Glacial lake sediments exposed at two sites in Skagit Valley, Washington, encase abundant macrofossils dating from 27.7 to 19.8 cal ka BP. At the last glacial maximum (LGM) most of the valley floor was part of a regionally extensive arid boreal (subalpine) forest that periodically included montane and temperate trees and open boreal species such as dwarf birch, northern spikemoss, and heath. We used the modern distribution and climate of 14 species in 12 macrofossil assemblages and a probability density function approach to reconstruct the LGM climate. Median annual precipitation (MAP) at glacial Lake Concrete (GLC) was ~50% lower than today. In comparison, MAP at glacial Lake Skymo (GLS) was only ~10% lower, which eliminated the steep climate gradient observed today. Median January air temperature at GLC was up to 10.8°C lower than today at 23.5 cal ka BP and 8.7°C lower at GLS at 25.1 cal ka BP. Median July air temperature declines were smaller at GLC (3.4°C–5.0°C) and GLS (4.2°C–6.3°C). Warmer winters (+2°C to +4°C) and increases in MAP (+200 mm) occurred at 27.7, 25.9, 24.4, and 21.2–20.7 cal ka BP. These changes accord with other regional proxies and Dansgaard–Oeschger interstades in the North Atlantic.

Some lateral moraines contain a rich record of Holocene glacial expansion. Previous workers have used such evidence to document glacial fluctuations in western Canada, Alaska, and the U.S. Pacific Northwest, but similar studies in Patagonia are uncommon. Here we report on the late Holocene behavior of Stoppani Glacier, a 75 km2 glacier sourced in the Cordillera Darwin, southernmost Patagonia. Based on radiocarbon-dated wood and organic material contained in the glacier's northeast lateral moraine, we infer that Stoppani Glacier advanced shortly before 3.8–3.6, at 3.2–2.8, 2.3–2.1, and 0.3–0.2, and possibly sometime before 1.4–1.3 and 0.8–0.7 cal ka BP. These advances culminated at 0.3–0.2 cal ka BP, when the glacier constructed a prominent end moraine, marking its greatest extent of the past 4000 years. Although the timing of several of the advances overlap with the age range of glacial expansion recognized elsewhere in Patagonia, some do not. Asynchronous behavior observed in the glacial record may arise from the type of evidence (e.g., lateral stratigraphy vs. end moraine) used to document glacial fluctuations or variations in climate or glacial response times. A significant difference between the Stoppani record and some other Patagonian records is that the former indicates general expansion of ice over the last 4000 years, whereas the latter indicate a net decrease in extent over that period.

Our study combines new geological and paleoecological information to reconstruct the glacial history and terrestrial paleoenvironments on Haida Gwaii during the advance phase of the Fraser glaciation (Marine Isotope Stage 2). At Cape Ball on eastern Graham Island, five accelerator mass spectrometry radiocarbon ages ranging from 23,200±280 to 26,650±390 14C yr BP (ca. 27,000–31,400 cal yr BP) record the earliest approach of mainland glaciers to Haida Gwaii. Abundant marine dinoflagellate cysts indicate isostatic depression by glacial ice in Hecate Strait to the east. At Mary Point on the north coast of Graham Island, similar outwash of a piedmont lobe advancing westward along Dixon Entrance preserves plant remains dated from 19,270±360 to 23,740±300 14C yr BP (22,500–28,600 cal yr BP). These sediments also contain marine indicators. Plant macrofossils, pollen, and invertebrates support the geological evidence of a proglacial environment under a colder-than-present macroclimate. Although some trees were likely present on Graham Island at this time, tundra-like plant communities dominated low-lying areas. A large area that appears to have been ice-free during this time is a portion of the continental shelf off the east coast of Moresby Island, referred to provisionally as the “Hecate Refugium.”

