Species Status Assessment Version 1.0

Florida Golden Aster (Chrysopsis floridana) Species Status Assessment Version 1.0

Photo by Alafia River State Park

July 2018

U.S. Fish and Wildlife Service

Region 4

Atlanta, GA

Table of Contents

ACKNOWLEDGEMENTS iv

EXECUTIVE SUMMARY iv

1 INTRODUCTION 1

1.1 Species Federal Status 3

2 SPECIES BIOLOGY 3

2.1 Species Description and Taxonomy 3

2.2 Life History and Demography 4

2.3 Habitat 6

2.4 Abundance and Distribution 8 2.4.1 Historical 8 2.4.2 Current 8

2.5 Genetics 10

3 SPECIES NEEDS FOR VIABILITY 11

3.1 Individual Level 11

3.2 Population Level 12

3.3 Species Level 12

4 INFLUENCES ON VIABILITY 13

4.1 Habitat Availability 14

4.2 Habitat Management 14

4.3 Introductions 15

4.4 Climate Change 16

5 CURRENT CONDITION 18

5.1 Delineating Populations 18

5.2 Current Resilience 18 5.2.1 Population Size 19 5.2.2 Habitat Protection 21 5.2.3 Habitat Area Available 22 5.2.3.1 Measuring Available Habitat 23 5.2.4 Classifying Resilience 27

5.3 Current Redundancy and Representation 32

6 FUTURE CONDITION 34

6.1 Future Considerations 34 6.1.1 Habitat Quantity 35 6.1.2 Habitat Quality 35 6.1.2.1 Development Risk Assessments 38

6.2 Future Scenarios 44 6.2.1 Status Quo 44 6.2.2 Pessimistic 45 6.2.3 Targeted Conservation 45 6.2.4 Likelihood of Scenarios 49

6.3 Future Resilience 49

6.4 Future Redundancy and Representation 53

LITERATURE CITED 54

APPENDIX 57

iii

ACKNOWLEDGEMENTS

This document was prepared by Stephanie DeMay (Texas A&M Natural Resources Institute), Todd Mecklenborg (U.S. Fish and Wildlife Service [Service]), Byron Hamilton (Service), and Michael Marshall (Service). External species expertise, guidance, and document reviews were provided by technical team members Eric Menges (Archbold Biological Station), Michael Jenkins (Florida Forest Service), Ann Johnson (Florida Natural Areas Inventory), and Cheryl Peterson (Bok Tower Gardens). Peer review was provided by Sheryl Bowman (Hillsborough County) and Jennifer Possley (Fairchild Tropical Botanic Garden).

EXECUTIVE SUMMARY

The Florida golden aster (Chrysopsis floridana) is a short-lived perennial endemic to scrub habitat and ecotones between scrub and flatwoods east and southeast of the Tampa Bay area of central Florida. The historical distribution of this plant is not well understood because of the speed at which its habitat was converted to residential, commercial, and agricultural uses after human settlement. It is currently known from 30 populations within 5 counties in Florida, and has been listed as federally endangered since 1986 (51 FR 17974). Of these, 25 populations occur entirely or mostly on lands protected for conservation, and 9 of these populations were introduced (3 extant populations introduced in the 1980s, 6 extant populations introduced since 2008). Over half of the known number of individuals occur in these introduced populations, illustrating the importance of captive propagation and introductions to the recovery of the species. Open habitat is important to the species, which was maintained historically by periodic natural burns and other natural processes (e.g., animal trails, trees blown over in storms). Presently, open habitat must be maintained by prescribed burning or mechanical treatment. Many details about C. floridana’s habitat requirements (at every scale, landscape to microhabitat) and demography are not yet known, although there is a range-wide study in progress (partnership between Archbold Biological Station, Bok Tower Gardens, and the U.S. Fish and Wildlife Service) to gain a better understanding of these aspects of C. floridana biology. For the following Species Status Assessment (SSA), we made many assumptions about C. floridana needs and responses to stressors based on currently available knowledge and input from species experts, but further study is needed to test whether these assumptions hold. We are clear and explicit in the SSA about where these assumptions were made and why.

We assessed current resilience of populations based on 3 factors: population size, habitat protection, and habitat area available. Resilience was tied primarily to population size, as large populations are better able to withstand demographic, environmental, and anthropogenic stochastic events. These resilience classes were as follows: < 100 plants = low resilience, 100- 500 plants = moderate resilience, 501-1000 plants = high resilience, and > 1000 plants = very high resilience. Habitat protection was a proxy for habitat management, under the assumption that populations on protected conservation lands are more likely to receive adequate habitat management (openness maintained) than those on private lands. Finally, we used a basic habitat model constructed from the current state of knowledge about the species biology and needs to calculate the amount of available habitat for each population to occupy, spread into, or shift into as current habitat becomes less suitable. Populations that were not on protected lands and/or had

a small amount of habitat available were assigned to a resilience class lower than their population size would otherwise warrant. Based on this resilience classification strategy, there are currently 7 very highly resilient populations, 11 highly resilient populations, 6 moderately resilient populations, and 6 populations with low resilience. The vast majority of highly and very highly resilient populations occur in the central portion of the species range in Hillsborough and Manatee Counties (13 populations), with 3 such populations farther west in Pinellas County, and 2 such populations farther east in Hardee and Highlands Counties. Besides these 3 geographic clusters that are isolated from each other, there was no evidence of any representative units based on genetic or ecological differences, although future research could reveal otherwise.

We assessed the future condition of C. floridana 20 years into the future under 3 scenarios: Status Quo, Pessimistic, and Targeted Conservation. These scenarios explored differences in habitat quantity (loss of habitat on private lands, acquisition of and/or introductions in unoccupied habitat) and habitat quality (interpreted as the ability to burn or maintain open habitat, influenced by the will and resources of managing entities and conditions that limit the ability and flexibility to burn such as urban development and roads). Under the Status Quo scenario, no new populations were established, and all were managed for stable populations (same resilience class as current condition) except for populations with a high risk of present and future development that will limit the ability to manage habitat; these fell one resilience rank. In the Pessimistic scenario, management effort on all populations decreased, presumably as an effect of wide-scale changes in management priorities and resources, resulting in a drop in resilience scores across the board, and all populations on non-protected lands were lost. Under the Targeted Conservation scenario, populations with high and very high resilience were managed to maintain their rank; in cases where populations had a high risk of development limiting the ability to manage, this involved an increase in management effort compared to what would be needed to maintain the same level of resilience for a population with a low risk of development impacts. Populations with currently moderate resilience on protected lands received management effort increases to either move them into the high resilience class (low risk from development) or maintain moderate resilience (high risk from development). Conservation resources were steered towards maintaining and growing these larger populations, and not as much towards rescuing populations that currently have low resilience. Additionally, 5 new sites were selected across the species range in which to introduce new populations, thus improving species redundancy. Of these 3 scenarios, the Status Quo scenario is the most likely to occur, although the Targeted Conservation scenario represents a likely future if both habitat-focused management (prescribed burning and mechanical or manual habitat management) by a variety of partners/managing entities and species-specific conservation (captive propagation and introductions) are prioritized and well-funded.

It is important to note that as we applied habitat management impacts to populations and projected them into the future, populations only increased or decreased by one level of resilience over 20 years. Populations can grow or decline very rapidly, but we do not fully understand what drives these dynamics, and do not have reliable estimates of population growth rates to more quantitatively project population sizes into the future. Thus, we are conservative in the magnitude of our projections, but more confident in the direction of the change. Under all 3 scenarios, 4-7 populations currently with low resilience become extirpated. Under the Status Quo scenario, the number of highly and very highly resilient populations drops from 18 currently to

12 in the future. Under the Pessimistic scenario, that number of highly and very highly resilient populations is 7. Under the Targeted Conservation scenario, the number of highly and very highly resilient populations increases to 27, with improved redundancy across the species range. In the Status Quo and Pessimistic scenarios, all populations are lost from Pinellas County, an area geographically separated from the two other general clusters of C. floridana populations.

1 INTRODUCTION

The Species Status Assessment (SSA) framework (USFWS 2016) is intended to support an in- depth review of the species’ biology and threats, an evaluation of its biological status, and an assessment of the resources and conditions needed to maintain long-term viability. The intent is for the SSA to be easily updated as new information becomes available and to support all functions of the Endangered Species Program from Candidate Assessment to Listing to Consultations to Recovery.

The Florida golden aster (Chrysopsis floridana) is endemic to xeric uplands east and southeast of the Tampa Bay area of central Florida. Chrysopsis floridana has been listed as endangered under the Endangered Species Act of 1973 (Act), since 1986 (51 FR 17974). This SSA for C. floridana is intended to provide the biological support for the decision on whether to reclassify the species and for potential future Act actions. Importantly, the SSA does not result in a decision by the U.S. Fish and Wildlife Service (Service) on whether this species should be proposed for reclassification under the Act. Instead, this SSA provides a review of the available information strictly related to the biological status of C. floridana. The reclassification decision will be made by the Service after reviewing this document and all relevant laws, regulations, and policies, and the results of a proposed decision will be announced in the Federal Register, with appropriate opportunities for public input. For the purpose of this assessment, we generally define viability as the ability of C. floridana to sustain populations in its range over time. Using the SSA framework (Figure 1), we consider what the species needs to maintain viability by characterizing the status of the species in terms of its resilience, redundancy, and representation (Wolf et al. 2015).

• Resilience describes the ability of populations to withstand stochastic events (arising from random factors). We can measure resilience based on metrics of population health; for example, birth versus death rates and population size. Highly resilient populations are better able to withstand disturbances such as random fluctuations in birth rates (demographic stochasticity), variations in rainfall (environmental stochasticity), or the effects of anthropogenic activities.

• Redundancy describes the ability of a species to withstand catastrophic events. Measured by the number of populations, their resilience, and their distribution (and connectivity), redundancy gauges the probability that the species has a margin of safety to withstand or return from catastrophic events (such as a rare destructive natural event or episode involving many populations).

• Representation describes the ability of a species to adapt to changing environmental conditions. Representation can be measured by the breadth of genetic or environmental diversity within and among populations and gauges the probability that a species is capable of adapting to environmental changes. The more representation, or diversity, a species has, the more it is capable of adapting to changes (natural or human caused) in its environment. In the absence of species-specific genetic and ecological diversity information, we evaluate representation based on the extent and variability of habitat characteristics across the geographical range.

Figure 1. Species Status Assessment Framework

To evaluate the biological status of C. floridana we compiled known information from the literature and species experts about the species biology and needs, and assessed the species’ resilience, redundancy, and representation (together, the 3 Rs) under current conditions and multiple plausible future scenarios. The format for this SSA includes: (1) species biology and needs, (2) influences on viability, (3) current conditions, and (4) future conditions. This document is a compilation of the best available scientific and commercial information and a description of past, present, and likely future risk factors to C. floridana.

1.1 Species Federal Status Chrysopsis floridana was listed as endangered in 1986 (51 FR 17974) under the Act. The two most recent 5-year reviews (2009 & 2017) both indicated that the species status was improving, assigned a Recovery Priority Number of 8 indicating moderate degree of threat and high recovery potential, and recommended downlisting to threatened. The recovery plan (USFWS 1988) provides that the species can be considered for downlisting when 10 self-sustaining populations exist in Hardee, Hillsborough, Manatee, and Pinellas counties, and can be considered for delisting when 20 or more self-sustaining populations occur in one or more of those same counties.

2 SPECIES BIOLOGY

2.1 Species Description and Taxonomy Chrysopsis floridana is a perennial herb with stems that are woody toward the base and non- woody above. The plants have basal rosettes (clusters of leaves at ground level) with leaves 4-10 cm long, 1.5-2.0 cm wide, and densely short-woolly pubescent. The stems grow upright from the rosettes or the bases of old shoots and are usually 0.3-0.4 m, up to 0.7 m tall. The stem leaves are nearly the same size from the top to the bottom of the stem. They are obovate-elliptic, slightly clasping the stem, entire, and densely short-woolly pubescent. The flower heads are grouped into a more or less flat-topped cluster of 1-25 heads at the top of the stem. Each head is slightly over 2.5 cm in diameter. Both the central disc and the rays are golden yellow. The species is distinguished from other members of the genus by the woodiness of its stems, the type of pubescence and shape of the stem leaves, and the way the flower heads are arranged in a flat-

topped cluster (Semple 1981, Wunderlin et al. 1981). Chrysopsis floridana can be distinguished from similar species C. highlandsensis by the densely short-pubescent hairs on mid-stem leaves, while the hairs on mid-stem leaves of C. highlandsensis are loosely lanate (Wunderlin and Hansen 2011).

The currently accepted taxonomic ranking for C. floridana is described below:

Kingdom: Plantae Subkingdom: Viridiplantae Infrakingdom: Streptophyta Superdivision: Embryophyta Division: Tracheophyta Subdivision: Spermatophytina Class: Magnoliopsida Superorder: Asteranae Order: Asterales Family: Asteraceae Genus: Chrysopsis (Nutt.) Elliot Species: floridana Small Common name: Florida goldenaster, Florida golden aster

*Retrieved 1/9/2018 from the Integrated Taxonomic Information System on-line database, http://www.itis.gov

2.2 Life History and Demography Chrysopsis floridana is a short-lived perennial herb that flowers in late November and December and sheds seeds from December through January. Chrysopsis floridana can spread vegetatively by forming new basal rosettes at the ends of rhizomes, but reproduction is primarily by seed. The entire genus has an out-crossing breeding system (self-incompatibility), preventing self- fertilization (Wunderlin et al. 1981). Specific pollinator and dispersal information is not available for C. floridana, but insects are the primary pollinators of members of the Asteraceae family (Mani and Saravanan 1999), and the seed morphology of many species within Asteraceae, including C. floridana (plumose structures on seeds that act as parachutes), is well adapted for wind dispersal (Andersen 1993, Lambert and Menges 1996), and seeds can also adhere to animal fur (or researcher clothing). Researchers have observed that wind-dispersed seeds often get caught in adjacent vegetation; seed dispersal is thus not typically very far from parent plants, and population footprints (specifically of introduced populations) are not quick to expand (C.

