Nain

Datasets for this study were drawn from sampling sites on the Labrador mainland near the Nain town site and in two adjacent large bays_ Nain 1, Nain 2, KAM and WB (Figures 1 and 2).

Nain 1 and Nain 2 are located in two parallel valleys bordering the community (Figure 2(a)); in this study, these two adjacent sites are combined into a single dataset (referred to henceforward as Nain). In 1771, a permanent settlement was established at Nain by Moravian missionaries and the location has seen permanent, year-round settlement since that date (Taylor 1974; Hiller 1977). The town site is located on a travel route used by Inuit during seasonal migrations between the coastal and interior caribou hunting grounds, which lead to the establishment of a number of other seasonal settlements in the area (Taylor 1974). Other similar sites used by diverse prehistoric populations extending back to Maritime Archaic populations of ca 7000–5000 BP have also been documented (Hood 2008).

The KAM site is located approximately 30 km south of Nain and approximately 1 km south of Kammasûk, a small abandoned settlement comprising three houses and a small Hudson's Bay Company trading post. Kammasûk is located on the sheltered side of a cape separating Voisey's Bay (Tasiujatsuak) and Anaktalak Bay, two ecologically rich fjord systems fed by substantial rivers draining long valleys extending deep into the Quebec-Labrador plateau (Figure 2(b)).

The Webb's Bay site, referred to here as WB, is a small settlement in the inner reaches of a large bay approximately 35 km north of Nain. It is located on the sandy banks and delta of a brook draining a valley extending into the highlands of the continental interior. This valley comprises a mix of spruce forest, lichen heath, lakes and peat bogs (Figure 2(c)).

South Aulatsivik Island

South Aulatsivik Island is located about 10 km from Nain; it is the largest island in the archipelago (456 km² in area) and on the Labrador Coast. The sampled site consisted of an open spruce forest colonising drained soils at about 25 m in altitude (Figure 2(d)). A short river about 3 km long originates in a small lake and flows east into the sea. The nearest archaeological site is located at about 2 km south of the sampling site on a low raised beach terrace (3 m asl). This site, named South Aulatsivik 6 (HdCi-20), includes the preserved ruins of at least eight house structures. These include semi-subterranean sod houses reflecting the architecture typical of winter iglosoak houses of the 18th to mid-19th centuries, with dugout floors, sleeping platforms and entrance tunnels, as well as hybrid house structures combining turf and wood plank constructions that are more typical of the late 19th to early 20th century. A dendrochronological date of 1875 (Woollett et al. 2019) was obtained for a post associated with the excavation of one of the latter houses, a date confirmed by the artifact assemblage which includes late 19th century industrially produced items and a cast iron stove. Two other significant Inuit settlements sites, dated to the 18th and 18th to 20th centuries, respectively, are located in protected bays 6 and 12 km to the north of the South Aulatsivik sampling site.

Dog Island

Dog Island is located about 35 km northeast of Nain. It is a large island characterised by high summits and deep-protected bays. The area consisting of Dog Island and its surrounding islands is one of the most completely surveyed archaeological landscapes in Nunatsiavut, comprising over one hundred discrete archaeological sites representing every major phase of Labrador culture history extending back to ca 7000 BP (Hood 2008; Woollett 2010). Tree samples were taken from the slope of Mount Alagaiai (see Roy et al. 2017) that shelters the bay and delimits the northern shore of Oakes Bay and the Oakes Bay 1 archaeological site (HeCg-08) (Figure 2(e,f)). The slope from about 8 to 180 m asl is colonized by an open forest of white spruce and some mesic taxa such as Betula glandulosa, Rhododendron groenlandicum.

Collection of demographic data

A reconstruction of the structure and scale of Inuit and Settler settlement in the study region was based on published historical sources derived from Moravian Mission records, Royal Navy surveys and other historical sources (Jenness 1965; Brice-Bennett 1977; Kaplan 1983). The settlement history of the Kammasûk site was documented through interviews and oral histories published in a local history magazine (McGrath 1997) while the history of the Webb's Bay site was reconstructed through interviews (Lemus-Lauzon et al. 2012). The settlement histories of the South Aulatsivik 6 and Dog Island sites were documented through historical archives, ethnohistorical sources and archaeological evidence (Taylor 1974; Kaplan 1983; Woollett 2003; Woollett et al. 2019).

