Introduction

Coastal Natives encountered by European and American voyagers in British Columbia (BC) and southeast Alaska in the late 1700s had a long-established coastal adaptation that made extensive use of wood and bark from the surrounding temperate rainforest. The Kwakiutl, Tsimshian, Haida and Tlingit Indians lived in large communal houses of hewn logs, traveled in long sea-going dug-out canoes, and made many wood and bark utensils from Picea sitchensis (Sitka spruce), Tsuga heterophylla (western hemlock) and especially Thuja plicata (western red cedar) and Chamaecyparis nootkatensis (yellow cedar). Bark was typically peeled from a live tree by making a horizontal cut near the base of the trunk with an adze, chisel or axe, then prying the sheet of bark upward to meet another horizontal cut marking the desired length. Or—especially in the case of Thuja and Chamaecyparis—the upper cut was omitted and the bark was stripped up the trunk along the natural fiber alignment to yield a piece sometimes over 10 metres long (Steward 1984). Usually the tree was not completely girdled, and the scar was left to heal.

On the Pacific coast of North America (Fig. 1) these culturally modified trees (CMTs) have been documented from Oregon through Washington and British Columbia to Kodiak Island in Alaska (Mack and Hollenbeck 1985; Arcas Associates 1984; Mobley 1984, 1989; Mobley and Eldridge 1992; Lewis and Mobley 1994), although most samples have been described only in technical reports. Tree-ring dated CMTs in Washington and especially British Columbia have documented patterns of traditional Native forest use during the last 400 years (Eldridge and Eldridge 1986; Gottesfeld 1992; Stryd and Eldridge 1993; Mack 1996; Prince 2001). The oldest dated scar in southeast Alaska is from a Thuja bark-stripped in a.d. 1686 (Mobley 2002), but most studies in Alaska have analyzed only the surface appearance of CMTs. Studies of Native forest use using tree-ring dated CMTs in the American Southwest (Swetnam 1984; Kaye and Swetnam 1999) and Scandinavia (Niklasson et al. 1994; Zackrisson et al. 2000; Östlund et al. 2003; Ericsson et al. 2003; Bergman et al. 2004), have developed technical approaches applicable to samples from North America’s Northwest Coast, including Alaska.

Fig. 1
figure 1

Map of the northwest coast of North America

Full size image

The physiology of tree scarring and natural healing is similar from species to species throughout the Pacific Northwest. After a tree is stripped, the bark at the scar edge grows over the scar face. Each new ring curls over the last to meet the scar face, creating a lobe of tissue on either side of the scar. With time the two lobes can grow together and hide the scar. Healing lobes on a CMT usually lack branches or knots and thus offer uniform fiber, so they were sometimes harvested repeatedly by Natives to produce a tree with multiple scars dating many years apart (Fig. 2).

Fig. 2
figure 2

Cross-section of CMT no. 46, a Chamaecyparis 423 years old. More than 75% of the bark was removed in a.d. 1745; the subsequent healing lobe at the top of the specimen had regained 75% of the tree’s circumference by a.d. 1783, less than 40 years later, when the tree was again stripped. The later bark-stripper took all the accessible bark on the tree—almost the whole healing lobe—leaving only two little strips where the lobes folded over the a.d. 1745 scar. Since a.d. 1783 the two strips have grown into two lobes (seen at the bottom of the specimen), which have been healing both the a.d. 1783 scar and the original a.d. 1745 scar

Full size image

Most CMTs are shells of living bark and good wood hiding a rotten core, with convoluted rings, so the thin column extracted by a tree increment-borer is unsuitable for dating cultural scars (Eldridge and Eldridge 1986 p. 4). Instead, a chainsaw is used to cut pie-shaped wedges that intersect one corner of the scar inside the lobe of a standing tree (Hicks 1985 p. 107), or the tree is cut down to remove a whole slice from the trunk (Niklasson et al. 1994 p. 185). Whole-round samples allow better ring counting and produce more accurate dates, they indicate the natural age of the tree, and they are the only way to discover completely embedded scars.