Peterson, personal comm.). Chrysopsis floridana has four main life stages: seed, basal rosette, basal rosette with vegetative stalks, and reproductive.

Investigations on C. floridana seed by Bok Tower Gardens (BTG) in captivity for propagation and introduction programs revealed that about 50% of seeds produced were good (full) seeds, with the other half being empty seeds or not fully formed (Campbell 2008). Germination rates of full seeds were approximately 50% in warm and cool growth chambers, but only 21% in greenhouse trials. In the growth chambers, the germination period was from day 6 to day 35, with germination starting 3 days earlier in the warm chamber compared to the cool chamber. The germination period for the greenhouse was day 13 to day 81. Germination peaked in both growth chamber treatments and greenhouse conditions during the 4th week. Seedling mortality was 9% of seedlings (warm chamber), 2% (cool chamber), and 19% (greenhouse). Germination rates did vary by site from which the seeds were harvested. Mortality of potted seedlings moved to the greenhouse approached 40% for seedlings germinated in the warm chamber of greenhouse, while mortality of seedlings germinated in the cool chamber was nearly half of that. After this, mortality showed a sharp increase in March, which then decreased as plants grew. Plant growth seemed to stop as temperatures rose and days lengthened, and did not resume in the greenhouse again until late October.

Germination in the wild is drastically lower than that for seeds planted in captivity. Of 48,000 seeds dispersed in a field experiment by Lambert and Menges (1996), only 232 (0.5%) emerged over the course of the two-year experiment, with 86.6% of those germinating during the second spring after they were distributed. Some seeds were still viable and sprouted in the third year, after the experiment concluded. When organic litter was absent and soil was disturbed, a positive effect on germination rates was evident that was greater than the additive effects of the two factors alone. Lambert and Menges (1996) found that while fire did not influence seedling survival rates, it did appear to stimulate reproduction; all flowering plants in their field experiment occurred in burned plots, and 10 out of 12 of the third-year seedling recruits were found in burned plots.

In June and August 2008, 410 seedlings grown at BTG were planted at two sites at Cordell East, a site in Manatee County owned by the Florida Fish and Wildlife Conservation Commission (Campbell 2008). Monitoring during the November following the introductions revealed the following population stage structure: 4% mortality, 33% basal rosette stage, 4% basal rosette with vegetative stalks, 59% with reproductive stalks. There were differences in relative percentages between the two sites, likely resulting from differences in timing of the plantings and habitat characteristics. Plants with reproductive stalks averaged 3.5 reproductive stems per plant, with values ranging from 1 to 17. There were differences in morphology and vital rates between plants from different source populations, with plants from Goldenaster Scrub and Bullfrog Creek lineages having the highest fitness.

Under a partnership between BTG, Archbold Biological Station (ABS), and the Service, additional research is currently in progress to examine vital rates (survival, growth, fecundity, and recruitment) and population stage structure of natural and introduced populations, with data collection expected to occur through November 2019.

2.3 Habitat Chrysopsis floridana occurs in open areas of Florida scrub and the adjacent sandhill communities and ecotone between scrub and flatwoods. The soils are classified as excessively- drained and well-drained that include Archbold fine sands, St. Lucie fine sands, Lakewood fine sands, Duette fine sands, and Pomello fine sands (Wunderlin et al. 1981). All of these soils are extremely nutrient-poor and composed primarily of siliceous sand that retains little water.

Vegetation frequently found growing in association with C. floridana include, in the overstory and shrub strata, Pinus clausa (sand pine), Quercus chapmanii (Chapman’s oak), Q. geminata (sand live oak), Q. inopina (scrub oak), Q. myrtifolia (myrtle oak), Q. pumila (running oak), Serenoa repens (saw palmetto), Ceratiola ericoides (Florida rosemary), Ximenia americana (hog plum), Lyonia fruticosa (coastalplain staggerbush), Bejaria racemosa (tarflower), and Asimina reticulata (netted pawpaw). An herbaceous component that could be considered an indicator of appropriate habitat conditions would consist of Pityopsis graminifolia (narrowleaf silkgrass), Balduina angustifolia (coastalplain honeycombhead), Carphephorus corymbosus (coastalplain

chaffhead), Polygonella ciliata (hairy jointweed), Liatris tenuifolia (shortleaf gayfeather), Stillingia sylvatica (queen’s delight), Licania michauxii (gopher apple), Aristida stricta (wiregrass), and Cladonia leporina (cup lichen) (Lambert and Menges 1996, Wunderlin et al. 1981).

The “openness” of the habitat is key for species persistence. To maintain openness, periodic disturbance is necessary, and can include fires (natural or prescribed), storm blowouts, tree falls, animal trails, gopher tortoise burrows, sand roads, plow lines, etc. In most habitat types, fire is the most important disturbance to maintain open conditions in all strata within C. floridana preferred habitat. Fire appears to benefit C. floridana (e.g., increased germination and flowering after a prescribed burn; Lambert and Menges 1996), and may do so by reducing plant competition, releasing nutrients, and/or increasing reproduction (Gates and Tanner 1988, Menges et al. 2017). Fire suppression leads to the closing of the habitat. Lambert and Menges (1996) noted increasing mortality and variable recruitment within fire-suppressed sites as canopy closure increased. Lower light conditions could induce additional stress on the plants by affecting their vegetative growth and reproduction. Additionally, the seeds of C. floridana have a pappus (tuft of hairs) of capillary bristles that aids with its dispersal away from the parent plant by enabling it to be dragged along the ground or lifted up into the air and carried by wind. Wind dispersal is most effective in open areas and only minimally effective in closed canopied systems (Burrows 1986).

Like many scrub endemics, C. floridana is associated with soil disturbance, which is important for regeneration. Previous C. floridana studies have demonstrated that seedling emergence was favored by disturbed soils, the absence of a litter layer and by their combination (Lambert and Menges 1996). These disturbances might benefit C. floridana reproduction by breaking seed dormancy, burying seeds at appropriate depths, triggering a release of nitrates within the soils, and reducing litter to expose bare soil (Lambert and Menges 1996, St. John 1987). Litter layers might inhibit germination not only by serving as a physical barrier that lightweight seeds cannot penetrate, but may also inhibit germination chemically (Karssen and Hilhorst 1992, Loydi et al. 2015, Richardson 1985). Because C. floridana favors disturbed open soil, several current and historical populations occur/occurred along roads, railroads, and other right of ways.

2.4 Abundance and Distribution

2.4.1 Historical The historical range of C. floridana is thought to span parts of Hillsborough, Manatee, Pinellas, Highlands, and Hardee County, but the true range is not known for certain because the ecosystems it occurs on were rapidly converted to residential, commercial, and agricultural uses after settlement of the region. The species was first collected and described from a specimen in Manatee County in early 1901, with subsequent collections in Pinellas and Hillsborough counties (1920s). No collections in Pinellas County since the 1920s have occurred. The last remaining natural population known to occur in Pinellas County was reported in 1983; however a housing development eliminated all of the available habitat by 1985. The initial Manatee County and Pinellas County populations occurred in coastal areas of Bradenton Beach and St. Petersburg Beach. These populations have also been extirpated.

When the species was listed as endangered in 1986, all nine known extant populations of the species occurred in five locations in southeastern Hillsborough County (Wunderlin et al. 1981). Since listing of the species, increased survey efforts have resulted in the discovery of additional populations, including occurrences further inland from the initial coastal populations. Many of the new locations discovered have now been acquired as conservation lands with active management activities implemented to improve the habitat conditions. Also benefiting the species, population introductions have occurred on conservation lands in Hardee, Hillsborough, Manatee, and Pinellas counties. Introductions here refer to the outplanting of plants grown from seed in captivity to suitable habitat within the historical range of the species. It is not known whether these introduction sites were historically occupied by C. floridana, or if so, how long ago they supported natural populations.

2.4.2 Current As of the most current surveys across the species range (2006 – 2018 depending on the population), there are 30 known extant populations, natural and introduced, occurring in five counties (Hardee, Highlands, Hillsborough, Manatee, and Pinellas; Figure 2). Populations for

this count were delineated using a 2 km separation distance between occurrences (more detail in Current Conditions section).

Figure 2. The five Florida counties where C. floridana occurs as of 2017 are highlighted in gray, with Hillsborough County shaded darker gray. At the time of listing in 1986, populations of C. floridana were only known to occur in Hillsborough Country.

Of these, 25 populations occur entirely or mostly on 22 protected sites, with “protected” referring to a site that has been fee simple acquired and placed into long-term conservation by a non- governmental, local, state, or federal entity, or a conservation easement or other binding land agreement by the site owner that shows a commitment to its conservation in perpetuity. In addition, each site should have a management agreement or plan developed and implemented. None of the lands occupied by C. floridana are federally owned or managed. The remaining 5 extant populations occur on private lands or along roadways or railroad lines. Abundance estimates from the most recent surveys (occurring between 2006 and 2018) give a total of approximately 50,000 individuals (these numbers can only give an approximation of the true current abundance of C. floridana because surveys were not conducted every year, were conducted differently by various biologists for different purposes, and population sizes fluctuate annually), with over 90% occurring on protected lands. Twelve of the 30 populations had greater

9 than 1,000 individual plants present when last observed. During the most recent surveys, just over half of C. floridana individuals occurred in nine introduced populations at eight sites. The earliest introductions were undertaken in 1986; three of those populations are still extant, while seven other introductions in Pinellas and Hillsborough counties failed. Introductions were again initiated during 2008-2013. Only introduced populations occur in Pinellas County; no natural populations are known to occur there. There is a 56 km gap between the easternmost naturally occurring population in Manatee County and the nearest natural population in Hardee County, and it is not presently known whether this gap is due to the lack of suitable habitat, lack of observation, a long-distance dispersal event, or fragmentation of a formerly continuous distribution.

2.5 Genetics A 1998 study used random amplification for polymorphism detection (RAPD) to compare the genetic variation of eight natural populations in Hillsborough County with a seed stock collection population at BTG (Markham 1998). The sampled populations exhibited relatively high genetic variation (gene diversity 0.242; Nei 1987) for an endemic species, and the seed stock collection at BTG had similar levels of genetic variation compared to natural populations. Natural populations did not exhibit genetic structuring; the majority of the variation was attributed to within-population differences with only 20% attributable to genetic variation among populations. In a follow-up study using another genetic analysis method on the same samples, inter-simple sequence repeat analysis (ISSR), it was again found that there was little genetic differentiation among populations, with >80% of the variation found within populations (Cochrane 2002).

There are two primary genetic threats to small populations, inbreeding and genetic drift. Inbreeding occurs when related individuals reproduce together, and can lead to the accumulation of harmful alleles that can cause decreases in fitness (Hedrick 2014). Genetic drift is the random change in allele frequencies in a population as genes are passed along each generation. Harmful alleles are more likely to become fixed in small populations compared to large populations. Both inbreeding and genetic drift can be accelerated when individuals are brought into captivity and propagated, and resulting offspring run the risk of having lower fitness than wild offspring as a

result of the reduced genetic diversity in captivity. Translocated individuals can also have lower fitness in a new environment if they lack local adaptations to that environment. The results from studies on C. floridana imply that, at the time of these studies, the populations were fairly homogenous, with gene flow likely occurring to maintain genetic similarities to counteract genetic drift. With little genetic differences identified between different populations, or between wild and captive seed stock populations, there was no reason for concern about the loss of genetic information caused by either propagating a captive population from a limited number of founders, or by introducing those individuals into new or established populations in the wild. Ongoing research under a partnership between BTG, ABS, and the Service seeks to compare demography and vital rates of wild and introduced populations, which may reveal whether individuals have different levels of fitness depending on their population origin, which could be based on genetic, habitat, management, or other factors.

It is also important to note that the above studies did not examine genetic variation known to be associated with any adaptive traits, and only used samples from Hillsborough County. Differences in vital rates and morphological measurements are apparent between plants grown at BTG and introduced to other sites from different source populations (Campbell 2008). Because plants from the different source populations were grown in common conditions, these differences might have a genetic basis.

3 SPECIES NEEDS FOR VIABILITY

3.1 Individual Level At the individual level, C. floridana plants require suitable habitat to flourish during each of their life stages, a dispersing and germinating seed, rosette, vegetative plant, and reproductive plant.

Habitat characteristics and requirements are discussed in detail in the next section as they relate to the population scale, but individual plants have similar needs, briefly: disturbance-maintained open habitat patches in Florida scrub and sandhill, preferably free of organic litter, with suitable soils, nutrients, and microclimates. Under a partnership between BTG, ABS, and the Service, research is currently in progress to further characterize microhabitat needs of C. floridana plants,

particular in terms of cover (e.g., tree, shrub, forb, grass, litter, ground lichen, bare sand) of C. floridana in five natural and five introduced populations.

