2 Background

One-hundred twenty-eight land snail collections from 18 sites are analyzed here. The land snails were collected during archaeological and paleontological investigations carried out during cultural resources management projects associated with construction of Barbers Point harbor and Ko Olina resort adjacent to the harbor (Fig. 1). The investigated sites include ten sinkholes without cultural modification, four sinkholes that appear to have been modified for use as garden pits by Hawaiians, two habitation sites, and two natural marshes. The minimum number of individuals (MNI) identified in the collections is 120,210, with individual collections ranging in size from 69 to 21,816 MNI. The number of collections that we analyze at each site ranges from 2 to 16. The sites with relatively few collections include two marsh sites, both habitation sites, and modified sinkhole sites. Several unmodified sinkholes, including especially sites 1710-1,³ 9574,⁴ 9647-2, 9659-1, and 9661-2, yielded nine or more stratigraphically ordered collections.

Figure 1 Locations of sites mentioned in the text. Topographic information from 1928 USGS quadrangle.

The potential of Kalaeloa land snail collections to contribute to environmental reconstruction was first recognized by Kirch (1978), who identified eleven species of land mollusks in three small collections from site 9574. The small sample of identified mollusks and circumstances of their collection precluded detailed interpretation. However, Kirch (1978) was able to set out field methods that subsequent investigators should follow: samples should be collected from each stratigraphic layer in a site, preferably at 5-10 cm intervals; fine-meshed screens should be used to maximize recovery of small individuals; and a survey of the contemporary land snail fauna at Kalaeloa should be made "to serve as a control for interpretation of the subfossil assemblages" (Kirch 1978:4).

The first major application of quantitative snail analysis in Hawai`i was carried out by Christensen and Kirch (1986), who identified a minimum number of 21,376 individual snails in 26 collections from five sites, including two unmodified sinkhole sites (9574, 9670-P1), two human habitation sites (2700-1, 2701-1), and one sinkhole site possibly modified for use as a pit garden (2701-3). Their analysis and interpretation of these materials strongly influenced subsequent studies.

Central to the analysis is division of taxa into ecological groups (Table 1). Among native land snail taxa, "native, extinct" taxa are presumed extinct, either globally or locally, and "native, extant" taxa are found today at Kalaeloa.⁵ In addition to native land snail taxa, other snails identified in Kalaeloa deposits include introduced taxa, most importantly the Polynesian introduction Lamellaxis gracilis and the ubiquitous post-Contact introduction Gastrocopta servilis, and various aquatic taxa, the most abundant of which, Assiminea nitida, is found in deposits that are dry today and is an important marker of wetter conditions.

Christensen and Kirch's analysis centered on sinkhole site 9574, whose 85 cm deep, three-layer, stratigraphic sequence suggested that it was "of critical importance in assessing long-term ecological change" (Christensen and Kirch 1986:72). The site showed striking evidence of change over time in the relative proportion of native, extinct taxa. The seven collections from the bottom 70 cm of the stratigraphic column yielded between 71% and 85% native, extinct taxa. Native, extinct taxa decline to 67% of the collection from 5-15 cm below surface and fall to 20% in the surface collection. None were found in the leaf litter. In the absence of radiometric dates for the stratigraphic sequence, an absence keenly felt (Christensen and Kirch 1986:76), Christensen and Kirch attempted to date the decline of native, extinct taxa through the stratigraphic distribution of temporal index sub-fossils. Taxa introduced to Hawaii by Polynesians in the first millennium A.D. and found in sinkhole deposits include the snail L. gracilis, scincid and gekkonid lizards, and Polynesian rat (Rattus exulans). Post-contact (A.D. 1778) introductions include the snail G. servilis, house mouse (Mus musculus), and rats other than Polynesian rat. They found L. gracilis, lizards, and R. exulans in the top 25 cm of the stratigraphic column, a distribution that correlates with the decline in relative proportion of native, extinct taxa. The unexpected presence of G. servilis throughout the stratigraphic column, often as live or freshly deposited shells, was believed to have resulted from contamination in the period between the site's excavation (Sinoto 1978:21-24) and sampling for snail collections. Thus, it appeared that decline of extinct taxa began during the Polynesian period and that Polynesians were responsible for "extinction of much of the native land snail fauna" (Christensen and Kirch 1986:76).

A decline over time in relative proportion of native, extinct taxa was found at all other sites (except site 2701-3, which was dominated by the aquatic snail A. nitida), but stratigraphic sequences differed from the sequence at site 9574. Variations from the pattern of change documented at site 9574 were not interpreted, however. These include very gentle relative decline of native, extinct taxa at sinkhole site 9670-P1, their strong presence (>40%) at the end of that sequence, and evidence for mixing of the index sub-fossil G. servilis at sites 2700 and 2701.

