The primary datum in the argument for extinction by Polynesians of land snails at Kalaeloa is a decline over time in relative proportion of 18 native taxa in excavated archaeological and palaeontological collections. This pattern was first noted by Christensen and Kirch (1986) at site 9574, where native, extinct taxa are absent in leaf litter on the surface of the sinkhole. Subsequent excavations and analyses failed to yield another site where the decline of native, extinct snails was complete. At several sites this decline is pronounced (e.g. 9647-2, 9661-2), but at others the decline is more moderate (e.g. 1710-1, 9659-1, 9670-P1). At most sites the relative proportion of native, extinct taxa never declines to nil. Native, extinct taxa are absent only in the cultural layer (II) at site 2701-1 and the leaf litter of sinkhole site 9574.
Given the interpretative difficulties inherent in use of relative proportions, we analysed change over time in diversity of land snail collections, using measures of richness and evenness to characterise diversity. Richness is a straightforward measure that can take on integer values between 0 and the maximum number of taxa in the collection. Evenness is more difficult to quantify and several different measures have been proposed (Bobrowsky and Ball 1989). The measure used here, called H/Hmax by Kintigh (Kintigh 1989, 29), can take on real values between 0.0 and a maximum that depends on the richness of the collection, but usually is between 2.0 and 3.0. In an extinction event, both richness and evenness decline to 0.
The richness and evenness measures on Kalaeloa snail collections estimate diversity of land snail faunas that existed within catchments of sites where they were deposited. Like richness measures, evenness measures are sensitive to sample size, and both measures can yield poor estimates of faunal diversity if sample sizes are small (Grayson 1984, 158 ff.). The Monte Carlo technique used to estimate expected richness, discussed above, can also estimate expected evenness (Kintigh 1989) and estimates of expected richness and evenness were calculated for each of the land snail collections, using the site collection as the parent population from which samples were drawn. Use of site collections, rather than the Kalaeloa collection, as the parent population, helps to control for differences in diversity of land snail faunas among different habitats.
Figures 4-9 compare relative proportions of native, extinct taxa over time with their expected richness and evenness at six sinkhole sites. These include two sinkhole sites analysed by Christensen and Kirch (1986) and four of eight sinkholes excavated by Davis (1990) for which land snails were identified (Christensen 1995); sediments of these four sinkholes were analyzed by Allen (1995) and it is possible to assign estimated ages to their depositional unit boundaries. Error bars for relative proportions are based on one standard deviation counting errors, calculated as square root of sample size. Richness and evenness are expressed on the figures as the difference between observed and expected values generated by the Monte Carlo method. Error bars for richness and evenness represent 80% confidence intervals. It is rare at Kalaeloa to find a collection that is as diverse as expected by the Monte Carlo method, and most of the plotted differences between observed and expected diversity measures are negative. This is because most sinkhole sites show evidence for change over time in the land snail fauna, including episodes of colonisation and extirpation of taxa, which affect richness, and changes in relative proportions of taxa, which affect evenness. Because any single collection contains material from only a portion of the time that materials were deposited in the sinkhole, use of the site collection as the parent population in the Monte Carlo method often overestimates diversity of individual collections.
The distributions of introduced taxa L. gracilis and G. servilis are illustrated in the bottom panel of figures because they were used by Christensen and Kirch (1986) as index sub-fossils, as discussed above. They are plotted on a logarithmic scale as MNI + 1, to compensate for wide variation in counts from one collection to the next, many of which include nil values.
Sinkhole site 9574 is the type site for a decline in native, extinct land snails at Kalaeloa. Christensen and Kirch (1986) divided the 85cm deep stratigraphic column into three layers and collected ten samples for land snail identification. Layer I, a thin, black, relatively loose A0 and A1 horizon yielded a single sediment sample for land snail identification; Layer II, a 30cm deep, brown, silt and sand deposit with angular fragments of limestone yielded three samples; and Layer III, a 50cm deep, reddish yellow deposit of limestone breccia in a matrix of silt and sand yielded five samples. A single sample collected from leaf litter above Layer I represents the modern land snail fauna of the sinkhole.
