Corps de l’article

Introduction

The postglacial history of mountain forest ecosystems is well documented throughout much of southern British Columbia (Mathewes, 1985; Hebda, 1995). It has generally been assumed that these postglacial dynamics primarily record responses to climatic change. Nevertheless, concerns persist that vegetation changes may have lagged behind climate change (Davis, 1989). Paleotemperature interpretations based on pollen alone have proven problematic in other regions, where megafossils do not necessarily correlate with pollen and macrofossil data (e.g., Barnekow, 1999; Kullman, 1999).

Sub-fossil midges have been accepted as robust paleotemperature indicators (Battarbee, 2000), and we have been using these as a primary tool for reconstructing Holocene paleotemperatures for British Columbia. Our midge-temperature inference model reveals a six degree mean summer temperature range over the last 13 000 years (Palmer et al., 2002; Rosenberg et al., 2004). This use of midges in multi-proxy studies facilitates detailed comparisons of the vegetation record (Williams et al., 2004) with independent paleotemperature estimates (Smith et al., 1998).

Here, we reconstruct vegetation change using pollen, spores, plant macrofossils and charcoal, and examine these changes relative to the midge-inferred paleotemperature record (Palmer et al., 2002; Walker and Pellatt, 2003; Rosenberg et al., 2004). We also support our observations of landscape change via sediment yield analyses, a useful indicator of land surface stability (Brunsden, 1990; Gorham et al., 2001). Used together, landscape stability and vegetation response provide reliable measures of ecological sensitivity. Furthermore, we compare the results from Lake of the Woods with data from several nearby sites in the Engelmann Spruce-Subalpine Fir forest to provide a comprehensive view of landscape sensitivity to climate change, and broaden our knowledge of high elevation forests in the region.

Study Area

Cathedral Provincial Park is a 33 000 ha park situated on the north-eastern slopes of the Cascade Mountains in British Columbia, between the Ashnola River and the International Boundary with the USA (Fig. 1). Much of the park is considered an excellent example of late-successional Engelmann Spruce-Subalpine Fir (ESSF) forest and alpine-tundra (AT) wilderness, however it extends down valleys into drier Interior Douglas Fir (IDF) and Ponderosa Pine (PP) ecosystems (Meidinger and Pojar, 1991). This site (Fig. 2) was chosen because of the park’s limited historical human impact (Ministry of Environment and Parks, 1987) and its location within the main ecological range of a typical ESSF xeric cold (xc) ecosystem (Coupé et al., 1991). Mean annual temperature and precipitation of the ESSFxc are 1.7 °C and 565 mm, respectively (Lloyd et al., 1990). Lake of the Woods (49° 03’ N, 120° 11’ W), is situated at an elevation of 2 060 m, downstream of Pyramid (2 070 m) and Glacier Lakes (2 200 m), and is approximately 3.6 m deep with a surface area of 2.5 ha. Sediments in the lake are rich in organic matter, unlike the relatively minerogenic sediments of nearby Glacier, Ladyslipper, and Pyramid lakes, and thus are more suitable for our analyses. Underlying geology consists of Permian aged sediments (including limestone), Triassic aged lavas, Jurassic aged granitic plutons, and Eocene aged lavas and sediments (Melcon, 1975).

Figure 1

Map showing Lake of the Woods sampling site in Cathedral Provincial Park, as well as the park location relative to other locations mentioned in the text.

Emplacement des sites d’échantillonnage dans le parc provincial Cathedral et localisation du parc par rapport aux autres lieux mentionnés dans le texte.

-> Voir la liste des figures

The surrounding forest consists of a mixture of Pinus contorta ex Loud. var. latifolia Engelm. (lodgepole pine), Picea engelmannii Parry ex Engelm. (Engelmann spruce), and Abies lasiocarpa Hook. (Nutt.) (subalpine fir) (Table I). Larix lyallii Parl. in DC. (alpine larch) is abundant at higher elevations, forming a parkland between the ESSF and AT. Treeline occurs over a narrow elevation gradient, forest canopy opening at about 2 130 m, and absolute tree limit at 2 290 m. Non-forested areas consist of scattered low shrubs and herbaceous, alpine meadows (Ratcliffe and Turkington, 1987, 1989) shaped by cold temperatures and strong winds (Saunders and Bailey, 1996).

Methods

Sampling

Lake of the Woods was cored during the summer of 1995 using a modified Livingstone piston sampler (Wright, 1967). Ten cores were analysed to determine sediment yield (Evans, 1997a, b; Evans and Slaymaker, 2004). Core A9 (3.86 m long), taken from the deepest part of the lake, was subsampled in intervals of 1 to 5 cm for macro- and micro-fossil analyses.

