Introduction

The ghost orchid, Dendrophylax lindenii (Lindl.) Benth. ex Rolfe, is a rare epiphytic leafless orchid found in the warm, humid climate of south- west Florida and Cuba (^{GBIF 2023}). In Florida, where it is state listed as endangered, the ghost orchid grows naturally in Lee, Collier, and Hendry Counties (http://florida.plantatlas.usf.edu) where it inhabits cypress swamps and wet hammocks (^{Wunderlin & Hansen 2011}). In Florida, a diversity of tree species has been documented as hosts to ghost orchids including red maple (Acer rubrum L.), pond apple (Annona glabra L.), pop ash (Fraxinus caroliniana Mill.), oaks (Quercus species), royal palm (Roystonea regia (Kunth) O.F.Cook), and bald cypress (Taxodium distichum (L.) Rich.) (^{Brown & Folsum 2002}). In Florida, despite a diversity of substrate availability, the orchids are reported to be found primarily on pop ash, and typically less than 3 m from the soil surface on trunks and main branches of all host species (^{Mújica et al. 2018}).

Seed dispersal and subsequent establishment of mature individuals are important contributors to orchid distribution within habitats. Orchid seeds are typically small, produced in large numbers, and wind-dispersed (^{Arditti & Ghani 2000}, ^{Yoder et al. 2010}). Fruit production by the ghost orchid is variable, including some years when no fruit is produced, and has been reported to be negatively impacted by hurricanes (^{Mújica et al. 2013}, ^{Raventós et al. 2015a}). The number of seeds per fruit in orchids can be quite high (^{Arditti & Ghani 2000}); however, ghost orchid seed production has not been quantified. Orchid seeds typically lack endosperm (^{Arditti & Ghani 2000}), enabling long-distance dispersal but necessitating an external nutrient source for germination (^{Yoder et al. 2010}). Seeds of epiphytic orchids require a physical surface that prevents subsequent dislodging of the seed, retains sufficient moisture, and a substrate containing suitable mycorrhizal fungal species capable of providing carbon, nutrients and water (^{Callaway et al. 2002}, ^{Hoang et al.2017}, ^{Mújica et al. 2018}). Different tree species likely have differing amounts of suitable microhabitat that meet the criteria for successful colonization and establishment by epiphytic orchids. Such variation in suitable microhabitat among potential host species (^{Mújica et al. 2018}), along with variability in the distribution of those species on the landscape would influence and correlate with the distribution of specific epiphytic orchid species.

Compared to animals, plants receive relatively little attention when it comes to conservation and protection. This is likely a function of several factors, such as plant blindness, and conservation efforts focused on charismatic megafauna species (^{Balding & Williams 2016}). Orchids are a plant group analogous to charismatic megafauna and one could argue that ghost orchids could be the “charismatic megaflora” poster child. All plants, including orchids, perform important ecosystem functions and services that are often ignored, such as serving as an energy source for pollinators, and in the case of ghost orchids, serving as recreational and tourism resources. Human activities have significantly impacted orchid populations such as ghost orchids (^{Wraith & Pickering 2018}). Historically, this species has been heavily affected due to poaching by orchid enthusiasts, logging of old growth cypress forests, land use change, and altered hydrology and fire regimes in southern Florida. Looking into the future, it will be further impacted by changes resulting from anthropogenic climate change. Ghost orchids will unlikely be able to sustain their populations given all these pressures from human activities. If they are to remain on the landscape, there is a strong need for conservation efforts focused on protecting and enhancing their populations. To promote successful conservation and/or recovery of this and other orchid species, the integration of multiple areas of research - genetic diversity, propagation science, pollination biology, mycology, ecology and distribution, and reintroduction - are required (^{Stewart 2007}). While work has been done on the biology of the ghost orchid - exodermis structure and its relationship to mycorrhizal fungi (^{Chomicki et al. 2014}), desiccation tolerance (^{Coopman & Kane 2019}), greenhouse acclimatization methods (^{Coopman & Kane 2018}), and germination and seedling development (^{Hoang et al. 2017}) - until recently little was known about its ecology (but see ^{Mújica et al. 2018}, ^{Ray et al. 2012}, ^{Zettler et al. 2011}, ²⁰¹⁹). Recent ecological studies have primarily focused on the orchid’s population biology and pollinator ecology (^{Danaher et al. 2020}, ^{Houlihan et al. 2019}, ^{Mújica et al. 2021}, ^{Sadler et al. 2011}, ^{Zettler et al. 2019}). This work examined a population of ghost orchids in Fakahatchee Strand Preserve State Park (FSPSP) in Collier County in southwest Florida to determine if ghost orchids in FSPSP exhibit host preference, vertical stratification, substrate diameter stratification, and a distribution pattern similar to that of their host plants. The data presented here will aid in the long-term quantification of both spatial and temporal changes in this orchid population (^{Zotz et al. 1999}) and afford a better understanding of this orchid’s population ecology, providing valuable information to those making management and conservation decisions.

