Tree mortality from a short-duration freezing event and global-change-type drought in a Southwestern piñon-juniper woodland, USA

Study area

The Chisos Mountains are a small rhyolitic mountain range located entirely within Big Bend National Park. Current forests are Pleistocene relicts, and their distributions are the product of species migrations from lowlands to uplands during early Holocene warming (Vandevender & Spaulding, 1979). The CM rise to 2300 m asl. They are bound at lower elevations by deserts dominated by shrub and succulent desert flora, where tree establishment and growth are inhibited due to high temperatures and moisture-limited conditions. The CM represent an ecological transition zone because of their position at the eastern edge of the Basin and Range Province and they share biological affinities with flora of the Rocky Mountains and the Sierra Madre Ranges (Muldavin, 2002). Soils are a mixture of mollisols and entisols. They are composed of moderately deep gravelly loam, which is well drained and non-calcareous (Carter, 1928). Runoff is moderate to rapid. Available water capacity is low.

Forests (above 1600 m asl) in CM are composed of piñon-juniper-oak, pine-oak, and mixed conifer woodlands. Piñon-juniper woodland is the dominant forest type which is comprised of Mexican piñon pine (Pinus cembroides Zuccarini), alligator juniper (Juniperus deppeana vonSteudal), gray oak (Quercus grisea Liebmann), Graves oak (Quercus gravesii Sudworth), Emory oak (Q. emoryi Leibmann), and weeping juniper (J. flaccida vonSchlechtendal) (Poulos & Camp, 2010). Lower elevations also contain small populations of one seed juniper (J. monosperma Englemann) and red berry juniper (J. pinchotii Sudworth) and oak shrublands that are dominated by Q. pungens Leibmann. Arizona pine (P. arizonica Englemann), Douglas fir (Pseudotsuga menziesii Mirbel), and Arizona cypress (Cupressus arizonica Greene) also have restricted populations in Boot Canyon in CM. Taxonomy follows Powell (1998).

The modern climate is arid, characterized by cool winters and warm summers. Precipitation is distributed bi-modally in late summer and winter with the majority of precipitation falling during summer storms as part of the North American Monsoon System. Mean annual precipitation for the Chisos Basin is 49.7 cm (range 10–135 cm). Mean January precipitation is 1.5 cm (range 0–2.5 cm) and is 8.0 cm (range 0.2–20.5 cm) in July. Mean monthly minimum temperatures are 1.8 °C in January and 17.0 °C in July. Maximum temperatures are 14.1 °C in January and 29.1 °C in July.

Field sampling

Thirty-six plots were established at low, middle, and high elevations (12 at each elevation) in the CM in June 2009 and I resampled them during the growing season in June 2012 after the drought. Low elevation plots were randomly placed in Green Gulch within 100 m of the edge of tree cover in CM. Middle elevation plots were randomly distributed across the Chisos Basin. High elevation plots were randomly distributed along the Southeast Rim. Plots were located so that they did not intersect trails, power lines, or archeological or cultural resources. The Southeast Rim was chosen for the high elevation sampling area because it had not previously burned in prescribed fires or wildfires. Trees > 5 cm diameter at breast height (dbh) were measured using 10 m radius (0.03 ha) fixed area plots. Seedlings (individuals < 5 cm dbh) were tallied by species in nested 5 m radius plots. Plot boundaries for both the tree and seedling plots were determined using a two-way ultrasonic rangefinder (Cptcam Inc., Shenzhen, China). The center point of each plot was marked with rebar and its location was recorded with a gps. Each tree was tagged with a uniquely numbered brass tree tag in 2009. I recorded the species, dbh, condition (live or standing dead), distance from the plot center and azimuth from north of each individual. Distance and azimuth measurements greatly assisted in relocating plot center. In 2012, plots were revisited and all trees from the 2009 inventory were resampled. Tree condition (live, recent snag, snag broken above dbh, snag broken below dbh, or clean snag) was noted. Trees lacking leaves or needles, with brittle and/or missing branches were classified as recent snags in the 2012 sampling interval. All recent snags were also checked for evidence of bark beetle infestation including presence of pitch tubes and beetle galleries.

Statistical analysis

I quantified differences in forest stand structure in 2009 and 2012 using linear mixed effects models to account for the repeated measures sampling design. I used the R Statistical Language (R Development Core Team, 2013) and the lme4 (Bates, Martin & Bin, 2012) and lmerTest (Kuznetsova, Christensen & Brockhoff, 2012) packages to perform linear mixed effects analyses of the temporal shifts in forest structure and species composition from the freeze and drought events. Timestep was designated as a fixed effect. Random effects were considered for the intercept, the sample plot, and the interaction of sample plot and timestep. The residuals of each model were inspected for deviations from homoscedasticity, and only models containing residuals without obvious deviations from normality were kept in the analysis. The final structure of the fixed-effects for each model was selected by sequentially dropping non-significant terms from the full model, by measuring changes in the significance of conditional F-tests for each term (Pinheiro & Bates, 2000). The intra-class correlation was also estimated for each model in order to assess the amount of variance in the response variable that can be attributed to the random effects in a model. The models describing the data most adequately were then selected using the Akaike Information Criterion (AIC) (Akaike, 1974). The significance of individual sites and site-year combinations was assessed after final model selection via the F statistic using the lmerTest package.

