Conservation Status and Management Insights from Tracking a Cryptic and Critically Endangered Species of Orchidaceae

Conservation status and management insights from tracking a cryptic and Critically Endangered species of Orchidaceae

T. KRAAIJ,J.A.BAARD and B . J . C RAIN

Abstract Ninety-five percent of orchid species associated To view supplementary material for this article, please visit with fynbos shrublands of South Africa’s Cape Floristic http://dx.doi.org/./S Kingdom have been assessed for the IUCN Red List, yet aspects of their demography and population biology remain poorly understood. We conducted a -year demographic study of the Critically Endangered Disa procera, a cryptic, terrestrial species from South Africa with a global popula- Introduction tion of c.  individuals known from a single location. We he Cape Floristic Kingdom in South Africa is a global aimed to provide management recommendations that biodiversity hotspot (Myers et al., ) that comprises would facilitate its persistence. Our findings indicate that T one of the most unique and threatened orchid floras the population of D. procera is larger than previously (McDonald, ; Linder et al., ). Endemism and rarity thought, and the species occurs at two distinct locations. are common amongst Orchidaceae of the region (Stewart These orchids exhibit high interannual variation in popula- et al., ; Linder et al., ; Liltved & Johnson, ); tion size and turnover of individuals, potentially indicative % are endemic and % are near-endemic (Liltved & of a species with a short life span, and still meet the criteria Johnson, ). Accordingly,  of the  orchid species as- for Critically Endangered status. The species benefits from sociated with fynbos, a primary vegetation type of the re- disturbances, such as brush cutting along trails, or fire, gion, are on the IUCN Red List (Raimondo et al., ). which open up clearances in the vegetation. However, phys- Nonetheless, relatively little is known about the fundamen- ical damage to plants during their aboveground growing tal ecology of many of the region’s orchid species, and with- season (September–January) is particularly detrimental out this information it is difficult for conservation managers and should be avoided in habitat management for the spe- and policy makers to understand how these species will re- cies. Fire had beneficial effects at the population and indi- spond to threats, and how to make effective decisions when vidual levels and is recommended at –-year intervals, working to protect them. outside the orchid’s growing season. The species exhibited Although for some orchid species rarity and endemism comparatively high rates of fruit set (%), suggesting that may be attributable to natural rarity (Stuckey, ), threats pollination limitation does not currently constrain its per- such as habitat loss or degradation, invasive exotic plants, formance. Its patchy distribution may, however, indicate illegal harvesting, and inappropriate burning regimes constraints on dispersal or recruitment. We recommend (mostly fire suppression) also play critical roles in the that management strategies should include continued pro- Cape Floristic Kingdom and elsewhere (Duncan et al., tection and monitoring of both populations, studies on pol- ; Raimondo et al., ; Bytebier et al., ; lination, habitat requirements and mycorrhizal associates, Whitman et al., ; Crain & Tremblay, ). In temperate and a prescribed disturbance regime. areas such as South Africa, the USA and Australia, for ex- Keywords Cape Floristic Kingdom, Critically Endangered ample, appropriate burning regimes are particularly import- species, Disa procera, fire regime, fynbos shrubland, man- ant for many orchid populations, as regularly occurring fires agement strategies, non-destructive sampling, population are known to be beneficial for a number of species in terms biology of reproduction and population growth (Bowles, ; Bytebier et al., , ; Coates & Duncan, ; Lamont & Downes, ). Fire is known to have a positive

T. KRAAIJ (Corresponding author) Nelson Mandela Metropolitan University, effect on orchid populations, as it removes competing vege- School of Natural Resource Management—Nature Conservation, Private Bag tation, releases nutrients and reduces harmful fungi and X6531, George, 6530, South Africa. E-mail [email protected] other pests and pathogens (Stuckey, ; Stewart et al., J. A. BAARD South African National Parks, Garden Route Scientific Services,    Knysna, South Africa ; Coates & Duncan, ; Liltved & Johnson, ). In South Africa several species of Disa P.J. Bergius orchids, a B. J. CRAIN Department of Ecology and Evolutionary Biology, University of Tennessee, Knoxville, USA charismatic group from the region, reproduce only after  Received  July . Revision requested  August . fires (Bytebier et al., ). Consequently, it is critical that Accepted  February . First published online  June . the effects of fire are assessed when making management

Oryx, 2017, 51(3), 441–450 © 2016 Fauna & Flora International doi:10.1017/S0030605316000272 Downloaded from https://www.cambridge.org/core. IP address: 170.106.202.8, on 26 Sep 2021 at 11:37:02, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0030605316000272 442 T. Kraaij et al.

