Human Impacts on the Nearshore Environment: an Archaeological Case Study from Kaua‘I, Hawaiian Islands1

Human Impacts on the Nearshore Environment: An Archaeological Case Study from Kaua‘i, Hawaiian Islands1

Alex E. Morrison2,3 and Terry L. Hunt2

Abstract: Archaeology provides a long-term framework to document prehistoric resource use and habitat modification. Excavation at Nu‘alolo Kai, Kaua‘i, yielded a large, well-preserved shellfish assemblage. Analysis determined the susceptibility of mollusk communities to human foraging pressures in the past. Some coral reef and intertidal species, such as Turbo sandwicensis and Strombus maculatus, declined in abundance as a result of heavy exploitation. In contrast, shoreline mollusk communities remained fairly stable through time. Archaeo- logical research provides baselines for modern conservation efforts and fisheries management.

The oceans in their expanse over the globe of environmental disturbances, long-term hu- provide over 99% of the habitable environ- man use, and postcolonial fisheries on marine ment for life on our planet (Woodward ecosystems. Jackson (2001) and Jackson et al. 2000). Yet even this vast and once believed in- (2001) showed how retrospective records exhaustible ‘‘last frontier’’ has suffered from from coastal ecosystems offer baselines that cases of overexploitation, depletion, and even contrast with modern field studies. ‘‘We can- ecological collapse. With concerns of im- not generate realistic null hypotheses about pending catastrophe, Woodward (2000) noted the composition and dynamics of ecosystems that marine biologists are frightened by the from our understanding of the present alone, scale and complexity of the emerging crisis. since all ecosystems have almost certainly Human influences are widespread, but the changed due to both human and natural envi- greatest impact has occurred in nearshore and ronmental factors’’ ( Jackson et al. 2001:630). coastal zones (0–50 m in depth) in close prox- Historical perspectives are essential not imity to the majority of the planet’s human only for establishing baselines and providing populations. a deeper view of ecosystem change, but also Although many impacts are recent, Jack- for constructing realistic models to serve con- son et al. (2001:629) argued that in a number servation biology (Lyman and Cannon 2004). of cases, ‘‘overfishing and ecological extinc- Discerning modern trends in many species is tion predate and precondition modern eco- often hindered by a lack of knowledge con- logical investigations and the collapse of cerning prehistoric human impacts and other marine ecosystems in recent times.’’ Those relevant factors that may have contributed to authors pointed out the importance of histor- the current demography of these resources ical data in better understanding the impacts (Wing and Wing 2001, Butler and Delacorte 2004). In the Hawaiian Islands, catch records for the endemic shellfish ‘opihi (Cellana spp.) show that since approximately 1900 human- 1 Manuscript accepted 9 July 2006. 2 Department of Anthropology, University of Hawai‘i induced pressures from heavy fishing have at Ma¯noa, 2424 Maile Way, Saunders Hall 346, Hono- led to a substantial decrease in the abundance lulu, Hawai‘i 96822-2223. of these shoreline fauna (Cobb 1905, Kay and 3 Correspondence: [email protected]. Magruder 1977). Consequently, legislation was developed in the late 1970s to enforce Pacific Science (2007), vol. 61, no. 3:325–345 ‘opihi fishing regulations (Hawai‘i Depart- : 2007 by University of Hawai‘i Press ment of Land and Natural Resources 1981). All rights reserved However, these regulative measures have

325 . 326 PACIFIC SCIENCE July 2007 failed to take into consideration the long- (2001), Thomas (2001, 2002), and McAllister term historical perspective offered by archae- (2003), among others, have provided the de- ology. tails of fish and shellfish use, including cases Here we develop a case study of human of relative resource stability or depression predator-prey relations for a marine ecosys- and habitat structure modification. However, tem over the long-term historical frame pro- despite the importance of shellfish in tradi- vided by archaeology. Our research examines tional Pacific Island subsistence, few studies approximately 600 yr of shellfish exploita- have addressed the question of human im- tion from Nu‘alolo Kai, Kaua‘i, extending rec- pacts, including resource depression, on ords of ‘opihi and other mollusk species use shellfish populations in the past. Two recent centuries before historic documentation. We studies by Thomas (2001, 2002) show the believe that zooarchaeological and paleoeco- promise of studying archaeological marine logical analyses are invaluable for properly mollusk assemblages from Pacific islands. modeling ecosystem change and habitat man- Documenting human predator-prey relations agement (Lyman and Cannon 2004). Analyz- with these often-abundant, protein-rich re- ing an archaeological assemblage of shellfish sources has important implications for other from a dated, stratified context, we use for- aspects of subsistence change in the Pacific aging theory models to test hypotheses for islands. human-induced resource change. By applying formal modeling developed specifically to Evolutionary Ecology, Foraging Theory, and examine spatial and temporal patterns in re- Resource Depression source use, this study is able to gain fine- grained resolution regarding habitat exploita- As a theoretical framework, evolutionary ecol- tion and human impacts. ogy allows us to test predictions regarding human behavior in a variety of ecological contexts. Predictions are typically derived Background from formal optimization models. The logic Over the past several decades, archaeolo- of foraging theory and other evolutionary gists, anthropologists, and others have written ecology models is based on fundamental con- about indigenous societies as conservationists cepts in Darwinian evolution (Pianka 1983). inhabiting pristine environments, or as agents Foraging models are usually constructed of widespread impacts and wholesale habitat in three parts: decisions, currencies, and con- destruction (Kirch 1997, Smith and Wishnie straints (Stephens and Krebs 1986). Decisions 2000). The latter perspective has recently are the alternative choices an individual been popularized and perhaps sometimes ex- faces while foraging, such as which resources aggerated (e.g., Diamond 2005). A dichotomy should be pursued while others are ignored. of conservation and wholesale destruction Once an item is chosen for processing and oversimplifies the complexity of human consumption, other prey are excluded. In for- predator-prey relationships in the historical aging theory, the decision component refers context of marine environments. to criteria that are important to prey choice. Several studies provide evidence for the Examples of these criteria are prey size, loca- nature and extent of human impacts on ma- tion, and ease of capture. rine ecosystems, including overfishing and its To evaluate the likelihood of alternative consequences (e.g., Hildebrandt and Jones decisions, a currency must be added to the 1992, Butler 2000, Grayson 2001, Mannino model as a measurement scale. The most and Thomas 2001, 2002, Lyman 2003, Er- common currency used in foraging models is landson et al. 2004, Reitz 2004, Steneck et al. net energy gained foraging ðRÞ, which can be 2004). In the Pacific Islands, Swadling (1976, expressed as the energy acquired from a prey ð Þ 1986), Anderson (1981), Kirch and Yen item per unit of time spent foraging Ft . (1982), Allen (1992, 2002, 2003), Butler Foraging time equals the time spent search- . Archaeological Case Study from Kaua‘i Morrison and Hunt 327

