Toward a geography of foodways in the southern Gulf Islands,Pacific Northwest Coast

Abstract

Fine-screen animal bone and Pacific salmon ancient DNA (aDNA) results from Northwest Coast shell midden sites, together with other kinds of material culture, can provide detailed information on foodways, site-specific activities, and sociality. Seasonal use of the landscape may also be revealed through an understanding of place in the southern Gulf Islands of British Columbia, Canada. New results from column sample faunal analysis at the Pender Canal site are considered in conjunction with previously identified fauna. Alongside site characteristics, zooarchaeological and aDNA species identification data are employed to help reconstruct activities that people undertook. These tasks and their social implications at Pender Canal are contextualized with a discussion of several similar data sets from contemporary sites in the region. Temporal patterns in small fish remains and ancient salmon DNA at Pender Canal correspond with region-wide changes in land use, helping us interpret the formation of Coast Salish social relationships and identities over millennia.

Keywords

Foodways zooarchaeology landscape archaeology ancient DNA Coast Salish Pacific Northwest Coast Pender Canal site environmental remains

Introduction

This study focuses on an analysis of DeRt-1, Pender Canal, located in the southern Gulf Islands on the Pacific Northwest Coast. Several First Nations have a formal interest in the Pender Canal area, but Pender Island is considered to be within WSÁNEĆ core territory (Elliott, 1990; Jenness,1934–1935; Suttles, 1974). Data that are available from Pender Canal allow an archaeological engagement with specific aspects of social activity. Material culture together with detailed zooarchaeological and genetic data facilitate a microscale study of foodways during the inhabitation of this site. Documented architectural features, site dimensions, and location help access the nature and scale of routine activities. Characterizing individual archaeological sites through fine-grained examination provides a database of known places in the social landscape. The connections among these places and our knowledge of some of the pathways between them are a basis for subsequent discussion of regional interaction that could take several interrelated forms, from trade through ritual. It is also possible to explore continuities and changes that are visible in these interactions during the past.

Pender Canal was almost certainly inhabited by a lineage-based group who shared a history and possibly a specific common ancestor or origin myth. The social group at Pender Canal was closely related, and this is visible in numerous aspects of life, including subsistence routine. There are also several temporal patterns evident in the faunal data from Pender Canal which not only differentiate it from other Gulf Islands sites but also provide clues to some of the longer term events that were underway in the southern Strait of Georgia. Trends in the use of small fish species at DeRt-1, as well as potential patterning in the species composition of the salmon assemblage as revealed by ancient DNA (aDNA) analysis, may be related to changes in regional land use that took place at approximately 650 BC and AD 650. Beginning at c. 650 BC, the inhabitants of the region established a number of large village sites, including aggregations of villages in at least two locales. By c. AD 650, many of these village occupations were abandoned, and settlement patterns shifted to emphasize an increase in seasonal mobility among various resource use sites. People appear to have traveled more frequently through the landscape in smaller groups, inhabiting more residential camps and fewer villages. This pattern is evaluated using additional new data from recent test excavation at the WJOLELP site in WSÁNEĆ territory.

Pender Canal

The Pender Canal (DeRt-1) shell midden site is situated at the north end of the small canal that separates North and South Pender Island (Figures 1 and 2). The canal was dredged from Bedwell Harbour to Port Browning in 1902–1903 and destroyed parts of both DeRt-1 and DeRt-2. DeRt-2 may be found at the southern entrance to the canal at the head of Bedwell Harbour. The current research is focused on DeRt-1, although DeRt-2 has been discussed alongside DeRt-1 in previous publications (Carlson, 1987, 1990; Carlson and Hobler, 1993; see also Carlson, 1994; Carlson et al., 2017 for examples based primarily on DeRt-2). DeRt-1 is a medium-sized shell midden by regional standards (Acheson et al., 1975), with recorded site dimensions of 100 × 20 m (2000 m²). According to excavation records, in the central site area, the depth of shell midden deposits extended up to approximately 3 m below ground surface. It was from these midden matrices that the faunal remains discussed below were recovered. However, the midden furthest inland was only 2 m deep, and near the beach, it reached a depth of over 4 m, only two of which were above the high tide line (Carlson, 1987: 3; Carlson and Hobler, 1993: 41).

Figure 1.

Location of Pender Canal and other southern Gulf Islands area archaeological sites mentioned in the text.

Figure 2.

Pender Canal (DeRt-1) in 2020, looking southeast from the beach toward the wave-cut bank (photograph: Paul Ewonus).

Approximately 60%, or 308 m³, of DeRt-1 was excavated between 1984 and 1986 under the direction of Roy Carlson and Philip Hobler (Carlson and Hobler, 1993; see Figure 3). The cultural deposits consist primarily of shell-rich midden containing the occasional human burial. The obvious exception to the generalized shell midden is a compact black matrix which accumulated above an artificially leveled area of glacial till. This extensive stratigraphic unit contained a rock slab hearth, a stone slab storage cyst, a rectangular cache pit, oval depressions, a flat stone anvil, and several possible post hole depressions. Carlson and Hobler (1993) interpret this context as the remains of a living floor, and radiocarbon dates place its age between approximately 450 and 200 BC. While based on current evidence the seasonal timing of the use of this floor surface is only partly known, site use following 200 BC is better understood.

Figure 3.

1971 and 1984–1986 excavations at DeRt-1, Pender Canal. Excavation units sampled for previous level bag animal bone identification and current column sample analysis are indicated (see Carlson and Hobler, 1993: Figure 3; Hanson, 1991: Figure 4.3.).

As will be discussed in the following sections, Pender Canal was a seasonal camp inhabited from the late winter or early spring into at least the early summer during this later time period. Prior to 200 BC, site use ended in the summer, possibly in the early summer as is also indicated for the later camp inhabitation, although when each year people moved to DeRt-1 between 450 and 200 BC is yet equivocal.

Chronology of site use

Sixteen radiocarbon dates were obtained from DeRt-1 and range from 2660 ± 350 to 360 ± 200 ¹⁴C years BP (1690 BC–AD 1950, at 2σ). An additional 62 dates have been run on excavated material from DeRt-2, the earliest of which was 4370 ± 220 ¹⁴C years BP¹ (3635–2480 BC, at 2σ) (Carlson and Hobler, 1993: Tables 1 and 2; Carlson et al., 2017; Burial 84–12, δ¹³C = –13.5‰). At both these adjacent sites, Carlson and Hobler (1993) interpret a change in site use at approximately 2200 ¹⁴C years BP (c. 200 BC). The types of artifactual remains recovered prior to this time suggested village deposits to them, while a seasonal resource camp was considered to have existed afterward. Most of the zooarchaeological data discussed below are associated with the post-2200 BP (c. 200 BC) seasonal camp. Indeed, the majority of the previously obtained faunal results from DeRt-1 are from 1500 ¹⁴C years BP (c. AD 550) and later. Carlson and Hobler (1993) note that the radiometric dates form a continuous series and suggest this indicates some use of both sites respectively during their entire occupation spans. There is a transition in site use from village to seasonal camp at DeRt-1, although the village inhabitation is not as well understood. Despite significant site destruction due primarily to canal construction and erosion at Pender Canal, the proximity of DeRt-1 to DeRt-2 suggests that the sites may have constituted a single occupation prior to c. 200 BC. Immediately following c. 200 BC, site use at DeRt-1 was reduced for a short period as indicated by stratigraphic data and radiocarbon dating results.

