Scientists estimate average annual precipitation in Georgia was just 15 inches during the Last Glacial Maximum (~24,000-~19,000 calendar years BP). I’ve kept a rain gauge in my backyard here in Augusta, Georgia for 17 years, and I’ve carefully recorded precipitation. Average annual precipitation in my backyard was 47.8 inches over this time period. The driest year in my records was 2010 when just 29.5 inches of precipitation fell, and the wettest year according to my records was 2017 when 69.8 inches of precipitation fell. The difference between present day precipitation totals and LGM precipitation suggests the floral composition must’ve been considerably different then, and many modern day species of plants must’ve retreated to small refugia. However, there is no evidence of this, and in fact a couple lines of evidence indicate species abundant today were just as widespread during the LGM.
Genetic evidence shows that species common in deciduous forests occurred all the way to the glacial boundary during the Ice Age, despite pollen records indicating spruce forests dominated the landscape from north Georgia to the Ice Sheet. DNA studies of eastern chipmunks, red and sugar maple, shagbark hickory, beech, and yellow birch suggest they all ranged right up to the edge of the glacier which expanded all the way to southern Ohio. The ranges of at least 17 species of trees including persimmon, sweet gum, and river birch still reach their northern limits at the ghost boundary of where the massive glacier advanced. (See: https://markgelbart.wordpress.com/2017/03/30/the-ghost-boundary-of-the-last-glacial-maximum-ice-margin/ ) Although genetic evidence reveals discontinuities between populations of species, these appear to be caused by geographical barriers such as major rivers and the Appalachian Mountains. The genetic evidence suggests multiple diffuse refuges during the LGM for species that are common today. It seems like a paradox, but I have an hypothesis that can explain this for the Georgia piedmont. I call it my Abundant Oases Hypothesis, and it can probably be applied north of Georgia as well.
First, temperatures were much cooler during the LGM, so an average annual precipitation total of 15 inches would have gone much further then. Evapotranspiration rates were much lower, especially during summer. 15 inches of rain and snow may have been the equivalent of 25 inches in today’s climate–similar to an average drought year today. Second, Georgia’s piedmont soils are mostly clay, and they don’t drain as well as sandy soils. Water was held longer near the surface when it did receive precipitation. Third, the hilly terrain of the piedmont was a factor in contributing to oases where flora and fauna could flourish. Rainwater flowed down hills (this is known as colluvial flow), so the bottom areas between them hosted more plant life. Although most creeks dried up, there were plenty of areas at the bottom of hills where the water table came close enough to the surface to form intermittent springs. Beavers dammed springs, making them deeper and helping them hold water longer. Fourth, major rivers didn’t dry out completely and provided plenty of mesic refuge for species that could expand into oases during phases when annual precipitation increased.

Topographical map of the Piedmont National Wildlife Refuge in Jones County, Georgia. Note the hilly terrain and abundant creeks. During the LGM colluvial slope flow, intermittent springs, and lower temperatures helped reduce the negative effect of droughts.
My Abundant Oases Hypothesis is speculative. There are no pollen studies dating to the LGM from piedmont Georgia. There is 1 site in Winder, Georgia known as Nodoroc where pollen was collected that dates to just before the LGM when temperatures were warmer and precipitation was higher. Oak and pine were the dominant species, and they co-occurred with hickory, fir, and spruce. Beech, chestnut, birch, and maple were present. The shrub layer consisted of hazelnut and blueberry/rhododendron. There is no sediment dating to between 28,000 years BP-5000 years BP, suggesting land was eroding rather than accumulating sediment during this time period.
This is how I envision the Georgia piedmont landscape during the LGM. The tops of hills were covered with widely spaced shortleaf pine and post oak. These were slow growing and ancient because reduced CO2 levels in the atmosphere caused slow plant growth. Grass, herbs, thorny patches with prickly pear cactus, exposed boulders and rocks, and bare earth occurred between the trees. Following rare rain events, the ground burst into flower, but most of the year it looked dull and brown. Deep gullies, red from exposed clay, were common on the hillsides–a result of erosion that commonly occurred due to storms and a lack of topsoil. The bottoms of hills stayed green longer, and in some low areas hosted springs surrounded by marshy vegetation and deciduous woods of oak, maple, and beech. These are the oases of my hypothesis. Megafauna game trails connected these oases with each other, and the river systems where even more deciduous woodlands existed. This system of oases is what supported the continued existence of species that were able to expand when climatic conditions improved.
Reference:
Soltis, et al.
“Comparative Phylogeography of Unclaciated North America”
Molecular Ecology 2006