Seedling Growth at Paired Fire Sites
To test whether seedlings have increased growth in recently burned sites, I measured vegetative production represented by terminal stem diameter and number of new needle bundles produced by whitebark pine seedlings in the Smith Creek 1880 burn and the adjacent Smith Creek 1988 burn. These two burn sites were located approximately 15 km west of Stevensville, Montana, USA, at 46° 28′ N, 114° 15′ W. The site elevation is 2100 m to 2150 m. Both sites burned in 1880, with a partial re-burn in 1988 up to a constructed fire line separating the two survey sites. Both burn sites had similar soils. The 1880 burn site had generally east- to southeast-facing aspects, with 0 % to 24 % slopes. The 1988 burn site also had east- to southeast-facing aspects, although more of it was southeast facing than in the 1880 burn site. Slopes in the 1988 burn were 1 % to 17 %. The 1988 burn was a high-severity burn, which removed all soil organic matter down to mineral soil, and killed all trees in the study area. The severity of the 1880 burn is uncertain, due to its age and lack of records. At the time of sampling, the 1880 burn site had a mature canopy consisting of subalpine fir, lodgepole pine (Pinus contorta Douglas ex Loudon), whitebark pine, and an occasional Englemann spruce. Whitebark pine regeneration was occurring beneath this canopy, with individuals ranging from <0.5 m in height to small trees 2 cm to 8 cm diameter at breast height (DBH) in light gaps. Although whitebark pine growth can be severely suppressed beneath a canopy (Arno and Hoff 1990, Keane et al. 1994, Campbell and Antos 2003), trees that are <0.5 m in height are unlikely to have been present prior to the 1880 fire, and so they were considered to be postfire regeneration. Blister rust was present on this site, but whitebark pine cone production still occurred.
I made vegetation production measurements on 23 seedlings per burn site in June 2000 consisting of: 1) terminal internode stem diameter from the top of the seedling, taken at the top internode and representing growth in 1999; and 2) number of new needle bundles in the top whorl of all branches, representing growth in 2000. I considered all trees <0.5 m in height to be seedlings for this study. I also sampled leaf tissue, from needle bundles produced in 1999, from each of these seedlings to measure total N content. The single, previous year measurement of stem diameter was used rather than overall height, since trees of the same height differed greatly in age between the 1880 and 1988 burns. Internode distances were tightly clumped in many of the seedlings in the 1880 burn, making it difficult to ascertain numbers of nodes and distances between them.
To control for possible confounding influences from adjacent understory plant effects, and to test for potential effects, I sampled seedlings growing with different common understory plants, as well as those growing on bare ground within the 1988 burn. Within the species associations, sampling was random. The associated species and sample sizes within the 1880 burn were: beargrass (Xerophyllum tenax [Pursh] Nutt.), n = 8; grouse whortleberry (Vaccinium scoparium Leiberg ex Coville), n = 10; and Hitchcock’s smooth woodrush (Luzula glabrata [Hoppe ex Rostk.] Desv. var. hitchcockii [Hämet-Ahti] Dorn), n = 5. Within the 1988 burn, the associated species and sample sizes were: beargrass, n = 4; grouse whortleberry, n = 9; bare ground, n = 10. No whitebark pine seedlings occurred adjacent to Hitchcock’s smooth woodrush within the 1988 burn, and fewer occurred adjacent to beargrass in the 1988 burn than in the 1880 burn. Due to the severity of the 1988 burn, much of the ground surface was bare, whereas vegetation covered nearly the entire 1880 burn area ground surface. To minimize the effect of size differences on growth rates, I selected seedlings 25 cm to 30 cm in height with three to five whorls of side branches per seedling, within each burn site.