Previous zooarchaeology

Previous zooarchaeological research at Pender Canal (DeRt-1), undertaken primarily under the direction of Diane Hanson (1991), included identification of vertebrate fauna recovered in ¼-inch hardware cloth screens from selected excavation units (2 through 7, 18, and 30). Hanson’s (1991, 1995) research also incorporated column sample analysis of bone and shell dated to between AD 550 and European contact from two excavation units, the larger 3.2 × 3 m Unit 18 and 1 × 2 m Unit 6 (Figure 3). A partial column sample from each unit was analyzed. The column samples were subdivided into 90% and 10% subsamples for 2.0 and 1.0 mm sieve fractions, and only 10% of fauna was identified from 1.0 mm fractions. The available reference collection at Simon Fraser University limited identifications of small fish bone recovered in the column samples.

The analysis of level bag fauna from several units and the majority of two column samples during and immediately following the excavation program at DeRt-1 has provided important information on the post-AD 550 occupation of the site (Garvin, 1987; Hanson, 1991, 1995). An important caveat is that representation of small fish could not be assessed with the comparative skeletal collections available to previous analysts. The present discussion, however, relies on the results of earlier DeRt-1 faunal work. An important goal of the current column sample analysis was to address some of the limitations inherent in the available zooarchaeological data from the site. Prior to presenting the results of this new analysis, it is important to outline the baseline of research that exists by summarizing work undertaken primarily by Hanson (1991). Most important to this discussion are the results of fish analysis, although mammal, bird, and invertebrate data will also be summarized.

Employing results presented by Hanson (1991), all previous fish data from DeRt-1 are considered together, including analysis of level bag fauna recovered in 6.4 mm (¼ inch) screens and identification of much of the fish bone from two column samples (see also Ewonus, 2011a). Fish were most abundant, by number of identified specimens (NISP), among identified vertebrates from the site. All identified level bag fish remains were associated with the uppermost precontact stratigraphic layer in excavation Unit 18, Stratum 6, dated to between 1580 and 160 radiocarbon years BP (c. AD 500–1800) (Carlson and Hobler, 1993; Hanson, 1991: 93, 97). Approximately half of the fish identified from level bags (total NISP 3578) are surfperch (NISP 1776). Pacific herring, though it is likely to be underrepresented in the recovered site fauna, comprises 15% of identified fish from the level bags. Nested screening of column samples (Units 6 and 18) resulted in a predominance of herring elements in 2 mm and 1 mm sieve samples. Results of the column sample analysis, for which 77% of identified fish were herring, indicate fine screening of a larger proportion of excavated sediment would certainly have increased the overall representation of this species. Rockfish (11%), cods (7%), salmon (5%), dogfish (4%), sculpin (4%), and other fish (4%), including greenling, lingcod, ratfish, flatfish, starry flounder, and prickleback, are also present in the Stratum 6 level bag sample. In the column samples, Pacific salmon is present in a NISP proportion identical to that identified from the level bags. The column samples did produce the first evidence for skate from the site, although the remaining taxa are also represented in the level bag faunal assemblage.

A total of 854 mammal bones were identified in the previous studies of DeRt-1 fauna, primarily associated with Stratum 6. Deer were the most common mammalian taxon by NISP, followed by canids, harbour seal, and delphinids. The delphinid family includes killer whales, dolphins, and porpoises, all common to the region historically (Banfield, 1974). Elk, sea lion, and beaver are also present, along with several smaller-sized rodents. Three small carnivores, mink, raccoon, and river otter, were identified in limited numbers. Overall, the mammal assemblage is similar to other sites in the region. The numbers of sea mammals present are notable for the Gulf Islands, although not numerous enough to be considered anomalous.

The small identified bird assemblage (NISP 49) from Stratum 6 includes a range of species. Various duck species predominate, although loons and birds of prey are also well represented. The golden eagle and turkey vulture, present at DeRt-1, are less common raptors than the bald eagle in Gulf of Georgia archaeological sites (see Ewonus, 2011a). Geese, gull, great blue heron, grouse, and a northern flicker were also identified. While mammal and bird bone was not identified in the column samples, representative information on invertebrates is available exclusively from the Unit 6 and 18 column samples (Hanson, 1991: Table 5.12, 1995: Table 5; Muckle, 1985). Bivalve clams are most abundant in the samples, followed by mussel, whelks, edible barnacles, and sea urchin. These data were collected primarily from analysis of the 6.4 mm (¼ inch) sieve fraction, and therefore, species with typically smaller shell fragments, such as mussel and sea urchin, are likely to be underrepresented.

According to Hanson’s (1991: 172, 1995: 46) analysis of post-AD 550 animal bone, DeRt-1 was inhabited primarily during the late winter, spring, and summer. The presence of herring and sea urchin suggested a late winter to spring season of site use. Sea urchins are commonly collected for their gonads, the weight of which is greatest prior to spawning in late winter to early spring (Hanson, 1991: 136; Himmelman, 1978; McLay, 1999: 47). Beyond these two seasonal indicators, several fish species available in the spring and summer were used to support the argument for continued site use into the summer. However, these species are also available, perhaps not as readily, in other seasons of the year. While the use of Pender Canal into the summer seems logical, only the late winter or spring seasons were firmly supported at the time.

Methods

Analysis of previously unstudied DeRt-1 column samples from 1 × 2 m excavation Units 3 and 4, which formed part of the 1-m wide trench excavated at the site in 1984, were chosen for study. Units 3 and 4 are centrally located within the site and, importantly, were accessible in the site material archive. Several column sample faunal bags from Unit 6 were encountered during the course of retrieving material from the other units, and these were included in the analysis.

Zooarchaeology

Vertebrate faunal identifications were made using a comparative collection which included 15 complete modern Northwest Coast fish specimens acquired specifically for this project and a loan of 11 further fish skeletons from the Zooarchaeology Laboratory Skeletal Comparative Collection in the Department of Anthropology at the University of Victoria, together with the existing collection housed in the Grahame Clark Laboratory for Zooarchaeology, Department of Archaeology, Cambridge University. Bone specimens that could not be satisfactorily identified to taxon in Cambridge due to limits of the available comparative collection were subsequently transported to the University of Victoria for complete identification using the Zooarchaeology Laboratory reference collection.

Identifications were performed according to the methods described in Ewonus (2011b). Faunal remains were identified to the most specific taxonomic category possible, given the limitations of observable morphological variation. The identifications of taxa and skeletal elements were made on the basis of direct comparisons with zoological reference specimens. A binocular microscope (up to 65× magnification) was used to assist with identifications. Faunal remains were quantified using standard zooarchaeological protocols (NISP, element, side, aspect, and bone weight). In a limited number of cases, aDNA analysis of Pacific salmon vertebrae allow species identification where only a determination to genus would otherwise have been possible (see below). NISP proportions, rather than counts, are used for intrasite comparisons at Pender Canal in order to help control for the absence of several individual sample bags from the archived material examined at Simon Fraser University.

Additionally, zooarchaeological results from column samples taken in 10 cm increments are combined into 20 cm intervals for summary data presentation and analysis. Results of the current column sample analysis are considered with reference to previous analyses but not combined for quantitative study. This is due to differences in analyst experience and the limits of the comparative collection available for the previous column sample analysis, specifically the absence at the time of many of the small fish species from the Simon Fraser University collection (Hanson, 1991).

These new column sample results are employed along with aDNA analysis of the salmon fishery at Pender Canal to further our understanding of site-level subsistence-related activity. Animal bone taxonomic representation and diversity are of primary interpretive significance, again with an emphasis on fish remains but also including mammal, bird, shellfish, and unidentified bone data. These zooarchaeological and genetic results are considered alongside other kinds of information about the site and its local environment in order to characterize site type and discuss seasonal inhabitation.