Chronology

We employed the sediment chronology previously established by Palmer et al. (2002) for core A9 from Lake of the Woods, which is based upon five radiocarbon dates and two chemically identified tephras (Table II). Calibration of these dates to calendar years was accomplished using CALIB 4.3 (Stuiver and Reimer, 1993). Pollen and charcoal accumulation rates were calculated using only the calibrated ages of the one conifer needle macrofossil and two tephras (Hallett et al., 1997), as an inversion in the early Holocene sediment chronology occurs with respect to the AMS dated macrofossil. Carbonates present in the bulk gyttja samples may have affected the dating (Table III), though the only apparent inaccuracy appears near the carbon plateau at 10 000 14C BP (Hughen et al., 1998).

Pollen, Charcoal, and Plant Macrofossil Analyses

Approximately 1 cm3 of sediment was processed for pollen and spores according to standard methods (Faegri and Iversen, 1989), and mounted on microscope slides in glycerine jelly. Lycopodium tablets were added to samples allowing influx determinations (Stockmarr, 1972). Palynomorph identifications were made using an unpublished key to British Columbia pollen and spores, and McAndrews et al. (1973). A minimum of 300 terrestrial pollen grains was counted from each sample, which was used as the base sum to calculate percentages. Percent (Fig. 3) and influx (Fig. 4) diagrams were produced using TILIA 2.0.b.4. (Grimm, 1993) and TILIAGRAPH 2.0.b.5. (Grimm, 1991). Zonation was facilitated by conducting stratigraphically constrained incremental sums-of-squares cluster analysis (CONISS) on square-root transformed percent data (Grimm, 1987).

Figure 2

Photograph of the study area, looking east from Quiniscoe Mountain. Quiniscoe Lake is at the bottom left, Lake of the Woods in the centre right, and Lakeview Mountain is in the immediate background.

Photographie de l’aire d’étude vue de l’ouest, depuis le mont Quiniscoe. Le lac Quiniscoe est situé en bas, à gauche, Lake of the Woods, à mi-hauteur, à droite, et le mont Lakeview, en arrière-plan.

-> Voir la liste des figures

Table I

Vegetation surrounding Lake of the Woods (percent cover by visual estimation)

Vegetation surrounding Lake of the Woods (percent cover by visual estimation)

-> Voir la liste des tableaux

Table II

Chronology of Lake of the Woods

Chronology of Lake of the Woods

-> Voir la liste des tableaux

Table III

Mean organic matter and biogenic silica proportions for correlated zones based on cores A1, A4b, A6, A7, A8, A9, A10, A11, A12 and A13 (Evans, 1997a)

Mean organic matter and biogenic silica proportions for correlated zones based on cores A1, A4b, A6, A7, A8, A9, A10, A11, A12 and A13 (Evans, 1997a)

-> Voir la liste des tableaux

Figure 3

Pollen and spore percent diagram from Lake of the Woods, Cathedral Provincial Park, British Columbia, plotted on a depth scale. Hollow exaggeration curves are X10. Constrained sum of squares (CONISS) analysis on upland taxa is shown on the right.

Analyst: M. Heinrichs

Diagramme des pourcentages de pollens et de spores de Lake of the Woods, au parc provincial Cathedral, en Colombie-Britannique, en fonction de la profondeur. Les courbes sont exagérées dix fois. La somme contrainte des carrés (CONISS) de l’analyse sur les taxons des hautes terres est présentée à droite.

-> Voir la liste des figures

Figure 4

Pollen influx diagram for selected taxa from Lake of the Woods, plotted on a time scale using a smoothed curve based on the one AMS macrofossil date and two tephra ages. Note that the accumulation rate scales differ among taxa.

Diagramme d’influx polliniques de taxons choisis de Lake of the Woods, selon une échelle de temps dont la courbe a été corrigée à l’aide d’un âge de macrofossile (défini par SMA) et deux âges de tephra. Notez que les échelles de taux d’accumulation diffèrent d’un taxon à l’autre.

-> Voir la liste des figures

Charcoal concentrations were determined for each subsample by extracting macroscopic charcoal from 1 cm3 of sediment that remained on a 180 µm mesh (Millspaugh and Whitlock, 1995; Whitlock and Larsen, 2001). This charcoal-bearing residue was backwashed into a watch-glass and tallied under 4X magnification; concentrations were converted to accumulation rates (pieces cm-2 yr-1) based on the chronology established using tephra and AMS dates. Plant macrofossils recovered from the remainder of the sediment in each subsample are presented as whole needles, needle fragments and seeds (Fig. 5). Plant macrofossils were identified using a key prepared by Dunwiddie (1985) and the reference collection of the Royal British Columbia Museum.

