Dryland forests face multiple climate change related threats, including more frequent droughts and shifts in the seasonality of precipitation. While dryland tree species tend to have adaptations for adult persistence during drought, recruitment is often episodic, occurring only in years with favorable conditions. As conditions become more arid, with fewer favorable periods for seedling establishment, variation in the presence of drought-adapted phenotypes among populations may determine recruitment potential and lead to different outcomes. We used a four-year common garden experiment to evaluate seedling survival and growth in response to varying precipitation timing and amount for 23 populations of Pinus monophylla, a tree species occurring in semiarid forests of southwestern United States. Previous research on P. monophylla showed that intraspecific phenotypic variation in maternal trees persists in seedling offspring, yet fitness implications are unknown. We sowed seeds at the dry edge of the species’ climatic distribution and used experimental watering treatments to simulate four seasonal precipitation regimes: ambient control (drought), spring supplemental watering, summer supplemental watering, and spring + summer supplemental watering. We used a randomized block design, sowing seeds under the canopy of the native shrub Artemisia tridentata to meet known shade requirements for seedling establishment. Seed source climate was associated with differences in recruitment. Seedlings sourced from more arid climates had higher survival and aboveground growth than seedlings from wetter climates under all simulated precipitation regimes. Climate transfer distance relationships were consistent with local adaptations to mean annual temperature and spring water availability, and supplemental summer water increased seedling survival. We found large differences in seedling establishment between sowing years, highlighting the importance of episodic recruitment under favorable conditions. Nurse shrub canopy cover was also a strong predictor of survival in the first two years of establishment. Our results show that intraspecific phenotypic variation mediates the effect of seasonal drought on dryland tree recruitment, suggesting that different populations may respond uniquely to changes in climate. The effects of precipitation may be contingent on shifts in seasonal timing and amount, and P. monophylla seedling survival may decline with increasing temperatures or decreased summer water availability.

Seed source locations: Seeds were collected in September 2019 from 23 populations of P. monophylla. We selected nine mountain ranges from across the species’ distribution that spanned regional gradients of mean annual precipitation and proportion of summer precipitation (July, August, September) to annual precipitation using 30-year climate means from PRISM (Daly et al., 1994). At each mountain range, we aimed to locate three seed collection sites along the elevational gradient of P. monophylla, ranging from the lowest- to highest-elevation stands of cone-bearing trees that were accessible from a road. Three mountain ranges had fewer than three sites, due to narrow woodland elevation ranges or a lack of available cones.

Common garden location: Our common garden was located near Carson City, NV, USA (39.0729°, -119.784°), at an elevation of 1448 m. The common garden location receives a mean of 245 mm of annual precipitation and has a mean annual temperature of 10.1 °C (30-year mean from 1991–2020; Daly et al., 1994). Compared to the seed source locations, the common garden’s climate is warm and has low precipitation, especially in the spring and summer.

Experimental design: We used a randomized block design, sowing filled seeds in the soil under the canopy of the native shrub Artemisia tridentata. Under the north side of each shrub canopy, we sowed one seed from each of the 23 populations in a randomized 5 × 5 grid (45 cm × 45 cm; 0.2 m²). We sowed seeds in two successive years to test whether population differences in recruitment were contingent on weather conditions in the study year. We sowed 288 blocks (6,624 seeds) in November 2019 (“cohort one”) and 128 blocks (2,667 seeds) in November 2020 (“cohort two”). In cohort two, only 22 of our 23 populations were included due to limited seed availability for one seed source.

Watering treatments: We used experimental watering treatments to simulate four seasonal precipitation regimes, which were designed based on the observed extremes of precipitation seasonality from the seed sources. The “control” treatment (W1) had no added water and thus experienced ambient, dry conditions of the common garden site. The “spring” watering treatment (W2) received additional water weekly from April through June. The “summer” watering treatment (W3) received additional water weekly from July through September. The spring + summer (“both”) watering treatment (W4) received additional water biweekly from April through September. Each watering addition simulated a rainfall event of approximately 10 mm. Treatment precipitation amounts used in analyses reflect the actual amount of water received.

Environmental data (extracted from geospatial datasets): Environmental conditions were obtained for the seed source locations and the common garden experiment site. We used 30-year means (1980–2009) for water balance variables (annual cumulative climate deficit and vapor pressure deficit), which were extracted from AdaptWest (Dobrowski et al., 2013). Available water capacity, which represents the amount of water the soil can make available for plant use, was extracted from POLARIS (Chaney et al., 2019). Temperature and precipitation data were extracted from PRISM (Daly et al., 1994) and summarized into the following variables: mean and minimum annual temperature, annual precipitation, and spring (April–June) and summer (July–September) precipitation. We used 30-year means (1991–2020) to characterize climate regimes in each location and monthly data for seasonal conditions during the study period (2020–2023).

Environmental data (measured): We installed environmental measuring equipment to monitor local conditions at the common garden site. A weather station (ATMOS 41; METER Group, Inc.) was mounted at a height of 2 m, collecting hourly data on air temperature, precipitation, solar radiation, and relative humidity from March 2020 through July 2023. “Precipitation” for the supplemental watering treatments was calculated as the observed ambient precipitation plus the water addition. We estimated canopy cover for each nurse shrub using hemispherical photographs. Two photos were taken near midday from perpendicular angles from the center of each seedling block. Percent shrub cover was calculated using the ForestCrowns program (Winn et al., 2016), and the average value from the two images was used for each seedling block.

Percent shrub cover was calculated using: Winn, M. F., Palmer, A. J., Lee, S. M., & Araman, P. A. (2016). ForestCrowns: a transparency estimation tool for digital photographs of forest canopies. e-Gen. Tech. Rep. SRS–215. Asheville, NC: U.S. Department of Agriculture Forest Service, Southern Research Station. 10 p.