Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Insectes soc. 47 (2000) 313– 316 0020-1812/00/040313-04 $ 1.50+0.20/0 © Birkhäuser Verlag, Basel, 2000 Insectes Sociaux Research article Ant nests in tank bromeliads – an example of non-specific interaction Nico Blüthgen 1, M. Verhaagh 2, W. Goitía 3 and Nils Blüthgen 4 1 2 3 4 Botanisches Institut der Universität Bonn, Meckenheimer Allee 170, D-53115 Bonn, Germany Current address: Rainforest CRC, P. O. Box 6811, Cairns, Qld 4870, Australia, e-mail: [email protected] Staatliches Museum für Naturkunde, Erbprinzenstrasse 13, D-76133 Karlsruhe, Germany, e-mail: [email protected] Universidad Simón Rodríguez, Apartado 47925, Caracas 1010, Venezuela, e-mail: [email protected] Theoretische Biologie, Humboldt Universität, Invalidenstrasse 43, D-10115 Berlin, e-mail: [email protected] Received 2 September 1999; revised 20 April 2000; accepted 10 May 2000. Summary. Four species of epiphytic tank bromeliads on an island in the Orinoco river in Venezuelan Amazonia were inhabited by 13 ant species from four subfamilies. None of these ant species are known as specialised plant-ants. A Monte Carlo randomisation test showed that ants were randomly distributed among host plants: (1) there was no association between particular ant species and bromeliad species, and (2) there was no vertical stratification of the ant community between bromeliads sampled on the ground and at two height classes in trees. This contrasts with the few published data on the distribution of ants on terrestrial myrmecophytes and epiphytes, respectively, to which we applied the same analytical method. liads are often inhabited by ant colonies (Schimper, 1888; Bequaert, 1922; Wheeler, 1942; Dejean et al., 1995), especially intermediate types with characteristics of both phytotelm and myrmecophytic bromeliads (e.g., Aechmea bracteata: Dejean and Olmsted, 1997). In such unspecific hosts, ant-inhabitants can be expected to be unspecific as well (Skwarra, 1934; Davidson and McKey, 1993). In four species of tank bromeliads of the lowland Amazonian rainforest corresponding to the “variable tank type” in Dejean et al. (1995), we examined whether ant inhabitants were randomly distributed or structured, using randomisation tests, and if ant taxa could be considered opportunistic or specialised. We use our study to recalculate and discuss other reported ant-plant matrices by applying the same statistical methods. Key words: Formicidae, Bromeliaceae, Monte Carlo methods, randomisation, commensalism. Materials and methods Introduction Myrmecophytes – plants offering highly modified structures that provide nesting space for ants – are widespread in tropical lowland rainforests and often regarded as classical example of co-evolution (Janzen, 1966; Davidson and McKey, 1993; Jolivet, 1996). Ant-inhabitants of myrmecophytes were found to be relatively host-specific at least on a local scale (Fonseca and Ganade, 1996). In less co-evolved systems, however, one would expect a relatively low degree of specificity between partners (Jordano, 1987). Bromeliads include a few “myrmecophytic” species with morphological modifications, but more commonly “non-myrmecophytic” species such as typical tank bromeliads (phytotelm types) (Benzing, 1970, 1990; Huxley, 1980; Davidson and Epstein, 1989). Nevertheless, tank brome- The study area is located on an island in the upper Orinoco river near La Esmeralda, Venezuela, two km south of the Surumoni crane plot (3°8¢ N and 65°38¢ W). Parts of the island and study area were seasonally flooded. Mean annual rainfall is 2000 mm (Szarzynski and Anhuf, in press). Our study took place between June and August 1998 during the wet season. In the wide regions of the terra firme rainforest that had been studied in the surrounding, the population density of bromeliads and other epiphytes is relatively low (Engwald et al., 2000), which seems typical for lowland Amazonian rainforests (Grubb et al., 1963). In