Materials and methods

Study site.- Fakahatchee Strand Preserve State Park (FSPSP), located in Collier County in southwest Florida (Fig. 1), experiences a humid, subtropical climate with a mild winter dry season and hot summer rainy season. Mean annual temperature is 23°C, ranging from a mean January temperature of 14°C to a mean August temperature of 28°C (^{Duever 1986}). Most of the region’s mean annual rainfall of 136 cm falls during the summer (^{Duever 1986}).

FSPSP comprises a complex mosaic of ecosystems, including a cypress strand that runs north to south through the center of the Preserve. The strand, regrown after extensive logging in the mid-20^th century and characterized by a 6 to 9-month hydroperiod (^{Ewel 1990}), is dotted with deep sloughs with a second growth bald cypress canopy and an understory dominated by pond apple and pop ash. Humidity from the deep waters of the central strand and the ponds within provide insulation for tropical epiphytic orchids during the infrequent cold events that occur in south Florida (^{Lodge 2016}, ^{Mújica et al. 2018}).

Data collection.- A line-transect of 90 m was established in a mixed bald cypress/pond apple/pop ash slough within FSPSP with a known population of ghost orchids. The geographic coordinates of this population are not described here due to the risk of poaching associated with the study species. Woody plant species within 15 m of the transect and with a dbh (diameter at breast height or 1.3 m) greater than 4 cm were identified, dbh recorded, and their locations mapped. The vertical stems of multi-stemmed trees, such as pond apple and pop ash were mapped individually, to provide a better indicator of habitat availability (^{Bennett 1986}). All ghost orchids were recorded, and orchid heights were quantified as above or below 3 m, based on a previous study (unpublished data) indicating the slough’s standing water provided thermal protection up to 3 m above the ground during irregular frost events. Ghost orchids co-occur with the leafless ribbon orchid, Campylocentrum pachyrrhizum (Rchb.f.) Rolfe, at this study site but were distinguishable bytheir significantly less compressed roots which have short white dashes that parallel the direction of growth.

Data analysis.- Likelihood Ratio χ² tests were conducted to determine if ghost orchids were randomly distributed among tree species, or at different heights, and on host plant stems of varying dbh. Pearson χ² tests were used when data were sparse (^{Sall et al. 2017}). We limited our diameter preference analysis to pop ash because most individual ghost orchids in this study were found on this host tree species. Because of its increased power compared to other indices (^{Goodall & West 1979}), the Hopkins Index of Aggregation (A) was computed for pop ash trees, individual pop ash vertical stems, and ghost orchids to determine whether their distributions on the landscape were significantly clumped (A greater than 1) compared to random (A equal to 1) and whether the dispersion pattern of the two species was different (^{Hopkins & Skellam 1954}). The Hopkins Index of Aggregation was computed in Microsoft Excel. All other statistics were conducted in R using package “RVAideMemoire” (^{Hervé 2019}) and in SAS-JMP^® Version 14.0 (SAS Institute Inc., Cary, NC).

Figure 1 Location of Fakahatchee Strand Preserve State Park in Collier County, Florida (^{Thomas 2009}). The state of Florida is in yellow with county boundaries in black. The insets are of Collier County, Florida in blue and the Fakahatchee Strand Preserve State Park in green. 

Results

A total of 695 stems were recorded among 13 woody plant species. Fraxinus caroliniana stems totaled 33.5% of all stems measured, bald cypress totaled 25.6%, and pond apple totaled 20.3% (Table 1). Laurel oak exhibited the highest mean dbh (28.7 cm) of all woody plant species while arthritis vine (Hippocratea volubilis L.), a woody vine, had the lowest (4.6 cm) (Table 1).

