Supporting information: Weather underground: subsurface hydrologic processes mediate tree vulnerability to extreme climatic drought.

S1: Study site tree community composition, species distributional associations and study design.

S2: Datasets
S2.1 Species geographic distributions
Species distributions were derived using vegetative distribution data from the California GAP Analysis database (Davis et al. 1998). The database uses the California Natural Diversity Database, or “Holland” system to describe actual vegetation at the community level. Database polygons, ranging from a few (16.5 ha) to thousands (40672.9) of hectares (mean =1409.8 ha), may contain up to three plant community types (primary, secondary and tertiary based on percent cover). We limited our definition of a species distribution to polygons in which the species was identified as the dominant or co-dominant species within the primary or secondary cover type.

S2.2 USDA Forest Service Aerial Detection Monitoring Campaign
To identify areas of regional-scale dieback in blue oak and grey pine, we identified locations within the mapped distributions that were surveyed by the 2015 USDA Forest Service Aerial Detection Monitoring campaign (USDA Forest Service 2015), which is regularly conducted to quantify mortality in tree species across California. In these surveys, observations of tree species and dieback (areas of bare or brown canopy) are recorded from a low flying airplane. The plane flies along pre-selected flight lines with surveyors visually assessing tree species and canopy condition from either side of the airplane (up to 3.2 km in distance either side) using digital aerial sketch mapping systems to record dieback.
Because mortality cannot be confirmed in oak species without ground surveying, especially given their capacity for subsequent crownsprouting and basal resprouting, presence of dieback is not a definitive indicator of mortality. Surveyed areas with no dieback were considered “absence”. From these observations we compiled datasets of presence/absence of dieback within surveyed areas for the four tree species.

S2.3 USGS California geology classifications
We used only high-level (Lith1) geological classifications for analyses, combining unconsolidated and sedimentary categories and igneous and metamorphic categories. A full taxonomy of mapped lithologies is provided in Table S1.

S3: Sampling and sample analyses details
S3.1 Isotope analysis

S4: Results
S4.1: Regional results
S4.2: Survey subset results
S4.3: Full survey results