|Home | About | Journals | Submit | Contact Us | Français|
Undisturbed forest habitat can be relatively impenetrable to invasive, non-native species. Orchids are not commonly regarded as invasive, but some species have become invasive and these generally depend on habitat disturbance. One of the most aggressive orchids is Oeceoclades maculata, a terrestrial species with remarkable ecological amplitude. Originally from tropical Africa, it is now widespread in the neotropics. By associating its local distribution with land-use history and habitat characteristics, it was determined whether O. maculata is dependent on habitat disturbance. It was also investigated whether this exotic orchid occupies the same habitat space as two sympatric native species.
Six 10 m × 500 m transects were censused in June 2007 on the 16-ha Luquillo Forest Dynamics Plot, located in the Luquillo Mountains, Puerto Rico. The plot had been mapped for historical land use, topography and soil type.
Oeceoclades maculata was the most abundant of three orchid species surveyed and was found in all four historical cover classes. In cover class 3 (50–80 % forest cover in 1936), 192 of 343 plants were found at a density of 0·48 plants per 5 × 5 m subplot. Over 93 % of the 1200 subplots surveyed were composed of Zarzal or Cristal soil types, and O. maculata was nearly evenly distributed in both. The orchid was most common on relatively flat terrain. The distribution and abundance of two sympatric orchid species were negatively associated with that of the invasive species.
Oeceoclades maculata does penetrate ‘old growth’ forest but is most abundant in areas with moderate levels of past disturbance. Soil type makes little difference, but slope of terrain can be important. The negative association between O. maculata and native species may reflect differences in habitat requirements or a negative interaction perhaps at the mycorrhizal level.
Spread of invasive species has become a global concern with a wide array of opinions on what, if any, actions should be taken to remove them from the landscape (Ewel and Putz, 2004; Lugo, 2004; Denslow and Johnson, 2006). Success of many invasive species can be directly tied to previous land-use history such as agriculture and logging, which open up seemingly impenetrable landscapes (Costa and Magnusson, 2002; Sax et al., 2002; Grau et al., 2003; Lindborg and Eriksson, 2004; Flinn and Velland, 2005). Abandonment of these radically altered lands, generally due to economic shifts, leaves open disturbed areas that are easily colonized by non-native species (Aide et al., 2000; Lugo, 2004). This may be particularly true for tropical islands where naturalized exotic plant species have approximately doubled the size of some floras (Sax et al., 2002; Denslow, 2003; Denslow and Johnson, 2006; Gimeno et al., 2006) and where the frequency of introductions is high (Lockwood et al., 2007).
The islands of the Caribbean have a long history of ecological disturbance and recovery, both from natural and human causes. Most of the Caribbean is subjected to the destructive power of hurricanes, but the most dramatic ecological alteration that has occurred since the 15th century has been the arrival of Europeans. In Puerto Rico, the landscape went from nearly 100 % forest cover to <6 % by the 1940s (Aide et al., 2000; Rudel et al., 2000; Lugo, 2004). In the years that followed, an exodus from rural farms to urban factories occurred as the island economy shifted to one more reliant on small industry. Abandoned farms were allowed to become secondary forests to such an extent that ‘from 1950 to 1990 proportionally more land in Puerto Rico had been reforested than anywhere in the world’ (Rudel et al., 2000). Although this transformation is currently unusual, it may portend changes elsewhere in the tropics as economies evolve worldwide.
Luquillo Experimental Forest of Puerto Rico was established in the 1930s and at the time consisted of a mosaic of land uses. It is now entirely forested, but variation in human use at the time of establishment has had a great effect on current tree species composition. Nevertheless, the forest is sufficiently mature to become largely impenetrable to non-native invasive tree species (Thompson et al., 2002). Herbaceous components of its understorey vegetation also show patterns of abundance associated with historical land use (Portugal Loayza, 2005; Bergman et al., 2006), but it is not known how impenetrable the forest is to invasion by non-native herbaceous species.