Sharp-crested moraines, up to 120 m high and 9 km beyond Little Ice Age glacier limits, record a late Pleistocene advance of alpine glaciers in the Finlay River area in northern British Columbia. The moraines are regional in extent and record climatic deterioration near the end of the last glaciation. Several lateral moraines are crosscut by meltwater channels that record downwasting of trunk valley ice of the northern Cordilleran ice sheet. Other lateral moraines merge with ice-stagnation deposits in trunk valleys. These relationships confirm the interaction of advancing alpine glaciers with the regionally decaying Cordilleran ice sheet and verify a late-glacial age for the moraines. Sediment cores were collected from eight lakes dammed by the moraines. Two tephras occur in basal sediments of five lakes, demonstrating that the moraines are the same age. Plant macrofossils from sediment cores provide a minimum limiting age of 10,550–10,250 cal yr BP (9230±5014C yr BP) for abandonment of the moraines. The advance that left the moraines may date to the Younger Dryas period. The Finlay moraines demonstrate that the timing and style of regional deglaciation was important in determining the magnitude of late-glacial glacier advances.

Two sand sheets underlying tidal marshes at Fair Harbour, Neroutsos Inlet, and Koprino Harbour on the northwestern coast of Vancouver Island, British Columbia, were probably deposited by tsunamis. The sand sheets become thinner and finer-grained landward, drape former land surfaces, contain marine microfossils, are locally graded or internally stratified, and can be correlated with earthquakes that generated tsunamis in the region. 137Cs dating and historical accounts indicate that the upper sand sheet was deposited by the tsunami from the great Alaska earthquake in 1964. Radiocarbon ages on plant fossils within and on top of the lower sand sheet show that it was deposited sometime after about A.D. 1660. We attribute the lower sand sheet to a tsunami from the most recent plate-boundary earthquake on the Cascadia subduction zone about 300 yr ago, extending the documented effects of this earthquake north of the Nootka fault zone. The 1964 tsunami deposits differ little in thickness and continuity among the three marshes. In contrast, the lower sand sheet becomes thinner and less continuous to the north, implying a tsunami source south of the study area.

New data from a deep-sea core in the eastern North Pacific Ocean indicate that the western margin of the Late Wisconsin Cordilleran Ice Sheet began to retreat from its maximum position after 15,600 yr B.P. Ice-rafted detritus is present in the core below the 15,600 yr B.P. level and was deposited while lobes of the Cordilleran Ice Sheet advanced across the continental shelf in Queen Charlotte Sound, Hecate Strait, and Dixon Entrance. The core data are complemented by stratigraphic evidence and radiocarbon ages from Quaternary exposures bordering Hecate Strait and Dixon Entrance. These indicate that piedmont lobes reached the east and north shores of Graham Island (part of the Queen Charlotte Islands) between about 23,000 and 21,000 yr B.P. Sometime thereafter, but before 15,000–16,000 yr B.P., these glaciers achieved their greatest Late Wisconsin extent. Radiocarbon ages of late-glacial and postglacial sediments from Queen Charlotte Sound, Hecate Strait, and adjacent land areas show that deglaciation began in these areas before 15,000 yr B.P. and that the shelf was completely free of ice by 13,000 yr B.P.

Models of late-glacial environmental change in coastal areas are commonly based on radiocarbon ages on marine shell and basal lake sediments, both of which may be compromised by reservoir effects. The magnitude of the oceanic reservoir age in the inland waters of the Georgia Basin and Puget Lowland of northwestern North America is inferred from radiocarbon ages on shell-wood pairs in Saanich Inlet and previously published estimates. The weighted mean oceanic reservoir correction in the early and mid Holocene is −720±90 yr, slightly smaller than, but not significantly different from, the modern value. The correction in late-glacial time is −950±50 yr. Valley-head sites yield higher reservoir values (−1200±130 yr) immediately after deglaciation. The magnitude of the gyttja reservoir effect is inferred from pairs of bulk gyttja and plant macrofossil ages from four lakes in the region. Incorporation of old carbon into basal gyttja yields ages from bulk samples that are initially about 600 yr too old. The reservoir age declines to less than 100 yr after the first millennium of lake development. When these corrections are accounted for, dates of deglaciation and late-glacial sea-level change in the study area are pushed forward in time by more than 500 yr.

