Bull. Fla. Mus. Nat. Hist. (2003) 44(1): 27-34 27 SCREEN SIZE AND THE NEED FOR REINTERPRETATION: A CASE STUDY FROM THE NORTHWEST COAST Kathlyn M. Stewarti and Rebecca J. Wigen2 There has been much discussion in the archaeologicalliterature on the utilization of different screen mesh sizes for recovering faunal elements, with many researchers decrying the use of the 6.4 mm (1/4") screen, which allows small elements to fall through the mesh. There has been less discussion on the merits of the data obtained through recovery of smaller elements, specifically, is the "new data" worth the extra time and labor? This paper examines faunal material from three sites from the Northwest coast of Canada. The material has been recovered using either 6.4 mm or 2.8 mm (1/8") mesh screens. The results suggest that elements of herring and other small fish have been greatly underestimated in Northwest coast sites, and that these formed an important part of the coast economy and subsistence. Where salmon, a large fish with often well preserved elements, has been seen as the mainstay of Northwest diet and economy, excavation with small-mesh screens may indicate a much greater importance for herring and other small fish on the coast. For accurate reconstruction of past lifeways, at least on the Northwest coast, screens with mesh 2.8 mm must be used for at least a substantial part of the matrix, in combination with the 6.8 mm mesh. Key words: screen gauge, fish bones, Northwest coast, sampling, North American prehistory The question of mesh size for screening archaeological In this paper, we analyze fauna from three matrix has been the focus of much debate (eg., Butler Northwest coast shell midden sites where either 6.4 mm 1993; Cannon 1999; Clason and Prummel 1977; Gordon or 2.8 mm (1/8") mesh screens were utilized. Depending 1993; Grayson 1984; James 1997; Lyman 1982; Payne upon which screen size was used, completely divergent 1972; Rick and Erlandson 2000; Schaffer 1992; Schaffer faunal results were obtained. These divergent results and Sanchez 1994). Traditionally the 6.4 mm (1/4") and mean that completely different representations of even larger meshes have been used in archaeological Northwest coast economy and subsistence can be excavations, but their utilization has come under attack inferred, depending upon which set of data is consulted. because of loss of small bone elements through the mesh Our study illustrates that, at least in Northwest coast and consequent loss of faunal data. But how important shell middens, accurate subsistence and economic are these lost data? While much debate focuses on the reconstruction is absolutely dependent on recovery of faunal elements lost and optimal strategies for recovery, the faunal elements through screens with 52.8 mm mesh. there is less discussion about the value of these elements We suggest that 50% of the matrix be screened through to our overall knowledge of the cultures studied or the 52.8 mm mesh. reconstruction ofthe associated environments and animal communities. Several studies have undertaken recovery METHODOLOGY of fauna with <6.4 mm mesh, and have added the"new" In 1994,1995, and 1997, the University of Victoria field taxa to their lists of resources exploited, and calculated school under the direction of Don Mitchell, Quentin resulting diversity indexes and/or edible meat weights Mackie, and Becky Wigen began excavating the (e.g., Gordon 1993; Rick and Erlandson 2000). Is this Kosapsom site (DcRu 4) located on southeastern additional knowledge worth the considerable time, labor, Vancouver Island along the Gorge waterway in the and cost needed for screening, sorting, and identifying present-day Victoria area (Fig. 1). The excavation was small faunal elements from small-mesh screens? in part a salvage recovery prior to future construction of a sidewalk. Thirty-eight 1xlm excavation units were ' Canadian Museum of Nature, PO Box 3443, Stn D, Ottawa, randomly and.subjectively selected and opened in a stripCanada. that more or less followed the shore along the Gorge.2 Dept of Anthropology, University of-Victoria, Victoria, BC, Canada. Although the deposits were somewhat mixed, there was For communication on this