Study area
The study sites are located in the SSPM National Park (31°37'N, 115°59'W) in north-central Baja California, approximately 100 km southeast of Ensenada, Mexico (Fig. 1). This area is identified as being in the North American mediterranean-climate zone (Minnich et al. 2000), although fire seasonality would suggest that the North American Monsoon System also influences this region (Skinner et al. 2008). There is no long-term weather information from the SSPM study area; average annual precipitation measured near our research sites (<5 km) with temporary weather stations on the northern plateau (2400 m elevation; 1989 to 1992) was 55 cm (Minnich et al. 2000). Climate data for the duration of the study is provided in Fig. 2 and is from the Potrero Remote Accessed Weather Station (NWS ID 045730), San Diego County, California, USA. We did not use weather data from the SSPM the temporal depth (length of record) was too short and they did not record snowfall (Dunbar-Irwin and Safford 2016).
Jeffrey pine dominates the conifer forests in the SSPM (Minnich et al. 2000). We quantified stand structure using two 4 ha stem maps from Fry et al. (2014). The SSPM granitic site (GRAN) is edaphically and structurally similar to portions of the eastern Sierra Nevada, Lake Tahoe Basin, and southern California mountains (Minnich et al. 1995, Stephens 2001, Barbour et al. 2002, Taylor 2004, Dunbar-Irwin and Safford 2016, van Wagtendonk et al. 2018; Fig. 2). Soils are shallow, well to excessively drained, and relatively acidic (Stephens and Gill 2005); chemical and textural properties are similar to typical granite-derived soils in comparable forests in California (Potter 1998). The SSPM metamorphic site (META), approximately 2 km north of GRAN, is a monotypic stand of Jeffrey pine with an understory of canyon oak (Quercus chrysolepis Liebm.) and peninsular oak (Quercus peninsularis Trel.). Soils are shallow and parent material is metamorphic quartz schist (R. Taskey, California Polytechnic State University, San Luis Obispo, California, USA, personal communication).
Fire is an important ecological process in Jeffrey pine-mixed conifer forests (Taylor 2004, North et al. 2009a), and occurred relatively frequently prior to the Park’s recent efforts at suppressing fires, with median fire return intervals of 5 to 13 yr, depending on the composite filter (Stephens et al. 2003). Road construction and limited summer and fall fire suppression efforts have increased fire intervals in most areas of the forest beginning in the early 1970s (Skinner et al. 2008). According to the fire-scar record, the last large wildfire was in 1946; smaller fires occurred in 1980 and 1962 at GRAN and META, respectively (Stephens et al. 2003). The SSPM has not been harvested for timber except for one 10 ha area that had overstory trees removed that is not near our research sites, although there is a history of livestock grazing at varying intensities, commencing in the late eighteenth century, to support the San Pedro Mártir Mission (Minnich and Franco 1998, Stephens et al. 2003).
Plot selection and mapping
At both sites, we used a 4 ha stem map on a uniform slope aspect and soils to describe forest spatial patterns and measure surface fuels. This scale included 1000 to 1500 trees, incorporating most of the stand-level spatial heterogeneity (Fry and Stephens 2010, Fry et al. 2014). In pine-dominated forests that once experienced frequent, low- to moderate-intensity fires, tree regeneration occurred in small openings (Stephens and Fry 2005, Larson and Churchill 2012), which were likely created by fine-scale canopy disturbances caused by fire, insects, disease, windthrow, or drought (Boyden et al. 2005). Using a grid of monument posts established within both 4 ha areas, we collected GPS coordinates and recorded species, status, diameter at breast height (DBH), and tree height and height to live crown base (live trees only) for every tree with DBH > 5 cm (Fig. 1). We used the stem map information in Spatial Analyst tool in ArcGIS v10 (ESRI, Redlands, California, USA) to calculate the total percent canopy cover, canopy cover by species, total basal area, and basal area by species for a 0.01 ha circular plot centered on surface-fuels measurement plots. We used this plot size to avoid multiple plots sampling the same tree. Overstory canopy cover was calculated using crown radius for each tree using species-specific allometric equations from Bechtold (2004) (see Fry et al. 2014 for description of canopy cover calculations).
Soil sampling and analysis at SSPM-META
We collected soils from the grid of monument posts at the META site, similar to previous work on the GRAN site (see Stephens and Gill 2005). Separated by approximately 25 m, each of the 27 soil samples were collected at a depth of 5 to 8 cm. Soil samples were sieved to <2 mm and coarse fragment content was determined on a weight basis. The air-dried moisture content was determined and all future measurements were corrected for soil moisture and were based on oven-dry weight. Particle size analysis was determined by the hydrometer method (Gee and Bauder 1986). Soil pH was determined in a 1:2 soil-to-solution mixture of 0.01 M CaCl2 (Kalra and Maynard 1991) using a glass electrode pH meter (Accumet 15, Fisher Scientific Hampton, New Hampshire, USA). Cation exchange capacity (CEC) was determined by the ammonium acetate (pH 7) method (Sumner and Miller 1996) and exchangeable bases in the ammonium acetate leachate were measured by atomic absorption spectrometry. Available phosphorus was determined by the Bray 1 (dilute acid-fluoride) method and a colorimeter. A subsample of each soil was ground in a ball mill to pass 60-mesh screen for total carbon and nitrogen determination by combustion gas analyzer (Kalra and Maynard 1991).