Twenty-two new radiocarbon ages from Skagit valley provide a detailed chronology of alpine glaciation during the Evans Creek stade of the Fraser Glaciation (early marine oxygen isotope stage (MIS) 2) in the Cascade Range, Washington State. Sediments at sites near Concrete, Washington, record two advances of the Baker valley glacier between ca. 30.3 and 19.5 cal ka BP, with an intervening period of glacier recession about 24.9 cal ka BP. The Baker valley glacier dammed lower Skagit valley, creating glacial Lake Concrete, which discharged around the ice dam along Finney Creek, or south into the Sauk valley. Sediments along the shores of Ross Lake in upper Skagit valley accumulated in glacial Lake Skymo after ca. 28.7 cal ka BP behind a glacier flowing out of Big Beaver valley. Horizontally laminated silt and bedded sand and gravel up to 20 m thick record as much as 8000 yr of deposition in these glacially dammed lakes. The data indicate that alpine glaciers in Skagit valley were far less extensive than previously thought. Alpine glaciers remained in advanced positions for much of the Evans Creek stade, which may have ended as early as 20.8 cal ka BP.

A peaty marsh soil is sharply overlain by a sand sheet and intertidal mud at tidal marshes near Tofino and Ucluelet, Vancouver Island, British Columbia. Foraminifera and vascular plant fossils show that the buried soil was submerged suddenly and was covered quickly by sand. Radiocarbon ages place this event between 100 and 400 yr ago. The coastal subsidence suggested by the submergence occurred in an area of net late Holocene emergence, perhaps during the most recent great earthquake on the northern part of the Cascadia subduction zone. The sand sheet overlying the peaty soil records the tsunami triggered by this earthquake. Similar stratigraphic sequences of about the same age have been reported from estuaries along the outer coasts of Washington and northern Oregon, suggesting that hundreds of kilometers of the Cascadia subduction zone may have ruptured during one, or a series of plate-boundary earthquakes less than 400 yr ago.

Two exposures of organic-rich interstadial sediments in central British Columbia provide information on middle Wisconsinan environments and climates near the center of the region subsequently covered by the late Wisconsinan Cordilleran Ice Sheet. Interstadial sediments at Bullion Pit overlie drift of early Wisconsinan or older age and underlie thick drift of late Wisconsinan age. Alluvium (or colluvium) and peat were deposited on the floor of the ancestral Quesnel River valley 46,000–40,000 14C yr ago when the vegetation consisted of spruce forest with dry openings and local fens and the climate was colder and perhaps drier than today. This is broadly consistent with paleoclimatic reconstructions for the same time interval for Babine Lake, 400 km to the northwest, and for Meadow Creek, 400 km to the southeast. Plant-rich pond sediments containing tephra layers and vertebrate remains are exposed between two drifts in a ravine at Mexican Hill, 30 km east of Quesnel. Although they may be contemporaneous with the Bullion Pit beds, the nonglacial sediments at Mexican Hill more likely were deposited sometime after the warmest part of the last interglaciation, but prior to 50,000 yr B.P. At that time, the vegetation at Mexican Hill probably was parkland. The present vegetation at Mexican Hill is boreal forest; thus, a drier and possibly cooler climate is indicated.

A peat bed on east-central Graham Island of the Queen Charlotte Islands occurs within a nonglacial fluvial succession that is both overlain and underlain by glacial deposits. Radiocarbon dates of 27,500 ± 400 and 45,700 ± 970 yr B.P. at the top and base of the peat, respectively, indicate that it was deposited during the mid-Wisconsin nonglacial interval. The peat is the first documented mid-Wisconsin organic deposit in northern coastal areas of British Columbia. Three local pollen zones are represented. The lowest zone (PM-1) is restricted to sandy silt directly underlying the dated peat. Very high Cyperaceae and moderate Poaceae pollen percentages characterize zone PM-1, and a variety of other herbs are common, suggesting an open landscape rather than a forested one. The middle zone (PM-2) is characterized by abundant pollen of Picea, Tsuga mertensiana, and Cyperaceae, and also contains pollen of Abies, a genus now absent from the Queen Charlotte Islands. Graham Island probably had extensive forests at this time, but abundant pollen and macrofossils of Cyperaceae and emergent aquatics such as Hippuris vulgaris, Veronica scutellata, Potentilla palustris, and Menyanthes trifoliata indicate that there also were open wetland areas. Zone PM-3 also contains abundant arboreal pollen. Large amounts of Sphagnum spores and Selaginella selaginoides megaspores indicate succession of the wetland area at the sample site to a peat bog. Paleoecological analysis of the data suggests that subalpine vegetation elements were depressed by at least 400 m, probably due to a cooler climate. Probable modern analogs in southeastern Alaska and the presence of Abies (probably A. amabilis) indicate that precipitation was higher on eastern Graham Island during the mid-Wisconsin than at present.

