Regeneration dynamics of Great Basin bristlecone pine in southern Nevada

The Great Basin bristlecone pine (Pinus longaeva D.K. Bailey) is the longest-living nonclonal plant, with some individual trees surviving more than 4000 years (LaMarche and Mooney 1972). Yet adults of this species have relatively thin bark and are considered sensitive to fire (Fryer 2004) and white pine blister rust, Cronartium ribicola J.C. Fisch. (Kinloch 2003; Vogler et al. 2006). To date, however, Great Basin bristlecone pine remains the only North American five-needle pine species with no confirmed incidence of white pine blister rust in wild populations (Bentz et al. 2017; Miller et al. 2017). As a subalpine tree, bristlecone pine is vulnerable to a warming climate as more competitive species move upslope. As a long-lived species, its regeneration is dependent on only a few seedlings surviving over the course of centuries to replace each adult. However, regeneration dynamics of bristlecone pine are poorly understood (Fryer 2004; Stritch et al. 2011), including seed bank dynamics, seed dispersal mechanisms and distances, and seedling microsite preferences. Anecdotal observations suggest that Clark’s nutcracker (Nucifraga columbiana (A. Wilson, 1811)) may play an important role in seed dispersal (Fryer 2004).

Great Basin bristlecone pine forms nearly pure stands with sparse understories at elevations above 2800 m in the Spring Mountains of southern Nevada, United States (Abella et al. 2012). In July 2013, the Carpenter 1 Fire burned 11 000 ha of subalpine forests in the Spring Mountains National Recreation Area (SMNRA) (Kallstrom 2013; Herrmann 2017), killing most trees within its perimeter. That recent disturbance provided an opportunity to examine regeneration dynamics of this slow-growing tree species with and without fire. This study reports the results of surveys for bristlecone pine seedlings along the entire elevation range of the species throughout the Spring Mountains, as well as in a large burned area 4 years after the fire. Our objectives were to characterize microhabitats and densities of bristlecone pine seedlings before and after fire and to understand the environmental factors contributing to seedling establishment. To our knowledge, this work constitutes the first quantitative comparison of pre- and postfire natural regeneration dynamics of Great Basin bristlecone pine (hereafter referred to simply as bristlecone pine).

In June and July 1992, we measured bristlecone pine regeneration along eight transects through unburned bristlecone pine forest and woodland in the Spring Mountains, generally to the west and northwest of the Mount Charleston townsite (Walker 1993). Bristlecone pine in this area is found at elevations from 2500 to 3500 m, which are occasionally associated with ponderosa pine (Pinus ponderosa Douglas ex P. Lawson & C. Lawson), limber pine (Pinus flexilis E. James), white fir (Abies concolor (Gordon & Glend.) Lindl. ex Hildebr.), or trembling aspen (Populus tremuloides Michx.) (Abella et al. 2012). Each transect, ranging in length from 100 to 720 m, was subdivided into continuous plots 6 m wide × 10 m long, beginning one plot below the lowest elevation of bristlecone pine occurrence and continuing upslope on a compass bearing until one plot above the upper tree line (always dominated by bristlecone pines) or until a tree-covered ridgetop was reached (total = 343 plots, 20 580 m²). The eight transects were positioned to represent the full range of aspects and elevations where bristlecone pines occur in the Spring Mountains, but they were also influenced by accessibility. Each contiguous 60 m² plot was considered an “extensive” plot, in which we measured aspect, elevation, and slope and the number, height, and diameter at breast height (DBH; breast height = 1.30 m) of every bristlecone pine individual, categorized as seedling (<50 cm tall), sapling (51–200 cm tall), juvenile (>200 cm tall, DBH ≤ 10 cm), or adult (DBH > 10 cm). We also estimated forest overstory cover in four categories (1 = no overstory; 2 = 1%–50% cover; 3 = 50%–90% cover; and 4 = >90% cover). The distance from each seedling to its nearest living bristlecone pine neighbor (of any size) was also recorded.

In one randomly located 20 m × 20 m “intensive” plot in each of seven of our eight transects, we counted all individual woody plants, tallied by species and size class (seedlings, saplings, juveniles, and adults), and tested for differences in soil conditions under adult bristlecone pines and in the open between canopies. We collected two 5 cm deep samples of mineral soil (in metal tins, following removal of surface organic matter) in each of the two habitats from five locations in each intensive plot (total = 140 samples). We also measured mineral soil depth (mean of 20 probes near each sample). After passing soils through a 2 mm sieve, we measured pH in a 1:2 soil:water paste and determined gravimetric soil moisture (percentage dry mass) by weighing samples before and after drying at 105 °C for 36 h. Soil organic matter was determined by mass loss on combustion at 550 °C for 4 h. We also estimated bulk density (in grams per cubic centimetre) using dry mass and volume of the sieved fines. This method represents an overestimate of actual field bulk density because pore spaces and rocks (14%–70% of volume) in these skeletal soils are not incorporated in the calculation (see discussion in Chapin et al. 1994).