decisions for rare and threatened orchid species in places slender, delicate and grass-like in appearance) makes vege- such as the Cape Floristic Kingdom. tative individuals practically indistinguishable from sympat- Despite the high level of rarity and threat pertaining to ric grasses and sedges when flowers are not present. orchids of the Cape Floristic Kingdom, including many spe- Furthermore, individuals are difficult to mark because cies of Disa, aspects of their demographics and the driving handling or tagging them can cause damage (particularly forces behind them are little studied and are mostly inferred to underground portions). The species’ transient nature pre- from analyses undertaken elsewhere (e.g. Stuckey, ; sents a considerable challenge to its study. Mehrhoff, , ; Hutchings, a,b). In terms of Monitoring of D. procera is therefore confined to flower- southern African orchids in general, the published literature ing individuals, in which form the species is identifiable and focuses on topics such as taxonomy and biogeography distinctive from congeners. Prior to initiation of the study, (Johnson & Bond, ; Steiner et al., ; Linder et al., surveys conducted by a group of volunteer citizen scientists ; Liltved, ; Bytebier et al., ; Johnson et al., experienced in plant identification (the Custodians of Rare ), and comprehensive information on the population and Endangered Wildflowers) indicated that the total popu- biology of many species is lacking. However, a sound under- lation of D. procera comprised c.  flowering individuals standing of the population biology of rare or threatened spe- during any particular growing season. All individuals oc- cies is a requisite for most conservation management efforts curred in – sub-locations (depending on the survey (Dixon & Cook, ), and a lack thereof is considered a year), most of which were associated with some form of dis- threat to a species’ persistence (Raimondo et al., ). turbance (e.g. trails and clearing) that reduced aboveground Accordingly, we studied the population dynamics and vegetation cover. habitat attributes of Disa procera H.P. Linder, a Critically Given the rare and threatened status of D. procera, the Endangered terrestrial orchid associated with fynbos shrub- primary objective of our study was to perform an up-to-date lands along the southern Cape coast of South Africa, to ob- survey of the species, to evaluate its demographic attributes tain information critical for the development of an effective and trends, and to determine how it is affected by habitat conservation strategy for the species. At the last assessment management. We were particularly interested in exploring of the species (Von Staden & Turner, ) D. procera was how fire may influence the health and recovery of the spe-  known from only a single location ,  km in extent and cies, given ongoing changes in natural burning regimes had a global population of c.  individuals. The main pres- where it occurs. Accordingly, our specific aims were to () sures leading to the decline of this species have been habitat record the current size of the population of D. procera, transformation as a result of coastal development, agricul- and assess interannual variation in the number of indivi- tural expansion and plantation forestry (Von Staden & duals and overall growth rates, () quantify measures of Turner, ). Furthermore, the population continues to plant vigour and fecundity as indicators of population per- face threats from inappropriate vegetation management ac- formance, and assess if these measures vary through time, tions (e.g. a lack of, or excessive, disturbance from fires, her- and () examine the effects of fire on plant vigour and fe- bivores and brush-cutting along roads and trails; illegal cundity as well as on population growth rates and persist- collection; and invasive alien plants; TK, JB, pers. obs.). ence probabilities. Ultimately we endeavoured to offer Consequently, the species is categorized as Critically management recommendations that would facilitate opti- Endangered on the IUCN Red List, based on its limited ex- mal population performance and thus perseverance of this tent of occurrence and continuing declines in its area of oc- Critically Endangered species. cupancy and in the number of mature individuals in the population, and because it occurs at only a single location     (IUCN criteria B. .a & B. .b(iii, v), B. .a & B. .b(iii, v), Study area and C..a(ii); Von Staden & Turner, ). Too little is known about the ecology of D. procera to know what con- Disa procera occurs along the southern Cape coast of South stitutes optimal environmental conditions for its persistence Africa. Its exact locality is not disclosed here, to prevent un- and proliferation. scrupulous visitation, disturbance and collection of plants. Disa procera is a slender, tuberous geophyte with narrow, Vegetation at the study site comprises southern Cape linear, upright leaves arranged in a rosette. Individual plants dune fynbos, a mixture of sandplain fynbos and subtropical can reach c.  cm in height. Reproductive individuals have thicket (Mucina & Rutherford, ). Fragmentation and cerise flowers arranged in a lax raceme (Kurzweil, ). The isolation of remaining undisturbed vegetation in the region aboveground vegetative growing season commences in early and suppression of natural fires in the surrounding land- spring (September), flowering occurs during October– scape have resulted in altered patterns of fire ignition and November, and capsules develop during November and spread, and a general lack of fire at the study site (Kraaij early December. By January, plants have dried out and are et al., ). The agency currently managing the study area no longer detectable. The species’ cryptic stature (it is does not routinely implement prescribed burning but