paper, please contact the senior evidence of three distinct cultural components, each of author. which concentrated in different areas of the site: Locarno 28 ZOOARCHAEOLOGY: Papers to Honor Elizabeth S. Wing flip' BRITISH ..rf '1 fr Irri U COLUMBIA '4 ' L« w Canada -7994 USA El, 11 #mkm -k A---/ =54 Kosapsom site . dIP I '1 Figure 1. The Kasapsom site, southern Vancouver Island. Beach Phase, dating from about 3500 to 2500 B.R, Gulf collections of the Canadian Museum of Nature in of Georgia Phase, dating from about 1500 B.R to contact, Ottawa, and the University of Victoria osteological and Historic, dating from 1855 A.D. to the present. collections. In this paper we present the results of the In the 1994 field season, all material recovered from Kosapsom faunal analysis, and also compare the results excavation by 5 cm levels was screened using a 2.8 mm with the fauna from two nearby and reasonably screen. Recovery of the small bones from the screens contemporaneous sites, Esquimalt Lagoon (DcRu 2) in the field was so time-consuming that the decision was and Fort Rodd Hill (DcRu 78). These sites are about 3 made to bag the material from the 2.8 mm screens and and 4.5 km, respectively, west of the Kosapsom site on finish the sorting in the lab. In addition, a 20 x 20 x 5 cm the Gorge waterway. sample was taken from a corner of each excavation At Esquimalt Lagoon, fourteen lx2 meter units unit. Retrieved material was taken to a laboratory at the were excavated and fauna collected by 10 cm levels. University of Victoria for sorting. Unfortunately, the time Matrix excavated was screened through a 6.4 mm mesh required for sorting the material into the categories of screen. All vertebrate material was identified to element, shell and vertebrates was enormous, and after four but bird elements were not identified to taxon (Hanson weeks of full-time work by several students,,little of the 1991). At Fort Rodd Hill, five 2x2 meter units were material had been sorted. With so little material excavated, and matrix screened through a 5 mm (1/5") processed, the decision was made to use a 6.4 mm screen mesh screen. All vertebrate fauna was analyzed (Hanson forrecovery in the new units opened during both the1995 1991). Comparisons with the Esquimalt Lagoon and Fort and 1997 field seasons. The 2.8 mm mesh was used for Rodd Hill faunas are based on numbers in Hanson (1991). column samples. All vertebrate material was identified by one author RESULTS (KS) and invertebrate material by the other (RW), To date, a total of 3707 elements have been analyzed although the latter is not discussed in this paper. from the Gulf of Georgia levels at Kosapsom, from three Vertebrate material was identified using both the classes (Table 1). As can be seen in the total numbers, STEWART and WIGEN: Screen Size and the Need for Reinterpretation 29 fish are most abundant, with considerable diversity, while Table 1. Numbers and percentages of faunal elements recovered mammalsare next in abundance, but with less diversity. from Gulf of Georgia levels at Kosapsom. Birds were poorly represented, with very little diversity. no. percent When the faurial analysis is sorted by screen size, Fish however, two very different pictures of the fauna emerge Squalusacanthiastdogfish) 41 4.5 (Fig. 2). In the fauna recovered using 2.8 mm mesh, fish Raja sp. (ray) 1 0.1 Clupea harengus (herring) 580 64.3is clearly the dominant class, comprising 75.8% of the Engraulis mordar (anchovy) 3 0.3 total number of elements, compared to less than half Oncorhynchus sp. (salmon) 187 20.8 (32%) recovered in.the 6.4 mm mesh. A similar reversal Gadus macrocephalus'(Pacific cod) 8 0.9 oceurs with mammalian elements, comprising only 19.6% Gasterosteus aculeatus of the 2.8 mm assemblage, but a much higher 62.1% in (3-spine stickleback) 10 1.1 Embiotocidae (surfperch) 18 2.0 the 6.4 mm assemblage. Birds were about equally, but Sebastes sp. (rockfish) 2 0.2 poorly, represented in both assemblages. Ophiodon e/ongams (lingcod) 16 1.8 Diversity and abundance also differ between Hexagrammidae (greenling) 2 0.2 screen-mesh sizes, particularly with fish taxa (Figs. 3, Cottidae (sculpin) 12 1.3 Hippoglossus stenolepis4). Thirteen different