THIS STUDY TESTS Grant’s hypothesis (1994) that northwestern Newfoundland experienced a 10-metre sea-level oscillation during the late Holocene,[1] against previous reconstructions that propose continuous emergence. This study also comprises part of a larger regional project that is reconstructing spatial and temporal patterns of postglacial sea-level history for Newfoundland and Labrador (e.g., Bell et al. 2001 this volume) and assessing its archaeological implications (e.g., Bell and Renouf 2003; Smith et al. 2003). The importance of palaeoshoreline position for understanding prehistoric coastal settlement is well established in areas of dynamic sea-level change (e.g., Labrador, Clark and Fitzhugh 1992; Newfoundland, Bell and Renouf 2003; Gulf of Maine, Sanger and Belknap 1987, Sanger and Kellogg 1989; British Columbia, Josenhans et al. 1997). Studies show that reconstruction of palaeoshoreline elevation and configuration is an essential tool for explaining the location of known archaeological sites and for developing search strategies for finding others (Bell and Renouf 2003; Renouf and Bell in press).

To our knowledge, this is the first study in Newfoundland and Labrador that employs diatoms and lake basin isolation history to address questions related to sea-level change. Similar studies, however, have been widely used elsewhere in the world, including Scandinavia (Corner et al. 1999; Eronen et al. 2001), Greenland (Long et al. 1999), and western Canada (Hutchinson et al. 2004). The methodology involves identifying litho- and bio-stratigraphic indicators that accord with changes between marine and lacustrine sediments, in basins that have been isolated (or transgressed) by changing sea levels. This study uses diatoms, forms of aquatic algae that are particularly sensitive to changes in salinity. Once marine-lacustrine boundaries have been identified, they are dated using the radiocarbon method. This allows researchers to plot the elevation of outlets of various basins (thresholds for isolation/transgression) against time, and thus reconstruct relative sea-level fluctuations.

Our current understanding of RSL history for the Port au Choix region (Figure1) is presented in a companion paper in this volume (Bell et al. this volume), while the spatial pattern of sea-level change along the west coast of the Great Northern Peninsula is under review (Bell unpublished). These studies propose that Port au Choix is located near the transition between a continuously falling postglacial sea-level trend (type-A, Quinlan and Beaumont 1981) and one in which the postglacial sea level initially falls below its present level and then rises again (type-B, Quinlan and Beaumont 1981; Figure 2). Grant (1994) proposed a third possible sea-level history for the region, a modification of the type-B model, in which postglacial sea level falls below its present level at approximately 9000 cal BP,[2] followed by a brief oscillation above present (reaching 3-4 m above sea level [asl]) between 3700 and 2000 cal BP, and a transgression to present sea level during the last several thousand years (Figure 2).

This theory was based on the occurrence of a low, horizontal relic sea cliff, which lies just above present tide level along the west coast of the Great Northern Peninsula (Figure 3; Grant 1994). The cliff is an erosional feature and so was not dated directly, but its age is apparently bracketed stratigraphically by 4500-year-old beaches at Port au Choix that it cuts into, and by 900-year-old beaches farther north that overlap it. Similar oscillations in sea level along the northeastern seaboard of North America have been attributed to the migration of a collapsing glacioisostatic forebulge (Barnhardt et al. 1995), eustatic changes related to mid-Holocene warming (Scott and Collins 1986) or, as Mörner (1976, 1992) has argued, to changes in the Earth’s geoid or rate of rotation.