undertakes trail maintenance approximately every  One month later, during fruiting periods, we revisited months, during which time brush cutting occurs. Thus, each plant and recorded the number of developed seed cap- the only known population of D. procera is confined to a sules. With this information we quantified the rate of fruit site that is regularly altered and disturbed by human set per plant (calculated as the percentage of flowers devel- activities. oping into capsules). To understand causes of reproductive failure we distinguished cases of physical damage to plants resulting in severance of the entire raceme and thereby pre- Methods venting fruit set (referred to as damaged plants) from failure to set seed in intact plants (probably indicative of pollin- We focused our survey efforts on previously identified sub- ation failure). Damage to plants may have resulted from nat- locations to cover the entire known flowering population, ural factors (e.g. animal activity), flower picking by people, but each year the authors and the group of volunteer citizen or trail maintenance (in the form of brush cutting) carried scientists conducted additional searches for flowering indi- out during the species’ aboveground growth phase. We viduals outside the known occurrences. We documented the therefore evaluated each individual for signs of physical entire detectable flowering population each year during damage resulting in raceme removal by means other than –, both during the peak flowering season (the normal senescence. We used Spearman rank correlations end of October) and at the culmination of fruit development to investigate relationships between the mentioned mea- (the beginning of December). At each sub-location where sures of fecundity, and we used Mood’s median test to detect multiple plants occurred together we recorded their exact differences in these measures among years, given that the locations as distances at right angles to, and along, perman- data for these measures were not normally distributed. For ently marked line transects. Where individuals occurred in comparisons of fecundity measures between two groups (e.g. groups of fewer than five we marked plants with metal pins damaged vs undamaged plants) we used Wilcoxon tests.  cm away from their stems in a specified compass direc- To assess the effects of fire, given that the vegetation had tion. Geographical coordinates were recorded for all line not burnt in over  years and had become moribund, pre- transects and for individual plants marked with metal scribed burning was undertaken in a portion of the study pins. By marking the flowering individuals we were able to site during February and April  in an attempt to stimu- relocate them during the subsequent fruiting period or dur- late growth of the study species. This facilitated the best pos- ing the next annual survey if they flowered again. sible assessment of the effects of fire on the population To assess turnover in individuals among years we calcu- because the limited population size, the species’ small area lated Sørensen similarity coefficients (Sørensen, ) for of occurrence, and its zero tolerance for mortality resulting the population for all combinations of years. We also calcu- from sampling precluded more extensive or replicated ex- lated the growth rate of the flowering population (λ) each perimentation. To determine if the burn had an effect on year as the number of flowering individuals recorded during population size, we compared the numbers of flowering  the survey year (Nt+) divided by the number of flowering plants in the burnt area during the post-fire years with individuals recorded during the previous survey (Nt) the numbers in the unburnt area during the same period, (Morris & Doak, ). We then calculated the mean expressed as proportions of the mean population sizes in λ  χ growth rate ( G) as the geometric mean of the annual each of the two areas during the pre-fire years. A test rates. We used the values of λ calculated for each year to was used to compare these relative population sizes in the project the sizes of flowering populations over the next  burnt and unburnt areas during the post-fire years (– years. To accomplish this we multiplied the size of flowering ), correcting for the effect of damage by other causes. population by randomly selected values of λ each year for Furthermore, we calculated annual growth rates (λ), as be- a period of  years. We performed , iterations of fore, for the subsets of the flowering population that were this process and calculated the mean flowering population burnt and unburnt so that the growth rates in each group size and the standard deviation for each year of the simula- could be compared. We also projected the mean population tion. Furthermore, for each year we calculated extinction sizes and standard deviations over  years for each group. probabilities as the number of iterations out of , in Likewise, we calculated annual extinction probabilities for which the flowering population size fell below zero. flowering plants in each group. Demographic analyses were conducted using the PopTools To investigate the effects of fire on measures of plant vig- add-on in Microsoft Excel (Hood, ). our and fecundity (plant height, number of flowers, and During surveys conducted in peak flowering periods we number of capsules) we employed a Wilcoxon test, compar- also recorded a number of physical attributes of each indi- ing pre- and post-fire plant performance in the burnt and vidual. We measured the aboveground height of each flow- unburnt areas, respectively. Statistical analyses were per- ering plant, and recorded the total number of flowers on formed in StatGraphics Centurion XVII (Statpoint each plant. Technologies Inc., Warrenton, USA).