genera/families of fish were (Pacific halibut) 1 0.1 recovered with the 2.8 mm screen, while the assemblage Pleuronectiformes (flatfish) 22 2.4 recovered from the 6.4 mm screen contains only 10 genera/families. In addition, the abundance differs Mammals Rodentia (rodents) 8 24.2markedly between assemblages, with herring dominating Canidae (dogs or wolves; mainly dogs) 10 30.3the 2.8 mm fauna at 72.4%, but comprising a mere 5.3% Enhydra lutris (sea otter) 4 12.1 of the total in the 6.4 mm mesh assemblage (Fig. 3). A Mustelids (otters, minks etc.) 2 6.1 similar reversal occurs with salmon, which comprise Odocoileus hemionus (deer) 3 9.1 53.6% of the larger 6.4 mm mesh assemblage, but only Cervidae (deer, elk, etc.) 6 18.2 16.1% of the smaller 2.8 mm fauna. Among the less Birds numerous fish taxa (Fig. 4), dogfish, lingcod, and flatfish Anatidae (ducks, geese, etc.) 3 42.9 are most common in the 2.8 mm screen, while dogfish, Laridae (gulls) 4 57.1 surfperch, and sculpin dominate the 6.4 mm fauna. An analysis of the estimated total lengths of the fish recovered in each screen size shows more dissimilarities other than salmon and herring (Fig. 8), but, in general, (Fig. 5), with a clear peak for smaller-sized fish the sites using the 6.4 mm mesh screen had poor or no recovered in the 2.8 mm screen, and a dominance of representation of the smaller fishes, including surfperch, larger-sized fish in the 6.4 mm assemblage. sculpins, sticklebacks, and anchovies. When we Mammalian taxa also show dramatic differences compared mammals between the three sites (not between the two screen mesh sizes (Fig. 6), with dogs, graphed), we found no discernable pattern in the elements rodents (including beaver), and deer predominating in recovered. The smallest elements, those of rodents, were the 6.4 mm assemblage, but rodents, sea otters, and dogs/ equally represented in all sites, regardless of screen mesh deer predominating in the 2.8 mm assemblage. There used. Birds were not compared among sites, as few are three taxa that are only represented in one or the bird elements have been recovered at Kosapsom, and other size group. birds were not identified beyond class in the Esquimalt Comparison of the fish fauna of Kosapsom, sorted Lagoon assemblage. by screen size (6.4 mm and 2.8 mm meshes), with the fish fauna of Fort Rodd Hill, screened through 5 mm DISCUSSION mesh, and Esquimalt Lagoon, screened through 6.4 mm KOSAPSOM mesh (Fig. 7), shows that salmon predominates in the The faunal data retrieved from the 6.4 mm screens sites utilizing the larger screen meshes. In contrast, and from the 2.8 mm screens present very different herring overwhemingly predominates in the Kosapsom pictures of subsistence and economy at Kosapsom. If levels screened with 2.8 mm mesh. only the results from the 6.4 mm screen are used, Trends are not as distinct when examining fish taxa mammals are twice as common in the site refuse as 30 ZOOARCHAEOLOGY: Papers to Honor Elizabeth S. Wing ~ 6.4 mm/1/4" ~ 6.4 mm/1/4" ~ 2.8mm/1/8" I 2.8mm/1/8- - lili ~ Fish ' Salmon 1 1 ~ Hemng1 1. 1.. Mammals .--11 1 Birds , 1--v - -1 1 -=* 41 ~ Other 0 20 40 60 80 0 20 40 60 80 % of all fish elements % of all fish elements Figure 2. Percentages of each class, of all elements, by screen Figure 3. Percentages of salmon and herring elements, of all size at Kosapsom. fish elements. by screen size at Kosapsom. Dogfish Lingcod Flatfish ~ Surfperch Cods ~ 1 Sculpin Greenling 0 0 C O. 1 3 Halibut Anchovy ~ 6.4 mm/1/4" ~ 2.8mni/1/8" Stickleback 0 Rockfish 0 Ray 0 5 10 15 20 25 30 35 40 % of all fish elements Figure 4. Percentages of "other" fish elements by screen size at Kosapsom. STEWART and WIGEN: Screen Size and the Need for Reinterpretation 31 fish, and, among fish, salmon predominate, with almost % of all fish elements 10060% of the remains, followed by dogfish, lingcod, and flatfish. The mammals represented were primarily dogs or wolves, but size suggests mainly dogs, which 80 ethnohistoric accounts suggest were kept as pets, not usually for food (e.g., Drucker 1963). Next most important 60at the site were rodents, which, other than beavers, may have been site intruders, as the Coast Salish did not A consume smaller rodents (Suttles 1974), followed in 40 abundance by deer. Deer are very common today on Vancouver Island, and were likely common in Kosapsom times as well. They were hunted for food and their hides 20 for clothing, and bone and antler were used for tools and 00 0.'