The implications of Grant’s (1994) sea-level oscillation for the Port au Choix regional archaeological record were intriguing (Bell and Renouf 1998). Prior to recent finds, the only Maritime Archaic Indian (MAI) site at Port au Choix was a cemetery, radiocarbon dated between 4690 and 3620 cal BP (Tuck 1976) at 6-10 metres above sea level. The search for a habitation site associated with the cemetery was the focus of many unsuccessful surveys in the 1960s, 1980s, and 1990s. These site surveys were predicated on two assumptions. First, that camps or longer-term settlements would likely have been situated near the contemporaneous shoreline, and therefore at an elevation of 6-8 metres above present sea level, similar to the cemetery; and second, that the relative sea-level history of Port au Choix followed a simple type-A model. The proposition of another possible sea-level history for the region presented alternative search strategies for locating the MAI habitation site. For example, Grant’s (1994) modified type-B sea-level curve suggests that the Groswater and Dorset Palaeoeskimo sites would have been located 9-10 metres above the shoreline at the time of occupation, whereas the MAI cemetery, at 14-15 metres above sea level, would have been even farther removed from the active beach during its time of use (Figure 2). In other words, the cemetery would have been on higher ground than the Palaeoeskimo settlements. If MAI and Palaeoeskimo settlements were similarly situated from the contemporaneous shoreline, for similar practical considerations such as view, shelter and access, then MAI settlements might have been closer to the active beach than the cemetery. If MAI settlements were situated within 10 metres of the contemporary shore, as were the Palaeoeskimo sites, these MAI sites today would be at, or below, present sea level. Therefore the need for an accurate sea-level history to develop a search strategy for MAI habitation sites highlighted the importance of testing Grant’s proposed sea-level oscillation. Previous studies elsewhere had demonstrated the potential of the lake isolation method in resolving sea-level fluctuations of this magnitude, and so we adopted the approach as one part of an integrated geo-archaeological initiative in the Port au Choix region (Renouf and Bell 2002).

Diatoms are a diverse group of microscopic (5-500 µ#xB5;m), mainly photosynthetic algae (class Bacillariophyceae) that are found almost universally, wherever moisture is present (i.e., lakes, rivers, oceans, hot springs, soil, etc.). The distribution of individual diatom species is controlled by a number of ecological parameters, including acidity, temperature, nutrients, and salinity. Particular to this group of organisms is a cell wall (frustule) composed of silica, which upon death remains well preserved in sediment. Diatom valves have two basic shapes: pennate (linearly and bilaterally symmetrical; Figure 4a) and centric (radially symmetrical; Figure 4b). They may exist in either solitary or colonial forms. The solitary forms are free-living cells that are found floating in water (planktonic forms; Figure 4b), or may be mobile or im-mobile on the basal material. Colonial forms are groups of cells attached to each other by either clasp-like spines or mucilage secretions, or to various plant (epiphytic forms) or mineral substrata (rocks — epilithic; sand grains — episammic; mud — epipelic; Barber and Haworth 1981; Smol 1987). Each diatom frustule is comprised of two separate valves, the morphology of which is unique and identifiable to the species level. Identifications are made with reference to standard taxonomic monographs, such as those of Hustedt (1930) and Patrick and Reimer (1966), as well as numerous smaller, often site-specific studies, i.e., Poulin et al. (1984a). The fact that diatoms can be identified to the species level, and have known ecological and habitat affinities, makes these organisms a strong (though underutilized) tool in archaeological investigations. Examples of diatom applications to archaeology include tracing the provenance of artifacts, assessing the origin, stratigraphy, and depositional environment of archaeological sediments, and reconstructing changes in climate and paleoenvironments (Battarbee 1988; Jug-gins and Cameron 1999). In the present study, however, diatoms are used to reconstruct changes in salinity as a reflection of glacioisostatic uplift and the progressive isolation of basins from open marine conditions to isolated freshwater lakes.