Genetic and osteometric analysis of salmon vertebrae

Sampling of salmonid vertebrae from the Pender Canal (DeRt-1) animal bone collection was random. All whole vertebrae from a single 1 × 2 m excavation unit, Unit 4, were measured for maximum transverse diameter, and a random sample of 10 was selected for aDNA analysis. Vertebrae of all diameters were included in the sample. A total of 43 whole salmonid vertebrae were identified from the examination of all Unit 4 level bag faunal material, which was recovered from 6.4 mm (¼ inch) screening of excavated unit sediments. This examination of DeRt-1 faunal level bags was undertaken at Simon Fraser University where the site material is archived. A single whole Pacific salmon vertebra was identified in the Unit 4 column sample. This vertebra was included in the sample population for genetic analysis. DNA extraction was conducted in the Simon Fraser University Ancient DNA Laboratory following a silica-spin column protocol (Yang et al., 1998) modified as described in Yang et al. (2008). Salmon species were identified through coamplification of salmonid d-loop (249 bp) and cytochrome b (cytb) (168 bp) mitochondrial DNA fragments using the approach described in Yang and Speller (2006), and the DNA authentication was conducted as in Ewonus et al. (2011).

In addition to the genetic analysis, the maximum transverse diameter of all whole salmonid vertebral centra was measured with digital calipers. This provides an approximate indication of the relative size of individual fish and may be used to structure DNA analysis sampling as well as provide information on the size composition of the salmonid sample from the site. The maximum transverse diameter was recorded for whole salmonid vertebrae encountered in column sample analysis as well as for Unit 3 and 4 level bag fauna. A selection of faunal level bags from Unit 3 was examined for whole salmonid vertebrae at Simon Fraser University. However, not all such bags from this excavation unit could be located among the archived collection, and therefore, Unit 3 salmonid vertebrae (n = 21 whole centra) were not included in aDNA analysis sampling.

Results of 2009 column sample analysis

The current column sample analysis provides additional animal bone data that have several advantages. Use of the skeletal comparative collection at the University of Victoria to supplement the collection established at Cambridge University has allowed identification of the full range of column sample taxa. As a result, numerous small fish specimens could be identified in the current analysis that were absent from earlier data sets. Furthermore, the current analysis can begin to address temporal patterns prior to AD 550, especially in fish remains. Together with previous results, the new faunal data can also illuminate the period after AD 550, a time when important shifts in region wide site use took place (see Ewonus, 2011a; Thompson, 1978). This section outlines the new results, emphasizing fish but also discussing bird and mammal remains identified in the columns.

Table 1.

Fish bone identified in 2009 column sample analysis (NISP) by depth below ground surface.

	20–40	40–60	60–80	80–100	100–120	120–140	140–160	160–180	180–200	200–220	220–240	240–260	260–280	280–300	Total	%Fish
Scorpaeniformes
Hemilepidotus hemilepidotus				2	1	7	1					1			12	0.8
Leptocottus armatus						1									1	0.1
Myoxocephalus polyacanthocephalus							3		1						4	0.3
Cottidae			1	1			2	1	1	1					7	0.5
Sebastes		1	1	11	2	4	3	2		3	1			1	29	1.9
Anoplopoma fimbria				1		1									2	0.1
Hexagrammos				3		1		1							5	0.3
Pleuronectiformes
Atheresthes stomias											1				1	0.1
Platichthys stellatus							2				1				3	0.2
Parophrys vetulus							2								2	0.1
Citharichthys sordidus						1									1	0.1
Gadiformes
Theragra chalcogramma					1					1					2	0.1
Salmoniformes
Oncorhynchus		1	2	2	5	3		9	1	5	10	2		11	51^a	3.3
Salmonidae			1			1	4		1						7	0.5
Clupeiformes
Clupea pallasii pallasii		41	94	37	123	281	100	58	124	145	115	14	9	26	1167	76.6
Engraulis mordax						14	1		12	14	4				45	3.0
Osmeriformes
Hypomesus pretiosus			2	1	9	17	3	9	6	26	12		1		86	5.6
Perciformes
Cymatogaster aggregata					1										1	0.1
Rhacochilus vacca		1				1									2	0.1
Embiotocidae		3	9	3	19	9	2	6	18	6	4				79	5.2
Batrachoidiformes
Porichthys notatus		1								1					2	0.1
Gasterosteiformes
Gasterosteus aculeatus aculeatus						4									4	0.3
Squaliformes
Squalus					5	1			1		1		1		9	0.6
Chimaeriformes
Hydrolagus colliei										1					1	0.1
Rajiformes
Rajidae					1										1	0.1
Miscellaneous
Osteichthyes	2	60	119	193	390	239	100	87	54	71	89	8	5	32	1449
Total	2	108	229	254	557	585	223	173	219	274	238	25	16	70	2973

^aFive Oncorhynchus vertebrae are greater than or equal to one-fourth of the centrum, and five Oncorhynchus elements are not vertebrae. The remaining 41 vertebrae are less than one-fourth of the centrum.

The fish identified in the current study increase the known richness of species, not only among small fish—northern anchovy (Engraulis mordax), surf smelt (Hypomesus pretiosus), and three-spined stickleback (Gasterosteus aculeatus aculeatus)—but also among medium-sized taxa such as sablefish (Anoplopoma fimbria), arrowtooth flounder (Atheresthes stomias), English sole (Parophrys vetulus), Pacific sanddab (Citharichthys sordidus), Alaska pollock (Theragra chalcogramma), shiner perch (Cymatogaster aggregata), and plainfin midshipman (Porichthys notatus) (Table 1). Most common in the column samples was Pacific herring (Clupea pallasii pallasii), representing 77% of identified fish by NISP, the same proportion evident in the earlier column analyses (Ewonus, 2011a). This correspondence suggests the two sets of data are comparable. There are some temporal fluctuations in the proportions of several identified fish, most noticeably herring (see Table 1). These are interpreted as sampling effects, either due to the limited volume of individual column samples or small changes in the use of space within the site. Surfperch (Embiotocidae) also show a similar periodic increase and decrease in their contribution to the fish assemblage. However, they represent a significantly smaller proportion of the fish (5.4%) than herring, emphasizing Hanson’s cautionary note regarding herring representation. The column sample data now clearly indicate that surfperch does not dominate fish remains from DeRt-1 and that smaller fish were likely more important to subsistence.

Second only to herring in NISP is another small-sized fish, surf smelt (5.6%). Northern anchovy contribute a further 3% to identified fish. While the small size of these fish species means that larger numbers are required to rival the caloric value of medium-sized and large fish, together herring, smelt, and anchovy represent 85% of the identified fish NISP. The distribution of northern anchovy and surf smelt through time is also unique. Herring, surfperch, salmonids (3.8% of fish NISP), and rockfish (Sebastes) (1.9% of fish NISP) are present throughout the column sample, while many other fish taxa appear intermittently in small numbers. Northern anchovy and surf smelt follow neither of these two basic patterns. They appear in the middle depths but are absent at the beginning and end of the sample. A single smelt vertebra appears in the 260–280 cm sample, although both smelt and anchovy first appear in larger numbers in the 220–240 cm sample. Based on radiocarbon dates from Units 3 and 4, these levels date to between 800 and 450 BC. Anchovy abruptly disappears from the column after 120 cm and smelt trails off to as high as 60 cm. These depths in Units 3 and 4 are associated with dates between approximately AD 450 and 700. Thus, the distribution of these two species corresponds well with the period of village aggregation discussed below. They are schooling fish best harvested using specialized techniques and equipment, which has implications beyond subsistence practice (see Discussion section below). Sculpins (Cottidae) account for 1.7% of the identified fish NISP, and their temporal distribution most closely follows the consistently present taxa (herring, surfperch, salmonids, and rockfish). The remaining 12 fish taxa together represent the final 2.2% of identified fish (Table 1). Figure 4 shows the proportionate contribution to the fish assemblage of all identified taxa with the exception of herring. This allows a survey of the range of species present in the column samples as well as providing an indication of the relative contributions of nonclupeid taxa to subsistence.

Figure 4.

