John T. Longino, Jonathan Coddington, and Robert K. Colwell. 2002. The ant fauna of a tropical rain forest: estimating species richness three different ways. Ecology 83:689–702.

Field Sampling and Sample Processing

Fogging. – Canopy fogging was carried out using the general procedures discussed in Erwin (1983a,b), Adis et al. (1984), and Stork (1988). In an individual fogging event, 40 1-m² funnels were suspended beneath the crown of a canopy tree, the crown was fogged with synthetic pyrethrins, and the funnels were harvested after a 2-hour drop time. The results in this paper are from 18 trees fogged in 1993-1994, and 6 trees fogged in 1999-2000. Additional details on these foggings (tree species, date, location, etc.) may be found in Longino and Colwell (1997) and the ALAS website.

The ALAS staff of four Parataxonomists processed the 1993-1994 samples. For each funnel sample, they sorted specimens in alcohol to temporary morphospecies (temporary in the sense of pertaining only to that sample, and not being cross-referenced to other samples). For each temporary morphospecies, they point-mounted and labeled a series of 1-10 specimens, depending on their confidence in species separation (more specimens were prepared when they had doubts about the uniformity of their morphospecies). Thus, at the time of processing, decisions on what material to prepare were based on within-sample sorting by trained non-specialists. The process was repeated for each new funnel sample. Not all funnels from each fogging event were processed: 10 funnel samples were processed from each of 12 trees, and 11, 11, 11, 17, 20, and 40 funnel samples from 6 additional trees, respectively.

Longino processed the 1999-2000 samples. Most ants were identified in alcohol, with point-mounting only of those requiring it for identification or otherwise of taxonomic importance. All 40 funnels for each fogging event were processed.

In this paper, one sample is one funnel.

Malaise. – Sixteen Malaise traps were placed in light gaps and potential flyways and maintained from March 1993 to March 1994, for a total of 13 months. At the beginning and the middle of each month, the collecting bottle with accumulated arthropods was removed and replaced with a fresh bottle of ethanol. Thus each sample is approximately 2-weeks' catch from one Malaise trap. For four of the traps, the ALAS staff processed the first sample from each month for each trap, resulting in 52 samples. Ten additional samples were processed, 2 from the above traps and 8 from 3 other traps. Processing followed the same protocol used for the fogging samples.

Berlese. – Berlese samples were taken by quickly placing a 14.5 cm inside diameter, 10 cm deep PVC ring on top of the leaf litter, driving it into the ground, dislodging it, and placing the approximately 1.5 L of soil and litter in a plastic bag. The samples were returned to the laboratory on the same day and placed in Berlese funnels. Sixteen sample points were chosen, distributed evenly with respect to alluvial versus residual soils, and old growth vs. second growth forest. Twice each month a sample was taken 10 m distance and a random direction from the sample point. After 3 months, 4 outlying points were moved to new points closer to the ALAS laboratory, to improve sampling efficiency. The ALAS staff processed at least the first sample from each month, and occasionally the second sample, for each sample point, using the protocol described above for the fogging samples.

Winkler. – Winkler samples are an efficient method of sampling leaf litter ants (Besuchet et al. 1987, Olson 1991, Agosti et al. 2000). In general the method involves sifting bulk samples of leaf litter and rotten wood by agitating them vigorously in a bag above a coarse mesh screen. Litter arthropods are concentrated in the finer "siftate" that passes through the screen. Arthropods are then extracted from the siftate by a passive extraction method, in which the siftate is placed in thin mesh sacks and then suspended and enclosed within an outer cloth "Winkler bag." The Winkler bag tapers to a cup of ethanol. After loading the sample the Winkler bag is closed at the top and suspended in a sheltered location. Arthopods fall from the litter and accumulate in the ethanol, and the sample is taken off after a fixed period of time. In the present study, samples were taken off after 3 days. Methodological variations involve how the bulk litter is selected. The method used here involved a relatively subjective selection of litter to sift, and was not based on a particular area or volume of bulk litter (thus they are not directly comparable to studies of Fisher [1996, 1998, 1999a,b] and Delabie et al. [2000]). The person taking the sample moved through the habitat selecting pockets of moist litter. These accumulations of litter and rotten wood occurred between buttresses, along the sides of rotten logs, around dead stumps, and in soil depressions. When approximately 6l of siftate were obtained, the material was placed in a cloth bag and returned to the lab.