ð Þ ing for a prey Ts , plus the time spent at the faunal remains in combination with ar- ð Þ handling the item Th . For archaeological tifact distributions. studies, Broughton (1994, 2001) suggested According to the prey choice model, for- that prey size can be used as an approxima- agers should include various prey types in ð Þ tion for energy gained per item Ef . The the optimal diet breadth as long as those currency rate ðRÞ can be expressed as: items have a net energy return ðRÞ greater than the average return rate of higher-ranked ¼ ð þ ÞðÞ R Ef / Ts Th 1 items. If the abundance of these items does not change and technological innovations do Measuring foraging time directly in ar- not occur, foraging return rates should re- chaeological applications is impossible be- ð Þ ð Þ main stable. Because changes in prey choice cause search time Ts and handling time Th are dependent upon the encounter rates of cannot be observed. Anderson (1981) noted high-ranked items, when these items decrease that for shellfish studies this does not present in abundance foragers may choose a wider ar- a substantial problem because mollusks are ray of lower-ranked items and foraging effi- relatively immobile and usually processed us- ciency will subsequently decrease. ing similar roasting methods. Consequently, Decreases in the encounter rates of high- individual foragers are likely to expend the ð Þ ranked items may be caused by a number of same amount of energy searching Ts and ð Þ factors. Environmental perturbations, such as handling Th prey regardless of their species. climate change and natural and/or human- If search time and handling time are assumed induced habitat modification, can potentially equal for all items, then the denominator of affect the distribution of prey species, ulti- equation 1 is irrelevant and the new equation mately decreasing the encounter rates of becomes: high-ranked items (Byers and Broughton ¼ ð Þ R Ef 2 2004, Wolverton 2005). More-favorable cli- matic conditions may also lead to higher Moreover, because an approximation for ca- abundances of certain species, leading to an ð Þ loric energy gained per prey item E f is the increase in foraging efficiency. size of the prey, the currency rate ðRÞ is sim- Human exploitation can result in declin- ply equal to the size of the resources. ing prey encounter rates as well. Resource In many ecological and archaeological depression due to overexploitation occurs studies potential prey are ranked according when diminishing high-ranked prey leads to the energy returned upon consumption of foragers to rely on smaller, more-diverse, the item. Although in most archaeological ap- less-profitable items (Grayson 2001). To plications size is used as an approximation for demonstrate adequately that changes in prey energy gained per unit of time spent forag- encounter rates were caused by human pre- ing, in many coastal contexts prey can poten- dation, it is necessary to rule out the pos- tially be captured en masse. When prey are sibility of either environmental change or mass captured, the acquired energetic unit new technologies as important factors. In becomes the group of items rather than coastal environments, sea-level fluctuations, the individual (Madsen and Schmitt 1998, coastal reconfigurations, and geomorpho- Nagaoka 2002, Butler and Campbell 2004, logic changes can depress mollusk popula- Ugan 2005). Consequently, taking of prey tions (Butler and Campbell 2004; A.E.M. items through mass-harvesting techniques and E. E. Cochrane, unpubl. data). In the may actually result in a higher harvesting absence of any well-documented paleoenvir- return rate than when they are acquired indi- onmental records showing dramatic change vidually. Changes in technology can also sub- for Nu‘alolo Kai over the last 600 yr, we stantially affect harvesting rates by decreasing must assume that declining abundance is best handling time and effort (Winterhalder 1981) explained by human foraging rather than en- and therefore should be assessed by looking vironmental factors. . 328 PACIFIC SCIENCE July 2007

Prey species are often located in spatially were formerly less advantageous will be for- distinct clumps, referred to as patches (or aged more often. habitats). Increased hunting within resource- rich habitats can result in declining prey materials and methods abundance and greater foraging effort in less- profitable areas. The patch choice model and The Nu‘alolo Kai Excavation the marginal value theorem predict that as Nu‘alolo Kai is located along the north shore productive habitats decrease in overall prey of Kaua‘i Island (Figure 1) in close proximity return rate, foragers will begin using less- to a highly productive marine ecosystem that productive areas (Charnov 1976). Conse- includes a fringing coral reef community. quently, the ratio of taxa from productive to At Nu‘alolo Kai there are several archaeo- less-productive habitats may indicate shifts in logical sites attesting to the location’s long- patch exploitation. term prehistoric occupation. In particular, The prey and patch choice models provide ancient constructed terraces protected by a some predictions for subsistence patterns. massive rock overhang have been the focus First, decreases in the abundance of large- of extensive archaeological research. From bodied prey and increases in smaller prey 1958 to 1964, Kenneth Emory and his asso- signify declining foraging efficiency. Because ciates conducted large-scale excavations and items with high energetic returns are likely recovered thousands of well-preserved arti- taken upon encounter, an increase in relative facts (see Graves et al. 2005 for an overview). abundance of less-profitable items signifies Although these finds are important, the stan- diminishing foraging return rates. At Nu‘a- dards of excavation were not the same as lolo Kai, tracking the relative relationship be- those today, and many of the assemblages, in- tween large-bodied and small-bodied mollusk cluding some shellfish, present problems for species is a useful method for measuring for- any quantitative analyses. In an effort to re- aging efficiency. Consequently, if decreased cover comparable materials under modern ex- foraging efficiency occurs, a general increase cavation and collection standards, one of us in the relative abundance of small-bodied (T.L.H.) directed fieldwork at Nu‘alolo Kai mollusks should be quantifiable. in 1990 (see Hunt 2005 for a summary). Second, as foraging efficiency declines In 1990 T.L.H. excavated a 2 by 1 m unit substantially, the diet breadth may widen at Site K3 in an area not previously excavated enough that pursuing lower-ranked items (details provided in Hunt 2005). Excavation upon encounter becomes advantageous. In yielded an abundance of shell, other fauna, this situation the subsistence routine expands and macrobotanical remains, as well as an to include a more-diversified, generalized abundance of artifacts. A.E.M. identified and diet. However, it is possible that foraging analyzed the shellfish assemblage reported efficiency may decline without leading to here. The 1990 excavations also resulted in a changes in diet breadth, particularly in situa- new suite of radiocarbon dates. T.L.H.’s ex- tions where processing and harvesting lower- cavation offers the possibility to understand ranked items do not result in a high mean re- temporal patterns in mollusk use because of turn rate. In the study reported here, richness the large amount of shellfish remains, the and evenness analyses were used to assess excellent preservation, and the chronological changes in diet breadth at Nu‘alolo Kai. Fi- control. New research can provide a deeper nally, declines in foraging efficiency may lead understanding into this remarkable archaeo- to the inclusion of lower-ranked habitats logical deposit. where smaller-bodied mollusk species are found. According to the patch choice model Chronology (Charnov 1976, Orians and Pearson 1979), if the mean foraging return rate of high-ranked Five samples of wood charcoal submitted habitats decreases extensively, patches that from the deepest levels of Emory’s early ex- Figure 1. Map of Kaua‘i with location of Nu‘alolo Kai on northern coast (adapted from Esh 2005). Figure 2. West wall of Nu‘alolo Kai excavation. . Archaeological Case Study from Kaua‘i Morrison and Hunt 331

TABLE 1 Relationship between Radiocarbon Dates and Analytic Units Developed for the Analysis

Analytic Unit Dates (Interval) Date Period

Zone A Present–(1860–1740) 1800–present Protohistoric–Historic Zone B (1860–1740)–(1620–1520) 1570–1800 Late Prehistoric–Protohistoric Zone C (1620–1520)–(1470–1350) 1410–1570 Prehistoric cavations have suggested to some a chronol- suggests that initial occupation at Nu‘alolo ogy of occupation at Nu‘alolo Kai that began Kai began around cal A.D. 1400 (with materi- in the twelfth century (Kirch 1985). How- als in Layer XI). The age of Layer X (three ever, concluding that the rockshelter dates to dates) may extend to cal A.D. 1600 or later. the twelfth century places great faith in a sin- The dates for Layer X suggest that the re- gle radiocarbon date (GaK-1343, 840 G 70 maining layers of the deposit formed over a B.P.). As Hunt (2005:253) pointed out, many period of about 200 yr. Historic artifacts of would reject this date because it has never Layers IV–V represent the beginning of ef- been replicated in additional dates from simi- fective European contact, in this area of lar stratigraphic context of the same deposit. Kaua‘i, probably after A.D. 1800. In sum, the Four other radiocarbon determinations from chronology of occupation at Nu‘alolo Kai the deepest cultural context of the deposit may have an early beginning at around A.D. overlap at one standard deviation, placing the 1400, but much of the cultural deposit dates earliest occupation of Nu‘alolo Kai Rockshel- to later times (after A.D. 1600 [Table 1, Fig- ter between about A.D. 1300 to 1500 (cali- ure 3]). brated [see Hunt 2005]). From the ﬁeldwork of 1990 T.L.H. se- Aggregation and Sample Size lected four wood charcoal samples from se- cure stratigraphic contexts (Figure 2, Layers In the 1990 Nu‘alolo Kai excavation, 11 X and XI) and submitted them to Beta Ana- stratigraphic layers were excavated in arbi- lytic. The results provide a chronology that trary levels to provide minimal analytic units

