented a general shift from open to closed diabase glades in North Carolina (Gustafson and habitats in the Nottingham barrens from 1937 to Latham 2005), and possibly several other endemic 1993. Nottingham barrens lost half of its savanna species (Harris and Rajakaruna 2009, Boufford et habitat during that period, with 30% shifting to al. 2014). There is also evidence of ecotypic oak-pine (Quercus-Pinus) woodland, 21% to pine differentiation on eastern North American serpen- woodland, and 8% to nonoak-hardwood forest tine soils (Burgess et al. 2015b). Many other (Arabas 2000). Historical, aerial photographs show plants, including some species whose main range a similar shift from savanna habitats to closed is in western prairies, are regionally restricted to forests in the Pilot barrens (Tyndall 1992b, Latham serpentine barrens and are, thus, rare or endan- 1993), New Texas barrens (Barton and Wallenstein gered at the regional, state, or global level (Latham 1997), and Cherry Hill barrens (Burgess et al. 1993, Rajakaruna et al. 2009). Based on both the 2015a). At Cherry Hill, dendrochronology further uniqueness of their vegetation and the rarity of shows an increase in tree recruitment from 1850 to individual species, serpentine barrens represent a 2000 and a shift from xerophytic oaks through high priority for conservation in eastern North conifers to mesophytic, shade-tolerant tree species America. Here, we focus on the State Line (Burgess et al. 2015a). Overall, Maryland serpen- serpentine barrens on the Pennsylvania-Maryland tine barrens lost 75% of their savanna area to tree border, the largest occurrence of serpentine barrens encroachment between 1938 and 1988 (Tyndall in the eastern United States and a prime target for 1992b). conservation efforts (Latham 1993, Rajakaruna et This loss of open habitat is generally thought to al. 2009). result from fire suppression, although the relative One challenge to the conservation of eastern importance of natural and human-set fires, as well serpentine barrens—hardly unique to these habi- as grazing, remains controversial (Tyndall 1992b, tats—is the spread of invasive species. Robinia Latham 1993, Barton and Wallenstein 1997, pseudoacacia L., a North American tree not native Tyndall and Hull 1999, Arabas 2000, Burgess et to eastern Pennsylvania or Maryland (Little 1971), al. 2015a). The rapid afforestation of the 20th has been considered a problem at some State Line century coincided with a period of fire suppression; sites (Latham 1993). A number of other widely problematic invasive species have been recorded at Nottingham, for example, active fire suppression as present at Nottingham barrens, which are the began with a 1936 fire and became increasingly largest barrens in the State Line group and which effective in the 1950s and 1960s (Arabas 2000). At has the largest remaining area of serpentine the same time, increasing fragmentation by savanna (R.E. Latham, personal communication). agriculture, development, and roads created inad- Several of these are shrubs with the potential to vertent firebreaks (Latham 1993). Evidence from change the physiognomy of serpentine grasslands natural and prescribed burns also supports the idea and savannas. However, before this study, we that fire can maintain the openness of serpentine lacked data on which invasive species were present vegetation (Miller 1981, Tyndall 1994, R.E. across the State Line serpentine barrens, how Latham, personal communication). For example, frequent and abundant they were, and what a series of natural fires effectively removed Pinus impacts they may have on serpentine vegetation. virginiana Mill. and Juniperus virginiana L. from Another threat to serpentine vegetation in this the Rock Springs barrens (Miller 1981). In fact, the area is tree encroachment into open habitats. ubiquity of tree encroachment on eastern serpen- Historical evidence indicates that the grasslands tine barrens in the absence of fire leads some to and savannas of the State Line serpentine barrens conclude that ‘‘evidence for an edaphic climax declined drastically in size during the 20th century, community is lacking’’ (Tyndall and Hull 1999, p. becoming denser woodlands or closed forests 71), or at least that disturbance, as well as unique (Tyndall 1992b, Latham 1993, Barton and Wallen- soil conditions, has an important role in shaping stein 1997, Tyndall and Hull 1999, Arabas 2000, serpentine plant communities (Latham 1993, Burgess et al. 2015a). Nottingham barrens have Arabas 2000, Burgess et al. 2015a). 2017] FLINN ET AL.: DIVERSITY AND COMPOSITION OF SERPENTINE BARRENS 127

Table 1. Locations of preserves in the State Line serpentine barrens, southeast Pennsylvania, USA, where we sampled plant diversity and community composition.

Location Latitude Longitude Elevation at sites (m) Chrome Barrens 39844015.8500 N75857010.8200 W 138 Goat Hill Barrens 39843040.5700 N76804016.8500 W 122 Nottingham County Park 39844007.1600 N76802031.2400 W 121 Rock Springs Preserve 39843034.6400 N76809057.3900 W 124

