The Effects of Fire and Shrub Cutting on Rare and Endangered Plants in the Ninigret National Wildlife Refuge

The decline in rare plant populations and the decrease in size of their native ranges have led to escalating losses of biodiversity. A combination of shrub cutting and fire was applied to an endangered species "hot spot" in southern Rhode Island, USA to determine the effects of these treatments on a group of eight rare herbaceous plant species exposed to ever-increasing encroachment by tall shrubs. Shrub cutting and fire were also applied to a shrub-dominated area adjacent to the hot spot to determine if it was possible to promote establishment and growth of the rare plants in areas in which they were not found in abundance. My analyses focused on the three rare species that were found in greatest abundance within the rare plant area; Aletris farinosa, Platanthera ciliaris, and Polygala cruciata. Fire appeared to have a positive effect on the reproductive output of the endangered orchid Platanthera ciliaris; the mean density of inflorescences doubled from 0. 70 inflorescences/m 2 in 2005 to 1.41 inflorescences/m 2 in 2006 in plots that were burned. However, this difference was not statistically significant. Significant differences in vegetation characteristics were observed between plots in which rare plants were present and plots in which rare plants were absent. Platanthera and Aletris had negative relationships with woody plant cover, and positive relationships with herbaceous plant cover. The density of Aletris was significantly correlated with the density of Drosera sp. (r=0.73, P<0.001), a species indicative of disturbed, nutrient poor environments. Fire had no effect on the cover of woody plants within the rare plant area, however, fire did have an effect on the cover and height of woody plants in adjacent experimental sectors. In 2006, all combinations of shrub cutting and fire decreased the cover and height of woody plants in these sectors . No rare plants were found in the sector plots adjacent to the rare plant area, but the cut treatment applied to one of these sectors caused the rapid proliferation of the invasive woody vine Celastrus orbiculatus. The density of Celastrus was an order of magnitude higher in the cut sector than in the control. In an adjacent sector that was both cut and burned, the native shrub Rhus copallinum produced a similar rapid response to the treatment. Continued management must now focus on limiting the potential harmful impacts of Celastrus while promoting growth and establishment of the rare plants at the site. Acknowledgments I thank Peter Keller for data collection in 2002-2004. I also thank Chelsea Dion for her help with field sampling and site preparation in the summer of 2005, and Rick Enser for his assistance with plant identification in the summer of 2006. I thank James Heltshe for his help and advice on statistical analyses. Thanks to Suz.anne Paton , Charlie Vandermeor, and staff from the U.S. Fish and Wildlife Service for providing the resources and support that made the burning and cutting manipulations for this study possible. I also thank Eric Morales for his assistance in excavating the scarified plots. Laura Meyerso°' Carol Tuomber, and Larry Englander provided helpful feedback and advice regarding this study. Finally, I thank Keith Killingbeck for his guidance and comments, which greatly improved this manuscript.