Figure 3. Probability distribution for four calibrated dates from Nu‘alolo Kai. . 332 PACIFIC SCIENCE July 2007

TABLE 2 TABLE 3 Shellﬁsh Weight in Grams by Natural Stratigraphic Revised Analytic Aggregates after Dealing with Sample Units Size Issues

Stratigraphic Unit Shellfish Weight (g) Time Stratigraphic Shellfish Period Unit Weight (g) I 736.61 II 476.91 Zone A Layer I–VIII 15,760.5 III 1,762.112 Zone B Layer IX–X 14,037.1 IV 2,941.22 Zone C XI 16,283.4 V 3,843.7 VI 2,254.31 VII 1,413.73 VIII 2,331.92 IX 2,418.9 X 11,618.2 patterns. A potential drawback when collaps- XI 16,283.4 ing stratigraphic layers into larger analytic Total 46,081.012 units is that the analyst must rely on fewer data points for comparative analyses. This can be problematic when assessing trends on a quantitative scale. However, we opted for for analysis (Table 2). Although fine-grained greater confidence in our results at the ex- temporal divisions would have been ideal, pense of a large number of comparative units. several issues regarding sample size emerged. The final analytic units are listed in Table 3 Analyses looking at the diversity of species and these correspond to the temporal desig- and the relative proportions of prey items nations shown in Table 1. may be seriously affected by differing sample sizes (Grayson 1984). For example, measures Comparative Indices and Resource Depression of relative abundances are often strongly correlated with the size of the subsamples from To test the hypothesis that foraging pressure which they were derived. This is particularly resulted in resource depression we first com- true when the assemblage is dominated by pared changes in the amount of large-bodied small and dramatically varying sample sizes. versus small-bodied prey items within each of When aggregated by natural stratigraphic two habitat types. In this paper we use com- layers, sample size is positively correlated parative indices to track these relationships. with relative abundance in several important X . X mollusk species. For example, as the size of large taxon ðlarge taxon the sample increased, so did the relative þ ÞðÞ ¼ : small taxon 3 abundance of Turbo sandwicensis (rs 909 < ¼ : sig .000) and Nerita picea (rs 0 836, Nagaoka (2002) employed comparative in- P <:001). To solve this problem, we com- dices in her study of New Zealand foraging bined samples from aggregations of layers to patterns. Index values simply turn the ratio form assemblages (analytical units) of near of a large to small-sized taxa into an easily equal sizes. Combining these layers into tractable number between 1 and 0. High in- larger assemblages decreased the significant dex values (close to 1) reflect higher pro- ¼ : ¼ correlation (Turbo sandwicensis rs 0 5, P portions of larger items, and a smaller index : ¼ : ¼ : 667; Nerita picea rs 0 5, P 667). Al- value indicates decreasing foraging efficiency though analyzing each layer independently (Nagaoka 2002). Using an assemblage from would have created a greater number of com- the Shag Mouth site, Nagaoka documented a parative points, the strong positive correla- decrease in the moa-quail index from 1.0 at tion between sample size and prey relative level 10 to 7.5 later in the sequence. Thus, as abundance would have limited the reliability large moa were declining, smaller quail were of our results and obscured the cause of these being taken more often. . Archaeological Case Study from Kaua‘i Morrison and Hunt 333

TABLE 4 Coral Reef Mollusk Species Recovered from the Nu‘alolo Kai Excavation

Coral Reef Species

No. Species Zone A Zone B Zone C Total

1 Turbo sandwicensis 2,926.50 7,244 10,296 20,466.5 2 Strombus maculatus 3,193.01 3,676.5 2,879.8 9,749.31 3 Isognomon sp. 2,205.10 1,132.9 1,102.2 4,440.2 4 Cypraea spp. 720.3 169.2 254.1 1,143.6 5 Conus spp. 623.9 219.1 56.2 899.2 6 Periglypta reticulata 158.52 144.2 48 350.72 7 Trochus intexus 47.31 135.2 129.7 312.21 8 Rhynoclavis sp. 13.9 — 35.8 49.7 9 Tellina spp. — 33.7 — 33.7 10 Codakia punctata 19.4 — — 19.4 11 Thais intermedia 15.4 — — 15.4 12 Modulus tectus — — 14.7 14.7 13 Vexillum sp. — 10.9 — 10.9 14 Tellina sp. 10.8 — — 10.8 15 Hydantia sp. 4.6 — — 4.6 16 Umbraculum sp. 0.01 — — 0.01 17 Anadara sp. 0.01 — — 0.01 18 Trachycaridum sp. 0.01 — — 0.01 19 Cerithium sp. 0.01 — — 0.01 Total 9,938.78 12,765.7 14,816.5 37,520.98

Note: Recorded by shell weight in grams (0.01 indicates presence but negligible weight).

Patch Classification relationship between large- and small-bodied prey. Cochran’s test for linear trend is a chi- A limitation of the prey choice model is the square test comprising three components: a assumption that all prey items are searched traditional chi-square test that tests the over- for simultaneously. However, in many cir- all relationship between the two variables, the cumstances, foragers allocate time in specific variation as a result of a linear trend, and the patches and therefore do not search for all departure from the linear trend. available prey species in the environment at one time. Smith (1991) suggested that to solve this problem it is necessary to apply the results prey choice model within patches. This cre- Foraging Efficiency in the Coral Reef Patch ates the possibility for analyzing encounter rates within spatially distinct habitats. For When compared by prey size, Turbo sandwi- Nu‘alolo Kai, we used two habitat classifica- censis is the largest coral-reef taxon, with tions based on prey locations: the shoreline an average size of 90 mm (Severns 2001). In and coral reef patches (Tables 4 and 5). contrast, another abundant yet smaller prey The statistical methods used in the analysis item is Strombus maculatus, which has an aver- consist of Spearman Rank Order Correlation age size of 32 mm (Severns 2001). If foraging Coefficient, which is a nonparametric test ap- efficiency declined in the coral reef patch plied to ordinal data. This test is commonly there should be a measurable decrease in the applied to archaeological data that are ordinal relative relationship between Turbo sandwicen- rather than interval, small sample sizes, or not sis and Strombus maculatus, indicated by de- normally distributed. clining index values. We also used Cochran’s test of linear Figure 4 demonstrates a decrease in the trend to test for a significant tendency in the proportion of the large Turbo sandwicensis . 334 PACIFIC SCIENCE July 2007

TABLE 5 Shoreline Mollusk Species Recovered from the Nu‘alolo Kai Excavation

Shoreline Species

No. Species Zone A Zone B Zone C Total

1 Cellana spp. 2,553.4 414.2 652.9 3,620.5 2 Nerita picea 2,447.3 426.5 520.3 3,394.1 3 Drupa ricina 206.5 123.7 59.4 389.6 4 Morula sp. 207.4 39.6 83.1 330.1 5 Littorina spp. 116.9 47.4 22.5 186.8 6 Purpura aperta 26.2 — 27.5 53.7 7 Hipponix foliaceus 15.9 — 3.8 19.7 Total 5,573.6 1,051.4 1,369.5 7,994.5

Note: Recorded by shell weight in grams (0.01 indicates presence but negligible weight).

initially within the coral reef patch, foragers were relying predominately on Turbo sandwicensis to the near exclusion of other taxa. Over time Turbo sandwicensis declined substantially, and Strombus maculatus increased during the late prehistoric period and remained fairly stable. The decreasing index values are indic- ative of a decline in foragers relying on Turbo sandwicensis.

