The Fox permafrost tunnel, which penetrates 110 m into frozen sediments of Gold-stream valley, provides a continuous exposure of fossiliferous silt and alluvium above schistose bedrock. Deposition of fluvial gravel was followed by a long interval of loess accretion and permafrost aggradation that was punctuated by episodes of thaw and of gullying and redeposition of silt.
Imbricated sandy gravel above the bedrock contains lenses of finer alluvium that contain wood fragments and some rooted stumps. Radiocarbon dates indicate that the gravel is older than 40 ka, but absence of mature soil and weathering profiles at its upper contact indicates that fluvial activity must have continued until shortly before loess accretion began at the tunnel site.
Silt is the most widespread depositional unit in the tunnel. This deposit is of eolian origin (loess), but some has been redeposited by slope processes. The silt units contain abundant ground ice as pore filings, lenses, wedges, and buried pond ice. Loess accretion was interrupted by a period when little loess accumulated and when large ice wedges formed in the lower loess unit and subsequently were truncated by thaw. Loess began forming sometime before 40 ka and was rapidly accreting by 39 ka under xeric conditions with open vegetation. A sharply decreased rate of loess accretion associated with local erosion and thaw between about 36 and 30 ka is marked by anomalous cation concentration values, lenses of buried sod, fossils indicative of moist to wet substrates, and truncated ice wedges beneath small frozen ponds or streamlets that occupied ice-wedge troughs. A later episode of rapid loess influx under drier conditions began after 30 ka and coincided with glacial advances of late Wisconsin age in the adjoining Alaska Range. Large ice wedges also formed in the upper loess unit, but only their bases are exposed in the tunnel, and their history of development is uncertain.
Fanlike deposits of poorly sorted debris near the tunnel portal formed between about 12.5 and 11 ka during deep erosion of loess slopes under moister conditions. The deposits locally form two subunits: the younger over- whelmed a stand of tall willows on the floor of Goldstream valley between about 11.3 and 11.1 ka; the older may have formed about 1,000 yr earlier.
Stratigraphic records elsewhere in central Alaska indicate variable middle Wisconsin environments followed by colder and drier conditions that began between 30 and 25 ka and persisted until perhaps 12.5 ka. Widespread loess erosion and redeposition subsequently occurred under moister and probably warmer conditions. Renewed early Holocene loess deposition may have been widespread, but its exact environmental controls are uncertain.
Our data challenge three generally accepted concepts of late Quaternary periglacial processes in central Alaska. We contend that (1) many ice-wedge systems may have formed under interstadial conditions rather than full-glacial conditions, (2) episodes of rapid loess influx may have been partly out of phase with episodes of glacier expansion, and (3) redeposition of loess by solifluction, sheetwash, and gully formation may have been episodic and required conditions moister than those under which the loess initially accreted.