Host species preference.- Twenty-five ghost orchids were found on three host plant species - 20 (80%) on pop ash, four (16%) on arthritis vine, and one (4%) on pond apple (Table 2). Although the sample size is small, by some estimates, it could represent over 1% of the entire Florida population (^{Zettler et al. 2019}). Given that most ghost orchids occurred on pop ash (20), we used a likelihood ratio χ² to determine whether this frequency was statistically greater on pop ash than other species based upon the host plant species composition of the forest (Fig. 2). This analysis indicated a statistically significant occurrence of ghost orchids on pop ash relative to all of the other species in the study area (χ² = 30.3, p<0.01).

Figure 2 Woody plant species composition of the forested study site and the composition of the ghost orchid host species, calculated as a percentage of all stems. 

To determine if the distribution on pop ash was greater than expected for just the species it occurred on in this study, the frequencies of D. lindenii were compared against the relative frequencies of only pond apple, pop ash, and arthritis vine. The Pearson χ² test indicated significantly more ghost orchids on pop ash and less on pond apple compared to what would be expected given their relative frequencies in the study area (χ² =9.9, p<0.01). However, comparing the frequencies of ghost orchids against the relative basal areas of the host plant species on which it occurred, arthritis vine, representing only 2% of the relative basal area of host species, accounted for significantly more of the ghost orchid occurrences (16%) than would be expected (χ² =14.6, p<0.001). Pond apple accounted for fewer than expected ghost orchids compared to its relative basal area contribution of 17.5% and pop ash accounted for 80% of the ghost orchids, matching its 80% contribution to the relative basal area of the species upon which ghost orchids were growing.

Vertical stratification.- Fifteen ghost orchids (60%) were found growing below 3 m and 10 (40%) were growing above 3 m (Table 2). A likelihood ratio χ² was used to determine if this differed significantly from a 50:50 ratio. The data in this study do not indicate a statistically greater number of ghost orchids occurring <3m from the forest floor compared to above that height (χ² =1.0, p=0.32).

Substrate diameter stratification.- To assess host plant diameter preference, we divided all pop ash stems into five dbh classes, each containing approximately equal numbers of stems (Table 3). Most ghost orchids growing on pop ash were found on stems between 14.1 cm and 17.2 cm dbh (Table 3). The χ² test indicated a weakly significant deviation from a random distribution of ghost orchids across the five dbh categories (χ² =9.07, p=0.06).

Dispersal.- The spatial distribution of individuals of both ghost orchids and pop ash were assessed in two separate ways. First, the Hopkins Index of Aggregation (A’) was computed for the locations of the pop ash trees, the individual pop ash stems, and the ghost orchids. The data in this study indicate the trees and stems of pop ash and the epiphytic ghost orchids demonstrate departures from random dispersal on the landscape (Table 4) with all of them being significantly clumped (p<0.01). The pop ash trees showed the lowest Index of Aggregation (3.0), and ghost orchids had the highest (9.0).

Additionally, we tested whether the presence of ghost orchids on a host plant significantly increased the probability of having multiple ghost orchids (13.0%) present on the same host compared to the probability of any stem in the study being host to at least one ghost orchids (2.9%) or not (97.1%). The Pearson χ² test on these data indicates a significantly increased probability of multiple individuals of ghost orchids occurring on a host plant if there is at least one individual present (p=0.004). This test is further evidence of increased clumping in ghost orchids compared to its host plants.

Discussion

Host species preference.- Pop ash served as the primary host for ghost orchids growing in a mixed cypress strand in FSPSP, as has been reported in nearby Florida Panther National Wildlife Refuge (^{Coopman & Kane 2019}, ^{Mújica et al. 2018}). Alternatively, ^{Sadler et al. (2011}) reported pond apple as the predominant host species in another ghost orchid population in Collier County. This work documents a greater utilization of pop ash than previously reported (^{Mújica et al. 2018}) and describes arthritis vine, a woody vine, as a new host species for ghost orchids in south Florida. Further, this work examined the distribution of ghost orchids in relation to the relative proportion of the woody plant species in the forested transect.

Although there were more pop ash stems within the transect than the other woody species, ghost orchids were not randomly distributed among all species, or randomly distributed among the stems of host species at the site. For example, pond apple made up 20.3% of all stems, but only one ghost orchid (4% of total individuals) was found on this species. In contrast, bald cypress made up 25.6% of all stems, but hosted no ghost orchids. The orchid’s distribution is likely explained by the synergy of several factors influencing host preference in epiphytes. However, as suggested by ^{Zettler et al. (2019}) this preference for pop ash by the ghost orchid must also be considered in the context of the invasion of the exotic emerald ash borer as a threat to this preferred host.