Foraging Efﬁciency in the Shoreline Patch Of the two patches, shoreline mollusks are Figure 4. Coral reef patch comparative index values the smallest, with most of the represented measuring the relative relationship of Turbo sandwicensis and Strombus maculatus. taxa having a mean size of under 15 mm. Thus, judging by prey size alone the shoreline patch appears to have the lowest overall foraging return rate of the two habitats in relative to the smaller Strombus maculatus this analysis. The limpets Cellana spp. (‘opihi) 2 ¼ <: (Xtrend 1,739.1, P 001). Weight mea- are the largest prey species in the shoreline surements also support this conclusion, with habitat (Kay 1979). In contrast, Nerita picea a substantial decrease in Turbo sandwicensis is a smaller prey, with a mean size of 14 mm. across the three analytic zones (Table 6). The index values for the shoreline patch The archaeological evidence suggests that do not suggest resource depression (Figure

TABLE 6 Absolute Abundance of Turbo sandwicensis and Strombus maculatus by Weight (Grams) from Zone A to Zone C (Abundance Results Conﬁrm Results from the Comparative Indices)

Species Zone A Zone B Zone C Total

Turbo sandwicensis 2,926.5 7,244.0 10,296.1 20,466.5 Strombus maculatus 3,193.01 3,676.5 2,879.8 9,749.31 . Archaeological Case Study from Kaua‘i Morrison and Hunt 335

Evaluating Changes in Habitat Use As foraging efﬁciency within a given patch declines to a point equal to or below the return rate for other exploitable patches, habitats will be added into the subsistence routine (Charnov 1976, Broughton 2001, Nagaoka 2002). At Nu‘alolo Kai this should be re- ﬂected in a shift from foraging in the coral reef patch to the shoreline habitat. To test this hypothesis it is useful to create comparative indexes between items that are found in each habitat. X . X Figure 5. Shoreline patch comparative index values mea- ðcoral reef taxaÞ ðshoreline suring the relative relationship between Cellana spp. and þ ÞðÞ Nerita picea. coral reef taxa 4 The results from the patch choice analysis demonstrates that foragers relied substantially on the coral reef patch during the prehistoric TABLE 7 and late prehistoric period, and then began Absolute Abundance of Cellana spp. and Nerita picea more use of the shoreline patch during the protohistoric period (Figure 6). Cochran’s Species Zone A Zone B Zone C Total test for linear trend supports this conclusion ( 2 ¼ 4,071.98, <:001). Cellana spp. 2,553.4 414.2 652.9 3,620.5 Xtrend P Nerita picea 2,447.3 426.5 520.3 3,394.1 Both the prey and patch indices suggest a decrease in the use of the coral reef patch. Accordingly, the shoreline patch appears to have been utilized substantially more by the 5). Foraging return rates remained fairly sta- protohistoric period. All but one shoreline ble through time. There is only a very slight species increased in abundance. Conse- drop in the index value during the late prehis- quently, the use of the shoreline patch was toric period. A lack of evidence for resource relatively stable through time. depression is also supported by weight measurements (Table 7). As Table 5 shows, all but one shoreline species increase dramatically up until the protohistoric period. The increase in the preva- lence of shoreline species correlates with a decrease in important coral reef taxa, such as Turbo sandwicensis and Strombus maculatus. Based solely on the prey choice index analyses from the coral reef and shoreline habitats, it appears that by the late prehistoric period, resource depression had occurred in the coral reef patch. However, because temporal patterns in habitat use are closely tied to both the productivity of the habitat being foraged and the productivity of nearby habitats, it is Figure 6. Habitat comparative index values measuring important to look also at the relative contri- the relationship between coral reef species and shoreline bution of prey species by habitat. species. . 336 PACIFIC SCIENCE July 2007

To provide more corroborative evidence for the trends measured by the prey and patch choice analyses, we also analyzed temporal patterns in assemblage diversity. Diver- sity measurements document changes in the number of taxa in the diet breadth as well as measure the proportional contribution of individual taxa to the entire assemblage.

Prey Diversity and Foraging Efficiency Archaeologists commonly use two different means for measuring prey diversity. The first Figure 8. Number of taxa (NTAXA) from the coral reef tracks the expansion of diet breadth by com- patch, indicating an increase in prey diversity during the paring the number of taxa (NTAXA) by time protohistoric period. period. A more-diversified, generalized diet breadth should show an increase in NTAXA. The second measure of prey diversity is evenness. Evenness is a more-sophisticated measure because it quantifies the proportional relationships among species, rather than fo- cusing simply on the number of taxa present. The total assemblage shows evidence for increased taxonomic richness during the protohistoric period (Figure 7). Both the prehistoric and late prehistoric deposits contain 17 taxa, but the protohistoric deposits include 28 taxa. To determine if expanding diet breadth is a general trend or occurs only within distinct habitats, we also assessed taxonomic rich- Figure 9. Number of taxa (NTAXA) from the shoreline ness for both the shoreline and coral reef patch, indicating stability in prey diversity. patches.

The evidence from the coral reef patch supports the trend from the entire assemblage (Figure 8). During the prehistoric and late prehistoric periods diet breadth was relatively stable. However, in the protohistoric deposits there is an increase in NTAXA from 9 to 16 items. The NTAXA assessments from the shoreline patch document a different trend (Figure 9). Diet breadth in the shoreline patch begins with seven prey items and then drops to five during the late prehistoric. By the protohistoric period there are seven items present again. Accordingly, diet breadth did not change substantially in the shoreline patch. Figure 7. Number of taxa (NTAXA) from the entire as- Measurements of NTAXA are limited be- semblage, indicating an increase in prey diversity during cause they do not reflect the contribution of the protohistoric period. individual taxa to the entire assemblage. For . Archaeological Case Study from Kaua‘i Morrison and Hunt 337 example, the inclusion of more species into the diet is evidence for an expanding diet breadth, but no details on the relative contribution of specific taxa are taken into consideration. An assemblage dominated by a few abundant taxa will have a low measure of evenness. In contrast, when all taxa are present in near-equal amounts, the assemblage is considered highly even. Measurements of evenness can supplement diet breadth analysis by offering researchers the ability to track the relative importance of specific species through time. As diet breadth expands to in- Figure 10. Prey evenness values for the entire assem- clude new items, it is necessary to look at the blage, indicating an increase in evenness. contribution of these taxa to the assemblage. In many ecological and archaeological studies Shannon’s Index is used to calculate evenness (Grayson 1984, Claassen 1998, Grayson and Delpech 1998, McAlister 2003). Shannon’s Evenness Index ðEÞ is calculated as: X ¼ ð ½ Þ ð Þ Evenness pi log pi /log NTAXA ð where pi is the proportional contribution of each itemÞð5Þ

