Demography of the narrow endemic mint thinicola: Patterns, drivers, and management recommendations based on 18 years of data from its largest wild population1 Authors: Eric S. Menges, Suzanne M. Kennedy, Stacy A. Smith, and Stephanie M. Koontz Source: The Journal of the Torrey Botanical Society, 146(3) : 155-165 Published By: Torrey Botanical Society URL: https://doi.org/10.3159/TORREY-D-19-00003.1

BioOne Complete (complete.BioOne.org) is a full-text database of 200 subscribed and open-access titles in the biological, ecological, and environmental sciences published by nonprofit societies, associations, museums, institutions, and presses.

Your use of this PDF, the BioOne Complete website, and all posted and associated content indicates your acceptance of BioOne’s Terms of Use, available at www.bioone.org/terms-of-use.

Usage of BioOne Complete content is strictly limited to personal, educational, and non-commercial use. Commercial inquiries or rights and permissions requests should be directed to the individual publisher as copyright holder.

BioOne sees sustainable scholarly publishing as an inherently collaborative enterprise connecting authors, nonprofit publishers, academic institutions, research libraries, and research funders in the common goal of maximizing access to critical research.

Downloaded From: https://bioone.org/journals/The-Journal-of-the-Torrey-Botanical-Society on 01 Oct 2019 Terms of Use: https://bioone.org/terms-of-use Access provided by Torrey Botanical Society Journal of the Torrey Botanical Society 146(3): 155–165, 2019.

Demography of the narrow endemic mint Dicerandra thinicola: Patterns, drivers, and management recommendations based on 18 years of data from its largest wild population1 Eric S. Menges,2,5 Suzanne M. Kennedy,3 Stacy A. Smith,4 and Stephanie M. Koontz2 2 Archbold Biological Station, 123 Main Drive, Venus, FL 33960 3 Florivista, Inc., Cocoa Beach, FL 4 Department of Agronomy, University of Florida, Gainesville, FL

Abstract. The mint genus Dicerandra is the rarest of any in the southeastern USA, and the narrow endemic D. thinicola H.A. Mill. is restricted to one wild population on public land and a few unprotected populations on private lands. From 2001 to 2017, we studied ~9,000 plants and 90,000 annual transitions in permanent plots in Florida scrub and roadsides to assess the health of this population and potential drivers of demographic change. numbers have fluctuated widely, largely due to variably large pulses of winter seedling recruitment, but the overall trend has been upward, especially along sandy roadsides, in scrub gaps, and in chopped/burned scrub. Across the data set, annual survival (mean 67%) varied among (being highest along roadsides) and among years (with decreased survival in some recent years). Nearly half of surviving vegetative plants advanced from vegetative to reproductive each year and most plants, once flowering, continued to flower each year. Growth in number of branches was consistently positive; only one-quarter of plants had reduced size each year. Relative growth rate was higher along roadsides than in other habitats. Reproductive output was lowest in scrub gaps and varied among years. Seedling recruitment was concentrated in winter months, varied widely among years, and was lower in the scrub matrix than in scrub gaps or roadsides. About half of seedlings died before their second year, the maximum observed life span was 13 yr, and fewer than 6% of plants survived 10 yr. Flowering began as early as age 2 (rarely as seedlings) and by age 4–7, 90% of surviving plants were reproductive. Land management (roller chopping and fire) in 2007 had profound effects on the population. Chopping killed 91% of plants and chopping followed by burning killed 100%. However, recruitment in these treated plots was 2.5–5.5 times higher in the 6 yr after treatments than in the 8 yr before treatments; no such differences were seen in untreated plots. Posttreatment plants grew faster and flowered earlier than other plants. Consequently, after several years, plant numbers in the treated areas had increased 4–8 times, whereas plant numbers in untreated areas changed little. This state-endangered plant is short-lived and depends heavily on disturbance (ideally fire) for recruitment and population growth. It is threatened not only by its narrow distribution but by insufficiently aggressive land management. Key words: Fire, Florida scrub, plant demography, population dynamics, prescribed burn, roller chopping, seedling dynamics

The mint genus Dicerandra is the rarest of any are narrowly endemic, allopatric, perennial spe- in the southeastern USA (Estill and Cruzan 2001). cies. A few of these species have received Currently, 10 taxa are recognized as occurring in considerable study. On the Lake Wales Ridge, D. Florida (http://florida.plantatlas.usf.edu/) and most frutescens Shinners has received over 30 yr of intensive demographic study (Menges 2008). Fire strongly affects D. frutescens populations, with 1 We thank the City of Titusville Water Resources and Fire departments for their cooperation in allowing access most vital rates (survival, growth, fecundity, and keeping us informed of routine management. Thanks recruitment) and population growth being greatest to the US Fish and Wildlife Service, Brevard County, the in the first 2 decades after fire (Menges et al. 2006; Florida Forest Service, and The Nature Conservancy for Evans et al. 2008, 2010). is their interest in protecting and managing this species. Zack Prusack of The Nature Conservancy, Rudy Khan of protected at two sites. Two closely related Titusville Water Resources, Ron Weiss of the Florida congeners also endemic to the Lake Wales Ridge, Forest Service, and CalLee Davenport along with staff D. christmanii Huck & Judd and D. modesta from the Brevard County Environmentally Endangered Lands Program were instrumental in working with 5 multiple agencies to accomplish the land management Author for correspondence: emenges@archbold- that has occurred. We thank our funders, including the station.orggene. National Science Foundation (DEB 1347843) and the doi: 10.3159/TORREY-D-19-00003.1 Florida Division of Plant Industry. We thank Stephanie ÓCopyright 2019 by The Torrey Botanical Society Power and many Archbold research assistants and Received for publication December 26, 2018, and in interns for their diligent fieldwork. This manuscript revised form February 28, 2019; first published was improved by comments from Cheryl Peterson. September 6, 2019. 155