Pop ash and pond apple are similar in architecture, often producing multiple stems per tree, while the other woody species in the study site tend to produce one main stem per tree. Multiple stems from the same root system would create proximal suitable habitat for wind-dispersed orchid seeds from parent plants. Although similar numbers of stems existed for pop ash and pond apple (Table 1), pond apple only served as host to one ghost orchid.

Age of the host tree may also influence host preference (^{Benzing 1980}). Successful pollination of ghost orchids is rare (^{Luer 1972}) as is successful seed germination (^{Zettler et al. 2011}). In addition, epiphytic orchids are slow-growing due to the resource-limited environments in which they live (^{Laube & Zotz 2006}). Therefore, an older substrate may have an increased probability of hosting ghost orchids than a younger substrate. For example, ^{Thomas (2009}) found that pop ash growing in another slough in FSPSP were older than bald cypress in the same slough by 30-40 years, likely due to successional processes during recovery following logging in the 1940s and 1950s.

No data were collected on bark roughness for this work, although bark texture is known to affect host preference in epiphytes (^{Benzing 1990}) and is suggested as critical to ghost orchids (^{Zettler et al. 2019}). A rough surface creates cracks and fissures where moisture can potentially be retained and seeds lodge until conditions are right for germination (^{Callaway et al. 2002}, ^{Hoang et al. 2017}, ^{Mújica et al. 2018}). ^{Zarate-García et al. (2020}) more specifically suggest bark water storage linked to microrelief characteristics influences phorophyte preferences of orchids. ^{Mújica et al. (2018)} characterize pop ash and pond apple bark as corrugated, with crevices between 0.5 and 2.0 cm deep, and suggest bald cypress has “smoother bark” than these species. ^{Brown (1984, p.18} ) describes bald cypress bark as “…an interwoven pattern of narrow flat ridges and narrow furrows,” which may suggest bald cypress as a suitable substrate for ghost orchid seeds. However, unlike pop ash and pond apple, bald cypress sheds its bark, perhaps explaining why no ghost orchids were found on this potential host species.

Recent research emphasizes the role of mycorrhizal fungi in the life cycle of epiphytic orchids. Orchids require these fungi to stimulate germination (^{Yoder 2010}). ^{Rasmussen (2002}) and ^{Gowland et al. (2011}) suggest host preference may be related to the mycobiont found on a host species. A strain of Ceratobasidium grows symbiotically with adult ghost orchids in Florida (^{Mújica et al. 2018}) and improves germination and seedling development (^{Hoang et al. 2017}). The presence of this strain or other fungal symbionts may be necessary for establishment of ghost orchids (^{Mújica et al. 2018}, ^{Hoang et al. 2017}). Suitability of host plants for ghost orchid establishment therefore may be determined partially by the fungal species present on the hosts. The endophytic fungal strains studied by ^{Hoang et al. (2017)} were collected from a ghost orchid population in the Florida Panther National Wildlife Refuge in Southwest Florida. ^{Hoang et al. (2017)} did not specify the host plants supporting the individual ghost orchids that they sampled but reported that most epiphytic orchids in their study area grew on pop ash and pond apple. While explored to some extent by ^{Johnson (2019}), the importance of fungal symbionts to the establishment of ghost orchids and other epiphytic orchids warrants further investigation into the spatial distribution and potential host plant preference of these fungal species.

Vertical stratification.- We anticipated a significant tendency for the orchid to occur below 3 m, as has been reported (^{Mújica et al. 2018}). Although most of the individuals we located were below 3 m (15 of 25), the trend was not significantly different from a random distribution above and below 3 m. Canopy height was not assessed in this study but may significantly impact vertical stratification of ghost orchids because of differences in solar radiation and vapor pressure deficit. Future investigation of the role of canopy height on orchid vertical stratification across populations from different locations would provide valuable insights into the observed differences across studies. It is difficult to quantify the available space for potential colonization, given the variable stem diameter and height of potential host species and the complex surfaces of these stems at greater height with numerous branches. It is possible that ghost orchid occurrence at higher positions in the canopy may be under- reported as seeing this leafless orchid in upper canopy levels would be difficult when not in flower. Alternatively, many epiphytes are sensitive to higher light levels found in higher positions in the canopy (^{Laube & Zotz 2006}). In addition, a pilot study indicated standing water in the swamp provides thermal protection of up to 2^oC as high as 3 m above the ground surface from irregular frost events (unpublished data). It may be that historical vertical distributions were driven in part by cold events, but these events are likely to decrease in the future due to climate change. ^{Mújica et al. (2013)} found no differences in the mean height of surviving and non-surviving ghost orchid individuals following Hurricane Ivan in Cuba, suggesting that hurricane damage may not differentially affect individuals based on vertical position on the host plant. However, continuing to study the vertical position of ghost orchids in the canopy will be important as climate change can alter the occurrence of both low temperature events and hurricanes, particularly in landscapes impacted by human activities that alter hydrology.