The corresponding value will be between 0 and 1. A value of 0 demonstrates an assem- Figure 11. Prey evenness values for the coral reef patch, blage with only a single taxon present, and a indicating an increase in evenness. value of 1 indicates that all items are represented in equal amounts. In contrast to the indices used for measuring prey choice, an trast, the shoreline patch is stable, with even- increase in evenness (and consequently evenness values fairly similar in all three zones ¼ : ¼ : ness index values) indicates resource depres- (rs 0 5, P 667 [Figure 12]). sion. Because large prey items are declining A principal limitation of the use of even- in abundance, the assemblage will incorporate ness indices is that similar index values do a higher number of small less-profitable taxa. not necessarily reflect similar resource use The result will be a more even assemblage. patterns. It is possible that proportional rela- When the entire shellfish assemblage is tionships between taxa stay the same while the taken into consideration, evenness increases taxa representing those proportions change ¼ : < significantly through time (rs 1 00, P (Nagaoka 2001). Analysis of prey rankings :001 [Figure 10]). However, as with NTAXA, and relative abundances are necessary for a it is useful to calculate evenness indices based finer understanding of resource exploitation on separate habitat types to gain spatial clar- patterns. Inspection of the absolute abun- ity. dances and comparative indices given here The evenness index values for the coral supports the results of the relative abun- reef patch are similar to the results from the dances and provides conclusive evidence for entire assemblage (Figure 11). There is a the trends discussed in this paper (Table 8). general trend toward increasing evenness Inspection of the relative shellfish abun- ¼ : <: through time (rs 1 00, P 001). In con- dances reveals a number of important trends. . 338 PACIFIC SCIENCE July 2007

discussion The role of shellfish in the diet of past coastal dwellers has been the subject of debate among archaeologists (Yesner 1984, 1987, Claassen 1986, 1998, Erlandson 1988, 1991, 2001, Glassow and Wilcoxon 1988). In particular, the ability for mollusk species to sustain high-density human populations has received considerable attention in past years. For example, Claassen (1986, 1998) chal- lenged the assertion that human predation led to substantial impacts on mollusk envi- Figure 12. Prey evenness values for the shoreline patch, ronments, except in cases of heavy exploita- indicating stability in evenness. tion in rocky shoreline habitats. In contrast, Swadling (1976, 1986), Botkin (1980), Ander- son (1981), Jerardino (1997), Thomas and First, the high-ranked Turbo sandwicensis de- Mannino (1999, 2001), Mannino and Tho- creases in abundance over time. Moreover, mas (2001, 2002), and Thomas (2001) argued as Turbo sandwicensis decreases, other smaller, that sustained human exploitation can indeed less-profitable prey increase. This is apparent severely affect mollusk populations. The re- in an increase in Strombus maculatus during sults from Nu‘alolo Kai complement those the late prehistoric period. Moreover, both studies by providing evidence for human im- shoreline taxa Cellana spp. and Nerita picea pacts on coral reef species while demonstrat- also increase, but the coral reef species de- ing stability in the use of shoreline mollusk crease. communities. Finally, the relative abundance measure- In the coral reef habitat, the large Turbo ments support the diet breadth and evenness sandwicensis decreased relative to the smaller analyses. For example, inspection of the pro- Strombus maculatus, suggesting that as the en- portional contribution of the five most counter rate of T. sandwicensis declined, for- common species demonstrates an increasingly agers relied more on smaller, abundant prey. greater contribution of other taxa to the total A similar trend was also documented by Raab assemblage. The total percentage value for (1992), who used foraging theory to explain a these five items decreases from 94.9% during shift in the exploitation of Haliotis spp. by in- the prehistoric to 84.5% during the protohis- habitants of San Clemente Island in southern toric. Moreover, the relative percentage con- California. Haliotis spp. or abalone are large tribution of these prey items becomes more mollusks that can be easily acquired. In con- even through time. trast, Tegula spp. are smaller and more diffi-

TABLE 8 The Five Most Abundant Taxa (Percentages Shown in Parentheses)

Rank Zone A Zone B Zone C

1 Strombus maculatus (20.25%) Turbo sandwicensis (51.5%) Turbo sandwicensis (63.23%) 2 Turbo sandwicensis (18.56%) Strombus maculatus (26.19%) Strombus maculatus (17.68%) 3 Cellana sp. (16.2%) Isognomon sp. (8.1%) Isognomon sp. (6.8%) 4 Nerita picea (15.52%) Nerita picea (3.0%) Cellana sp. (4.0%) 5 Isognomon sp. (14.0%) Cellana sp. (2.9%) Nerita picea (3.2%) Total 84.53% 91.69% 94.9% . Archaeological Case Study from Kaua‘i Morrison and Hunt 339 cult to process. Raab suggested that the in- that capture methods can affect the return creased abundance of Tegula spp. and de- rate of certain animals. In her archaeological creasing use of abalone is evidence for study of prehistoric European rabbit (Orycto- overexploitation. lagus cuniculus) hunting, Jones showed that The diversity analysis from Nu‘alolo Kai the mass capture of rabbits through the use also demonstrates a general expansion of of warrens might have substantially increased diet breadth by the protohistoric period. The the energetic return rate of these prey items. number of taxa (NTAXA) foraged and even- Current research on the distribution of ness for the entire assemblage increase sig- bird bone artifacts at Nu‘alolo Kai (Esh nificantly through time, indicating a more- 2005; K. S. Esh and A.E.M., unpubl. data) diversified diet relying less on large mollusk will prove useful for better understanding species and more on an array of smaller the relationship between technological inno- prey. Similarly, at Palliser Bay, New Zealand, vation and the exploitation of certain marine Anderson (1981) found that heavy predation species. In the past, bone artifacts have been pressure during prehistory led to a shift from interpreted as tools for extracting the fleshy a more-specific subsistence strategy focused meat of shellfish prey (Kirch 1985). If on large shellfish species to wider diet breadth changes in the presence of bone tools can be including smaller shellfish. securely tied to the extraction of mollusk Botkin (1980) studied changes in the rela- meat, then the expansion of this technology tive importance of Mytilus californianus and may have increased the foraging return rates Protothaca staminea at the Malibu site along of less easily procured species. Research on the southern California coast by looking at these dynamic processes is needed to assess differences in the natural distribution and the relationship between Polynesian subsis- harvesting techniques of the two species. He tence economies and the technological inno- reasoned that because M. californianus has a vations documented by artifact distributions. lower procurement cost and larger size than Recent ethnoarchaeological research on P. staminea it should dominate the earlier shellfish procurement processes and transport deposits. The archaeological evidence dem- by Bird and Bliege Bird (1997, 2000) and onstrates that M. californianus was indeed Thomas (2002) highlights the potential biases heavily foraged during the early occupation when using the archaeological record to com- period. A reversal in the relative abundance pare relative abundances of large and small of these two species in later archaeological mollusk species. Bird and Bliege Bird (1997, deposits suggests resource depression and 2000) showed that among contemporary Mer- overuse of M. californianus. iam foragers of the Torres Strait, Australia, At Nu‘alolo Kai, we assume that Turbo large mollusk species with low processing sandwicensis and Strombus maculatus had simi- costs tend to be handled in the field to de- lar handling and procurement costs because crease travel time. In contrast, many smaller there is no conclusive evidence to suggest species with high procurement costs are often that harvesting technologies were different taken back to a base camp for processing for the two mollusk species. As a result, prey (Bird and Bliege Bird 1997), and when chil- size at Nu‘alolo Kai continues to be the best dren forage there is an increase in the indicator of energetic return. Because har- number of small species gathered (Bird and vesting technologies affect the overall return Bliege Bird 2000). Thomas (2002:191) also rate by decreasing energy spent acquiring suggested that ‘‘because resources often con- and preparing items, capture and procure- tain parts of high and low utility, other for- ment techniques can ultimately have an effect aging models, such as prey choice, may not on energetic return. Madsen and Schmitt anticipate variability in archaeological as- (1998) suggested that researchers should pay semblages: the mismatches between behavior more attention to technological innovations and its archaeological correlates.’’ developed for the mass capture of small- Although we are in agreement with Bird bodied prey. Jones (2004, 2006) also noted and Bliege Bird (1997, 2000) and Thomas . 340 PACIFIC SCIENCE July 2007