Ordination analysis was implemented to objectively determine if patterns of change exist within the pollen data (Fig. 6), as outlined in Heinrichs et al. (2002). A pilot detrended correspondence analysis (DCA), using the computer program CANOCO 3.12 (ter Braak, 1991), indicated a principal gradient length of 0.8; thus, principal components analysis (PCA) was selected as an appropriate ordination method for subsequent analysis (Jongman et al., 1987; ter Braak and Prentice, 1988).

Midge-Inferred Temperatures

A detailed description of the midge analysis and paleotemperature inference procedure is found in Palmer et al. (2002) and Rosenberg et al. (2004). Midge-inferred temperature records are presented for Lake of the Woods and the southern interior region of British Columbia based on five high-elevation lakes, including Lake of the Woods (Fig. 7a).

Figure 5

Macrofossil record, composed of needles, needle fragments, and seeds for Picea engelmannii, Pinus contorta, and Abies lasiocarpa from Lake of the Woods.

Analyst: M. Heinrichs

Relevé de macrofossiles composé d’aiguilles, de fragments d’aiguilles et de graines de Picea engelmannii, de Pinus contorta et d’Abies lasiocarpa de Lake of the Woods.

-> Voir la liste des figures

Figure 6

Ecosystem trajectory, composed of axis 1 vs. axis 2 sample scores from a principal components analysis (PCA). Zones are clustered using rounded shapes.

Trajectoire de l’écosystème telle que définie par les axes 1 et 2 de l’analyse en composantes principales (ACP). Les zones sont circonscrites à l’intérieur d’ensembles.

-> Voir la liste des figures

Sediment Yield

Sediment yield is a quantitative estimate of coarse (>8 mm) and fine (<8 mm) inorganic material generated through weathering and dustfall, deposited or transferred through an aquatic system (Caine, 1974); it provides an indication of landscape sensitivity to climate change (Gerrard, 1990). Sediment yield to Lake of the Woods (Fig. 8) was calculated using ten cores, correlated using volume magnetic susceptibility measurements made by whole core scanning, and with reference to the core stratigraphy (Evans, 1997a; Evans and Church, 2000; Evans and Slaymaker, 2004). Correlated horizons were assigned ages using the radiocarbon dates and two identified tephras. Sediment depth for each of the five correlated units was converted to sediment volume by multiplying by the area of a thiessen polygon constructed around the core site. Sediment volumes were converted to mass using measured sediment densities, corrected for averaged concentrations of organic matter (Bengtsson and Enell, 1986) and biogenic silica (Engstrom and Wright, 1984) (Table III). Sediment yields were normalised by the basin area and expressed as tonnes per kilometre squared per year (t km-2 yr-1). Confidence intervals around estimated sediment yields were derived using a regression based technique described in detail in Evans and Church (2000).

Results

Stratigraphy

Bluish clay dominates from 386 to 353 cm; however a distinct coarse-grained (2-3 mm) sand layer was noted from 378.5 to 378 cm. A transition to light brown gyttja begins at 353 cm and extends to 346.5 cm. Gyttja occurs from 346.5 to 196 cm, where it begins to be mixed with tephra. The mixture gradually becomes dominated by tephra, and from 189 to 185 cm consists entirely of tephra. This tephra is attributed to the 6730 ± 40 14C BP eruption of Mount Mazama, which is widespread throughout southern mainland British Columbia (Nasmith et al., 1967). This tephra layer has been identified in nearby lakes as the Mazama tephra (Evans, 1997a), and is positioned between radiocarbon dates of 5450 ± 90 and 8580 ± 80 14C BP. Gyttja occurs from 185 cm to the top of the core. A distinct tephra layer interrupts the gyttja from 87 to 82 cm; it is identical in appearance and stratigraphic position to tephra identified by X-ray fluorescence as Mount St. Helens Yn (3390 ± 130 14C BP) in Quiniscoe and Glacier Lakes (Evans, 1997a).

Figure 7

a

Temperature Change (°C)

Temperature Change (°C)

b

Axis 1, PCA

Axis 1, PCA

c

Axis 2, PCA

Axis 2, PCA

d

Picea/Pinus ratio

Picea/Pinus ratio

Depth plots for Lake of the Woods of a) midge-inferred temperature changes for Lake of the Woods adapted from Palmer et al. (2002) and regional consensus reconstruction modified from Rosenberg et al. (2004), b) principal components analysis axis 1 sample scores adapted from Heinrichs et al. (2002) c) principal components analysis axis 2 sample scores, and d) Picea/Pinus pollen ratios.

Graphiques a) des changements de température déduits à partir de l’étude de moucherons (d’après Palmer et al. [2002]) et d’une reconstitution des températures régionales issue d’un consensus (modifiée de Rosenberg et al. [2004]), b) des résultats de l'axe 1 de l'analyse en composantes principales (adaptés de Heinrichs et al. [2002]), c) des résultats de l'axe 2 de l'analyse en composantes principales et d) des ratios de pollen de Picea/Pinus, selon la profondeur, à Lake of the Woods

-> Voir la liste des figures

Figure 8

Postglacial changes in sediment yield in Lake of the Woods derived from ten cores, correlated using volume magnetic susceptibility measurements. Error bars were estimated using a regression technique (Evans and Church, 2000).