In September and October 2017, we sampled areas burned by the 2013 Carpenter 1 Fire in the SMNRA west and southwest of Mount Charleston townsite, 1.5 to 8 km from areas sampled in 1992, focusing on the pure bristlecone pine forest above 3000 m in elevation (Supplementary Fig. S1¹ ). We recorded postfire bristlecone pine seedlings, defined as those with basal diameter < 5 mm and height < 12 cm, determined (on the basis of woodiness absence and few bud scale scars) to be newly emerged since the 2013 fire instead of fire survivors; this differs from our 1992 definition of seedlings. All five-needle pine seedlings were considered to be Great Basin bristlecone pine, as the burned woodland and surrounding forest consisted almost exclusively of this species, and all three excavations of seedling caches revealed ungerminated seeds that were <7 mm in length, which was too small to be limber pine, the other five-needled pine in the area. A total of 25 transects, ranging from 36 to 186 m in length, were positioned in any direction that maximized overlap with burned forest. Most transects ran along a contour, and most transects had <5% cover of postfire herbaceous vegetation.

In each transect, we looked for bristlecone pine seedlings in plots 6 m wide × 10 m long (total = 246 plots, 14 760 m²). At each transect end plot (n = 50) and in each additional plot containing seedlings (n = 21), we took the following measurements: Universal Transverse Mercator (UTM) coordinates, aspect, elevation, slope, distance to nearest adult tree (live or dead), distance to forest edge (using Global Positioning System (GPS) way points collected in the field and digitized live tree positions visible in postfire Google Earth aerial imagery (imagery date 8 September 2016), analyzed using QGIS version 3.10 (QGIS Development Team 2009)), and burn severity (low = most adult trees in the plot survived; moderate = most adult trees were killed but mostly scorched; and severe = no surviving trees, strongly charred). For plots in which no seedlings were found, elevation, aspect, slope, and distance to forest edge (green adult trees) were estimated by linear interpolation of values found in the nearest plots with measurements.

Although omitted from the 1992 survey, signs of white pine blister rust were searched for on seedlings and surviving trees encountered in and around the 2017 survey area. In both years, where multiseedling clusters were found, we recorded the number of seedlings per cluster.

Separate statistical analyses were conducted to assess environmental effects on seedling presence or absence, as well as seedling density, before and after the fire. The effects of elevation, aspect, slope, cover of mature trees, and distance to forest edge were analyzed as single and multiple factors for their effects on seedling presence or absence (where seedling caches were treated as one regeneration microsite location) at the plot level using logistic regression (LOGISTIC procedure; SAS Institute Inc. 2012). Aspect was evaluated separately (as difference from south, 180° azimuth, or difference from southwest, 225°) and in combination with slope to calculate potential incident radiation and heat load (McCune and Keon 2002). Burn severity effects on 2017 regeneration microsite presence or absence were determined using both ground-based descriptions of burn severity ranked on a three-point scale and satellite-based mapping of burn severity ranked on a four-point scale (see Supplementary Material A¹). As most plots had no seedlings, zero-inflated Poisson regression was used to test for seedling density response to aspect (both years) and to the distance from intact forest (2017) using the GENMOD procedure in SAS (SAS Institute Inc. 2012). We also calculated categorical means for stem densities by size class and environmental categories and Spearman rank correlations of seedling density with soil attributes (using the MEANS and CORR procedures, respectively; SAS Institute Inc. 2012).

The survey of unburned forest in 1992 encountered a total of 76 seedlings (<50 cm tall) over a sample area of 20 580 m², for an overall density of 36.9 seedlings·ha^–1. Adult trees were more abundant than seedlings or saplings in this forest (Fig. 1, inset). Bristlecone pine regeneration was concentrated in a few locations: 38 seedlings were found as the only ones in 38 plots, another 38 seedlings shared 16 plots with at least one other seedling, and 289 plots had no seedlings at all. In the extensive unburned plots, logistic regression of bristlecone pine seedling presence or absence showed a significant negative response to solar radiation (p = 0.0001), which was mostly due to the influence of aspect (p < 0.0001; Table 1). The strongest predictor of preburn seedling density was incident solar radiation (in megajoules per square centimetre per year), for which nonzero density (in stems per hectare) = e^{(5.7876 – 0.4429 solar radiation)} (Wald χ² = 43.27, p < 0.0001), with the odds of density = 0 being e^{(–1.5986 + 4.1081 solar radiation)} (Wald χ² = 15.45, p < 0.0001). Elevation, slope, and overstory cover had no significant value as predictors of seedling presence (Table 1). Only one multifactor logistic regression model was significant, in which elevation and overstory density both exhibited a negative influence on seedling presence, but the interactive effect of these two factors was positive (Table 1). Plots with seedlings were most likely to be encountered at lower elevations with denser tree cover and on steeper north-facing slopes (Supplementary Table S2¹).