(2002), we suggest that the problems associ- The changes demonstrated in the Nu‘alolo ated with the nature of archaeological mate- Kai shellfish assemblage point to a few im- rial do not negate the usefulness of foraging portant issues. First, it seems likely that theory in the Nu‘alolo Kai assemblage. Bird human-induced resource depression in the and Bliege Bird’s research indicates that coral reef patch led to an increased use of biases created by child foraging and differen- the shoreline patch simply because the dis- tial field processing would create a more tribution of coral reef species declined and stable frequency of lower-ranked resources foragers exploited new areas and species. to higher-ranked ones. In contrast, ‘‘if inten- However, an alternative explanation is that sification accounts for the differences, over gathering shoreline species may have re- time the introduction of lower ranked prey quired less time investment because of close in the assemblage should correlate with a de- proximity to habitation areas as well as their pression in higher ranked resources’’ (Bird susceptibility to mass capture. Moreover, and Bliege Bird 2000:472). Mannino and coral reef foraging may have been abandoned Thomas (2002:458) also noted that Bird and in favor of shoreline harvest to allocate more Bliege Bird’s model ‘‘could account for the time to gardening and terrestrial resources. species composition of the latter midden de- Allen (1992, 2002) argued on Aitutaki, Cook posits studied by both Anderson (1981) and Islands, that when productive land became Raab (1992), although it cannot explain why scarce, offshore fishing may have decreased higher-ranked species dominated the earlier to allocate more time to protecting agro- phases.’’ nomic plots, which included domesticated In our analysis we suggest that changes in animals and gardens. A similar explanation the Nu‘alolo Kai assemblage are more ade- might account for the switch from coral reef quately explained by foraging intensification to shoreline harvesting at Nu‘alolo Kai. and resource depression rather than differen- The trends documented in the Nu‘alolo tial field processing, although the two are not Kai mollusk assemblage are also important mutually exclusive (see Cannon 2003). First, when viewed in combination with other because we take multiple lines of evidence predator-prey studies from the Pacific islands. into consideration when testing for resource A number of archaeological analyses have depression, we avoid the problems of making documented long-term human exploitation general conclusions based on the relative rela- of nearshore reef fishes and mollusks that in tionships of large- and small-bodied mollusks many instances led to measurable environ- alone. Second, although documenting pat- mental impacts (e.g., Swadling 1976, 1986, terns in the largest and highest-ranked Anderson 1981, Spennenman 1987, Allen shellfish taxa in the prehistoric diet would 1992, 2002, 2003, Kirch et al. 1995). It is ultimately facilitate our understanding of the probable that the decline in marine resources overall temporal changes in the assemblage, documented by those studies is also closely our analysis is based on relative relationships connected with other temporal changes in the of prey items already in the diet breadth. It overall Pacific island subsistence economy. is hard to see how differential field processing could result in any of the trends documented conclusion in our assemblage. For example, we demonstrate a significant decline in the abundance Evidence from our excavation and analysis at of the large gastropod Turbo sandwicensis cor- Nu‘alolo Kai suggests that mollusk species, related with an increase in the presence of the like other nearshore marine resources in the small Strombus maculatus. Differential field Pacific, were indeed susceptible to human processing could indeed account for the in- overuse. Previous subsistence studies in Ha- crease in Strombus maculatus but fails to pro- wai‘i and Polynesia have emphasized the vide insight into the decreasing abundance importance of mollusks and other inshore of Turbo sandwicensis. marine species in prehistoric diets as well as . Archaeological Case Study from Kaua‘i Morrison and Hunt 341 the role of environmental variability in struc- across marine environments in the Hawaiian turing resource use patterns (e.g., Kirch 1982, Islands; instead, detailed archaeological his- 1985, Dye and Steadman 1990). Yet little re- tories of ecological variation will be essential. search has focused specifically on how mollusk use varied through both space and time. Literature Cited By separating mollusk taxa by habitats, we are able to distinguish which species and habitats Allen, M. S. 1992. Dynamic landscapes and were more susceptible to human exploitation, human subsistence: Archaeological investi- as well as provide important supplementary gations on Aitutaki Island, Cook Islands. records regarding resource use predating Ph.D. diss., University of Washington, historical fisheries catch accounts. Because Seattle. modern conservation of mollusk species in ———. 2002. Resolving long-term changes Hawai‘i continues to be an important concern in Polynesian marine fisheries. Asian Per- for biologists (Gulko 2004, 2005, Meadows spect. 41 (2): 195–212. et al. 2005), our research has implications for ———. 2003. Human impact on Pacific applied habitat management. nearshore marine ecosystems. Pages 317– Further studies on human impacts to near- 325 in C. Sand, ed. Pacific archaeology: shore environments in Hawai‘i and elsewhere Assessment and prospects (2003). Proceed- in the Pacific should also take into consider- ings of the International Conference for ation long-term patterns in inshore and off- the 50th Anniversary of the First Lapita shore fish use. Data from a number of studies Excavation, Kone´, Noumeá, 2002. Les elsewhere in Oceania suggest that fishing in Cahiers de l’Archaeólogie en Nouvelle general decreased through time (Allen et al. Cale´donie. Vol. 15. Service des Museés et 2001, Allen 2002, 2003, Fitzpatrick and Ka- du Patrimonie de Nouvelle-Cale´donie, taoka 2005). Currently, the timing and cause Noumeá. of this decline is not well understood; however Allen, M. S., T. N. Ladefoged, and J. J. Wall. it is plausible that overexploitation of wild 2001. Traditional Rotuman fishing in tem- varieties of shellfish and fish species might poral context. Int. J. Osteoarchaeol. 11:56– have led to increased dependency on more 71. reliable and predictable terrestrial food re- Anderson, A. J. 1981. A model of prehistoric sources (Kirch 1973, Fitzpatrick and Kataoka collecting on the rocky shore. J. Archaeol. 2005). Although substantive data to test this Sci. 8:109–120. hypothesis are indeed lacking for Hawai‘i, Bird, D. W., and R. L. Bliege Bird. 1997. the ongoing development of prey and patch Contemporary shellfish gathering strat- choice models integrating the entire spec- egies among Meriam of the Torres Strait trum of subsistence strategies practiced across Islands, Australia: Testing predictions of a the archipelago will enhance the opportunity central place foraging model. J. Archaeol. to examine subsistence change with a well- Sci. 24:39–63. developed theoretical framework. ———. 2000. The ethnoarchaeology of juve- Finally, the results of our study show that nile foragers: Shellfish strategies among human exploitation of shellfish follows a Meriam children. J. Anthropol. Archaeol. complex trajectory that affected the marine 19:461–476. ecosystem at Nu‘alolo Kai. The changes we Botkin, S. 1980. Effects of human predation detect are relatively subtle, suggesting that on shellfish populations at Malibu Creek, human predation did not have dramatic con- California. Pages 121–139 in T. K. Earle sequences such as wholesale resource deple- and A. L. Christenson, eds. Modeling tion, extinctions, or habitat destruction. The change in prehistoric subsistence econo- complexity of change also argues for variabil- mies. Academic Press, New York. ity in cases of human predation and the im- Bronk Ramsey, C. 2005. OxCal program. pacts. Thus, it would be naive to generalize v3.10. . 342 PACIFIC SCIENCE July 2007