Introduction 1
Site Description 6 Methods 9

Results 15
Discussion 23 Appendix A 55 AppendixB 56 AppendixC 58 AppendixD 60 Bibliography 63 V List of Tables   Table I: Estimated population sizes of rare plants  34   Table 2: Maximum temperatures reached in burned plots 35 Table 3: Density of stems, density of inflorescences and percent cover of Aletris, Platanthera, and Polygala 36 Table 4: Density and/or percent cover of all other habitat variables sampled 37 Table 5: Density of stems, density of inflorescences and percent cover of Aletris, Platanthera, and Polygala in unburned plots 38   The significant loss of rare plant populations and the decrease in size of their historical ranges is a problem both regionally in New England (Farnsworth and Ogurcak 2006) and throughout the world (Kirkpatrick and Gilffeder 1995, Norton and De Lange 2003, Wotavova et al. 2004). There are, however, extant sites that still support relatively large numbers of rare plant populations, thus contributing to locally high levels of biodiversity (Farnsworth and Ogurcak 2006). These areas are often called "hot spots" of biodiversity and are critical to conservation efforts because preservation and management of relatively small areas of land can protect disproportionately large numbers of rare species (Dobson et al. 1997).
One such endangered species "hot spot" exists within the Ninigret National Wildlife Refuge (NNWR) in Charlestown, Rhode Island (Killingbeck et al. 1998).
An area just 2,300 m 2 in size supports an assemblage of at least eight species of plants considered to be rare or endangered in the state of Rhode Island and found at few other localities in the state (Enser 2002). The "endangered species hot spot" at Ninigret contains more individuals of the state-endangered yellow-fringed orchid, Platanthera ciliaris, than all of the rest of New England combined (Killingbeck et al. 1998). When the site was first discovered in 1993, it was clear that the relatively small area provided habitat for a variety of biologically important plant species.
When earlier field studies were initiated at the site, it was apparent that, while rare plant density was extremely high within the 2,300m 2 core of the hot spot, relatively few rare plants existed beyond the boundaries of this area The curious presence of an abrupt, definitive boundary coupled with an equally abrupt difference in soil organic matter suggested that anthropogenic disturbance played a key role in the development of vegetation at this site (Killingbeck et al. 1998). However, the limited disturbance history of the site offers limited clues to deduce the profound landscape changes that have occurred in the past 100 years. Additionally, there is even a high degree of uncertainty regarding the species composition and structure of all coastal New England vegetation in presettlement times (Day 1953, Foster and Motzkin 2003, Marks 1983).
While disturbance can lead to the outright destruction and elimination of communities, it can also promote variation in the landscape and, in turn, provide conditions necessary to enhance species richness (Kirkpatrick and Gilfedder 1995).
Cultural practices such as mowing, plowing, and burning have perpetuated grasslands and shrub lands for hundreds of years by preventing their progression to forested systems (Foster and Motzkin 2003). However, changes in land use have led to the cessation of certain historical disturbance regimes and, in turn, the preservation of habitats that support many rare and unique species of plants (Hall et al. 2002, Latham 2003. It is necessary to reconcile the historical and environmental influences that have impacted the landscape in order to actively conserve these important habitats (Foster 2002, Foster and Motzkin 2003, Neill et al. 2007).
Rare plants are :frequently found within anthropogenically-disturbed landscapes (Latham 2003, Zaremba 2004, and much work has been done to explore the relationships between rare plants and disturbance history (Baskin et al. 1997, Clarke and Patterson 2006, Vickery 2001. For many rare North Atlantic coastal plant species, disturbance is necessary to provide the environmental conditions that favor their existence at a particular site (Clarke andPatterson 2006, Motzkin andFoster 2002). Fire functions as one such critical disturbance in early successional communities and has been shown to promote growth and flowering in some herbaceous plants (Jacobson et al. 1991). Many studies have shown that flowering in orchids and other geophytes is enhanced by fire (Lamont and Runciman l 993, Whelan 1995), and consequently, larger populations are observed after burning (Lunt 1997, Norton and de Lange 2003, Stuckey 1967. Fire can also have indirect effects, one of which is the removal of competing vegetation (Jacobson et al. 1991 ), which increases both light availability and openings for seedling establishment (Gankin and Major 1964).
Tree and shrub clearing has also been used as a method to promote or restore biodiversity to a system facing increased encroachment of vegetation capable of outcompeting target species (Barbaro et al. 2001). Many rare, herbaceous plants depend on open, short vegetation with patches of bare ground for establishment (Dolman and Southerland 1992). Habitat for low-growing species is often maintained by disturbance or grazing, however, restoration efforts, such as shrub cutting, can simulate such processes (Zobel et al. l 996). In the absence of alternative management, shrubs and trees are likely to dominate communities that were formerly composed of species of shorter stature (Bakker et al. 1996, Bullock andPakeman 1997).
Previous studies at the Ninigret site offered evidence that shrub cutting and/or fire had a positive effect on at least four species of rare plants (Keller and Killingbeck 2004), yet additional experiments were needed to determine whether continued frequent, periodic fire would elicit a similar response. To understand how the practice of prescribed burning might change . the population dynamics of the rare plants at the site, experimental manipulations including a combination of shrub cutting and fire were utilized at the Ninigret Endangered Species Area (NESA) to reduce light limitation due to shading.
An additional goal of the study was to determine whether areas beyond the margins of the rare plant "hot spot" could be manipulated to provide the conditions necessary for establishment and growth of the rare plants. The experimental design included clearing a shrub-dominated area adjacent to the "endangered species hot spot" with three combinations of shrub cutting and fire to determine how areas harboring few, if any, rare species would respond. The discovery of a number of stems of Platanthera ciliaris in 2004 in a mowed firebreak adjacent to the hot spot suggested that such manipulations could be instrumental in increasing the population sizes of at least some of the rare plant species beyond the core area.
The main objective of the study was to determine the efficacy of using shrub cutting and fire to enhance populations of rare plants th.at might otherwise face shadeinduced reductions, or extirpation. This study aimed to address the following specific Island state concern), and Aristida longespica Poiret. (slimspike three-awn; Rhode Island state concern) (Enser 2002). Additionally, Platanthera ciliaris is considered to be regionally rare in New England (Division 2, State Endangered 1, Brumback and Mehrhoff 1996).
Aletris farinosa is one of just five species in the genus Aletris that occurs in North America (Sullivan 1973). Aletris , part of the family Liliaceae, is a geophyte with fleshy, underground tissues that serve reproductive and storage functions (Rutters et al. 1993 (Killingbeck et al. 1998).