our study site, however, trees were bearing a conspicuously high density of epiphytes, possibly due to the moist conditions on the island. This situation enabled us to examine 77 bromeliads from four species, all sampled within an area of approximately 0.1 ¥ 1 km along the river bench of the Orinoco. (This area contained a considerably higher density of bromeliads, but many of them were not accessible to us). The species were Guzmania lingulata, Tillandsia adpressiflora, Vriesea procera, and V. rubra (Bromeliaceae: Tillandsioideae). We sampled bromeliads that were fallen on the ground (17 plants, most commonly large plants of V. procera that survived and sometimes flowered on the ground) as 314 N. Blüthgen et al. Ants in bromeliads well as epiphytes from 23 trees or shrubs that were partly accessed with the aid of the single-rope technique. For each bromeliad, size parameters were recorded (number, length and width of leaves, height and width of the rosette). All leaf bases were checked for ant nests, and ant colony size was roughly estimated. Ants were recorded only when eggs or larvae were observed, collected in alcohol and deposited in the National Museum in Maracay (Venezuela) and Natural History Museum in Karlsruhe (Germany). For analysing the level of species sorting (compartmentalisation) of the ant-plant matrix, we used a randomization method that is directly comparable with the study by Fonseca and Ganade (1996) on terrestrial myrmecophytes (chi-square statistics were not permitted, since most expected cell entries were too low, see Fowler and Cohen, 1996). For this analysis, 25000 random matrices were generated with the same row and column totals as in the empirical matrix (computer program based on the algorithm from Patefield, 1981). For each matrix with r rows and c columns, the test statistics, T = ∑ (a r, c · log a r, c ) was obtained, including all cell entries ar, c (the equation is derived from log-likelihood statistics). Highly structured matrices (implying a high degree of compartmentalisation between species, see Fonseca and Ganade, 1996) result in higher values of T as random matrices, while overdispersed matrices yield the lowest T. We compared the statistics of our empirical data (Tobs) with the distribution of the statistics of all generated random matrices (Tran). The significance level p of the difference between Tobs and the random sample was calculated as p = min (PL , PU ), where PL is the proportion of all Tran being equal or smaller than Tobs , and PU is the proportion of all Tran being equal or higher than Tobs (Manly, 1997). (The computer program is available from Nils Blüthgen or under http://itb.biologie.hu-berlin.de/~nils/stat/). Results Of 77 individual bromeliads studied here, 43 (56%) were lacking ant nests. In 25 bromeliads, we found nests of a single ant species, while in seven cases two ant species co-occurred in different regions of the same bromeliad, and three versus four different ant species in one plant were found a single time each, respectively. Ant-free bromeliads were found in all four species, but occurred most frequently in Tillandsia adpressiflora (in 24 of 34 individuals). Overall, 13 ant species were nesting in the plants (Table 1). They belonged to eight genera from four subfamilies: Cyphomyrmex, Pheidole, and Wasmannia (Myrmicinae), Dolichoderus (Dolichoderinae), Paratrechina (Formicinae), Pachycondyla, Platythyrea, and Odontomachus (Ponerinae). All frequent ant species (> 2 nests) occurred in at least two different bromeliad species. In some cases, ant species were found to occupy two or three bromeliads on a single tree or within a small area on the ground. This might be due to multiple nests of a single colony and cannot be considered independent; thus replicates within the same area were excluded from statistical analyses of the matrix (see legend in Table 1). More commonly, however, ant inhabitants varied when epiphytes were sampled from the same tree. Ant species were not compartmentalised between bromeliad species. The test statistics (Tobs.) of the ant-by-plant matrix (Table 1a, without numbers in brackets) was 26.8 and was similar to the statistics obtained by random simulations (mean Tran.