Downloaded From: https://bioone.org/journals/The-Journal-of-the-Torrey-Botanical-Society on 01 Oct 2019 Terms of Use: https://bioone.org/terms-of-use Access provided by Torrey Botanical Society 156 JOURNAL OF THE TORREY BOTANICAL SOCIETY [VOL. 146

(Huck) Huck, are protected at a single site. Other Adjacent to the Titusville Wellfield is the sole rare Dicerandra perennial species (D. cornutissima publicly owned conservation site, the Dicerandra Huck, D. immaculata Lakela, D. thinicola) are Scrub Sanctuary, managed by Brevard County’s endemic to other small ridges along the east coast Environmentally Endangered Lands Program. of Florida or in the Ocala area. During 2002, a survey of its namesake conserva- Levels of genetic variation differ among Dicer- tion-acquired land, Dicerandra Scrub Sanctuary, andra species (McDonald and Hamrick 1996, found two adult D. thinicola on the 44-acre Menges et al. 2001, Oliveira et al. 2007) and are sanctuary. An introduction of seeds from the unrelated to current and past landscape patterns or Wellfield to the Sanctuary was made in 2002–03 population sizes (Menges et al. 2010). Different (Menges and Kennedy 2007) and plants from genotypes of D. immaculata var. savannarum had subsequent generations persist where introduced. differential success in experimental introductions As for other perennial members of the genus (e.g., (Peterson et al. 2013). Dicerandra species are D. frutescens, D. christmanii; Menges 2007), D. either tetraploid or hexaploid (Huck and Chambers thinicola occurs in Florida scrub on yellow sands. 1997). Dicerandra thinicola sites are dominated by Dicerandra species share similar ecological and other (e.g., Q. myrtifolia, C. floridana, traits. They grow in Florida scrub on yellow sands Q. geminata, Q. chapmanii Sarg., Serenoa repens, (Menges et al. 2007; termed -hickory scrub or Ximenia americana L.) that resprout and grow yellow sand scrub; Menges 2007), often in areas rapidly after fires and other disturbances. dominated by myrtle oak (Quercus myrtifolia The ecology or management of wild populations Willd.), sand live oak (Quercus geminata Small), of D. thinicola have not yet been investigated. scrub hickory (Carya floridana Sarg.), and pal- Although we hypothesize that this species parallels mettos (Serenoa repens (W.Bartram) Small, Sabal other Dicerandra species in its dependence on fire etonia Swingle ex Nash). Florida scrub is fire and gaps for healthy populations, this hypothesis maintained, with oak-hickory scrub being charac- has not yet been tested. On the basis of 16 yr of terized by intense fires occurring every 5–12 yr demographic data, our goals are to describe the life (Menges 2007). Studied Dicerandra species are history and demography of this plant species, gap specialists (Menges 1992, Menges et al. 1999, assess the trends in the Titusville Wellfield III Peterson et al. 2013, Richardson et al. 2013). population, evaluate fire and weather as potential Dicerandra frutescens and D. christmanii are self- drivers of demography, and make recommenda- compatible, require insect visitation for seed set, tions for this endangered plant species’ conserva- and show evidence of inbreeding depression and tion and management. limitation (Evans et al. 2004). Diceran- dra immaculata var. immaculata is also self- Methods. STUDY SITE AND PLOT SETUP. Demo- compatible, with most seeds produced by insect- graphic data have been collected since 2001 in mediated outcrossing (Richardson et al. 2016). habitats across the Titusville Wellfield III (referred Dicerandra frutescens depends on a single bee-fly to as Titusville Wellfield throughout the rest of this for most of its (Deyrup and Menges document). Specifically, we established permanent 1997), although other Dicerandra species have monitoring plots in four types: oak- different visitors (Richardson et al. 2016). The two dominated Florida scrub matrix, oak-dominated subspecies of D. immaculata and D. christmanii Florida scrub gaps, sandy roadsides, and sand pine have received considerable conservation work ( (Chapm. ex Engelm.)Vasey ex including introductions and augmentations (Peter- Sarg)-dominated Florida scrub. Plots along road- son et al. 2013, Richardson et al. 2013, Menges sides were chosen randomly. Most roadside plots unpublished data). were 4 m wide and extended from the roadsides for Among the narrowly endemic Dicerandra 50 m into adjacent Florida scrub. Additional plots mints, the Titusville mint, D. thinicola, has one in nonroadside habitats were chosen subjectively of the narrowest ranges, known from only 10 sites to encompass patches of plants in a variety of (Kennedy 2003). The largest population (Titusville conditions. These plot sizes were variable depend- Wellfield III) comprises tens of thousands of ing on the size and shape of the D. thinicola patch. plants. The site is owned by the City of Titusville We established 16 plots in 2001 and added three and managed for water recharge and distribution. additional plots in 2008. Sampling occurred