FIG. 1 Total numbers of Disa procera plants recorded per annum FIG. 2 Projected trend of the entire flowering population of Disa during –, categorized according to whether the plants procera over  years. The starting population size was based on were undamaged and produced seed capsules, undamaged but the total number of flowering individuals recorded during the failed to produce seed capsules, and damaged after flowering, final survey of this analysis, in . Projected population sizes and thus unable to produce seed capsules. (± SD) are based on mean values from , iterations of population projections calculated using randomly selected values of λ, which were calculated from observed annual growth rates Results in the population. During the study period we recorded  occurrences (plant × year combinations) of D. procera, comprising  expected to increase over the next  years (Fig. ). individuals (Supplementary Fig. S). At the original site Nevertheless, stochasticity in the projected population where D. procera was known to occur the annual flowering sizes as a result of variation in annual growth rates suggests population size was – individuals (coefficient of vari- that the population also has significant potential to reach ation among years = %; Fig. ). The median annual flow- zero flowering individuals in ,  years. ering population was  individuals. The smallest numbers In terms of attributes of the plants themselves, median of individuals were recorded during  and , both plant height during the study period was . cm (range years in which trail maintenance was done during the spe- .–. cm, n = ). Plant height was relatively stable cies’ aboveground growth phase (see Methods). A second among years, except in  and , when plants were sig- population of c.  individuals was discovered by volun- nificantly smaller than in most other years (Fig. ; Table ). teers in a suburban area a few kilometres from the study The median number of flowers per plant was  (range –, site. As this population was discovered in the final year of n=; Fig. ) and did not differ significantly among years the study, however, no physical or demographic data (Table ). A third of all recorded plants were damaged after could be collected from these individuals and they were the survey of flowering, and were thus unable to produce not included in further analyses presented here. capsules. Damaged plants (n = ) did not differ signifi- Turnover in flowering (i.e. detectable) individuals was cantly from undamaged plants (n = ) in terms of height high among years (Supplementary Fig. S), with only four or the number of flowers per plant (Table ). The median plants encountered four times during the -year study per- number of seed capsules per undamaged plant was seven iod,  plants encountered three times, and  plants en- (range –,n=) and did not differ significantly countered twice. The mean similarity in individuals among years (Fig. ). The median proportion of capsule among all years (i.e. the likelihood of a plant being re- set (percentage of flowers developing into capsules per un- encountered in any other year) was %(n=). The damaged plant) differed significantly among years, with mean similarity between populations occurring in consecu- measures in  and  significantly lower than mea- tive years was %(n=), between those occurring  years sures in ,  and  (pairwise comparisons lacked apart %(n=), and between those occurring  years apart the power to identify significant between-group differences) %(n=). Therefore the likelihood of a plant reappearing (Fig. ). In considering the total reproductive output of the declined with increasing time after first detection. population over the study period, %( of ) of un- In terms of population growth the overall number of damaged plants produced at least one capsule, and %of flowering individuals increased in the time between the ini- all flowers of undamaged plants developed into capsules.   ,   tial survey and the final survey (Fig. ). Annual population The number of flowers (rs = . ,P . ) and cap-   ±   −   ,   sizes (mean . SD . , range ) and annual popu- sules (rs = . ,P . ) per plant were both significantly λ   ±    −  lation growth rates ( G = . SD . , range . . ) correlated with plant height, and the number of capsules were highly variable, however. The population projections was significantly correlated with the number of flowers   ,   indicate that on average the overall flowering population is (rs = . ,P . ). The proportion of fruit set was not

On the burnt site, projections of population sizes after  years were smaller when based on observed growth rates of plants before the burn (mean =  ± SD ; Fig. a) than when based on observed growth rates of plants after the burn (mean = , ± SD ,; Fig. b). Conversely, on the unburnt sites, population projections were larger when based on observed growth rates prior to the burn (mean = , ± SD ,; Fig. a) than after the burn (mean =  ± SD ,; Fig. b). Thus, fire had a positive effect on predicted population sizes for the next  years. Analyses of extinction probabilities for flowering individuals supported the results of the population projections. Projections for flowering plants on the burnt site based on observed annual growth rates from pre-burn years indicated there was a significant probability of reaching  individuals within  years (. %of, iterations; Fig. a), whereas projections based on observed annual growth rates from post-burn years showed markedly reduced probability of reaching  flowering individuals within  years (%of , iterations). Conversely, the population on the unburnt site had a low probability of reaching  flowering individuals within  years, based on projections using pre-burn growth rates (, %of, iterations), whereas extinction probabilities increased (. %of, iterations) when using growth rates from post-burn years (Fig. b). Fire also had positive effects on some, but not all, measures of plant vigour (Table ). Fire resulted in a significant increase inmedianplantheightintheburntarea(preburn.cm,post burn . cm), with a concomitant decrease in median plant height in the unburnt area (pre burn . cm, post burn . FIG. 3 Annual medians of (a) plant height (all plants), (b) cm).Furthermore,themediannumberof flowers per plantde- number of flowers per plant (all plants), (c) number of capsules clined in the unburnt area between the pre-burn (. flowers) per plant (undamaged plants only), and (d) percentage capsule and post-burn (.) periods but did not change significantly set (percentage of flowers developing into capsules per  undamaged plant). Error bars indicate % confidence limits (Table ) over the same period in the burnt area (pre burn based on Mood’s median test order statistics (Table ). .,postburn.). No effect was detected in terms of the number of capsules produced per plant. Nevertheless, plants   significantly correlated with plant height (rs = . , on the burnt site were larger and produced more flowers   –  P= . ) or number of flowers per plant (rs = . , after the burn treatment than those on the unburnt site, and P=.). thus fire appeared to be beneficial to D. procera. Results from the post-burn surveys indicated that fire had an effect on several of the variables measured. We recorded an increase in the population size of orchids in the Discussion burnt area (Fig. ). During the first  post-fire years the number of plants in the burnt area as a proportion of that Disa procera is a Critically Endangered, cryptic and transi- prior to the burn was significantly higher than expected ent orchid species, and therefore we employed non- based on a comparison with the unburnt area (Table ). destructive methods to mark and monitor these plants Fire also corresponded with an increase in observed over successive years to help improve management efforts population growth rates: mean rates increased in the burnt for the only known extant populations. Given the overall λ   ±   population after the fire ( G = . SD . pre burn; rarity of the species, very little disturbance from survey λ   ±   G = . SD . post burn). In contrast, population growth and sampling techniques could be permitted. The methods rates decreased in the unburnt population after fire employed facilitated evaluation of the current status of the λ   ±   λ   ±   ( G = . SD . pre burn; G = . SD . post burn). species, and projection of its future demographic trends. Population projections for plants on burnt vs unburnt Our results confirmed that the study species has a di- sites supported results demonstrating the benefits of fire. minutive population size but revealed that it comprises

