The fauna from the 2010 Seven Site excavations was identified using comparative collections held in the Zooarchaeology Laboratory at the University of Alabama at Birmingham. The comparative collection includes hundreds of skeletons of marine fishes from the tropical Pacific Islands, including materials from the geographic areas of Micronesia, Polynesia, and Fiji. Multiple individuals of various size classes of the most common species have facilitated identification and the development of secondary data such as biomass and size classes. The collection contains numerous individuals in the following families: Belonidae, Scaridae, Labridae, Serranidae, Carangidae, Acanthuridae, Scombridae, Lethrinidae, Lutjanidae, Mullidae, Siganidae, and Holocentridae.

Zooarchaeological methods follow techniques developed by Reitz and Wing (1999) and are explained elsewhere in detail (see O’Day 2001). All efforts were made to identify faunal specimens to the lowest taxonomic level possible, that is, with as much specificity as possible. Fish taxonomy followed Myers (1991) and Froese and Pauly (2004). In addition to the texts, I consulted Fishbase (http://www.fishbase.org/) for a variety of biological information on the identified species.

All faunal material was counted and weighed and modifications such as cut marks or burning were recorded. The number of identified specimens (NISP) is the basic specimen count used. Minimum number of individuals (MNI) were determined by paired elements and estimated size-class for fishes. Following Reitz and Wing (1999:194), MNI was defined as the smallest number of individuals that is necessary to account for all skeletal specimens of a given species within the assemblage. Initially, when I estimated the MNI, each provenience (cluster) was considered separately and the summary tables list the totals listed by fish taxa. This is a conservative method of estimation that results in fewer MNI for the entire Seven Site fish assemblage. For comparative purposes I also estimated the MNI from the combined fish taxa identified from all the clusters (Table 1). The total estimated MNI of 85 for fish combines the MNIs from each cluster. I feel that this is a safe and relatively accurate representation of the actual fishes that were consumed given the bones present in the assemblage.

Estimating Fish Size

The Seven Site assemblage contained a total of 650 fish vertebrae. A sample of the anterior widths of the vertebral centra were measured as a proxy for fish size, following the assumption that the fish vertebrae (identified and unidentified) provide a representative cross-section of the species in the assemblage (O’Day 2001; Wing 1998). The vertebral centra width data, based on measurements of whole vertebrae with the anterior centra intact (N = 577), are used herein to estimate the average body size of fish in the assemblage.

Estimates of Dietary Contribution

Calculated sample biomass was used to estimate the relative dietary contribution from the skeletal weight of archaeological specimens. These calculations provide information on the quantity of meat potentially supplied by an animal based on allometric principles that an animal’s body mass, skeletal mass, and skeletal dimensions change in proportion with body size increases (Reitz et al. 1987). Biomass estimates provide information that cannot be ascertained from specimen weights alone, such as those presented in Tables 2 – 8.

The following allometric equation was applied to data from the Seven Site assemblages in order to describe the relationship between body weight and skeletal weight, estimating the amount of meat or soft tissue related to the archaeological materials:

Y = aX^b or log₁₀ Y= a + b (log₁₀X)

where:

Y = estimated sample biomass (kg) contributed by the archaeological specimen for a taxon

X = specimen weight (kg) of the archaeological specimen for a taxon

a = the Y-intercept of the linear regression line

b = slope of the regression line

Sample biomass, or soft tissue weight, was predicted for fishes using specimen weight for X, in the above allometric formula. The allometric constants, a and b, were obtained from Reitz and Wing (1999:72) who originally derived the constants from a large set of reference specimens (Florida Museum of Natural History, University of Florida) with known whole body weights and dimensions of the skeletal elements.

Results

TIGHAR’s 2010 Seven Site faunal assemblage contained a total of 2824 fish bones, including a small number of shark remains and thousands of bones from boney fishes. In addition, a total of 7 rat bones (Rattus exulans) were recovered from clusters 2, 4, and 6. A total of 85 fish MNI were estimated for Clusters 2-7 and X and the fish bone material weighed a total of almost 280 grams. Unidentified fish bones made up the majority of the fish bone material in terms of count (NISP), weight (grams), and biomass. However, this is not unusual in Pacific island fish assemblages. In sum, the majority of the fish bones across clusters are represented by a suite of taxa including (in decreasing order of abundance) Scarids, Acanthurids, Serranids, Lutjanids, Lethrinids, Myrpristis sp., Labrids, and Carangids. These fishes are common in Pacific Island assemblages.