Methods
Data previously obtained at this site (in 1996, 2003Keller and Killingbeck 2004) were collected from 61 0.5-m 2 permanent plots distributed throughout the NESA core along 13 transects (Killingbeck et al 1998). · Transects were established perpendicular to a northeast/southwest line that bisected the NESA core, and were positioned in randomized locations within eight-meter intervals.
Along each of the 13 transects, permanent sampling plots were established in randomized locations within four-meter intervals in the spring of 2005.
Randomization within each four-and eight-meter segment was achieved by dividing each interval into one-meter units and then choosing by random draw the meter segment at which each plot would be located. All 87 original plots were sampled for the 2005-2006 study.
In 2005, the area of dense woody shrubs beyond the NESA core was divided by five-meter wide :firebreaks into four sectors and each sector was given a separate experimental treatment (Figure 1 ). This experimental design included plots that were cut and burned (Sector D), cut only (sector C), burned only (Sector B), and left untreated ( control, Sector A, Figure 2). Treatments for the four sectors were assigned randomly. The original 13 transects were extended approximately 30-40m to the northwest and ended at the perimeter :firebreak that surrounds the site. Fifteen new plots were established in each of the four sectors along three expanded transect lines in each sector. The protocol for plot placement explained above was also used to expand the already existing transects and establish a total of 60 new 0.5-m 2 plots in the densely dominated shrub area beyond the original NESA core.
Five plots on each transect line were distributed randomly at four-meter intervals. Due to constraints imposed by the shape of the cut and burn sector, plot placement deviated slightly from the above design (see Figure 1 ). If a four-meter interval fell within one meter of a tree, the four-meter interval was skipped and the plot was established within the next four-meter interval.
Geographic coordinates were taken directly over each permanent plot marker, firebreak boundaries and experimental sector boundaries with a Garmin handheld GPS unit, and were imported into ArcMap. All other geospatial data used for the site map were obtained from the Rhode Island Geographic Information System server (RIGIS 1989 andRIGIS 2005).
Fire and shrub-cutting treatments were formally approved by Suzanne Paton  (9,10,18,19,21,22,31,32,34,35,45,46,48,49,60,61), were also dominance (percent cover), numbers of stems with inflorescences, and maximum height of inflorescences. All plots were also assessed for the number and percent cover of the carnivorous sundew Drosera spp., percent cover of mosses, percent cover of lichens, percent cover and mean height of herbaceous and graminaceous species, percent cover and mean height of woody species, and percent cover of unvegetated soil. Collectively, these latter parameters were thought to be indicative of the recent disturbance history of this site (Brewer 1999, Eldridge et al 2000, Juniper et al 1989, Sedia and Ehrenfeld 2003. Sampling of the scarified plots took place in the summer of 2006 to determine if scarification on the periphery of the NESAcore would promote the establishment of rare plant species. All plant nomenclature follows Gleason and Cronquist (1991)