TABLE 1 Test results of the effects of year (–), damage from various causes, and fire (in early ) on various measures of Disa procera.

Measure Effect Test Test statistic Direction of effect Plant height Year Mood’s 30.5*** See Fig. 3 Number of flowers per plant Year Mood’s 5.6 Number of capsules per undamaged plant Year Mood’s 4.1 Rate of capsule set Year Mood’s 91.5*** See Fig. 3 Plant height Damaged vs undamaged Wilcoxon 9,102.0 Number of flowers per plant Damaged vs undamaged Wilcoxon 9,477.0 Number of plants during post-fire years 2012–2014 Burnt vs unburnt area χ2 320.6* Burnt . unburnt Plant height in unburnt area Pre fire vs post fire Wilcoxon 4,688.0** Pre fire . post fire Plant height in burnt area Pre fire vs post fire Wilcoxon 1,343.0* Post fire . pre fire Number of flowers per plant in unburnt area Pre fire vs post fire Wilcoxon 4,641.0** Pre fire . post fire Number of flowers per plant in burnt area Pre fire vs post fire Wilcoxon 1,856.5 Number of capsules per plant in unburnt area Pre fire vs post fire Wilcoxon 1,988.5 Number of capsules per plant in burnt area Pre fire vs post fire Wilcoxon 538.5

Significantly different at *, P , .; **, P , .; ***, P , .

FIG. 4 The annual population size of Disa procera in the areas burnt and unburnt in the  fire. Population size is expressed as a percentage of the mean number of plants occurring during the pre-fire years – in the burnt and unburnt areas. An estimated correction (based on field notes of plant locations where there was noticeable mechanical disturbance) is shown for the effect of brush cutting during  in the unburnt area.

two subpopulations, one of which was undocumented prior to our surveys. The existence of a second subpopulation at a distinct location considerably enhances the species’ pro- FIG. 5 Projected sizes of the flowering population of Disa procera spects for persistence, although the current Red List categor- on the burnt site over  years. Starting population sizes were ization of Critically Endangered remains applicable because based on the total number of flowering individuals recorded (a)  of the limited area of occupancy, fragmentation and decline prior to the burn treatment (during the survey) and (b)  in quality of the habitat, and the small size of the overall after the burn treatment (during the survey). Projected population sizes (± SD) are based on mean values from , population. Thus, protection and appropriate management iterations of population projections calculated using randomly of the new site must be a priority. selected values of λ, which were calculated from observed annual The subpopulation that was the focus of this analysis is growth rates in the population before or after the burn subject to high interannual variation in total size. High turn- treatment. over of individuals among years and decreasing likelihood of plant reappearance over time suggests that the species has a relatively short life span (cf. Hutchings, a), al- was not predictable over successive years, and because our though our evidence is limited to the -year study period. surveys were limited to flowering plants this suggests that Notably, however, flower production by individual plants the total population size could exceed the mean annual