Substrate diameter stratification.- The distribution of ghost orchids among pop ash stems of varying diameter was not random. Rather, ghost orchids exhibited a weak preference for larger stems, specifically those between 14.1 and 17.2 cm dbh. ^{Abe et al. (2018}) found a similar trend among the Okinawa dendrobium (Dendrobium okinawense), an epiphytic orchid from Japan. Larger, and hence older, trees often have rougher bark, creating crevices for orchid seeds, symbiotic fungi, and sufficient moisture to support them both. In addition, older trees would give the slow-growing orchids a longer opportunity to establish and grow.

We did not collect data on the size of the ghost orchid individuals. However, individuals were found on the smallest size class of pop ash (<9.2 cm dbh) suggesting relatively recent establishment, possibly indicating this population is not ‘senile’ (^{Zettler et al. 2019}) but may be actively recruiting.

Dispersal.- In terms of aggregation of ghost orchids, the data from this study indicated individuals of their primary host species, pop ash, have a clumped spatial arrangement on the landscape. As would be expected, the stems of pop ash with ghost orchids appear to be even more clumped. Such a spatial arrangement on the landscape may be a function of dispersal (e.g. ^{Trapnell et al. 2013}) or appropriate microhabitat conditions during establishment. The clumped arrangement of pop ash likely is contributing to the clumping observed in ghost orchids. However, our analysis suggests additional dispersal and/or microhabitat limitations on ghost orchids may be further influencing its dispersion pattern. Many variables determine the suitability of a host plant species, including bark stability and texture, age and architecture, and bark chemistry (^{Migenis & Ackerman 1993}, ^{Zettler et al. 2019}). In addition, abiotic and biotic factors including fungal associations, temperature, irradiance, relative humidity, water availability, wind speed and direction during seed dispersal events, may collectively contribute to the clumping patterns observed in this species, and all warrant further investigation.

Establishment of new populations may be driven primarily by a few founding individuals who later reproduce, with nearby establishment of offspring, or by the concurrent arrival and establishment of many founding individuals. ^{Trapnell et al. (2013}) inferred the former pattern in the epiphytic Lady of the Night orchid, Brassavola nodosa (L.) Lindl. It is not known which pattern is typical of ghost orchid populations. However, the population dynamics of ghost orchids suggest that recruitment of new individuals, along with reproductive success, is especially important in maintaining a stable population (^{Raventós et al. 2015b}). Population decline has been reported in ghost orchid populations affected by hurricanes (^{Mújica et al. 2013}, ^{Raventós et al. 2015a}), which are predicted to become more intense under climate change projections (^{Knutson et al. 2010}). Hurricane events may negatively impact host plants, and they can influence orchid-pollinator species interactions (^{Ackerman & Moya 1996}, ^{Mújica et al., 2018}, ^{Zettler et al. 2019}). However, unpublished data from Fakahatchee Strand Preserve State Park suggests increased seed capsule production following hurricane disturbance. This dichotomy may suggest the response of this species varies according to hurricane intensity and ecosystem structure. Therefore, an improved understanding of host plant preference, microhabitat requirements, disturbance impacts and distribution of ghost orchids is crucial for the conservation of this charismatic epiphytic orchid.

Conclusion

Conservation of the ghost orchid in southern Florida must include both protection and additional re- search to better understand its ecology. This work adds to the understanding of ghost orchid spatial distribution and suggests potential drivers of that distribution. Long-term monitoring of this population should be continued to quantify spatial and temporal changes, particularly in a time of changing climate. However, due to the risk of poaching, this should be done with the utmost of discretion. At the time of this writing, several of the orchids within this population have been removed by humans since the fieldwork was completed.