-- - - personal ornaments (e.g., Stewart 1987). Salmon, the 0 dominant fish represented in the 6.4 mm screenings, can < 10 10-30 30-50 50-100 >100 achieve a total length of 147 cm (Hart 1980) and were Estimated length of fish probably caught On spawning runs up rivers and streams. 6.4 mm (1.2") 2.8 mm (1.8") Dogfish, caught for food and for their skin, used for woodworking, can grow to about 130 cm (Hart 1980). Figure 5. Estimated length of fish by screen size at Kosapsom. Lingcod, a large greenling (up to 152 cm TL), is a valued food fish caught in shallow bottom waters. Flatfish, a 1980). After herring and salmon, dogfish is best diversified group, are well knownfood fish, ranging from represented in the 2.8 mm screening, followed by 50 to 267 cm, depending on species (Hart 1980). surfperch, sticklebacks, and sculpins, the majority being If only elements retrieved from the 2.8 mm screen smaller fish of <30 cm total length. are examined, a completely diffefent picture emerges Based on different screen sizes, in the 6.4 mm of Kosapsom subsistence and economy from that from assemblage, Kosapsom inhabitants were mammal hunters the 6.4 mm assemblage: fish elements outnumber and salmon eaters, while in the 2.8 mm assemblage, mammalian elements by almost4 to 1; among fish, herring inhabitants consumed mainly herring, with smaller amounts elements outnumber salmon, again almost 4 to 1. Among of salmon, rarely consuming mammals. We asked, why mammals, rodents are most numerous, again possibly such discrepancies? One scenario might be that, because intruders into the site. Sea otter elements, the next most the remains derive from different squares, but the same abundant, surprisingly are completely absent from the cultural levels, the creators of the 6.4 mm faunal 6.4 mm screen elements. Whether a rare inclusion, or assemblage were indeed primarily mammal and salmon only represented by small elements (teeth), the sea otter eaters, while the 2.8 mm fauna reflected different was valued for food, fur, and teeth for ornaments (Stewart preferences. This seems highly unlikely, however, because 1987). Dog and deer are next most abundant, but are the squares were all located in the same area, often not well represented in the 2.8 mm fauna. contiguous with each other, and such consistently different Among the fish in the 2.8 mm assemblage, herring, proportions are unlikely given the contemporaneity and notsalmon, is by far best represented, comprising 70.1% proximity of the squares to each other. of all fish, mammal, and bird elements identifiable to order A far more reasonable explanation is the obvious or lowen Herring generally grow to a maximum length one: that the larger, 6.4 mm screen allowed most of the of 25 cm in British Columbia (Hart 1980) and have been smaller fish bones, particularly the herring elements, to much prized in historic times for food and oil (Hart 1980). drop through. The smaller, 2.8 mm mesh screen retrieved Northwest coast cultures so revered herring, along with a more representative proportion of the fauna in the salmon and eulachon, thatthey performed annual rituals levels, particularly herring. to ensure good caches (Drucker 1963). The Gorge waterway, where Kosapsom is located, has historically ESQUIMALT LAGOON AND FORT RODD HILL been a herring fishery, and is fished today for herring. When the Kosapsom fauna are compared with the Herring spawn in late winter, most heavily in March (Hart Esquimalt Lagoon and Fort Rodd Hill site faunas, all 32 ZOOARCHAEOLOGY: Papers to Honor Elizabeth S. Wing Dog -t ~//p ./I'll// 9/Il/9/""Ilf-nor -WIdRodent -~I- il 7/plilt. 