In addition to suitable habitat, individual plants depend on pollinators (likely bees and other insects) to successfully reproduce. Individual plants are susceptible to mortality from lack of sufficient resources (due to poor habitat, competition, etc.), herbivory, and natural or anthropogenic disturbance (e.g., extreme rain or drought, vehicle traffic).

3.2 Population Level For resilient populations to persist, open habitat must be maintained. Lambert and Menges (1996) recommend prescribed burning that mimics the historical burn pattern (frequent low intense fires in sandhill, less frequent burns in scrub, with fires primarily in late spring and summer) and periodic mechanical disturbance of the ground cover during late winter or early spring when seeds are dispersed. In the absence of fire, habitat openness can be maintained with mowing, hand removal of trees and shrubs near plants, or other mechanical treatments; populations have persisted along periodically mowed right of ways (e.g., underneath powerlines, along roads and railroads) for decades without a prescribed burn program.

Populations must be suitably large and connected to provide a reservoir of individuals to cross- pollinate with, as plants will not self-fertilize, and to maintain levels of genetic diversity high enough to prevent harmful consequences from inbreeding depression and genetic drift (Ellstrand and Elam 1993).

3.3 Species Level For the species to be viable, there must be adequate redundancy (suitable number, distribution, and connectivity to allow the species to withstand catastrophic events) and representation (genetic and environmental diversity to allow the species to adapt to changing environmental conditions). Redundancy improves with increasing numbers of populations, and connectivity (either natural or human-facilitated) allows connected populations to “rescue” each other after catastrophes. Representation improves with increased genetic diversity and/or environmental conditions within and among populations.

4 INFLUENCES ON VIABILITY

Viability of C. floridana has been and will continue to be impacted both negatively and positively by a number of anthropogenic and natural influences (Figure 3). Historically, the primary threats to C. floridana were habitat loss as a result of human development and habitat degradation due to lack of adequate habitat management. As threats to habitat have been alleviated via habitat protection and management, recovery has been further bolstered by captive propagation followed by introduction into unoccupied sites. Finally, climate change will almost certainly influence C. floridana into the future by influencing habitat suitability and the ability to manage habitat with prescribed fire. Other influences not discussed in detail here, either because they are not thought to be a major threat or there is little information available, include invasive plant species like Cogongrass (Imperata cylindrica), and future genetic consequences of small and/or translocated populations.

Figure 3. Influence diagram illustrating relationships between key habitat and population factors, influences on those factors, and species viability. 4.1 Habitat Availability When the species was listed in 1986, C. floridana was only known to occur on private property in nine populations in Hillsborough County. The two largest populations occurred in residential subdivisions on vacant lots. Other populations were in scrub vegetation grazed by cattle, on an abandoned railroad embankment, and in recently burned sand pine scrub. Many of these sites were eventually lost to development. It is unknown how much historical habitat was eliminated that existed prior to the widespread development of the Tampa Bay region. When the recovery plan was written in 1988, additional populations had been discovered, but only two populations occurred on public lands: one at a roadside park and one at a county boat landing, neither of which was being actively managed to benefit the species (USFWS 1988).

Populations occurring on private lands were, and remain, subject to adverse human activity including mowing, dumping, off-road recreational vehicles use, and land clearing. As a result, one of the primary recovery actions for C. floridana has been the acquisition and protection of conservation lands. Currently, about 83% of C. floridana populations and 92% of individual plants occur on protected lands, and at least one protected population occurs in each of the five counties where C. floridana is found. Acquiring lands for conservation has benefitted natural populations (16 populations), and provided unoccupied sites for 9 extant introduced populations. Currently protected lands occupied by C. floridana are owned by the Florida Department of Environmental Protection, Hillsborough County, Florida Fish and Wildlife Conservation Commission, Manatee County, Southwest Florida Water Management District, City of St. Petersburg, and Pinellas County. As mentioned previously, there are no populations on federal lands.

4.2 Habitat Management Ideal habitat management is generally regarded as prescribed burning that mimics the historical burn pattern (frequent low intensity fires in sandhill, less frequent burns in scrub, with fires primarily in late spring and summer) and periodic mechanical disturbance of the ground cover during late winter or early spring when seeds are dispersed (Lambert and Menges 1996). Initial burning to restore the openness of degraded habitat involves frequent intense fires, after which

burning can be less intense and frequent to simply maintain the habitat. Failing to maintain open scrub habitat can disrupt C. floridana reproduction, survival, and dispersal (Lambert and Menges 1996).

Historical and current lack of management, especially the absence of periodic fire and/or mechanical treatment, has led to habitat degradation throughout the C. floridana range. As with habitat destruction and modification, this threat is mainly a concern on private and non- conservation lands. Populations that occur on public conservation lands are often being managed to maintain optimal open scrub habitat. However, budget constraints, manageability, conflicting priorities, and other factors may preclude proper management activities even on conservation lands. Additionally, proximity to urbanized areas can limit the number of days available for prescribed burns, and urbanization in the Tampa Bay area is increasing rapidly (Xian et al. 2005). Optimal burn days have wind speeds and wind directions that do not unduly burden urbanized areas with smoke. For this reason, large rural tracts of habitat are easier to burn than small tracts tucked into developed areas. Increasing development could lead to further decreases in prescribed burning in the future, which may or may not be replaced with adequate habitat management by other means, which are more expensive than using fire. The type of development also factors into management ability and flexibility, with major roads, schools, hospitals, retirement homes (places with vulnerable populations) weighing more heavily on the decision of if/when to burn than other types of development (B. Camposano, personal communication).

4.3 Introductions Introductions into unoccupied but suitable habitat have been a major contributor to C. floridana recovery. As above, an introduction for our purposes is defined as the outplanting of plants grown from seed in captivity to suitable habitat within the historical range of the species. It is not known whether these sites were historically occupied by C. floridana, or if so, how long ago they supported natural populations. Of 10 introductions initiated in the 1980s, 7 failed, and 3 populations are still extant in Hillsborough and Pinellas counties, supporting a total of 1126 plants as of the most recent censuses. From 2008 to 2013, BTG has introduced 6 additional populations in Hardee, Manatee, and Pinellas counties, which contain 24,825 plants (as of the most recent censuses, with about half of this number in one population) where prior to 2008

there were none. All were planted as rosettes with vegetative stalks, as they were planted during the rainy season before the reproductive season (C. Peterson, personal communication). Follow- up monitoring has demonstrated that introductions can be effective and successful. For example, the Weedon Island population increased from 922 introduced plants in 2010 to 5024 plants in 2015; the McKay Creek population increased from 613 introduced plants in 2009 to 3900 in 2015; the Duette Preserve populations increased from 321 introduced plants in 2013 to 1416 in 2015; and the Paynes Creek Historic State Park population increased from 199 introduced plants in 2011 to 1732 in 2015. Currently, over half of known C. floridana plants are found in introduced, rather than natural, populations.

4.4 Climate Change In the future, changing climatic conditions will likely impact C. floridana. The Intergovernmental Panel on Climate Change (IPCC) concluded that warming of the climate system is unequivocal (Pachauri et al. 2014). Species that are dependent on specialized habitat types, limited in distribution (e.g., C. floridana), or at the extreme periphery of their range may be most susceptible to the impacts of climate change (Byers and Norris 2011; Anacker et al. 2013). There is evidence that some terrestrial plant populations have been able to adapt and respond to changing climatic conditions (Franks et al. 2014). Both plastic (phenotypic change such as leaf size or phenology) and evolutionary (shift in allelic frequencies) responses to changes in climate have been detected. Given enough time, plants can alter their ranges, resulting in range shifts, reductions, or increases (Kelly and Goulden 2008, Loarie et al. 2008).

The climate in the southeastern United States has warmed about 2 degrees F from a cool period in the 1960s and 1970s, and is expected to continue to rise (Carter et al. 2014). Projections for future precipitation trends in the Southeast are less certain than those for temperature, but suggest that overall annual precipitation will decrease, and that tropical storms will occur less frequently, but with more force (more category 4 and 5 hurricanes) than historical averages (Carter et al. 2014). Sea levels are expected to rise globally, potentially exceeding 1 m of sea level rise by 2100 (Figure 4; Reynolds et al. 2012). Local sea level rise impacts depend not only on how much the ocean level itself is increasing, but also on land subsidence and/or changes in offshore currents (Carter et al. 2014), and impacts on terrestrial ecosystems can occur via

submergence of habitat during storm surges or permanently, salt water intrusion into the water table, and erosion.

Figure 4. From Reynolds et al. 2012; global mean sea level and future projections (with uncertainty) based on three climate models from the Intergovernmental Panel on Climate Change (IPCC 2007). The purple projection represents a scenario without scaled-up ice sheet discharge, blue represents a scenario with scaled-up ice sheet discharge, and orange is based on a different semi-empirical projection approach (Rahmstorf 2007).

Of the current populations of C. floridana, only one (Fort De Soto County Park, Pinellas County) is directly vulnerable to inundation from 1 foot of sea level rise, a reasonable estimate of sea level rise by 2050 (Figure 4). Hotter and drier conditions in the future could lead to fewer days with optimal conditions for prescribed burning, which could lead to reduced habitat quality if land managers are unable to make up for the lack of burning with adequate mechanical treatment. It is possible that there will be increases in the number of lightning strikes and sizes and severities of resulting wildfires, which could have a positive or negative effect on specific C. floridana populations. Hurricanes similarly could have positive or negative effects on the species. Prolonged flooding could harm populations, but the mechanical disturbance of trees being uprooted could improve habitat for colonizing species like C. floridana (E. Menges and A. Johnson, personal communication). We have no additional information or data regarding effects of climate change with respect to C. floridana populations into the future; further research will

17 be helpful to determine how this species responds directly to changes in temperature and water availability.

5 CURRENT CONDITION

5.1 Delineating Populations We delineated populations using a 2-km separation distance rule based on species expert opinion, resulting in 30 populations across 5 counties. This strategy differs from the 1-km separation distance rule that was used in the most recent 5-year review, which was based on NatureServe’s default criteria for defining plant populations (NatureServe 2004). The team of species experts providing input on this SSA suspected that 1 km is likely an underestimate of the distances that gene flow can regularly occur via pollination. While the exact insect pollinators of C. floridana are not known, studies on multiple bee species (major plant and Chrysopsis pollinators) demonstrate foraging distances that regularly exceed 1 km (Greenleaf et al. 2007, Hagler et al. 2011).

5.2 Current Resilience Resilience was assessed for each population using 3 factors: population size, habitat protection, and area of available habitat. Other factors were considered that likely contribute to population resilience, but data were not available to assess them over all or most of the populations. Examples of these factors not included in the final resilience classification strategy are: explicit measures of habitat quality (e.g., vegetation composition or structure), prescribed fire regime, time since fire, non-fire habitat management, and whether there is a land management plan in place that is being followed. We considered using population trend as a factor to assess resilience; population growth indicates conditions (that we cannot explicitly assess) that favor population persistence, while conditions that negatively influence populations (e.g., lack of or improper habitat management) would be expected to generate population declines. While some past survey data are available for many populations, species experts did not feel comfortable comparing population counts across time periods. In many cases, differences in population sizes were likely not a result of increasing populations, but rather of differences in survey methodology, number of surveyors, and/or areas searched (e.g., surveyors more likely to visit

known patches and not find new patches; alternately, a bias toward larger counts over time as old patches are revisited and additional patches are found). Past population counts, when available, are included in Table A1 in the Appendix, but were not used to assess resilience.

In the following discussion of the factors that were used to determine resilience, as well as the factors that were not ultimately used, it must be emphasized that this species has not been extensively studied, and quantitative data on habitat needs and population dynamics are only now being collected. There is great uncertainty in precisely how these factors influence C. floridana population resilience.

5.2.1 Population Size Population size is both a direct contributor to resilience and an indirect indicator of resilience. Small populations are more susceptible to demographic and environmental stochastic events than larger populations. Small populations are also more likely to suffer from Allee effects or decreased fitness as a result of low genetic diversity from inbreeding or genetic drift (Willi et al. 2005). Large populations are more buffered from the effects of prescribed burning or other disturbances, which are necessary to maintain open habitat, but can temporarily reduce population sizes by killing plants. Indirectly, large population sizes are likely indicative of other conditions that contribute to population resilience. For example, we do not have adequate data to assess habitat quality and the quality of management at all C. floridana populations for this SSA, but large population sizes likely generally reflect good habitat quality and management (among other factors) compared to smaller populations, though this assumption may not hold in all cases.

We categorized populations into 4 size classes: < 100 individuals, 100 – 500 individuals, 501 – 1000 individuals, and > 1000 individuals. Each population size class was associated with a baseline resilience class, respectively low, moderate, high, and very high.

The population size threshold between high and very high resilience of 1,000 individuals was chosen because it is the typical population size used to rank element occurrences as having “excellent viability” and likely to persist for the next 20 – 30 years (NatureServe 2008). This is a generic population size limit that is not specifically tailored to C. floridana with empirical data.

Further support for using 1,000 individuals as the threshold for the highest resilience category comes from a study of 10-year extirpation rates for populations of varying sizes of 8 short-lived plant species in Germany (Matthies et al. 2004). In this study, for 7 of 8 species, the probability of population persistence increased with population size, and all populations of more than 1,000 individuals (flowering plants) persisted for the duration of the 10-year study (Figure 5).