Variations postglaciaires de la production des sédiments de Lake of de Woods obtenues à partir de l’étude de dix carottes. Les corrélations ont été établies à l’aide de mesures de susceptibilité magnétique. Les intervalles d’erreur ont été estimés par le biais d’analyses de régression (Evans et Church, 2000).

-> Voir la liste des figures

Pollen, Charcoal, and Plant Macrofossils

Zone I: 386 to 357 cm, 11 200 to 10 800 cal BP, Pinus-Cupressaceae-Picea-Alnus-Artemisia

This zone is characterised by the highest non-arboreal pollen (NAP) percentage values, beginning at >20 % and decreasing to <10 % by the end of the zone. Pinus pollen values increase from 60 % at 386 cm to 80 % at the end of the zone. Pollen derived from the Cupressaceae, Betula, Alnus, Salix, Poaceae and Artemisia is more common on a percentage basis in Zone I than in all subsequent levels of the Lake of the Woods core. The total pollen accumulation rate increases in this zone from ~650 to 7 600 grains cm-2 yr-1. No charcoal was observed. Picea/Pinus ratios are <0.10 (Fig. 7d).

Zone II: 357 to 198.5 cm, 10 800 to 8000 cal BP, Pinus-Picea-Alnus

The proportion of arboreal pollen (AP) is very high and relatively constant in this zone, around 95 %. Pinus values comprise 95 % of the pollen assemblage at the beginning of the zone, but decrease to 85 % by the zone end. Picea increases from about 2 to 10 % through the zone, with a period of lower values occurring around 270 cm. Cupressaceae, Betula, and Abies pollen are present throughout the zone, making up <1, <1, and <5 %, respectively. Alnus varies between 5 and 10 %, and Poaceae and Artemisia occur at <3 %. Isoëtes is more abundant in Zone II than at any other time. Pollen influx is up to ten times greater than in the previous zone, ranging from ~1 300 to 57 000 grains cm-2 yr-1. Picea/Pinus ratios are slightly higher than in Zone I, as several intervals exceed 0.10.

Charcoal accumulation is low with small peaks observed at 346 and 247 cm. Needle fragments and a seed from Picea engelmannii were recovered in this zone, together with needles of Pinus contorta and Abies lasiocarpa. Increased amounts of organic matter were deposited, as determined by the change in sediment from clay to gyttja and the corresponding increase in loss-on-ignition.

Zone III: 198.5 to 0 cm, 8000 cal BP to present, Pinus-Picea

The AP values are >95 % for much of this zone, and Pinus values vary between 75 and 90 % of the pollen assemblage. Picea values range between 5 and 20 %, generally increasing upwards in the core. Tsuga heterophylla and Pseudotsuga/Larix pollen is uncommon (<1 %), but is more abundant in Zone III than elsewhere in the record. Abies also tends to be more abundant than in Zones I or II. Isoëtes spores, and Cupressaceae and Shepherdia pollen decline or disappear early in this zone, whereas Cyperaceae pollen becomes increasingly common in late Zone III. Pollen influx in this zone ranges from ~520 to 4 500 grains cm-2 yr-1, however, less Pinus pollen was deposited than previously. Picea/Pinus ratios are significantly higher in this zone, exceeding 0.25. Charcoal accumulation values are also greater than in previous zones. Needle fragments of Picea engelmannii, Abies lasiocarpa, and Pinus contorta were recovered from the sediments.

Midge-Inferred Temperatures

The midge stratigraphic record for Lake of the Woods has been published in Palmer et al. (2002), thus we provide only a brief summary here. The midge-paleotemperature record was incorporated into a consensus record as developed by Rosenberg et al. (2004) for southern British Columbia (Fig. 7a).

The basal chironomid zone, containing a mix of cold and warm-adapted midges, has inferred temperatures indicating cold mean July air temperatures initially, ca. 8.7 °C, but subsequent warming to 13.1 °C. A major shift to dominance by warm-adapted midges occurs at 357.5 cm, coincident with the pollen zone change. Accumulation of organic material also began at this time, indicating changes in terrestrial and aquatic productivity, likely associated with the warmer temperatures.

Temperate midges, dominating throughout pollen Zone II (Palmer et al., 2002), yielded warm inferred-temperatures ranging between 12.5 and 16.4 °C for this period (Fig. 7a). Increased abundance of cold stenothermous taxa (i.e., Sergentia), and decreased abundance of temperate taxa (i.e., Microtendipes) in Zone III generated cooler inferred-temperatures than earlier, ranging from 13.1 to 8.7 °C.