Downloaded From: https://bioone.org/journals/The-Journal-of-the-Torrey-Botanical-Society on 01 Oct 2019 Terms of Use: https://bioone.org/terms-of-use Access provided by Torrey Botanical Society 2019] MENGES ET AL.: DEMOGRAPHY AND MANAGEMENT OF A RARE MINT 157

quarterly (November, February, May, August) tion. Habitats had previously been classified as beginning in November 2001, then was reduced roadside, scrub gap, scrub matrix, and sand pine to three times per year by dropping the August scrub. We did not use the last category in analyses sample in 2005, then reduced to twice annually because of low sample sizes in recent years. With (May, November) in 2012, and finally reduced to so many years in the data set, we considered year annually in November in 2014. Most analyses in as a continuous variable (although results were this paper deal with annual data, but we also focus similar analyzing year as a factor). We used binary on seedling recruitment during periods of more logistic regression to analyze annual survival and frequent sampling. We work with data sets transitions among life-history stages (seedling to comprising 9,039 individual plants and 90,135 flowering, vegetative to flowering, flowering to annual observations. vegetative). We used general linear models to Nomenclature follows the Florida plant atlas; analyze relative growth rate (RGR; for each annual http://florida.plantatlas.usf.edu/. transition, based on number of branch tips) and reproductive output. To compare habitats, we PLANT MEASUREMENTS AND SITE HISTORY. Within conducted Tukey’s honestly significant difference each plot, we marked each plant with a stake flag (HSD) pairwise comparisons after one-way AN- and a metal tag. New plants and seedlings were OVA. We also analyzed the effect of management marked with a small plastic toothpick (coded to treatments in 2007 on population size, survival, playing-card suit and color) and groups of seeding recruitment, RGR, and flowering. We seedlings were also marked with a metal tag and generally compared vital rates between the 6 yr stake flag. Plants were also mapped within plots to before 2007 and the 8 yr after 2007 in plots that the nearest centimeter using X and Y coordinates. were set up in 2001. At each sampling time, we recorded survival and We also considered weather as a potential driver recruitment. Recruits were classified as seedlings of D. thinicola demography. From the nearby (on the basis of the presence of cotyledons or small Titusville, FL weather station, we collected data on size) or larger plants missed in prior sampling. temperatures (low, high, mean, minimum warm, Each November (the time of peak flowering), we maximum cold) and precipitation (average mini- categorized that year’s life-history stage (seedling, mum, average maximum, total precipitation, vegetative, flowering adult, flowering seedling) maximum daily precipitation), summarized for and measured size (number of branch tips 2 cm years corresponding to our annual demographic long or greater) and flowering (number of sampling (November through October). Because flowering branch tips). Measurements were con- of collinearity (independent variables highly cor- sistent with those used in demographic studies for D. frutescens and D. christmanii by the authors, related) and a small number of years of demo- and for D. immaculata (Peterson et al. 2007). graphic data (13), we analyzed effects of total In 2007, a 10-ha block at the Titusville Wellfield precipitation and mean annual temperature, the received a combination of roller chopping (using a uncorrelated weather variables with the strongest drum with blades to break up ground vegetation) relationships to vital rates. Finally, we evaluated and prescribed fire, with the goals of reducing fuel effects of four 2004 hurricanes on marked D. loads and the dominance of woody vegetation. The thinicola plants. roller chopping occurred in January, whereas the Results. POPULATION TRENDS AND VITAL RATES. fire was delayed to August 7, 2007 because of lack The overall population trend through most of this of sufficient rain and suboptimal conditions for study was upward, with pronounced fluctuations burning. The fire was patchy, with substantial (Fig. 1). A noticeable peak in numbers occurred in unburned areas remaining. After both roller 2005, followed by 4 yr of population decline. chopping and fire, we visited each plot and made Since 2009, the trend has been generally upward observations on whether plots were affected. We until a large drop after firelane maintenance that also reconstructed plant locations in plots and disturbed some plots in March 2016. Until assessed survival and subsequent recruitment. recently, the increases have been most consistent STATISTICAL ANALYSES. All analyses were made in two habitats: sandy roadsides and gaps in using SPSS version 22. We analyzed how vital Florida scrub (Fig. 2). Populations in the scrub rates varied with habitat, year, and their interac- matrix have been relatively stable, whereas

