4.0 Habitats of Kalaeloa land snails

The `Ewa plain, of which Kalaeloa is part, is often described as featureless, a desolate, barren expanse of limestone. From a land snail's perspective, however, this could be rough and variable terrain. An obvious limiting factor is water; Kalaeloa is the driest corner of O`ahu with a median annual rainfall of less than 60cm (Giambelluca et al. 1986) and no running streams. The surface of the land is for most of the year dry, hot and uninviting. But this is only skin deep. Not far beneath the surface the water table fills cavities in the limestone, creating subterranean pools. In some places sinkholes are deep enough to reach the water table, and water standing in them creates a unique habitat. Even where sinkholes do not extend down to the water table, sediment trapped in them provides fertile soil for vegetation, relatively close to a ready supply of water and with a degree of protection from wind.

The distribution of land snails at Kalaeloa indicates that, for many taxa, this was a patchy habitat. This shows most clearly at the low-lying sinkhole sites 2701-3, 2700-18 and 2701-8 (see map), where the aquatic taxon A. nitida predominates to the near exclusion of all other taxa. It shows more generally in differences in ubiquity of taxa among excavated collections. Eleven of 28 land snail taxa are widely distributed at Kalaeloa, occurring in at least 13 of 18 site collections (Table 3). They include four abundant native, extant taxa (excluding only Lyropupa (Mirapupa) spp. among native, extant taxa), four native, extinct taxa, and the historically introduced G. servilis. A good example of the ability to live in a wide range of environments is the native, extant snail Succinea caduca, which is often found in 'extremely arid conditions' (Christensen and Kirch 1986, 60). S. caduca appears to thrive in wet conditions, as well. It comprises nearly the entire collection recovered from marsh site 1716 and makes up an important component of the collection from marsh site 1715-1.

Five taxa have restricted distributions at Kalaeloa, occurring in 7-12 site collections. These include three native, extinct taxa, the aquatic snail, A. nitida, and the Polynesian introduction, L. gracilis. In spatial terms, the Kalaeloa invasion of the Polynesian introduction L. gracilis appears to have been much more tentative than that of the historic introduction G. servilis. L. gracilis is absent from aquatic sinkholes, marshes, habitation site 2701-1, and sinkhole 1710-1. In contrast, G. servilis is found in every site except the marshes, and where it co-occurs with L. gracilis it is always found in greater numbers.

The twelve remaining taxa have very restricted distributions at Kalaeloa, occurring in 1-6 site collections. These are nearly all native, extinct taxa, most of which are rare in Kalaeloa collections, although the common taxon L. ovatula is also very restricted in its distribution.

The habitat most favoured by land snails at Kalaeloa was the sinkhole. The most diverse collections are from sinkhole sites, with habitation sites, marshes, and 'aquatic' sinkholes yielding markedly less diverse collections (Table 4). Richness, or the number of taxa in a collection, is the measure of diversity used here. Richness is strongly correlated with sample size (see Grayson 1984) and a straight comparison of richness might be misleading because sinkhole sites generally yielded the largest collections (Table 4). A Monte Carlo method has been proposed for estimating expected richness for a collection of a given size drawn from a specified parent population (Kintigh 1989) and this method has been applied to each of the sites, using all land snails identified at Kalaeloa as the parent collection. Controlling for influence of sample size in this way does little to alter the relative diversity of sites: sinkhole sites retain their place as the site type that yields the richest collections (Table 4).

Most sites yielded collections that are less rich than expected by the Monte Carlo procedure, as indicated by negative values in the columns labelled 'Range (o-e)' (observed value minus expected value) in Table 4. The small collection from marsh site 1716, for example, was expected to yield as many as 15 taxa or as few as 12, but in fact yielded only two, a difference of 10 to 13 taxa. Only sinkhole site 9647-2, with 22 taxa, yielded a collection as rich as expected by the Monte Carlo procedure. This result is due to the uneven distribution of taxa among excavated sites. The parent collection in the Monte Carlo procedure included all land snails recovered at Kalaeloa, which comprises collections from the full range of sampled snail habitats. Many excavated sites represent habitats that fall outside the environmental ranges of one or more taxa. Thus, the parent collection used in the Monte Carlo procedure represents an unrealistically rich population for comparison with individual sites, and the procedure over-estimates expected number of taxa at most sites. This feature of the procedure does not diminish its utility in a comparative context and it does highlight the degree of land snail habitat variability at Kalaeloa.

The diversity of land snail collections from sinkhole sites is partially due to the presence of several taxa that are absent from the collections collected at other site types. Most of these are rare taxa with very restricted distributions, including Pacificella sp., Achatinella mustelina, Amastra subrostrata, A. umbilicata, and the sinistral Lyropupa. However, it also includes two abundant taxa, Pupoidopsis hawaiiensis and Nesopupa newcombi. Other taxa, including Cookeconcha sp. and L. gracilis are only rarely found outside of sinkholes. For these taxa, sinkholes appear to have been islands of suitable habitat in a sea of harsh, forbidding terrain.

This inference that sinkholes were islands of suitable habitat is based on a model that posits primarily local deposition of land snails in sinkholes, with relatively little input from the surrounding landscape. There are two lines of evidence that support this model. The first comes from the two aquatic sinkhole sites 2701-3 and 2701-8, which contain collections composed almost exclusively of the aquatic snail, A. nitida. At site 2701-3, A. nitida makes up between 89% and 98% of collections (Christensen and Kirch 1986, 70), and at site 2701-8 proportions range from 78% to 99% (Christensen 1995, 219). Clearly, input from the surrounding landscape is minimal in these instances; some terrestrial taxa are expected to live on sinkhole walls and on vegetation that grows there. A second line of evidence comes from Layer II at habitation site 2700-1 and from the partially mixed Layer II at habitation site 2701-3, which both yielded collections of low diversity composed primarily of abundant, widespread taxa. Both of these layers represent pre-habitation deposits, and their land snail collections presumably reflect the nature of land snail faunas on the open plain. Although these collections are small, and thus stand a good chance of underestimating faunal diversity, they do not contradict the inference that the land snail fauna of the open plain was not diverse and was composed of taxa that are widely distributed through a range of environments. We believe this evidence supports the plausibility of the hypothesis that land snail collections from sinkholes are derived primarily from the fauna of the sinkhole itself, with little input from the surrounding landscape. However, we recognise the possibility of a washed-in component, and point to the continued desirability of detailed study of depositional processes at work in sinkholes.

© Internet Archaeology URL: http://intarch.ac.uk/journal/issue10/dye/kalaeloase4.html
Last updated: Tue May 29 2001