Ordination Analyses

The plot of PCA species scores depicts a strong separation of open vegetation versus closed forest species along axis 1 (Heinrichs et al., 2002). Artemisia, Alnus, Poaceae, Salix, and Cupressaceae (probably Juniperus, based on current Cupressaceae distribution in British Columbia) have high axis 1 species scores whereas species with low (negative) scores include Picea, Abies and Tsuga. On axis 2, Artemisia and Picea have high species scores, and Pinus and Potamogeton have negative scores.

The plot of PCA sample scores provides an ecosystem trajectory (Fig. 6). The three pollen zones identified via cluster analysis (Zones I, II, and III) represent three biogeochrons: 1) an open steppe during the initial revegetation stage, close to the late-glacial/Holocene boundary, 2) open Pinus parkland forests and Abies- and Picea-dominated stands in the early Holocene, and 3) the modern closed ESSF forest in the middle- to late Holocene.

The plot of axis 1 sample scores versus depth (Fig. 7b, reproduced from Heinrichs et al., 2002) potentially provides a reconstruction of “openness” of the vegetation. The high scores in Zone I indicate that the vegetation likely comprised an open steppe or alpine tundra. Somewhat lower scores in Zone II may reflect open forest parkland. The negative scores of Zone III indicate a more closed ESSF forest, perhaps reflecting cooler, moister conditions in the late Holocene. There is a similarity in the shape of this plot to that of the Picea/Pinus ratio (R2 = 0.74). This is not surprising, as Beaudoin (1986) found that high Picea/Pinus ratios indicate forested vegetation and low ratios more open conditions. We also note an inverse correspondence (R2 = 0.32) between the axis 2 sample scores (Fig. 7c) and the midge-inferred paleotemperature records (Fig. 7a); thus, axis 2 likely relates to the paleotemperature signal.

Sediment Yield

Average sediment yields to Lake of the Woods range between 4.6 and 6.2 t km-2 a-1 over the course of the Holocene. The highest yields are recorded for the period preceding the Mazama tephra, corresponding with pollen Zone II. However, the estimated Holocene variance of sediment yield is within the 95 % confidence intervals of the individual points so that statistically, the sediment yield is constant throughout the Holocene.

Discussion

Late-Pleistocene–Holocene Transition (>10 900 Cal BP)

The high relative abundance of Poaceae, Artemisia and other non-arboreal pollen types clearly indicates that the freshly deglaciated landscape near Lake of the Woods was dominated by open vegetation, perhaps an Artemisia-Poaceae steppe or tundra, with scattered Abies and Picea krummholz or distant forested stands. Similar high elevation late-glacial communities have been documented elsewhere in the southern interior (Fig. 9), at Stoyoma Mountain (Pellatt et al., 1998, 2000), Mount Kobau (Heinrichs et al., 2001a), and Crater Mountain (Heinrichs et al., 2002).

High Cupressaceae values in the earliest part of the Lake of the Woods record suggest that Juniperus communis may have been a prominent plant in this early landscape. It is currently common on poor, mineral soils with a dry, cold climate, such as exposed mountain ridges (Ringius and Sims, 1997). Why this plant seems to have occurred abundantly around Lake of the Woods, but apparently not at other sites nearby, such as Crater Mountain (Heinrichs et al., 2002), is not clear.

The low pollen influx reflects the open character of the vegetation (Heinrichs, 1999), and the midge-inferred summer temperatures are up to 3 °C colder than today. Low temperatures and drought likely restricted tree growth. Although Pinus pollen is abundant, these trees are prolific pollen producers. Much, if not all of the Pinus pollen, was likely derived via long-distance transport, creating the illusion of a treed landscape when few, if any, trees were actually present.

The dominance of minerogenic sediments (Table III) reflects the initial unstable, sparsely-vegetated landscape with low aquatic and terrestrial productivity. The coarse sand layer probably represents a single, large sediment pulse, associated with either high discharge or a mass movement in the unstable land surface. Fires were not an important feature of the landscape during the late-glacial period, probably due to a low fuel load and unsuitable weather conditions, which has been documented in boreal forests (Flannigan et al., 2000).

Climate was changing rapidly at this time, with summer insolation approaching its postglacial maximum. The high albedo of the Cordilleran ice sheet would have reinforced the initially cold climate, and an anticyclone over the ice would have promoted an easterly flow of cold, dry air from the continental interior (Bartlein et al., 1998). As the ice-sheet rapidly waned, the decreasing albedo and high summer insolation promoted increasing temperatures, and westerly flows from the Pacific began to dominate the regional climate (Kutzbach and Guetter, 1986; Barnosky et al., 1987; Anderson et al., 1988). This dramatic shift in temperatures is seen in the midge-paleotemperature reconstructions (Fig. 7a). Climate-induced changes in hydrology are also apparent in the shift from freshwater to saline conditions at nearby Kilpoola and Mahoney Lakes (Heinrichs et al., 2001b).