Downloaded From: https://bioone.org/journals/The-Journal-of-the-Torrey-Botanical-Society on 01 Oct 2019 Terms of Use: https://bioone.org/terms-of-use Access provided by Torrey Botanical Society 158 JOURNAL OF THE TORREY BOTANICAL SOCIETY [VOL. 146

FIG. 1. Number of plants in each November sample (2001–17) for 16 plots set up in 2001. Some plots were damaged by fire-lane maintenance in March 2016.

populations in Florida scrub dominated by sand seedling mortality). Although roadsides often had pine have showed consistent declines. Rapid much higher survival than scrub (e.g., 2005–06, declines after 2015 in the firelane populations 2006–07), this pattern was reversed in some years was a result of firelane maintenance, although (2007–08, 2009–10, 2015–16), in some cases by other habitats also showed decreases (Fig. 2). road and firelane maintenance inside monitoring Overall annual survival has been moderate plots. (67.3%), with substantial variation among years Surviving plants tended to advance among life- and habitats. In logistic regression, individual history stages. Most seedlings became vegetative annual survival (n ¼ 8,211) varied significantly plants the following year (89%), but some by year (Wald ¼ 35.705, d.f. ¼ 1, P , 0.001), flowered in their second year (11%). Vegetative habitat (Wald ¼ 41.630, d.f. ¼ 2, P , 0.001), and plants were slightly more likely to remain their interaction (Wald ¼ 41.518, d.f. ¼ 2, P , vegetative (54.3%) than advance to a flowering 0.001). Survival was significantly higher for plants stage (45.7%). Once plants were flowering, they growing along roadsides (unweighted mean almost always flowered in the successive year 70.9%) than those in either scrub gaps (66.1%) (93.8%). Flowering seedlings comprised 0.017% or the scrub matrix (66.1%; Tukey’s HSD, P ¼ of the database. Of the flowering seedlings, most 0.010). Survival also was low for sand pine scrub occurred shortly after fire. Seedling-to-flowering (60.7%), which was not used in analyses because transitions were unrelated to habitat, year, or their of low sample sizes in later years. Annual survival interaction (logistic regression, P . 0.12). Like- was lowest in 2011–12 (50%) and 2015–16 (57%) wise, vegetative-to-flowering transitions were un- and highest in 2012–13 (83%); other years varied related to habitat, year, or their interaction from 65% to 78%. Annual survival showed an variables (P . 0.5). Flowering plants to vegetative overall downward trend, with variation related to reversions were affected by year (Wald ¼ 12.104, years with many seedlings (and subsequently high d.f. ¼ 1, P ¼ 0.001) but not by habitat or the

Downloaded From: https://bioone.org/journals/The-Journal-of-the-Torrey-Botanical-Society on 01 Oct 2019 Terms of Use: https://bioone.org/terms-of-use Access provided by Torrey Botanical Society 2019] MENGES ET AL.: DEMOGRAPHY AND MANAGEMENT OF A RARE MINT 159

FIG. 2. Number of plants in each November sample (2001–17) for plots set up in 2001, divided by four habitat types.

interaction of year and habitat. These reversions was significantly lower in scrub gaps (mean 6.62 were much higher in 2015–16 (12.8%) and 2016– flowering branches) than in the scrub matrix 17 (18.9%) than in other years. (10.48) or roadsides (7.78). Overall, reproductive Most plants grew in number of branches from output declined during the last 4 yr of the study. year to year. Only 25% of plants regressed and Reproductive output in roadsides was higher in the 10% had the same number of branch tips, whereas earlier years of the study, whereas reproductive 65% of plants grew larger. The mean RGR was output in the scrub matrix was higher in later years 0.43, with a similar median (0.41). RGR varied by of the study, accounting for the significant habitat (F ¼ 14.194, d.f. ¼ 2, P , 0.001), by the interaction. Reproductive output was closely interaction of habitat and year (F ¼ 14.292, d.f. ¼ correlated with plant size (r2 ¼ 0.724; both 2, P , 0.001), and less so by year (F ¼ 7.266, d.f. variables natural log transformed). ¼ 1, P ¼ 0.002). RGR was significantly higher Seedling recruitment has been concentrated along roadsides (0.55) than in scrub gaps (0.38) or during winter months, as evidenced by consistent scrub matrix (0.33; Tukey’s HSD, P , 0.05). peaksinseedlingnumbersfoundduringthe Growth was highest in 2010–11 (0.83), 2014–15 February sampling (Fig. 3). Recruitment also (0.66), and 2013–2014 (0.64), 4–8 yr after the varied substantially among years, with notable 2007 chopping and fire. Individual plant growth peaks in 2005 (264), 2011 (364), and 2013 (421). was, on average, positive in every year of the In contrast, 2009 (2) had little recruitment. study. Seedling recruitment across all years was highest Reproductive output, measured as the natural in scrub gaps (1,476 in 81 m2; 1.07 seedlings m2 log of the number of flowering branches, varied by yr1), and roadsides (842 in 64 m2; 0.78 seedlings year (F ¼ 21.712, d.f. ¼ 1, P , 0.001), habitat (F ¼ m2 yr1) and lower in the scrub matrix (362 in 957 5.768, d.f. ¼ 2, P ¼ 0.003), and their interaction (F m2; 0.02 seedlings m2 yr1) and sand pine scrub ¼ 5.759, d.f. ¼ 1, P ¼ 0.003). Reproductive output (126 in 46 m2; 0.16 seedlings m2 yr1).