Statistical Analyses
Within the NESA core, plots were segregated between those burned and those not burned in 2005. Plots considered to be burned were those with a cover class ofB (at least 5% charring) or higher and one or more melted Tempilaq paint strips. Plots considered to be unburned in 2005 were those with a charring class of A (no charring) or B (less than 5% charring), and unmelted Tempilaq paints.
Associated organisms and plot characteristics were analyzed based on two categories of plots: plots in which rare plants were present, termed occupied plots, and plots in which rare plants were absent, termed unoccupied plots. All burned and unburned plots were included in this analysis.
The Lilliefors test was used to determine whether data were normally distributed. Probabilities of differences among treatments were generated with the Kruskal-Wallis statistic to test for differences among bum cover classes and maximum temperatures attained during the bum. Tests for an interaction between year and treatment were generated by comparing the change in burned plots between years and the change in unburned plots between years using the Wilcoxon rank-sum test. The Wilcoxon rank-sum test was also used to test for pairwise differences among burned and unburned plots and habitat characteristics in plots occupied and unoccupied by rare plants.
Percent cover of lichens, moss, unvegetated soil, and Drosera (referred to collectively as "indicators") were thought to be potential indicators of past disturbance that may have facilitated the colonization of the NESA core by rare plants. Cover and mean height of both herbaceous and woody plants were referred to collectively as ''vegetation characteristics." Correlations between the rare plants and indicators were generated from a Pearson correlation matrix. Regression analyses were completed on woody and herbaceous vegetation and rare plant density, number of inflorescences, and cover, and adjusted coefficients of determination were reported.
Statistical analyses of the rare plant data were completed for Aletris, Platanthera and Polygala ( Figure 3). All other species were not conducive to statistical testing due to their extremely low numbers or absence within the NESA core plots. Data on the abundance of these species were generated by active searches within the entire NESA core.
Statistical analyses were completed with SYSTAT software (Wilkinson 1992).

Estimated population sizes
Six of the eight rare species known to exist at the site were documented during the 2005 and 2006 growing seasons (Table 1 ) plots that were unburned had a higher coverage than plots that were burned, however, this difference was not statistically significant ( Figure 6). For plots that were burned, the percent cover of Platanthera was higher in 2006 (3.6% cover) than in 2005 (1.2% cover; Table 3). Polygala had a relatively low coverage in both burned and unburned plots, and differences were not significant in either year of the study (P>0.05 Wilcoxon Rank Sum Test).
Significant year-to-year differences were observed in both the density of Aletris and the percent cover of Platanthera (Table 3). The percent cover of  (Table 4). In 2006, Drosera cover in burned plots was not significantly different than in unburned plots ( Figure 8). However, in 2005 Drosera cover was significantly higher in unburned plots than in burned plots (P=0.024; Wilcoxon Rank Sum Test).
Differences in percent cover of lichens and moss were not significant between any combination of year and treatment (P>0.05; Wilcoxon Rank Sum Test).
The between year decrease in cover of Drosera in unburned plots was significantly greater than the between year decrease in burned plots (P=0.01; Wilcoxon rank sum test, Figure 9). The between year decrease in unvegetated soil in burned plots was significantly different than the between year increase in unburned plots (P=0.001, Wilcoxon Rank Sum Test).
Herbaceous cover increased more from 2005 to 2006 in burned than in unburned plots (P=0.001; Wilcoxon Rank Sum Test, Figure 9). Herbaceous cover was significantly lower in burned than in unburned plots in 2005 (P=0.001; Wilcoxon Rank Sum Test), however, that effect had disappeared by 2006 ( Figure 10). In   (Table 6).

Vegetation and habitat characteristics inside the NESA core
The mean percent cover of Drosera, lichens and moss was significantly different in plots occupied by Aletris and Platanthera than in plots unoccupied by those species (P<0.01; Wilcoxon Rank Sum Test, Figure 11). Cover of unvegetated soil did not differ significantly between occupied and unoccupied plots for any of the rare plants. In plots unoccupied by Aletris, mean Drosera and lichen cover was near  (Table 7). In 2006, woody cover accounted for 31 % of the variance in Aletris density, and 21 % of the variance in Platanthera density. Aletris density, number of inflorescences and cover all were positively related to herbaceous plant cover (Table   7). Platanthera density and cover were positively related to herbaceous cover.
Herbaceous cover explained 30% of the variance in Aletris density and 21 % in  Figure 13). In 2006, the density of Celastrus in the cut sector (11.6 stems/m 2 ) was an order of magnitude higher than in the control sector The woody shrub Rhus copallinum (winged sumac) was extremely abundant in the cut and burn sector, and density was significantly higher there than in any other sector ( Figure 13). In the cut and burn sector, the density of Rhus

Response of rare plants to fire
Past studies on geophyte populations and their response to fire have described increases in flowering or aboveground stem production following a bum (Lamont andRunciman 1993, Verboom et al. 2002). Specifically, fire has been utilized to promote flowering in forbs such as Aletris lutea in the southeastern U.S. (Carter et al. 2004, Platt et al. 1988. Such a response from Aletris farinosa at Ninigret was not discerned in 2005 or 2006, as no significant increase in stem density, cover, and number of inflorescences of Aletris was detected in burned plots.