Here, ecological correlates of distribution and abundance of an invasive, terrestrial orchid, Oeceoclades maculata are examined in the Luquillo Mountains of Puerto Rico. The species was first described from Brazil in 1829. It is thought to have originated in the tropical regions of Africa because it is the only species in the genus that exists outside of Africa, Madagascar and several small adjacent islands (Dod, 1986; Stern, 1988). Today, O. maculata is found throughout the neotropics, making it one of the most successful invasive plant species (Dod, 1986; Stern, 1988). The species is autogamous, at least in Puerto Rico, which may help explain its march through the neotropics (González-Díaz and Ackerman, 1988). Among invasive orchids, reproductive systems vary from apomixis and autogamy to obligately outcrossing systems with deceptive pollination systems (Ackerman, 2007). Autogamy, then, is not likely to be the sole reason for the success of O. maculata.
Oeceolades maculata was first noted in Puerto Rico in the mid 1960s and has rapidly spread throughout the island (González-Díaz and Ackerman, 1988). Based on the definitions of Richardson et al. (2000), it is both invasive and naturalized. The species occurs in a wide range of habitats from the northern mogotes (karst hills), where it is most abundant in the moist to wet understorey of secondary forests, to the southern dry forests of Guánica (González-Díaz and Ackerman, 1988). The rainforest of the Luquillo Mountains is one of the last areas of Puerto Rico to be invaded by O. maculata; the first collection was made in 1987 (Ackerman 2384, UPRRP). Here it is investigated how land-use history may be associated with the distribution of invasive O. maculata, particularly whether the orchid has established in old-growth habitats. Potential effects of other ecological conditions such as soil type and slope of the terrain are explored further. Finally, the relationship between distributions of native terrestrial orchids in the forest and that of O. maculata are examined.
A census of three species of Orchidaceae was conducted with the primary focus being on the distribution of the invasive Oeceoclades maculata (Lindl.) Lindl. Wullschlaegelia calcarata Benth. and Prescottia stachyodes (Sw.) Lindl. were assessed to establish the current relationships with O. maculata and provide a baseline to determine if the invasive species will have an effect on their distributions in years to come. All three have minute, dust-like, wind-dispersed seeds typical of the family. They are also entirely dependent on mycorrhizal associations for successful seed germination. Voucher specimens for the three species were deposited at the University of Puerto-Río Piedras herbarium (UPRRP; O. maculata: Cohen 169; P. stachyodes: Cohen 170; W. calcarata: Cohen 171).
Oeceoclades maculata is found in shaded, often disturbed habitats ranging from dry to wet forests and from sea level to 750 m throughout Puerto Rico but is most abundant in the northern karst region. Plants are sympodial and caespitose with short pseudobulbs, and each shoot bears one to three distinctively mottled green leaves. Inflorescences arise from the base of pseudobulbs (Ackerman, 1995). Flowers are autogamous but less than half produce fruit. Fruit production is resource limited (González-Díaz and Ackerman, 1988).
Wullschlaegelia calcarata is a saprophytic, achlorophyllous orchid found in the understorey of tropical rainforests of the West Indies, Central America and South America (Born et al., 1999). It grows in the montane forests of Puerto Rico between 250 m and 750 m (Ackerman, 1995) where densities may reach over 1300 plants ha−1 (Bergman et al., 2006). A single, erect, leafless shoot arises from a cluster of short fusiform roots embedded in partially decomposed leaf litter or just below the soil surface. The whitish plants reach a height of 15–55 cm and have a racemose inflorescence of several to many, cleistogamous flowers. The small fruits mature rapidly (Ackerman, 1995).