Bar graph of percent NISP contribution to total identified nonclupeid fish from DeRt-1 2009 column sample analysis, including salmon vertebral fragments representing less than one-fourth of the centrum. Miscellaneous fish specimens are not included in the total.

While mammal and bird bone identified in the column samples appear in numbers too low to address changes through time (Ewonus, 2011a), they do enhance the information available from the level bag analysis. American mink (Mustela vison) is the most common mammal (60% of NISP), followed by small rodents (16% of NISP). Deer (Odocoileus) make a small contribution to the total of identified mammals, and despite the size of the matrix samples, sea lion (Eumetopias jubatus or Zalophus californianus) was identified from 140 to 160 cm near the middle of the column. A total of 67 mammal bones were identified, while only 14 bird bones were present. These include single duck (Anatidae) and crow (Corvus) elements, again from the 140 to 160 cm column sample.

The DeRt-1 salmon fishery

Salmonids were identified from throughout the column sample, absent only from 260 to 280 cm. Of a total of 51 Pacific salmon elements, 41 represent vertebral fragments smaller than one-fourth of the centrum. Thus, their importance to percent NISP should be considered inflated, although an adjustment will not be made in this instance (Figure 4). Table 2 presents the results of aDNA analysis of 10 randomly selected whole salmon vertebrae from Unit 4. The sampled vertebrae originate from throughout the inhabitation of DeRt-1, although there exist two broad groups. Vertebrae from levels 24 to 18 range in age from approximately 800 BC to AD 450, while specimens from levels 9 to 5 date to the period following AD 550. The sample sizes for these two periods individually are too small to be statistically significant. However, it is apparent that qualitatively speaking a tentative pattern is visible. Prior to AD 450 pink salmon (Oncorhynchus gorbuscha) is most common followed by sockeye (Oncorhynchus nerka) (2:1 ratio), although there are only six DNA results for this thousand year period. After AD 550, or perhaps as late as AD 650, a similar ratio exists (2:1) in an even smaller three specimen sample, with sockeye again identified as the secondary species. However, during this period, chum salmon (Oncorhynchus keta) replaces pink as the most common species identified.

Table 2.

Results of ancient DNA analysis of 10 randomly selected Pacific salmon (Oncorhynchus) whole vertebral centra from 1 × 2 m excavation Unit 4, DeRt-1, Pender Canal.

Sample	aDNA Identification	Diameter (mm)	Vertebra type	Unit	Level
1	Sockeye	7.54	Thoracic	4	5
2	Chum	10.93	Thoracic	4	7
3	Chum	8.07	Thoracic	4	9
4	Failed	8.22	Thoracic	4	18
5	Sockeye	9.43	Caudal	4	18
6	Pink	7.72	Caudal	4	22
7	Pink	6.90	Caudal	4	23
8	Pink	8.79	Thoracic	4	24
9	Pink	6.91	Caudal	4	24
10	Sockeye	8.63	Precaudal	4	24

aDNA: ancient DNA.

We suggest two observations are justified by these genetic results. First, sockeye appears as the secondary species in the salmon assemblage from DeRt-1 throughout the sampled sequence. Second, a shift in the most abundant species of salmon is suggested by the small data set to have occurred at c. AD 550–650. While further testing is desirable to confirm these results, qualitative interpretation of the DNA data indicate this may have been from a fishery where pink salmon was first targeted to one more reliant on chum salmon. This hypothesized change in the species composition of the salmon assemblage takes place at the same time as many of the large villages in the southern Strait of Georgia appear to be abandoned (Ewonus, 2011a).

When the genetic data are considered as a whole irrespective of temporal variation, the salmon fishery does not appear to be intensively focused on any one species. Rather, the three most common species in the region today (Groot and Margolis, 1991) are evident in the random sample. From highest to lowest NISP, they are pink (4), sockeye (3), and chum (2). While chinook and coho salmon are absent from the sample, the overall impression is a relatively generalized pattern of salmon harvesting. An approximately equal number of salmonid vertebrae measure greater than 8.5 mm maximum diameter as measure less than this arbitrary value (Table 3). The implication of these observations for season of site use is that DeRt-1 may not have been occupied late into the summer. The pink and chum salmon are likely to have been brought to the site as stored food, preserved from fishing near salmon streams such as the Fraser River during the fall. The higher fat content sockeye are more likely to represent summer or early fall catches consumed either fresh or stored for a shorter length of time (see Ewonus et al., 2011; Rozen, 1985).

Table 3.

DeRt-1 salmonid vertebral centrum maximum diameters, Units 3 and 4.

≥8.5 mm (n = 31)

11.65, 11.41, 11.35, 10.93, 10.85, 10.41, 10.05, 10.05, 10.04, 10.01, 9.88, 9.60, 9.59, 9.55, 9.47, 9.43, 9.39, 9.33, 9.11, 9.09, 9.03, 8.83, 8.81, 8.79, 8.78, 8.78, 8.73, 8.73, 8.63, 8.61, 8.57

<8.5 mm (n = 37)

8.44, 8.41, 8.35, 8.25^a, 8.22, 8.22, 8.21, 8.17, 8.07, 8.01, 7.90, 7.90, 7.79, 7.78, 7.74, 7.72, 7.72, 7.63, 7.54, 7.53, 7.47, 7.41, 7.27, 7.27, 7.26, 7.24, 7.13, 6.99, 6.96, 6.91, 6.90, 6.88, 6.83, 6.79, 6.27, 5.98, 2.73

Total = 68 measurements

Only vertebrae intact enough to permit a complete maximum diameter measurement are included.

^aUnit 6, level 220–230 cm below ground surface.

Alternatively, there may have been less focus on the summer sockeye fishery at Pender Canal than at other Gulf Islands sites. Another potential interpretation is that there was a change in the length of time the site was inhabited each year which corresponded with the change in site function from village to camp suggested by Carlson and Hobler (1993). However, material culture evidence indicates this change in site use took place at approximately 200 BC, while neither the column sample data nor the salmon DNA evidence show a shift at this time. Therefore, while subsistence-related changes appear to have occurred both earlier and later than 200 BC at DeRt-1, evidence for continuity in foodways at this time supports the interpretation that site seasonality did not change significantly over the period of occupation.

Discussion

The two sets of zooarchaeological results, gathered in the mid to late 1980s and 2008 to 2009, show a high degree of similarity. While the 2009 column sample results represent the fish assemblage most accurately, they are nicely complemented by the previous column sample and level bag results. In particular, the level bag fish bone adds several further species not present in the column samples. Three cods, Pacific cod, Pacific hake, and Pacific tomcod, as well as striped surfperch, lingcod, buffalo sculpin, cabezon, and the prickleback family (Stichaeidae) are present only in the level bag sample. Tomcod is somewhat of an exception, as it was identified by Hanson (1991) in both the level bag and Unit 18 column sample. The 2009 column sample analysis provided evidence of fish from 17 families, with only the pricklebacks (NISP 1) absent from among the site total of 18 fish families. The fish taxon most notably absent from the column samples was Pacific hake, representing 4.6% of the previous identified fish assemblage (level bag NISP 213). The combination of level bag and column sample analysis has resulted in the quantification of a large number of fish species, indicating high richness (Table 4). As expected, mammals and birds are best represented in the level bag results, although the column samples analyzed in 2009 contained two additional species.

Table 4.

Summary of archaeological site characteristics.