Regardless of its cause, tree encroachment conservation issues, our vegetation survey of two threatens serpentine vegetation because most of serpentine plant communities provides a much- the species of special concern are shade-intolerant needed, quantitative characterization of these plants that do not occur in forests (Rhoads and unique habitats. Block 2007). In addition, Pinus virginiana inva- sion ameliorates the soil in ways that facilitate Materials and Methods. STUDY AREA.We invasion by other nonserpentine species, creating a conducted this study in the State Line serpentine positive feedback loop (Barton and Wallenstein barrens, an archipelago of sites spread over 155 1997, Cumming and Kelly 2007). Tree clearing km2 on the Pennsylvania-Maryland border, in the increased the cover of certain species in a Piedmont physiographic province, about 15 km 0 0 Maryland barren (Tyndall 1994, Tyndall 2005), east of the Susquehanna River (39843 –39844 N, 0 0 but overall, we do not know how tree encroach- 75857 –76809 W; 75–135 m elevation). The region ment affects plant diversity or community compo- has a growing season of about 23 wk, with a mean sition. In fact, quantitative characterizations of annual temperature of 118C and a mean annual plant communities in eastern serpentine barrens are precipitation of 110 cm (Custer 1985). Barton and largely lacking (Rajakaruna et al. 2009; but see Wallenstein (1997), Casper et al. (2008), and Pope et al. 2010). Likewise, it is unknown whether Burgess et al. (2015a) characterized soils in the invasive species outcompete characteristic serpen- State Line serpentine barrens and found mean pH tine plants; affect light, water, or soil conditions; or values of 6.4–6.6, low nutrient (N–P–K) levels, change community characteristics in eastern ser- extremely low Ca:Mg ratios of 0.15–0.23, and pentine barrens. Invasion and tree encroachment extremely high Ni and Cr levels. We sampled nine could also have interactive effects if, for example, sites across four preserves (Table 1). an invasive species flourished more in woodlands FIELD SAMPLING. We located each site at a than in savannas. Although invasion and tree relatively discrete boundary between open savanna encroachment could be seen as the same process— and woodland vegetation. Savannas were defined unwanted plants increasing in abundance—here, as having 10–25% tree cover and woodlands we distinguish invasion by species not native to the having 25–60% tree cover (Latham and McGeehin region, regardless of their growth form, from tree 2012, Burgess et al. 2015a). We located boundar- encroachment, the spread of trees native to the ies between these two habitats where tree density region. and understory vegetation changed. Perpendicular We explored potential impacts of invasion and to this boundary, we sampled transects (5 m wide tree encroachment on serpentine vegetation by and 50 m long) into each habitat, recording the comparing savannas and woodlands across the presence of all vascular plants in 10 contiguous 5 State Line serpentine barrens. To begin to 3 5-m plots. In those plots, we also recorded the understand how invasive species might change species and number of all trees 1 m tall. We serpentine vegetation, we identified which invasive established a 1 31-m plot within each 5 3 5-m plot species were most frequent in each habitat, to record the percentage of cover by all vascular compared invasive cover between habitats, and plants , 1 m tall, on a scale of 0 to , 1, 1 to , 5, related invasive cover to diversity patterns. To 5to, 25, 25 to , 75, and 75–100. Nomenclature suggest how conversion to woodland might follows Rhoads and Block (2007). We omitted 12 change savanna communities, we compared rich- woodland plots because they contained thickets of ness, evenness, alpha and beta diversity, and Smilax rotundifolia L. that we would have had to community composition between savannas and destroy to enter. Thus, our sample comprised 168 adjacent woodlands. Regardless of these specific plots. 128 JOURNAL OF THE TORREY BOTANICAL SOCIETY [VOL. 144

Aerial photographs from 1937 show that the al. 2007, Chase 2010, Anderson et al. 2011). Thus, woodland areas were then savannas, lacking closed Raup-Crick beta diversity ranges from 0 to 1, with canopies. Their transition from savanna to wood- higher values indicating greater beta diversity. We land is representative of sites throughout the State used the R package (R Foundation for Statistical Line serpentine barrens, in which 50–75% of Computing, Wien, Austria) vegan (Oksanen et al. savanna habitats converted to woodlands during 2014) to calculate Raup-Crick beta diversity for all the 20th century (Tyndall 1992, Latham 1993, pairs of plots in savannas located at different sites Barton and Wallenstein 1997, Arabas 2000, and for all pairs of plots in woodlands located at Burgess et al. 2015a). Many savannas similar to different sites (i.e., plots were paired among sites our savanna sites became pine- and oak-dominated within community types). The resulting 6,250 woodlands similar to our woodland sites (Arabas measures of Raup-Crick beta diversity were 2000). This historical context and the direct pairwise, not independent. Thus, we used a t test evidence from aerial photographs justify viewing to determine whether Raup-Crick beta diversity savannas and woodlands as successional stages. differed between savannas and woodlands, but Comparing directly adjacent habitats is the best with only 166 degrees of freedom, because there design for a chronosequence approach because it were 168, 1 3 1-m plots. The significance level of minimizes any factors other than succession that the result was unchanged under any number of may have contributed to the compositional differ- degrees of freedom from 6 to 6,248. ences between successional stages. Soil develop- Community Composition. To describe patterns ment likely accompanies this chronosequence in community composition, we performed a because tree colonization of serpentine savannas changes soil properties (Barton and Wallenstein nonmetric multidimensional scaling (NMS) ordi- 1997, Cumming and Kelly 2007, Burgess et al. nation on the percentage of cover in 1 3 1-m plots 2015a). in PC-ORD (MjM Software Design, Gleneden Beach, OR) (McCune and Grace 2002, McCune STATISTICAL ANALYSES. Plant Diversity. We com- and Mefford 2006). The ordination used Sørensen pared the species richness, evenness, and Shannon (Bray-Curtis) distance, a random starting point, 50 diversity of 1 3 1-m plots in savanna versus runs with real data, and 249 runs with randomized woodland using two-way ANOVAs with habitat data. We found a two-dimensional solution and (savanna or woodland) and site (N ¼ 9) as varimax-rotated it. A Monte Carlo test was used to predictors, with habitat as a fixed effect and site evaluate whether the ordination extracted stronger as a random effect. For all ANOVAs, we report a axes than expected by chance. Wald v2 test of the type III sums of squares for the To test whether community composition dif- predictor of interest, which was habitat (savanna or fered between savannas and woodlands, we used a woodland). For each habitat, we constructed rank- multiresponse permutation procedure (MRPP) abundance curves—that is, plots of species’ log- based on Sørensen (Bray-Curtis) distance in PC- transformed abundance against their rank in ORD (McCune and Grace 2002, McCune and abundance—to compare the distribution of abun- Mefford 2006). To specify how species composi- dance between savannas and woodlands (Whit- taker 1965). We quantified evenness as Shannon tion differed between savannas and woodlands, we diversity divided by the natural logarithm of tested whether savannas and woodlands differed species richness (Pielou 1969). To compare beta along each ordination axis using ANOVAs with diversity among savannas versus among wood- axis scores as response variables and habitat lands while controlling for any differences in alpha (savanna or woodland) and site (N ¼ 9) as diversity, we used the probabilistic measure of predictors. Habitat was a fixed effect, and site Raup and Crick (1979). That measure can be was a random effect. We also conducted an interpreted as the probability that two samples indicator species analysis (ISA) to identify species share fewer species than expected for samples associated with each habitat (Dufreneˆ and Legen- drawn randomly from the species pool, given their dre 1997). This method combined information on existing differences in richness; it is calculated as species’ concentration of abundance in each the proportion of pairs of communities generated habitat and their faithfulness of occurrence in each by the null model that shares the same number or habitat. We used a Monte Carlo test with 4,999 more species than the actual samples (Vellend et randomizations to evaluate the significance of the 2017] FLINN ET AL.: DIVERSITY AND COMPOSITION OF SERPENTINE BARRENS 129 indicator values, also in PC-ORD (McCune and Grace 2002, McCune and Mefford 2006). Because the native greenbrier S. rotundifolia had extremely high cover in many plots, we considered it separately from other native species in the following analyses. We calculated the percentage of total plant cover in each habitat comprising species in four categories: native species, except S. rotundifolia; S. rotundifolia; exotic, but not invasive, species; and exotic, invasive species. We classified exotic species as invasive if they were listed as such in the US Department of Agriculture (USDA) PLANTS database (USDA Natural Resources Conservation Service 2015). Otherwise, species not native to the region were classified as exotic but not invasive. We compared savannas and woodlands in the percentage of total plant cover comprising these groups using ANOVAs with habitat (savanna or woodland) and site (N ¼ 9) as predictors, with habitat as a fixed effect and site as a random effect. We tested whether the cover of exotic, invasive species correlated with species richness, evenness, Shannon diversity, or the cover of native species, except S. rotundifolia, using Pearson correlations.