The extent and timing of the bum in 2005 may offer explanations as to why
Aletris did not respond positively to this trea1ment. It is possible that the fire was applied too late in the growing season for Aletris to respond in 2005. At the time of · the May burn, the basal rosettes of the plant were already above ground and it may have been too late for the plants to immediately benefit from the effects of the fire.
Charring was observed on the basal rosettes of some Aletris stems, but such ·charring was minimal and it did not seem as though these individuals were killed by the bum.
It is likely that a combination of both uneven distribution of the fire relative to the distribution of the plants within the NESA core and the timing of the bum contributed to the lack of an increase of Aletris in burned plots during this study.
The high heterogeneity influenced the ability to determine significant changes in Platanthera as a result of the burning treatment. The increases in reproductive output and percent cover of Platanthera were not statistically significant, however, trends overall suggest that burning may have had a positive effect. This would support similar results in other investigations in the response of rare orchids to fire (Norton andDelange 2003, Wotavova et al. 2004). Like Aletr~s, Platanthera is a geophyte with underground storage for future growth and reproduction. It is believed that disturbances such as fire can stimulate growth and reproduction in orchids (Stuckey 1967, Whelan 1995. Due to the high incidence of rarity in the family Orchidaceae, not only in New England (Brumback andMehrhoff 1996, Enser 2002) but also in many other parts of the world (Kirkpatrick and Gilfedder 1995, Lunt 1997, Silvertown et al. 1994, the suggestion that fire may stimulate reproductive output is an important consideration. Research on fire and rare plants has been increasing since the 1980's and more rapidly since the early 1990's (Hassl and Spackman 1995). Such research is vital to the conservation of some rare plant populations and fire is a recommended tool for increasing rare plant populations (Jacobson et al. 1991), yet it is difficult to elucidate effects without long-term monitoring of at least three years (Menges 1986).
The best course of action toward the recovery of an endangered plant species is a comprehensive management plan that asseses the biological status and demographics of the plant in question, and also includes guidelines for management and recovery (Schemske et al. 1994). Presently, one such plan exists for Platanthera (Sharp 2004).
Once found in greater numbers in New England, only 10 extant sites with populations of Platanthera still remain (out of 41 historic populations). Of the 10 populations still considered extant in New England, two are relatively stable, two have shown decline, and six have not'been found in recent surveys (Sharp 2004). Fire may be an important tool in efforts to restore vigor to those populations facing extirpation.

Response of NESA core vegetation to fire
In examining the effects of fire on the herbaceous and woody vegetation inside the NESA core, it was apparent that the burn did not reduce cover of woody plants or height of herbaceous plants. Differences that were observed in burned plots, specifically cover of herbaceous plants and height of woody plants, were short-lived.
By the year after the burn, cover of herbaceous plants and height of woody plants had both increased to effectively negate the differences observed in 2005 after the burn.
This suggests that the fire did not kill a significant amount of the woody plants inside the NESA core. Results from this study and others (Briggs et al. 2002, Glasgow and Matlack 2007, Heisler et al. 2003 suggest that responses of woody shrubs to fire are much weaker by the second year after a prescribed burn, necessitating a continuation of prescribed fire if long-term reductions in woody vegetation are a goal.
Legacy effects from shrub-cutting treatments in fire exclosed plots in 2003 were not significant in 2005 or 2006 for herbaceous or woody vegetation. This suggests that the shrub cutting treatments, which were applied once before the initiation of this study, were not effective in controlling coverage or height of the vegetation inside the NESA core in the long-term. Others have observed this effect after shrub-cutting treatments were applied in an effort to reduce competition between rare plants and competing vegetation (Barbaro et al. 2001, Gordon 1996, Muller 2002). However, the mean height of the vegetation was significantly lower in the unburned plots than in the burned plots, despite the fact that no past treatments had taken place in 13 of the unburned plots. This limited the ability to test for differences between burned and unburned plots within the NESA core, as these groups were inherently different before the treatment was applied. The fact that the differences existed despite the fact that no manipulations had occurred in many of the unburned plots gives merit to the theory that disturbance history has played a role in the present day distribution of plants and habitat characteristics at the site.