Broughton, J. M. 1994. Declines in mamma- rine shellfish-species use along the Atlantic lian foraging efficiency during the late coast in the southeastern United States. Holocene, San Francisco Bay, California. Southeast. Archaeol. 5:120–137. J. Anthropol. Archaeol. 13:371–401. ———. 1998. Shells. Cambridge Manual in ———. 2001. Resource intensification and Archaeology, Cambridge, United King- late Holocene impacts on Pacific bird dom. populations: Evidence from the Emeryville Cobb, J. N. 1905. The commercial fisheries. shellmound avifauna. Pages 251–279 in T. Pages 317–765 in D. S. Jordan and B. W. L. Hunt, C. P. Lipo, and S. L. Sterling, Evermann, eds. The aquatic resource of eds. Posing questions for a scientific ar- the Hawaiian Islands. Part II. Bull. U.S. chaeology. Bergin and Garvey, Westport, Fish. Comm. 23. Connecticut. Diamond, J. 2005. Collapse: How societies Butler, V. L. 2000. Resource depression on choose to fail or succeed. Viking Books, the northwest coast of North America. New York. Antiquity 74:649–661. Dye, T., and D. W. Steadman. 1990. Polyne- ———. 2001. Changing fish use on Mangaia, sian ancestors and their animal world. Am. southern Cook Islands: Resource depres- Sci. 78:207–215. sion and the prey choice model. Int. J. Os- Erlandson, J. 1988. The role of shellfish in teoarchaeol. 11:88–100. prehistoric economies: A protein perspec- Butler, V. L., and S. K. Campbell. 2004. Re- tive. Am. Antiq. 53:102–109. source intensification and resource depres- ———. 1991. Shellfish and seeds as optimal sion in the Pacific Northwest of North resources: Early Holocene subsistence on America. J. World Prehist. 18 (4): 327– the Holocene coast. Pages 89–101 in J. 405. Erlandson and R. H. Colten, eds. Hunter- Butler, V. L., and M. G. Delacorte. 2004. gatherers of Early Holocene coastal Cali- Doing zooarchaeology as if it mattered: fornia. Perspectives in California Archae- Use of faunal data to address current issues ology. Vol. 1. Institute of Archaeology, in fish conservation biology in Owens University of California, Los Angeles. Valley, California. Pages 25–44 in R. L. ———. 2001. Aquatic adaptations: Paradigms Lyman and K. P. Cannon, eds. Zooarch- for a new millennium. J. Archaeol. Res. aeology and conservation biology. Univer- 9:287–350. sity of Utah Press, Salt Lake City. Erlandson, J., C. Rick Torben, R. Vellano- Byers, D. A., and J. M. Broughton. 2004. weth, and T. Largaespada. 2004. Human Holocene environmental change, artio- impacts on ancient environments: A case dactyl abundances, and human hunting study from California’s northern Channel strategies in the Great Basin. Am. Antiq. Islands. Pages 51–83 in S. M. Fitzpatrick, 69:235–255. ed. Voyages of discovery: The archaeology Cannon, M. D. 2003. A model of central of islands. Praeger, New York. place forager prey choice and an applica- Esh, K. S. 2005. Human impacts on Pacific tion to faunal remains from the Mimbres seabirds: An analysis of resource use at Valley, New Mexico. J. Anthropol. Ar- Nu‘alolo Kai, Kaua‘i. M.A. thesis, Univer- chaeol. 22:1–25. sity of Hawai‘i at Ma¯noa, Honolulu. Catterall, C. P., and I. R. Poiner. 1987. The Fitzpatrick, S. M., and O. Kataoka. 2005. potential impact of human gathering on Prehistoric fishing in Palau, Micronesia: shellfish populations with reference to Evidence from the Northern Rock Islands. some NE Australian intertidal flats. Oikos Archaeol. Oceania 40:1–13. 50:114–122. Glassow, M. A., and L. R. Wilcoxon. 1988. Charnov, E. L. 1976. Optimal foraging: Coastal adaptations near Point Conception The marginal value theorem. Popul. Biol. California, with particular regard to shell- 9:129–136. fish exploitation. Am. Antiq. 53:36–51. Claassen, C. 1986. Temporal patterns in ma- Graves, M. W., J. M. Field, and W. K. . Archaeological Case Study from Kaua‘i Morrison and Hunt 343

McElroy. 2005. An overview of site 50-30- dolfi, C. H. Peterson, R. S. Steneck, M. J. 01-196, Nu‘alolo Kai, Kaua‘i: Features, ex- Tegner, and R. R. Warner. 2001. Histori- cavation, stratigraphy, and chronology of cal overfishing and the recent collapse of historic and prehistoric occupation. Pages coastal ecosystems. Science (Washington, 149–187 in M. T. Carson and M. G. D.C.) 293:629–637. Graves, eds. Na Mea Kahiko o Kaua‘i: Ar- Jerardino, A. 1997. Changes in shellfish spe- chaeological studies in Kaua‘i. Special cies composition and mean shell size from Publication No. 2, Society for Hawaiian a late Holocene record of the west coast of Archaeology, Honolulu, Hawai‘i. Southern Africa. J. Archaeol. Sci. 24:1031– Grayson, D. 1984. Quantitative zooarchaeol- 1044. ogy: Topics in the analysis of archaeologi- Jones, L. E. 2004. Broad spectrum diets and cal faunas. Academic Press, New York. the European rabbit (Oryctolagus cuniculus): ———. 2001. The archaeological record of Dietary change during the Pleistocene- human impacts on animal populations. J. Holocene transition in the Dordogne, World Prehist. 15 (1): 1–68. southwestern France. Ph.D. diss., Univer- Grayson, D., and F. Delpech. 1998. Chang- sity of Washington, Seattle. ing diet breadth in the Early Upper Paleo- ———. 2006. Prey choice, mass collecting, lithic of southwestern France. J. Archaeol. and the wild European rabbit (Oryctolagus Sci. 25:1119–1129. cuniculus). J. Anthropol. Archaeol. 25:275– Gulko, D. 2004. Hawaiian marine species for 289. ESA Candidate listing revised Candidate Kay, A. 1979. Hawaiian marine shells. Reef list. State of Hawai‘i Division of Aquatic and Shore Fauna of Hawaii, Section 4: Resources, Honolulu. Mollusca. Bernice P. Bishop Musem Press, ———. 2005. Hawaii endemic species status Honolulu, Hawai‘i. chart spreadsheet. State of Hawai‘i Divi- Kay, A., and W. Magruder. 1977. The biol- sion of Aquatic Resources, Honolulu. ogy of opihi. Department of Planning Hawai‘i Department of Land and Natural and Development, State of Hawai‘i, Ho- Resources. 1981. Protected marine fish- nolulu. eries resources. Chapter 92, ‘‘Opihi.’’ In Kirch, P. V. 1973. Prehistoric subsistence State of Hawai‘i Administrative Rules, patterns in the northern Marquesan Is- Honolulu. lands, French Polynesia. Archaeol. Phys. Hildebrandt, W. R., and T. L. Jones. 1992. Anthropol. Oceania 8:24–40. The evolution of marine mammal hunting: ———. 1982. The ecology of marine exploi- A view from the California and Oregon tation in prehistoric Hawaii. Hum. Ecol. coasts. J. Anthropol. Archaeol. 11:360–401. 10:455–476. Hunt, T. L. 2005. Archaeological stratigra- ———. 1985. Feathered gods and fishhooks: phy and chronology at Nu‘alolo Kai, Na An introduction to Hawaiian archaeology Pali District, Kaua‘i. Pages 236–258 in M. and prehistory. University of Hawai‘i T. Carson and M. G. Graves, eds. Na Mea Press, Honolulu. Kahiko o Kaua‘i: Archaeological studies ———. 1997. Microcosmic histories: Island in Kaua‘i. Special Publication No. 2 Soci- perspectives on global change. Am. ety for Hawaiian Archaeology, Honolulu, Anthropol. 99 (1): 30–42. Hawa‘i. Kirch, P. V., D. W. Steadman, V. L. Butler, Jackson, J. B. C. 2001. What was natural in J. Hather, and M. I. Weisler. 1995. Prehis- the coastal oceans? Proc. Natl. Acad. Sci. tory and human ecology in eastern Polyne- U.S.A. 98:5411–5418. sia: Excavations at Tagatatu rockshelter, Jackson, J. B. C., M. X. Kirby, W. H. Berger, Mangaia, Cook Islands. Archaeol. Oceania K. A. Bjorndal, L. W. Botsford, B. J. Bour- 30:47–65. que, R. H. Bradbury, R. Cooke, J. Erland- Kirch, P. V., and D. Yen. 1982. Tikopia: Pre- son, J. A. Estes, T. P. Hughes, S. Kidwell, history and ecology of a Polynesian out- C. B. Lange, H. S. Lenihan, J. M. Pan- lier. Bernice P. Bishop Mus. Bull. 238. . 344 PACIFIC SCIENCE July 2007