Fuel deposition and decomposition
Using the GPS-based monument posts that were used to develop the stem maps (Fry et al. 2014), we selected 30 posts (plots) that were systematically distributed throughout the area at each site to measure surface fuel deposition and decomposition. Mean distance between adjacent plots was 27 m (SD ± 2.6 m). We installed 0.12 m2 wire collection baskets (mesh opening of 2 cm2) to intercept fuels. This mesh size appeared sufficient for collecting canopy fuels, including male cones and white fir needles. At 10 of these plots, we measured woody fuels deposition by measuring material that fell within a 3 m radius (28.3 m2) area centered on plot center. Upon establishment, we removed all woody fuels from the immediate area and collected and measured accumulated fuels every year, between the last week in May to the first week in June from 2010 to 2013, replacing baskets disturbed by wildlife or fallen trees and branches.
We sorted all woody fuels by size class and type: 1-hour (0 to 0.6 cm branch diameter), 10-hour (0.6 to 2.5 cm), and 100-hour (2.5 to 7.6 cm), female cones (separated by current year’s drop [brown] and greater than one year old [grey]), needles, and other (i.e., male cones, bark, cone scales, and hardwood leaves). All fuels were weighed in the field, and a subsample of each category was transported to the laboratory, oven dried at 100 °C for 24 hours, and weighed. We used the ratio of dry weight to wet weight of the subsampled fuels to calculate moisture content of each fuel class. Our values were similar to the previously established estimates of 4, 6, and 8% moisture content for 1-, 10-, and 100-hour fuels, respectively, so we used these values to derive the dry weight of all fuels (Stephens 1998). For analytic purposes, we considered litter as 1-hour fuels and cones as 10-hour fuels.
At these same 10 plots, we used litter bags to estimate the rate of decay for Jeffrey pine needles. The bags, made of nylon window screen (about 225 cm2) with a mesh size of about 2 mm, sewn closed with fishing line, held a field-measured 50 grams of recently dead needles collected from nearby live trees. We laid the five litter bags on the forest floor near the litter baskets, and then anchored them to the plot post. Bags were installed in 2009 and collected at the same time as described above in 2010, 2011, and 2015. Decomposition was measured over six years by collecting one bag from each plot, drying in an oven at 50 °C for 24 hours, then calculating weight prior to submitting a subsample for chemical analysis at a soil processing lab.
Analyses
For all surface fuels, soil characteristics, and pine needle decay, we summarized mean plot-level values by year and site. To determine the potential for spatial dependence, we tested each variable for spatial autocorrelation using Moran’s I test (Moran 1950). Moran’s I is a weighted correlation coefficient that detects departures from spatial randomness, and ranges from +1 (positive correlation) to −1 (negative correlation). For surface fuels, we examined how a set of variables varied among sites, years, and with forest structural characteristics (plot-level canopy cover, basal area, and tree density). The surface fuel variables (Mg ha−1) tested included total weight in litter basket; litter weight in the litter basket; total fuel weight on the surface plots; total cone weight on surface plots; and 1-hour, 10-hour, and 100-hour fuels on the surface plots. All response variables were normalized using a square root function. We collected data on 1-, 10- and 100-hour fuel loads within the litter baskets, but these data were highly zero-inflated and skewed by a few plots in which woody fuels happened to fall within the baskets, so we did not analyze these variables. Total fuel loads in baskets across our study sites were primarily driven by litter accumulation rather than woody fuels.
We tested for a significant effect of site (GRAN versus META) on total fuel loads in both baskets and on surface plots, for each year, using a simple linear mixed-effects model,
$$ y=B\ast site+\left(1| year\right)+E $$
(1)
where y is the response variable of interest (basket- or plot-based surface fuel load), B is the fixed effect of site, and (1|year) is a random intercept for year, allowing for total fuel loads to vary among years, and E is the random error assumed to be independently distributed with a normal distribution. We found no significant effect of site on any surface fuel variable, so for subsequent analysis of forest structure effects, we combined the two sites. We tested the effects of local canopy cover, basal area, and tree density (0.01 ha circular plots) on the surface fuel variables described above, using a simple linear model in which each year was analyzed separately.
To compare the change in pine needle weight over time, we used the exponential decay function (e), which quantifies decay of organic material for longer decay processes, and the parameter k may be used to compare decay rates (Keane 2008),
$$ \frac{A_{\mathrm{t}}}{A_0}={e}^{- kt} $$
(2)
where A is the amount of material at time zero (A0) and at time t (At). We fit a curve defined above to each plot with at least three data points (initial and two subsequent measurements) using non-linear least squares estimation in R (R Development Core Team 2014). The fitted value of k can therefore be used to calculate the percent change in needle decay between any two given years, and to compare decomposition in the SSPM to other studies that have also calculated k within different systems. Initial values of needle weight were determined by sampling recently dead needle material from 5 trees per site in 2009.
We modeled variation in decay rate k as a function of site, using the k parameter value calculated for each plot across the six years, using a simple linear model. As site did not have a significant effect on k (see Results), we pooled both sites and tested for effects of forest structure on decomposition rate using a simple linear model of k conditional on canopy cover, basal area, and density.