Downloaded From: https://bioone.org/journals/The-Journal-of-the-Torrey-Botanical-Society on 01 Oct 2019 Terms of Use: https://bioone.org/terms-of-use Access provided by Torrey Botanical Society 160 JOURNAL OF THE TORREY BOTANICAL SOCIETY [VOL. 146

FIG. 3. Number of seedlings found in original 2001 plots, 2001–11, by census number and month of census. February censuses are filled bars, May censuses are white bars, and November censuses are gray bars. August sampling did not occur after 2004 and is not shown, but recruitment was minimal in August sampling dates.

LIFE SPAN AND AGE AT FLOWERING. Because we Most D. thinicola plants were short-lived, with could identify seedlings in the field, especially nearly all plants living for a decade or less (Fig. 4). when sampling quarterly, we could assess life span For all seedlings in our data set, 26% still were and age of first flowering. On the basis of large alive in 2017, with a median life span of 2 yr. For seedling cohorts from 2002 to 2005 (n ¼ 122, 106, seedlings born 2005 or earlier, about 3% were still alive. For this group, 21.0% had survived 5 yr or 201, and 264 respectively), we found that the longer and 5.4% survived 10 yr or longer. Among median life span (measured from the plant’s first plants in the group that died, the median life span November) was 1 yr, with the maximum life span was 1 yr, 17.8% survived 5 yr or longer, and only over 13 yr. All plants from the 2002 cohort died 2.2% survived 10 yr or longer. before 2015; the oldest plant lived 11 yr, dying in Flowering occurred as early as age 2 yr and 2013. From the 2003 cohort, all plants died, with increased with age. By age 4–7 yr, depending on the oldest plant living 13 yr, dying by 2017. From the cohort, over 90% of surviving plants had the 2004 and 2005 cohorts, some plants in each flowered (Fig. 5). Among these four cohorts, only cohort (2.5%, and 1.1% respectively) were still 1–7% of plants survived for 10 yr.

living in 2017, having survived 12–13 yr. Most EFFECTS OF ROLLER CHOPPING,FIRE, AND WEATH- plants died before becoming reproductive and over ER. Management treatments (chopping and pre- 75% of plants sampled from seedlings onward did scribed burn) affected many of our plots in 2007. not live beyond 2 yr. Three plots were chopped, two plots were chopped

Downloaded From: https://bioone.org/journals/The-Journal-of-the-Torrey-Botanical-Society on 01 Oct 2019 Terms of Use: https://bioone.org/terms-of-use Access provided by Torrey Botanical Society 2019] MENGES ET AL.: DEMOGRAPHY AND MANAGEMENT OF A RARE MINT 161

FIG. 4. Percentage of known seedlings from 2005 (n ¼ 267) surviving to subsequent years, and flowering in those years. Sample sizes after 2015 are too low to estimate percent flowering reasonably.

and then burned, and one plot had a mixture of the several years through about 2013, but since 2013, two treatments. Survival in plots that were the effects of the treatments on population numbers chopped but not burned was only 9.4%, as seem to have abated. compared with 75.8% survival for plants not Plant growth was higher in treated areas after affected by either chopping or burning. No plants the disturbances. Relative growth rates before survived the combination of chopping and burn- 2007 were variable and the areas that would later ing. be disturbed did not support particularly high or After the treatments, new seedlings began low growth. However, later RGRs were notably recruiting into chop and burn plots. In chopped higher for both disturbed types relative to the plots, seedling recruitment increased 5.5-fold from control. In a general linear model, the interaction a mean of 5.3 seedlings yr1 in the 6 yr before of disturbance type and whether years were within chopping to 29.3 seedlings yr1 in the 8 yr after 6 yr before or within 8 yr after 2007 was highly chopping. Numbers peaked in 2013 (6 yr after significant in predicting RGR (F ¼ 30.101, d.f. ¼ 2, disturbance). In the plots that were chopped and P , 0.001), whereas neither factor alone was a burned in 2007, the increase was 2.5-fold, from significant predictor (P . 0.263). 10.8 to 26.0 seedlings yr1, peaking in 2010 (3 yr Flowering occurred earlier in seedlings recruit- postdisturbance). This substantial increase con- ing post-disturbance. The percentage of plants of trasts with only a 1.1-fold increase in seedling two-seedling cohorts with reasonable sample sizes recruitment in plots not disturbed over the same (n ¼ 159 for 2010, n ¼ 70 for 2011) flowering in time period. their second year was generally higher for The overall effect of the treatments was to seedlings recruiting in areas either chopped increase population size (Fig. 6). The response in (28.6%, 7.7%) or chopped and burned (41.7%, the burned areas was quicker than in areas that 50.0%) relative to unburned areas (17.6%, 13.2%). were only chopped. The response occurred over Likewise, over 70% of seedlings from cohorts that