	Approx. site area (×100 m²)	Calibrated ¹⁴C age^a	House features	Food processing features	Mortuary features	Comparative artifact richness^b	Five most common fish taxa (%fish)^c	Three most common mammal taxa (%mammal)^c	Three most common bird taxa (%bird)^c	Interpretation of seasonal site use
DeRt-1, Pender Canal	2.83	c. 800–200 BC	Y?	N	Y	Unknown	Herring (73.9), smelt (9.3), salmon (6.7), northern anchovy (4.3), surfperch (2.4) [14]	Mink (100) [?]	Unknown	Late winter/early spring to summer village or camp (minimally)
DeRt-1, Pender Canal	2.83	c. 200 BC–AD 1750	N	N	Y	Moderate	Herring (76.6), smelt (5.6), surfperch (5.4), salmon (3.8), northern anchovy (3.0) [28]	Deer (31), canids (17), harbour seal (8) [12]	Ducks (65), loons (12), golden eagle (6) [14]	Late winter/early spring to summer camp
Dionisio Point	70	c. AD 200–650	Y	N	Y	Moderate-high	Herring (71.7), dogfish (6.6), rockfish (6.3), salmon (3.2), sculpins (3.0) [33]	Canids (18), deer (8), elk (4) [7]	Ducks (29), alcids (10), geese (5) [8]	Early spring to summer aggregate village
Harbour House	150	380 BC to AD 330 & later burials	N	N	Y	Moderate	Herring (90.0), cods (3.6), surfperch (2.4), salmon (2.2), rockfish (0.8) [9]	Artiodactyls (66), canids (24), black bear (2) [6]	Ducks (50), grouse (28), coot/rail (6) [5]^d	Spring and early summer camp
Cable Bay	75	780 BC–AD 320; AD 1290–1480	N	N	Y	Moderate	Salmon (28), rockfish (20), dogfish (13), flatfish (10), cod (10) [17]	Deer (35), elk (15), canids (14) [7]	Ducks (50), gull (18), loon (9) [7]	Fall or winter camp
WJOLELP	265	795 BC–AD 230, up to historic period	Y	Y	Y	Moderate-high	Herring (69.3), surfperch (8.5), salmon (7.8), northern anchovy (5.8), dogfish (4.0) [19]	Deer (74), harbour seal (9), northern sea lion (6) [6]	Ducks (50), great blue heron (17), grebe (17) [5]	Winter village from AD 100 onward

^aWhere necessary, radiocarbon dates include isotopic fractionation and marine reservoir corrections (ΔR = 350 years; Fedje et al., 2009).

^bEstimated with reference to a sample of 30 regional sites where similar analysis has been undertaken (Ewonus, 2011a).

^cBracketed values indicate taxonomic richness (NTAXA) tallied at the genus level. Percentages are calculated based on NISP, including only identified taxa.

^dOf the 22 identified bird elements recovered from Harbour House, 4 may represent species which were introduced during the postcontact period (Columbidae: doves and pigeons). Data summarized here do not include these 4 potentially nonindigenous specimens, resulting in a total identified bird NISP of 18. Emberizidae (American sparrows and relatives), NISP 2, were the third most common bird taxon identified, but Rallidae (rails and coots), NISP 1, were likely more important to subsistence and are therefore listed third.

There are several implications of these zooarchaeological results for understanding foodways at Pender Canal. Certain species of fish can be important to subsistence because of their own behavior. When taxa such as salmon and herring school tightly in confined waters, either oceanic or riverine, they may be harvested in very large numbers. Both these species were important to the Indigenous inhabitants of the southern Strait of Georgia, and their seasonal aggregation in preparation for spawning makes them potential clues to the timing of site use. Two other schooling species, surf smelt and northern anchovy, spawn multiple times during the year and are therefore not useful for estimating seasonality (Hart, 1973). However, during various portions of the year, they do school in large numbers near the shoreline. Anchovy school in bays and inlets while surf smelt prefer to spawn on beaches (Hunter and Goldberg, 1980). This makes them attractive resources because they can be taken in high numbers. In order to fish these species, the inhabitants of Pender Canal would have used specialized technology. Species such as rockfish, dogfish, and surfperch could be taken with hook and line or by spearing, as they are typically larger and do not normally school. Small schooling species, in contrast, are best harvested with nets, rakes, or perhaps traps (Rozen, 1978). The use of nets would be an inherently social practice. Depending on the target fish species, anywhere from 2 to 20 people might be involved in the endeavour (Suttles, 1974). Gillnets, which were used ethnographically for salmon fishing, might also have been used for smelt and anchovy. The preferred method of fishing for herring during the ethnographic period was using a specially constructed herring rake. This was usually a long wooden pole, flattened on two sides, into which a large number of bone or wood points were secured, creating a very long, sharply toothed rake. Herring were scooped up into a waiting canoe with the aid of these rakes. The manufacture and maintenance of this specialized fishing gear would also be an opportunity for social interaction. In good weather, this was likely an outdoor group activity, perhaps undertaken near canoes raised along the beach. In poor conditions, there was space to work inside plank houses or more temporary shelters, although for fine work the lighting would not be as good.

The introduction of smelt and anchovy fishing at DeRt-1 between 800 and 450 BC is suggestive of an increase in social interaction at the site. There is no reason to suspect that these species were absent from the area prior to this time. Thus, it is unlikely a case of a new resource becoming available. New technology might have been required to make it worthwhile fishing for these species, but similar techniques were already being used to take herring and salmon at the site and at numerous other sites in the Gulf of Georgia (Ewonus, 2011a; Stevenson, 1998). Technology for net making certainly existed prior to this time, although the history of development of the gillnet is entirely unknown. What we suggest is more plausible is that the labor requirements for harvesting these two small fish species became justified and easily facilitated at DeRt-1 some time around 650 BC. If this was a time of population aggregation at the site and throughout the region, then the numbers of people required to effectively fish for smelt and anchovy would have been available. Moreover, with a larger population living seasonally at the site, it might have been advantageous to include a wider array of taxa in the subsistence quest, utilizing especially those fish that could be harvested in quantity. The phasing out of these fish from the diet around AD 550–650 at DeRt-1 also coincides with regional changes in land use. A return to smaller groups inhabiting a larger number of sites might make it difficult, and perhaps no longer necessary, to undertake group fishing activities for all but the most important fish.

This brings us to a discussion of two of these species at DeRt-1, species that were important to regional subsistence to the extent that seasonal schedules and ceremonial activity revolved in part around their harvest (Suttles, 1990). Pacific herring was obviously an important food to the site’s occupants, accounting for 77% of fish NISP. However, the range of other fish taxa together with various mammal and bird species indicates that while herring was important, it was not a dominant focus of subsistence practice. Pacific salmon was not a focal species, but it was an important resource in the broader region (Coupland et al., 2010; Ewonus, 2011a). Both species were fished using specialized technology, the production, use, and maintenance of which were primarily communal activities. A number of small bone points, suitable for use in herring rakes as well as other fishing gear, were recovered from DeRt-1. Salmon fishing was almost exclusively a communal activity in the Gulf Islands and nearby Fraser River delta. In ethnohistoric accounts, an important method of harvesting these fish was by employing a gillnet, either spread out between two large canoes or anchored to objects near the mouth of the Fraser River and other regional streams (Barnett, 1955; Suttles, 1974). Whether gillnets might have been used during part of the sequence at DeRt-1 cannot be assessed with the current evidence, although some type of nets were undoubtedly in use from the site’s earliest inhabitation.

The change in the species composition of the small salmon DNA sample from pink and sockeye to chum and sockeye, in a 2:1 ratio, between approximately AD 550 and 650 is intriguing. Does it suggest a change in peoples’ use of the broader landscape? Because pink and chum are more likely to have been taken near salmon streams on the mainland or southeastern Vancouver Island, the changing species composition may indicate a change in another aspect of the, “off site,” seasonal round. While prior to this time those living at DeRt-1 fished for pink salmon in a certain location during a specific time of the year, likely before coming to Pender Canal, there was a change to a different combination of location and timing when chum became most common in the site deposits after AD 550–650. The temporal correspondence with significant land use changes in the southern Strait of Georgia supports this interpretation. An alternative might be a change in the spawning patterns of the individual species in the location frequented, or possibly owned, by those who later traveled to Pender Canal. Currently, no ancient population genetics data exist to attempt to test this hypothesis, although it would be very difficult to assess. Generally speaking, a more likely confounding explanation is simple sampling bias. Nevertheless, this pattern remains of interest and warrants further testing.