For the goals of this study, the scale of settlement at specific sites was extrapolated from various sources of historical and archaeological documentation, using numbers of households as a proxy. The settlement history of the Kammasûk site was documented through interviews (see Lemus-Lauzon et al. 2012) and oral histories published in a local history magazine (McGrath 1997) while the history of the Webb's Bay site was reconstructed through interviews (Lemus-Lauzon et al. 2012). Historical descriptions of these sites and of their inhabitants permit the attribution of dates and numbers of contemporaneously occupied houses which can be confirmed by archaeological inventories of standing houses and house ruins.

The Nain town site is a very complex palimpsest where modern occupations cut into and mask traces of previous settlement. Census data compiled by Moravian missionaries after 1774, Royal Navy surveys and various other colonial agents, for Nain and for some other sites described in Moravian records, were used in this study (Jenness 1965; Taylor 1974; Brice-Bennett 1977; Kaplan 1983). Absolute population sizes are recorded in these records but numbers of households are not consistently reported. For this reason, the number of households present at Nain was extrapolated, presuming an average of 20 persons per household. This estimation was based on data from 18th and early 19th century (see Taylor 1974). Plank houses of the 19th century were smaller than their 18th century predecessors and so this study probably underestimates the total number of occupied houses in Nain in the 19th and 20th centuries and hence the number of households. Finally, some archaeological sites (notably Oakes Bay 1 and South Aulatsivik 6) have little or no historical documentation. At these sites, the number of households is extrapolated by number of discrete inventoried house ruins whose occupation could be attributed to general (25 years) chronological intervals.

Collection of field data for dendroecological analysis

Fieldwork at the study sites on the mainland around Nain involved mapping, stump sampling and tree coring conducted in the summers of 2010 and 2011. From each living tree selected for sampling at each site, two cores were extracted perpendicular to the trunk using a Pressler increment border. Cores were inserted at about 25 cm from the root collar of living trees while cross-sections from dead trees were taken at the root collar, using a saw. Sample transects established at each site were oriented perpendicular to their valleys (except for the WB site), starting at the altitudinal tree line on one side of the valley and crossing the valley to the tree line on the opposite slope. Each transect had a width of 5 m. The Nain1 and Nain2 transects were oriented east–west, whereas the KAM transect was oriented northeast–southwest. Because of its geomorphological setting on a wide and flat sandy terrace, the sampling strategy of the WB site included three small transects, two of which were oriented east–west and the other north–south. At these three sites, transect sampling followed a systematic scheme in which every fifth living tree having a diameter ≥ 5 cm was sampled. All trees within a transect that had a diameter ≥ 20 cm were also sampled. As well, each cut stump observed along these transects was noted and described (i.e., position, height, maximum diameter, and presence or absence of bark) and sampled. A total of 308 tree samples were recovered including 242 core samples of living trees and 66 deadwood sections.

At South Aulatsivik Island, no formal transects were used because trees were distributed in clusters of clones, each one of which comprised an isolated tree stand. The density of trees in each cluster varied between 0.23 and 1.5 per hectare. The sampling area was a pair of linear terraces located on either side of a stream draining a wide valley on the island's southeast side. The samples were collected in an initial study undertaken in the summers of 2018 and 2019 and include both cores extracted from the trunks of live trees and cross-sections sawn from cut stumps. In total 19 samples were recovered including 13 core samples of living trees and six cut stumps.

At Dog Island, sampling was carried out in the summer of 2010 near Oakes Bay, using twenty 10 m-wide transects spaced every 100 m along a linear distance of 2 km (Figure 1). Each individual tree and krummholz stand observed along each 10 m-wide transect was measured and recorded, but only those having a diameter greater than 15 cm were sampled. Each dead tree with a stump that was not buried or decomposed was sampled. A total of 133 samples were recovered, including 41 core samples of living trees and 92 dead-wood sections. In addition, four pieces of spruce wood exposed during the excavation of House 2 at the Oakes Bay 1 archaeological site were collected for analysis.