Lyman, R. L. 2003. Pinneped behavior forag- Raab, M. 1992. An optimal foraging analysis ing theory and the depression of metapop- of prehistoric shellfish collecting on San ulations and non-depression of a local Clemente Island, California. J. Ethnobiol. population on the southern northwest 12:63–80. coast of North America. J. Anthropol. Ar- Reimer, P. J., M. G. L. Baillie, E. Bard, A. chaeol. 22:376–388. Bayliss, J. W. Beck, C. Bertrand, P. G. Lyman, R. L., and K. P. Cannon. 2004. Blackwell, C. E. Buck, G. Burr, K. B. Cut- Zooarchaeology and conservation biology. ler, P. E. Damon, R. L. Edwards, R. G. University of Utah Press, Salt Lake City. Fairbanks, M. Friedrich, T. P. Guilderson, Madsen, D. B., and D. N. Schmitt. 1998. K. A. Hughen, B. Kromer, F. G. McCor- Mass collecting and the diet breadth mac, S. Manning, C. Bronk Ramsey, R. model: A Great Basin example. J. Ar- W. Reimer, S. Remmele, J. R. Southon, chaeol. Sci. 25:445–455. M. Stuiver, S. Talamo, F. W. Taylor, J. Mannino, M. A., and K. D. Thomas. 2001. van der Plicht, and C. E. Weyhenmeyer. Intensive Mesolithic exploitation of coastal 2004. IntCal04 Terrestrial radiocarbon resources? Evidence from a shell deposit age calibration, 26–0 ka BP. Radiocarbon on the Isle of Portland (southern England) 46:1029–1058. for the impact of human foraging on pop- Reitz, E. J. 2004. ‘‘Fishing down the web’’: A ulations of intertidal rocky shore mollusks. case study from St. Augustine, Florida, J. Archaeol. Sci. 28:1101–1114. USA. Am. Antiq. 69:63–83. ———. 2002. Depletion of a resource? The Severns, M. 2001. Hawaiian seashells. Island impact of prehistoric human foraging on Heritage, Hawai‘i. intertidal mollusk communities and its sig- Smith, E. A. 1991. Injuamiuat foraging strat- nificance for human settlement mobility egies: Evolutionary ecology of an Artic and dispersal. World Archaeol. 33:452– hunting economy. Aldine De Gruyter, 474. New York. McAllister, A. 2003. Prehistoric fishing at Fa- Smith, E. A., and M. Wishnie. 2000. Conser- kaofa Atoll, Tokelau. M.A. thesis, Univer- vation and subsistence in small scale soci- sity of Auckland, Auckland, New Zealand. eties. Annu. Rev. Anthropol. 29:493–524. Meadows, D., A. L. Kane, C. Mitchell, and C. Spennenman, D. H. R. 1987. Avaliability of Ogura. 2005. Technical Report X. Hawai‘i shellfish resources on prehistoric Tonga- Statewide Aquatic Wildlife Conservation tapu, Tonga: Effects of human predation Strategy. Pacific Cooperative Studies Unit, and changing environment. Archaeol. Oce- University of Hawai‘i at Ma¯noa, Hono- ania 22:81–96. lulu. Steneck, R. S., J. Vavrinec, and A. V. Leland. Nagaoka, L. 2001. Using diversity indices to 2004. Accelerating trophic-level dysfunc- measure changes in prey choice at the tion in kelp forest ecosystems of the west- Shag River Mouth site, southern New Zea- ern North Atlantic. Ecosystems 7:323– land. Int. J. Osteoarchaeol. 11:101–111. 332. ———. 2002. The effects of resource depres- Stephens, D. W., and J. R. Krebs. 1986. For- sion on foraging efficiency, diet breadth aging theory. Princeton University Press, and patch choice use in southern New Princeton, New Jersey. Zealand. J. Anthropol. Archaeol. 21:419– Swadling, P. 1976. Changes induced by hu- 442. man predation in prehistoric shellfish pop- Orians, G. H., and N. E. Pearson. 1979. On ulations. Mankind Q. 10:156–162. the theory of central place foraging. Pages ———. 1986. Lapita shellfishing: Evidence 154–177 in D. J. Horn and R. D. Mitchell, from sites in the Reef/Santa Cruz Group, eds. Analysis of ecological systems. Ohio Southeast Solomons. Pages 137–148 in State University Press, Columbus. Traditional fishing in the Pacific. Anthro- Pianka, E. R. 1983. Evolutionary ecology. pological Records 37. Bernice P. Bishop Harper and Row, New York. Museum, Honolulu, Hawai‘i. . Archaeological Case Study from Kaua‘i Morrison and Hunt 345

Thomas, F. R. 2001. Mollusk habits and ﬁsh- Wing, S. R., and E. S. Wing. 2001. Prehis- eries in Kiribati: An assessment from the toric ﬁsheries in the Caribbean. Coral Gilbert Islands. Pac. Sci. 55:77–97. Reefs 20:1–8. ———. 2002. An evaluation of central place Winterhalder, B. 1981. Foraging strategies in foraging among mollusk gathers in west- the boreal forest: An analysis of Cree ern Kiribati, Micronesia: Linking behavior hunting and gathering. Pages 66–98 in ecology with ethnoarchaeology. World Ar- B. Winterhalder and E. A. Smith, eds. chaeol. 34:182–208. Hunter-gatherer foraging strategies. Uni- Thomas, K. D., and M. A. Mannino. 1999. versity of Chicago Press, Chicago. The bioarchaeology of the Culverwell Wolverton, S. 2005. The effects of the hypsi- shell midden. Pages 94–114 in S. Palmer. thermal on prehistoric foraging in Mis- Culverwell Mesolithic site habitation site, souri. Am. Antiq. 70:91–106. Isle of Potland: Excavation report and re- Woodward. C. S. 2000. Ocean’s end: Travels search studies. Archaeological Reports, through endangered seas. Basic Books, Oxford, United Kingdom. New York. ———. 2001. The exploitation of inverte- Yesner, D. R. 1984. Population pressure in brates and invertebrate products. Pages coastal environments: An archaeological 427–440 in D. R. Brothwell and A. M. test. World Archaeol. 16:108–127. Pollard, eds. Handbook of archaeological ———. 1987. Life in the ‘Garden of Eden’: sciences. Willey, Chichester, United King- Causes and consequences of the adoption dom. of marine diets by human societies. Pages Ugan, A. 2005. Climate, bone density, and 285–310 in M. Harris and E. B. Ross, eds. resource depression: What’s driving varia- Food and human evolution: Towards a tion in large and small game in Fremont theory of food and human habits. Temple archaeofaunas? J. Anthropol. Archaeol. University Press, Philadelphia, Pennsylva- 24:227–251. nia.