Associations and Assembly Rules of a Vascular Epiphyte Community in Monteverde, Costa Rica Ivan D. Welander Department of Botany University of Wisconsin Madison _____________________________________________________________________ ABSTRACT The existence of "assembly rules" and the importance of deterministic versus stochastic processes in ecological communities have been important subjects of debate in community ecology for over 20 years. To add to the debate, I tested for associations in Zo ne Three vascular epiphyte communities of Ficus tuerckheimii in Monteverde, Costa Rica. Community composition of the 102 recorded species varied widely between trees but tended to be more similar between trees of similar height and location. Also, a negati ve association was found between epiphytic Araceae and Orchidaceae P < 0.016, which I suggest is due to abiotic factors rather than interaction. The results are inconclusive in deciding the importance of stochastic versus deterministic processes but lay the foundation for future research. RESUMEN La existencia de "reglas de asamblea y la i mportancia de procesos determinanticos y fortuitos en comunidades ecolÃ³gicas han idos sujetos importantes de debate en ec ologÃa comunitaria por los 20 aÃ± os pasados. Para contribuir al debate, examinÃ© as ociaciones en comunidades de epÃ fitas de Zona Tres de Ficus tuerckheimii en Monteverde, Costa Rica. ComposiciÃ³n comunitaria de l os 102 especies recordados variÃ³ mucho entre Ã¡rboles pero habÃa mÃ¡ s similar entre Ã¡ rboles de la misma altura y lugar. TambiÃ©n una asociaciÃ³n negativa habÃa encontrado entre Araceae y Orchidaceae P < 0.016 que sugiero que es por razones abiÃ³ticos y no es interacciÃ³n. Los resultos son inconcluyentes para decidir la importancia de proceso s deterministicos y fortuitos pero construyen una fundaciÃ³n para estudios del futuro. INTRODUCTION For the relatively young science of community ecology, one of the most interesting topics of debate remains a question of self identity: what is an ecological community? Since its infancy, the definition of a community has been guided by the dichotomy between deterministic equilibrium and stochastic non equilibrium models. Roughgarden 1989 stated that this "central question of community ecology was posed decades ago: Do the populations at a site consist of all those that happened to arrive there, or of only a special s ubset Â€ those with properties allowing their coexistence?" Though not explicitly so named, a deterministic model of community structure was championed by Elton while Gleason favored stochastic processes Roughgarden 1989. In support of deterministic models of community structure Gilpin and Diamond 1984 defended Diamond's 1975 set of "assembly rules" that govern species composition of a community, emphasizing interspecific interactions such as competition. The debate over the validity of assembly rules ha s generated supportive laboratory and experimental field evidence as well as criticism of their consistency and predictive ability,
and it remains an issue "whether assembly rules can be inferred from nonexperimental data, specifically from combinations of coexisting species, usually on islands" Gotelli and Graves 1996. Nonetheless, observational data may be useful in testing predicted patterns of association. Literature is sparse on the subject with regard to existing plant communities, even though epiph yte communities in forest canopies may be good island analogies and would prove interesting subjects in tests of association and assembly rules. Wolf 1995 studied non vascular epiphytes in the canopy of a Colombian Upper Montane Rain Forest and found ass ociations for four groups of species at different positions from outer to inner canopy. She suggested that competitive interactions were at play because of decreased diversity from outer to inner canopy despite longer possible colonization time in the inne r canopy. Catling and Lefkovitch 1989 studied vascular epiphytes in a Guatemalan Cloud Forest and found small, closely related species to be associated with small, young branches and larger, unrelated species associated with larger, older trunks. They co nclude that both non equilibrium and equilibrium processes, respectively, are at play Catling and Lefkovitch 1989. In a continuation of this type of study, I conducted an inven tory of vascular epiphytes at multiple sites near Monteverde, Costa Rica. H ypothesizing that I would find none beyond what could be explained by chance; I tested for associations at the species and family levels, then analyzed possible causes of observed patterns in light of assembly rules, ultimately to better understand stoc hastic and deterministic models of community organization. METHODS Study Site: The study site was in a cattle pasture located on the Campbell property in the community of Monteverde, west of the Monteverde Cloud Forest Reserve on the Pacific slope of the Cordillera de TilarÃ¡ n, Costa Rica. Lying at 1540 m elevation and receiving approximately 