Xerothermic Interval (10 900-8000 Cal BP)

Midge-inferences indicate early Holocene summer temperatures up to 4 °C warmer than present. At this time, the increasing arboreal pollen percentages and influx, reflect the first establishment of forests, with Pinus, Picea and Abies all contributing to the pollen rain. Present-day forests surrounding Lake of the Woods also consist of a mixture of Pinus contorta var. latifolia, Picea engelmannii, and Abies lasiocarpa. It therefore appears that this dominant aspect of the vegetation was already established in the early Holocene. Macrofossils of all three species were recovered from the early Holocene sediments, confirming this inference. Pinus percentages were higher in the early Holocene than in surficial sediments at Lake of the Woods, whereas Picea and Abies percentages were lower. In addition, the influx of important non-arboreal pollen indicators, such as Poaceae, Artemisia, Alnus and Cupressaceae (probably Juniperus), was particularly high in the early Holocene. Thus, the fossil pollen assemblages reflect a more open parkland-type forest.

Figure 9

Comparison of pollen and midge zones and midge-inferred mean-July air temperature deviations (where available) from Crater Lake (Heinrichs et al., 2002), Lake of the Woods (this paper), Cabin Lake (Pellatt et al., 1998, 2000; Smith et al., 1998), Eagle Lake (Rosenberg et al., 2004) and Buckbean Bog (Heinrichs et al., 2001a). Calendar year ages were estimated using Calib 4.3 (Stuvier and Reimer, 1993).

Comparaison des zones définies à l’aide des pollens et de celles obtenues par l’étude des moucherons, et variations de la température moyenne de juillet déduite à partir de l’étude des moucherons (là où l’information était disponible) au lac Crater (Heinrichs et al., 2002), à Lake of the Woods (présente étude), au lac Cabin (Pellatt et al., 1998, 2000; Smith et al., 1998), au lac Eagle (Rosenberg et al., 2004) et à la tourbière Buckbean (Heinrichs et al., 2001a). Les âges calendaires ont été estimés à l’aide du logiciel Calib 4.3 (Stuvier et Reimer, 1993).

-> Voir la liste des figures

The charcoal record indicates that fires burned regularly throughout the early Holocene, likely supported by a greater fuel supply and the warmer, drier summer climate. This pattern has been observed in other mountainous forests (Hanson and Weltzin, 2000). These early fires at Lake of the Woods were important determinants of forest structure and composition, maintaining a vegetation mosaic dominated by successional forest stands and favourable for shade-intolerant species (Hebda, 1998).

This early Holocene phase is commonly referred to as the xerothermic interval in British Columbia. The high midge-inferred summer temperatures (Fig. 7a) approximately coincide with the Holocene maximum for summer solar insolation. Winter temperatures are difficult to infer from our records. However, the Holocene winter minimum for insolation also occurred at this time (Berger and Loutre, 1991; Webb and Bartlein, 1992). Thus, it is probable that the hot, dry early Holocene summers alternated with severely cold winters. The implications of such cold winters, in terms of forest structure and composition, are unclear, but severe winters are recognised as a key factor limiting infestations by mountain pine beetle (Bentz and Mullins, 1999) and other forest pathogens in present-day forests of interior British Columbia.

Despite the warm, dry summer conditions, sediment yield upstream in Glacier Lake was at its maximum at this time (Evans, 1997b). This may be in part due to an abundance of unconsolidated material higher in the catchment which was left over from deglaciation and available for transport. In the Coast and Rocky Mountains, sediment yield typically increases while passing through the system, whereas the pattern at Lake of the Woods is more similar to that of low-slope systems (Evans, 2000; Schiefer et al., 2001). In Colorado, the period of maximum sediment yield was also associated with maximum warmth, but much of the sediment was air-borne material (Andrews et al., 1985). At Glacier Lake, particle size determination indicated that most of the material consists of silts and clays and is not eolian (Evans, 1997a). Since Glacier Lake and Lake of the Woods are situated in close proximity, the sediment is also unlikely to be of eolian origin at Lake of the Woods. Furthermore, the warm conditions of the xerothermic allowed trees to establish at Lake of the Woods and stabilize the landscape, limiting overall sediment yield (Evans, 1997a; Evans and Slaymaker, 2004).

Mesothermic Interval (8000-5000 Cal BP)

Midge-inferred summer temperatures gradually declined as summer insolation decreased through the middle Holocene. Hebda (1995) refers to this time as the Mesothermic interval.

