DRAFT
MAY 26, 1998
PLEASE DO NOT CITE
Abstract. 1. Here we use a new approach of comparing a well-known regional fauna (Costa Rica) with samples from distant sites (Barro Colorado Island, Panama, and the Chiquibul Forest Reserve, Belize) to estimate beta diversity - the turnover of species with distance.
Key words. beta diversity; biodiversity; parasitoids; Pimplinae
Introduction
Estimates of the number of species on Earth depend on understanding beta diversity - the turnover of species composition with distance. One must know whether species have wide-spread or localized distributions. In recent years there has been much speculation on the number of insect species on Earth. Wilson (1988) estimated that there are between 5 and 30 million species. These numbers are extrapolated from the number of species already described, and the higher estimate is based largely on the work of Erwin on the species-rich beetle fauna of Central and South America (Erwin & Scott, 1980, Erwin, 1982). Erwin's data show that only 1 percent of species of Coleoptera are shared among four different forest types near Manaus, Brazil (Erwin, 1983). Erwin's (1988) data also show that only 2.6 percent of the Coleoptera species are shared between an upland forest in Manaus, Brazil and a similar site that is about 1,500 kilometers away in Tambopata, Peru. Erwin (1982, 1988, 1997) has estimated from these data that there may be 30 million, and perhaps even as many as 50 million, species of insects throughout the world. However, May (1988, 1992), Stork (1988, 1997), and Gaston (1991) concluded that more quantitative studies within and between sites, as well as across taxa, must be completed before these estimates can be accepted. Here, we report one such study.
Presently there are few estimates of insect beta diversity (Hespenheide, 1994). A major problem is attempting to compare species assemblages across sites. It is difficult to find sites that are similar enough in regards to climate, habitat type, and other environmental factors, while also being geographically distant enough to be relevant. Problems also arise in sampling species-rich faunas thoroughly enough for appropriate statistical comparisons to be made. For many species-rich insect taxa, it is a monumental task to collect all species from a site. Moreover, it is a far from trivial exercise to know what proportion of a site's fauna has been sampled as a function of collecting effort and method (Brown and Feener, 1995; Colwell and Coddington, 1994; McGowan, 1996). Previous studies of beta diversity, such as that of Erwin (1988), have usually compared one site with another, i.e. one localized habitat with another. The current study presents a new approach to estimating beta diversity - comparing a localized "site" with a larger "region" that includes a number of sites. Here we compare the Pimplinae (Hymenoptera: Ichneumonidae) from two sites (Barro Colorado Island (BCI), Panama and Las Cuevas, Belize) with a well-collected region (Costa Rica). Thus, we circumvent much of the difficulty of exactly matching BCI and Las Cuevas with ecologically equivalent, distant sites in Costa Rica.
Materials and Methods
Barro Colorado Island is located at 9° 9' North, 79° 51' West and was created in 1914 when the Chagres River was dammed during the construction of the Panama Canal. It is now completely isolated from the mainland and consists of approximately 1,500 hectares. The dry season on BCI lasts from late December or early January until late April or early May. Over 90 percent of the total annual rainfall, averaging 2,600mm, falls during the wet season, which comprises the rest of the year. For a complete site description of BCI see Leigh and Wright (1990). BCI is located approximately 600 km from the centre of Costa Rica.
Las Cuevas is a site in the centre of the Chiquibul Forest Reserve, Belize,
located at approximately 9° North, 83° West. This
area has a mean annual rainfall of 1,500 mm, with most falling between May and January and
a drier season from February until April. The climate and forest are described by Johnson and
Chaffey (1973). Las Cuevas is approximately 1000 km from the centre of Costa Rica.
For this study, we used 2 Malaise traps, designated Trap 319 and 320, on BCI. We located the traps in old-growth tropical moist forest, with Trap 319 being located under a Terminalia canopy tree and Trap 320 being located under an Anacardium tree. The specimens were collected from these traps from June of 1992 to September of 1994. We mounted and identified all pimpline wasps to species. For a complete listing of methods see Bartlett (1997).