Figure 5. Figure from Matthies et al. (2004), showing the proportion of populations of 8 plant species in Germany that persisted for 10 years from 1986 to 1996, as a function of population size (# flowering plants) in 1986. Data points shown are for populations of size: > 6 plants, 6 – 25 plants, 26 – 50 plants, 51 – 100 plants, 101 – 1000 plants, and > 1,000 plants. Numbers above the x-axis represent sample sizes (number of populations) for each point.

The other threshold population sizes for this SSA were chosen with guidance from other plant SSAs (e.g., whorled sunflower and dwarf-flowered heartleaf) and EO ranking criteria. Populations with fewer than 1,000 but greater than 500 individuals were still considered to have high resilience, as there is evidence from historical survey data of populations with fewer than 1,000 individuals persisting for decades. There are presently no empirical estimates of minimum viable population sizes for C. floridana. Demographic and stage structure data are being collected at over half of the delineated C. floridana populations; when available, this information should be used in future updates of this SSA to refine the population size categories and their

20 implications for population resilience. There are examples of small populations persisting for decades (e.g., Highlands Hammock State Park).

We obtained “current” population size data for all populations, with data collected as recently as 2018 for some populations, and no older than 2006 for any population (Appendix, Table A1). Population sizes have undoubtedly changed since the last surveys for those populations that have not been surveyed as recently, as populations fluctuate in response to management actions, time since management, environmental events, stochastic demographic processes, etc. Thus, the reported numbers reflect best available estimates for population sizes, rather than precise counts meant to represent actual current population sizes. For the purposes of this SSA, population sizes included all plants counted, whether flowering or not. Survey data for some populations provided separate counts for each life stage, but for many populations, survey data were simply numbers with no information about whether that number was only flowering plants, or all plants. In using total plant numbers, and assuming that ambiguous counts were minimum counts of total plants in each population, we were conservative in our population counts. The alternative of assuming that ambiguous counts were of only flowering adult plants, when they may have included basal rosettes, would inflate population sizes in cases where the assumption was wrong.

An important caveat to consider for this resilience factor is that the current population size shows only a snapshot in time. In the case of this SSA, the snapshot shown for each population falls between 2006 and 2018. The recently introduced populations have been monitored annually, but the vast majority of naturally occurring populations have not. Chrysopsis floridana populations can fluctuate over time, particularly in response to disturbance. A disturbance (e.g., fire, grazing) that will ultimately improve habitat and induce population growth might reduce populations in the short term. Snapshots of population size taken at different times before or after disturbance will look very different from each other, and for this SSA, we do not have population-by- population data about time since disturbance or other clarifying information.

5.2.2 Habitat Protection The second factor that fed into resilience was habitat protection. Land was considered “protected” if it was fee simple acquired and placed into long-term conservation by a non-

governmental, local, state, or federal entity or a binding land agreement. Protected sites should have management plans developed and being implemented. Information was not available to assess the quality of habitat management for C. floridana at each protected population, as site- specific levels of management vary over time with changes in personnel, funding, management priorities, issues at urban interfaces, and other influences. Given the uncertainty about levels of habitat management at occupied sites, and uncertainty about what exactly constitutes optimal habitat for C. floridana, we considered habitat protection (protected vs. not protected) to be a proxy for habitat management. As more is learned about the needs of this species, collecting data about habitat management actions and outcomes can refine future assessments of habitat and management quality.

5.2.3 Habitat Area Available Chrysopsis floridana population sizes fluctuate up and down, and can occur in high densities in small patches of habitat. For a given population size, 1,500 plants for example, a population covering a large area will be more resilient than a population covering a small area. A perturbation of the same size will have a proportionally larger effect on small-area populations than large-area populations. In assessing population resilience, we considered the amount of habitat available rather than the amount of habitat occupied for two reasons. First, the amount of area occupied is very uncertain for most populations. Surveys are likely to return to known patches of C. floridana, but new patches can be easily missed and it is likely that the data we have underestimates the true amount of area occupied by C. floridana. Adding to the uncertainty, the most current spatial data for some populations comes from 2006, 12 years ago, and may no longer reflect the current distribution at those sites. Second, population footprints are not always static across available habitat; C. floridana can spread into unoccupied areas as populations grow, and/or shift across a landscape as different areas become more or less suitable (i.e., habitat becomes more open as habitat is disturbed or managed to maintain openness, and subsequently becomes less open as time since disturbance increases). As an example, at Moody Branch Preserve, a repeat survey by FNAI in 2017 revealed new C. floridana sites within 300m of 2006 locations, possibly as a result of prescribed burning to maintain habitat that has occurred since 2006 (and also potentially as a result of an increase in survey effort; A. Johnson, personal comm.). For this reason, we used the amount of habitat available for populations to occupy

currently, grow into, and/or shift into as a factor contributing to population resilience. Using GIS as described in the following section, we identified available habitat within a 2 km radius around known occurrences, consistent with the assumption we made about pollinator movements when delineating populations. We characterized the available habitat for populations as small or large, with 35 acres as the threshold between the two groups. This value was selected based on natural breaks in the data and expert input. Although experts believe that the area of habitat is an important piece contributing to resilience, there is no empirical data showing what an appropriate cut-off value might be, and this value should be refined as more information is gathered.

5.2.3.1 Measuring Available Habitat Using spatial locations of C. floridana occurrences, GIS, and expert input, we modeled C. floridana habitat within the 5 Florida counties in which it occurs. Spatial occurrence data came from 2 sources: sampling locations from the ongoing demographic study being conducted by ABS, BTG, and the Service, and Element Occurrence (EO) data maintained by the Florida Natural Areas Inventory (FNAI). For the following exercise, we used only FNAI EO polygons that were classified as extant and had ‘High’ or ‘Very High’ spatial accuracy in order to minimize the amount of non-habitat included in the assessment. Demographic study points represented 2x1m study plots, and were converted to 2-m radius circular polygons for the following analysis. Both natural and introduced populations were used for this analysis. Although introduced populations do not necessarily represent natural habitat conditions for this species, some historical and all recent introductions have been successful (although the Cordell West site, which makes up a portion of the larger Cordell population, has experienced a population decline from 577 to 270 plants from 2010 to 2015, the Cordell east population has flourished to over 12,000 plants), and over half of current C. floridana individuals occur in introduced populations, illustrating the suitability and importance of the habitat they occupy. We incorporated land cover type, soil type, normalized difference vegetation index (NDVI), elevation, and slope into this model as described below.

• Land Cover Type We used Cooperative Land Cover, Version 3.2 (2016) data developed by the Florida Fish and Wildlife Conservation Commission (FWCC) and FNAI based on 2010 aerial imagery

(http://myfwc.com/research/gis/applications/articles/Cooperative-Land-Cover) and extracted a list of land cover types C. floridana occurs on (Appendix, Table A2). Nineteen land cover types were identified. Using expert opinion, we ranked these habitat types by suitability for C. floridana. ‘Scrub’ and ‘Scrubby Flatwoods’ were given the highest rank (4), ‘Xeric Hammock’ was given a moderate rank (3), and ‘Sandhill’ and ‘Mesic Flatwoods’ were given a rank of 2. All other land cover types were given a rank of 1 and were not considered to be quality habitat even though we have data indicating that C. floridana occur there. These are places where C. floridana can exist and persist given that open habitat is maintained, but are not considered desirable or natural for C. floridana (e.g., ‘Improved Pasture’, ‘Transportation’, ‘Rural Open Forested’). Some are also the result of the temporal mismatch between the occurrence data and current land cover; one ‘extant’ occurrence record appears in what is now a mowed grass residential front yard.

• Soil Type As in our treatment of land cover type, we identified the NRCS soil types on which C. floridana occurs, and considered all equally suitable with no associated ranking (Appendix, Table A2). (Data updated in 2015, accessed from Florida Geographic Data Library https://www.fgdl.org/)

• Elevation and Slope Using 30-m resolution Digital Elevation Model (DEM) data, we found the maximum elevation (39 m) and slope (2.7 degrees) where C. floridana occurs (maximum 64 m elevation and 14.9 degree slope for the entire 5-county area).

• NDVI NDVI quantifies green leafy cover from remote sensing data by comparing near-infrared (reflected by green vegetation) with red (absorbed by vegetation) light. Higher NDVI values indicate more vegetation, and we used it as a measure of habitat openness. We calculated NDVI from 2017 USDA NAIP imagery at a 5 m resolution and overlaid C. floridana occurrences. Initial exploration with the FNAI EO data revealed an artificially

inflated distribution of NDVI values (mean 0.27, maximum 0.51) as a result of large EO polygons that encompass densely vegetated non-habitat in between occupied patches. As discussed before, there is also a large temporal mismatch between 2017 (NDVI data) and FNAI data for many populations, some of which have not been surveyed since 2006. Consequently, we used only the more recent BTG/ABS/Service study location points (Figure 6; mean NDVI 0.17, maximum 0.44, n = 168 points across 17 study sites). Points from introduced populations might be biased toward low NDVI (open habitat) because open areas were initially targeted for plantings.

Frequency

NDVI Figure 6. Frequency distribution of NDVI values from C. floridana study locations. Lower values indicate more open habitat.

We cut off the highest 5% of NDVI values (NDVI > 0.36), assuming from the shape of the distribution that these were areas not highly beneficial for C. floridana. Areas with NDVI ≤ 0.36 were given a rank of 1 (suitable), and areas with NDVI > 0.36 were given a rank of 0 (unsuitable).

We combined the above factors to identify potential quality habitat across the species range. We excluded all areas that were not of a land cover or soil type where C. floridana is known to occur and all areas that exceeded the maximum elevation and slope where C. floridana is known to occur. For remaining areas, we multiplied land cover rank (1-4) by NDVI rank (0-1) to produce a final habitat quality score ranging from 0 to 4. To assess the ability of this score to provide information about population resilience, we calculated a mean habitat quality score for each population. We pooled together the two highest and two lowest population size classes to

25 increase sample sizes for the comparison (the lowest population size class had only a sample size of 2) and found that on average C. floridana populations with over 500 plants (n = 19) are found in habitat with median habitat quality score about 1 rank higher than those with fewer than 500 plants (n = 7, Figure 7). However, there was wide variation in habitat quality scores within population size groups, and the difference was not statistically significant (ANOVA, F(1,24)=0.236, p = 0.63). Based on these results, all land cover types with a rank of 2 or higher (‘Scrub’, ‘Scrubby Flatwoods’, ‘Xeric Hammock’, ‘Sandhill’, and ‘Mesic Flatwoods’) were considered potential quality habitat that can support resilient populations, and we created a final habitat layer of these land cover types in otherwise suitable areas (suitable soil, NDVI, elevation, and slope). This is the available potential habitat that we measured within a 2-km radius of C. floridana occurrences as a factor contributing to resilience. We stress that this is potential habitat based on current expert knowledge and occurrence data, and that much remains unknown about specific habitat requirements; there are ample scrub and scrubby flatwoods patches that do not currently support C. floridana for unknown reasons (A. Johnson & C. Peterson, personal comm.). However, this basic model represents the current state of knowledge.

Figure 7. Average habitat quality scores of C. floridana populations of different sizes

5.2.4 Classifying Resilience We combined the above factors to create resilience classes as shown in Table 1.

Table 1. Strategy for assigning current resilience scores to populations of C. floridana.

Population Size Habitat Habitat Habitat Area (# plants) Protected Not Protected Available Small <100 LOW Large

Low Low Small 100-500 Moderate Low Large Moderate Moderate Small 501-1000 High Moderate Large

High High Small >1000 VERY HIGH High Large

Resilience classes are based primarily on population size as described above, with 4 resilience classes corresponding to 4 population size categories. Populations with fewer than 100 individuals were assigned low resilience. Within the three higher population size categories (100-500, 501-1000, and > 1000 plants), populations were assigned a baseline resilience score associated with their population size (moderate, high, or very high, respectively). This baseline score could then be lowered by either of the two other factors, habitat protection and habitat area available.

Populations that occur on non-protected lands were assigned to the resilience class one step lower than they would if they were on protected lands. By doing so, we do not mean to discount the importance of populations on non-protected lands to the viability of the species or imply that owners of these parcels are managing the land poorly or are harming C. floridana. Large populations of C. floridana can be supported on private lands. For example, when private landowners burn pasture to improve forage for cattle, they may incidentally improve habitat for C. floridana. However, even large populations of fire-adapted scrub plants can crash quickly due to poor management (e.g., Polygal lewtonii, Weekley and Menges 2012; Warea carteri, Quintana-Ascenscio et al. 2011) and these lands that are not protected for conservation run a higher risk of changes in management or land use that could harm C. floridana populations. For populations that extended across property boundaries and contain individuals occurring on both protected and non-protected lands, we used the protection status that applied to the majority of individuals to classify the entire population.

Populations occupying and/or surrounded by a small area of available habitat were assigned to the resilience class one step lower than they would if they existed within a larger area of available habitat, as they are less able to withstand and recover from perturbations or shift across a landscape as habitat quality changes. For any populations experiencing both of these resilience- reducing conditions (small habitat area on non-protected lands), their resilience score was only reduced one step rather than being demoted twice, once for each factor.