Approximately 20% of the fish bone material was burned and highly fragmentary; note that this stands in stark contrast to the material identified from the 2007 excavations of the Seven Site (Features SL-2, SL-3, and WR-1), where more than 90% of the fish bones displayed evidence of burning. A list of common names for the identified species is provided in Appendix A.

The Clusters

I analyzed each cluster separately and then amalgamated the data to estimate the total assemblage’s overall secondary data, including biomass (Table 1). Summaries of the identified fauna from each cluster are presented in the tables below (Tables 2-8).^[1]

Table 2. Cluster 2 faunal summary.

Taxa	NISP	wt. (g)	MNI
Acanthuridae	6	6.5	2
Carangidae	2	0.3
Caranx sp.	1	0.8	1
Epinephelus sp.	1	0.7	1
Labridae	1	0.5	1
Lutjanidae	8	3.6	1
Lutjanus sp.	1	0.1	1
Muraenidae	1	0.1	1
Myrpristis sp.	11	2.2	3
Naso sp.	1	1.1	1
Perciformes	2	0.4
Scaridae	43	2.6	1
Scombridae	1	2.2	1
Serranidae	6	2.9	1
Sphyraena sp.	1	0.2	1
UID fish	302	34.1
TOTAL	388	58.3	16
Rattus exulans	3	0.2	1

The material from Cluster 2 (Lanes A and B) was largely comprised by unidentified fish bone fragments (Table 2). The classification, “unidentified fish,” in all of the units refers to bone that was primarily composed of fragments of elements, or of elements that are not identifiable to family, genus, or species (for example, spines and rays, which are copious in fish skeletons). The unidentified fish was often broken and fragmentary, therefore making identification of some of the material impossible.

The fishes represented by the bone in Cluster 2 include, in decreasing order of abundance: Scarids (Parrotfishes), Holocentrids (Soldierfishes), Acanthurids (Tangs), Lutjanids (Snappers), and Serranids (Groupers). Less than 2 grams of the fish bones were burned. The majority of the Parrotfish remains consisted of scales. Three bones of the Pacific Rat, Rattus exulans, were also identified from Cluster 2 (Lane A).

Table 3. Cluster 3 Faunal Summary.

The Cluster 3 fauna are entirely made up of fish bone and includes 214 specimens. Carangids (Carangidae) were the most frequently identified taxa in terms of count and weight. The general classification of “Perciformes” was used to identify bones from this cluster, and other, when I was unable to determine what family or genus to classify the bone, but I was certain that it belonged to one of the closely related (and similar in terms of some bone morphology) families including, Carangidae, Lutjanidae, Serranidae, or Lethrinidae. A large portion of the Cluster 3 bones were burned and heavily charred; the burned bone included 187 bone fragments (87% of the total bone), and 20.6 grams (80% of the total weight).

The Cluster 4 fish fauna included a total of 890 bone specimens, weighing 47.8 grams and making up an estimated 15 MNI. Additionally, 3 rat bones (MNI=1) were identified from Cluster 3 (this cluster also contained bird bone that looks like chicken—I am sending this to Dr. King for further analysis along with the other non-fish/bird bones). The fish bones included 44 specimens that were burned and which contribute about 16% of the total bone mass for this cluster. This assemblage yielded a small fragment of a drilled tooth-isolate from a reef shark (Carcharhinidae). The vast majority of the identified fish material is made up of Parrotfish (Scaridae and Scarus sp.) bone fragments and scales, representing an estimated 4 MNI. The next most commonly identified taxa include 2-estimated MNI each of Myrpristis sp. (Soldierfish) and Lutjanids. Other notable but less frequently identified fishes include Scombrids (Tunas) and Exocoetids (flying fish).