Prescottia stachyodes is a terrestrial orchid (rarely rupicolous or epiphytic) of variable size, often >50 cm tall, found in the understorey of wet forests in Brazil, Venezuela, Colombia, Central America, Mexico and the West Indies. In Puerto Rico it occurs in moist to wet montane forests between 450 m and 800 m elevation (Ackerman, 1995) where densities may reach 960 plants ha−1 (M. Whitman and J. Ackerman, unpubl. res.). Plants have shallow, thick, fleshy roots produced from a short, rhizome. The few persistent, basal leaves are elliptical and dark green to dark purplish green. The erect inflorescence produces numerous flowers in a dense raceme. In Brazil, flowers are pollinated by pyralid moths (Singer and Sazima, 2001). Pollinators and breeding system are unknown in Puerto Rico, but morphological examination of flowers along the inflorescence axis suggests that autogamy is likely (A. Cuevas, pers. comm.). Fruit set generally approaches 100 % (Ackerman, 1995).
Field work took place on the 16-ha Luquillo Forest Dynamics Plot (LFDP, south-west corner 18°20′N, 65°49′W), which is near El Verde Field Station (University of Puerto Rico) in the Luquillo Mountains. The site lies within the subtropical wet life-zone (Holdridge system; Ewel and Whitmore, 1973) and averages approx. 3500 mm of rainfall annually (at least 200 mm per month), with March and April being the driest months (Thompson et al., 2002). The LFDP is in the north-western part of the 19 648-ha Luquillo Experimental Forest and is approx. 1·2 km from the nearest forest boundary and ranges in elevation from 333 m to 425 m (Thompson et al., 2002). It was divided into four hundred 20 m × 20 m plots, each of which was further subdivided into sixteen 5 m × 5 m subplots.
Thompson et al. (2002) mapped the LFDP into four canopy cover classes based on historical records including aerial photographs from 1936. Today, the current canopy coverage is nearly 100 % in the LFDP, but effects of previous land use can still be seen today in the composition and distribution of tree species (Thompson et al., 2002). Canopy cover classes are as follows: cover class 1 (1·16 ha, <20 % canopy cover), cover class 2 (3·96 ha, 20–50 % canopy cover), cover class 3 (5·64 ha, 50–80 % canopy cover). These first three cover classes comprise the northern half of the LFDP; they were all logged and used for various agricultural activities including coffee, mangos and bananas. Canopy class 4 (5·24 ha, >80 % canopy cover) makes up the southern portion of the LFDP and was never clear-cut or used for agriculture, but it was subject to some selective logging in the 1940s.
To determine the association of land-use history and distribution and abundance of O. maculata within the LFDP, six 500-m transects running south to north were made crossing into the historic canopy cover classes. Each transect was 10 m wide (two adjacent 5 m × 5 m subplots) with 30 m between each transect except for transects 1 and 6 which were 15 m from the eastern and western boundaries of the LFDP. Because the LFDP is representative of the surrounding forest, transects 1 and 6 suffer no edge effect. Twelve hundred 5 m × 5 m subplots were surveyed to measure abundance of the three orchid species during June 2007.
Neither Prescottia stachyodes nor O. maculata was in flower at the time of survey, but they were easily identified using leaf characters. Plants of both species are easily spotted in the forest in the non-flowering condition because the leaves are the most conspicuous part of the plants. Wullschlaegelia calcarata was past flowering but, during the study, shoots were still visible as capsules were maturing and dispersing seed. Plants of W. calcarata are only visible when inflorescences or infructescences are present. To enhance independence among subplots for statistical analyses, 100 of 101 subplots from cover class 1 and 100 subplots from the three other cover classes were randomly selected. It was investigated whether abundance of O. maculata was associated with any of the four classes of historical canopy coverage using the nonparametric Kruskal–Wallis test to compare numbers of orchids per subplot and obtain a comparison among means. It was also examined whether frequency of plots with at least one O. maculata plant was independent of cover class using a chi-square test of independence. These and all other analyses were performed using the statistical package JMP (version 4·04; SAS Institute, Cary, NC, USA).