The level of Kluane Lake, the largest lake in Yukon Territory, was lower than at present during most of the Holocene. The lake rose rapidly in the late seventeenth century to a level 12 m above present, drowning forest and stranding driftwood on a conspicuous high-stand beach, remnants of which are preserved at the south end of the lake. Kluane Lake fell back to near its present level by the end of the eighteenth century and has fluctuated within a range of about 3 m over the last 50 yr. The primary control on historic fluctuations in lake level is the discharge of Slims River, the largest source of water to the lake. We use tree ring and radiocarbon ages, stratigraphy and sub-bottom acoustic data to evaluate two explanations for the dramatic changes in the level of Kluane Lake. Our data support the hypothesis of Hugh Bostock, who suggested in 1969 that the maximum Little Ice Age advance of Kaskawulsh Glacier deposited large amounts of sediment in the Slims River valley and established the present course of Slims River into Kluane Lake. Bostock argued that these events caused the lake to rise and eventually overflow to the north. The overflowing waters incised the Duke River fan at the north end of Kluane Lake and lowered the lake to its present level. This study highlights the potentially dramatic impacts of climate change on regional hydrology during the Little Ice Age in glacierised mountains.

Relative sea level at Vancouver, British Columbia rose from below the present datum about 30,000 cal yr B.P. to at least 18 m above sea level 28,000 cal yr B.P. In contrast, eustatic sea level in this interval was at least 85 m lower than at present. The difference in the local and eustatic sea-level positions is attributed to glacio-isostatic depression of the crust in the expanding forefield of the Cordilleran ice sheet during the initial phase of the Fraser Glaciation. Our findings suggest that about 1 km of ice was present in the northern Strait of Georgia 28,000 cal yr B.P., early during the Fraser Glaciation.

The ages of some tsunami deposits can be determined by optical dating, a key requirement being that the deposits are derived from sediment that was reworked and exposed to daylight by tidal currents, waves, wind, or bioturbation during the last years before the tsunami. Measurements have been made using 1.4 eV (infrared) excitation of K-feldspar grains separated from samples of prehistoric tsunami sand sheets and modern analogs of tsunami source sediments at four sites in Washington state and British Columbia. Source sands gave equivalent doses indicative of recent exposure to daylight. Tsunami sand at Cultus Bay, Washington, yielded an optical age of 1285 ± 95 yr (calendric years before A.D. 1995, ±1σ). At 2σ, this age overlaps the range of from 1030 to 1100 yr determined through a combination of high-precision radiocarbon dating and stratigraphic correlation. Tsunami sands at three sites near Tofino and Port Alberni on Vancouver Island, British Columbia, have optical ages of 260 ± 20, 325 ± 25, and 335 ± 45 yr. Historical records and radiocarbon dating show that the sand at each of the three sites is between 150 and 400 yr old. These optical ages support the hypothesis that the Vancouver Island sands were deposited by a tsunami generated by a large earthquake on the Cascadia subduction zone about 300 yr ago.

We reconstructed late Holocene fluctuations of Kluane Lake in Yukon Territory from variations in bulk physical properties and carbon and nitrogen elemental and isotopic abundances in nine sediment cores. Fluctuations of Kluane Lake in the past were controlled by changes in climate and glaciers, which affected inflow of Slims and Duke rivers, the two largest sources of water flowing into the lake. Kluane Lake fluctuated within a narrow range, at levels about 25 m below the present datum, from about 5000 to 1300 cal yr BP. Low lake levels during this interval are probably due to southerly drainage of Kluane Lake to the Pacific Ocean, opposite the present northerly drainage to Bering Sea. Slims River, which today is the largest contributor of water to Kluane Lake, only rarely flowed into the lake during the period 5000 to 1300 cal yr BP. The lake rose 7–12 m between 1300 and 900 cal yr BP, reached its present level around AD 1650, and within a few decades had risen an additional 12 m. Shortly thereafter, the lake established a northern outlet and fell to near its present level.