Christensen and Kirch (1986) argue for a period of increased change and decline in relative proportion of native, extinct taxa beginning in Layer II and continuing through to their extirpation. This argument is based on the following data (Fig. 4, top). Native, extinct taxa make up 71±3% of the land snail collection in the bottom sample from Layer III at the base of the stratigraphic column and rise to a maximum of 85±6% in the top two samples from this layer. This high level is maintained through the bottom two samples in Layer II, but it declines significantly in the top sample to 67±2%. This proportion is not significantly different from the proportion at the base of the stratigraphic column and the relative proportion of native, extinct taxa in Layers II and III can be characterised as fluctuating slightly between 65% and 91% of the total collection. The relative proportion of native, extinct taxa does decline significantly in Layer I, where they make up only 20±1% of the collection. There are no native, extinct taxa in the leaf litter (native, extinct taxa were found in leaf litter at sites 9647-2 (Fig. 8) and 9661-2 (Fig. 9)).
Figure 4: Native, extinct land snail taxa at sinkhole site 9574. Top, relative proportion with 1
errors; second from top, difference of observed and expected evenness with 80% confidence intervals; third from top, difference of observed and expected richness with 80% confidence intervals; bottom, frequency of introduced taxa (note logarithmic scale)
Analyse the data using XLisp-Stat
The diversity measures for native, extinct taxa at site 9574 show no sign of a significant decline through three layers of the stratigraphic column, and it is not until the leaf litter, when both measures decline to 0, that diversity measures indicate any stress in the native, extinct land snail fauna.
It is not possible to correlate stratigraphic layers with Allen's depositional units and there are no 14C dates from the site, so absolute dates cannot be assigned to changes in land snail collections. Christensen and Kirch (1986) believed that Layer II was deposited, at least in part, during the Polynesian period, based on occurrence of small numbers of L. gracilis shells in the top two samples from the layer. This inference is by no means secure, however, as discussed below. The only layer that is securely dated is Layer I, whose rich collection of Polynesian and historically introduced land snail shells indicate that it was laid down after Polynesian colonisation and into the post-Contact period.
Christensen and Kirch (1986) correlated the onset of decline in relative proportion of native, extinct taxa with Polynesian influence. This inference was based on the stratigraphic distribution of various animals introduced by Polynesians, including the land snail L. gracilis. As the bottom panel in Figure 4 shows, however, correlation of the decline with introduction of L. gracilis is not particularly clear. L. gracilis, is first present in Layer II, but it occurs as five or fewer individuals in two Layer II collections in which it appears, and is not well-represented until Layer I, when the historically introduced taxon G. servilis is also first present in large numbers. Given pervasive mixing of small numbers of land snails down through the stratigraphic column at sites excavated subsequently (Christensen 1995), the presence of a few shells of L. gracilis in Layer II does not necessarily indicate this snail's colonisation of the sinkhole at the time Layer II sediments were deposited. The distribution of G. servilis down to the base of the stratigraphic column is apparently not the result of mixing, and was explained as contamination of sampled sediments by living snails that colonised the excavated portion of the sinkhole before land snail samples were collected (Christensen and Kirch 1986, 63). (Curiously, L. gracilis is absent in the leaf litter along with native, extinct taxa. The other introduced taxon, G. servilis, is the dominant component of the leaf litter collection.)
The history of the native, extinct land snail fauna that emerges from this analysis is very different from the history posited by Christensen and Kirch (1986) based on a consideration of relative proportion. The population of native, extinct land snails at site 9574 is stable until the loose sediments of Layer I were deposited, at which time the relative proportion of native, extinct taxa falls to levels not experienced previously. This is stratigraphically coeval with colonisation of the sinkhole by the historically introduced G. servilis, and possibly by the Polynesian introduction L. gracilis, as well. The diversity of native, extinct land snails shows no stress in the population at this time, but declines suddenly after Layer I was deposited. This indicates that a drastic change leading to extirpation of native, extinct taxa at the site belongs either to the late pre-Contact period, or more likely to the post-Contact period. There are no data here that indicate changes in the native, extinct land snail population were due to Polynesian settlement, rather than to major anthropogenic environmental changes to the region documented historically.
Sinkhole site 9670-P1 was also analysed by Christensen and Kirch (1986). The deposits at site 9670-P1 are much shallower than at site 9574, reaching a maximum depth of 27cm. The stratigraphic sequence at the site is divided into three layers. One sediment sample for land snail identification was collected from Layer I and five samples were collected from Layer II. No samples were taken from Layer III.