9.I Deer r 1Artiodactyl r Carnivore ....7=--9 1 0 1 Mustelids El_-_ul 0 r ~ 6.4 mm/1/4"0 ~ 2.8 mm/1/8" Sea Otter Dog or Wolf ' 0 5 10 15 20 25 30 % of elements Figure 6. Percentages of mammalian elements by screen size at Kosapsom. located less than 5 km from each other. clear differences exist in the fish exploited. The Esquimalt Lagoon and F-1 Kosap 6.4 mm Fort Rodd Hill inhabitants procured salmon and herring F7 Esqui 6.4 mm in similar proportions to that of the 6.4 mm fish fauna at ~ Ft. Rodd 5 mm Kosapsom, while the smaller-mesh fauna from I Kosap 2.8 mrn Kosapsom show salmon and herring in reversed proportions, perhaps the result of cultural differences between the sites, but given the sites' proximity and the Salmon I * 1. - similarity of results in sizes of screen meshes, it seems that mesh size is the important variable. Further, the coastal area. including Esquimalt Lagoon, is a well known ' spawning area for herring in both historic and modern Herring , times. It would be highly unusual if herring were not also exploited in prehistoric times, given its value among the historic Coast Salish (e.g„ Suttles 1974) and its ease of procurement when spawning (e.g., Drucker 1963). The Other ~ low proportions of herring at the Fort Rodd Hill site and. especially, the Esquimalt Lagoon site seem best explained by the use of the 6.4 and 5 mm meshes. 0 20 40 60 80 % of all fish elements CHANGING INTERPRETATIONS OF SUBSISTENCE AND EcoNOMY ON THE NORTHWEST COAST Figure 7. Percentage of salmon, herring, and other fish ele- The primary purpose of archaeological excavation.ments at Kosapsom. Equimalt Lagoon, and Fort Rodd Hill, with screen size. recovery. and analysis is to reconstruct past lifeways. A STEWART and WIGEN: Screen Size and the Need for Reinterpretation 33 long-standing view of the prehistoric Northwest coast ~--1 Kosap 6.4 mm cultures has been their overwhelming reliance on salmon, El Esqui 6.4 mm a view derived from historic ethnographies and ~ Ft. Rodd 5 inm archaeological reports. The often large size and I Kosal) 2.8 nirn distinctiveness of salmon bones, in particular vertebrae, have enhanced their preservation in archaeological sites, Doefish s .while remains of smaller or less robust fish either have -I.-4-not been preserved or are less visible. When only larger- mesh screens are used. it is predictable that salmon Flatfish 0 remains dominate to the near-exclusion of all but the ~IZZ] ~ largest of other fish species. Surfperch ~ While salmon was certainly a valued food fish and, indeed. a staple among Northwest coast cultural groups, Cod ~ ._ -~- reliance on it may not have been as complete as was 3 , I once thought. particularly in areas where other resources Sculpin ~ were easy or easier to procure. In a point relevant to this paper, Hart (1980:111) notes that pink salmon Greenling -6* (Oncorhvncus gorbuscha), the most abundant salmon I in British Columbia. spawn in a "very great number of Anchovy ~~ coastal streams and in all the major rivers with the 0 1 exception of those along the southeast part of Vancouver Stickleback 0 1 Island," where Kosapsom. Esquimalt Lagoon, and Fort 0 1 Rodd Hill are located. Other less abundant salmonid Other ~2--~~~ species spawn up the coastal rivers of southeastern Vancouver Island, so alternate resources were probably 0 10 20 30 40 50 utilized. % of all fish elements Because most archaeologists face time and labor Figure 8. Percentage of "other" fish elements at Kosapsom,constraints, this study does not advocate that screens Equimalt Lagoon, and Fort Rodd Hill, with screen size. with mesh 52.8 mm be used for all archaeological matrices. We recommend, however. that more than a column sample must be screened through small meshes. in press). because considerable numbers of herring For recognition of small fish, we suggest that 50% of vertebrae fall through the 2.8 mm screens. Although the matrix be screened through 6.8 mm and 50% through increased time and labor is involved, the data support a 52.8 mm mesh. more accurate reconstruction of coastal subsistence and economy. CONCLUSIONS The importance of herring to Northwest coast The Kosapsom data highlight the differences in data from peoples has been discussed in this paper, but, according faunal assemblages depending on screen size used in to both historic and ethnographic records, other small excavations. We believe that the importance of small fish, including eulachon. anchovy, smelt. and pilchard, fish may be highly underestimated in archaeological were highly valued. Being similar in size to