Results. PLANT DIVERSITY. Overall we found 162 plant species, of which, 132 were native; 20 exotic, but not invasive; and 10 were exotic and invasive (Appendix 1). Understory species richness ranged from 1 to 15 species m2 in savannas and from 0 to 10 species m2 in woodlands. Savannas had 44% greater species richness than woodlands did (Wald v2 ¼ 60.13, d.f. ¼ 1, P , 0.001; Fig. 1A). The two habitats had similar evenness (Wald v2 ¼ 2.28, d.f. FIG. 1. Plant diversity of savannas versus woodlands in the State Line serpentine barrens, ¼ 1, P ¼ 0.131; Fig. 1B). Savannas had 36% higher southeast Pennsylvania, USA: Mean 6 SE. (A) 2 Shannon diversity than woodlands had (Wald v ¼ Species richness (species m2), (B) Evenness, and 27.22, d.f. ¼ 1, P , 0.001; Fig. 1C). According to (C) Shannon diversity. rank-abundance curves, savannas appeared to have a more even distribution of abundance than woodlands had (Fig. 2). Raup-Crick beta diversity Microstegium vimineum (Trin.) A. Camus, which was greater among savannas than among wood- occurred in 39% of savanna plots and 61% of lands (t ¼ 6.152, d.f. ¼ 166, P , 0.001). woodland plots. The native trees J. virginiana, Acer rubrum L., Pinus rigida Mill., and Prunus COMMUNITY COMPOSITION. Of the 162 species we serotina Ehrh. ranked next in frequency. The near- found across all plots, 81 occurred in both savannas and woodlands, 51 in savannas only, endemic forb S. depauperatum (Gustafson and and 30 in woodlands only. The most frequent Latham 2005) was also among the most frequent species were the native greenbrier S. rotundifolia, species, occurring in 62% of savanna plots and 4% which occurred in 80% of savanna plots and 90% of woodland plots. Next-most frequent were the of woodland plots; the native Rubus allegheniensis native grasses Sorghastrum nutans (L.) Nash and Porter, which occurred in 64% of savanna plots Schizachyrium scoparium (Michx.) Nash. After M. and 35% of woodland plots; and the invasive grass vimineum, the most-frequent invasive species were 130 JOURNAL OF THE TORREY BOTANICAL SOCIETY [VOL. 144

FIG. 2. Rank-abundance curves (Whittaker 1965), of species’ log-transformed abundance against their rank in abundance, for savannas and woodlands in the State Line serpentine barrens, southeast Pennsylvania, USA. the woody vine Lonicera japonica Thunb., in 19% rotundifolia. Seventeen species emerged as signif- of savanna plots and 24% of woodland plots, and icant indicators of savanna and five of woodland the shrub Elaeagnus umbellata Thunb., in 22% of within the barrens (Table 2). The strongest savanna plots and 11% of woodland plots. indicators of savanna were the native grasses S. Savannas had a tree density (mean 6 SE) of 1.2 scoparium and S. heterolepis, whereas the stron- 6 0.23 trees 1 m tall per 25 m2, and woodlands gest indicator of woodland was S. rotundifolia had 3.3 6 0.28 trees 1 m tall per 25 m2. The (Table 2). most common trees were A. rubrum, J. virginiana, In savannas, most of the plant cover (mean 6 P. rigida, and P. serotina. SE, 67 6 3.8%) consisted of native species other The NMS ordination had a final stress of 24.67; than S. rotundifolia, whereas in woodlands, only a similar final stress was unlikely to have been 32 6 3.7% of cover consisted of native species, obtained by chance (P ¼ 0.0040). Axis 1 explained other than S. rotundifolia, and S. rotundifolia 21% of the variation in community composition, comprised 52 6 4.5% of cover (Fig. 4). Exotic and axis 2 explained an additional 22%. Axis 1 invasive species comprised 16% of cover across represented a gradient from the invasive species M. both habitats. The percentage of total plant cover vimineum and L. japonica, at lower values, to the comprising native species, except S. rotundifolia, native grasses Sporobolus heterolepis (A. Gray) A. was twice as high in savannas as it was in Gray and S. scoparium and the native forbs woodlands (Wald v2 ¼ 81.60, d.f. ¼ 1, P , 0.001). anonyma (A.W. Wood) W.A. Weber & Smilax rotundifolia represented four times more of A´ .L¨ove and S. depauperatum, at higher values the total plant cover in woodlands than it did in (Fig. 3A). Axis 2 represented a gradient from the savannas (Wald v2 ¼ 119.96, d.f. ¼ 1, P , 0.001). native greenbrier S. rotundifolia, at lower values, The percentage of cover comprising exotic, but not to a number of species, including the native grass invasive, species was slightly higher in savannas S. nutans, at higher values (Fig. 3A). than it was in woodlands (Wald v2 ¼ 4.55, d.f. ¼ 1, Savannas and woodlands differed in species P ¼ 0.033). Exotic, invasive species comprised a composition according to MRPP (A ¼ 0.05734, P similar proportion of cover in both habitats (Wald , 0.001). Savannas had higher scores than v2 ¼ 0.090, d.f. ¼ 1, P ¼ 0.765). The cover of woodlands along both axis 1 (Wald v2 ¼ 88.94, exotic, invasive species did not correlate with d.f. ¼ 1, P , 0.001) and axis 2 (Wald v2 ¼ 98.42, species richness, evenness, or Shannon diversity d.f. ¼ 1, P , 0.001) of the NMS ordination (Fig. (all P . 0.18). However, the cover of native 3B). This indicates, for example, that savannas had species other than S. rotundifolia declined with greater cover of many native grasses and forbs, increasing cover of exotic, invasive species (r ¼ whereas woodlands were more dominated by S. 0.272, P , 0.001). 2017] FLINN ET AL.: DIVERSITY AND COMPOSITION OF SERPENTINE BARRENS 131