Figure 5: Native, extinct land snail taxa at sinkhole site 9670-P1. Top, relative proportion with 1 errors; second from
top, difference of observed and expected evenness with 80% confidence intervals; third from top, difference of observed and expected richness with 80% confidence intervals; bottom, frequency of introduced taxa (note logarithmic scale)
Analyse the data using XLisp-Stat
Christensen and Kirch (1986) hypothesized, on the basis of the distribution of L. gracilis to the base of the stratigraphic column, that the stratigraphic sequence correlated with the upper portion of the sequence at site 9574 (Christensen and Kirch 1986, 72). The land snail sequence at site 9670-P1 differs from the upper portion of the sequence at site 9574, however, in that the relative proportion of native, extinct taxa declines moderately up through the stratigraphic column and native, extinct taxa never disappear from collections. Instead, they represent 40±3% of the total land snail collection in Layer I. There is a slight, but significant, decline in richness between the sample from the base of the stratigraphic column and two samples from the top of the column, but there is no significant change in evenness.
As was the case with site 9574, it is not possible to correlate the layers at site 9670-P1 with the depositional units of Allen (1995). The distribution of introduced taxa L. gracilis and G. servilis through the stratigraphic column is consistent with the relative timing of their introductions, and supports the inference that Layer II was laid down from the Polynesian period through to the early post-Contact era, although the small number of L. gracilis shells in collections from the bottom half of Layer II suggest some interpretative caution (pace Christensen and Kirch 1986, 64). A plausible argument could be made for deposition of the bottom half of Layer II during pre-Polynesian times, with subsequent downward mixing of a few L. gracilis shells. Lacking 14C dates it is not possible to discriminate between these alternative hypotheses.
It is difficult to see in these data any effect on native, extinct taxa that can be attributed to Polynesian influence. Christensen and Kirch (1986) chose to focus their analysis on the decline in relative proportion of just two native, extinct taxa (O. uberta and E. kalaeloana) and increases in three native, extant taxa (Lamellidea spp., Tornatellides spp., and S. caduca), citing these trends as supporting evidence for the decline in native, extinct taxa documented at site 9574. The implication of this choice is that differences in the depositional histories of land snails at sites 9574 and 9670-P1 reflect a deviation at site 9670-P1 from a widespread pattern, rather than simply an expression of general variability in land snail faunas among Kalaeloa sinkholes.
Site 9659-1 was excavated by Davis (1990). His analysis indicated that all of the sinkholes he excavated at Kalaeloa 'yielded similar patterns of change in the snail assemblage' (Davis 1990, 197) and that this change was that described by Christensen and Kirch (1986) at the land snail 'type site', sinkhole site 9574. Davis (1990) illustrated the pattern of change with the land snail sequence from an excavation under the overhang at sinkhole site 9659-1, which he believed showed the pattern most clearly. The ~50cm thick sediment deposit was excavated in seven arbitrary 5cm and 10cm levels that were later assigned to four layers (Davis 1990, 182). Allen (1995) recognised three depositional units in the four layers. A total of ten sediment samples (Christensen 1995, 245) was collected for land snail identification. (The bar graphs in Davis (1995, 199) and Davis (1990, 438) indicate 15 and 13 samples, respectively, but there is no published record of samples identified in addition to the ten identified by Christensen (1995).) The ~21cm thick gravelly loam of the transported sediment deposit (Layers I and II) yielded three sediment samples for snail identification; four samples were taken from the 15-20cm of very gravelly, coarse sandy loam of the structural collapse deposit (Layer III); and three samples were collected from the 5-10cm of gravelly, very coarse sandy loam of the basal diagenetic deposit (Layer IV and Layer V from Square 4 in the exposed portion of the sinkhole).
There are two periods of marked decline in relative proportion of native, extinct taxa in the collections (Fig. 6), although the sequence itself is characterised by a first-order decrease in relative proportion over time. The first period of decline occurs at the base of the stratigraphic column where the proportion drops from 100-11% to 76±4%. The second begins at the top of the structural collapse deposit, where the relative proportion is 65±3%, and continues through to the base of the transported sediment deposit where relative proportion drops to 22±1%. Native, extinct taxa are well-represented at the top of the stratigraphic column, making up 19±2% of the collection there. (Davis describes this collection as consisting 'entirely of native survivors and more recently introduced snails' (Davis 1990, 198).) Use of the term 'survivors' is potentially confusing here, as it seems to indicate that only native, extant taxa remain. In fact, both native, extinct taxa and native, extant taxa are present in the collection.