The decreasing abundance of Cupressaceae and Shepherdia pollen, together with the increases in Picea and Abies, presumably reflect an increase in effective soil moisture, the development of an ESSF-type forest composed of Abies lasiocarpa and Picea engelmannii, and generally more mature, closed forest stands. The decreased pollen influx likely reflects the decreased importance of Pinus, and increased proportions of Picea and Abies; Picea and Abies trees produce much less pollen than Pinus (Faegri and Iversen, 1989; Hebda and Allen, 1993).

Middle Holocene increases in Tsuga heterophylla, and Pseudotsuga/Larix pollen were also noted. Tsuga heterophylla is unlikely to have ever grown near this dry interior site; thus, the Tsuga pollen probably reflects long-distance transport from coastal forests, >50 km to the west. Rosenberg et al. (2003) note Tsuga heterophylla pollen normally constitutes at least 5 % of the pollen rain at sites where the trees are actually present.

The pollen of Pseudotsuga and Larix cannot reliably be distinguished. The increase in Pseudotsuga/Larix pollen may date the arrival of Larix lyalli at the Cathedral Lakes, but this possibility needs confirmation via plant macrofossil or stomate evidence.

Charcoal accumulation increased during the middle Holocene. We suggest this reflects the greater density of the forests and increased fuel supply, rather than an increase in fire frequency due to the increased effective moisture. It is, however, also possible that increased thunderstorm activity may have accompanied the gradual shift to a moister hydrological regime. Similar middle Holocene increases in charcoal accumulation have been documented nearby at Crater Mountain (Heinrichs et al., 2002), at Dog Lake in the Kootenay region of southeastern British Columbia (Hallett and Walker, 2000) and at Sheep Mountain Bog, Montana (Mehringer, 1996).

These patterns of moisture-related landscape change are similar to those documented at other sites in the dry, southern interior of British Columbia. Increasing moisture during the middle Holocene at Mount Kobau, was accompanied by an increase of Pinus on the landscape, rather than Picea and Abies (Heinrichs et al., 2001a). There, fires may have become more intense, destroying mature Pinus trees and/or stands of trees, promoting regrowth of Pinus, and generating abundant charcoal. At Crater Lake, only 60 m higher in elevation and <20 km to the north of our study site, Pinus remained a dominant tree on the landscape, with true ESSF forest only developing ~4000 years later than at Lake of the Woods (Heinrichs et al., 2002). Picea also appears to have been a less important element of the forests at Stoyoma Mountain (Pellatt et al., 1998, 2000), as compared to Lake of the Woods.

The statistically unchanged sediment yield at Lake of the Woods in the middle Holocene is contrary to expectation, since more moisture was available for weathering and transport of material. It is unlikely that exhaustion of sediment sources occurred (Ballantyne, 2002). Evans (1997b) shows an increasing sediment yield at Glacier Lake, which likely intercepted material that otherwise would have been deposited in Lake of the Woods.

Neoglacial Interval (5000 Cal BP to Present)

By 4000 to 5000 BP summer insolation and midge-inferred temperatures were at, or very close to, modern values, and little subsequent change in the vegetation occurred. Picea/Pinus ratios were high, reflecting the cooler climate and moister soils. An increase in Cyperaceae suggests that an extensive marginal sedge fen developed at Lake of the Woods around 4000 cal BP, and provides additional evidence of the increased moisture availability.

Curiously, charcoal accumulation rates were higher during this interval than at any other time. We are not suggesting that fire frequency increased with the decreasing temperatures. The increase may reflect 1) different charcoal fragment preservation characteristics, or number of fragments produced by Picea and Abies relative to Pinus, 2) a change in fire intensity/fuel load, or 3) the influence of first nations (Turner, 1994; Clark and Royall, 1995).

At Mount Revelstoke, Rosenberg et al. (2003) note the arrival of Tsuga heterophylla and T. mertensiana populations after 3500 cal BP, providing additional evidence of the cooler, moister environment. Although late-Holocene glacial advances, and increased sediment yield are evident at many British Columbia sites (e.g., Souch, 1994; Desloges, 1999), this is not apparent at Lake of the Woods. The lack of sediment yield response throughout the Holocene, but especially in the neoglacial period, demonstrates how insensitive Lake of the Woods is to climate-driven, landscape weathering.

Implications

Differences in timing and severity of landscape or vegetation response to climate change between Lake of the Woods and other dry ESSF sites in the southern interior are likely the result of differences in aspect, elevation (Lavender, 1990; Stathers et al., 1990), relief, and specific landscape features — e.g., streams (Schiefer et al., 2001). Lake of the Woods is in a large, north-facing basin surrounded by high mountain ridges, thus air mass movement and temperatures at Cathedral Provincial Park are likely affected by cold air drainage, consequently winter snow-packs remain longer and less drying occurs during the summer. Lake of the Woods receives water from a 2.2 km2 catchment area that includes Pyramid and Glacier Lakes, thus short-term droughts may be mediated by stream inflow, soil moisture, and ground-water seepage (Winter, 1978).