2.5 m of rainfall per year, the site is classified as Lower Montane Wet Forest in t he Holdridge life zone system Holdridge 1967. The dry season is buffered by frequent misting from the Atlantic side of the Continental Divide, which maintains an abundant epiphyte load in 30 40 m primary forest canopies Nadkarni and Wheelwright 2000. Logistics: I chose to study a group of Ficus tuerckheimii Moraceae for its abundance, ease of canopy access, favorable climbing and epiphyte supporting architecture, and to minimize variation between tree species that could complicate tests of associatio n among epiphytes. I accessed the canopy using single rope t echniques as described by Nadkarn i 1984, or by free climbing when possible. Only Zone Three was inventoried, which is defined here as the area within a 2.5 m radius circle centered upon the bole of the tree at its first major branching and generally spanned one third to one fifth the diameter of the crown Fig. 1; Johansson 1975. Presence of all vascular epiphyte species located in Zone Three of each tree was noted and samples were taken for later identification. Unidentifiable species were classified as specifically as possible and given a unique morpho species label. I also recorded diameter at breast height DBH and estimated height to the beginning of Zone Three HZ3 for each tree. Analysis: Richness values of vascular epiphytes were calculated at family level with Division Pteridophyta lumped as one, and species level for each and all trees combined. To test for associations between different taxa, I c ompiled the data into a presence absence matrix for
all taxa on all trees and performed Q mode and R mode analyses Gotelli and Graves 1996. In Q mode analysis, compositional similarity of epiphyte communities was measured by calculating Jaccard similari ty index scale 0 to 1, dissimilar to similar, respectively; Gotelli and Graves 1996 wit hin taxonomic levels for all tree pairs n = 28 comparisons for eight trees. Jaccard index values were compared for a variety of dichotomous groupings of trees such a s upper pasture location versus lower, near forest edge versus far, large crown versus small, shady versus sunny, and tall accessed by ropes, HZ3 = 5.0 to 12.5 m versus short free climbed, HZ3 = 2.1 to 4.4 m. Jaccard indices for comparisons within the same group e.g. tall tall were tested for variation from comparisons between groups e.g. tall short by Mann Whitney U test. Opposing within group comparisons e.g. tall tall versus short short were also tested for variation by Mann Whitney U test. Add itionally, Jaccard indices for each tree pair were regressed versus the difference in height to Zone Three between trees. In R mode analysis, the richness values of the five most species rich families plus Division Pteridophyta were compared together acros s all trees in a 6x8 contingency table and pairwise across all trees in 2x8 contingency tables, using Chi square values to test significance of associations. RESULTS Two hundred forty one vascular epiphytes comprising 102 species from 20 angiosperm families and the Division Pteridophyta were documented in Zone Three of eight F. tuerckheimii individuals Fig. 2. Distribution of species richness among the most species r ich families except Bromeliaceae, for which identification to species was mostly unsuccessful closely followed previously described trends for the Monteverde area Ingram et al. 1996 and for worldwi de vascular epiphyte diversity Fig. 3; Kress 1986. Si milarity between vascular epiphyte communities increased at higher taxonomic levels with Jaccard indices between trees ranging from 0.09 to 0.42 for species comparisons, and 0.29 to 0.75 for families, though the significance of variation between species an d family levels was not tested. At the species level there was a marginal negative correlation between compositional similarity and the difference in height to Zone Three between tree pairs Fig. 4; R 2 = 0.131, P < 0.059. Species composition was more diff erent for tall short tree comparisons than for tall tall and short short tree comparisons Mann Whitney U = 26; P < 0.002; n = 28, while similarity for tall tall and short short comparisons n = 12 was not significantly different. Also, species compositi on compared between upper lower pasture was more different than when compared for upper upper and lower lower pasture Mann Whitney U = 31.5; P < 0.003; n = 28, but similarity did not vary significantly between upper upper and lower lower comparisons n = 13. Jaccard indices for species in other dichotomous groupings and all family groupings did not vary significantly. No associations were found among the most common epiphyte families when richness values for all families were compared for all trees in a 6x8 contingency table Chi square = 34.93; df = 35; P > 0.471. However, pairwise comparisons of family richness across all trees in 2x8 contingency tables revealed a negative association between Araceae and Orchidaceae Fig. 5; Chi square = 17.27; df = 7; P < 0.016.
