The North American mid-continent was transformed at the beginning of the Holocene by a regional drying:  dunes mobilized, rates of loess deposition increased, C₄ grasses increased in abundance, the eastern and northern prairie-forest borders shifted eastward, lakes levels dropped, and lake salinity levels increased.  This progressive drying provides an instructive model, although not a perfect analog, for the regional drying trends projected for the next several decades.  Here we synthesize and map a variety of vegetational, limnological, speleothem and aeolian paleorecords from the North American mid-continent in order to better understand 1) the timing and rapidity of aridity onset and 2) how different ecological and physical systems responded to this drying.  Paleovegetational reconstructions are based on quantitative reconstructions of fractional tree cover from fossil pollen records, and so improve upon the standard usage of percent arboreal pollen.
Results/Conclusions

In this synthesis, individual sites often abruptly change (<200 years) from a ‘wet state’ to a ‘dry state’ (e.g. from forest to prairie, C₃ to C₄, or increased salinity) but the timing of change is not synchronous.  Sites begin to dry out by the beginning of the Holocene (11.5 ka) and drying continues until 8 ka.  Some of this age range may be due to dating uncertainty, but even when the dataset is constrained to a subset of well-dated sites, it shows a range of ages with a peak in frequency around 8 ka.  This pattern of abrupt but non-synchronous changes is likely caused by the combination of 1) a regional drying trend that lasts several thousand years and 2) abrupt and non-linear site responses in which the tipping point for each site is determined by local hydrological factors.  The cluster of site-level changes at ca. 8ka might be caused by an accelerated rate of regional drying, caused by the collapse of the Laurentide Ice Sheet, but could also be caused by the clustering of site-level tipping points around some regional threshold.  Our results suggest that local responses to regional drying trends may be highly non-linear and that the location of local tipping points may be difficult to predict.