Laboratory analyses

Tree growth patterns of the five study sites were analysed in order to detect growth release events. Lemus-Lauzon et al. (2018) published an initial analysis of data from the Nain, KAM and WB sites, these data are re-examined in the present study. Similarly, data used in the course of a similar study regarding Dog Island, published by Roy et al. (2017) have been re-examined and re-analysed here. The South Aulatsivik data represent a summary of the preliminary results of an ongoing research project.

Tree-ring widths were measured with a Velmex micrometer (precision of 0.002 mm) under a binocular microscope at 40x magnification. Diagnostic light rings, that is, growth rings with exceptionally few latewood cells, were used to date and cross-date living- and dead-wood sections following protocols of Filion et al. (1986). Archived light-ring chronologies derived from the L2 tree-ring sequence from Okak Bay (1680–1922, Payette 2007) and from Oakes Bay on Dog Island (1620–1978, compiled by Natasha Roy), which are available at the Dendrochronology Laboratory at the Centre d'études nordiques and Faculté de foresterie, de géographie et géomatique, Université Laval, were also used as validation tools. Cross-dating was also verified using the program COFECHA (Holmes 1983; Grissino-Mayer 2001).

Wood harvesting in forest stands improves growth conditions for neighboring trees. Forest openings reduce competition for resources such as light and water and can substantially modify local growth conditions. Improved growing conditions are reflected by a sudden increase in radial growth of surviving trees, which is maintained for several years. In closed forests, the increase in radial growth is normally recorded by the majority of trees surrounding an opening created by fallen or cut trees (Payette 2010). In open forests, it is also possible to use sharp growth increases of surviving trees to date tree harvesting (Roy et al. 2017; Lemus-Lauzon et al. 2018). However, only trees located in the immediate environment of the harvested tree will record significant growth release.

To distinguish growth changes due to factors other than disturbance events, we used the boundary line approach (see Black and Abrams 2003, 2004; Black et al. 2009), a criterion that responds to the suddenness and the duration of the growth changes unique to canopy releases (Lorimer and Frelich 1989; Nowacki and Abrams 1997). This method scales each release by its species maximum physiological potential, as defined by a normalized growth rate. The radial growth increase over a given year was calculated as follows: (M2-M1)/M1, where M1 is the average growth over the 10-year period prior to the given year and M2 is the average growth over the subsequent 10-year period (Nowacki and Abrams 1997). The relationship between prior growth and release potential is specific to species. For this reason, we calculated boundary lines using data from documented 64,851 white spruce growth rings from the Labrador area (ring width data from Payette 2007). The white spruce boundary line corresponds to an exponential curve (968.22^–2.393x) calculated from maximum increase recorded for each growth class (M1). Growth releases that exceed 20% of the boundary line and lasted a minimum of 10 years are reported. We used this method to identify and date growth releases on living trees and on dead trees.

Although growth releases persist over many years, only the first year of a release event was considered in this analysis. Also, when there was at least a two-year gap between two growth releases, those were considered as distinct events. Afterwards, the total growth releases of each tree were arranged in 10-year classes. Because the frequency of growth releases decreases backward in time due to the decreasing number of trees included in the analysis, we fitted a log-linear regression to remove the exponential trend in the number of detected growth releases (Morneau and Payette 2000).

The Labrador coast is strongly influenced by natural climate forcing such as the North Atlantic Oscillation (NAO) and the El Niño–Southern Oscillation (ENSO) cycle. In order to distinguish disturbance events (e.g., fires, insect outbreaks) from climate forcing, the 10-years mean window used here to calculate growth release is long enough to smooth the periodicity of the ENSO cycle, which is 3.5 years, and short enough to avoid the NAO signal, which is 40 to 60 years (D'Arrigo et al. 2003; Berg et al. 2006).

Demographic data and household evolution

In Figure 3, population levels are represented in terms of number of households and, at archaeological sites, are extrapolated from number of house ruins dated to approximate time intervals. Oral histories give unambiguous estimates of number of households at certain sites. Censuses realized at intervals in the Nain region recorded numbers of resident individuals in the town and in certain outlying settlements (Curtis 1774, cited in Jenness 1965; Taylor 1974; Kaplan 1983). These population numbers were converted into number of hypothetical households assuming an average of 20 persons per household (average household size between 1783 and 1900) (Figure 3).