FIG. 6 Projected sizes of the flowering population of Disa procera  on the unburnt site over  years. Starting population sizes were FIG. 7 Projected extinction probabilities over years for based on the total number of flowering individuals recorded (a) flowering individuals of Disa procera on (a) the burnt site, based prior to the burn treatment (during the  survey) and (b) on observed annual growth rates from pre-burn years and (b) after the burn treatment (during the  survey). Projected the unburnt site, based on observed annual growth rates from population sizes (± SD) are based on mean values from , post-burn years. Probabilities are calculated as the number of   iterations of population projections calculated using randomly times the population reached zero flowering plants out of , selected values of λ, which were calculated from observed annual iterations for each year. growth rates in the population before or after the burn treatment. fragments isolated by surrounding indigenous forest in the southern Cape, possibly as a result of pollinator limitation size of the detectable aboveground population. Although (Bond et al., ). Our study site is similarly isolated monitoring all life stages of this species could improve our from larger stretches of fynbos but given the study species’ understanding of its demographics, aboveground markers comparatively high rate of seed set it is unlikely that a lack of to obtain such data are undesirable as metal pins can dam- pollination currently constrains population growth. The age underground plant parts and increase the visibility of pollinators of D. procera are unconfirmed; a brown monkey plants that are already prone to illegal collection. beetle Pachycnema marginella was observed inside a flower Nevertheless, the number of reproductive individuals in and may contribute to pollination but the species is prob- the study population is an important indicator of the overall ably adapted for bee pollination (Johnson & Steiner, ; health of the species as these individuals are responsible for Johnson et al., ; Steiner, ; S.D. Johnson, pers. future recruitment. Accordingly, continued monitoring of comm.,  January ) and more information is needed reproductive individuals is an important management to determine if pollinator presence is truly a limiting factor. protocol. There are several alternative explanations for the scarcity Encouragingly, fruit set recorded for the study species of D. procera. One explanation may be the aged and over- (% of flowers producing capsules) was comparable to or grown nature of the vegetation where D. procera occurs. higher than that of many other South African orchids (Disa Likewise, the species’ patchy distribution and absence from ferruginea c. %, D. atricapilla c. %, D. bivalvata c. %, considerable tracts of seemingly suitable habitat, even within D. uniflora c. %, D. grandiflora c. %, Herschelianthe gra- the study area (authors, pers. obs.), may indicate constraints minifolia c. %; Johnson & Bond, ; Johnson, , ; to dispersal, recruitment or symbiotic mycorrhizae. Steiner et al., ), and higher than the global mean for nat- Furthermore, physical damage to plants during their ural fruit set in non-autogamous orchids (. ± SE .%; aboveground growing season may be limiting the recovery Tremblay et al., ). Orchids are scarce in fynbos of these orchids. Plant heights were significantly reduced

in  when brush cutting occurred late into the growing successive years (Stuckey, ; Stewart et al., ;Willson season, potentially with a legacy effect in . Brush cutting et al., ). In the absence of naturally occurring fires, how- in  and  during the vegetative growth phase and ever, prescribed burning (outside the orchid’s growingseason) prior to flowering also coincided with the smallest recorded is advisable at intervals of – years and intensities charac- population sizes and suppressed rates of fruit set within teristic of the natural disturbance regime in coastal fynbos those seasons. Damage incurred between the flowering (Kraaij et al., ). These prescribed burns should have a posi- and fruiting stages through brush cutting and other agents tive impact on plant vigour, flower production and population likewise had a negative effect on the proportion of fruit set of growth rates, and significantly bolster the recovery of the damaged individuals and thus the potential for population species. growth. Nevertheless, brush cutting at appropriate times is Despite high levels of flower production, high capsule set, beneficial to some orchids and could have a positive impact and projections suggesting the population should be in- on D. procera as well (Steiner, ). creasing over time, the population remains extremely The study species’ close association with hiking trails also small, and accordingly conservation managers need to con- predisposes it to flower picking and illegal collection and sider carefully the various factors that could be limiting the could hinder its recovery. However, damaged and undam- population size. The existence of an additional subpopula- aged plants were not discernible on the basis of plant height tion potentially facilitates more manipulative studies of the or number of flowers per plant, which suggests that the effects of brush cutting, fire and other factors that may be agents responsible for damage were impartial to plant size limiting D. procera. Analysis of dormancy periods, mycor- or showiness. rhizal associates, soil seed banks and seed longevity should The local fauna could also be contributing to the scarcity also help to elucidate remaining uncertainties regarding the of D. procera, as animal activity was often evident (e.g. shal- species’ longevity, turnover of individuals, recruitment and low soil disturbance and severance of racemes a few centi- the role of vegetative reproduction (Tamm, ; Stewart metres aboveground) in the immediate vicinity of study et al., ; Hutchings, a,b; Whigham et al., ). plants. These signs suggest that flowers may have been da- Consequently, in addition to the management strategies re- maged and removed by tortoises and rodents, including commended based on the results of this study, we advocate porcupines Hystrix africaeaustralis and dune mole-rats further studies of this species along with continual monitor- Bathyergus suillus, which are abundant in the area. These ing of both populations to ensure full species recovery. animals can also disturb topsoil and possibly harm under- Overall, the species’ small global population, the large inter- ground plant parts that are important during dormancy per- annual variance in its population size, the short life span of iods. Damage to flowers and seed capsules by invertebrates individual plants, and the improbable existence of a long- was also common in the study population. It is likely that lived soil seed bank (Whigham et al., ) all increase plants were damaged to varying degrees by a variety of herb- the species’ susceptibility to extinction, and proactive man- ivorous animals, and future analyses on the effects of the agement efforts will be necessary to increase the likelihood local fauna could be beneficial to managers. of this species’ long-term survival. Despite the limitations and disturbances that threaten D. procera, our results indicate that fire has a favourable effect on the species at the population and individual levels, an Acknowledgements outcome that is in line with evidence from other studies South African National Parks funded this research. We    (Stuckey, ; Stewart et al., ; Bytebier et al., , thank Ismail Ebrahim (Threatened Species Programme,    ; Coates & Duncan, ; Lamont & Downes, ). South African Biodiversity Institute) for initial assistance Like many other orchids, D. procera seems to benefit from with development of the field survey protocol; the dedicated disturbances that create open and well-lit conditions, as evi- Outramps group of the Custodians of Rare and Endangered dent from its habit of growing along trails and its enhanced Wildflowers for assistance with in-field searches for add- performance after fires. Accordingly, prescribed burns itional plants, and the discovery of a second population of could be an important management strategy for this species. the study species; and protected-area management staff However, because physical damage to the plants during for undertaking prescribed burning. their aboveground growing season is particularly detrimental, management agencies should refrain from any activity (including trail maintenance and prescribed burning) that References may cause disturbance to the species’ habitat during its above-  ground growing season (September–January). Although dis- BOND, W.J., MIDGLEY,J.&VLOK,J.( ) When is an island not an island? Insular effects and their causes in fynbos shrublands. turbances may enhance growing conditions, inappropriate Oecologia, , –. timing and types of disturbances may have dire consequences BOWLES, M.L. () The tallgrass prairie orchids Platanthera and potentially affect the population’s performance over leucophaea (Nutt.) Lindl. and Cypripedium candidum Muhl. ex