We would avoid suggesting this change in the salmon assemblage indicates a change in the season of site use at DeRt-1. The available zooarchaeological results do not provide any compelling evidence for a change in seasonality during the site’s occupation, although we lack a sizable sample of mammals and birds from the deposits older than AD 550. The secondary nature of sockeye among the salmon, however, may indicate that the summer inhabitation of Pender Canal was more limited than evident at the Dionisio Point site (DgRv-3), located on Galiano Island (Ewonus et al., 2011; see below). What is unclear at present is when during the year people came from other locations to DeRt-1. This would have occurred some time between the late fall and early spring, but no clear line of environmental evidence is currently available which allows us to pinpoint this move. Using the zooarchaeological data, we can infer site use from the herring fishery in late winter to early spring into at least the early summer; thus, Hanson’s hypothesis of a spring to summer seasonality for the post-AD 550 period at DeRt-1 is well supported. Material culture evidence helps resolve this issue following 200 BC. It indicates the site was clearly not a winter village during this later period and therefore is unlikely to have been inhabited during this season.

Prior to c. 200 BC, part of the picture remains unclear. Artifacts, mortuary evidence, and features associated with the buried occupation surface indicated to Carlson and Hobler (1993) that the site was a village, possibly a winter village. It is not improbable that once villagers had completed salmon fishing and storage activities at another location, likely near a salmon stream somewhere nearby in the Salish Sea, they traveled with some of those stores to Pender Canal in the late fall or early winter. Due to the destruction of a significant portion of DeRt-1 by wave erosion resulting from recent sea level rise of approximately 1.5 to 2 m (Fedje et al., 2009) and the dredging of the canal itself, any evidence for the extent of house structures that may have existed at the site has been destroyed. Only the single occupation surface, interpreted by Carlson and Hobler (1993) as a house floor, remained by the 1980s, and further structures are unlikely to be found in the unexcavated portion of the site. As one partial house floor does not constitute a village in its own right, it is more difficult to make the argument that DeRt-1 was a winter village. However, the large number of human burials uncovered at DeRt-2, which has also been heavily impacted by erosion and canal construction, is suggestive of the rear of a major village site. It is not uncommon for shell midden burials to be situated in older midden accumulations at the back of large villages, when these deposits are preserved (Arcas Consulting Archeologists Ltd., 1999: 41–43; Borden, 1970; Carlson and Hobler, 1993: 36). Therefore, while we know that DeRt-1 was inhabited from the spring into the summer throughout its known use and that it was potentially a village of some kind prior to approximately 200 BC and a seasonal camp afterward, we do not know whether the village occupation began in the winter or the spring. A method of evaluating this data gap would be to examine the stable isotope profiles of shellfish from the site, most notably the butter clam (Saxidomus gigantea) (Burchell et al., 2013; Hallmann et al., 2009).

Despite the uncertainty over the yearly initiation of occupation at DeRt-1 prior to 200 BC and the limited, although not unsubstantial, evidence for a possible village inhabitation between 450 and 200 BC, a great deal is known about the site. It existed within a landscape populated with other sites and places, the most influential of which for the present study is the DeRt-2 shell midden, situated approximately 100 m to the south. It is possible that a similarly dated house floor deposit (2200 ± 120 BP, c. 200 BC) was uncovered prior to the 1984–1986 excavations at DeRt-2 and was subsequently completely destroyed by wave erosion (Carlson and Hobler, 1993: 39). The date on this dense black layer, 2 m below the ground surface and associated with a further small sandstone slab storage cyst, was obtained in 1963 by Wilson Duff, then British Columbia Provincial Archaeologist, on a sample recovered during test excavation in the 1950s by the Provincial Museum. Approximately 80% of DeRt-2 was excavated as part of the same project that investigated DeRt-1 between 1984 and 1986, in the process cataloguing 3719 artifacts and recording over a hundred intact human burials. Thus, the pre-200 BC village hypothesis is strengthened by the results of work at DeRt-2. However, no significant faunal or floral analyses have been initiated, which would allow for further comparison. Looking instead more closely at the post-200 BC inhabitation of Pender Canal, which chronologically overlaps the use of several additional Gulf Islands sites, the parallels and differences in foodways among these sites help enhance their overall interpretation.

Contemporaneous Gulf Islands sites

Excavation results from the Harbour House site (DfRu-3), situated in Ganges Harbour on Saltspring Island, suggest it was a seasonal camp focused on shellfish gathering and herring fishing (Arcas Consulting Archeologists Ltd., 1993). Despite a small volume of hand excavated sediment and the resulting small artifact assemblage, a detailed faunal analysis provides useful data from the site (Table 4). Fish at Harbour House comprises 1160 of a total vertebrate NISP of 1311 and are dominated by Pacific herring remains (90% of identified fish). Cods are next most abundant (by NISP) at 3.6%, followed by surfperch at 2.4% and Pacific salmon at 2.2%. The remaining 1.8% of identified fish includes rockfish, Pleuronectidae (righteye flounders), dogfish, greenling, and ratfish. Significant amounts of shell observed in the midden deposits together with a subsistence focus on herring fishing suggested to the excavators that this part of the site was likely a spring and early summer camp. A comparatively limited representation of salmon and the characteristics of the recovered material culture support this inference.

Cable Bay (DfRu-74) is located on the northeastern shore of Galiano Island facing the mainland of British Columbia. It is one of the more unusual sites in the southern Strait of Georgia. Faunal data suggest that the site was most likely inhabited some time during the fall or winter (Wilson, 1992). However, there is no known evidence for houses, and the site is not large (Table 4). Pacific salmon account for 28% of the identified fish from Cable Bay, while herring contribute only 5%. Only a part of the excavated site sediment was screened through 3.2 mm (1/8 inch) mesh, while the majority was passed through 6.4 mm (1/4 inch) screens. Regardless of the potential for underrepresentation of herring, salmon appears to have been a major food species at the site. It is likely that such numbers of salmon could only have been caught close to, or along, a spawning stream, preserved and transported to the site. Following salmon, the next most abundant identified fish were rockfish (20%) and dogfish (13%).

Harbour House and Cable Bay are two southern Gulf Islands seasonal camp sites that were contemporaneous with the Pender Canal seasonal camp. An examination of foodways at these sites indicates three significantly different engagements with local Gulf Islands environments. The social character of these interactions would also have varied. Beyond seasonal camps, village occupations such as Dionisio Point were equally variable in nature. Both Pender Canal and Dionisio Point were likely inhabited during a similar time of year, spring and summer, although Dionisio Point was a village and Pender Canal a seasonal camp. Village architecture is not in evidence at Pender Canal during the time that Dionisio Point was occupied, and there are some corresponding differences in other types of material culture, such as artifacts. The majority of variation in the faunal assemblages from the two sites is visible in fish remains (Table 4). We observe a similar degree of difference at Harbour House and Cable Bay, in material culture remains and importantly, fish relative abundance.