In contrast to the moderate first-order decline in relative proportion of native, extinct taxa, richness and evenness measures indicate an increase in diversity through to the base of the transported sediment deposit. Richness declines in the transported sediment deposit and evenness dips deeply in the middle of the deposit due to a sudden peak in N. newcombi shells, which make up 25.5% of the collection (Christensen 1995, 246). These data indicate that there was some stress in the population of native, extinct taxa by the middle of the transported sediment deposit.
This sinkhole was dated by Davis and age estimates can be assigned directly to depositional unit boundaries (Section 5.1). The distributions of introduced taxa appear to substantiate the chronology derived from 14C dates. G. servilis is found only in the transported sediment deposit, and in significant quantities only in the top-most collection, so the lower boundary of the unit appears to be late pre-Contact in age. L. gracilis also occurs most frequently in the transported sediment deposit, but is found in scant numbers in the structural collapse and basal diagenetic deposits as well. Presence of this taxon in deposits laid down in the era before Polynesian colonisation, and its discontinuous distribution through the stratigraphic column (Christensen (1995, 245) reports non-apical fragments of L. gracilis from the lower two samples in the structural collapse deposit) suggest that the few shells in the basal diagenetic deposit are intrusive. Although the four MNI of L. gracilis at the top of the structural collapse deposit might indicate when this snail colonised sinkhole 9659-1, it is also possible that these few shells, like the few shells in the basal diagenetic deposit, are intrusive. If this is the case, then L. gracilis was first deposited in transported sediments, either late in the pre-Contact era or early in the post-Contact period.
Figure 6: Native, extinct land snail taxa at sinkhole site 9659-1. Top, relative proportion with 1 errors; second from top, difference of observed and expected evenness with 80% confidence intervals; third from top, difference of observed and expected richness with 80% confidence intervals; bottom, frequency of introduced taxa (note logarithmic scale)
Analyse the data using XLisp-Stat
The timing of changes in land snail collections at site 9659-1 casts doubt on the proposition that Polynesian activities were responsible for the decline of native, extinct taxa. First, the relative proportion of these taxa declines over the entire stratigraphic sequence. Onset of this first-order decline is likely to pre-date the fifth century AD (Fig. 3). Second, diversity of the native, extinct land snail population does not decline until after deposition of the transported sediments began. This depositional unit began either very late in the pre-Contact era or in the post-Contact period (Fig. 3). Finally, regardless of when the population of native, extinct taxa began to decline, either late in the pre-Contact era or in the post-Contact period, the cause of decline was insufficient to extirpate them at the site. At the top of the stratigraphic column, the population of native, extinct taxa is as diverse as it was in collections from both Polynesian and pre-Polynesian periods.
Sinkhole site 1710-1 was excavated in four layers to a maximum depth of 76cm, comparable to the deep deposit in site 9574 (Davis 1995). Land snails were identified in sixteen sediment samples from the site. The transported sediment deposit (Layers I and II), a dark, organic loam, yielded four samples; the structural collapse deposit (Layers IIIa and IIIb) yielded nine samples; and the basal diagenetic deposit yielded three samples. Avian bones from the site were dated and boundaries between three depositional units can be assigned absolute dates (Section 5.1).
Figure 7: Native, extinct land snail taxa at site 1710-1. Top, relative proportion with 1 errors; second from top, difference of observed and expected evenness with 80% confidence intervals; third from top, difference of observed and expected richness with 80% confidence intervals; bottom, frequency of G. servilis (note logarithmic scale). L. gracilis was not recovered from this site
Analyse the data using XLisp-Stat
Collections from the basal diagenetic deposit are characterised by an abundance of the aquatic snail, A. nitida, and this accounts for the low relative proportion of native, extinct taxa in these collections (Fig. 7, top). Davis believed that the sinkhole was at least partially flooded throughout the period represented by this unit (Davis 1990, 200), and basal diagenetic deposit collections strongly resemble collections from low-lying, 'aquatic' sinkholes. The structural collapse deposit marks the beginning of terrestrial land snail dominance in deposits, but native, extinct taxa never dominate collections here as they do in older collections of other sinkholes. Native, extinct taxa make up between 35±2% and 51±2% of the structural collapse deposit collections, with a small, but significant dip in the top two collections. The population of native, extinct taxa does undergo a change at the boundary of transported sediments, where their relative proportion drops to approximately 22% of the collection. It remains at this level to the top of the stratigraphic column.