This paper used tree-ring dating to investigate the CMTs at the Ship Island site (Site no. CRG-357) in southeast Alaska. The purpose of the study was to evaluate the nature and timing of local Native tree exploitation, and to explore a possible link with the legendary last battle between the Stikine Tlingit Indians of Alaska and the Tsimshian Indians of northern British Columbia. The Ship Island results are compared with the Meares Island and Newcastle Block samples reported from British Columbia (Stryd and Eldridge 1993) and the Gifford Pinchot National Forest sample from Washington State (Mack 1996) because they have large sample sizes and show different patterns of Native forest use through time.

Natural environment

Southeast Alaska is a fjord-fringed, island-choked strip along the north Pacific coast (Fig. 1), where annual rainfall of over 400 cm supports a temperate rainforest. The Ship Island site is on the mainland 50 km northwest of Ketchikan. There the exposed shoreline provides little shelter from the treacherous waters of Clarence Strait, and the almost tree-less Ship Island marks a small indentation or bight where small craft can be beached. The difficult shore access at the Ship Island site played a part in the oral history of the last battle between the two Native groups. It also inhibited commercial logging (thus leaving more large old trees to be included in the CMT sample) and created difficulties in obtaining the tree-ring samples. Around most of the bight is an arc of relatively level forest up to 300 m wide, bounded by a rock scarp marking a transition to a bench about 30 m in elevation that leads gradually up to steep mountains 1 km away. Isolated stumps mark sawn trees selectively cut in the early half of the twentieth century, amid mostly mature trees hundreds of years old.

History

Following visits by European and American sailing ships in the late 1700s, southeast Alaska Natives became involved in the commercial fur industry managed by Russian administrators during the first half of the nineteenth century. In 1867 the United States paid Russia for the rights to Alaska, and the Tlingit Indians of southeast Alaska gradually abandoned more traditional practices to participate in the cash economy brought about by the subsequent century of commercial fishing, logging and mining. During the twentieth century most of southeast Alaska was managed by the U.S. Forest Service.

Traditionally the area was the territory of the Cape Fox Tlingit (Goldschmidt and Haas 1998). Under the Alaska Native Claims Settlement Act of 1980, title to the land in the Ship Island vicinity was acquired by Sealaska Corporation—a regional Native corporation representing Tlingit, Haida and other Alaskan Natives living in southeast Alaska. In 1988 Mobley (1989 pp. 79–82) recorded burned and bark-stripped trees on the mainland at Ship Island.

Materials and methods

An expedition to obtain tree-ring samples from the Ship Island site was launched by helicopter in April of 1998 after attempts by boat and floatplane the year before failed due to adverse weather. In five days Charles M. Mobley and an experienced logger mapped 46 CMTs and cut samples from 23 (Fig. 2). CMTs to be tree-ring dated were selected on the basis of safety, since some examples were rotten, some were leaning, and some grew from slopes affording little footing for the logger. Nine round samples and 14 wedge samples were obtained from 23 trees. At the University of Alaska Museum in Fairbanks the 300 kg collection of wood samples was dried by Michael Lewis to enhance the visibility of the rings. Then, beginning at the bark (representing 1998), the rings were counted inward under illuminated magnification to reach the year in which the bark was stripped. On whole-round samples both lobes of each scar were counted independently for comparison.

Results

Morphology of CMTs

The 46 CMTs at the Ship Island site consisted of 41 Thuja, two Chamaecyparis, four Picea, and one unidentified charred stump. Despite the few examples, it is evident that Chamaecyparis was handled much like Thuja, while Picea was handled differently. Three of the four Picea CMTs had two bark-strips each, totaling seven scars at the small end of the length range (Fig. 3). The two Chamaecyparis CMTs were relatively small in diameter, but each had two triangular bark-strips ranging from 1.5 to 6 m long.

Fig. 3
figure 3

Graphs plotting scar length, and tree diameter at chest height, for all three species

Full size image

The 41 Thuja CMTs showed the most variability. The trees ranged from 0.35 to an estimated 2.30 m in diameter (Fig. 3). The scars ranged in length from 30 cm to over 7 m. Most bark-strips—including all the longer ones—were triangular scars formed by peeling upward along the natural tissue alignment to a point of convergence. In contrast, the seven rectangular scars—formed when a tool was used not only to score the bottom of the removal but the top as well (Fig. 4)—displayed a narrower range in length of 85–185 cm. Some scars were so hidden by their healing lobes as to show only a crease in the live bark. Of seven Thuja that were burned, two were bark-stripped prior to burning, and three were hacked in search of tinder for fire lighting. Hack-marks were also recorded on three bark-stripped CMTs, and five CMTs showed only hack marks, making in total eleven hacked CMTs. An axe was used to bark-strip or hack some trees; blade mark widths of 4 and 5 cm on other trees indicate use of an adze or chisel.