The WJOLELP site

We can test the hypothesis that regional subsistence changes took place in the southern Strait of Georgia at c. AD 650 with recent results from excavation of the WJOLELP site (DdRu-6) (Ewonus and Bowman, 2020). The WJOLELP site is situated in WSÁNEĆ territory, as is Pender Canal, but is located on Tsartlip South Saanich IR1 north of Brentwood Bay, BC (Figure 1). During the historic period, the WJOLELP site was a named village (Elliott, 1990). In 2018, test excavation of the main site area was directed by Paul Ewonus, in collaboration with the Tsartlip First Nation. Four radiocarbon dates on wood charcoal (D-AMS 031780-83; including macrobotanical identifications) provide an age range at 2σ from 795 BC to AD 230, with median calibrated ages from 640 BC to AD 160, although site use continued after this time (Table 4). An initial Marpole phase occupation is indicated, possibly including a late Locarno Beach phase component. Site use continued during most of the Marpole phase and likely later into the historic era. Archaeological results from the site thus far do not include house structural elements, such as large post molds, but excavation has been limited in extent. During the excavation project, two hearth features were encountered (one was in secondary context and could also represent a roasting pit). Evidence for large-scale shellfish processing, including burning, and a variety of artifact types, including antler wedges and sandstone abraders, was observed. In conjunction with oral histories of village inhabitation, these data suggest the site was a village prior to the colonial era. Whether the initial use of the site was a village occupation or some other type of habitation is not certain at this time.

Fish taxonomic richness is moderate in the recovered assemblage (fish NISP = 745). Taxonomic evenness is low to moderate, as a number of taxa are present in small quantities, while six species are much more abundant. Faunal data indicate an abundance of herring and salmon and suggest a winter to spring site occupation, at minimum, particularly from the late Marpole phase onward. The earlier evidence for seasonality is equivocal, with the exception of a probable spring occupation revealed by high herring relative abundance. Temporal trends in fish representation suggest a shift in subsistence patterns at approximately AD 650. Pacific herring, northern anchovy, and surfperch decrease in taxonomic abundance around this time, while salmon and rockfish increase in abundance. The significant decrease in relative abundance of small fish species occurs at the same time that small fish disappear from the assemblage at Pender Canal. The initial occupation of the WJOLELP site was likely at least a century later than at Pender Canal. As a result, we may not have evidence from a period where these small fish were absent from the WJOLELP assemblage. While the sample size is modest, the results of test excavation and zooarchaeological analysis at WJOLELP support the pattern established based on work at Pender Canal.

Conclusion

Much of the variability that exists among the four Gulf Islands sites may be explained by differences in social dynamics. Perhaps more, we would suggest, than the effects of resource distributions or other ecological influences. The social nature of the group inhabiting Dionisio Point would have been noticeably different than that of their contemporaries at Pender Canal (Ewonus, 2017). It is difficult to address whether there were differences in population numbers between the four sites, due to the destruction of part of DeRt-1 over the past few thousand years and limits to archaeological investigations at Harbour House and Cable Bay. Nonetheless, the nature of the social aggregate at Dionisio Point would have been different than what was probably a lineage-based group at Pender Canal. This group would most likely have arrived at the DeRt-1 camp together as a unit in the late winter or early spring. During the summer, they may have left the locale again as a group, moving toward a more productive salmon fishing location. There are a number of significant salmon spawning streams in the southern Strait of Georgia. Therefore, it is also possible that the Pender Canal lineage group split into smaller related family groups to join with relatives in several places within the region.

While they were living at Pender Canal, however, these individuals would have had very similar interests and needs. Following the winter season, rich with ceremony and social activities although much more limited in terms of the availability of nonstored food resources, spring would have been an important time for undertaking subsistence activities. The inhabitants of Pender Canal would have worked together to fish first for herring and perhaps to harvest sea urchin but also to obtain a wide variety of animal and plant foods in the newly revitalized environment. Harvesting these resources meant working together for many regular and periodic activities as well as employing knowledge of the surrounding landscape and the specific locations of animals and plants within it. Both these kinds of food classes were found at different times in different places, and experience, both personal and learned from oral tradition, the experience of one’s ancestors, was important to successful fishing, hunting, and plant gathering. Moving through the landscape on a regular basis was also necessary in order to fish and hunt. Whether people traveled long distances or focused on the landscape around Pender Island, they would have made these trips primarily together in small groups. Because these people shared a common history and broadly speaking, genealogy, their preferences for and traditions associated with different animal foods would have been similar.

It was at this quotidian scale that group identities developed in the southern Gulf Islands. As time passed, traditions developed from routines and daily practice, formalizing shared repertoires into communities (Wenger, 1998). Regional shifts in land use and foodways, such as occurred at c. AD 650, were based on the culmination of small changes at Pender Canal and elsewhere. Subsistence pattern shifts in small fish remains and possibly salmon aDNA at this time that are observed at Pender Canal are visible at the contemporaneous WJOLELP site in the relative abundance of small fish species. As in all large-scale processes, variability may be seen in specific locales and at individual archaeological sites. While results are preliminary, the WJOLELP village does not appear to have been abandoned at c. AD 650, as were a number of other villages in the southern Strait of Georgia. Nevertheless, at approximately this time people’s foodways changed at the site, in concert with regional trends. The consistency of this pattern in this part of the Salish Sea is of significant interest, and additional research is necessary to more fully understand its meaning.

Footnotes

Acknowledgments

We wish to sincerely thank the following First Nations with territory in the southern Gulf Islands for this opportunity to work in their territories: Tsawout First Nation, Tsartlip First Nation, Tseycum First Nation, Pauquachin First Nation, Cowichan Tribes, Penelakut Tribe, Lyackson First Nation, and Stz’uminus First Nation. Special thanks to Councillor Simon Smith Jr and Karen Harry of Tsartlip First Nation. We also wish to thank Rebecca Wigen for kindly providing lab space and access to the University of Victoria Zooarchaeology Laboratory skeletal comparative collection. Shannon Wood was most helpful in providing access to DeRt-1 faunal material and column samples archived at Simon Fraser University. We are grateful to Arcas Consulting Archeologists Ltd. and Ian Wilson for making available the unpublished results of their investigations at Harbour House and Cable Bay.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Research funding was provided by the Social Sciences and Humanities Research Council of Canada, the Cambridge Commonwealth Trust, and Vancouver Island University.

Note

Author Biographies

Paul A Ewonus received his PhD in Archaeology from Cambridge University. He is currently a senior archaeologist at a consulting company in Victoria, BC and an honorary research associate in the Department of Anthropology at Vancouver Island University. His research interests include landscape archaeology, social zooarchaeology, and community-based archaeology.

Camilla F Speller received her PhD in Archaeology from Simon Fraser University. She is currently an assistant professor in the Department of Anthropology at the University of British Columbia. Her research interests include molecular environmental archaeology, ancient DNA, and ancient proteomics.

Roy L Carlson received his PhD in Anthropology from the University of Arizona. He is currently an emeritus professor in the Department of Archaeology at Simon Fraser University. His research interests include cultural history, precontact Indigenous art, and Northwest Coast and Southwest prehistory.

Dongya Y Yang received his PhD in Anthropology from McMaster University. He is currently a professor in the Department of Archaeology and director of the Ancient DNA Laboratory at Simon Fraser University. His research interests include ancient DNA, physical anthropology, and environmental bioarchaeology.

References

Acheson

Cassidy

Claxton

(1975) Report of the archaeological survey of the Southwestern Gulf of Georgia. Report submitted to the British Columbia Archaeology Branch, Victoria. Unpublished Permit Report (1975-6).

Arcas Consulting Archeologists Ltd. (1993) Archaeological investigations at the Harbour House site (DfRu 3), Saltspring Island, B.C. Report submitted to the British Columbia Archaeology Branch, Victoria. Unpublished Permit Report (1993-83).

Arcas Consulting Archeologists Ltd. (1999) Archaeological investigations at Tsawwassen, B.C.: Volume IV, interpretation. Report submitted to the British Columbia Archaeology Branch, Victoria. Unpublished Permit Report (1989-41 and 1990-2).

Banfield

AWF

(1974) The Mammals of Canada. Toronto: University of Toronto Press.