Trends in vegetation and habitat characteristics in the NESA core
Discovering what makes a particular environment suitable for the rare plants that inhabit it has always been a challenge (Clarke and Patterson 2006, Schemske et al. 1994 (Brewer 1999) and slightly acidic, hydric environments (Juniper et al. 1989). Such environmental characteristics may be important for the establishment, growth, or persistence of Aletris and Platanthera.
Overall, plots containing 40-50% cover of herbaceous plants and less than 30-40% cover of woody plants were most likely to have at least one rare plant species growing in them. Both Aletris and Platanthera had positive relationships with cover of herbaceous plants and negative relationships with cover of woody plants. These trends underscore the potential influence of common species on the abundance and distribution of rare species.
It has been observed that light limitations and competitive interactions with surrounding species have been important in excluding many orchid species from sites in which they have occurred historically (Wotavova et al. 2004). At Ninigret, there was no difference in percent cover of woody vegetation in plots that were occupied by Platanthera than in plots that were not occupied by Platanthera. This suggests that the mere presence of woody vegetation did not preclude the presence of Platanthera. However , the mean height of the woody vegetation was significantly lower in plots that were occupied by Platanthera than in plots that were not occupied by Platanthera. This result supports the theory that disturbances which limit the height of surrounding vegetatio~ including certain historic management regimes , are vital to the persistence of the rare orchid species found within disturbed sites (Lunt 1997  The properties of the soil inside the NESA core are most likely one of the keys to explaining the unique nature of the NESA core. For example, the presence of Drosera is indicative of recently disturbed habitats with high light availability and low soil fertility (Brewer 1999, Givnish et al. 1984. The tight association between Drosera and bothAletris and Platanthera may be a function of both high light and low fertility. Both Platanthera and Scleria sp. were associated with nutrient poor environments in conservation plans for each species (Sharp 2004, Zaremba 2004). In addition, organic matter in the soil of the NESA core was 50% lower than in areas adjacent to the NESA core (Killingbeck et al. 1998 Greenberg et al. (2001) have proposed a "sit and wait" strategy for the mechanism by which Celastrus establishes and spreads. This strategy involves growing and persisting within an undisturbed, closed canopy until a disturbance opens up the canopy, allowing for rapid growth and proliferation (Greenberg et al. 2001).
The introduction of disturbance to the experimental sectors initiated a rapid increase in growth of Celastrus and clearly indicates that cutting shrubs in the absence of fire may contribute to future increases in this nonnative species. Studies suggest Celastrus will spread rapidly after a sufficient disturbance (Silveri et al. 2001) and can be highly correlated with scarified, exposed soils (McNab and Loftis 2002, Silveri et al. 2001). Celastrus has a rapid, prolific, and consistent annual rate of production of seeds that have both high viability and high germination rates (Ellsworth et al. 2004 Some shrub species are typically the first to grow back after disturbance due to their ability to re-sprout and spread by rhizomes (Glasgow and Matlack 2007).
Consequently, plant species with clonal growth forms are common in many disturbed habitats (Hartnett 1987). Although it is unknown whether the profusion of Rhus was primarily from seed germination or re-growth from rhizomes, the very low densities of Rhus in the control sector suggest that germination from seed may have been a major source of this species in the cut and burned sector.
Toe lack of Celastrus stems in the cut and burn sector is difficult to explain.
There seems to be no published literature focused on the effects of fire on Celastrus, but in this study, the differences in Celastrus density between the burned sectors and the cut-only sector suggest that fire may have a negative impact on this species. One possible mechanism for such an effect would be a reduction in seed viability caused by high temperatures.
While major differences in the cover of woody plants between burned and unburned plots did not occur in the NESA core, major differences were found in the The scarified plots introduced into the cut and burn sector had no immediate impact on rare plant growth or establishment However, it is possible that such growth may take longer than the two growing seasons of the study.

Management implications
Continued prescribed burning is recommended inside the NESA core.