Downloaded From: https://bioone.org/journals/The-Journal-of-the-Torrey-Botanical-Society on 01 Oct 2019 Terms of Use: https://bioone.org/terms-of-use Access provided by Torrey Botanical Society 162 JOURNAL OF THE TORREY BOTANICAL SOCIETY [VOL. 146

FIG. 5. Percentage of seedlings surviving for different number of years. The final pair of bars (labeled as SA) refers to plants that are still alive (various ages for all cohorts, at least 10 yr old for 2001–05 cohorts).

were in disturbed areas flowered by their third Discussion. LIFE HISTORY OF DICERANDRA THINI- year, in comparison with less than 60% of COLA. Our long-term study (2001–17; about 9,000 seedlings in undisturbed areas. plants and 90,000 annual transitions) provides the Four hurricanes (Charley, Frances, Ivan, and data from which to characterize the life history of Jeanne) affected the Titusville Wellfield in 2004 D. thinicola as a short-lived perennial with (August 13, September 5, September 16, and consistent individual growth, early reproduction, September 26). These storms had major effects on and frequent seedling recruitment during the the site, especially in tipping or uprooting many winter months. Experiments show that this species sand pines. However, only 6 of 481 living D. forms a persistent seed bank of at least 2 yr thinicola plants were affected, mainly by being (Menges and Kennedy 2007, unpublished data). buried in live debris from sand pines. Of these This species has modest annual survival (mean plants, three survived. 67%). Annual survival is comparable with its No weather variables affected annual survival congeners D. frutescens (mean 58%, Menges, (r2 ¼ 0.070 with total precipitation and mean unpublished data) and D. christmanii (76%, temperature). Seedling recruitment quantified Menges unpublished data). Growth from year to through November each year increased marginally year is consistent, with about half of vegetative with total precipitation (r ¼ 0.500, d.f. ¼ 13, P ¼ plants advancing to flowering each year. Once 0.08) and decreased with mean temperatures (r ¼ flowering, surviving plants continue to flower. 0.592, df ¼ 13, P ¼ 0.03). However, when Growth in branch numbers is also consistently combined in a single analysis, neither weather positive; only one quarter of plants show reduced variable was significant (P . 0.1, r2 ¼ 0.36). size from one year to the next.

Downloaded From: https://bioone.org/journals/The-Journal-of-the-Torrey-Botanical-Society on 01 Oct 2019 Terms of Use: https://bioone.org/terms-of-use Access provided by Torrey Botanical Society 2019] MENGES ET AL.: DEMOGRAPHY AND MANAGEMENT OF A RARE MINT 163

FIG. 6. Percent change in number of plants relative to 2006, for plants subjected to chopping, chopping followed by burning, or neither (no treatment) in 2007.

Dicerandra thinicola is not long-lived. About EFFECTS OF HABITAT MANAGEMENT (DISTURBANC- half of seedlings died before their second year, the ES) ON DEMOGRAPHY. Land management (roller maximum observed life span was 13 yr, and fewer chopping, fire) in 2007 had profound (and mainly than 6% of plants survived 10 yr. Flowering began positive) effects on the D. thinicola population. as early as age 2 (rarely as seedlings) and by age Chopping killed 91% of plants and chopping 4–7, 90% of surviving plants were reproductive. followed by burning killed 100%. However, these The short life span and rapid growth to reproduc- treatments had positive effects on D. thinicola tive status is typical of other Dicerandra species demography for many subsequent years. Seedling (Menges, unpublished data, Peterson, unpublished recruitment in these treated plots was 2.5–5.5-fold data). higher in the 6 years after treatments than the 8 yr before treatments; no such differences were seen in EFFECTS OF HABITAT ON DEMOGRAPHY. Roadsides untreated plots. Posttreatment plants also grew and scrub gaps were favorable microhabitats for D. faster and flowered earlier than other plants. thinicola, in comparison with closed conditions in Responses to chopping and burning were quicker the matrix. Annual survival varied among than responses to roller chopping only (with peak habitats (being highest along roadsides) and numbers in 3 vs. 6 yr after disturbances). The among years (with decreased survival in some slower response to chopping may be due to slow recent years). Relative growth rate also was higher litter decomposition in areas not burned, which along roadsides than in other habitats. Reproduc- would inhibit seedling recruitment. tive output was lowest in scrub gaps and varied The benefits of fire on population dynamics of among years. Seedling recruitment was concen- D. thinicola is shared by other Dicerandra species. trated in winter months, varied widely among Higher vital rates and population growth rates in years, and was lower in the scrub matrix than in the years after fire have been demonstrated for D. either roadsides or scrub gaps. frutescens (Menges et al. 2006; Evans et al. 2008,