= 25.4, ± 2.6 S.D.). 29% of all values for Tran were above or equal to Tobs and 71% below or equal to Tobs , hence the observed matrix was not significantly different from random matrices (p = 0.29). Furthermore, the ant community was not significantly structured between epiphytes on the ground or two height classes in trees. The statistics of the ant-by-height matrix (Table 1b) did not differ significantly from random matrices (Tobs = 26.7; Tran = 24.0, ± 2.3; p = 0.12). Ant colony sizes ranged from 50 to 200 ants, as in most species, to approximately 1000 ants (Pheidole sp. G). There were weak, but significant positive correlations between ant colony size and certain size parameters of the bromeliads (Spearman’s rank correlation between colony size and Table 1. Ant nests (a) in four bromeliad species, and (b) in different height levels above ground. Numbers are given for the occurrence of interactions on different trees or regions on the ground. If interactions were repeated on the same tree or ground site, the total of recorded interactions is shown in brackets. In cases marked with *), two different epiphyte species occurred in close proximity and were not considered independent in (b), so that the total in (a) is higher than in (b). Reproductive ants (alates) occurred in ant species marked with R). Ant species Cyphomyrmex minutus R) Cyphomyrmex rimosus Dolichoderus laminatus R) Pachycondyla villosa Paratrechina sp. A Paratrechina sp. C R) Pheidole sp. D R) Pheidole sp. E Pheidole sp. F R) Pheidole sp. G Platythyrea angusta Odontomachus haematodus R) Wasmannia auropunctata (a) Bromeliad species (b) Height above ground Guzmania lingulata Tillandsia adpress. Vriesea procera Vriesea rubra on ground 1–4 m 5 – 18 m – – – – – 2 3 – – – – – – 3 *) – – – 1 – 3 *)*) 2 *) – – – 1 *) 1 (2) 2 *) (3) 2 – – – 1 5 *) (7) 1 – – – 3 – – – 1 1 – 1 1 *) (2) 1 1 (2) 1 (2) 1 1 *) – 3 (4) 1 – – 1 – 3 (5) 2 (3) 1 (2) – – 3 (4) – – 1 – – – 3 6 (8) 1 – – – 1 – 1 (2) – 1 1 – 1 2 – – 1 (2) 1 – 1 (2) Insectes soc. Vol. 47, 2000 number of leaves (r = 0.41), or colony size and plant height (r = 0.43; p < 0.05). Some ant nests were found in relatively dry conditions between the outer leaves, others occurred in wet parts of the leaf rosette. Reproductive alates were found in colonies of six ant species during the study (Table 1). Discussion In tank bromeliads as a model for non-specific host plants, ant inhabitants were found to be non-specific as well and probably highly opportunistic in respect to host plant species and height above ground. Among the ant species or genera recorded here, there are no obligate or common inhabitants of myrmecophytes (Benson, 1985; Davidson and Epstein, 1993; some uncertainty remains for morphospecies of Pheidole). The two small species of fungus growing Cyphomyrmex (tribe Attini) are known to nest in soil or rotten wood on the ground as well as in suitable arboreal substrates like epiphytes or epiphytic humus. The very common Wasmannia auropunctata is an opportunistic coloniser of a broad scale of natural cavities including litter, dead wood, bark, epiphytes etc. throughout the whole vertical height range. Most Paratrechina and Pheidole species in the neotropics nest in the soil or in rotten wood on the ground, but nests of different species have also been found in the lower vegetation and in tree crowns including epiphytes. It is unclear if the encountered species are true arboreal ones. A ground dwelling species is Odontomachus haematodus nesting in the soil or between plant material near the ground while nests of Platythyrea angusta are most commonly found in rotten wood or cavities in living or dead trees. The only known true arboreal species in our survey