Table 4. Cluster 4 Faunal Summary.

Table 5. Cluster 5 Faunal Summary.

The Cluster 5 fauna consist entirely of fish bones, including 509 specimens and an estimated MNI of 13. After the unidentified bones, material from Perciformes, Serranidae, Myrpristis sp., and Mullidae (Goatfish) make up the majority of the fish bones. Bones from Serranids, especially cranial elements, were common in Cluster 5 and these produced an estimated MNI of 5, a notably large estimate relative to the both this assemblage and others from the 2010 excavations. More than 90% of this assemblage is burned and many of the bones are highly fragmentary. Not surprisingly, this material looks like the other material from WR-units. It stands in stark contrast to the other clusters, in this analysis, in terms of the character, color, and fragmentary nature. This material appears to have been cooked on an open fire.

The Cluster 6 fauna include 475 specimens of fish material, weighing almost 80 grams and including an estimated MNI of 17; this assemblage contains more fish material than the other clusters. One bone of a rat is included in this assemblage. The vast majority of the identified fish bone, by all measures, represents Acanthurids (Acanthuridae and Acanthurus sp.). Muraenids (Moray eels), Serranids, Lethrinids (Emperorfish), and Scarids also occur in relatively high frequency. In terms of elements present, fish vertebrae make up a large part of the overall assemblage (68% by weight and 32% by count). Approximately 15% of the Cluster 6 fauna is burned (by both NISP and weight measurements).

Table 6. Cluster 6 Faunal Summary.

The Cluster 7 assemblage is made up of 268 reef fish bone elements and fragments. Bones from relatively small-bodied Serranids occur in high frequency (17 NISP, 5.2g, MNI 4) and are represented by a variety of cranial elements and jaw elements in particular. Most of the bones in this assemblage are burned and heavily charred (94% by NISP and 74% by weight). Bones from Perciformes are also common in this assemblage.

Cluster X fauna is summarized in Table 8 and consists entirely of fish; 80 specimens weighing 19.4 grams and an estimated 9 MNI are included in this assemblage. Acanthurids and Serranids are the most frequently identified taxa. Approximately 42% (by NISP, and 38% by weight) of the X Cluster fauna is burned. Some of this bone is charred black, white, and blue, suggesting that they were cooked in some kind of fire feature. However, unlike much of the material analyzed from the 2007 excavations, the 2010 assemblages did not have an abundance of ash and charcoal on the exterior.

A total of 577 fish vertebrae were measured in order to estimate the average size of the fishes represented in the assemblage. The vertebrae ranged from 1 to 16.7 mm in anterior width, representing fishes from about 8 cm in total length (TL) to about 62 cm TL. The average body size of the reef fishes represented in this assemblage is 28.5 cm TL (as estimated from the average size of the vertebral centra in the assemblage). The standard deviation of the vertebral centra widths is 8.7 mm.

Table 9. Summary of fish vertebral measurements

A total of 659 fish body elements and 158 cranial elements, not including scales (N=415), were identified from the Seven Site assemblage (Table 5). The majority of body elements are vertebrae (N=650). If the fish scales and the unidentified fragments are eliminated from the analysis, there are a total of 817 identified elements; of that approximately 19% are cranial and 80% are body elements. The average reef fish has about 56 cranial bones, between about 24 to 36 vertebrae, and numerous additional postcranial elements, including hundreds of bony spines and rays; note that this is an estimate, and depending on the fish species the number of bones overall and per body region varies widely (this estimate also does not include scales). Regardless of the species of fish, the postcranial elements far out number the cranial elements by at least 3.5 times in the average fish skeleton.

In both the 2010 assemblages and in the 2007 assemblages, the identified fishes include a portion of cranial bones that approximates the natural distribution of cranial to postcranial bones in the average fish. I interpret this pattern as reflecting a relatively unbiased deposition of fish in the archaeological record. As I said in my last report, this depositional pattern is different from the pattern I have generally observed in Fiji where cranial bones are deposited differently than post-cranial bones, due to preference for the head and particular eating habits (Jones 2011). Moreover, fauna from archaeological sites I have worked on elsewhere in the Pacific Islands typically occur in midden contexts or earth ovens. Pacific Islanders generally do not cook fish on an open fire; when they do, the material deposited often reflects a biased deposition of bone. That is, the pattern of deposited bones results from preference, the way Pacific Islanders eat the whole fish and relish the head, and the habit of throwing bone refuse in areas that are distinct from the place where the fish is cooked.