In the unburned forest, nearest neighbors within 1 m of seedlings were three times more likely to be other seedlings (n = 15) than larger bristlecone pines (n = 5); at distances of >1 m, the nearest neighbors to seedlings were less likely to be seedlings (n = 5) than larger pines (n = 51). Nearest neighbor seedling distance (mean ± standard error) to other bristlecone pine seedlings was 1.03 ± 0.29 m (n = 21), distance to saplings was 2.42 ± 0.27 m (n = 22), distance to juveniles was 2.38 ± 0.45 m, and distance to adults was 3.01 ± 0.41 m (n = 24). Only one seedling cluster was found, with six seedlings within 5 cm of each other, which was suggestive of a seed cache. In the intensive plots in unburned forest, soils under adult bristlecone pines were deeper (F = 8.01, p = 0.005), wetter (F = 4.81, p = 0.030), had higher soil organic matter (F = 8.10, p = 0.005), and had lower bulk density (F = 20.28, p < 0.001) than soils away from bristlecone pine canopies, whereas pH did not differ significantly (F = 0.16, p = 0.689) (Table 2). Adult bristlecone pine trees were positively associated with wetter sites (ρ = 0.75, p = 0.03), whereas nonadults were positively associated with low pH soils (ρ = 0.71, p = 0.05).

In the burned forest surveyed in 2017, 72 new bristlecone pine seedlings (<12 cm tall) were encountered over a sample area of 14 760 m², for an overall density of 48.8 seedlings·ha^–1. No living saplings or juvenile trees were found in moderately or severely burned forest. No confirmed symptoms of white pine blister rust were observed on any bristlecone pine trees in the area. Most seedlings (n = 39; 54%) were found in lightly burned forest, with other plant cover averaging <1% overall. Most seedlings (n = 49; 68%) were found <100 m from the edge of the unburned forest, but some (n = 15; 21%) were found >300 m from living adult trees (Fig. 2; Supplementary Table S3¹). Seedlings were often found in clusters (30 of 72 new seedlings, in seven caches of two to nine seedlings each). Three of these clusters were excavated, and we confirmed that each consisted of individual seedlings, not sprouts from a single root crown (Fig. 3). Observations further suggest that cache locations are concentrated on south-facing slopes near ridge crests, where snow cover is potentially lower.

As in the preburn survey, postburn seedling presence was negatively related to heat load index (p < 0.0001), solar radiation (p = 0.0002), and southwest-facing aspects (p < 0.0001) (Table 3). Seedlings were generally found at lower elevations (p = 0.0036) and on steeper slopes (p = 0.0210) within the area sampled in 2017 (Table 3). Seedling occurrence was negatively associated with the burn severity as described in the field (p = 0.0001) but bore no significant relationship to burn severity based on satellite mapping (p = 0.6427) (Table 3). The distance of individual bristlecone pine seedlings to an adult tree (live or dead) averaged 2.47 ± 0.38 m. Distance to intact forest emerged as a significant predictor of seedling density only after the large number of zero densities was accounted for by the effect of aspect (Fig. 2).

Seedlings of Great Basin bristlecone pine were relatively rare in both burned and unburned habitats. Baker (1992) likewise found fewer seedlings and saplings than adults in five of six sites in Colorado that supported the related Rocky Mountain bristlecone pine, Pinus aristata Engelm. It is widely recognized that infrequent recruitment events and low densities can still be sufficient to maintain populations of long-lived trees (Platt et al. 1988; Wiegand et al. 2004). If the overall size class structure shown in Fig. 1 is stable, it suggests there are constraints that limit the recruitment of saplings into juvenile size classes.

Seedlings in the unburned forest, though found on diverse physical microsites, tended to be clustered (at least at the plot level) and constrained by factors associated with aspect and solar radiation (Table 1). This result may partially explain their observed affinity for the shelter of adult trees, where soil moisture, soil depth, organic matter (and presumably shade) were greater and bulk density was lower (Table 2). The role of adult trees in providing protective shade and soil enrichment may be more important at high elevations (Weiss et al. 2005), as indicated by the positive interactive effects of elevation and overstory density in the preburn forest (Table 1). Similar facilitation effects for Great Basin bristlecone pine seedlings were observed in the White Mountains of California by Maher et al. (2015) and for postfire recruitment of Rocky Mountain bristlecone pine by Coop and Schoettle (2009). Proximity to adult trees and a clustering of seedlings may also reflect dispersal limitations.