Nain

Inuit families settled in Nain in only modest numbers prior to about 1820, after which date a wave of religious conversions and more active participation in the trading post economy run by the Moravian missionaries encouraged an increase in Inuit settlement there (Taylor 1974; Brice-Bennett 1977, 1990). A set of episodic Missionary and military surveys and censuses provide demographic information for Nain. In the late 18th century, the village's population numbered 63 people, 168 by about 1820, 298 in 1840, 314 in 1850, 277 in 1850, 282 in 1876, and 263 in 1890 (Jenness 1965; Kaplan 1983). This corresponds to an increase in extrapolated household numbers from 5 to 15 from the late 18th century to the late 19th century. Then, population raised sharply during the last century; Nain is now a community of 1200 residents, of whom 90% identify as Inuit (Statistics Canada 2017).

Kammasûk

The site Kammasûk was settled by Amos Voisey, an English fisherman and trapper, in the mid-19th century (approximately 1845; McGrath 1997). By the early 20th century, Kammasûk comprised three houses belonging to three distinct households (Figure 3). A simple saw mill was set up in nearby Anaktalak Bay in the late 19th century and a modest wood-cutting operation including saw pits was also installed at Kammasûk in the early 20th century (McGrath 1997). These families had largely relocated to Nain by the 1960s. Presently, one house remains standing at the site.

Webb's Bay

The location was settled in 1859 by James Webb, an English settler, fisherman and trapper. His descendants maintained a permanent, year-round habitation there, incorporating a single household, into the mid-20th century (Henry Webb, Personal communication 2020) (Figure 3). Subsequent generations of the family have built seasonal houses at Webb's Bay since the mid-20th century, although the site has not seen permanent, year-round habitation since the 1960s. A small historic period Inuit winter settlement site and a series of prehistoric sites are located on islands and points at the entrance to Webb's Bay, approximately 10 km east of the sampling site.

South Aulatsivik Island

This large island, located at the interface of continental and marine environmental zones north of Nain is characterised by varied topography. It currently hosts dense and tall spruce forests and large lakes on its western portions, with thin and forested tundra including open spruce woodlands and some krummholz individuals, peat bogs and lichen barrens on its eastern side. The island has seen ongoing human occupation since its initial settlement by Maritime Archaic people about 7000 BP (Fitzhugh 1978; Hood 2008). Sites of each major culture-history phase are identified on the island, clustering locations at ecological interfaces such as fluvial valleys connecting the interior to the sea, and capes extending into marine hunting areas.

Figure 3.

Estimation of Inuit settlement size at the five study sites based on archaeological, oral history and textual historical sources. Population levels are represented in terms of number of households and are extrapolated from historical records and number of archaeological house ruins dated to approximate time intervals.

The ruins found around the sampled site are consistent with late 18th–19th century Inuit winter house architecture and are not noted in Moravian censuses between 1773 and 1820; for this study the site is assumed to date to the mid-19th century and to represent the maintained occupation of a single household. Three isolated settlement sites consisting of single house structures or ruins are documented in coves between Long Pond Brook and South Aulatsivik 6 (Figure 1). All of these date between the late 19th and late 20th centuries. Three other isolated settlement sites are documented on the southwest shoreline of South Aulatsivik Island, all consist of single structures constructed in the early to mid-20th century.

Structure of forest stands in the Nain Archipelago

Tree-ring width analysis shows contrasting growth patterns of inland and island forest stands. Mean ring-widths for 50-year intervals of mainland forests show a higher growth rate during the 20th century than the island forests (Table 1). For the period of 1850–1900, all sites show a mean ring-width distribution lower than that of the 20th century, with the exception of Kammasûk, which is located further to the south and in an area well protected from strong northeastern winds.