Drier ESSF sites, such as Crater Lake (2 120 m asl), approximately 15 km north of Cathedral Provincial Park, display greater changes in vegetation with Holocene climate change. Fire-dependant Pinus-dominated parkland occurred until about 7500 cal BP, and Abies and Picea were lesser components of a Pinus forest until about 4200 cal BP. Increasing moisture during the neoglacial resulted in modern ESSF composition around 1700 cal BP (Heinrichs et al., 2002). Mount Kobau (1 830 m asl) also responded later than Lake of the Woods to early Holocene warming, shifting from open grasslands to Pinus parkland with increasing mesothermic moisture, and Pinus dominated-ESSF forest late in the neoglacial (Heinrichs et al., 2001a). Though Mount Kobau, Crater Mountain, and Cathedral Provincial Park are classified into the same variant of the ESSF biogeoclimatic zone (xc, or xeric cold), the sites clearly do not share a similar vegetation history, nor do they respond ecologically in the same manner to climate change. The vegetation at Mount Kobau is highly sensitive to climate change, showing the greatest sensitivity, and therefore should be expected to respond significantly to future climate change and disturbance. The forests at Lake of the Woods were much less sensitive to past climate change, suggesting that they will be more resilient to future climate change. Stoyoma Mountain (ESSFdc - dry cold) and Mount Revelstoke (ESSFvc - very wet cold), were similar to Cathedral Provincial Park in the degree of vegetation change relative to climatic change, suggesting that in the southern interior of British Columbia, differences in precipitation regime at higher elevations may be more important than temperature in regulating vegetation changes. Precipitation was also found to have significant influence on subalpine vegetation in Washington state, United States (Fagre et al., 2003).

Site sensitivity to climate change and associated natural disturbance (Parminter, 1998) is relevant to biodiversity and natural resource management, and is critical in selecting sites for habitat preservation. Sites less sensitive to disturbance, including climate change, may be ideal for the preservation of existing ecosystems. These sites may also be suitable for selective forestry, although changes in wildlife habitat may have further implications related to conservation issues or recreational opportunities (Wall, 1998). Since not all sites respond to climate and disturbance in the same manner, better knowledge of the forest history in all parts of the western Cordillera is invaluable in understanding and preserving biodiversity for the future, especially with future climate change (Laprise et al., 2003). It is recommended that paleoecological analyses be used routinely to determine vegetation histories of designated park sites and to plan timber harvesting operations. Sensitive sites may require more consideration in terms of current and future anthropogenic impact such as range use, recreational development, or preservation strategies.

Conclusions

The vegetation history of the Cathedral Provincial Park region consisted of three biogeochrons: open steppe with krummholz spruce and fir occurred during the cool late-glacial interval, pine forest with some spruce and fir and frequent fires occurred in the warm early Holocene, and ESSF forest occurred with increasing moisture after 8000 cal BP, closing after 4000 cal BP under cooler temperatures. Relatively little change in vegetation occurred at this site compared to other dry ESSF sites, such as Mount Kobau or Crater Mountain. Wetter sites, such as Stoyoma Mountain and Mount Revelstoke were also less sensitive than the two drier sites. Midge-inferred temperatures from Lake of the Woods correspond closely with other quantitative records in southern British Columbia. Sediment yield results also suggest a stable environment throughout the Holocene and that Lake of the Woods has a high climatic threshold to geomorphic change. Thus terrestrial ecosystems at Cathedral Provincial Park have a low potential sensitivity to climate change.

The results of this study provide information valuable to site selection for habitat preservation and biodiversity management in the ESSF forests of the southern interior. Low sensitivity to natural disturbance made this site an ideal, and likely fortuitous, choice for the preservation of existing ecosystems and biodiversity, especially under the scenario of future climate change (Zwiers and Kharin, 1998; O’Brien and Leichenko, 2000; Laprise et al., 2003). Conservation strategies in all parts of the Canadian Cordillera should be enhanced by considering vegetation history and aquatic faunal responses to disturbance, including future climate change, because, as this study shows, these responses differ at sites within the same biogeoclimatic zone. This proposal is likely applicable to other geographically diverse regions that are facing increasing pressure of development or tourism. It is recommended that paleoecological analyses be routinely used to determine site-specific vegetation histories. These should be incorporated into the planning for designated park sites, timber harvesting operations, fire management strategies, range use, recreational development, and other important resource management decisions. Sites sensitive to natural disturbance may require more consideration of current and future anthropogenic impacts to ensure biodiversity conservation.