DISCUSSION The results of this study show that community composition among vascular epiphytes varies widely by location, even for a single host tree species at small spatial scales. Jaccard indices at the family level show relatively high similarity, with approximately 30 to 75% of families shared between trees. However, familial similarity does not necessarily translate to species similarity. The two most similar communities at the species level shared roughly 40% while the least similar ha d only 10% shared species, indicating very high turnover for species from host to host. Despite high turnover, the variation in species composition shows some predictable patterns. When the height or location of compa red trees is similar, the epiphyte communities found in them are also relatively similar and when pairs of trees differ more in height or location, they differ more in species composition. The relative importance of each of these factors to the similarity of species found in trees is confused due to the fact that the short trees were all in one location lower pasture and three of the four tall trees were in the other location upper pasture. Even without knowing whether height or location of trees is mor e important in shaping species composition, this information may help demonstrate the importance of deterministic versus stochastic processes. For deterministic models, height of the host tree may dictate how epiphyte community composition is affected by b iotic factors such as incidence of accidental epiphytes, secondary hemiepiphytes, climbers and vines, as well as ease of access to the canopy by mammals all qualitatively observed. For stochastic models, variations in abiotic factors like microclimatic c hanges, including light, temperature, and exposure to mist and wind, may be caused by differing height or location. Parameters important to colonization such as tree "island" size and distance to epiphyte source populations MacArthur and Wilson 1967 may also be affected by height and location. The observed associations may be explained using the previously mentioned combined model proposed by Catling and Lefkovitch 1989. The model suggests that early on in a tree's life its epiphyte load consists mainly of colonizers that arrive as a result of stochastic processes, while deterministic processes become important as the tree ages, epiphytes have a chance to interact, and the community reaches equilibrium with larger dominant species. This could explain dif ferences between tall and short tree communities because of differing age if a relationship between tree height and age is assumed, and near and far communities because of differing colonizer availability. However, since height and location are related h ere these effects are indistinguishable. The study also reveals a negative association at the family level between Araceae and Orchidaceae. The relationship is not the checkerboard pattern observed in some negative associations, but a pattern of inverse pr oportions of species richness by family at each location. Neither family was observed overtly dominating space or resources, but abundance data were not recorded, which would have been useful in assessing the ability of each family to compete with the othe r. It is possible that the negative association is a result o f interspecific interactions, as usually asserted by assembly rules. However, a more likely scenario is that abiotic factors, determine whether orchids or aroids are able to establish and survive at a given site. Interestingly, causes of epiphyte community composition often cited in the past, such as bark texture and canopy architecture, are ruled out in this case since the assoc iation was found within one tree species. I assert that shade and moisture tolerance along with
dispersal limitation cause the negative association. Superficially, the tree with the most species of aroids and no orchids in Zone Thr ee had a small, shady, cl osed canopy, and high bryophyte cover, while the three trees in which orchids were most species rich had large, sunny, open canopies, and little bryophyte cover. This trend is consistent with Orchidaceae being high sun and drought resistant while Araceae i s shade tolerant and moisture limited. Indeed, Orchidaceae originated from and still contains many "sun lovin g" plants Rundel and Gibson 1996 , while Araceae contains many understory and shade loving plants. Additionally, in the drier sites across rainfal l regimes e piphytic aroids are absent while orchids represent one of few groups of epiphytes present Gentry and Dodson 1987. The observed association is also consistent with the dispersal syndrome of each family in that bird dispersed aroid seeds may rea ch more crowded inner parts of the canopy and wind dispersed orchid seeds would reach more windy sites, as suspected by Delacroix 2001. Overall, this study generated some interesting results that further the