The association of O. maculata, W. calcarata and P. stachyodes with soil type and slope of the terrain in the LFDP was measured using the nonparametric Kruskal–Wallis test. Soil and slope analyses were done using soil maps from the US Department of Agriculture soil survey (Soil Survey Staff, 1995). The soil data are based on 20 m × 20 m plots and were interpolated for each of the 5 m ×5 m subplots that make up a 20 m × 20 m plot. The LFDP is separated into five major soil types: Zarzal, Prieto, Cristal, Coloso and Fluvaquents, but only Zarzal and Cristal covered sufficient area (93 % of the 1200 subplots surveyed) for the present analysis. Both are volcanic clay soils; they differ in that Zarzal is slightly drier than Cristal (Soil Survey Staff, 1995). The LFPD has three slope classes: 3–15 %, 15–30 % and 30–60 % slope, all of which cover sufficient area to be included. All slope categories were represented by at least 300 subplots from which 200 were randomly selected for analysis.
The relationship between distribution of O. maculata and W. calcarata and P. stachyodes was assessed using the nonparametric Spearman's correlation coefficient. Because many 5 × 5 m subplots had no plants of a given species, population counts from two side-by-side 5 m × 5 m subplots were combined, creating a 10 m × 5 m plot. Subplots with no data for either of the native species or O. maculata were omitted because such plots contain no information regarding a potential interaction between the three species. Then 100 of the 10 m × 5 m subplots were randomly selected for analyses.
Oeceoclades maculata is present in all four canopy cover classes, but it is most abundant in cover class 3 (Fig. 1), which is the second least disturbed cover class with an estimated forest cover in 1936 between 50 % and 80%. Here, 192, which is 56 % of the total population sampled, were counted. Cover class 3 had a mean of 0·74 O. maculata plants per 5 m × 5 m subplot, or approx. 370 plants ha−1. Cover classes 1, 2 and 4 had far fewer orchids than cover class 3 (Table 1; cover class 1 mean = 0·18, cover class 2 mean = 0·25, cover class 4 mean = 0·19). The number of O. maculata plants differed significantly among the four cover classes (chi-square test of independence: χ2 = 15·96, d.f. = 3, P = 0·0012; Table 1).
The frequency of plots that contained at least one orchid plant also showed that cover class 3 was significantly preferred over the other three cover classes (chi-square test of independence: χ2 = 26·3, P < 0·0001) (cover class 1 = 9 %, cover class 2 = 11 %, cover class 3 = 22 %, cover class 4 = 11 %).
Distribution of O. maculata was not associated with a particular soil type. Average number of O. maculata plants per 5 m × 5 m plot was 0·29 for Zarzal soils, and 0·27 for Cristal soils (Kruskal–Wallis test: χ2 = 0·33, d.f. = 1, P = 0·57). Wullschlaegelia calcarata, on the other hand, was associated with Zarzal (Zarzal av. = 0·37; Cristal av. = 0·09; Kruskal-Wallis test: χ2 = 9·38, d.f. = 1, P = 0·002), as was P. stachyodes (Zarzal av. = 0·15; Cristal av. = 0·06; Kruskal–Wallis test: χ2 = 7·15, d.f. = 1, P = 0·007).
The results for slope of the terrain were significant for all three species. Oeceoclades maculata and P. stachyodes were most frequent on slopes 3–15 % (Kruskal–Wallis tests: O. maculata: χ2 = 6·27, d.f. = 2, P = 0·04; P. stachyodes: χ2 = 13·05, d.f. = 2, P = 0·0001). Average number of plants for both species was highest in this slope class (Table 2; O. maculata mean = 0·41 per 5 m × 5 m subplot, P. stachyodes mean = 0·22). Distribution of the achlorophyllous W. calcarata among slope classes was different from that of the other two terrestrial species in the LFDP by its preference for slopes 15–30 % (Kruskal–Wallis: χ2 = 6·97, d.f. = 2, P = 0·03), where it averaged 0·37 plants per plot.