Table 10. Identified fish bone elements from 2010 excavations of the Seven Site, Nikumaroro.

Habitat and Ecology

The suite of taxa identified from Clusters 2-7 and X are commonly identified Pacific Island taxa. These fishes will generally all take a hook, but most could also have been collected with a spear or net (e.g., a cast net, scoop net, or gill net). The assemblage includes a variety of common reef-associated species that would also inhabit lagoon habitats, especially when they are young and small-bodied. The majority of the fishes in the 2010 assemblage are small-bodied fishes that would inhabit the lagoon and near-shore reef. Acanthurids, Serranids, Scarids, and Carangids are the most abundant species identified by all measures, making up 11, 13, 7 and 7 MNI, respectively (Table 1).

Many taxa contributed a small number to the overall NISP, weight, and MNI in the 2010 assemblage. Among these taxa are Lutjanids, Lethrinids, Scombrids, Holocentrids, Labrids (Wrasse), Muraenids, Belonids (Needlefish), and Mullids. As I stated in my last report, marine Biological surveys of the Phoenix Island Group, including Nikumaroro, have found that the Gymnosarda unicolor (Dogtooth tuna) was especially abundant prior to 2001 (representing up to 75% of the large fish sampled) as was the reef shark population (Uwate and Teroroko 2007). Dogtooth tuna is known to occur around coral reefs; this species is generally solitary or found in schools of six or less (Froese and Pauly 2004). Given their common occurrence on coral reefs and in the inshore area, it is therefore not surprising that both tuna (Scombridae) and reef shark (Carcharhinidae) were identified from the assemblage. Either could easily have been taken with a hook and handline or with a spear.

Comparison of Fish Assemblages from 2001, 2007, and 2010 Field Seasons

As with the material analyzed previously, when comparing the material I identified from the 2007 and 2010 excavations to that from the 2001 excavations, I notice a number of trends. First, all of the fish faunal assemblages are similar in character, make up, and abundance (with the exception of Cluster 5). Second, both the 2001 and 2007 excavations contained an abundance of Carangids and Holocentrids. These two families were identified from every feature/unit excavated in 2001. The material from the 2010 excavations did not contain an abundance of either Carangids or Holocentrids. Third, the 2010 material is less burned and fragmentary than the 2007 bones (with the exception of Cluster 5, which was much like the 2007 material in character and nature). Fourth, The assemblages or clusters from all the excavations are of relatively small sample sizes. It is possible that if these small assemblages are spatially distinct they could represent single incidents of eating and cooking. Cluster 5 looks as if it could represent a cooking and eating incident.

Important Questions

A number of important questions were posed in the TIGHAR preliminary report on the excavations of the Triangle Site and the Seven Site. Below I address the questions to which the available data can contribute insights, using data from my analysis of the 2010 faunal materials and drawing from my experiences identifying zooarchaeological assemblages and fishing with Pacific Islanders in Micronesia, Fiji, and Polynesia.

1. How many individuals are represented?

The fish bones from the 2010 excavations represent an estimated MNI of 85, that is 85 individuals, according to my conservative estimate.

2. What species are represented?

Seventeen distinct taxa were identified (> 16 boney fishes and reef shark; Table 1). This estimate is based on the bone I was able to identify to Family, genus, or species, but it is likely a vast underestimate of the actual diversity contained in the assemblage.

3. Where and how easily could the taxa represented in the assemblage have been procured?

The diversity and the body sizes of the represented fishes suggest that a net could have been used for collecting the fish, and/or that the collector gathered broadly in the inshore area and/or the lagoon reef (either inside or outside). An assortment of relatively small-bodied fishes is a common outcome of fishing that utilizes nets. The fish species represented could have been captured with a spear or taken with a hook and hand line as well as net technologies. All these technologies require some skill, but it could be said that nets produce a more consistent return in optimal environments.