One of the strongest controls on diatom distribution is their tolerance of different salinities. Recognizing this trait, diatoms have become a key tool in reconstructing sea-level changes, particularly in formerly glaciated regions, where they mark the change in salinity as a basin became isolated from the sea due to postglacial uplift (Figure 5; Stabell 1987; Pienitz et al. 1991). While the salinity tolerance of some di-atoms is very specific, many diatoms can exist under a range of different salinities, and therefore their distribution may span the breadth of the isolation phase. Thus, while individual species can only give a sense of salinity changes, marked changes in the overall diatom assemblage can be used to collectively estimate the point in a core at which the basin became isolated (or transgressed). This “isolation contact” can then be dated by radiocarbon methods (e.g., Stabell 1987; Pienitz et al. 1991; Denys and de Wolf 1999; Eronen et al. 2001). Isolation of a basin itself can take tens to hundreds of years and is dependent on many factors, including uplift rate and tidal regime. Other considerations include basin and catchment morphometry, marine incursions subsequent to isolation (i.e., storm surges), marine aerosol deposition, and leaching of salts from the surrounding catchment. Thus, recently isolated basins that lie only slightly above sea level can still remain brackish, or retain a lower brackish water layer overlain by freshwater. Such basins are given the name “glo” in Finland, while entirely freshwater basins are called “flada” (Munsterhjelm 1987). Diatom assemblages from glo basins can be expected to exhibit both freshwater and brackish floras.

In a very generalized fashion, the halobian, or salinity, classification for diatoms can be broken down into four categories: (i) Halophobous [strictly freshwater], 0l (parts per thousand) salt, (ii) Oligohalobous (indifferent (ind.) or halophilic) [fresh-weakly brackish], .0.2l salt, (iii) Mesohalobous [brackish], ~30 to 0.2l salt, and (iv) Polyhalobous [marine], /30l salt. Of these three, the mesohalobous, or brackish, phase is the most difficult to define in both a literal and numerical sense, as any divisions are sure to be artificial rather than defined according to biological thresholds. While several halobian classification schemes have been developed, those that combine salinity ranges with habitat, or life forms, seem to provide the greatest insight into past environmental changes, and have thus been utilized here.

In order to preserve a record of recent marine submergence by a sea-level rise of approximately 3-4 metres, a lake basin threshold must lie between the present shoreline and at most 3 metres above sea level. Candidate basins in the Port au Choix region that tentatively met this requirement were identified on aerial photographs and 1:50,000 topographic maps (10 m contour interval), based on their close proximity to the present coastline or their situation at elevations below 10 metres above sea level, respectively. Lake basins located near coastal towns were checked on 1:2500 or 1:5000 community maps (contour interval 2 m), on which surveyed lake levels were provided. Summer field reconnaissance eliminated lakes dammed by beach berms and extremely shallow or deep lakes, where sediment thicknesses were limited or sediment reworking was likely. Lake outlets were also examined and surveyed.

Two lakes met the criteria for sampling. Otter Pond is located within the town limits of Hawke’s Bay at 1.5± metres above sea level and within 450 metres of the present coastline (Figure 1). The lake is approximately 50 hectares in area and mostly 1 metre or less in depth. The coring location (50.592EN; 57.195EW) was 90 metres out from the midpoint of the southern shoreline in 1.3-metre water depth. Historically, the lake was used to store logs for the spring drive through the outlet channel to Hawkes Bay. As a result, the outlet channel has been straightened but there is no apparent evidence for deepening of the channel, which is floored by large boulders.

Gilmores Pond is the local name for a 25-hectare lake located between the communities of Brig Bay and Plum Point (51.062EN; 56.891EW; Figure 1). The lake outlet is at an elevation of approximately 0.3 metres above sea level and is separated from Brig Bay by a broad shallow inlet (500 m long) that is inundated by the sea during high water levels. The coastal road crosses the outlet and bedrock is encountered at the surface nearby, suggesting that the basin is a bedrock hollow. The lake is more than 10 metres deep, with the deepest area near its southeast corner. The community of Brig Bay had considered using the lake as a domestic water source but encountered brackish water at the bottom (indicating this is a glo basin).

Coring was carried out in the winter, enabling the lake ice to be used as a coring platform. A hole was drilled through the ice cover and then sediment cores were extracted from the deepest basin of each pond, using a simple percussion corer (Reasoner 1993). Retrieved core lengths were 1.8 metres in Gilmores Pond and 2.4 metres in Otter Pond. Core tubes were capped and sealed in the field, and then returned to the laboratory where they were split lengthwise and subsampled for diatoms and other sedimentological and biological characteristics.