FIG. 3. Nonmetric multidimensional scaling (NMS) ordination of plant community composition (percentage of cover) in savannas and woodlands in the State Line serpentine barrens, southeast Pennsylvania, USA. (A) Joint plot showing relationships between ordination axes and individual plant species. The angles and lengths of the vectors indicate the direction and strength of Pearson correlations between species’ axis scores and their percentage of cover. The plot shows only species with correlations jrj . 0.30. Species abbreviations are the first three letters of genus and species names (full names in Results, Table 2, and the Appendix). (B) Locations of savanna and woodland plots (N ¼ 168) on ordination axes, with 68% confidence ellipses.

Table 2. Indicator values (Dufreneˆ and Legendre 1997) of plant species for savannas and woodlands in the State Line serpentine barrens, southeast Pennsylvania, USA. The values range from 0 (no indication) to 100 (perfect indication—the species is always present in that habitat and never present in the other habitat). Within each habitat, species are ranked by strength of indication. The table shows only species with indicator values significant at P , 0.05. The table also indicates species’ origin as native (N), exotic but not invasive (E), or exotic invasive (EI). We classified exotic species as invasive (EI) if they were listed as such in the USDA PLANTS database (USDA Natural Resources Conservation Service 2015). Otherwise, species not native to the region were classified as exotic but not invasive (E).

Species Origin Savanna Woodland P values Indicators of savanna Schizachyrium scoparium (Michx.) Nash N 41 0 0.0002 Sporobolus heterolepis (A. Gray) A. Gray N 32 0 0.0002 Rubus allegheniensis L. N 30 7 0.0142 Symphyotrichum depauperatum (Fernald) Nesom N 28 0 0.0002 Packera anonyma (A.W. Wood) W.A. Weber & A´ .L¨ove N 26 0 0.0004 Potentilla canadensis L. N 25 1 0.0002 Solidago nemoralis Aiton N 20 0 0.0002 Pycnanthemum tenuifolium Schrad. N 20 0 0.0002 Conyza canadensis (L.) Cronquist N 15 0 0.0016 Ambrosia artemisiifolia L. N 15 0 0.0002 Viola sagittata Aiton N 13 0 0.0006 Achillea millefolium L. E 13 1 0.0254 Setaria parviflora Herrm. N 10 0 0.0058 Oxalis stricta L. N 9 0 0.0220 Rosa multiflora Thunb. ex Murray EI 8 0 0.0372 Muhlenbergia glomerata (Willd.) Trin. N 7 0 0.0308 Sabatia angularis (L.) Pursh N 7 0 0.0340 Indicators of woodland Smilax rotundifolia L. N 15 57 0.0002 Smilax glauca Walter N 0 13 0.0044 Lindera benzoin (L.) Blume N 0 11 0.0090 Vaccinium angustifolium Aiton N 0 10 0.0058 Spiraea latifolia (Aiton) Borkh. N 0 8 0.0372 132 JOURNAL OF THE TORREY BOTANICAL SOCIETY [VOL. 144

FIG. 4. Percentage of total plant cover (mean 6 SE) comprising species in four categories: native species, except Smilax rotundifolia L.; S, rotundifolia; exotic, but not invasive, species; and exotic, invasive species in savannas and woodlands in the State Line serpentine barrens, southeast Pennsylvania, USA. We classified exotic species as invasive if they were listed as such in the USDA PLANTS database (USDA Natural Resources Conservation Service, 2015). Otherwise, species not native to the region were classified as exotic but not invasive.