Downloaded From: https://bioone.org/journals/The-Journal-of-the-Torrey-Botanical-Society on 01 Oct 2019 Terms of Use: https://bioone.org/terms-of-use Access provided by Torrey Botanical Society 164 JOURNAL OF THE TORREY BOTANICAL SOCIETY [VOL. 146

2010). In particular, time-since-fire reduced ger- CONSERVATION IMPLICATIONS AND RECOMMENDA- mination, reproductive status, and growth of most TIONS. Dicerandra is the rarest of any plant genus life-history stages (Evans et al. 2010). For in the southeastern USA (Estill and Cruzan 2001), introductions and augmentations of D. christmanii, and the narrow endemic D. thinicola is restricted to burned gaps tended to have higher survival and one wild population on public land (Kennedy growth than unburned gaps (Menges and Koontz, 2003). Our research shows that this state-endan- unpublished data). Fire in these Florida scrub gered plant is short-lived and depends heavily on habitats may create and expand gaps (as in disturbance (ideally fire) for recruitment and rosemary scrub; Menges et al. 2008, 2017), which population growth. It is threatened not only by would benefit gap-loving Dicerandra species its narrow distribution and limited protection on (Menges 1992, Menges et al. 1999, Peterson et extant wild sites, but by insufficiently aggressive al. 2013, Richardson et al. 2013). Given the land management. Consistent fire management relatively strong demographic performance of D. will be needed on Florida scrub sites supporting thinicola in scrub gaps, it seems that fire’s role on D. thinicola, especially for its largest population at gap dynamics may explain why this population did the Titusville Wellfield. so well after fire. Although fire management is crucial to the The effects of other disturbances (such as roller conservation of D. thinicola, additional measures chopping and firelane maintenance) on Dicerandra are also needed. These include additional intro- species are less well documented, although their ductions to suitable sites, seed banking, federal consistently higher densities in these areas (espe- listing, and continued collaboration among agen- cially in the context of fire-suppressed scrub) cies and organizations. suggest that mechanical treatments could be helpful in maintaining larger populations. Howev- Literature Cited er, the relatively delayed response of D. thinicola DEYRUP,M.A.AND E. S. MENGES. 1997. Pollination to roller chopping without fire, compared with the ecology of the rare scrub mint Dicerandra frutescens quicker response to the combination of these (). Fla. Sci. 60: 143–157. treatments, suggests the importance of adding fire ESTILL,J.C.AND M. B. CRUZAN. 2001. Phytogeography of to mechanical treatments. This combination can rare plant species endemic to the southeastern United States. Castanea 66: 3–23. also be beneficial to management of Florida scrub EVANS, M. E. K., K. HOLSINGER, AND E. S. MENGES. 2008. and other ecosystems in Florida (Menges and Modeling the effect of fire on Dicerandra frutescens Gordon 2010). spp. frutescens (Lamiaceae), an endangered plant In contrast to fire, hurricanes had little effect on endemic to Florida scrub. Popul. Ecol. 50: 53–62. D. thinicola demography, despite causing consid- EVANS, M. E. K., K. HOLSINGER, AND E. S. MENGES. 2010. erable damage to sand pines at the Titusville Fire, vital rates, and population viability: A hierarchi- cal Bayesian analysis of the endangered Florida scrub Wellfield. The lack of effects on D. thinicola is mint. Ecol. Monogr. 80: 627–649. consistent with past research on Florida scrub EVANS, M. E. K., E. S. MENGES, AND D. R. GORDON. 2004. plants, as most shrubs and herbs in the Florida Mating systems and limits to seed production in two scrub were little affected by prior hurricanes Dicerandra mints endemic to Florida scrub. Biodivers. (Menges et al. 2011). Conserv. 13: 1819–1832. HUCK,R.B.AND H. L. CHAMBERS. 1997. Polyploidy: A Weather variables had weak effects on D. factor in the evolution of Dicerandra Benth. (Labi- thinicola demography in our analyses, although atae). Edinburgh J. Bot. 54: 217–229. there was some indication that recruitment might KENNEDY, S. 2003. Dicerandra thinicola Miller, Titusville increase with precipitation and decrease with mint, in Brevard County, Florida. Final Report to U.S. temperature. These results are consistent with data Fish & Wildlife Service, Jacksonville, FL. MCDONALD,D.B.AND J. L. HAMRICK. 1996. Genetic from D. immaculata, whose plants grew slower in variation in some plants of Florida scrub. Am. J. Bot. hot weather, and produced more flowers and 83: 21–27. recruited more seedlings in wetter years (Peterson MENGES, E. S. 1992. Habitat preferences and response to and Zaya, unpublished data). Seedling recruitment disturbance for Dicerandra frutescens, a Lake Wales increases with precipitation are roughly consistent Ridge (Florida) endemic plant. Bull. Torrey Bot. Club 119: 308–313. with some other winter-recruiting Florida scrub MENGES, E. S. 2007. Integrating demography and fire and sand-hill species (e.g., Polygala lewtonii management: An example from Florida scrub. Aust. J. Small; Weekley and Menges 2012). Bot. 55: 261–272.