The overall forest composition of the island sites contrasts with the continental sites in terms of age and composition of forest stands. The Nain, Kammasûk and Webb's Bay sites all have young age profiles with very few trees older than 100 years old of age. They are also composed of a mix of white and black spruce and larch. This contrasts with the island forests of South Aulatsivik and Dog Island, where old growth trees (older than 100 years) represent 63% and 81% of samples, respectively (Table 2). The South Aulatsivik and the Dog Island forest stands are almost entirely composed of white spruce trees. During fieldwork, no larch trees have been observed at South Aulatsivik while only three young larch were noted at Dog Island. Finally, an important observation, cut stumps were plentiful at each study site and no dead wood was observed lying on the ground.

Table 1.

Mean ring width (mm) of Picea sp. for 50-year time periods at Nain, Webb's Bay, Kammasûk, South Aulatsivik and Dog Island.

Growth release analysis

Forest-disturbance events and human occupation before 1900

Isolated growth release events were identified prior to the late 19th century in all study areas. The residuals of many of these release events are not significant, nevertheless they point to the possibility of ongoing human impacts on patches of spruce forest extending back as far as the 17th century, well prior to the use of cast iron stoves or construction of buildings and infrastructures that demanded great quantities of wood. The appearance of release events coincides with human occupation in nearly all of the sample areas except Nain, where local wood harvesting is the most extensive.

At the Nain site, the scarcity of growth release events before 1900 could be related to the young age of the forest, which contains few if any trees old enough to document forest dynamics prior to the mid-19th century. Since the establishment of a permanent Moravian mission at Nain in 1771, the town has maintained a nucleus of Inuit and Euro-Canadian inhabitants (Taylor 1974). In addition, some English settlers established themselves in the peripheries of the Nain region during the 19th century, and they adopted elements of traditional Inuit land use. All these inhabitants harvested wood from the forest around Nain for their domestic needs (fuel for heating) as well as for their episodic raw material needs (e.g., building houses). In interviews, elders from Nain related that standing dead wood was specifically sought out as firewood because of its ease of combustion, leading Lemus-Lauzon et al. (2012) to conclude that the lack of dead wood was the result of intensive harvesting.

A second factor that could explain the absence of old trees and dead wood is the high decomposition rate of wood caused by significant soil humidity. A much different pattern was observed at Dog Island, where living old trees were common and dead trees were frequently preserved (Table 2). This study area is much more exposed to cold currents and generally receives less snow cover, which tends to favor better preservation of wood. In addition, its isolation as an island environment well separated from the mainland protects its forests from fires and insects, a condition demonstrated elsewhere by Payette (1988) and Zasada (1984).

Occasional disturbance events were identified at the South Aulatsivik site between 1650 and 1710, a period where there is, as yet, no archaeological evidence of local habitation (Figure 5). The cause(s) of these disturbances are hard to explain, however the sampling area offers easy access to wood for people living in and traveling amongst the surrounding islands, especially when travelling on winter fast ice with dog teams. Williamson (1997) noted that contemporary winter travel routes pass north-south along the eastern side of South Aulatsivik Island, directly in front of this sampling site and in direct sight of it. A dozen significant winter settlements dated to the 18th and 19th centuries are documented in the islands surrounding the South Aulatsivik site (Kaplan 1983) many of which currently have no local available wood; this small forest therefore represents a potentially significant source of wood for the region.

Figure 4.

For each of the five study sites: (a) number of disturbance events (left, bars) and number of trees (right, line) per decade; (b) residuals of the log-linear regression of the age frequency distributions of release events.

No growth disturbances were observed at South Aulatsivik between 1790 and 1880, despite the presence of dwellings dated to the mid to late 19th century (Figure 5). The lack of concordance between growth release and occupation records may be due to the very small sample size of this preliminary data collection (19 samples). Nevertheless, the beginning of observed disturbance at South Aulatsivik corresponds to a period of more constant occupation of the adjacent settlement site between 1870 and 1900, as inferred by archaeological research to date (Woollett et al. 2019). Ongoing fieldwork in this region will assure access to a more reliable data set.

The Dog Island site shows a continuous series of moderate disturbance events since about 1550 (Figure 5). The sampling site is adjacent to the Oakes Bay 1 archaeological site which had episodes of occupation from the 16th to late 18th century (Woollett 2010; Roy et al. 2012; Foury 2016). Additional similar archaeological sites are located on nearby islands, such as September Harbour 2 (17th century), Koliktalik 6 (1780s) and Evilik Bay 2 (mid 19th to 20th centuries). Many other cabins were inhabited throughout these islands during the late 19th and 20th centuries (Brice-Bennett 1977; Kaplan 1983; Williamson 1997).