Willd.: some aspects of their status, biology, and ecology, and LINDER, H.P., KURZWEIL,H.&JOHNSON, S.D. () The Southern implications toward management. Natural Areas Journal, , –. African orchid flora: composition, sources and endemism. Journal BYTEBIER, B., ANTONELLI, A., BELLSTEDT, D.U. & LINDER, H.P. () of Biogeography, , –. Estimating the age of fire in the Cape flora of South Africa from an MCDONALD, D.J. () Montane flora of the southern Langeberg, orchid phylogeny. Proceedings of the Royal Society B, , –. South Africa: a checklist of the flowering plants and ferns. Bothalia, BYTEBIER, B., OLIVER, E.G.H. & LILTVED, W.R. () Disa , –. linderiana (Orchidaceae), a new orchid from the Western Cape of MEHRHOFF, III, L.A. () Pollination in the genus Isotria South Africa. South African Journal of Botany, , –. (Orchidaceae). American Journal of Botany, , –. COATES,F.&DUNCAN,M.() Demographic variation between MEHRHOFF, L.A. () The dynamics of declining populations of an populations of Caladenia orientalis—a fire-managed threatened endangered orchid, Isotria medeoloides. Ecology, , –. orchid. Australian Journal of Botany, , –. MORRIS,W.F.&DOAK, D.F. () Quantitative Conservation CRAIN, B.J. & TREMBLAY, R.L. () Update on the distribution of Biology: Theory and Practice of Population Viability Analysis. Lepanthes caritensis, a rare Puerto Rican endemic orchid. Sinauer Associates, Inc., Sunderland, USA. Endangered Species Research, , –. MUCINA,L.&RUTHERFORD, M.C. (eds) () The Vegetation of DIXON, P.M. & COOK, R.E. () Science, planning, and the recovery South Africa, Lesotho and Swaziland. Strelitzia, Volume . South of endangered plants. Endangered Species UPDATE, , –. African National Biodiversity Institute, Pretoria, South Africa. DUNCAN, M., PRITCHARD,A.&COATES,F.() Major threats to MYERS, N., MITTERMEIER, R.A., MITTERMEIER, C.G., DA FONSECA, Endangered orchids of Victoria, Australia. Selbyana, , –. G.A.B. & KENT,J.() Biodiversity hotspots for conservation HOOD, G.M. () PopTools version ... Http://www.poptools.org. priorities. Nature, , –. HUTCHINGS, M.J. (a) The population biology of the early spider RAIMONDO, D., VON STADEN, L., FODEN, W., VICTOR, J.E., HELME, orchid, Ophrys sphegodes Mill. I. A demographic study from  to N.A., TURNER, R.C. et al. (eds) () Red List of South African . Journal of Ecology, , –. Plants. Strelitzia, Volume . South African National Biodiversity HUTCHINGS, M.J. (b) The population biology of the early spider Institute, Pretoria, South Africa. orchid, Ophrys sphegodes Mill. II. Temporal patterns in behaviour. SØRENSEN,T.() A method of establishing groups of equal Journal of Ecology, , –. amplitude in plant sociology based on similarity of species content JOHNSON, S.D. () Carpenter bee pollination of Herschelianthe and its application to analyses of the vegetation on Danish graminifolia (Orchidaceae) on the Cape Peninsula. Flora, , commons. Biologiske Skrifter / Kongelige Danske Videnskabernes –. Selskab, , –. JOHNSON, S.D. () Evidence for Batesian mimicry in a STEINER, K.E. () The evolution of beetle pollination in a South butterfly-pollinated orchid. Biological Journal of the Linnean African orchid. American Journal of Botany, , –. Society, , –. STEINER, K.E., WHITEHEAD, V.B. & JOHNSON, S.D. () Floral and JOHNSON, S.D. & BOND, W.J. () Habitat dependent pollination pollinator divergence in two sexually deceptive South African success in a Cape orchid. Oecologia, , –. orchids. American Journal of Botany, , –. JOHNSON, S.D., HOBBHAHN,N.&BYTEBIER,B.