Subsequent investigations at Kalaeloa greatly augmented the number of land snail collections available for analysis, identified six taxa not found previously, and provided the basis for radiometric dating of certain sinkhole sequences. Davis' (1990) excavations at twelve sites, including eight sinkholes (1710-1, 2700-18, 2701-8, 2706-6E, 2711-28, 9647-2, 9659-1, and 9661-2), two modified sinkholes (2705-7 and 2706-8B), and two marshes (1715-1 and 1716), yielded 97 collections⁶ and a total of 96,388 identified individuals (Christensen 1995). Six snail taxa not previously identified at Kalaeloa were found in small numbers. These include an amphibious taxon,⁷ one historically introduced taxon, and three native, extinct taxa, including Pacificella spp., Achatinella mustelina, and Amastra (Metamastra) cf. subrostrata, the latter two of which are generally restricted to moister, forested regions. Their presence, along with the presence of a sinistral Lyropupa, well outside their known environmental ranges led Christensen to sound a note of caution about the possible precision of ecological inferences drawn from sub-fossil land snail collections. He also recognized that snails used as index sub-fossils by Christensen and Kirch (1986) were regularly found at depths far too great for their known ages. Although he provided some general guidelines for distinguishing in situ occurrences of these taxa from secondarily deposited specimens, it is not possible to apply these guidelines to the data as they are reported. In any case, the availability of ¹⁴C dates at some sites diminished the importance of index sub-fossils as a basis for chronological inferences.

Davis' dating program focused on sinkhole sites, whose long stratigraphic sequences provide information on change that is not yielded by habitation sites, marshes, or modified sinkholes. Davis recognized that sinkholes showed stratigraphic regularities and that these might be used to posit a regional depositional sequence that would apply generally to sinkholes. These stratigraphic regularities were described by Allen (1995), who defined depositional units based on formative conditions of sediments and proposed a three-phase stratigraphic sequence. At the base of the stratigraphic column is a "basal diagenetic deposit characterized by carbonate silts weathering directly off the reefal substrate" (Davis 1990:182).⁸ The basal diagenetic deposit was buried by sediments derived from "mass wasting and major episodes of structural collapse of the sinkhole walls" (Davis 1990:182). These structural collapse deposits were buried in turn by a dark-colored, loamy deposit "formed in transported sediments" (Davis 1990:182). Davis hypothesized that "(c)hange in the dominant mode of deposition through the column provides a tentative model of ecological change" (Davis 1990:182) for the region.

Davis concentrated his dating efforts at sites 9659-1 and 1710-1, submitting seven avian bones (procellariid) for age determinations. Bones came from each of the depositional units, with the structural collapse deposit providing five dates, and the transported sediment and basal diagenetic deposits one each. ¹⁴C ages returned by the laboratory (discussed below) are consistent both within and between the two sites, and offer general support for Davis' hypothesis of a regional sinkhole depositional sequence. However, calibration of the ¹⁴C ages on procellariid bone is not a straightforward matter. Procellariid birds feed exclusively on sea creatures and thus take in their carbon from the ocean reservoir, which has an apparent ¹⁴C age of approximately 400 years. Calibrating with an atmospheric curve would yield calibrated ages that are too old by several hundred years. Christensen (1995) reported "uncorrected" calendar dates derived by subtracting the conventional ¹⁴C age from 1950, a practice roughly equivalent to calibrating with an atmospheric curve.⁹ Davis recognized the problem with dating materials from the ocean reservoir, but instead of following calibration procedures established by the ¹⁴C dating community (Stuiver et al. 1986), chose instead to devise his own calibration curve based on ¹⁴C dates on paired samples of different materials from his Kalaeloa excavations. Despite the differences in calendar dates, with those reported by Christensen (1995) averaging about 500 years older than those reported by Davis (1990), both authors interpreted the ¹⁴C dates as supporting the hypothesis that decline of native, extinct taxa was due to Polynesian influence. Christensen (1995: 254) concluded that the first evidence for decline in land snails dated to A.D. 690, a time believed by many prehistorians to be early in the Polynesian era (Kirch 1985; Hunt and Holsen 1991). Davis (1990: 333) dated the period of greatest change in the land snails to A.D. 1200-1500, a period generally regarded as characterized by rapid population growth and expansion of settlements (Kirch 1985).

Recently, Cowie (1992) identified 1,968 MNI in six collections from modified sinkhole site 2717-23. These collections showed a decline in relative proportion of native, extinct taxa over time, which Cowie, following Christensen and Kirch (1986), interpreted as due to "the decline of native plant communities and the increasing dominance of introduced plants combined with increased human activity" (Cowie 1992:J-4).