Diversity measures indicate that native, extinct taxa were not under stress during the period represented by basal diagenetic and structural collapse deposits. Both diversity measures rise slightly over this time. The transition to transported sediments saw a marked decline in diversity, with both measures showing a significant drop from previous levels.
The decline in diversity occurs in either the late pre-Contact or post-Contact periods; 14C dating does not distinguish between these possibilities (Fig. 3). Distributions of introduced snails are also little help in this regard. The Polynesian introduction L. gracilis is absent from the site. The occurrence in small numbers of historically introduced G. servilis in most structural collapse deposit collections is almost certainly a result of mixing, and gives some indication of the potential magnitude of this problem. The taxon is present throughout the transported sediment deposit and this might be interpreted as an indication that these deposits are post-Contact in age. Although absolute numbers of this taxon rise a bit in the lower two collections from the transported sediment deposit, their relative proportion does not increase until the top two collections. Thus, an argument could also be made that G. servilis shells found below the top two collections of the transported sediment deposit are intrusive.
Excavations at site 9647-2 revealed 65cm of deposit in four layers beneath a cobble fill (Davis 1995). The transported sediment deposits (Layers II, III, IVa, and a portion of Layer IVb) yielded nine sediment samples for land snail identification; four samples were taken from structural collapse deposits (Layer IVc and a portion of Layer IVb); and a single sample was collected from the basal diagenetic deposit (Layer V) (this sample was taken adjacent to the sample column (Christensen 1995, 243)). (The layer designations used here are those reported by Davis (1995) and by Allen (1995). They designate rock fill at the site as Layer I; the first sediment deposit is Layer II. Layer designations used by Christensen (1995) ignore rock fill and begin with Layer I at the top of the sediment deposit. The Layer IVb sample was taken adjacent to the sample column (Christensen 1995, 243).) A sample was also taken from leaf litter at the surface of the deposit.
Collections from site 9647-2 reveal two periods of rapid decline in relative proportion of native, extinct taxa. At the base of the stratigraphic column, native, extinct taxa make up 96±4% of the collection from the basal diagenetic deposit (Fig. 8). This is followed by a large drop to 50±3% at the base of the structural collapse deposit. The structural collapse deposit is characterised by increases in relative proportion of native, extinct taxa, which reach 97±6% at the top of the unit. Relative proportions decline slowly through the bottom three collections in the transported sediment deposits. The collection from 22-24cm below surface yielded 59±3% native, extinct taxa, a proportion slightly greater than the collection from the base of the structural collapse deposit. The second period of rapid decline brings the proportion to 14±1%, and it remains in the range 7-17% through to the top of the stratigraphic column and into the leaf litter.
The first period of rapid decline in relative proportion of native, extinct taxa is mirrored by a decline in diversity. The collection of native, extinct taxa from the basal diagenetic deposit is the most diverse from the site. Richness and evenness both decline significantly through the transition from basal diagenetic to structural collapse deposits. The second period of rapid decline in relative proportion of native, extinct taxa is accompanied by a slight drop in richness, but this is barely detectable with the materials and techniques at hand. Both diversity indices indicate little change through the transported sediment deposit, which yields values of the indices that are within ranges seen previously in structural collapse deposits.