are Dolichoderus laminatus and Pachycondyla villosa, neither known as obligate epiphyte inhabitant, although at least the latter is often found nesting in certain epiphytes (Brown, 1975; Dejean and Corbara, 1990; MacKay, 1993; Dejean et al., 1995; Verhaagh, unpubl. data). All ant genera and many of the ant species recorded in our study, overlap with other reports of ant faunas in bromeliads (Skwarra, 1934; Wheeler, 1942; Dejean et al., 1995; Dejean and Olmsted, 1997). Furthermore, co-inhabitation of two or more ant nests within a single plant was frequent, also reported by other authors (Skwarra, 1934; Dejean et al., 1995; Dejean and Olmsted, 1997). The inclusion of both tree- and ground-dwelling ants indicates that bromeliads might function as refuges for partially flooded terrestrial habitats as in the study area here. We found no statistical evidence for stratification between ant species in bromeliads of different height levels (unlike Brühl et al., 1998; Longino and Natkarni, 1990 for other ant faunas). The only quantitative community analysis covering various myrmecophytes so far, was carried out by Fonseca and Ganade (1996) who focused on terrestrial myrmecophytes in the Amazonian rainforest near Manaus. Their ant-plant species matrix was shown to be significantly structured. Data by Cabrera and Jaffé (1994) in Venezuelan rain- Research article 315 forests lead to the same conclusion that terrestrial myrmecophytes were compartmentalised: when the same randomisation test is applied to their data, the statistics of the observed matrix is significantly larger than the statistics of the random matrices (Tobs = 242.9; Tran = 206.8 ± 4.0; p < 0.001). In a study of bromeliads from a low inundated forest in Mexico by Dejean et al. (1995), three Tillandsia species can be a posteriori classified as “myrmecophytic” or “myrmecophytic-like” (> 50% of individual bromeliads inhabited by ants), and four species as “variable tank type” (< 40% inhabited). Recalculation of their matrix and application of the same statistics as above shows that ant inhabitants in myrmecophytic-like and myrmecophytic bromeliads are highly structured (Tobs = 259.0; Tran = 197.5 ± 4.8; p < 0.001). Bromeliads of “variable tank type” can also be considered structured, although to a lesser extent (Tobs =17.2; Tran =13.1 ± 2.0; p = 0.04). These species – in contrast to the “variable tank” bromeliads in our study – may exhibit a greater interspecific variation of microhabitat conditions that could account for the species sorting of ants. In conclusion, all studies of myrmecophytic plants showed a structured ant community in contrast to our findings of a random structure in non-myrmecophytic plants. However, none of the cited studies treated the problem of overlapping ant colonies in neighbouring plants (leading to pseudoreplicates in the ant-plant matrix), hence the degree of compartmentalisation might be overestimated. To evaluate, if the degree of species sorting could be a result of specialisation in ants and/or plants, more quantitative community studies are needed, particularly about commensalistic, nonspecific ant-plant systems. Acknowledgments Thanks are due to Deutsche Forschungsgemeinschaft (DFG) (Ba 605/ 6-2) and Studienstiftung des deutschen Volkes for financial support, to the Ministerio del Ambiente y de los Recursos Naturales Renovales in Venezuela (MARHR and SADA) and the Austrian Academy of Science for logistical support of the project, to Silke Engels and Arnhild Althof for field assistance, and to an anonymous referee for valuable comments on the statistics used. References Benson, W.W., 1985. Amazon ant-plants. In: Key Environments – Amazonia (G.T. Prance and T.E. Lovejoy, Eds.). Pergamon, Oxford, pp. 239–266. Benzing, D.H., 1970. An investigation of two bromeliad myrmecophytes: Tillandsia butzii Mez., Tillandsia caput-medusae E. Morren, and their ants. Torrey Bot. Club Bull. 97: 109–115. Benzing, D.H., 1990. Vascular Epiphytes, General Biology and Related Biota. Cambridge Univ. Press, Cambridge, pp. 354. Bequaert, J., 1922. Ants in their diverse relations to the plant world. Bull. Am. Mus. Nat. Hist. 45: 333–584. Brown, W.L. Jr., 1975. Contributions toward a reclassification of the Formicidae. V. Ponerinae, tribes Platythyreini, Cerapachyini, Cylindromyrmecini, Acanthostichini, and Aenictogitini. Search Agriculture (Entomology) 15: 1–115. Brühl, C.A., G. Gunsalam and K.E. Linsenmair, 1998. Stratification of ants (Hymenoptera: Formicidae) in a primary rain forest in Sabah, Borneo. J. Trop. Ecol. 14: 285–297. 