herring, their reports from the Northwest coast because of faunal elements are only recovered in small-mesh screens. recovery strategies, both in past excavations where matrix Because several of these taxa are localized or only was not screened, thus small elements were lost. or in seasonally available, certain cultural groups on the coast more recent excavations where 6.4 mm screens have had access to them. while others did not. Some of these been used and. again, small elements have fallen through. fish were valuable. for example. eulachon for its oil. Screening matrices through 52.8 mm mesh will certainly Those groups with access to such fish could profit increase recovery and awareness of small faunal handsomely from trade. Use of fine-mesh screens and elements. Use of Still smaller meshes, e.g., 1.4 mm recovery of small elements could contribute an entirely ( 1/16") screens is even more preferable (Stewart et al. new set of data to be used for interpretation of Northwest 34 ZOOARCHAEOLOGY: Papers to Honor Elizabeth S. Wing coast subsistence and economy, and it is essential that of Georgia region of the Northwest Coast. Unpublished archaeologists make the effort to recover small elements. Ph.D. Dissertation, Simon Fraser University, Vancouver. Hart, J. L. 1980. Pacific Fishes of Canada. Ottawa: Fisheries Research Board of Canada.ACKNOWLEDGMENTS James, S.R. 1997. Methodological issues concerning screenI [KS] first met Elizabeth Wing at the ICAZ Fish Working size recovery rates and their effects on archaeofaunal Group meeting in York, England, in 1988. She was and interpretation. Journal of Archaeological Science 24: continues to be an inspiration to my work in 385-397. zooarchaeology and to everyone else who has met her Lyman, L. 1982. Archaeofaunas and subsistence studies. Pp. 331- or read her work. 393 in M. B. Schiffer, ed. Advances in Archaeological Method We are indebted to the editors of this volume for the and Theory, vol. 5. New York: Academic Press. invitation to contribute and for all ther work in putting it Payne, S. 1972. Partial recovery and sample bias: Theresults of together. KS is also grateful for a Canadian Museum of some sieving experiments. Pp. 46-64 in E. S. Higgs, ed. Papers in Economic Prehistory. London: CambridgeNature research grant and to the University of Victoria University Press.for use of their osteological collection. Rick, T.C. and Erlandson, J.M. 2000. Early Holocene fishing strategies on the California coast: Evidence from CA-SBA- REFERENCES CITED 2057. Journal ofArchaeological Science 27: 621-633. Butler, V. L. 1993. Natural versuscultural salmonid remains: Origin Schaffer, B.S. 1992. Quarter-inch screening: Understanding ofThe Dallesroadcutbones, ColumbiaRiver, Oregon, U.S.A. biases in recovery of vertebrate faunal remains. American Journal ofArchaeological Science 20: 1-24. Antiquity 57:129-136. Cannon,M.D. 1999. A mathematical modeloftheeffects of screen Schaffer, B.S. and Sanchez, J. L. J. 1994. Comparison of 1/8- size on zooarchaeological relative abundance measures. inch and 1/4-inch mesh recovery of controlled samples of Journal ofArchaeological Science 26(2): 205-215. small- to medium-sized mammals. American Antiquity 59: Clason, A. T., and W Prummel. 1977. Collecting, sieving and 525-530. archaeozoological research. Journal of Archaeological Stewart, H. 1987. The Adventures and Sufferings of John R. Science4: 171-175. Jewitt. Vancouver: Douglas and Mcintyre. Drucker, R 1963. Indians of the Northwest Coast. New York: Stewart, K.M., G Coupland, and D. Naughton. (in press). Effects The Natural History Press. of screen size on fish element recovery in northern Gordon, E. A. 1993. Screen size and differential faunal recovery: A Northwest coast middens. In K. Stewart, ed. Transitions Hawaiianexample. JoumalofFieldArchaeology 20: 453-460. in Zooarchaeology: New Methods and New Results. Grayson, D. K. 1984. Quantitative Zooarchaeology: Topics in the Ottawa: Canadian Museum of Nature. Analysis 6fArchaeological Faunas. Orlando: Academic Press.. Suttles, W. 1974. The Economic Life of the Coast Salish of Hanson, D. K. 1991. Late prehistoric subsistence in the Strait Haro and Rosario Straits. New York: Garland Publishing.