Resilience refers to the ability of populations to withstand stochastic events, whether demographic, environmental, or anthropogenic. For this SSA, empirical data are not available to associate resilience categories with specific quantitative extinction risks or probabilities of persistence. Rather, we are limited to providing qualitative definitions of each resilience category. Populations with low resilience are highly vulnerable to stochastic events and face a high risk of extirpation within the next few decades. Populations with moderate resilience are less likely to be extirpated within the next few decades, but require additional growth (with help of regular habitat management and/or restoration) to become more self-sustaining and resilient to stochastic events. Populations with high resilience are unlikely to be extirpated within the next few decades in the absence of catastrophes or significant declines in the quality of habitat

28 management. Populations with very high resilience are the most robust and resistant to stochastic fluctuations.

Summaries of the 30 delineated populations and their resilience scores are provided in Table 2.

Table 2. Summary of C. floridana populations, including population name and Florida Natural Areas Inventory (FNAI) Element Occurrence (EO) number, if any; county; ownership; origin (natural or introduced) and first observation of the population; last observation of the population; population size class during the last observation; habitat protection status, area of available habitat (acres) and current resilience score. Population sizes come primarily from Bok Tower Gardens, FNAI, and Hillsborough County unpublished data. For habitat protection and habitat area, values that are beneficial to resilience are shaded green, and values that are detrimental are shaded red.

Population EO # County Ownership Origin Last Population Size Habitat Habitat Resilience (First Observed) Observed Class (# plants) Protected Available (acres) Alafia River State Introduced Park 47 Hillsborough FDEP 1986, 1987 2011 501 – 1,000 Yes 2 – 101 Moderate Alafia Scrub Natural Preserve 36 Hillsborough HC 2006 2010 > 1,000 Yes 20.2 High 3 (includes Balm Boyette Scrub former EOs Natural Preserve A 19 and 41) Hillsborough HC 2006 2006 501 – 1,000 Yes 762 High Balm Boyette Scrub Natural Preserve B 18 Hillsborough HC 2005 2008 100 – 500 Yes 428 Moderate Balm Boyette Scrub Natural Preserve C 39 Hillsborough HC 2006 2006 100 – 500 Yes 299 Moderate Balm Scrub Natural Preserve 40, 42 Hillsborough HC 2006 2006 501 – 1,000 Yes 670 High

Bell Creek Preserve 28 (Bell Natural 2010 (Bell) and Rhodine Scrub Creek), 2 2004 (Bell) 2008 Preserve (Rhodine) Hillsborough HC 1976 (Rhodine) (Rhodine) > 1,000 Yes 341 Very High Boyd Hill Nature Introduced Preserve 49 Pinellas CSP 1986, 1987, 1989 2010 501 – 1,000 Yes 65.6 High Bullfrog Creek Natural Scrub Preserve 30 Hillsborough FWC 2004 2006 100 – 500 Yes 508 Moderate Duette Preserve Introduced 2015 North 54 Manatee MC 2013 501 – 1,000 Yes 1086 High Duette Preserve Introduced South 53 Manatee MC 2013 2015 501 – 1,000 Yes 502 High Natural Fish Hawk 14 Hillsborough HC 1989 2010 100 – 500 No 88 Low Fort De Soto County Introduced Park 56 Pinellas PC 1986, 1987, 1989 2009 < 100 Yes 0 Low Yes Golden Aster Scrub 9, 38 Natural *EO 24 Preserve (Golden HC (Golden 1998 2008 (103 and I-75 exit for Aster) Aster) 2006 2009 plants) not County Road 672 24 (1-75) Hillsborough FDOT (1-75) 1997 2010 > 1,000 protected 853 Very High Highlands Hammock State Natural Park 48 Highlands FDEP 1986 2016 100 – 500 Yes 149 Moderate

Natural Johnston Road and 34 1987 Littell Property (Johnston) Hardee Private 2008 2010 501 – 1,000 No 120 Moderate Lake Manatee State Natural Park 16 Manatee FDEP 1987 2014 > 1,000 Yes 315 Very High Little Manatee Natural River 5 Hillsborough SWFWMD 1980 2017 < 100 Yes 59.2 Low Little Manatee River (Cordell) Southfork East and Introduced West 51, 46 Manatee SWFWMD 2008,2009 2015 > 1,000 Yes 478 Very High Natural Little Manatee 1998 River State Park 31, 44 Hillsborough FDEP 2006 2009 > 1,000 Yes 751 Very High Lucky Lonesome Trail (aka Natural Singletary Road) 35 Hardee Private 1998 2018 100 – 500 No 96 Low Introduced McKay Creek 52 Pinellas PC 2009 2015 > 1,000 Yes 0 High Moody Branch Natural Mitigation Park 29 Manatee FWC 2004 2017 > 1,000 Yes 758 Very High Paynes Creek Introduced Historic State Park 50 Hardee FDEP 2011 2015 > 1,000 Yes 17 High Natural Stephens Road 5 Hillsborough Private unknown 2010 100 – 500 No 59.2 Low Sun City Heritage Natural Park Addition 6 Hillsborough HC 1990 2017 501 – 1,000 Yes 59.2 High Natural Sweetwater Road 13 Hardee Private 1987 2010 > 1,000 No 152 High SWFWMD, some Upper Little FDOT or on Natural Manatee River A 32 Hillsborough private land 1998 2010 < 100 Yes 165 Low Upper Little Natural Manatee River B 43 Hillsborough SWFWMD 2003 2009 > 1,000 Yes 328 Very High Introduced Weedon Island 55 Pinellas PC 2010 2015 > 1,000 Yes 11.3 High

BTG – Bok Tower Gardens, CSP – City of St. Petersburg, FDEP – Florida Department of Environmental Protection, FDOT – Florida Department of Transportation, FNAI – Florida Natural Areas Inventory, FWC – Florida Fish and Wildlife Conservation Commission, HC – Hillsborough County, MC – Manatee County, PC – Pinellas County, SWFWMD – Southwest Florida Water Management District 1 The habitat model identified 2.1 acres of potential habitat at Alafia Scrub Preserve, but more recent aerial photography shows the presence of ~8 more acres of scrub that may not have been apparent when the land cover map was made from 2010 aerial imagery.

Approximately two thirds of the assessed populations were classified as having high or very high resilience (Table 3). Populations on protected lands were more likely to have high or very high resilience (17 populations) than low or moderate resilience (8 populations), while populations on non-protected lands were more likely to have low or moderate resilience (4 populations) than high resilience (1 populations).

Table 3. Summary of resilience scores tallied across all populations of C. floridana.

Resilience Class All Populations Protected Not Protected Very High 7 7 0 High 11 10 1 Moderate 6 5 1 Low 6 3 3

5.3 Current Redundancy and Representation Redundancy for C. floridana is inherently fairly low compared to wide-ranging species because it is an endemic species with a narrow range around the Tampa Bay region in Florida and Hardee County farther inland (with one population just across the border in Highlands County). The entire species range spans 5 counties, with half of the populations occurring in Hillsborough County. The longest distance between 2 populations is 131 km. However, conceding that this is a narrow-ranging endemic, the spatial distribution of populations across its range does confer a moderate amount of redundancy, defined as the ability of the species to withstand catastrophic events. Catastrophic events could include, among others, too frequent fires, droughts, disease outbreaks, or hurricanes with prolonged flooding, each of which have impacts at a different spatial scale. It is worth noting that no information is currently known about seedbank resilience in the soil for this species; without knowing this, it is difficult to predict long term impacts of catastrophes.

The 30 assessed populations are distributed in 3 main groupings (Figure 8, with the number of highly and very highly resilient populations displayed for each grouping). There is about 20 – 30 km between each of the groupings, providing a buffer around each that may protect them from catastrophic events affecting the others (e.g., disease outbreak, depending on transmission type and vectors). Within each geographic cluster, there are at least 2 highly or very highly resilient

populations, which could serve as sources to naturally recolonize populations lost to catastrophic events. The Hardee-Highlands cluster has the lowest redundancy (2 resilient populations, 6 populations total) and is the most isolated from the other clusters. The Pinellas cluster has the next lowest redundancy of resilient populations (3 resilient populations, 4 populations total), and the Hillsborough-Manatee cluster has the highest redundancy (13 resilient populations, 25 populations total). Another factor contributing to redundancy is the wide range of property ownership; with so many managing entities, the species as a whole is buffered against poor management of any one entity (e.g., due to budget issues or changing priorities). Based on the spatial distribution of resilient populations managed by a variety of entities across a narrow range, current redundancy is considered qualitatively to be low-moderate. Rather than placing a lot of stock in this rather subjective classification, the value of characterizing current redundancy is most useful in comparison to redundancy under the future scenarios.

Figure 8. Spatial distribution of Chrysopsis floridana populations in three main geographic clusters across 5 labelled counties. The number of populations with high and very high resilience is shown within each cluster.

Representative units for this species could not be defined based on available data, with representation defined as the ability of the species to adapt to changing environmental conditions. Species experts contributing to this SSA suspect that there might be representative units with different genetic adaptations associated with soil differences, elevation above the water table, fire regime, or habitat structure. However, there are no data currently to confirm or refute these hypotheses. Genetic studies have found little to no genetic clustering among populations, with 80% of observed genetic variation occurring within populations, and only 20% of the variation attributable to between-population differences (Markham 1998). These results support the existence of a single representative unit for the species. However, that study did not examine genetic markers known to be associated with adaptive traits. Vital rates and morphology were observed to differ between individuals from different source populations that were grown at BTG and introduced to other sites (Campbell 2008). This provides evidence that there might be adaptive differences between different “types” of C. floridana across the species range. However, without any firm evidence to define representative units, we refrain from doing so here. Future research on C. floridana genetics, and life history and habitat differences can provide a more definitive basis for defining representative units in future iterations of the SSA.

6 FUTURE CONDITION

6.1 Future Considerations We developed 3 hypothetical future scenarios under which to assess the future viability of C. floridana in terms of resilience, redundancy and representation. Based on expert opinion, the life span of C. floridana, ideal fire return intervals (at least every 10 years) uncertainty about future conditions, and lack of knowledge about many basic aspects of C. floridana ecology, we chose to project populations 20 years into the future under each scenario, but no further. The most important factors identified by species experts to consider into the future were habitat quantity and quality.

6.1.1 Habitat Quantity Habitat quantity can be negatively impacted by development or land use change (particularly on private lands) or positively impacted by land acquisition, restoration, and/or introductions into unoccupied sites that already have presumably suitable habitat.

6.1.2 Habitat Quality Habitat quality is closely tied to active habitat management to maintain openness, either by prescribed burning or other types of management. In constructing our scenarios, we considered two avenues by which future habitat management can be influenced. First, the managing entities can choose their desired level of management effort, by implementing (or not) a management plan, allocating funding or personnel to or away from habitat management among competing priorities and limited resources, etc. For our scenarios, we allowed for 3 levels of habitat management effort by managing entities. The first was management for stability, a moderate level of management that would be expected to maintain populations at their current size. The other two management levels were an increase, or a decrease, compared to management for stability. An increase in management effort would be expected to grow populations, while a decrease in management would be expected to cause population declines. We unfortunately do not have the information to compare these somewhat vague future management effort levels to levels of management effort that are currently being implemented at each population.

The second avenue by which future habitat management can be influenced is development, particularly major roads and types of development associated with “vulnerable” human populations (e.g., schools, hospitals). This kind of development surrounding habitat limits management via prescribed burns by limiting the days that burns can take place; weather conditions have to align to ensure proper smoke management. For example, if a population is surrounded by nearby development to the north and west, it can be burned when the wind is blowing to the south and east. As more development hems in populations, there is less flexibility in when they can be burned. In trying to determine the appropriate radius around populations within which development might impact management, the most common answer we encountered was “it depends”. It depends for each burn on the type of development, temperature, humidity, wind conditions, size of the planned burn, risk tolerance of those implementing the burn, and

other factors. Answers (beyond “it depends”) that we received from Florida Forest Service employees (B. Camposano, J. Saddler) ranged from 0.5 km up to 5 miles. We chose an intermediate value, 5 km, in which to examine current and predicted future development. In choosing this concrete value, we concede that it depends, and some burns will need to consider areas greater or less than 5 km away, but this allows us to gain a general understanding of the risks of development on management surrounding populations.

Within a 5-km radius around C. floridana occurrences, we used GIS to examine current and projected urbanization and roads. Urbanization data came from the SLEUTH model (Slope, Land use, Excluded, Urban, Transportation and Hillshade; Jantz et al. 2010), and road data was available from the Florida Department of Transportation. The SLEUTH model has previously been used to predict probabilities of urbanization across the southeastern US in 10-year increments, and the resulting GIS data are freely available (Belyea and Terrando 2013). For our 20-year future projection, we used the SLEUTH data sets from the years 2020 (closest to current year 2018) and 2040 (closest to 20 years in the future), and examined the area predicted, with at least 80% probability, to be urbanized. Our assessment was both quantitative, calculating the area within the 5-km buffer surrounding each population that was urbanized at each time point, and qualitative, inspecting the distribution of urbanization and major roads within that area (e.g., is the urbanization concentrated to one side of the population or completely surrounding it?). With this quantitative and qualitative assessment, we categorized populations as having either low risk or high risk of development impacting their ability to manage. We define high risk of impacting management as >50% chance of negatively impacting management for C. floridana (more likely than not), and vice versa for low risk. Based on all of the “it depends” factors discussed above, we did not try to estimate risk more precisely than this. Descriptions of how we arrived at each population’s risk level are found in the following section. Populations that were classified as having low risk from development averaged 7.9% developed area within the 5-km buffer by 2040, with a range of 0% to 39% developed. Populations that were classified as having high risk from development averaged 45.5% developed area within the same buffer, ranging from 23% to 85%. For three populations with a percent of developed area in the overlapping range between the two categories (23% to 39% developed), the deciding factor between low risk and high risk was the distribution of development and roads around the population.