4. How were they prepared and cooked?

The evidence of burning on some of the bone (approximately 20% of the entire 2010 assemblage) is suggestive of cooking methods that involved the fishes being placed directly on the fire, or possibly the fishes were cooked in an oven-feature context. Both cooking on an open fire (roasting) and cooking in an earth oven will produce the type of burning exhibited by some of the 2010 bones (Cluster 5 in particular). The material from 2007 appeared to come from an open fire context, as the bone had much ash and charcoal adhering to it, in addition to being burned and highly fragmentary. In general, the material from 2010 does not display this type of evidence (again, Cluster 5 is an exception).

5. What parts were used and not used?

The bone elements represented by the identified fish bone are listed in Table 10. As with the 2007 material, the 2010 material occurs in proportion to that naturally found in the fish skeleton. Approximately 19% of the fish bone comes from the cranial region, and 80% of the bone comes from the post-cranial body region (excluding scales). Vertebrae comprise the majority of the non-cranial elements. It appears that the people who consumed the fishes represented in the 2010 assemblage deposited the bones of the consumed fishes in or around the place where they were prepared and/or consumed. In the case of the 2007 material, most of that bone appeared to have been deposited in or on the fire features where the fishes were cooked. From the limited information I have about the context of the bones excavated in 2010, it is less clear how deposition occurred. Nevertheless, based on the elements that are present in the assemblage, it is possible that most of the fishes was utilized for food, or that the unused portions were deposited in the same place as the portions that were used.

6. Generally, are the species represented, and the way in which they are prepared, more consistent with traditional Micronesian and Polynesian subsistence practices or with those of Europeans camping out?

The 2010 fish material does not provide strong evidence to make an argument about the ethnic identity of the people that consumed the fishes for food. Among the 2010 assemblages, Cluster 5 is the most unique. Based on my analysis of the 2007 bones (WR-1 and SL-units 2 and 3), I argued that the character and nature of the bone material suggested collection by non-native Pacific Islanders.

References Cited

Froese, R. and D. Pauly (editors) 2004
	Fishbase. World Wide Web electronic publication, http://www.fishbase.org/search.cfm, version 03/2004, accessed June-July, 2008.
Jones, S. 2011
	Contemporary Subsistence and Foodways in the Lau Islands Fiji: An Ethnoarchaeological Study of Non-optimal Foraging and Irrational Economics. In, Ethnozooarchaeology: The Present Past of Human-animal Relationships, edited by U. Albarella. Oxbow Books, Oxford, UK.
Krebs, Charles J. 1989
	Ecological Methodology. New York: Benjamin Cummings.
Myers, Robert F. 1991
	Micronesian Reef Fishes: A Practical Guide to the Identification of the Coral Reef Fishes of the Tropical Central and Western Pacific. Second Edition. Coral Graphics, Guam.
O’Day, S. and Jones 2001
	Excavations at the Kipapa Rockshelter, Kahikinui, Maui, Hawai’i. Asian Perspectives 40 (2):279-304.
Reitz, Elizabeth J. and Elizabeth S. Wing 1999
	Zooarchaeology. Cambridge: Cambridge University Press.
Reitz, Elizabeth J., Irvy R. Quitmyer, H. S. Hale, Sylvia J. Scudder, and Elizabeth S. Wing. 1987.
	Application of allometry to zooarchaeology. American Antiquity 52(2):304-317.
Uwate, K. Roger and Tukabu Teroroko 2007
	The Phoenix Islands Protected Area Management Plan, First Draft 2-5-07. Phoenix Islands Protected Area Ministry of Environment, Lands, and Agricultural Development, Government of Kiribati.
Wing, E. S. 1998
	The Sustainability of Resources used by Native Americans on five Caribbean Islands. Paper presented at the ICAZ meeting in Victoria, B.C., Canada, August 1998.

Appendix A.

Common names of the identified fishes from the 2010
excavations of the Seven Site, Nikumaroro; the fishes are listed in taxonomic order.

[1] The fauna are in alphabetical order in tables 2 – 8.