Discussion. This study provides much-needed species added another 2%, levels comparable to information about several conservation challenges other local habitats (e.g., Flinn et al. 2014). Similar in eastern serpentine barrens. First, certain invasive to our study site, Harrison et al. (2006a) found no species are a cause for concern. In particular, the detectable effects of exotic cover on native species invasive grass M. vimineum occurred in almost richness in California serpentine soils, but they did half of all plots and, where present, comprised 41 not examine relationships with native cover or 6 4.7% (mean 6 SE) of total plant cover. The community composition. The full impacts of invasive woody vine L. japonica and the invasive invasive species on serpentine communities, both shrub E. umbellata also occurred in about one fifth direct and through their modifications of light, of all plots. Future research and management water, or nutrient availability, remain to be seen. efforts should focus on those species. In the NMS The functional similarity of invaders to natives ordination, the cover of both M. vimineum and L. may determine both their invasion success and japonica was negatively associated with the cover their effects on serpentine communities, so func- of many native grasses and forbs, including the tional comparisons of serpentine natives and globally rare, near-endemic S. depauperatum invasive species are particularly needed (Hooper (Gustafson and Latham 2005). Although invasive and Dukes 2010, Harrison and Rajakaruna 2011). cover did not relate to diversity patterns, it did Our comparison of savannas and adjacent appear to reduce the cover of native species other woodlands suggests the potential effects tree than S. rotundifolia. In addition, invasion and tree encroachment has on plant diversity and commu- encroachment likely have interactive effects; nity composition. If savannas become woodlands, although, overall, invasive species cover was they will likely lose diversity. Species richness similar between habitats, M. vimineum was much may decline by almost one half. Tree encroach- more frequent in woodlands, suggesting that tree ment may also homogenize community composi- encroachment may facilitate its spread. tion across landscapes because woodlands are In California, serpentine soils are considerably more similar to each other than savannas are to less invaded than nonserpentine soils, apparently each other. Particular species of concern may be because of edaphic conditions, but they still have excluded as tree encroachment proceeds because an average exotic herbaceous cover of 20% 31% of the species we found were unique to (Huenneke et al. 1990, Harrison et al. 2001, savannas. Reductions in the abundance of native Harrison et al. 2006a, Vallano et al. 2012). At our plants with narrowly restricted distributions are State Line sites, invasive species comprised 16% especially concerning because those populations of total plant cover, and exotic, but not invasive, are vulnerable to extirpation from chance events 2017] FLINN ET AL.: DIVERSITY AND COMPOSITION OF SERPENTINE BARRENS 133 and normal fluctuations and the genetic conse- savannas was unusually high (Harrison and quences of small population sizes. In many eastern Inouye 2002). The State Line serpentine barrens serpentine barrens, fighting tree encroachment is share several dominant native grasses and forbs likely among the most effective ways to preserve with the other remaining serpentine barrens in diversity. Pennsylvania and Maryland (Tyndall 1992a, The most obvious cause of woodlands’ low Tyndall and Hull 1999). Together, these habitats diversity is the overwhelming dominance of a undoubtedly raise the beta diversity of the entire single species, the native greenbrier S. rotundifolia, region because many of the species we found do which comprised more than one half of the total not occur in other local habitats (Keever 1973, plant cover in woodlands. This species’ dominance Rhoads and Block 2007). The contribution of was largely responsible for the lower evenness of serpentine barrens to the biological diversity of abundance in woodlands, according to rank- eastern North American landscapes is unique, abundance curves. In the NMS ordination, the valuable, and now imperiled by multiple threats. cover of S. rotundifolia was negatively associated with the cover of many native species. These Literature Cited results confirm quantitatively what is readily ANDERSON, M. J., T. O. CRIST,J.M.CHASE,M.VELLEND,B. apparent in the field—the understory of many D. INOUYE,A.L.FREESTONE,N.J.SANDERS,H.V. serpentine woodlands is a dense, impenetrable CORNELL,L.S.COMITA,K.F.DAVIES,S.P.HARRISON, thicket of S. rotundifolia, in which few other plants N. J. B. KRAFT,J.C.STEGEN, AND N. G. SWENSON. 2011. can grow. Thickets of Smilax spp. have been a Navigating the multiple meanings of b diversity: a component of eastern serpentine barrens for at roadmap for the practicing ecologist. Ecol. Lett. 14: least 100 yr. Describing Pennsylvania barrens in 19–28. ANDERSON, R. C., J. S. FRALISH, AND J. M. BASKIN, eds. 1903, Harshberger (p. 342) wrote, ‘‘[w]hen the 1999. Savannas, Barrens, and Rock Outcrop Plant growth of these trees is dense the serpentine areas Communities of North America. Cambridge Universi- are rendered impenetrable in many places by the ty Press, Boston, MA. green briars. . . which festoon the trees and ARABAS, K. B. 2000. Spatial and temporal relationships intertwine with each other to form a dark gloomy among fire frequency, vegetation, and soil depth in an forest. . ..’’ The shade of increasing tree densities eastern North American serpentine barren. J. Torrey Bot. Soc. 127: 51–65. may allow S. rotundifolia to outcompete plants BARTON,A.M.AND M. D. WALLENSTEIN. 1997. Effects of more characteristic of open areas. Currently, in invasion of Pinus virginiana on soil properties in eastern serpentine barrens, this aggressive native serpentine barrens in southeastern Pennsylvania. J. species may pose a greater threat to plant diversity Torrey Bot. Soc. 124: 297–305. than the invasive species. BOUFFORD, D. E., J. T. KARTESZ,S.SHI, AND R. ZHOU. 2014. Aside from conservation threats, this study Packera serpenticola (; ), a documented the high diversity and uniqueness of new species from North Carolina, U.S.A. Syst. Bot. 39: 1027–1030. eastern serpentine vegetation. Serpentine savan- BURGESS, J., K. SZLAVECZ,N.RAJAKARUNA,S.LEV, AND C. nas had 47% higher species richness per square SWAN. 2015a. Vegetation dynamics and mesophication meter than local mesic forests had (Flinn et al. in response to conifer encroachment within an 2014). California serpentine had 63.2 6 20.2 ultramafic system. Aust. J. Bot. 63: 292–307. species per 1,000 m2 (mean 6 SD; Harrison et al. BURGESS, J., K. SZLAVECZ,N.RAJAKARUNA, AND C. SWAN. 2006b), but our serpentine savannas had 43.8 6 2015b. Ecotypic differentiation of mid-Atlantic Quer- 9.0 species in only a quarter of that area. Many of cus species in response to ultramafic soils. Aust. J. Bot. 63: 308–323. these species are rare, of special concern, or CASPER, B. B., S. P. BENTIVENGA,B.JI,J.H.DOHERTY,H. regionally restricted to serpentine barrens M. EDENBORN, AND D. J. GUSTAFSON. 2008. Plant-soil (Rhoads and Block 2007; Harris and Rajakaruna feedback: testing the generality with the same grasses 2009; Rajakaruna et al. 2009; R.E. Latham, in serpentine and prairie soils. Ecology 89: 2154–2164. personal communication). Symphyotrichum de- CHASE, J. M. 2010. Stochastic community assembly causes pauperatum, which occurred in 62% of our higher biodiversity in more productive environments. Science 328: 1388–1391. savanna plots, is globally imperiled and near CUMMING,J.R.AND C. N. KELLY. 2007. Pinus virginiana endemic to Pennsylvania-Maryland serpentine invasion influences soils and arbuscular mycorrhizae barrens (Gustafson and Latham 2005). Similar of a serpentine grassland. J. Torrey Bot. Soc. 134: 63– to California, the beta diversity of our serpentine 73. 134 JOURNAL OF THE TORREY BOTANICAL SOCIETY [VOL. 144