Figure 8: Native, extinct land snail taxa at sinkhole site 9647-2. Top, relative proportion with 1 errors; second from top, difference of observed and expected evenness with 80% confidence intervals; third from top, difference of observed and expected richness with 80% confidence intervals; bottom, frequency of introduced taxa (note logarithmic scale). The sample from the basal diagenetic deposit was taken adjacent to the sampling column at a depth of 45-60cm below surface. This sample has been given an arbitrary depth of 70cm below surface to keep the deposits in their correct stratigraphic order
Analyse the data using XLisp-Stat
Under the assumption that depositional units are penecontemporaneous among sinkholes, the first period of decline in the population of native, extinct taxa at site 9647-2 took place in the early centuries of the first millennium AD and the second in the late pre-Contact or post-Contact period. The stratigraphic distribution of the Polynesian introduction L. gracilis appears to contradict the estimated age of the second period of decline, however. Interpretation of the stratigraphic distribution of introduced snails is not straightforward, and the collections yield evidence for disturbance down to the base of the stratigraphic column. A single L. gracilis individual is present in the collection from the basal diagenetic deposit (non-apical fragments of G. servilis were also found in this collection (Christensen 1995, 240)), and it is found with G. servilis deep in structural collapse deposits. These contexts are stratigraphically inferior to four collections in which both taxa are absent. Single individuals of L. gracilis are found midway through transported sediment deposits, but it is not until the top of the stratigraphic column that either introduced taxon is found in significant numbers. The mode of L. gracilis distribution is earlier than that of G. servilis, but both modes are later than the second period of decline in relative proportion of native, extinct taxa at the site. If the assumption of penecontemporaneity in depositional unit boundaries holds, then L. gracilis was not established at site 9647-2 until very late in the pre-Contact period, or more likely the post-Contact period. (Allen suggests that the formation of the site 9647-2 sinkhole might have been later than other sinkholes in the region, based on the low density of bird bones and the apparent on-going structural collapse at the site (Allen 1995, 43). One or two 14C dates from the base of the stratigraphic column would likely provide the evidence needed to discriminate between this hypothesis and the hypothesis that depositional regimes were penecontemporaneous throughout the region.) If the penecontemporaneity assumption is violated at site 9647-2, then all that can be said is that the second period of decline pre-dated establishment of an L. gracilis population at the site.
The sediment deposits of site 9661-2 were excavated in six arbitrary 6cm and 10cm levels (Davis 1995, 331) (sediment deposits at site 9661-2 were capped by limestone rubble that was excavated as Level 1 and assigned to Layer I). Six sediment levels were later assigned to four natural layers and grouped into two depositional units. Basal diagenetic and structural collapse deposits are present at the site, but transported sediment deposits are absent. Seven samples for land snail identification were recovered from the structural collapse deposit and a single sample was taken from the basal diagenetic deposit. A sample was also collected from leaf litter at the surface of the sediment deposit.
The relative proportion of native, extinct taxa declines steadily from the base of the stratigraphic column, where they make up 76>±5% of the collection, to its top, where the proportion is reduced to 9±1% (Fig. 9). The collection from leaf litter shows a slight, though significant, increase in relative proportion to 20±4%.
Richness and evenness measures indicate a significant decline in diversity of native, extinct land snail population near the top of the structural collapse deposit. The population of native, extinct taxa never regains its richness, but evenness recovers to previous levels by the top of the deposit and maintains this level into the leaf litter. Fluctuations in diversity in the early portion of the sequence are insignificant.
Figure 9: Native, extinct land snail taxa at site 9661-2. Top, relative proportion with 1 errors; second from
top, difference of observed and expected evenness with 80% confidence intervals; third from top, difference of observed and expected richness with 80% confidence intervals; bottom, frequency of introduced taxa (note logarithmic scale). The site lacks a transported sediment deposit.
Analyse the data using XLisp-Stat
Both introduced land snails are found throughout the structural collapse deposit, and a few individuals of G. servilis were recovered from the basal diagenetic deposit. If depositional units at this site were correctly identified, then large numbers of G. servilis in the structural collapse deposit indicate substantial mixing of the stratigraphic column (cf. Christensen 1995, 293). Otherwise, occurrence of G. servilis through most of the structural collapse deposit indicates a post-Contact age for the upper boundary of the depositional unit, and a very recent age for decline in richness of native, extinct taxa at the site.
Although the timing of events at site 9661-2 is problematic, the site shows a common pattern of change in native, extinct land snails. Against a background of steady decline in their relative proportion beginning in the basal diagenetic deposit, diversity indices remain stable until the recent end of the sequence. Decline in diversity at site 9661-2 takes place at a date that is likely to fall in the post-Contact period, if large numbers of G. servilis are any indication of age, but the decline cannot be dated with sufficient precision to rule out a date in the pre-Contact era.
© Internet Archaeology
URL: http://intarch.ac.uk/journal/issue10/dye/kalaeloasu2.html
Last updated: Tue May 29 2001