In Las Cuevas, Belize, we ran 6 Malaise traps over the period September to December 1997 (18 trap-months of effort). The traps were in semi-evergreen forest, 25-30 m tall. We identified and tabulated all pimpline specimens to species.
The specimens from Costa Rica were collected as part of extensive efforts to inventory the
Hymenoptera of Costa Rica (Gaston et al., 1996; Hanson & Gauld, 1995). They were collected at
numerous sites throughout Costa Rica, over extended time periods, and using Malaise traps and
a variety of other collecting
methods. A total of 182 species of Pimplinae are known from Costa Rica.
Of these, 144 species are considered by Gauld (1991) and 38 new species, by Gauld, Ugalde, and
Hanson (1998). These references also list the sites, time periods, and collecting methods for the
Pimplinae in Costa Rica.
Results
Table 1 gives a complete listing of the species identified from BCI and Las Cuevas. In total, we identified 29 species of Pimplinae from 957 specimens collected on BCI. Of these, 28 have been found in Costa Rica. Only one Panamanian specimen (identified in Table 1 as Neotheronia sp. 1) appeared to be of a previously unidentified species.
Similarly, we report that 21 of the 25 pimpline species sampled from lowland forest in Belize occur in Costa Rica. Our sampling yielded 400 specimens of 25 species. Of these 21 have been collected in Costa Rica, one is a common Mexican species not found in Costa Rica, and three appeared to be of previously unidentified species (identified in Table 1 as Neotheronia sp. 2, Neotheronia sp. 3 , and Clistopyga sp. 1).
Gauld (1991) and Gauld, Ugalde, and Hanson (1998) give a complete listing of species identified in Costa Rica.
Discussion
Thus, based our results from Panama, Belize, and Costa Rica,
we conclude that pimpline species from lowland tropical forests overall have extensive
ranges throughout Central America.
Without an accurate estimate of the species richness for at least one area in a beta diversity study, estimates of beta diversity depend on sampling effort. In an extreme case, imagine that 20,000 to 40,000 beetle species are all present at both of two sites. Randomly sampling 1,000 species at each site would in all likelihood yield results similar to Erwin's, not because of a high turnover in species between sites but because of high overall species richness. In contrast, the proportional overlap in species expected between a site and a well-known fauna changes little as a function of site sampling effort.
Estimates of beta diversity based on between-site comparisons are fraught with intrinsic site-specific differences that confound distance effects. Sites matched at one level are likely to differ on finer examination. While Erwin sampled upland forest in both Brazil and Peru, his sites likely differed in one or more of the myriad of important edaphic, biologic, and climatic variables. In contrast, a site-region comparison, such as for the Pimplinae, or a region-region comparison, minimizes this problem. Consequently, we expect site-region studies to produce lower, more accurate estimates of beta diversity than site-site comparisons.
The results we present cast doubt on the higher estimates of the global number of insect species. Either (1) wasp and beetle species differ in turnover with distance, (2) Erwin (1988) sampled beetles across major geographic regions while our wasp study within Mesoamerica did not, or (3), as we fear, there are methodological problems with many previous site-site studies that confound our understanding of the distribution of species. Future efforts to refine biodiversity estimates should focus on site-region or region-region comparisons rather than on site-site comparisons. In-depth, regional studies across taxa, such as the planned All Taxa Biodiversity Inventory of the Great Smoky Mountains National Park (Kaiser, 1997), will provide a foundation for ultimately resolving this issue.
Acknowledgements. We thank Paul Hanson, Bonifacio de Leon, Saturnino Martinez, Pam Mitchell, Elizabeth Skillen, Jesus Ugalde, Sondra Ward, Joe Wright, and Lisa Wright for help with sampling and sorting. We thank Patty Gowaty, Mark Hunter, Dan Janzen, Mike Kaspari, Albert Meier, and Elizabeth Skillen for comments on previous drafts of the manuscript. This work was supported in part by the U. S. National Science Foundation (DEB 9522581 & DEB 9642121 awards to John Pickering), a Mellon Award to John Pickering and Michael Sharkey from the Organization of Tropical Studies and the Smithsonian Tropical Research Institute, and the Smithsonian Institution's Environmental Sciences Program.