Past climatic conditions on the Baja California peninsula and choice of GCM
We consulted geological literature to determine which GCM was best supported by paleoprecipitation proxies, which included alluvial fan, pollen, lake core record, and midden data across southern California and the BCP (Supporting Information). Unlike global climate, literature indicates that the BCP was generally wetter than present at the LGM (Lora, 2018; Lyle et al., 2012), culminating with an El Niño-like wettest period during the Late Pleistocene-Holocene transition ~14.5-10 ka (Antinao & McDonald, 2013). This was driven by strengthening East Pacific (westerly) precipitation that affected mainly southern California and the northern BCP in the form of winter rain and less seasonality (Antinao & McDonald, 2013; Antinao et al., 2016). On the other hand, while the south would have been drier overall after LGM, there is evidence from alluvial fan deposits and offshore cores that despite less precipitation, the precipitation came as stronger, and possibly more frequent, tropical cyclones reaching the southern BCP (Antinao et al., 2016). The Pacific High weakened ~17-11 kya and allowed warm tropical waters to move northward, partially strengthening the North American Monsoon (NAM) and leading to increasingly wetter conditions in the south comparable to what is observed today (Antinao et al., 2016). After that, the southerly NAM was stable but weaker, contributing less precipitation at least in part due to cooler Gulf of California sea surface temperatures 11–8 ka and a limited northward progression of those storms due to presence of the Pacific High.
Considering total annual precipitation (bio12), CCSM shows drying in the southern peninsula and increased wetness in the north (Figure 1b), which is in agreement with the paleoprecipitation data, whereas MIROC shows substantially wetter conditions throughout the peninsula (Supporting Information). Furthermore, the precipitation of the wettest month (bio13) shows a drying in the CCSM models in the south and slightly increased precipitation in the north, but the MIROC model shows wetter conditions throughout (also evident in bio18). CCSM shows a slight strengthening of seasonality (bio15), while MIROC shows patchy decrease of moderate magnitude, the latter of which is more in line with the paleoprecipitation proxy data. Despite this last observation, we chose the CCSM GCM as the overall best model for our analyses because it appears to be more consistent with paleoprecipitation data particularly as it relates to the increased precipitation in the north and decreased effect of NAM precipitation and tropical storms in the south.