Oeceoclades maculata was more abundant than W. calcarata in 124 of 224 of the 10 m × 5 m subplots, and this was true for P. stachyodes where 132 of 191 of the subplots contained more O. maculata (Fig. 1). All three species had a non-normal distribution (Shapiro–Wilk test: P < 0·000), so Spearman's rank correlation coefficient was used to determine whether the distributions of the two native orchids were correlated with that of O. maculata. The distributions of both W. calcarata and P. stachyodes were negatively correlated with O. maculata (W. calcarata and O. maculata: rs = –0·63, P < 0·0001; P. stachyodes and O. maculata: rs = –0·43, P < 0·0001; Fig. 1).
Tropical island ecosystems are at a high risk for species invasions and subsequent naturalization by non-native species. Their high invasibility is facilitated by a combination of heavy traffic in introductions and high levels of natural and anthropogenic habitat disturbances (Foster et al., 1999; Sax et al., 2002; Denslow, 2003; Gimeno et al., 2006). Orchids are not often the focus of invasive species studies, regardless of geography. From floristic works and miscellaneous literature, it is known that Hawai'i has at least five alien species established on the Big Island, and Puerto Rico has nine (Wagner et al., 1990; Caccia, 2005; Ackerman, 2007). Invasive orchids and plants of other families are most commonly seen in disturbed habitats, usually those caused by human activities. The forests where Oeceoclades maculata is abundant are in various successional stages. Some of these recovering forests superficially resemble old growth forests, so it is not always obvious whether O. maculata is affected by land-use histories. It is known that O. maculata has a broad habitat range, amply demonstrated in Puerto Rico by its presence in lowland cactus thorn-scrub and in broadleaf, montane rainforests. It has certainly found a suitable home in the LFDP, with a population density of approx. 114 plants ha−1. Nevertheless, plants are not evenly distributed among the cover classes. It was shown that O. maculata is most abundant in cover class 3, which was the second least disturbed part of the LFDP (50–80 % forest cover in 1936). The orchid is also scattered throughout the three other cover classes but in densities less than half of that for cover class 3 (Table 1).
Unaltered forest habitats are generally thought to resist establishment of invasive species (Denslow, 2003), and this seems true for the LFDP where non-native tree species, of which there are plenty in Puerto Rico, infrequently appear in the forest after a hurricane but fail to persist (Thompson et al., 2002). Invasive herbaceous herbs, however, may behave differently since the old growth forest of LFPD was not immune from invasion by O. maculata, nor were undisturbed bushland areas a barrier to an invasive orchid, Disa bracteata, in Australia (Bonnardeaux et al., 2007).
The first record of O. maculata in the Luquillo Mountains dates to 1987, and in the 20 years since then this orchid has become common in the LFDP and other parts of the Luquillo Forest (J. D. Ackerman, unpubl. obs.). Like some invasive herbs in other forest types, including old-growth temperate forests, time since initial colonization may be more important than distance from human disturbance (Wiser et al., 1998; Gilbert and Lechowicz, 2005). Although O. maculata appears to prefer parts of the forest that have been moderately modified by human activities, it is expected that, over time, differences in abundance of this orchid will diminish among the four cover classes.
Local distribution of orchids can be affected by a number of environmental factors, some of which may be independent of land-use history. Intensive agriculture can severely alter soil characteristics, and invasive species that follow abandonment can further alter soil features (Noble et al., 2000; Flinn and Vellend, 2005; Hawkes et al., 2005; Kulmatiski et al., 2006; Wolfe and Klironomos, 2005). In the LFDP, farming and logging were generally small-scale activities so damage to the soil was not on the magnitude of large-scale farming or clear-cut logging, nor was it widespread. Tree species composition in the LFPD was largely explained by land-use histories, not soil types (Thompson et al., 2002).
In contrast, forest understorey herbs, including the two native orchid species in this study, appeared to be a little more sensitive to soil type than tree species were (Portugal Loayza, 2005). Both native orchids preferred the drier Zarzal soils. Soil preference of W. calcarata is most likely related to its affinity for undisturbed locations, like cover class 4, which was the least disturbed part of the LFDP and is almost entirely composed of Zarzal soil (Bergman et al., 2006). Soil preference of P. stachyodes could be due to factors that were not part of this study such as soil pH, moisture content and, perhaps most importantly, mycorrhizal relationships. On the other hand, the alien O. maculata showed no preference between Zarzal and Cristal soil types, perhaps reflecting the unusual breadth of habitat tolerance in this species.