Experimental Seed Planting
The effects of fire on natural whitebark pine regeneration are confounded by blister rust reduction of seed availability (Tomback et al. 1995) and Clark’s nutcracker caching preferences (Tomback et al. 1993). To test the effects of fire on whitebark pine seedling recruitment, survival, and growth without these confounding factors, I performed a seed planting experiment at Beaver Ridge Experimental Site, Bitterroot Mountains, Idaho, USA (Keane and Arno 2001, Keane and Parsons 2010b). As part of a larger whitebark pine natural regeneration study, a portion of this site was prescribed burned to mimic a wildfire in September 1999, with fuel added to increase burn severity (R. Keane, RMRS Forestry Sciences Lab, Missoula, Montana, USA; personal communication). The fire severity was sufficient to kill nearly all trees on the site and much of the understory species aboveground plant material, and produced patches of hydrophobic soil. A large area of nearby mature forest, last burned in 1910, served as an experimental control (Keane and Arno 2001). No records exist of the 1910 fire severity on this site but, at the time of the experiment, a mature canopy of whitebark pine, subalpine fir, and Englemann spruce covered much of the site, interspersed with open patches of varying size. The understory plant community was well developed, dominated by common subalpine species including beargrass, grouse whortleberry, elk sedge (Carex geyeri Boott), and Hitchcock’s smooth woodrush. The proximity of the two fire ages at the same site provided me with the opportunity to study fire effects while minimizing other site effects. This site is located at an elevation of 2050 m to 2200 m, along a south-facing slope. Soils are granitic, derived from the Idaho Batholith. The site receives heavy winter snowfall and is usually snow free from early July through early October (R. Keane; personal communication).
Two burn treatments were used: high-severity burn (prescribed burn with slash fuel added, September 1999) and unburned control (last fire occurred in 1910). For simplicity, these treatments will be referred to as burned and unburned. To control for understory plant species effects, I planted whitebark pine seeds next to four major understory plant species, with 10 plots of each per burn treatment: grouse whortleberry, beargrass, elk sedge, and Hitchcock’s smooth woodrush. I selected these species because they showed either positive (whortleberry and woodrush) or negative (beargrass and sedge) association with whitebark pine seedlings at vegetation survey sites (J.L. Perkins, University of Montana, Missoula, Montana, USA, unpublished data). In the burned site, I located an additional 10 plots in patches of bare ground. I established a total of 90 1 m × 1 m plots, with 50 plots in the burned area and 40 plots in the unburned control. Plots were located along two bands, perpendicular to the slope, in parallel with each other and approximately 50 m apart in elevation. Plant association plots were spread evenly across the sites. Plots were located away from bases of mature trees and snags to minimize effects on light availability. Mean overhead canopy of unburned plots did not differ significantly from that of burned plots, as measured with a concave spherical densiometer (t-test: mean canopy <2 %, P = 0.446, n = 90). However, unburned plots received less light on a full-day basis than burned plots due to morning and afternoon shade from neighboring trees. Average overall understory plant canopy within the burned plots was 33 %, and within the unburned plots was 75 %. Plots were protected from rodent predation by hardware cloth exclosures following the design of McCaughey (1990).
I planted a total of 3330 whitebark pine seeds (37 seeds per plot) on 27 and 28 June 2001. The whitebark pine seeds came from seed lot WBP-14-7425, collected at 2800 m elevation from Union Pass, Shoshone National Forest, Wyoming, USA, in September 1999. The distance of the seed source from the experimental site might have affected overall germination and seedling success, but did not prevent testing of treatment effects within the site. Seeds were processed by the USFS-Coeur d’Alene Nursery, Coeur d’Alene, Idaho, USA, following the procedures of Burr et al. (2001), and remained in cold storage at the USFS-Coeur d’Alene nursery until preparation for planting. Whitebark pine seeds have delayed embryo maturation, and usually require two to three cold-warm stratification cycles to reach maturity (Tomback et al. 2001). The seeds used in this experiment were x-rayed at the nursery to check that embryo development was sufficient for planting (K. Burr, USFS-Coeur d’Alene Nursery, Coeur d’Alene, Idaho, USA; personal communication). Seeds are normally cached by Clark’s nutcrackers in late August (Tomback 2001), but the unusually long fire season of 2000 followed by early October snows curtailed site access and prevented fall seed planting. To compensate for the delayed planting date, I provided seeds with one warm stratification, following the guidelines of Burr et al. (2001), in September 2000, then returned them to cold storage for the 2000–2001 winter. I re-warmed the seeds and placed them in a 48 hr running water soak immediately prior to planting. I planted seeds within or immediately adjacent to target understory plant species, at 2 cm depth to mimic nutcracker cache depths (Tomback 2001). Due to the delayed planting date (27 and 28 June 2001, after snowmelt), I simulated snowmelt conditions by hand-watering plots with on-site spring water every three days for the first two weeks after planting. I placed two one-gallon plastic water jugs, with pin-prick holes along their bottom edges, on the uphill side of each plot to provide slow water seepage into the plots between hand-watering.