316 N. Blüthgen et al. Ants in bromeliads Cabrera, M. and K. Jaffé, 1994. A trophic mutualism between the myrmecophytic Melastomataceae Tococa guianensis Aublet and an Azteca ant species. Ecotropicos (Soc. Venez. de Ecol.) 7: 1–10. Davidson, D.W. and D. McKey, 1993. The evolutionary ecology of symbiontic ant-plant relationships. J. Hymenopt. Res. 2: 13–83. Davidson, D.W. and W.W. Epstein, 1989. Epiphytic associations with ants. In: Vascular plants as epiphytes (U. Lüttge, Ed.), Springer, Berlin, pp. 220–233. Dejean, A. and B. Corbara, 1990. Predatory behavior of a neotropical arboricolous ant: Pachycondyla villosa (Formicidae: Ponerinae). Sociobiology 17: 271–286. Dejean, A. and I. Olmsted, 1997. Ecological studies on Aechmea bracteata (Swartz) (Bromeliaceae). J. Nat. Hist. 31: 1313–1334. Dejean, A., I. Olmsted and R.R. Snelling, 1995. Tree-epiphyte-ant relationships in the low inundated forest of Sian Ka’an Biosphere Reserve, Quintana Roo, Mexico. Biotropica 27: 57–70. Engwald, S., V. Schmit-Neuerburg and W. Barthlott, 2000. Epiphytes in rain forests of Venezuela – diversity and dynamics of a biocenosis. In: Results of worldwide ecological studies. Proc. 1st Symp. A.F.W. Schimper-Foundation (S. W. Breckle, B. Schweizer and U. Arndt, Eds.), G. Heimbach Verlag, Stuttgart-Hohenheim, pp. 425–434. Fonseca, C.R. and G. Ganade, 1996. Asymmetries, compartments and null interactions in an Amazonian ant-plant community. J. Anim. Ecol. 65: 339–347. Fowler, J. and L. Cohen, 1996. Practical Statistics for Field Biology. Wiley and Sons, Chichester. 227 pp. Grubb, P.J., J.R. Lloyd, T.D. Pennington and T.C. Whitmore, 1963. A comparison of montane and lowland rain forest in Ecuador. I. The forest structure, physiognomy, and floristics. J. Ecol. 51: 567–601. Huxley, C.R., 1980. Symbiosis between ants and epiphytes. Biol. Rev. Camb. Phil. Soc. 55: 321–340. Janzen, D.H., 1966. Coevolution of mutualism between ants and acacias in Central America. Evolution 20: 249–275. Jolivet, P., 1996. Ants and Plants – an Example of Coevolution. Backhuys Publishers, Leiden, 303 pp. Jordano, P., 1987. Patterns of mutualistic interactions in pollination and seed dispersal: connectance, dependance asymmetries, and coevolution. Am. Nat. 129: 657–677. Longino, J.T. and N.M. Nadkarni, 1990. A comparison of ground and canopy leaf litter ants (Hymenoptera: Formicidae) in a neotropical montane forest. Psyche 97: 81–93. MacKay, W.P., 1993. A review of the New World ants of the genus Dolichoderus (Hymenoptera: Formicidae). Sociobiology 22: 1– 148. Manley, B.F.J., 1997. Randomization, Bootstrap and Monte Carlo Methods in Biology. Chapman & Hall, London, 399 pp. Patefield, W.M., 1981, An efficient method of generating random R x C tables with given row and column totals. Applied Statistics 30: 91–97. Schimper, A.F.W., 1888. Die Wechselbeziehungen zwischen Pflanzen und Ameisen im tropischen Amerika. Bot. Mitteil. Tropen 1: 1–98. Skwarra, E., 1934. Ökologie der Lebensgemeinschaften mexikanischer Ameisenpflanzen. Z. Morph. Ökol. Tiere 29: 306–373. Szarzynski, J. and D. Anhuf, in press. Micrometeorological conditions and canopy energy exchange of a neotropical rain forest (Surumonicrane-project, Venezuela). Pl. Ecol. Wheeler, W.M., 1942. Studies of Neotropical ant-plants and their ants. Bull. Mus. Comp. Zool. Harv. Univ. 90: 1–262. To access this journal online: http://www.birkhauser.ch