Fig. 4
figure 4

Logger Jay Sipe inspects CMT no. 1, a 427-year old Thuja with a rectangular scar bark-stripped in a.d. 1779

Full size image

Dating the CMTs

Twenty-three of the 46 recorded CMTs were dated using tree-rings (Fig. 5). The 23 trees showed 29 scars in the field and revealed seven more hidden scars, resulting in 36 dated scars. The two lobes of a single scar should yield the same bark-stripping date, though ring deformation and compaction due to stress—from differential light and slope as well as repeated cultural stripping—may suppress the visibility of rings from lobe to lobe. In three of the nine paired counts the same date was obtained, but for the other six pairs it was not, and recounting did not resolve the differences. The six conflicting pairs had count discrepancies of one, two, three, four, five, and ten years, respectively (Table 1), lessening the precision implied by the remaining single-lobe dates.

Fig. 5
figure 5

Date distribution for the Ship Island site compared with three Gifford Pinchot samples (Mack 1996 p. 28) and the Meares Island and Newcastle Block samples (Stryd and Eldridge 1993 pp. 217–219). To include in the Ship Island sample examples for which the same scar yielded two different dates, the average of the 2 years was plotted

Full size image
Table 1 Tree-ring dates from the Ship Island CMT sample
Full size table

The natural age of eight trees was determined (Table 1). Seven Thuja trees ranged from 329 to 470 years old, while one Chamaecyparis was 423 years old. The diameters of the trees do not correspond to their relative ages, with several old trees having diameter-at-breast-height estimates of only 60 cm, and one of 35 cm.

The bark-strip dates in the Ship Island sample span a range of almost 200 years, from 1718 to 1912 (Fig. 5). Most of the bark-stripping took place between 1850 and 1900; only three scars date from 1800 to 1850. A smaller mode in the distribution appears earlier, with six scars from the 8 years between 1779 and 1787.

Distribution of CMTs

The distribution of CMTs by species corresponded to the natural distribution of the trees. Clusters of CMTs were mapped at the site, but these did not correlate with specific date ranges or other variables. During the early twentieth century selective logging with hand saws for such uses as cannery docks and floating fish traps focused on Picea near the shoreline, perhaps resulting in fewer Picea CMTs today. An area of about 80 hectares was surveyed to record the 46 CMTs, yielding an average CMT density of 0.58 per hectare. The field survey was sufficient to suggest that CMT density decreased inland and along the shoreline to the south, but not necessarily along the shoreline to the north.

Discussion

Access to the Ship Island site requires landing at a small beach in the bight. Arcing around the bight is a low bench offering good camping, with dips inland where game-trails naturally lead the bark-stripper up small valleys and ridges, and it is in this zone of efficient exploitation that most CMTs were located. The bark of Thuja was most used, with lesser use of Chamaecyparis and Picea bark. The bight was used intermittently—in at least 29 different years—during that 194 year period. Most of the trees were bark-stripped in the 41 years between a.d. 1852 and 1893. A lesser burst of activity occurred earlier during the 9 years between a.d. 1779 and 1787, with eight scars representing five different years. Those scars showing tool marks suggested use of an axe, except for three examples with adze or chisel marks. At least some charred trees are probably of cultural origin and all were recorded, but—even given the rain forest environment—a natural fire cannot be ruled out. No aboriginally logged trees—a CMT type common in British Columbia (Arcas Associates 1984 p. 34) and less so in Alaska (Mobley 2002 pp. 19–22) and reflecting traditional Native exploitation of wood rather than bark—were noted at the Ship Island site.

The rectangular scars at the Ship Island site have a relatively small and narrow length range of 0.85–1.85 m, which fits well the mode for scar length in the Gifford Pinchot samples (Mack 1996 p. 7), and are the right size for handy containers and other utensils. The purposes to which the longer triangular bark pieces were put may have been similar, but their greater length expanded the range of applications to include things like cordage and shelter roofs.