I monitored seedling emergence and survival weekly for the first month after planting, and every two weeks for the remainder of the first growing season and through the second growing season (mid-July through early September 2002). No new germination occurred during 2003, but I continued to monitor survival monthly through the summer growing season. I marked seedlings with color-coded toothpicks and located them on plot maps upon their emergence in order to monitor survival. I calculated survival as total surviving seedlings relative to total emergence. Emergence included all seed germinants, even if they produced no more than cotyledons.
Seedling Growth and Nutrient Concentrations
At the end of the experiment, I removed all remaining surviving seedlings on 25September 2003 by digging a plug 20 cm in diameter and 35 cm deep around each seedling and lifting it from the ground, trying to keep the roots intact. Due to the rocky nature of the soil and fragility of the roots, some root damage occurred on most of the seedlings, but the bulk of the roots were removed intact. I kept seedlings refrigerated in their soil plugs until the roots were cleaned. After cleaning, I clipped seedlings at the top of the root mass to separate above- and belowground biomass, and dried each for 48 hr at 80 °C. I weighed aboveground and belowground biomasses separately. For plots with more than one surviving seedling at harvest, I used averages of all seedlings per plot for biomass quantification and leaf N analysis.
After weighing, I separated all dried seedling needles from fascicles and ground them to a fine powder with steel ball bearings in microcentrifuge tubes placed on a paint shaker. I weighed powdered needle samples of 3 mg in tin capsules on an analytical balance, and sent them to the UC Davis Stable Isotope Facility for total N and δ15 N analyses.
Soil Nutrients
I assessed soil NH4
+, NO3
−, and available P at the Beaver Ridge experimental site using ionic resin capsules (Unibest, Bozeman, Montana, USA) buried (one per plot) in mineral soil at 5 cm depth for one year to capture nutrients moving through by mass flow and diffusion (Binkley and Vitousek 1989). I installed one capsule per plot, in both burned and unburned treatments, on 16 July 2001 and removed them on 4 August 2002. I extracted ionic nutrients within 24 hours of removal by three successive 30 min agitated rinses with 2 M KCl (Kjønaas 1999, Morse et al. 2000). I centrifuged the decanted KCl extract from the three successive rinses at 3000 RPM for 10 min. I measured NH4
+ nitrogen and NO3
− nitrogen using a Technicon Autoanalyzer™ II single-channel colorimeter system (Technicon Instruments Corp., Tarrytown, New York, USA) following standard methodology for this instrumentation. I measured PO4
−3 phosphorous by the ascorbic acid colorimetric method (Page et al. 1982).
Statistical Analysis
I analyzed seedling vegetation production differences between the Smith Creek 1880 burn and 1988 burn survey sites with t-tests, and plant association effects within burn sites using one-way ANOVA. I analyzed leaf nitrogen differences with a Kruska-Wallis test due to violation of homogeneity of variances for my data. I used one-way ANOVA, with plant association as the main factor, to analyze plant association effects on emergence, aboveground growth, and total leaf nitrogen in the burned plots at Beaver Ridge. By the time seedlings were harvested, only six seedlings remained alive in the unburned treatment, and due to both low germination and low survival, plant species effects within the unburned site could not be determined. Only one seedling that germinated in 2001 remained alive at harvest, and was not included in biomass calculations to avoid confounding the data. All biomass means are for seedlings germinating in 2002. I pooled data from all plots within each of the burn treatments and used t-tests to test the effects of burn treatment on total number of seedlings emerged, aboveground biomass, total plant leaf nitrogen, and nutrients. I analyzed survival, belowground biomass, and percent leaf nitrogen of whitebark pine seedlings at Beaver Ridge using a Kruska-Wallis test, because survival and belowground growth violated assumptions of variance homogeneity. I normalized soil nitrate and available phosphorus data with ln transformations. I used t-tests to compare burn treatment effects on soil nutrients, and two-way ANOVA, with burn treatment and plant association as main factors, to test plant association effects. For all statistical tests, I adopted an α of 0.05 as my basis for mean separation.