During her dissertation in historical archaeology at UC Berkeley, my wife, Cheryl Smith-Lintner, studied the communities and foodways of the Spanish, Mexican, and early American periods (1776-1870) in Northern California. Her research specialty, zooarchaeology, dealt with the bone remains of animals; through statistical and diagnostic analyses of these remains, she drew inferences about the procurement, processing, and disposal of food as well as the nonfood uses of animals (e.g., hide and tallow) among California's presidio and rancho era communities from the late 18th through mid-19th centuries. One aspect of Cheryl's research centered on the availability of and human access to animals; significantly, a number of factors—socioeconomic, political, and environmental—influenced the quantity and quality of animals available to these communities. One question that resonated—and continues to resonate—with my interests is how climate and climate variability impact access and availability.

The instrumental record of climate variability in Northern California dates back to ~1850, i.e. just after the beginning of the American period. On the other hand, the Spanish presence in California began some 75 years earlier; while instrumental observations are not available for the period 1776-1849, the historic record from the Spanish and later Mexican periods does contain observations, accounts, and inferences about climate variability in Northern California, particularly regarding dry and wet episodes. However, interpretation of such records can be difficult, given their sporadic and qualitative nature and the personal (and unknown) biases of the authors.

It is possible to infer climate variability from noninstrumental sources, including tree rings. The figure above illustrates a tree ring record, from Napa County (approximately 40 km northeast of San Francisco), of reconstructed precipitation variability over the period 1775-1924. Note the overlap with the precipitation rain gauge record from 1850-1924 and the strong correlation between them. During (and slightly after) the 75-year interval from 1775-1849, we have indicated on this plot information regarding "floods" and "droughts" extracted from historic accounts. The agreement between the tree-ring timeseries and the available historic records is highly encouraging: all years of "flood" correspond to positive annual-mean reconstructed rainfall, while all years of "drought" (except one) correspond to negative reconstructed rainfall anomalies. It is interesting that the exception, 1820, occurs near the end of a 9 year interval in which every year, according to the reconstruction, experienced above normal precipitation. This is, perhaps, an example of "weather memory bias", i.e. the observer's experience with the previous interval of above normal rainfall conditions may have led to a perception of dry conditions in 1820. It is worth noting that the historic accounts possibly reflect conditions as perceived over relatively short durations (especially for flooding episodes), while the tree-ring record represents conditions integrated over many months. The correspondence between the historic records and the reconstructed timeseries is thus sensitive to how representative the tree-ring record is of the possibly episodic events that may be present in the historic accounts.