Following a visual description of the sediment cores, samples from halved sections were taken every 10 centimetres for bulk density, organic and carbonate content, and every 5 centimetres for diatom identification. Sediment colour was recorded using a Munsell soil colour chart. Standard analytical procedures using loss-on-ignition (LOI) methods were followed for bulk density, and organic and carbonate content (Berglund 1986). Samples (1 cc) for diatom analysis were first subjected to a digestion in 10% hydrochloric acid (HCl) to remove carbonates, and then 30% hydrogen peroxide (H₂O₂) to remove organic matter. Cleaned samples were then dispersed onto coverslips and mounted onto glass slides using Meltmount⁸(refractive index = 1.704) according to standard diatom procedures (Battarbee 1986). A total of 500 diatom valves were counted on each slide under oil immersion at x1250 magnification.

Species identification and salinity classification was made with reference to Bérard-Therriault et al. (1986, 1987), Campeau et al. (1999), Cardinal et al. (1984, 1986), Foged (1978), Krammer and Lange-Bertalot (1986, 1988, 1991a,b), Mölder and Tynni (1969, 1972), Peragallo and Peragallo (1897-1908), Poulin et al. (1984a,b,c), van Dam et al. (1994), and Vos and de Wolf (1988, 1993). Fragilaria nomenclature sensu Hustedt (1930) and Patrick and Reimer (1966) is used in this study.

Visual scans at x1250 magnification of prepared diatom samples (visually assessed every 5 cm and counted every 10-cm interval) revealed that the sediments of the two ponds were dominated by marine and brackish flora for almost their entire length. Freshwater (oligohalobous (ind.)) taxa were only prominent in the uppermost (<10 cm) sediment, wherein the cores were subsampled and diatoms counted at 1-centimetre intervals. Halophobous taxa were collectively very rare (<1%), which likely reflects the recent emergence of these basins and their proximity to the sea. Diatom abundance was high in all samples, and while counts from each slide commonly contained more than 40 different species, most individual species were too sparse to be meaningfully represented. Percentage abundances of the dominant species in Gilmores and Otter ponds are presented in Figures 6 and 7.

The diatom records presented from Gilmores and Otter ponds (Figures 6 and 7), and visual scans of samples at 5-centimetre intervals from throughout the total core lengths, indicate that these coastal basins only recently became isolated from the sea. Despite the absence of basal radiocarbon dates that would conclusively demonstrate the lengths of record in each core, we are confident that in consideration of low sedimentation rates recorded in similar basins and environmental settings (e.g., Pienitz et al. 1991; Anderson and Macpherson 1994), both cores extend into the early Holocene, and hence would incorporate a late Holocene sea-level oscillation had one occurred. Further evidence in support of this assertion is the observation in both cores, that with decreasing core depth, there is a progressive increase in epiphytic diatom taxa and an accordant decrease in epipelic and epipsammic species. This likely indicates an increase in the amount of plant material, and would be expected as the basin was raised further into the optimal photic zone. Similarly, the sedimentology and marine macrofossil records (Figures 6 and 7) support the assertion of progressive uplift of marine basins and emergence only in the uppermost sediment. Thus, the pattern of progressive shallowing and absence of oligohalobous (ind.) diatom taxa throughout all but the uppermost sediment is fully in agreement with the notion that these basins were being uplifted under a type-A sea-level history, and therefore disputes Grant’s (1994) notion of a late Holocene sea-level oscillation, or modified type-B curve.