Aspect effects on seedling establishment also prevailed after the 2013 fire. In the postburn environment, heat load and aspect differences from southwest were slightly more important than incident solar radiation and aspect differences from south, which prevailed in the preburn forest. The greater importance of heat load avoidance in the distribution of seedlings sampled in 2017 may be associated with greater overall openness and warmer conditions than experienced in the 1980s and early 1990s. ClimateWNA (Wang et al. 2016) interpolations were performed for the 4 years prior to postburn sampling (2014–2017) and for 4- and 10-year windows (1989–1992 and 1983–1992) prior to preburn sampling. The 2014–2017 conditions exhibited mean annual temperatures that were >2 °C warmer than those of the earlier periods, but climate moisture deficit was estimated to be slightly less prior to 2017 than that prior to 1992.

With many plots hundreds of metres from the nearest living adult tree, distance to forest also emerged as a significant limitation to postfire bristlecone pine regeneration, though it was still contingent on aspect effects (Fig. 2). Coop and Schoettle (2009) likewise found that Rocky Mountain bristlecone pine regeneration was concentrated near burn edges and surviving adult trees. It appears that the legacy of even dead trees persists in the burned landscape, with most seedlings found within 3 m of live or dead mature trees. Whether the importance of these trees is through the provision of shade, less compact soils, nutrients, or perches inducing local disperser activity remains to be determined.

Our findings are consistent with an interpretation of the importance of animal dispersal in the regeneration of Great Basin bristlecone pine (Lanner 1988). Most seedlings in 2017 were found within 100 m of the forest edge, yet individual postfire seedlings were found >300 m from the nearest intact forest edge (Fig. 2). Although those distant seedlings were not identified as emerging from seed caches, it is possible that they were associated with either seeds that did not germinate or seedlings that had died. Many (30 of 72) of the postfire seedlings encountered were found in caches, which may have been created by Clark’s nutcracker or the endemic Palmer’s chipmunk (Neotamias palmeri (Merriam, 1897)), both of which were observed foraging in the study area. Clark’s nutcracker, in particular, is known to be an important disperser of other five-needle pines (Hutchins and Lanner 1982; Coop and Schoettle 2009). Palmer’s chipmunk is also known to consume and cache seeds (Hirshfeld 1975), or it may facilitate secondary dispersal in the process of pilfering nutcracker caches (Pansing et al. 2017). Clark’s nutcracker may also contribute to some of the microsite differences observed in Great Basin bristlecone pine regeneration, if indeed it preferentially caches seeds in some locations over others. The fact that more seed caches were found in postfire surveys (7 of 21 regeneration microsites) than in the wide-ranging prefire surveys further supports the interpretation that Clark’s nutcracker preferentially caches seeds in recently burned areas and that fires are important for five-needle pine forest renewal (Coop and Schoettle 2011). Although only one of the 54 unburned regeneration microsites was identified as a cache, some of the larger seedlings encountered in 1992 may have been the sole survivors of those emerging from seed caches.

Our extensive survey of 35 340 m² in sample plots concludes that aspect, the influence of adult trees (live or dead), and dispersal can all be important to the regeneration of Great Basin bristlecone pine. With sparse regeneration encountered 4 years after fire, we conclude that postfire recovery of Great Basin bristlecone pine forests can be a lengthy process but is gradually achieved through natural processes. The severe environment encountered at high elevations in semiarid regions accentuates the importance of microsite protection, moisture availability, and nutrients. Caching by birds or rodents can ameliorate some of the environmental challenges for Great Basin bristlecone pine regeneration by burying seeds, may reflect site selection in the caching process, and is important for forest regeneration after large wildfires. Protecting healthy populations of these dispersers is essential, and managers may wish to emulate some of the microsite selection patterns documented here and in related studies when undertaking restoration efforts.

Joanne Baggs and Espen Walker assisted with collection of the 1992 data, which was supported by the University of Nevada Las Vegas. Carla Burton and Billy Blanchar assisted with collection of the 2017 data, which was partially supported by a sabbatical from the University of Northern British Columbia. Kristen Waring (Northern Arizona University) and Nicholas Wilhelmi (U.S. Forest Service) reviewed photographs showing potential blister rust infection. We thank Kristen Waring, Alana Clason, Vern Peters, the Associate Editor, and two anonymous reviewers for constructive comments on the manuscript.