In the present study, 13 samples were taken by Longino between 1986 and 1993, 20 were taken by the ALAS staff in June and July 1999, and 6 were taken by the ALAS staff in December 1999. The 13 Longino samples and the December 1999 samples were processed by Longino. The 20 samples from June and July 1999 were processed by the ALAS staff, using the protocol described above for the fogging samples.

Barger. – N. Barger carried out a baiting study during April and May 1994. At 20 sites in mature rain forest, a pair of bait transects was established. All sites were at least 50 m apart. One transect of a pair was along the edge of cleared trail. La Selva trails are 2-3 m wide, usually either bare soil or with a thin layer of leaf litter. In mature forest they do not interfere with canopy closure. The other transect of a pair was perpendicular to the trail, beginning about 3 m away from the trail. Each transect contained 20 baits placed at 1-m intervals. The baiting platform was an 8-cm diameter plaster of paris disk. The bait was a mixture of honey and solid vegetable oil (manteca vegetal). Baits were left for 30-60 minutes, long enough for ants to locate the baits and establish recruitment trails. Ants were harvested by quickly placing the bait station in a plastic tub and then collecting the ants into ethanol. For the present study, each transect of 20 baits is one sample.

Thompson. – C. Thompson carried out a study of leaf litter ant nests during July 1995. Four belt transects, each 15 m long by 0.5 m wide, were established. They extended in the 4 cardinal directions from La Selva GIS post 700/350. The leaf litter was carefully searched by turning all leaves and dissecting all pieces of dead wood and seed husks, and all ant nests were collected. Each ant nest collection was one sample. Each sample was placed in a Berlese funnel and all ants extracted (occasionally yielding more than one species of ant).

Longino. – This is a heterogeneous set of collections made by J. Longino between 1985 and the present. The methods were the visual search techniques used by myrmecologists, and involved no quantification. Nest collections were made by looking under stones, dissecting dead wood and stems in the leaf litter, looking for nest entrances in the soil, finding ants at food resources or attracting them to baits, looking for nests in or on low vegetation and tree trunks, cracking dead and live plant stems, searching for nests in recent treefalls, and searching at night for army ant columns and other nocturnally foraging species. Some collections are single workers or nests, others contain multiple species (e.g., scattered workers collected in a recent treefall, workers collected haphazardly from the trail during a day's collecting, individual ants scattered in a rotten log). The collections were made in a variety of habitats: mature and second growth forest, open road edges and the clearings around buildings, inside the buildings themselves (to sample pest ants), and in specialized antplants. After initial collections, common or easily recognized species were ignored, and an eye was maintained for species new to the inventory. At the time of collection samples were given sequential collection numbers. For the present study, each collection number is a sample.

Other. – These collections are miscellaneous other sources of information on the ants of La Selva. They include collections made by other researchers, either in the context of other studies or haphazard collections of interest brought to the ALAS lab. The majority were collected after the inception of the ALAS project (1991), but some were collected as early as 1972. These collections have individual collection codes in the ALAS database, representing either individual specimens or "lots" (sets of specimens with the same collection data). For the present study, each collection code is a sample.

Specimen Identification

One species in the dataset, Gnamptogenys simulans, was reported from La Selva by Lattke (1995), and has not been subsequently observed. All other species identifications reported here are based on specimens examined and identified by Longino. Nomenclature follows Bolton (1995). Voucher specimens are available for each species, deposited in the Costa Rican Instituto Nacional de Biodiversidad (INBio) and other museum collections. Detailed specimen data (barcode, species, determination history, repository) are maintained in the ALAS database. Species boundaries are poorly understood for the genera Solenopsis, Tapinoma, and Paratrechina, and, where questions arose, we attempted to err on the side of lower species numbers. In each of these genera distinctive species can be recognized, leaving one highly variable "species" that may be more than one cryptic species.

In many cases morphospecies code numbers are used for species that cannot be named at the present time, due to their being new and undescribed species, or more often due to the lack of taxonomic research that would allow one to determine whether a species was new or not. As names change in the future, the synonymy should be traceable through the determination histories in the ALAS database, and the Ants of Costa Rica web site. Manuscript names (those for which the author is followed by "ms") have no taxonomic standing and are not made taxonomically available by appearing here.

Literature Cited