CUSTER, B. H. 1985. Soil Survey of Lancaster County LATHAM, R. E. 1993. The serpentine barrens of temperate Pennsylvania. USDA Soil Conservation Service, eastern North America: critical issues in the manage- Washington, DC. ment of rare species and communities. Bartonia 57S: DUFRENEˆ ,M.AND P. L EGENDRE. 1997. Species assemblages 61–74. and indicator species: the need for a flexible asym- LATHAM,R.E.AND M. MCGEEHIN. 2012. Unionville metrical approach. Ecol. Monogr, 67: 345–366. Serpentine Barrens Restoration and Management Plan. FLINN, K. M., J. L. BECHHOFER, AND M. MALCOLM. 2014. Continental Conservation, Rose Valley, PA, and Little impact of the invasive shrub Japanese barberry Natural Lands Trust, Media, PA. (Berberis thunbergii DC) on forest understory plant LITTLE,E.L.JR. 1971. Atlas of United States Trees. communities. J. Torrey Bot. Soc. 141: 217–224. Volume 1: Conifers and Important Hardwoods. USA FLINN, K. M., M. J. LECHOWICZ, AND M. J. WATERWAY. 2008. Department of Agriculture Publication 1146, Wash- Plant species diversity and composition of wetlands ington, DC. within an upland forest. Am. J. Bot. 95: 1216–1224. MACLACHLAN, C. A. 1993. Nottingham County Park: the GUSTAFSON,D.J.AND R. E. LATHAM. 2005. Is the serpentine restoration and management of a temperate eastern aster, Symphyotrichum depauperatum (Fern.) Nesom, serpentine barrens. M.Sc. Thesis. University of a valid species and actually endemic to eastern Georgia, Athens, GA. serpentine barrens? Biodivers. Conserv. 14: 1445– MARKS, P. L., F. R. WESLEY, AND S. GARDESCU. 2008. The 1452. diversity of the Finger Lakes region of GUSTAFSON, D. J., G. ROMANO,R.E.LATHAM, AND J. K. central New York state: changes in the 1800s and MORTON. 2003. Amplified fragment length polymor- 1900s. J. Torrey Bot. Soc. 135: 53–69. phism analysis of genetic relationships among the MCCUNE,B.AND J. B. GRACE. 2002. Analysis of serpentine barrens endemic Cerastium velutinum Ecological Communities. MjM Software Design, Rafinesque var. villosissimum Pennell (Caryophylla- Gleneden Beach, OR. ceae) and closely related Cerastium species. J. Torrey MCCUNE,B.AND M. J. MEFFORD. 2006. PC-ORD: Bot. Soc. 130: 218–223. Multivariate Analysis of Ecological Data, Version HARRIS,T.AND N. RAJAKARUNA. 2009. Adiantum viridi- 5.31. MjM Software Design, Gleneden Beach, OR. montanum, Aspidotis densa, Minuartia marcescens, MILLER, G. L. 1981. Secondary succession following fire and Symphyotrichum rhiannon: additional serpentine on a serpentine barren. Proc. Pa. Acad. Sci. 55: 62–64. endemics from eastern North America. Northeast. Nat. OKSANEN, J., F. G. BLANCHET,R.KINDT,P.LEGENDRE,P.R. 16: 111–120. MINCHIN,R.B.O’HARA,G.L.SIMPSON,P.SOLYMOS, HARRISON,S.AND B. D. INOUYE. 2002. High b diversity in M. H. H. STEVENS, AND H. WAGNER. 2014. Vegan: the flora of Californian serpentine ‘islands.’ Biodivers. community ecology package. Version 2.2-0. Re- Conserv. 11: 1869–1876. trieved August 1, 2015, from Comprehensive R HARRISON,S.AND N. RAJAKARUNA. 2011. Serpentine: The Archive Network. ,http://CRAN.R-project.org/ Evolution and Ecology of a Model System. University package¼vegan.. of California Press, Berkeley, CA. PIELOU, E. C. 1969. An Introduction to Mathematical HARRISON, S., K. RICE, AND J. MARON. 2001. Habitat Ecology. John Wiley and Sons, New York, NY. patchiness promotes invasion by alien grasses on POPE, N., T. B. HARRIS, AND N. RAJAKARUNA. 2010. serpentine soil. Biol. Conserv. 100: 45–53. Vascular plants of adjacent serpentine and granite HARRISON, S., J. B. GRACE,K.F.DAVIES,H.D.SAFFORD, outcrops on the Deer Isles, Maine, U.S.A. Rhodora AND J. H. VIERS. 2006a. Invasion in a diversity hotspot: 112: 105–141. exotic cover and native richness in the Californian RAJAKARUNA, N., T. B. HARRIS, AND E. B. ALEXANDER. serpentine flora. Ecology 87: 695–703. 2009. Serpentine geoecology of eastern North Amer- HARRISON, S., H. D. SAFFORD,J.B.GRACE,J.H.VIERS, AND ica: a review. Rhodora 111: 21–108. K. F. DAVIES. 2006b. Regional and local species RAUP,D.M.AND R. E. CRICK. 1979. Measurement of richness in an insular environment: serpentine plants faunal similarity in paleontology. J. Paleontol. 53: in California. Ecol. Monogr. 76: 41–56. 1213–1277. HARSHBERGER, J. W. 1903. The flora of the serpentine RHOADS,A.F.AND T. A. BLOCK. 2007. The Plants of barrens of southeast Pennsylvania. Science 18: 339– Pennsylvania, 2nd ed. University of Pennsylvania, 343. Philadelphia, PA. HOOPER,D.U.AND J. S. DUKES. 2010. Functional TYNDALL, R. W. 1992a. Herbaceous layer vegetation on composition controls invasion success in a California Maryland serpentine. Castanea 57: 264–272. serpentine grassland. J. Ecol. 98: 764–777. TYNDALL, R. W. 1992b. Historical considerations of HUENNEKE, L. F., S. P. HAMBURG,R.KOIDE,H.A.MOONEY, conifer expansion in Maryland serpentine ‘‘barrens.’’ AND P. M. VITOUSEK. 1990. Effects of soil resources on Castanea 57: 123–131. plant invasion and community structure in Californian TYNDALL, R. W. 1994. Conifer clearing and prescribed serpentine grassland. Ecology 71: 478–491. burning effects to herbaceous layer vegetation on a KEEVER, C. 1973. Distribution of major forest species in Maryland serpentine ‘‘barren.’’ Castanea 59: 255–273. southeastern Pennsylvania. Ecol. Monogr. 43: 303– TYNDALL, R. W. 2005. Twelve years of herbaceous 327. vegetation change in oak savanna habitat on a 2017] FLINN ET AL.: DIVERSITY AND COMPOSITION OF SERPENTINE BARRENS 135