These two aspects of future management, (1) management resources and will from the management entity, and (2) impacts of development on management, interacted in our future scenarios in the following way (Table 4). With decreases in management effort (compared to management for stable populations), population resilience decreases one level. With management for stability, population resilience stays the same as the current condition resilience when there is low risk of development impacts, but where there is a high risk, resilience decreases one level, reflecting that management will be more challenging with higher risk from development. With increases in management effort, population resilience increases when there is low risk of development impacts, but stays the same when there is a high risk; the increased management effort cancels out the increased difficulty caused by development.

Table 4. Interaction of two aspects of future habitat management in determining future resilience; management effort and risk of development impacts on management.

Decreased Effort Management for Increased Effort Stability Development - Low Risk Resilience decrease No change Resilience increase Development – High Risk Resilience decrease Resilience decrease No change

Our consideration of development impacts on management focused entirely on management via prescribed burning, which is not the only management option available. For small populations or populations where burning is not an option, mechanical or manual habitat management can be highly effective (M. Jenkins, personal communication). However, these alternate management methods are more costly than burning over large areas, and may create additional risks, such as increased invasion by non-native species. If development renders prescribed burning unfeasible for a population and the managers switch to another method of habitat management, habitat quality might still suffer if the increased cost limits the amount of habitat that can be maintained.

As we applied these management impacts to populations and projected them into the future, populations only increased or decreased by one level of resilience. Populations can grow or decline very rapidly, but we do not fully understand what drives these dynamics, and do not have reliable estimates of population growth rates to use to more quantitatively project population

37 sizes into the future. Thus, we are conservative in the magnitude of our projections, but more confident in the direction of the change.

6.1.2.1 Development Risk Assessments • Alafia River State Park There is a low risk of present and future development limiting the ability to manage habitat at Alafia River State Park. Less than 1% of the 5km area around known C. floridana occurrences is predicted to be developed by 2040 and there are few major roads surrounding the population.

• Alafia Scrub Preserve There is a high risk of present and future development limiting the ability to manage habitat at Alafia Scrub Preserve. By 2040, 66% of the 5 km area around the known C. floridana occurrences is predicted to be developed. Current and future predicted development hem in the property on all sides, as do numerous major roads.

• Balm Boyette Scrub Preserve A There is a low risk of present and future development limiting the ability to manage habitat at Balm Boyette Scrub Preserve A. While development within 5 km of C. floridana occurrences is predicted to increase from 9% to 14% of the area, the plants are found in the interior of a large protected area, and nearby development is restricted primarily to two directions, both factors that contribute more flexibility for prescribed burns.

• Balm Boyette Scrub Preserve B There is a low risk of present and future development limiting the ability to manage habitat at Balm Boyette Scrub Preserve B. While development within 5 km of C. floridana occurrences is predicted to increase from 21% to 29% of the area, the large area of the rest of the preserve to the east provides some flexibility to burn.

• Balm Boyette Scrub Preserve C There is a low risk of present and future development limiting the ability to manage habitat at Balm Boyette Scrub C. Less than 6% of the 5-km area around known C. floridana occurrences is predicted to be developed by 2040, the population is surrounded by a large quantity of protected area, and there are few major roads surrounding the population.

• Balm Scrub Preserve There is a high risk of present and future development limiting the ability to manage habitat at Balm Scrub Preserve. By 2040, the amount of development within the 5-km area around the known C. floridana occurrences is predicted to increase from 16% to 23% developed. Major roads occur within a few km of the population in multiple directions.

• Bell Creek Preserve and Rhodine Scrub Preserve There is a high risk of present and future development limiting the ability to manage habitat at Bell Creek and Rhodine Scrub Preserves. By 2040, the amount of development within the 5-km area around the known C. floridana occurrences is predicted to increase from 47% to 51% developed, and Bell Creek Preserve is adjacent to a school. Current and future predicted development hem in the property on all sides, as do numerous major roads.

• Boyd Hill Nature Preserve There is a high risk of present and future development limiting the ability to manage habitat at Boyd Hill Nature Preserve. By 2040, 54% of the 5-km area around the known C. floridana occurrences is predicted to be developed, a marginal difference from the current condition because there is no room for further development; the remaining 46% that is currently undeveloped is primarily open water. Current and future predicted development hem in the property on all sides, as do numerous major roads.

• Bullfrog Creek Scrub Preserve There is a high risk of present and future development limiting the ability to manage habitat at Bullfrog Creek Scrub Preserve. By 2040, the amount of development within the 5-km area around the known C. floridana occurrences is predicted to increase from 31% to 39% developed. Current and future predicted development hem in the property on nearly all sides, as do numerous major roads, including Interstate 75. However, Hillsborough County owns about 800 undeveloped acres along the preserve’s eastern border, providing some buffer and ability to burn with winds blowing west to east.

• Duette Preserve North There is a low risk of present and future development limiting the ability to manage habitat at Duette Preserve North. None of the 5-km area around known C. floridana occurrences is predicted to be developed by 2040, occurrences are tucked into a large protected area, and there are few major roads surrounding the population.

• Duette Preserve South There is a low risk of present and future development limiting the ability to manage habitat at Duette Preserve South. None of the 5-km area around known C. floridana occurrences is predicted to be developed by 2040, occurrences are tucked into a large protected area, and there are few major roads surrounding the population.

• Fish Hawk Creek Preserve There is a high risk of present and future development limiting the ability to manage habitat at Fish Hawk Creek Preserve. By 2040, the amount of development within the 5- km area around the known C. floridana occurrences is predicted to increase from 48% to 49% developed. Current and future predicted development hem in the population on all sides, as do numerous major roads.

• Fort De Soto County Park There is a low risk of present and future development limiting the ability to manage habitat at Ford De Soto County Park. Less than 2% of the 5-km area around known C.

floridana occurrences is predicted to be developed by 2040 and the population is surrounded primarily by water.

• Golden Aster Scrub Preserve, County Road, & 1-75 Interchange There is a high risk of present and future development limiting the ability to manage habitat at Golden Aster Scrub Preserve and surrounding C. floridana occurrences. By 2040, the amount of development within the 5-km area around the known C. floridana occurrences is predicted to increase from 39% to 45% developed. Current and future predicted development hem in the property on all sides, as do numerous major roads, including Interstate 75.

• Highlands Hammock State Park There is a low risk of present and future development limiting the ability to manage habitat at Highlands Hammock State Park. Less than 4% of the 5-km area around known C. floridana occurrences is predicted to be developed by 2040 and there are few major roads surrounding the population.

• Johnston Road and Littell Property There is a low risk of present and future development limiting the ability to manage habitat at the Johnston Road and Littell Property population. None of the 5-km area around known C. floridana occurrences is predicted to be developed by 2040. The areas where C. floridana occurs here are pasture, and as long as they continue being used and managed as they are now, habitat will likely retain the same suitability for C. floridana. The uncertainty about whether this will happen is already reflected in the current conditions resilience score, where the population is reduced one rank for being on non- protected lands.

• Lake Manatee State Park There is a low risk of present and future development limiting the ability to manage habitat at Lake Manatee State Park. Less than 6% of the 5-km area around known C. floridana occurrences is predicted to be developed by 2040 and there are few major roads

surrounding the population, although State Road 64 does border the park’s southern boundary.

• Little Manatee River, Sun City Heritage Park Addition and Stephens Road There is a high risk of present and future development limiting the ability to manage habitat at these three nearby populations. By 2040, the amount of development within the 5-km area around the known C. floridana occurrences is predicted to increase from 24% to 32% developed. Current and future predicted development hem in the populations on all sides, as do numerous major roads, including Interstate 75.

• Little Manatee River (Cordell) East and West There is a low risk of present and future development limiting the ability to manage habitat at Alafia River State Park. None of the 5-km area around known C. floridana occurrences is predicted to be developed by 2040 and there are few major roads surrounding the population.

• Little Manatee River State Park There is a high risk of present and future development limiting the ability to manage habitat at Little Manatee River State Park. By 2040, the amount of development within the 5-km area around the known C. floridana occurrences is predicted to increase from 40% to 46% developed. Current and future predicted development hem in the property on all sides, as do numerous major roads.

• Lucky Lonesome Trail/Singletary Road There is a low risk of present and future development limiting the ability to manage habitat at the Lucky Lonesome Trail/Singletary Road population. Less than 3% of the 5- km area around known C. floridana occurrences is predicted to be developed by 2040. The areas where C. floridana occurs here are on private non-scrub land, and as long as they continue being used and managed as they are now, habitat will likely retain the same suitability for C. floridana. The uncertainty about whether this will happen is already reflected in the current conditions resilience score, where the population is reduced one

rank for being on non-protected lands.

• McKay Creek There is a high risk of present and future development limiting the ability to manage habitat at McKay Creek. By 2040, the amount of development within the 5-km area around the known C. floridana occurrences is predicted to increase marginally to 85%, with practically no more urbanization possible. Current and future predicted development hem in the property on all sides, as do numerous major roads.

• Moody Branch Preserve There is a low risk of present and future development limiting the ability to manage habitat at Moody Branch. Less than 1% of the 5-km area around known C. floridana occurrences is predicted to be developed by 2040 and there are few major roads surrounding the population.

• Paynes Creek Historic State Park There is a low risk of present and future development limiting the ability to manage habitat at Paynes Creek Historic State Park. By 2040, the amount of development within the 5-km area around the known C. floridana occurrences is predicted to increase from 14% to 20% developed, with nearly all development occurring along the US Highway 17 corridor to the west. While this will limit burning when smoke could move in that direction, the configuration of development and roads still provides some management flexibility.

• Sweetwater Road There is a low risk of present and future development limiting the ability to manage habitat at the Sweetwater Road population. None of the 5-km area around known C. floridana occurrences is predicted to be developed by 2040. The areas where C. floridana occurs here are on non-protected land along a rural road, and as long as the land continues being used and managed as it is now, habitat will likely retain the same suitability for C. floridana. The uncertainty about whether this will happen is already

reflected in the current conditions resilience score, where the population is reduced one rank for being on non-protected lands.

• Upper Little Manatee River A There is a low risk of present and future development limiting the ability to manage habitat at Upper Little Manatee River A. Only 4% of the 5-km area around known C. floridana occurrences are predicted to be developed by 2040 and there are few major roads surrounding the population, although the population does border County Road 579.

• Upper Little Manatee River B There is a low risk of present and future development limiting the ability to manage habitat at Upper Little Manatee River B. By 2040, the amount of development within the 5-km area around the known C. floridana occurrences is predicted to increase from 10% to 13% developed, but the development and few major roads surrounding the population are concentrated to a few directions.

• Weedon Island There is a low risk of present and future development limiting the ability to manage habitat at Weedon Island. By 2040, the amount of development within the 5-km area around the known C. floridana occurrences is predicted to increase marginally to 39%, with practically no more urbanization possible. Urbanization and roads however are concentrated to the west of the population, and with a prevailing eastward sea breeze, management should not be severely limited.

6.2 Future Scenarios Below we present 3 hypothetical future scenarios for C. floridana.

6.2.1 Status Quo Under the Status Quo scenario, no new protected areas are acquired and no new populations are introduced. Management effort for all populations is moderate for population stability, assuming that the ability to do so will not be hampered by funding or political issues, climate change, or

44 other factors. As a result, resilience scores remain stable unless there is a high risk of development impacts on management. Of the most recent batch of introductions since 2008, all have reached sizes > 1000 plants except for the populations at Duette Preserve (2 populations, North and South). Given that the Duette populations were the most recently introduced populations (2013), have been growing rapidly, and are surrounded by ample habitat and little to no development, these two populations are projected to increase from high to very high resilience.

6.2.2 Pessimistic Under the Pessimistic scenario, management effort on all populations decreases, presumably as an effect of a wide-scale change in priorities and/or resources, resulting in a drop in resilience scores across the board. Additionally, there is uncertainty in whether populations on non- protected lands will continue to be managed in a way that is compatible with continued C. floridana persistence. In this scenario, all populations on non-protected lands are lost, due to presumed land use or management change. No new protected areas are acquired, and no new populations are introduced.

6.2.3 Targeted Conservation Under the Targeted Conservation scenario, conservation resources are focused on keeping highly resilient populations highly resilient, and strengthening moderately resilient populations on protected lands.

• All populations that are highly or very highly resilient currently are either managed for stability if they are under a low risk from development, or receive increased management to counteract a high risk from development, with an end result that they all remain in the same category. The exceptions are Duette Preserve North and South, which increase from high to very high for the reasons described in the Status Quo scenario.

• Populations with currently moderate resilience are treated differently based on their habitat protection status. On non-protected lands, management is for stability, so they either remain moderately resilient, or drop to low resilience if under high risk from

development. On protected lands, management is increased, resulting in a shift to high resilience for populations under low risk from development. Populations under high risk from development remain moderately resilient.

• Populations with low resilience currently are managed for stability, and any under high risk from development are extirpated. Resources are not poured into rescuing these populations, but are instead conserved for improving the status of other populations, and introducing new ones. These populations could also serve to provide material for ex situ, augmentation, and introduction efforts.