() Ancestral deceit STEWART, J., LINDER, H.P., SCHELPE, E.A. & HALL, A.V. () Wild and labile evolution of nectar production in the African orchid Orchids of Southern Africa. Macmillan, Johannesburg, South Africa. genus Disa. Biology Letters, , . STUCKEY, I.H. () Environmental factors and the growth of native JOHNSON, S.D. & STEINER, K.E. () Pollination by megachilid bees orchids. American Journal of Botany, , –. and determinants of fruit-set in the Cape orchid Disa tenuifolia. TAMM, C.O. () Survival and flowering of some perennial herbs. II. Nordic Journal of Botany, , –. The behaviour of some orchids on permanent plots. Oikos, , –. JOHNSON, S.D., STEINER, K.E., WHITEHEAD, V.B. & VOGELPOEL,L. TREMBLAY, R.L., ACKERMAN, J.D., ZIMMERMAN, J.K. & CALVO, R.N. () Pollination ecology and maintenance of species integrity in () Variation in sexual reproduction in orchids and its co-occurring Disa racemosa L.f. and Disa venosa SW. (Orchidaceae) evolutionary consequences: a spasmodic journey to diversification. in South Africa. Annals of the Missouri Botanical Garden, , Biological Journal of the Linnean Society, , –. –. VON STADEN,L.&TURNER,D.() Disa procera H.P.Linder. KRAAIJ, T., COWLING, R.M. & VAN WILGEN, B.W. () Past National Assessment: Red List of South African Plants version approaches and future challenges to the management of fire and .. Http://redlist.sanbi.org/redcat.php [accessed  May ]. invasive alien plants in the new Garden Route National Park. South WHIGHAM, D.F., O’NEILL, J.P., RASMUSSEN, H.N., CALDWELL, B.A. & African Journal of Science, , article no. ,  pp. McCORMICK, M.K. () Seed longevity in terrestrial orchids— KRAAIJ, T., COWLING, R.M., VAN WILGEN, B.W. & SCHUTTE-VLOK, potential for persistent in situ seed banks. Biological Conservation, A.-L. () Proteaceae juvenile periods and post-fire recruitment as , –. indicators of minimum fire return interval in eastern coastal fynbos. WHITMAN, M., MEDLER, M., RANDRIAMANINDRY, J.J. & Applied Vegetation Science, , –. RABAKONANDRIANINA,E.() Conservation of Madagascar’s KURZWEIL,H.() Orchidaceae. In Plants of the Greater Cape granite outcrop orchids: the influence of fire and moisture. Floristic Region. : The Core Cape Flora (eds J. Manning & Lankesteriana, , –. P. Goldblatt), p. . South African National Biodiversity Institute, WILLSON, G., PAGE, M.J. & AKYUZ, F.A. () Precipitation and fire Pretoria, South Africa. effects on flowering of a rare prairie orchid. Great Plains Research, LAMONT, B.B. & DOWNES, K.S. () Ecology of plant resprouting in , –. fire-prone ecosystems. Plant Ecology, , –. LILTVED, W.R. () The ‘moonlight acrolophia’ recent observations on the orchid Acrolophia lunata, in the southern Cape Floristic Biographical sketches Region. Veld & Flora, , –.  LILTVED, W.R. & JOHNSON, S.D. ( ) The Cape Orchids: A Regional T INEKE K RAAIJ’s research interests include fynbos ecology, with Monograph of the Orchids of the Cape Floristic Region. Sandstone emphasis on fire ecology in the eastern part of the Cape Floristic Editions, Noordhoek, South Africa. Kingdom, the biology of invasive alien plants, conservation of

rare and threatened plant species, ecosystem restoration, and information systems to document and analyse biodiversity patterns plant−herbivory interactions. Much of her research has applied and processes. BENJAMIN J. CRAIN’s primary research interests in- value for biodiversity conservation in protected areas. J OHAN clude conservation biology, ecology of threatened species, and biogeog- A. BAARD’s research interests include indigenous forest and fynbos raphy. His research involves population viability modelling, spatial ecology, plant inventories, ecology of invasive alien plants, conserva- analyses, and biodiversity assessments, particularly for threatened spe- tion of rare and threatened plant species, and the use of geographical cies and their habitats.