The program of quantitative sampling was carried out in February, March, and April of 2001. The methods used were Malaise traps, flight intercept traps, Berlese sampling, Winkler sampling, blacklight sampling, and hand collecting by Longino. Return to the 2001 expedition home page for details of the sampling program.
A total of 175 ant species was captured and identified or sorted to morphospecies. Unambiguous new taxa include a new species of Lenomyrmex (and the first record of this genus for Costa Rica), two new species of Pyramica, a new species of Gnamptogenys, and a new species of Simopelta.
|
|
|
Lenomyrmex sp. nov.
The trapping methods that are directly comparable between the 1070m site and La Selva are Malaise traps and Berlese sampling. Winkler samples are not directly comparable because samples from La Selva were "grab" samples, taken haphazardly until approximately 6l of sifted litter were obtained, while samples from the 1070m site were replicate samples of 1 square meter plots, using the "miniWinkler" method of Fisher. However, one "grab" Winkler sample taken at the same site in 1986 yielded 27 species. A smoothed species accumulation curve for the 1070m miniWinkler samples reached 26.7 species after three samples. Thus a rough approximation is to assume the equivalence of one grab sample and three miniWinkler samples.
Inventory of hyperdiverse taxa is usually incomplete, and species accumulation curves usually are closely approximated by logarithmic curves:
in which S(t) is the number of species, t is a measure of effort (number of samples, number of species occurrences), and a and z are curve fitting parameters. Logarithmic curves have no asymptote, so how do you compare species richness for two communities for which total richness is unknown (and perhaps unknowable)? A measure of relative species richness of two sites might then be the ratio of species richness for equal sampling effort. I have found that this ratio is not stable with increasing sample size. However, the ratio appears to approach an asymptote as sample size increases. This makes sense if the curves are closely fit by logarithmic curves, because as t becomes very large, the ratio of two logarithmic curves becomes the ratio of the two z parameters, a constant. Thus a novel approach to comparing species richness between two sites is to calculate the ratio of the z parameters of the fitted logarithmic curves.
When comparing Malaise trap samples, La Selva is more diverse than the 1070m site (Figure 1A,B). The ratio of z parameters suggests that the 1070m site is about half as species rich as La Selva. The results for Berlese are similar but show the 1070m site being about 70% as rich (Figure 2A,B). The Winkler results show the 1070m site being about 50% as rich (Figure 3A,B). Finally, when all methods are combined, La Selva appears more than twice as rich as the 1070m site (Figure 4). Traditional rank abundance plots also show that the 1070m site is distinctly less diverse (Figure 5), with a distinctly steeper overall slope (rather than the 1070m curve simply being a truncated version of the La Selva curve due to undersampling).
Subfamily composition differed slightly between the 1070m site and La Selva. Myrmicinae and Ponerinae had proportionally more species at the 1070m site, while Dolichoderinae, Formicinae, and Pseudomyrmecinae had proportionally fewer.
The complementarity between two sites is the sum of the number of species unique to each site divided by the total number of species known from both sites. If there is no species overlap complementarity is 1; if there is total overlap it is 0. A problem with complementarity is that it is not stable with sampling effort and does not incorporate abundances. A circumstance where all species are widespread but differ dramatically in abundance can result in complementarity values that are high early in a sampling program, but decrease with increasing sampling effort.
As a way to see this phenomenon, I have plotted complementarity between the 1070mm site and La Selva as a function of an "abundance cut-off value." For each community sample I calculated complementarity based only on those species with proportional abundance greater than a cut-off value in both commmunities, for cut-off values 0 (all species), 0.001, 0.002, 0.004, 0.008, and 0.016 (Figure 6). What this reveals is that for the common species at each site there is essentially no overlap. However, with increased sampling overlap increases because species common at one site begin to show up as rare species at the other site.
The ratio of z values from logarithmic curves may also be a useful measure of complementarity between two sites, by comparing within-site curves to the combined curve for both sites. This "combined curve" approach was developed in Longino and Colwell (1997) as a method of assessing complementarity of sampling methods, but could also be used to compare communities. For example, this approach shows that the 1070m site is a distinct community because the combined curve is definitely higher than the La Selva curve (Figure 4).
Figure 1. Species accumulation curves for ants in Malaise trap samples, contrasting La Selva and the 1070m site on the Volcan Barva transect. A. Curves based on number of samples. B. Curves based on number of species occurrences.
Figure 2. Species accumulation curves for ants in Berlese samples, contrasting La Selva and the 1070m site on the Volcan Barva transect. A. Curves based on number of samples. B. Curves based on number of species occurrences.
Figure 3. Species accumulation curves for ants in Winkler samples, contrasting La Selva and the 1070m site on the Volcan Barva transect. A. Curves based on number of samples. B. Curves based on number of species occurrences.
Figure 4. Species accumulation curves for ants, all methods combined. The two lower curves are for the 1070m site and La Selva, the upper curve is for both sites combined.
Figure 5. Rank abundance plots for ants at the 1070m site and La Selva. Abundances are number of species occurrences in all samples and all sampling methods.
Figure 6. Complementarity between the 1070m site and La Selva, based on species with proportional relative abundance greater than a cut-off value in both sites.
Page author:
John T. Longino, The Evergreen State College, Olympia WA 98505 USA.longinoj@evergreen.edu