If the horizontal sea cliff along the west coast of the Great Northern Peninsula is therefore not of late Holocene age, when was it formed? One possible answer is that it dates to the last interglacial period about 120,000 years ago. Assuming there have been no net changes in crustal position or oceanographic conditions over the last glacial period, then the last interglacial shore should be coincident or parallel to the present shore. For example, a relic shore platform has been observed fairly consistently at about 4-6 metres above its modern counterpart on the Burin Peninsula, the Avalon Peninsula, and parts of southwest Newfoundland (Grant 1989). A similar feature and related deposits elsewhere in Atlantic Canada have been attributed to the last interglacial (Grant 1989), and therefore it is not unreasonable to consider the raised sea cliff on the Great Northern Peninsula a northeastern extension of this interglacial shore.

On the basis of the confirmed sea-level history and reconstructed palaeoshoreline configuration for Port au Choix, we were able to establish that the MAI cemetery was located on an ancient island, and to identify a prominent raised marine terrace between 6 and 8 metres above sea level, on what would have been a sheltered “mainland” shoreline, as the strongest candidate for a MAI living site. We selected the most likely location for MAI habitation on this ancient shoreline by identifying critical site selection factors such as proximity to fresh water, accessibility by boat, a well-drained substrate, and a flat broad terrace. Employing this site prediction strategy, intensive and sustained trenching through 1.4-metre-thick peat in a wooded area was conducted, and eventually yielded a MAI non-mortuary site (the Gould Site, EeBi-42, Renouf and Bell 2002). Without this study’s integrated assessment of sea-level history, coastal geomorphology, and archaeology, it is unlikely that we would have otherwise persisted with such a time-consuming testing programme, and achieved the success that we did.

Having demonstrated the utility of the isolation basin approach for testing sea-level histories in the Port au Choix area, we can project its potential use elsewhere in Newfoundland and Labrador in the following circumstances: Where raised marine sediments and organic remains suitable for radiocarbon dating are either absent or poorly preserved. Much of coastal Labrador is proposed as a continuously emergent landscape (type-A sea-level history; Clark and Fitzhugh 1992), but the bedrock-dominated coastline rarely preserves raised marine deposits, except in sheltered inlets and estuaries (e.g., Porcupine Strand, Smith et al. 2003), and carbonate fossils (e.g., shells) are commonly dissolved in the acidic groundwater.

RSL curves therefore are poorly constrained and rely on modeled uplift parameters (e.g., exponential decay constant “k,” Clark and Fitzhugh 1992). Numerous rock basins below marine limit, however, could potentially preserve a marine-lacustrine isolation sequence that could be radiocarbon dated using plant macrofossils (see, for example, Jordan 1975). Selective application of the isolation basin approach would thus generate sea-level data in support of archaeological investigations, and further constrain modeled sea-level history.

Where much of the Holocene sea-level record is submerged in the shallow offshore. For most of the Island of Newfoundland, sea level has been rising from an early to mid Holocene lowstand. The depth of the lowstand generally decreases towards the interior and the Great Northern Peninsula, with maximum values (-30 m) off the south and southwest coasts. The lowstand sea level has rarely been dated, but is likely to have occurred earlier in the offshore and outer coast than in nearshore areas. Bell and Renouf (2003) argued that this variable and complex sea-level history caused the uneven coastal distribution of late MAI sites (6290-3340 cal BP), and the apparent absence of early MAI sites (8860-6290 cal BP), despite their presence in nearby southern Labrador. They proposed that in areas where the lowstand is greater than -20 metres water depth, late MAI sites older than 3200 cal BP are today submerged, whereas in all areas outside the Great Northern Peninsula (which has experienced continuous emergence), early MAI sites would similarly be under water.

Testing this hypothesis initially requires detailed sea-level reconstruction from submerged records of former coastal environments. Unfortunately, the marine transgression across these environments can rework and destroy the landform and sedimentary evidence. One exception is the sedimentary record preserved in transgression basins, which acted as freshwater lakes when sea level was lower, but subsequently were transgressed by salt water as sea level rose. Such basins have been identified in the nearshore off northeast Newfoundland and are the focus of current coring initiatives.