Barnett

(1955) The Coast Salish of British Columbia (University of Oregon Monographs, Studies in Anthropology No. 4). Eugene: University of Oregon Press.

Borden

(1970) Culture history of the Fraser Delta region: An outline. In: Carlson RL (ed) Archaeology in B.C., New Discoveries: BC Studies. Vol. 6–7 (Special Issue), pp.95–112.

Burchell

Cannon

Hallmann

, et al. (2013) Refining estimates for the season of shellfish collection on the Pacific Northwest Coast: Applying high-resolution stable oxygen isotope analysis and sclerochronology. Archaeometry 55: 258–276.

Carlson

(1987) The 1986 excavations at the Canal site (DeRt 1 and DeRt 2). Report submitted to the British Columbia Archaeology Branch, Victoria. Unpublished permit report (1986-7).

Carlson

(1990) The Northwest Coast before A.D. 1600. In: Crownhart-Vaughn

(ed) The North Pacific to 1600: Proceedings of the Great Ocean Conference. Portland: Oregon Historical Society Press, Vol. I, pp.109–136.

10.

Carlson

(1994) Trade and exchange in prehistoric British Columbia. In: Baugh

Ericson

(eds) Prehistoric Exchange Systems in North America. New York: Plenum Press, pp.307–361.

11.

Carlson

Hobler

(1993) The Pender Canal excavations and the development of Coast Salish culture. BC Studies 99: 25–52.

12.

Carlson

Szpak

Richards

(2017) The Pender Canal site and the beginnings of the Northwest Coast cultural system. Canadian Journal of Archaeology 41: 1–29.

13.

Chisholm

(1986) Reconstruction of prehistoric diet in British Columbia using stable-carbon isotopic analysis. Unpublished PhD Dissertation, Simon Fraser University, Burnaby, BC.

14.

Coupland

Stewart

Patton

(2010) Do you never get tired of salmon? Evidence for extreme salmon specialization at Prince Rupert Harbour, British Columbia. Journal of Anthropological Archaeology 29: 189–207.

15.

Elliott D (1990) Saltwater People: As told by Dave Elliott Sr. A Resource Book for the Saanich Native Studies Program. Poth J (ed), 2nd ed. Saanich, BC: School District 63 Native Education.

16.

Ewonus

(2011a) Social landscapes of the Southern Strait of Georgia, Pacific Northwest Coast. Unpublished PhD Thesis, Department of Archaeology, University of Cambridge.

17.

Ewonus

(2011b) Social zooarchaeology of a Northwest Coast house. The Journal of Island and Coastal Archaeology 6: 72–97.

18.

Ewonus

(2017) Assessing internal household relationships and site use: Zooarchaeological evidence from Dionisio Point, Galiano Island, Canada. North American Archaeologist 38: 349–393.

19.

Ewonus

Bowman

(2020) Tsartlip IR1 road and drainage improvements: A report on site alterations to the WJOLELP site (DdRu-6). Non-permitted report submitted to the Tsartlip First Nation, and on file at the British Columbia Archaeology Branch, Victoria.

20.

Ewonus

Cannon

Yang

(2011) Addressing seasonal site use through ancient DNA species identification of Pacific salmon at Dionisio Point, Galiano Island, British Columbia. Journal of Archaeological Science 38: 2536–2546.

21.

Fedje

Sumpter

Southon

(2009) Sea-levels and archaeology in the Gulf Islands National Park Reserve. Canadian Journal of Archaeology 33: 234–253.

22.

Garvin

(1987) Vertebrate remains and stratigraphic zone correlations at the Pender Canal sites, DeRt-1 and DeRt-2. In: Carlson

(ed) Studies of Faunal Remains from the Pender Canal Excavations: DeRt 1 and DeRt 2. Unpublished report submitted to the British Columbia Archaeology Branch, Victoria, pp.49–100.

23.

Groot

Margolis

(eds) (1991) Pacific Salmon Life Histories. Vancouver: UBC Press.

24.

Hallmann

Burchell

Schöne

, et al. (2009) High-resolution sclerochronological analysis of the bivalve mollusk Saxidomus gigantea from Alaska and British Columbia: Techniques for revealing environmental archives and archaeological seasonality. Journal of Archaeological Science 36: 2353–2364.

25.

Hanson

(1991) Late prehistoric subsistence in the Strait of Georgia Region of the Northwest Coast. Unpublished PhD Dissertation, Simon Fraser University, Burnaby, BC.

26.

Hanson

(1995) Subsistence during the late prehistoric occupation of Pender Canal, British Columbia (DeRt-1). Canadian Journal of Archaeology 19: 29–48.

27.

Hart

(1973) Pacific Fishes of Canada. Ottawa: Fisheries Research Board of Canada Bulletin 180.

28.

Himmelman

(1978) Reproductive cycle of the green sea urchin, Strongylocentrotus droebachiensis. Canadian Journal of Zoology 56: 1828–1836.

29.

Hunter

Goldberg

(1980) Spawning incidence and batch fecundity in northern anchovy, Engraulis mordax. Fishery Bulletin 77: 641–652.

30.

Jenness

(1934–1935) The Saanich Indians of Vancouver Island. Typescript of unfinished manuscript on file in Anthropological Collections Section, Royal British Columbia Museum.

31.

McLay

(1999) The diversity of Northwest Coast shell middens: Late pre-contact settlement-subsistence patterns on Valdes Island, British Columbia. Unpublished MA Thesis, University of British Columbia, Vancouver.

32.

Muckle

(1985) Archaeological considerations of bivalve shell taphonomy. Unpublished MA Thesis, Simon Fraser University, Burnaby, BC.

33.

Rozen

(1978) The ethnozoology of the Cowichan Indian people of British Columbia I: Fish, beach foods, and marine mammals. Unpublished report on file at the British Columbia Archaeology Branch, Victoria.

34.

Rozen

(1985) Place names of the Island Halkomelem Indian people. Unpublished MA Thesis, University of British Columbia, Vancouver.

35.

Southon

Fedje

(2003) A post-glacial record of 14C reservoir ages for the British Columbia coast. Canadian Journal of Archaeology 27: 95–111.

36.

Stevenson

(1998) Wet site contributions to developmental models of Fraser River fishing technology. In: Bernick

(ed) Hidden Dimensions: The Cultural Significance of Wetland Archaeology. Vancouver: UBC Press, pp.220–238.

37.

Suttles

(1974) Coast Salish and Western Washington Indians I: The Economic Life of the Coast Salish of Haro and Rosario Straits. New York: Garland Publishing.

38.

Suttles

(1990) Central Coast Salish. In: Suttles

(ed) Handbook of North American Indians, vol. 7, Northwest Coast. Washington, DC: Smithsonian Institution, pp.453–475.

39.

Thompson

(1978) Prehistoric settlement changes in the Southern Northwest Coast: A functional approach. Reports in Archaeology 5. Seattle, WA: University of Washington.

40.

Wenger

(1998) Communities of Practice: Learning, Meaning, and Identity. Cambridge, England: Cambridge University Press.

41.

Wilson

(1992) Excavations at DfRu 74, Cable Bay, Galiano Island. Report submitted to the British Columbia Archaeology Branch, Victoria. Unpublished Permit Report (1991-76).

42.

Yang

Eng

Waye

, et al. (1998) Improved DNA extraction from ancient bones using silica-based spin columns. American Journal of Physical Anthropology 105: 539–543.

43.

Yang

Liu

Chen

, et al. (2008) Wild or domesticated: DNA analysis of ancient water buffalo remains from north China. Journal of Archaeological Science 35: 2778–2785.

44.

Yang

Speller

(2006) Co-amplification of cytochrome b and D-loop mtDNA fragments for the identification of degraded DNA samples. Molecular Ecology Notes 6: 605–608.