Bark-strippers at Ship Island used small-diameter trees, a tendency also noted by researchers in British Columbia. One tree first bark-stripped in 1833 when only 20 cm in diameter had grown to 90 cm in diameter and consisted of 90% healing lobes around its circumference.

The paired count discrepancies in the Thuja CMTs may be the result of environmental stresses on a species living near the northern extent of its range (all six pairs of conflicting dates are from the recent end of the date range). If such ring inconsistencies are a regional phenomenon, cross-ring dating using particular marker rings may be necessary to date scars to the exact year (Niklasson et al. 1994 p. 186); and studies focusing on general date distribution patterns may be more informative than investigations seeking exact year dates. Over a third of the bark-strip dates from the Meares Island sample had an error margin of more than 6 years, leading Stryd and Eldridge (1993 p. 217) to simply exclude them from the plotted distribution (paired Ship Island dates with a discrepancy were averaged for plotting in Fig. 5). Mack (1996 p. 17) independently compared counts from a wedge sample and a round sample taken later across the same scar on a Thuja and found a discrepancy of 5 years.

The Ship Island date distribution of 1718–1912 shows both similarities and differences with selected samples from British Columbia and Washington. Mack’s (1996) Gifford Pinchot sample of 183 dated CMTs in Washington was judged on the basis of scar length and ethnographic documentation to reflect Native manufacture of bark baskets for collecting huckleberries, and different date distributions for three separate areas are interpreted as the result of changing huckleberry availability and exploitation patterns (Fig. 5). All three Gifford Pinchot sub-samples bracket a shorter period of time than the Ship Island sample, and two—Areas 1 and 2—continue long past 1912 when bark-stripping at Ship Island stopped. The Ship Island date distribution also differs from that of Newcastle Block in British Columbia (Stryd and Eldridge 1993 p. 219), which has a narrower range of dates and no dates after 1850. The Ship Island sample is most like the Meares Island sample, although the latter includes both older and younger CMTs (Fig. 5). Stryd and Eldridge (1993 p. 216) interpret the fewer dates prior to 1800 in the Meares Island sample as “almost certainly the result of decreasing preservation with increasing age,” which also explains the few Ship Island trees bark-stripped before 1770. The Ship Island and Meares Island samples each have bimodal distributions with the major mode between 1850 and 1900, but Ship Island experienced a modest burst of bark-stripping around 1780 whereas Meares Island received one between 1790 and 1830. The last bark-stripping at Ship Island was done over 95 years ago, while Natives continued exploiting the tree bark of Meares Island for traditional uses well into the twentieth century (Stryd and Eldridge 1993 p. 217).

Conclusions

Of note are the 15 CMTs in the total sample of 46 that had multiple scars—as many as five per tree—indicating repeated stripping of individual trees at Ship Island. This intensity of Thuja bark harvesting observed at Ship Island may reflect the sustained use of a relatively valuable product near the northern limit of its natural range, such that the difficult access was not enough to deter local Native exploitation. Alternatively, the repeated bark-stripping of individual trees in the sample may be the result of recurrent forced encampments by Native parties seeking shelter from the stormy Clarence Strait. As few as ten or as many as 134 years elapsed between bark harvests on the same tree, with the average being 55 years between stripping.

Smithsonian anthropologist Waterman (1922 p. 51) recorded a local Tlingit story about the last battle between the Stikine Tlingit and the Tsimshian of northern British Columbia, in which afterwards ten large dugout canoes of Tsimshian women paddled north toward the battle site, waited out a long storm at Ship Island, then returned home with loads of cedar bark collected there. The battle was not noted by Russian officials, so the event must be earlier than 1800. Given the potential ring-count discrepancy of up to ten years in the method used in this study, the small burst of bark-stripping between 1779 and 1787 could reflect the Tsimshian women’s camp event, rather than be representative of a larger regional pattern.

The Ship Island sample is the first large group of CMTs to be tree-ring dated in southeast Alaska, so it is not yet clear how representative the pattern may be of the region. The difficult access of the shore has been consistent through time, so the bimodal intensity of bark-stripping at the Ship Island site may match regional fluctuations in Native forest use—heavier tree bark exploitation around 1780 and again between 1850 and 1900. The Ship Island study thus forms a basis for comparing subsequently dated CMT samples in Alaska.