Maryland serpentine barren after Virginia pine remov- BELLEMARE,O.HONNAY,J.BRUNET,M.WULF,F. al. Castanea 70: 287–297. GERHARDT, AND M. HERMY. 2007. Homogenization of TYNDALL,R.W.AND J. C. HULL. 1999. Vegetation, flora, forest plant communities and weakening of species- and plant physiological ecology of serpentine barrens environment relationships via agricultural land use. J. of eastern North America, p. 67–82. In R. C. Ecol. 95: 565–573. Anderson, J. S. Fralish and J. M. Baskin [eds.], WALKER, R. B. 1954. The ecology of serpentine soils, II: Savannas, Barrens, and Rock Outcrop Plant Commu- factors affecting plant growth on serpentine soils. nities of North America. Cambridge University Press, Ecology 35: 259–266. Boston, MA. WHITTAKER, R. H. 1954. The ecology of serpentine soils, I: USDA NATURAL RESOURCES CONSERVATION SERVICE. 2013. introduction. Ecology 35: 258–259. The PLANTS Database. Retrieved August 1, 2015, from National Plant Data Center, Baton Rouge, LA. WHITTAKER, R. H. 1965. Dominance and diversity in land ,http://plants.usda.gov.. plant communities: numerical relations of species VALLANO, D. M., P. C. SELMANTS, AND E. S. ZAVALETA. express the importance of competition in community 2012. Simulated nitrogen deposition enhances perfor- function and evolution. Science 147: 250–260. mance of an exotic grass relative to native serpentine WRIGHT, J. P., C. J. JONES, AND A. S. FLECKER. 2002. An grassland competitors. Plant Ecol. 213: 1015–1026. ecosystem engineer, the beaver, increases species VELLEND, M., K. VERHEYEN,K.M.FLINN,H.JACQUEMYN, richness at the landscape scale. Oecologia 132: 96– A. KOLB,H.VAN CALSTER,G.PETERKEN,B.J.GRAAE,J. 101.

Appendix Vascular plant species that were found in the State Line serpentine barrens, southeast Pennsylvania, USA. Nomenclature follows Rhoads and Block (2007). The Appendix Table indicates species’ origin as native (N), exotic but not invasive (E), or exotic invasive (EI). We classified exotic species as invasive (EI) if they were listed as such in the USDA PLANTS database (USDA Natural Resources Conservation Service 2015). Otherwise, species not native to the region were classified as exotic but not invasive (E). Species are listed in order of their frequency, the percentage of plots in which the species occurred (N ¼ 168, 5 3 5-m plots). A few taxa were identified to genus, and for a few species, we specified the variety, but we call the taxa ‘‘species’’ throughout the article because most of the taxa are species.

Family Species Origin Frequency Smilacaceae Smilax rotundifolia L. N 85 Rosaceae Rubus allegheniensis Porter N 51 Poaceae Microstegium vimineum (Trin.) A. Camus EI 49 Cupressaceae Juniperus virginiana L. N 48 Sapindaceae Acer rubrum L. N 47 Asteraceae Symphyotrichum depauperatum (Fernald) Nesom N 35 Pinaceae Pinus rigida Mill. N 35 Rosaceae Prunus serotina Ehrh. N 35 Poaceae Sorghastrum nutans (L.) Nash N 32 Poaceae Schizachyrium scoparium (Michx.) Nash N 31 Poaceae Sporobolus heterolepis (A. Gray) A. Gray N 30 Rosaceae Potentilla canadensis L. N 27 Lamiaceae Pycnanthemum tenuifolium Schrad. N 26 Poaceae Dichanthelium microcarpon (Muhl. ex Elliott) Mohlenbr. N 25 Asteraceae Solidago nemoralis Aiton N 24 Lauraceae Sassafras albidum (Nutt.) Nees N 24 Asteraceae Achillea millefolium L. E 24 Asteraceae Packera anonyma (A.W. Wood) W.A. Weber & A´ .L¨ove N 23 Asteraceae Solidago altissima L. N 23 Caprifoliaceae Lonicera japonica Thunb. EI 21 Poaceae Eragrostis capillaris (L.) Nees N 21 Asteraceae Ambrosia artemisiifolia L. N 20 136 JOURNAL OF THE TORREY BOTANICAL SOCIETY [VOL. 144

Appendix Continued.

Family Species Origin Frequency Ericaceae Vaccinium angustifolium Aiton N 19 Asteraceae Conyza canadensis (L.) Cronquist N 18 Smilacaceae Smilax glauca L. N 18 Oxalidaceae Oxalis stricta L. N 18 Elaeagnaceae Elaeagnus umbellata Thunb. EI 17 Fagaceae Quercus marilandica Mu¨nchh. N 17 Apocynaceae Asclepias verticillata L. N 16 Caprifoliaceae Lonicera morrowii A. Gray EI 16 Anacardiaceae Rhus copallina L. N 15 Asteraceae Ageratina aromatica (L.) Spach N 15 Caryophyllaceae Cerastium velutinum Raf. N 14 Gentianaceae Sabatia angularis (L.) Pursh N 14 Poaceae Setaria parviflora (Poir.) Kergue´len N 14 Rosaceae Rosa multiflora Thunb. ex Murray EI 14 Lauraceae Lindera benzoin (L.) Blume N 13 Anacardiaceae Toxicodendron radicans (L.) Kuntze N 13 Fagaceae Quercus prinoides Willd. N 11 Fagaceae Quercus stellata Wangenh. N 11 Rosaceae Spiraea latifolia (Aiton) Borkh. N 11 Violaceae Viola sagittata Aiton var. sagittata N11 Rosaceae Rubus occidentalis L. N 10 Asteraceae Euthamia graminifolia (L.) Nutt. N 9 Onagraceae Oenothera fruticosa L. N 8 Vitaceae Parthenocissus quinquefolia (L.) Planch. N 8 Pinaceae Pinus virginiana Mill. N 8 Apocynaceae Apocynum cannabinum L. N 7 Fagaceae Quercus palustris Mu¨nchh. N 7 Poaceae Dichanthelium clandestinum (L.) Gould N 7 Polygalaceae Polygala verticillata L. var. isocycla Fernald N 7 Asteraceae Vernonia noveboracensis (L.) Michx. N 6 Cyperaceae Scleria triglomerata Michx. N 6 Poaceae Agrostis gigantea Roth E 6 Poaceae Muhlenbergia glomerata (Willd.) Trin. N 6 Asteraceae Cirsium altissimum (L.) Spreng. E 5 Asteraceae Sonchus arvensis L. E 5 Cyperaceae Bulbostylis capillaris (L.) C.B. Clarke N 5 Fabaceae Robinia pseudoacacia L. E 5 Rhamnaceae Ceanothus americanus L. N 5 Asteraceae Cirsium discolor (Muhl.) Spreng. E 5 Caryophyllaceae Minuartia michauxii (Fernald) Farw. N 5 Poaceae Tridens flavus (L.) A. Hitchc. N 5 Rosaceae Rosa Carolina L. N 5 Anacardiaceae Rhus typhina L. N 4 Asteraceae Erechtites hieraciifolius (L.) Raf. ex DC. E 4 Juncaceae Juncus tenuis Willd. N 4 Polygalaceae Polygala verticillata var. verticillata L. N 4 Aquifoliaceae Ilex opaca Aiton N 4 Asteraceae Solidago juncea Aiton N 4 Asteraceae Symphyotrichum laeve (L.) A´ .L¨ove & D. L¨ove N 4 Dioscoreaceae Dioscorea villosa L. N 4 Rosaceae Rubus idaeus L. N 4 Vitaceae Vitis L. sp.N4 Berberidaceae Berberis thunbergii DC. EI 3 Fagaceae Quercus rubra L. N 3 Poaceae Danthonia spicata (L.) P. Beauv. ex Roem. & Schult. N 3 Polygonaceae Fallopia scandens (L.) Holub N 3 Polygonaceae Persicaria maculosa S.F. Gray E 3 Polypodiaceae Polystichum acrostichoides (Michx.) Schott N 3 Apiaceae Daucus carota L. E 2 2017] FLINN ET AL.: DIVERSITY AND COMPOSITION OF SERPENTINE BARRENS 137