The period of most intense occupation of the Oakes Bay 1 (17th century to 1772) does not coincide with statistically significant occurrences of growth releases in the local forest as demonstrated by residuals. Several growth occurrences were observed scattered throughout this time period, suggesting a minor level of ongoing harvesting, but their impact is not statistically demonstrated. Given the scale of the archaeologically documented occupation of the site, this minor level of forest impacts suggests that wood harvesting prior to the 19th century was limited. Quantities of wood, including spruce trunks of >20 cm diameter were used in the construction of the site's dwellings, as was well demonstrated in the archaeological excavations (Woollett 2010).

Wood appears to have been of relatively minor importance as a fuel during this time period, however. Seal oil, burned in steatite lamps, was the principal fuel for lighting and heating poorly-ventilated winter houses and for cooking into the 19th century (Taylor 1974). Wood was used principally as fuel for well-ventilated fires outside houses. Archaeological excavations at Oakes Bay 1 provided evidence consistent with these habits. Floor deposits from house interiors at the site included numerous fragments of steatite vessels and lamps with burnt fat residues as well as discarded masses of charred fat resulting from the cleaning of lamps (Woollett 2010; Couture et al. 2016; Foury 2016). Thus, a possible hypothesis explaining the relative lack of forest impacts prior to the 19th century is that ‘traditional’ Inuit modes of land use did not require intensive wood harvesting except during periods of house construction. A second important possibility is that the relatively poor preservation of trees of this age hinders the evaluation of forest disturbances prior to about 1800.

A period of multiple significant growth release events was noted at Dog Island between 1790 and 1870. Historical and archaeological data suggests that Oakes Bay 1 was no longer occupied as a winter settlement during this time period. However, other winter settlements with traditional sod house dwellings came in use in the region, Koliktalik 6 by 1780 and Evilik Bay 2 in the mid-19th century. It is probably that these disturbances are related to harvesting of wood for the construction of dwellings at these sites, they lack local sources of wood and are only 3 to 5 km from the sampling area (Figure 1), close enough to transport wood from the Dog Island forest via dog sled.

Forest-disturbance events and human occupation after 1900

A period of widespread, constant and significant growth release events was recorded at all sampling sites between 1890 and the present day. These show a clear correspondence with an increased population until ca 1950, as extrapolated by the number of households in the Nain archipelago. During this period, wood was used universally in the construction of plank houses and hybrid sod and wood houses (Figure 6). These structures require significant amounts of wood for construction and consumption of much more fuel (wood) as they were heated with cast iron wood stoves (Kleivan 1966; Brice-Bennett 1977; Kaplan 1983). Both Nain and the outlying settlements around it were permanent sites used in multiple seasons if not year-round and over multiple years. This pattern of settlement would inevitably have produced a constant demand for wood and a significant and maintained level of local wood harvesting.

Permanent settlements were established in localities around of all of the study's sampling sites during this time period and they all saw their maximum historical populations in the 20th century (Figure 5). The smaller study sites (DI, SA, KAM, WB) all had populations resident within 0 to 2 km from the study sites with additional small isolated settlements in a radius of 10 km. As a general trend, people living in outlying settlements relocated to Nain between 1950 and 1970, a process which added significantly to the population of Nain and which increased local wood harvesting. According to interviews performed with the residents of Nain (Lemus-Lauzon et al. 2012), wood harvesting conducted throughout the locality, on the mainland and on the adjacent islands, is still a major ongoing activity during the winter and spring months.

Figure 5.

Correspondence between growth release events (left, bars) and the scale of local human settlement in the five study sites in terms of extrapolated number of households (right, line) for the last 500 years. The shaded area shows regional forest disturbance between 1900 and 1950. Note that the scale for number of households (right) is not the same for all studied sites. Dotted lines indicate periods when the extrapolated number of households is based exclusively on the number of documented house ruins at archaeological sites, in absence of detailed historical data.