understanding of how communities are structure d. Specifically, community composition of vascular epiphytes can be highly variable, but the variability can be somewhat predicted by similarity of host tree height or location. In addition, the two epiphytic families Araceae and Orchidaceae tend to be mor e species rich on different individuals of the same host tree species, probably due to interaction with abiotic factors rather than interaction. These results prevent definitive assertions as to the nature of the processes that shape community composition, whether stochastic or deterministic and assembled. However, this study does lay the foundation for a more conclusive one that would include quantitative observations of species abundance and variation in environmental conditions. ACKNOWLEDGEMENTS Thanks to Alan, Mauricio, Andrew and the rest of the CIEE crew for sharing your lives with me and each other. I especially thank Karen Masters for all the help and support throughout the project and for putting up with identifying my orchids while teaching me to love ferns. Thanks to Bill Haber and Willow Zuchowski, without whom I would have been lost in a pile of inadequate field guides, to Will Wieder and Erin Kurten for spicing up long days in the field, and to the Campbell family who care for such a spe cial place as the Bullpen. Above all, I apologize to the epiphytes who I so ruthlessly stripped of their lives I hope your families forgive me and thank the trees who continue to inspire childlike wonder and amazement in me. LITERATURE CITED Catling, P. M., and L. P. Lefkovitch. 1989. Associations of vascular epiphytes in a Guatemalan cloud forest. Biotropica 21l:35 40. Delacroix, D. 2001. Community structure of vascular epiphytes in a Costa Rican elfin forest. CIEE Program. Spring 2001. Diamond, J. M. 1975. Assembly of species communities. Pages 342 444 in: M. L. Cody and J. M. Diamond, editors. Ecology and Evolution of Communities. Harvard University Press, Cambridge. Gentry, A. H., and C. H. Dodson. 1987. Diversity and biogeography of neotropical vasc ular epiphytes. Ann. Missouri Bot. Gard. 74: 205 233. Gilpin, M. E. and J. M. Diamond. 1984. Are species co occurrences on islands non random, and are null hypotheses
useful in community ecology? Pages 297 315 in: D. R. Strong, D. Simberloff, L. G. Abele, A. B. Thistle, editors. Ecological communities: Conceptual issues and the evidence. Princeton University Press, New Jersey. Gotelli, N. J., and G. R. Graves. 1996. Null models in ecology. Smithsonian Institution Press, Washington. Holdridge, L. R. 1967. Li fe zone ecology. Tropical Science Center, San Jose, Costa Rica. Ingram, S. W., K. Ferrell Ingram, and N. M. Nadkami. 1996. Floristic composition of vascular epiphytes in a neotropical cloud forest, Monteverde, Costa Rica. Selbyana 17: 88 103. Johansson, D. R. 1975. Ecology of epiphytic orchids in West African rain forests. Am. Orch. Soc. Bull. 44: 125 136. Kress, W.J. 1986. The systematic distribution of vascular epiphytes: An update. Selbyana 9: 2 22. Mac Arthur, R.H., and E.O. Wilson. 1967. The theory of i sland biogeography. Princeton University Press, New Jersey. Nadkarni, N.M. 1984. Epiphyte biomass and nutrient capital of a neotropical elfin forest. Biotropica 16 4: 249 256. Nadkarni, N.M. and N. T. Wheelwright, editors. 2000. Monteverde: Ecology and c onservation of a tropical cloud forest. Oxford University Press, Oxford. Roughgarden, J. 1989. The structure and assembly of communities. Pages 203 226 in: J. Roughgarden, R.M. May, and S.A. Levin, e ditors. Perspective in ecological theory. Princeton Unive rsity Press, New Jersey. Rundel, P.W., and A.C. Gibson. 1996. Adaptive strategies of growth form and physiological ecology in neotropical lowland rain forest plants. Pages 33 71 in: A.C. Gibson, editor. Neotropical biodiversity and conservation. University of California, Los Angeles, California. Wolf, J.H. D. 1995. Non vascular epiphyte diversity patterns in the canopy of an upper montane rain forest 2550 3670m, central cordillera, Colombia. Selbyana 16 2: 185 195.
Figure 1 . A tree divided into zones by number. Zone Three is defined here as the area within a 2.5m radius circle centered upon the hole at its first major branching. Modified from Johansson 1975.
Figure 2 . Species and family richness of vascular epiphytes by location. Figure 3 . Species richness of vascular epiphytes by family.
Figure 4. Jaccard index by species versus difference in height to Zone Three RÂ² = 0.131, P < 0.059. Points labeled ÂInÂ‚ were grouped in the same height category i.e. tall Âƒ tall or short short comparison and ÂOutÂ‚ were from different categories i.e. tall short used in the Mann Whitney U test U = 26; P < 0.002. Figure 5 . Comparison of sp ecies richness for the negatively associated families Orchidaceae and Araceae by tree. Canopy of tree 2 is superficially small and closed, while trees 4, 5, and 6 have large, open canopies.