I used plots as the repeated sampling unit and the sampling year as the treatment representing pre- and post-drought sampling intervals. I compared tree basal area, live seedling and tree density by species, and differences in forest size structure for the two sampling years. I also used mixed effects models to investigate how the drought affected tree populations across the elevation gradient by evaluating changes in tree density and species composition in response to the drought. I evaluated the trend in tree mortality by size by performing a regression analysis comparing the percentage mortality at 1.0 cm size-class intervals.

Preferential mortality of small trees

With increasing tree size, mortality rate commonly decreases (Lorimer, Dahir & Nordheim, 2001; Palahi et al., 2003). The pattern of higher mortality of smaller trees in CM was consistent with the recent die off event of Pinus edulis between 2002 and 2004 Arizona, New Mexico, Colorado, and Utah, although Mueller et al. (2005) observed the opposite pattern during the 1996 and 2002 acute droughts in piñon-juniper woodlands of northern Arizona. My results in the CM were consistent with the trend of high seedling and sapling mortality under drought relative to larger trees that, with their deeper root systems and larger carbon stores, were able to survive those same drought events (Mendel et al., 1997; Mueller et al., 2005; Lopez & Kursar, 2007; Ganey & Vojta, 2011). The lack of evidence of bark beetle infestation in trees that died over the sampling interval also suggests that the high mortality of small-diameter trees was not related to insect attack.

Differential tree mortality by species

Although Mexican piñon pine is a site generalist in west Texas (Poulos & Berlyn, 2007), the increased mortality of piñon pine relative to other tree species was consistent with the patterns of recent mass tree mortality in the Southwest in 1996 and 2002 where piñon pine was more severely affected by drought than juniper (Mueller et al., 2005; Breshears et al., 2009). Junipers are typically more drought tolerant than pines in the American Southwest (Breshears et al., 2009; McDowell et al., 2008 but see Bowker et al., 2012). So while junipers in CM did experience significant mortality from the 2011 drought, they were probably less affected than the piñon pines because of their higher drought hardiness.

Emory oak was also significantly affected by the drought, and large stands of this species were completely killed in CM. Although southwestern oaks can survive over two months of severe moisture stress under experimental conditions (Poulos & Berlyn, 2007; Ehleringer & Phillips, 1996), little is known about the mechanisms of oak drought and freezing tolerance in the American Southwest (but see Neilson & Wullstein, 1985; Davis, Sperry & Hacke, 1999). Oaks in this region likely display considerable variation in drought and freezing tolerance, but their large tracheid diameters may have led to greater freeze-induced cavitation vulnerability relative to other tree species (Davis, Sperry & Hacke, 1999). Emory oaks experienced lower mortality than piñon pines and junipers in this study, yet, there remains a need for more information about the range of variability in oak drought tolerance mechanisms in the Southwest as they represent a major component of Madrean Sky Island systems.

Shifts in forest stand structure and species composition

Although the mortality event will undoubtedly provide new nesting sites for cavity-nesting birds in CM, the higher mortality of smaller trees, the loss of over half of the piñon pines in my monitoring plots, and the death of piñon pine and entire stands of Emory oak across all elevations could result in major shifts in forest stand structure and species composition. Since 2011, CM has moved out of the drought and is experiencing normal temperature and precipitation levels. The return to normal climatic conditions could have a positive effect on surviving trees by releasing them from competition for moisture and bolstering their survival potential in subsequent droughts (Bowker et al., 2012) since water use efficiency in piñon-juniper woodlands can be associated with stand density (Lajtha & Getz, 1993). Nonetheless, surviving trees in CM may have experienced permanent losses in xylem conductivity in 2011, which could result in delayed tree mortaility (i.e., Bigler et al., 2007) or predispose them to succumb to future acute droughts, especially if these events are coupled with winter freezes. While many piñon pines survived the 2011 drought, future global-change-type droughts could shift CM species towards dominance by junipers and more drought-tolerant oaks.

Mortality patterns across the elevation gradient

The pattern of increased tree mortality with increasing elevation was surprising and contradictory to other prior landscape-scale accounts of drought-induced tree mortality (Allen & Breshears, 1998; Gitlin et al., 2006; McDowell et al., 2008) and canopy dieback (D Schwilk, 2013, unpublished data). The increased tree mortality at higher elevations in CM is probably related to the southerly exposure of the high elevation plots that were located on mesas of the southeast rim at the edge of high elevation forest cover and the exacerbation of the drought effects by the February freeze-thaw cycle. While high elevations of CM are cooler and more humid than low elevations, the South Rim is exposed to high incident solar radiation due to its southerly aspect, as well as high winds and temperature fluctuations because it forms the southern edge of forest cover where the rim drops from 1981 m asl down to the desert floor. Higher elevations also probably experienced the lowest temperatures during the short-duration freeze event in 2011, although cold air drainage also contributes to low temperatures at low elevations (D Schwilk, 2013, unpublished data). This may have stimulated greater damage to high elevation trees through freezing-induced xylem cavitation in high elevation trees which may have let to higher mortality during the course of the drought.