Analysis for an association with slope showed that O. maculata and P. stachyodes prefer flat terrain. Both species have shallow root systems (Ackerman, 1995), and being in flat places decreases the chances of getting swept away by heavy rains or landslides. This may explain the abundance of O. maculata in cover class 3, which is the flattest part of the LFDP (Thompson et al., 2002). Wullschlaegelia calcarata, on the other hand, preferred steep locations, which is also probably more related to its distribution and abundance in cover class 4, the steepest part of the LFDP (Thompson et al., 2002). Ironically, slope of the terrain acted as a barrier and probably prevented many areas on the southern and western ends of the LFDP from being farmed or logged because they were simply too difficult to work (Foster et al., 1999).
Distributions of both native orchids were negatively correlated with O. maculata (Fig. 1). The invasive species was more abundant than W. calcarata in 124 of 224 of the 10 × 5 m subplots, and the same was true of P. stachyodes for which 132 of 191 of the subplots contained more O. maculata (Fig. 1). This could be due to different niche requirements or negative interactions involving their mycorrhizal symbionts. The orchids may be competing for fungal associates or, if their associations are specific, their fungi may compete for resources among themselves altering their relative abundances. Oeceoclades maculata is the most abundant of the three orchid species censused in the LFDP, which could spell trouble for the native species if it continues to spread and competitive interactions actually do exist. These same concerns have been expressed for Australian orchids in face of the rapidly spreading South African species, Disa bracceata (Bonnardeaux et al., 2007).
Why specific invasive species do better than others is a hotly debated topic. They may simply out-compete native species, or it may be a case of being in the right place at the right time (Wiser et al., 1998; Sax et al., 2002; Denslow, 2003; Gilbert and Lechowicz, 2005; Gimeno et al., 2006; Thuiller et al., 2006). There is some evidence that invasive species are governed by general patterns rather than being idiosyncratic (Arim et al., 2006). It is not clear, though, what general patterns apply to invasive orchids. All orchids have two symbiotic interactions that may represent life-history bottlenecks: pollination and mycorrhizal fungi. One would predict that invasive species might be autogamous, like O. maculata, or apomictic like Zeuxine strautematica or Disa bracteata (Sun, 1997; Bonnardeaux et al., 2007), but most may actually have pollinator-dependent breeding systems and even low fruit production typical of most orchid populations (Ackerman, 2007; Tremblay et al., 2005). As for mycorrhizal associations, one may expect that invasive species specialize on a widespread fungus, or that they exploit a broad spectrum of species. Unfortunately, there are not yet sufficient data on these species, but the African Disa in Australia seems to be a generalist (Bonnardeaux et al., 2007).
Regardless of what the general reasons and patterns are for successful invasions, O. maculata has proven to be a hearty weed that has colonized a large portion of the neotropics, and in the LFDP all historical cover classes were invaded, including old growth forests with minor human impacts. Where O. maculata was most abundant, native orchids were less common, prompting the launch of a long-term study to assess potential negative interactions.
We thank the LFPD committee for the opportunity to work on the ‘Big Grid’, Jill Thompson for logistical support and suggestions for the analysis part of the paper, and Jess Zimmerman, Chris Bloch and Chris Higgins for assistance with statistical analyses. This project was supported by funding from NSF-Research Experience for Undergraduates program at El Verde Field Station, University of Puerto Rico, NSF grant number DBI-0552567, A. Ramírez, PI. The LFDP has been funded by NSF grants BSR-8811902, DEB 9411973, DEB 0080538 and DEB 0218039 to the Institute for Tropical Ecosystem Studies, University of Puerto Rico, and to the International Institute of Tropical Forestry, USDA Forest Service, as part of the Long-Term Ecological Research Program in the Luquillo Experimental Forest.