• Five new populations are introduced within the 20-year projection time. In selecting areas for new populations, we chose sites a) with ample suitable habitat (as identified in the habitat modelling exercise described in the Current Conditions section), b) at low risk of development impacts on management, and c) to improve redundancy (distribution and connectivity) across the species range. All introduced populations are projected to reach high resilience by the end of the projection time, justified by the success of the populations reintroduced from 2008-2013. These sites were selected based on GIS investigations and expert input and serve as potential new sites; they should be visited and assessed more rigorously before any acquisitions or introductions are planned. These sites are:

o Alafia River State Park, Hillsborough County – There is already an introduced population in the northeastern corner of the park, but there is another patch of suitable habitat in the northwestern corner, 2.9 km away from the first, separated by the Peace River Floodplain and reclaimed mine lands. Introducing a new population here will help fill in an area that is less dense with populations than farther west in Hillsborough County.

o Edward Chance Reserve, Manatee County, west of Duette Preserve – This is an area surrounded by ample habitat, and fills in gaps in the species distribution. This site is west of existing populations on Duette Preserve and south of the Cordell population.

47 o Edward Chance Reserve or Well Field Scrub-Jay Habitat, Manatee County, South of Duette Preserve – This is an area surrounded by ample habitat, and is south of the flourishing populations in Duette Preserve, but potentially outside of the species historical range.

o Southfork State Park, Manatee County – This park contains a large patch of habitat, and could serve to connect the Moody Branch population to the northeast and the Cordell population to the southeast. Element occurrences on Moody Branch are immediately adjacent to the property boundary, so it is possible that C. floridana already exists on this property.

o Private Land, Hardee County – Suitable habitat in Hardee County is limited, but we selected a site (~21 acres of scrub) on private land (presently owned by WALCO Leasing LLC), for potential acquisition and introduction to increase redundancy in this far eastern and isolated portion of the species range. Other nearby populations in Hardee and Highlands Counties are pictured in the below map as well.

6.2.4 Likelihood of Scenarios Status Quo is the most likely scenario given present listing status, conservation funding, and the pace of development. The Pessimistic scenario is unlikely; given that C. floridana populations span so many different ownerships, it is unlikely that all of the different managing entities will decrease their management effort, especially when there are other co-occurring threatened, endangered, and candidate species occupying the same habitat (e.g., Florida Scrub-jay, Aphelocoma coerulescens; eastern indigo snake, Drymarchon couperi; gopher tortoise, Gopherus polyphemus). The Targeted Conservation scenario is not likely with current conservation resources, but reflect a likely future if both habitat-focused management (e.g., prescribed burning) by a variety of partners/managing entities and species-specific conservation (e.g., captive propagation and introductions) are prioritized and well-funded.

6.3 Future Resilience Future resilience in 20 years of C. floridana populations under 3 scenarios are shown in Table 5 and summarized in Table 6 and Figure 9.

Table 5. Chrysopsis floridana populations and associated resilience under current conditions and 3 hypothetical future scenarios, Status Quo, Pessimistic, and Targeted Conservation. Development management risk refers to the risk that development around the population will negatively impact the ability to manage the habitat. Management effort associated with each scenario refers to the will and resources applied to habitat management by the managing entities, relative to management for stable populations. For habitat protection, development management risk, and management effort, values that are beneficial to resilience are shaded green, and values that are detrimental are shaded red.

Future – Status Quo Future – Pessimistic Future – Targeted Conservation

Population EO # County Population Habitat Current Development Management Future Management Future Management Future Size Class Protected Resilience Management Risk Effort Resilience Effort Resilience Effort Resilience (# plants) Alafia River State Park 47 Hillsborough 501 – 1,000 Yes Moderate Low Risk Stability Moderate Decrease Low Increase High Alafia Scrub Preserve 36 Hillsborough > 1,000 Yes High High Risk Stability Moderate Decrease Moderate Increase High 3 (includes Balm Boyette Scrub former EOs 19 Preserve A and 41) Hillsborough 501 – 1,000 Yes High Low Risk Stability High Decrease Moderate Stability High Balm Boyette Scrub Preserve B 18 Hillsborough 100 – 500 Yes Moderate Low Risk Stability Moderate Decrease Low Increase High Balm Boyette Scrub Preserve C 39 Hillsborough 100 – 500 Yes Moderate Low Risk Stability Moderate Decrease Low Increase High

Balm Scrub Preserve 40, 42 Hillsborough 501 – 1,000 Yes High High Risk Stability Moderate Decrease Moderate Increase High Bell Creek Preserve 28 (Bell and Rhodine Scrub Creek), 2 Preserve (Rhodine) Hillsborough > 1,000 Yes Very High High Risk Stability High Decrease High Increase Very High Boyd Hill Nature Preserve 49 Pinellas 501 – 1,000 Yes High High Risk Stability Moderate Decrease Moderate Increase High Bullfrog Creek Scrub Preserve 30 Hillsborough 100 – 500 Yes Moderate High Risk Stability Low Decrease Low Increase Moderate Duette Preserve South 54 Manatee 501 – 1,000 Yes High Low Risk Stability High Decrease Moderate Stability Very High

Duette Preserve North 53 Manatee 501 – 1,000 Yes High Low Risk Stability High Decrease Moderate Stability Very High Likely Likely Likely Fish Hawk 14 Hillsborough 100 – 500 No Low High Risk Stability Extirpated Decrease Extirpated Stability Extirpated Likely Likely Likely Fort De Soto County Park 56 Pinellas < 100 Yes Low High Risk Stability Extirpated Decrease Extirpated Stability Extirpated Yes Golden Aster Scrub *EO 24 Preserve 9, 38 (Golden (103 plants) and I-75 exit for County Aster) not Road 672 24 (1-75) Hillsborough > 1,000 protected Very High High Risk Stability High Decrease High Increase Very High Highlands Hammock State Park 48 Highlands 100 – 500 Yes Moderate Low Risk Stability Moderate Decrease Low Increase High Johnston Road and Littell Likely Property 34 (Johnston) Hardee 501 – 1,000 No Moderate Low Risk Stability Moderate Decrease Extirpated Stability Moderate Lake Manatee State Park 16 Manatee > 1,000 Yes Very High Low Risk Stability Very High Decrease High Stability Very High Likely Likely Likely Little Manatee River 5 Hillsborough < 100 Yes Low High Risk Stability Extirpated Decrease Extirpated Stability Extirpated Little Manatee River (Cordell) Southfork East and West 51, 46 Manatee > 1,000 Yes Very High Low Risk Stability Very High Decrease High Stability Very High

Little Manatee River State Park 31, 44 Hillsborough > 1,000 Yes Very High High Risk Stability High Decrease High Increase Very High

Lucky Lonesome Trail (aka Likely Singletary Road) 35 Hardee 100 – 500 No Low Low Risk Stability Low Decrease Extirpated Stability Low McKay Creek 52 Pinellas > 1,000 Yes High High Risk Stability Moderate Decrease Moderate Increase High Moody Branch Mitigation Park 29 Manatee > 1,000 Yes Very High Low Risk Stability Very High Decrease High Stability Very High Paynes Creek Historic State Park 50 Hardee > 1,000 Yes High Low Risk Stability High Decrease Moderate Stability High Likely Likely Likely Stephens Road 5 Hillsborough 100 – 500 No Low High Risk Stability Extirpated Decrease Extirpated Stability Extirpated Sun City Heritage Park Addition 6 Hillsborough 501 – 1,000 Yes High High Risk Stability Moderate Decrease Moderate Increase High Sweetwater Road 13 Hardee > 1,000 No High Low Risk Stability High Decrease Moderate Stability High Upper Little Manatee Likely River A 32 Hillsborough < 100 Yes Low Low Risk Stability Low Decrease Extirpated Stability Low Upper Little Manatee River B 43 Hillsborough > 1,000 Yes Very High Low Risk Stability Very High Decrease High Stability Very High Weedon Island 55 Pinellas > 1,000 Yes High High Risk Stability Moderate Decrease Moderate Increase High Alafia River State Park, New second population population Hillsborough NA Yes NA High Edward Chance Reserve New west of Duette Population Manatee NA Yes NA High New Hardee Co. Patch New Not population Hardee NA presently NA High Southfork State Park New population Manatee NA Yes NA High Well Field Scrub-Jay Habitat or Edward Chance New Reserve south of Duette population Manatee NA Yes NA High

Table 6. Summary of resilience scores tallied across all populations of C. floridana for the current condition and future condition under 3 hypothetical scenarios, Status Quo, Pessimistic, and Targeted Conservation.

Future – Future - Future - Targeted Resilience Class Current Status Quo Pessimistic Conservation Very High 7 4 0 9 High 11 8 7 18 Moderate 6 11 11 2 Low 6 3 5 2 Likely Extirpated NA 4 7 4

10 Number of Populations 5

0 Current Status Quo Pessimistic Targeted Cons.

Very High High Moderate Low

Figure 9. A graphical representation of the number of C. floridana populations in each resilience class in the current condition and 3 hypothetical future scenarios.

As implied by the scenario name, resilience of populations under the Pessimistic scenario is predicted to be poor, with only 7 highly resilient populations, a decrease from 18 currently highly or very highly resilient populations. Under the Status Quo scenario, resilience is expected to drop (12 highly or very highly resilient populations) due solely to the effect of development limiting the ability to adequately manage habitat. Under the Targeted Management scenario, focused management and conservation efforts to counteract detrimental effects of urbanization,

grow existing populations, and introduce new populations is expected to result in significant gains in resilient populations, with 27 expected highly or very highly resilient populations.

6.4 Future Redundancy and Representation Redundancy 20 years in the future is expected to decrease compared to current condition under the Status Quo and Pessimistic Scenarios (Table 7). In all scenarios, the majority of highly and very highly resilient populations are found in Hillsborough and Manatee counties. All redundancy of highly resilient populations in Pinellas County and the Hardee and Highlands county cluster are lost under the Pessimistic scenario. In the Status Quo scenario, where drops in resilience are due to development risks to management, no highly resilient populations remain in the heavily urbanized Pinellas County. Even in the Targeted Conservation Scenario, redundancy within Pinellas County does not improve, but both the number and distribution of highly resilient populations in the other two clusters does improve. As in the current condition section, we do not assess representation in the future due to a present lack of information needed to delineate representative units.

Table 7. Number of populations with ‘High’ or ‘Very High’ resilience in the 3 spatial groupings of C. floridana populations in the current condition and 3 hypothetical future scenarios.

Future – Geographic Future - Future - Targeted Cluster Current Status Quo Pessimistic Conservation Pinellas County 3 0 0 3 Hillsborough and Manatee 13 10 7 20 Counties Hardee and Highlands 2 2 0 4 Counties

LITERATURE CITED

Anacker, B.L., M. Gogol-Prokurat, K. Leidholm, and S. Schoenig. 2013. Climate change vulnerability assessment of rare plants in California. Madroño, 60:193-210. Andersen, M. 1993. Diaspore morphology and seed dispersal in several wind-dispersed Asteraceae. American Journal of Botany 80:487-492. Belyea, C. M. and A. J. Terrando. 2013. Urban growth modeling for the SAMBI Designing Sustainable Landscapes Project. North Carolina State University. http://www.basic.ncsu.edu/dsl/urb.html. Accessed 10 March 2018. Burrows, F. 1986. The aerial motion of seeds, fruits, spores and pollen. In: Murray, D. (ed.) Seed Dispersal. Academic Press, San Diego, CA. 47 pp. Byers, E., and S. Norris. 2011. Climate change vulnerability assessment of species of concern in West Virginia. West Virginia Division of Natural Resources, Elkins, West Virginia. Campbell, C. 2008. Florida goldenaster (Chrysopsis floridana) population introduction project. Annual report prepared for U.S. Fish and Wildlife Service. North Florida Field Office, Jacksonville, Florida. Carter, L. M., J. W. Jones, L. Berry, V. Burkett, J. F. Murley, J. Obeysekera, P. J. Schramm, and D. Wear. 2014. Ch. 17: Southeast and the Caribbean. Climate Change Impacts in the United States: The Third National Climate Assessment, J. M. Melillo, T.C. Richmond, and G. W. Yohe, (eds.), U.S. Global Change Research Program, 396-417. doi:10.7930/J0NP22CB. Cochrane, B. 2002. Genetic variation in Chrysopsis floridana. Progress report to the Endangered Plant Conservation Program, Florida Division of Forestry. University of South Florida, Tampa, FL. Gates, C. and G. Tanner. 1998. Effects of prescribed burning on herbaceous vegetation and pocket gophers (Geomys pinetis) in a sandhill community. Florida Scientist 51:129-139. Greenleaf, S. S., N. M. Williams, R. Winfree, and C. Kremen. 2007. Bee foraging ranges and their relationship to body size. Oecologia, 153: 589-596. Ellstrand, N. C. and D. R. Elam. 1993. Population genetic consequences of small population size: Implications for plant conservation. Annual Review of Ecology and Systematics 24:217-242. Franks, S.J., J.J. Weber, and S.N. Aitken. 2014. Evolutionary and plastic responses to climate change in terrestrial plant populations. Evolutionary Applications 7:123-139. Hagler, J. R., S. Mueller, L. R. Teuber, S. A. Machtley, and A. Van Deynze, 2011. Foraging range of honey bees, Apis mellifera, in alfalfa seed production fields. Journal of Insect Science 11:144. Hedrick, P. 2014. Conservation genetics and the persistence and translocation of small populations: bighorn sheep populations as examples. Animal Conservation 17:106-114. IPCC (Intergovernmental Panel on Climate Change). 2007. Climate change 2007: The physical science basis. Contribution of working group I to the Fourth Assessment Report of the IPCC. Cambridge University Press. 996 pp. Jantz, C. A., S.J. Goetz, D. Donato and P. Claggett. 2010. Designing and Implementing a Regional Urban Modeling System Using the SLEUTH Cellular Urban Model. Computers, Environment and Urban Systems 34:1-16. Johnson, A. F., A. C. Cox, P. F. Russo, and A. M. Jenkins. 2006. Status survey for Florida goldenaster. Florida Natural Areas Inventory. Tallahassee, Florida.