Appendix Continued.

Family Species Origin Frequency Apocynaceae Asclepias viridiflora Raf. N 2 Asteraceae Eupatorium rotundifolium L. N 2 Betulaceae Betula lenta L. N 2 Brassicaceae Arabis lyrata L. N 2 Campanulaceae Lobelia inflata L. N 2 Convolvulaceae Ipomoea pandurata (L.) G. Mey. E 2 Cyperaceae Carex hirsutella Mack. N 2 Fabaceae Baptisia tinctoria (L.) R. Br. N 2 Fabaceae Chamaecrista fasciculata (Michx.) Greene N 2 Fabaceae Strophostyles umbellata (Muhl. ex Willd.) Britton N 2 Fagaceae Quercus velutina Lam. N 2 Hypericaceae Hypericum canadense L. N 2 Hypericaceae Hypericum gentianoides (L.) Britton, Sterns & Poggenb. N 2 Hypericaceae Hypericum mutilum L. N 2 Orobanchaceae Agalinis purpurea (L.) Pennell N 2 Poaceae Poa compressa L. E 2 Rosaceae Pyrus calleryana Decne. EI 2 Violaceae Viola L. sp.N2 Adoxaceae Sambucus canadensis L. N 2 Alliaceae Allium vineale L. E 2 Apocynaceae Asclepias syriaca L. N 2 Asteraceae Eutrochium fistulosum (Barratt) E.E. Lamont N 2 Asteraceae Heliopsis helianthoides (L.) Sweet N 2 Colchicaceae Uvularia perfoliata L. N 2 Cornaceae Cornus florida L. N 2 Fabaceae Desmodium obtusum (Muhl. ex Willd.) DC. N 2 Iridaceae Sisyrinchium angustifolium Mill. N 2 Ophioglossaceae Botrychium virginianum (L.) Sw. N 2 Phytolaccaceae Phytolacca americana L. N 2 Poaceae Anthoxanthum odoratum L. N 2 Poaceae Setaria faberi Herrm. E 2 Polypodiaceae Asplenium platyneuron (L.) Britton, Sterns & Poggenb. N 2 Ruscaceae Polygonatum biflorum (Walter) Elliott N 2 Verbenaceae Verbena urticifolia L. N 2 Asteraceae Centaurea nigra L. E 1 Asteraceae Eupatorium perfoliatum L. N 1 Campanulaceae Lobelia spicata Lam. N 1 Celastraceae Celastrus orbiculatus Thunb. EI 1 Convolvulaceae Cuscuta L. sp.N1 Cyperaceae Carex annectens (E.P. Bicknell) E.P. Bicknell N 1 Cyperaceae Carex swanii (Fernald) Mack. N 1 Fabaceae Amphicarpaea bracteata (L.) Fernald N 1 Fagaceae Quercus alba L. N 1 Magnoliaceae Liriodendron tulipifera L. N 1 Poaceae Aristida dichotoma Michx. N 1 Poaceae Panicum anceps Michx. N 1 Polygonaceae Persicaria sagittata (L.) H. Gross N 1 Polypodiaceae Onoclea sensibilis L. N 1 Polypodiaceae Pteridium aquilinum (L.) Kuhn N 1 Rubiaceae Galium triflorum Michx. N 1 Ruscaceae Maianthemum racemosum (L.) Link N 1 Salicaceae Populus grandidentata Michx. N 1 Smilacaceae Smilax herbacea L. N 1 Adoxaceae Viburnum dentatum L. N 1 Asteraceae Ageratina altissima (L.) R.M. King & H. Rob. N 1 Asteraceae Bidens cernua L. N 1 Asteraceae Hieracium flagellare Willd. E 1 Asteraceae Lactuca canadensis L. N 1 Asteraceae Symphyotrichum undulatum (L.) G.L. Nesom N 1 138 JOURNAL OF THE TORREY BOTANICAL SOCIETY [VOL. 144

Appendix Continued.

Family Species Origin Frequency Asteraceae Taraxacum officinale F.H. Wigg. E 1 Asteraceae Vernonia gigantea (Walter) Trel. N 1 Brassicaceae Alliaria petiolata (M. Bieb.) Cavara & Grande EI 1 Cannabaceae Humulus japonicus Siebold & Zucc. EI 1 Caryophyllaceae Silene stellata (L.) W.T. Aiton N 1 Cistaceae Lechea minor L. N 1 Cyperaceae Carex leptonervia (Fernald) Fernald N 1 Ericaceae Monotropa uniflora L. N 1 Fabaceae Desmodium paniculatum (L.) DC. N 1 Fabaceae Gleditsia triacanthos L. E 1 Fagaceae Fagus grandifolia Ehrh. N 1 Juglandaceae Carya glabra (Mill.) Sweet N 1 Malvaceae Tilia americana L. N 1 Myrsinaceae Lysimachia L. sp.N1 Poaceae Agrostis scabra Willd. N 1 Poaceae Bromus japonicus L. E 1 Poaceae Dichanthelium linearifolium (Scribn.) Gould N 1 Poaceae Leersia virginica Willd. N 1 Polypodiaceae Adiantum pedatum L. N 1 Polypodiaceae Polypodium virginianum L. N 1 Rosaceae Amelanchier arborea (Michx. f.) Fernald N 1 Ulmaceae Ulmus americana L. N 1