Downloaded From: https://bioone.org/journals/The-Journal-of-the-Torrey-Botanical-Society on 01 Oct 2019 Terms of Use: https://bioone.org/terms-of-use Access provided by Torrey Botanical Society 2019] MENGES ET AL.: DEMOGRAPHY AND MANAGEMENT OF A RARE MINT 165

MENGES, E. S. 2008. Demography of the endemic mint demography in fire-controlled ecosystems. Biotropica Dicerandra frutescens in Florida scrub. Ecology 89: 43: 450–458. 1474 (data paper; Ecol. Arch. E089–088). MENGES, E. S., C. W. WEEKLEY,S.I.HAMZE´ , AND R. L. MENGES,E.S.AND D. R. GORDON. 2010. Should PICKERT. 2007. Soil preferences for listed plants on the mechanical treatments and herbicides be used as fire Lake Wales Ridge in Highlands County, Florida. Fla. surrogates to manage Florida’s uplands? A review. Fla. Sci. 70: 24–39. Sci. 73: 145–172. OLIVEIRA, L. O., R. B. HUCK,M.A.GITZENDANNER,W.S. MENGES,E.S.AND S. KENNEDY. 2007. A successful JUDD,D.E.SOLTIS, AND P. S. SOLTIS. 2007. Molecular introduction of the rare endemic Dicerandra thinicola phylogeny, biogeography, and systematics of Dicer- to a protected site (Florida). Ecol. Restor. 25: 222–223. andra (Lamiaceae), a genus endemic to the southeast- MENGES, E. S., S. J. H. CRATE, AND P. F. Q UINTANA- ern United States. Am. J. Bot. 94: 1017–1027. ASCENCIO. 2017. Dynamics of gaps, vegetation, and plant species with and without fire. Am. J. Bot. 104: PETERSON, C. L., C. CAMPBELL, AND B. ROBERTSON. 2007. 1825–1836. An Integrated Conservation Program for the Protection MENGES, E. S., R. W. DOLAN,R.PICKERT,R.YAHR, AND D. of Florida’s Rare and Endangered Flora. Final Report R. GORDON. 2010. Genetic variation in past and current to FDOCS for #011298. landscapes: Conservation implications based on six PETERSON, C. L., G. S. KAUFMANN,C.VANDELLO, AND M. L. endemic Florida scrub plants. Int. J. Ecol. 2010: 1–12. RICHARDSON. 2013. Parent genotype and environmental doi:10.1155/2010/503759. factors influence introduction success of the critically MENGES, E. S., R. W. DOLAN,R.YAHR, AND D. R. GORDON. endangered savannas mint ( 2001. Comparative genetics of seven plants endemic to var. savannarum). PLoSOne:https://doi.org/10.1371/ Florida’s Lake Wales Ridge. Castanea 66: 98–114. journal.pone.0061429. MENGES, E. S., P. J. MCINTYRE,M.S.FINER,E.GOSS, AND RICHARDSON, M. L., C. P. KEATHLEY, AND C. L. PETERSON. R. YAHR. 1999. Microhabitat of the narrow Florida 2016. Breeding system of the critically endangered scrub endemic , with compar- Lakela’s mint and influence of plant height on isons to its congener D. frutescens. J. Torrey Bot. Soc. and seed output. Popul. Ecol. 58: 277–284. 126: 24–31. RICHARDSON, M. L., M. L. J. WATSON, AND C. L. PETERSON. MENGES, E. S., P. F. QUINTANA-ASCENCIO,C.W.WEEKLEY, 2013. Influence of community structure on the spatial AND O. G. GAOUE. 2006. Population viability analysis and fire return intervals for an endemic Florida scrub distribution of critically endangered Dicerandra im- mint. Biol. Conserv. 127: 115–127. maculata var. immaculata (Lamiaceae) at wild, MENGES, E. S., A. WALLY,J.SALO,R.ZINTHEFER, AND C. W. introduced, and extirpated locations in Florida scrub. WEEKLEY. 2008. Gap ecology in Florida scrub: Species Plant Ecol. 214: 443–453. occurrence, diversity, and gap properties. J. Veg. Sci. WEEKLEY,C.W.AND E. S. MENGES. 2012. Burning creates 19: 503–514. contrasting demographic patterns in an endangered MENGES, E. S., C. W. WEEKLEY,G.L.CLARKE, AND S. A. perennial herb: A cradle-to-grave analysis of multiple SMITH. 2011. Effects of hurricanes on rare plant cohorts. Aust. J. Bot. 60: 347–357.

Downloaded From: https://bioone.org/journals/The-Journal-of-the-Torrey-Botanical-Society on 01 Oct 2019 Terms of Use: https://bioone.org/terms-of-use Access provided by Torrey Botanical Society