Karssen, C. and H. Hilhorst. 1992. Effect of chemical environment on seed germination. Pages 327-348. In Finner, M. (ed.). Seeds: The Ecology of Regeneration in Plant Communities. CAB International, Wallingford, Oxon, UK. Kelly, A., and M. Goulden. 2008. Rapid shifts in plant distribution with recent climate change. Proceedings of the National Academy of Sciences, 105:11823-11826. Lambert, B., and E. Menges. 1996. The effects of light, soil disturbance and presence of organic litter on the field germination and survival of the Florida goldenaster, Chrysopsis floridana Small. Florida Scientist 59:121-137. Loarie, S., B. Carter, K. Hayhoe, S. McMahon, R. Moe, C. Knight, and D. Ackerly. 2008. Climate change and the future of California's endemic flora. PloS one 3: e2502. Loydi, A., T. Donath, R. Eckstein, and A. Otte. 2015. Non-native species litter reduces germination and growth of resident forbs and grasses: allelopathic, osmotic, or mechanical effects? Biological Invasions 17:581-595. Mani, M. and J. Saravanan. Pollination Ecology and Evolution in Compositae (Asteraceae). Science Publishers Inc. Enfield, NH, USA. 1999. 166 pp. Markham, L. 1998. Genetic variation in Chrysopsis floridana Small, the endangered Florida golden aster, as revealed by random amplification for polymorphism detection (RAPD). University of South Florida, Master’s Thesis. 53 pp. Matthies, D., I. Bräuer, W. Maibom, and T. Tscharntke. 2004. Population size and the risk of local extinction: empirical evidence from rare plants. Oikos 105:481-488. Menges, E., S. Crate, and P. Quintana-Ascencio. 2017. Dynamics of gaps, vegetation, and plant species with and without fire. American Journal of Botany 104:1825-1836. NatureServe. 2004. A habitat-based strategy for delimiting plant element occurrences: Guidance from the 2004 working group. Available http://www.natureserve.org/library/deliminting_plant_eos_Oct_2004.pdf. Accessed 12 January 2018. NatureServe. 2008. Ranking species occurrences- A generic approach. Available http://explorer.natureserve.org/eorankguide.htm. Accessed 13 February 2018. Nei, M. 1987. Genetic variation within species. In Molecular Evolutionary Genetics. Columbia University Press, New York. 512 pp. Pachauri, R.K., M.R. Allen, V.R. Barros, J. Broome, W. Cramer, R. Christ, J.A. Church, L. Clarke, Q. Dahe, P. Dasgupta, and N.K. Dubash. 2014. Climate change 2014: synthesis report. Contribution of Working Groups I, II and III to the fifth assessment report of the Intergovernmental Panel on Climate Change. IPCC. Quintana-Ascencio, P. F., E. S. Menges, C. W. Weekley, M. I. Kelrick, and B. Pace-Aldana. 2011. Biennial cycling caused by demographic delays in a fire-adapted annual plant. Population Ecology 53:131-142. Rahmstorf, S. 2007. A semi-empirical approach to projecting future sea-level rise. Science 315:368-370. Reynolds, M. H., P. Berkowitz, K. N. Courtot, and C. M. Drasue (eds.). 2012. Predicting sea- level rise vulnerability of terrestrial habitat and wildlife on the Northwestern Hawaiian islands. US Geological Survey Open-File Report 2012-1182. 139 pp. Richardson, D. 1985. Allelopathic effects of species in the sand pine scrub of Florida. Ph.D dissertation. University of South Florida, Tampa, FL. Semple, J.C. 1981. A revision of the golden aster genus Chrysopsis (Nutt.) Ell. nom. cons. (Compositae-Astereae). Rhodora 83:323-384.

St. John, T. 1987. Mineral acquisition in native plants. Pages 529-536. In: Elias, T. (ed.). Conservation and management of rare and endangered plants; Proceedings from a conference of the California Native Plant Society. California Native Plant Society. Sacramento, California. U.S. Fish and Wildlife Service (USFWS). 1988. Florida Golden Aster Recovery Plan. Atlanta, Georgia. 15 pp. USFWS. 2016. USFWS species status assessment framework: an integrated analytical framework for conservation. Version 3.4, August 2016. Xian, G., Crane, M. and Steinwand, D., 2005. Dynamic modeling of Tampa Bay urban development using parallel computing. Computers & Geosciences 31:920-928. Weekley, C. W., and E. S. Menges. 2012. Burning creates contrasting demographic patterns in Polygala lewtonii (Polygalaceae): a cradle-to-grave analysis of multiple cohorts in a perennial herb. Austratlian Journal of Botany 60:347-357. Willi, Y., J. Van Buskirk, and M. Fischer. 2005. A threefold genetic Allee effect: Population size affects cross-compatibility, inbreeding depression and drift load in the self-incompatible Ranunculus reptans. Genetics 169:2255-2265. Wolf. S., B. Hartl, C. Carroll, M.C. Neel, and D.N. Greenwald. 2015. Beyond PVA: why recovery under the Endangered Species Act is more than population viability. BioScience 65:200-207. Wunderlin, R., D. Richardson, and B. Hansen. 1981. Chrysopsis floridana. Status report prepared for U. S. Fish and Wildlife Service. Jacksonville Field Office, Jacksonville, Florida. Wunderlin, R. and B. Hansen. 2001. Guide to the Vascular Plants of Florida, Third Edition. University Press of Florida, Gainseville, FL. 783 pp. Wunderlin, R.P. and Hansen, B.F., 2011. Guide to the Vascular Plants of Florida, Third Edition. University Press of Florida, Gainseville, FL. 783 pp.

APPENDIX

Table A1. Resilience table with values for “current” population sizes used to assign populations to resilience categories. When available, data from historical surveys are provided, and are from a range-wide survey in 2006 in many cases or the number of introduced plants for recently introduced populations. These data are shown here, but growth rates from historical to current surveys were not used to assess resilience. In many cases, differences in population sizes were likely not a result of increasing populations, but rather of differences in survey methodology, number of surveyors, and/or areas searched.

Population EO # County Current Year Current Size Historical Year Historical Size

586 flowering Alafia River State Park 47 Hillsborough 2011 (actual)

> 3,000 1781 flowering + 424 rosettes Alafia Scrub Preserve 36 Hillsborough 2010 (estimate) 2006 (FNAI, estimate) 3 (includes 373 flowering + > 294 former EOs 19 rosettes +12 plants Balm Boyette Scrub Preserve and 41) Hillsborough 2006 (estimate)

100 Balm Boyette Scrub Preserve 18 Hillsborough 2008 (actual) 2005 > 100 (FNAI, estimate) 167 flowering 285 rosettes Balm Boyette Scrub Preserve 39 Hillsborough 2006 (actual) 528 flowering (actual) + Balm Scrub Preserve 40, 42 Hillsborough 2006 > 155 rosettes (estimate) 28 (Bell Bell Creek Preserve Creek), 2 2010 (Bell) > 2172 1215 flowering + 434 rosettes and Rhodine Scrub Preserve (Rhodine) Hillsborough 2008 (Rhodine) (estimate) 2006 (FNAI, actual)

Boyd Hill Nature Preserve 49 Pinellas 2010 > 500 (estimate) 400-600 flowering + rosettes Bullfrog Creek Scrub Preserve 30 Hillsborough 2006 (estimate) Duette Preserve 2015 665 plants North 54 Manatee (actual) 2013 156 plants (BTG, actual) Duette Preserve 751 plants South 53 Manatee 2015 (actual) 2013 165 plants (BTG, actual) 250 Fish Hawk 14 Hillsborough 2010 (estimate) 2005 100-150 (FNAI, estimate)

40 Fort De Soto County Park 56 Pinellas 2009 (estimate)

9, 38 (Golden 2008 2006 Golden Aster Scrub Preserve Aster) 2009 2006 1,067 flowering and I-75 exit for County Road 672 24 (1-75) Hillsborough 2010 1081 (estimate) 2004 (FNAI)

Highlands Hammock State Park 48 Highlands 2016 300 2014 23 plants (FNAI, estimate)

Johnston Road and Littell 550 Property 34 (Johnston) Hardee 2010 (estimate) 2006 (Johnston Rd) 13 (Johnston Rd) 1,257 flowering + rosettes Lake Manatee State Park 16 Manatee 2014 (actual) 2004 > 100 (FNAI, estimate)

Little Manatee River 5 Hillsborough 2017 25 (actual) 2004 > 15 (FNAI, estimate) Little Manatee River (Cordell) 12,743 plants Southfork East and West 51, 46 Manatee 2015 (actual) 2008-2009 1,455 plants (BTG,actual)

2832 flowering + 477 Little Manatee River State Park 31, 44 Hillsborough 2009 1787rosettes (actual) 2006 (FNAI, actual)

> 300 > 100 mature & rosettes (FNAI, Lucky Lonesome Trail 35 Hardee 2018 (estimate) 2006 estimate) 3,900 adult plants McKay Creek 52 Pinellas 2015 (actual) 2009 613 (BTG, actual) 2655 most flowering Moody Branch Mitigation Park 29 Manatee 2017 (actual) 2006 212 (FNAI) 1,732 plants Paynes Creek Historic State Park 50 Hardee 2015 (actual) 2011 199 (BTG, actual) 150 Stephens Road 5 Hillsborough 2010 (estimate) 663 Sun City Heritage Park Addition 6 Hillsborough 2017 (estimate) 2005 10-20 plants (estimate)

> 3,000 > 100 flowering and rosettes Sweetwater Road 13 Hardee 2010 (estimate) 2006 (FNAI, estimate)

< 25 Upper Little Manatee River 32 Hillsborough 2010 (estimate) 2005 20 plants (FNAI, estimate) > 2,000 Upper Little Manatee River 43 Hillsborough 2009 (estimate) 2004 or 2003 15 plants (FNAI) 5,034 plants Weedon Island 55 Pinellas 2015 (actual) 2010 922 (BTG, actual)

Table A2. Land cover (derived from 2010 aerial imagery) and NRCS soil types (2015 data) where C. floridana occurred in the spatial data set from an ongoing Bok Tower Garden/Archbold Biological Station/Service study, and Florida Natural Areas Inventory element occurrences with ‘High’ or ‘Very High’ spatial accuracy. The quality rank for land cover types was developed from expert opinion, where 4 indicates highest quality, 2 indicates lowest quality, and 1 indicates unsuitability. This data set includes populations successfully introduced into habitat selected for the plants by biologists.

Land cover site name Area in Data (m2) Land Cover Rank Soil component name Scrub 86197 4 ADAMSVILLE FINE SAND Scrubby Flatwoods 17724 4 ARCHBOLD FINE SAND Xeric Hammock 4391 3 ARENTS, NEARLY LEVEL Mesic Flatwoods 2688 2 BASINGER, HOLOPAW, AND SAMSULA SOILS, DEPRESSIONAL Sandhill 1182 2 CANDLER FINE SAND, 0 TO 5 PERCENT SLOPES Improved Pasture 4942 1 CASSIA FINE SAND Mesic Hammock 56 1 DUETTE FINE SAND, 0 TO 5 PERCENT SLOPES Mixed Hardwood-Coniferous 2269 1 IMMOKALEE FINE SAND Mixed Wetland Hardwoods 63 1 IMMOKALEE FINE SAND, 0 TO 2 PERCENT SLOPES Residential, Low Density 1800 1 IMMOKALEE SOILS AND URBAN LAND Residential, Med. Density - 2-5 Dwelling 63 1 JONATHAN SAND Units/AC Rural Open 1372 1 MYAKKA FINE SAND, 0 TO 2 PERCENT SLOPES Rural Open Forested 47114 1 MYAKKA FINE SAND, 2 TO 5 PERCENT SLOPES Shrub and Brushland 2662 1 MYAKKA SOILS AND URBAN LAND Transportation 3118 1 ORLANDO FINE SAND, 0 TO 5 PERCENT SLOPES Unimproved/Woodland Pasture 2274 1 PAOLA AND ST. LUCIE SOILS AND URBAN LAND Urban Open Forested 125 1 POMELLO FINE SAND, 0 TO 2 PERCENT SLOPES Urban Open Land 38 1 POMELLO FINE SAND, 0 TO 5 PERCENT SLOPES Wet Prairie 327 1 POMELLO SOILS AND URBAN LAND, 0 TO 5 PERCENT SLOPES SMYRNA SAND, 0 TO 2 PERCENT SLOPES ST. JOHNS FINE SAND ST. LUCIE FINE SAND, 0 TO 2 PERCENT SLOPES TAVARES SAND, 0 TO 5 PERCENT SLOPES