Effects of nurse shrubs and biochar on planted conifer seedling survival and growth in a high-severity burn patch in New Mexico, USA
Data files
Apr 03, 2023 version files 248.03 KB
-
biochar_shrub_cuml_diam_CHM_na.csv
-
biochar_shrub_cuml_height_CHM_na.csv
-
biochar_shrub_survival.csv
-
README.md
Abstract
The synergistic effects of widespread high-severity wildfire and anthropogenic climate change are driving large-scale vegetation conversion. In the southwestern United States, areas that were once dominated by conifer forests are now shrub- or grasslands after high-severity wildfire, an ecosystem conversion that could be permanent without human intervention. Yet, the reforestation of these landscapes is rarely successful, with a mean planted seedling survival of just 25 %. Given these low rates, we carried out a planting experiment to quantify the impacts of biochar as a soil amendment and shrubs as nurse plants on planted conifer seedling survival and growth following high-severity wildfire. We planted 1200 seedlings of three species (Pinus ponderosa, P. strobiformis, and Pseudotsuga menziesii) in a 2-ha area within the footprint of the Las Conchas fire in New Mexico, USA. We used four treatments: under shrubs, or in the open and with or without biochar in a full-factorial design. We found that planting tree seedlings underneath shrubs increased tree seedling survival by 46 % after 3 years, with some marginal evidence that shrubs inhibited seedling diameter growth (mean R2 = 0.08). The addition of biochar increased seedling survival by 11 % but had no effect on seedling growth. Our study suggests that planted seedling survival in post-wildfire areas can be increased by planting under shrubs in soil amended with biochar. The widespread adoption of these methods may improve the success rates of post-wildfire reforestation efforts in semi-arid areas, regaining some of the ecosystem services lost to high-severity wildfire.
Methods
In 2011, the Las Conchas fire burned approximately 63,130 ha in the Jemez Mountains of New Mexico, USA, ∼30 % of which was high-severity as defined by Eidenshink and colleagues, (2007). The Las Conchas fire burned over footprints from five previous wildfires (Coop et al, 2016). This repeated high-severity burning has led to a large-scale vegetation shift, from ponderosa pine-dominated forest to a Gambel oak (Quercus gambelii) and New Mexico locust (Robinia neomexicna) shrub-dominated landscape (Coop et al., 2016; Kaufmann et al., 2016).
We selected a 2 ha area within the footprint of the Las Conchas fire for our planting experiment (lat: 35.785, long: -106.415), due to its location within an area classified as having experienced high-severity wildfire, its elevation, 2520 m above sea level, (close to the mean elevation of the Las Conchas fire footprint, 2541 m) and proximity to a road to enable planting efforts. We planted all 1200 seedlings during July 15–17 of 2019, between 2516 m and 2524 m in elevation, with the maximum distance of 174 m between seedlings to ensure climatological and meteorological influences on seedlings were similar. 400 seedlings of each species (ponderosa pine, southwestern white pine, and Douglas-fir) were planted equally across four treatment types; in the open, in the open and with biochar amended to the soil, under shrub canopy, under shrub canopy and with biochar amended to the soil, so that 100 seedlings of each species were in each treatment type. Seedlings planted in the open treatments were ≥ 2 m from the nearest shrub, and planted in grids of nine seedlings, three individuals of each species planted 30 cm apart. We co-located the grids of individuals of each treatment (open, open with biochar) to minimize potential microtopographical variance which may impact experiment results (Marsh et al., 2022b). Seedlings planted beneath shrubs (shrub, shrub with biochar) were planted in groups of six (one individual of each species and treatment combination), each 30 cm apart, so that they were subject to similar conditions (Supplementary Fig. 1). All seedlings in the “under shrub canopy” were planted beneath Gambel oak shrubs and were located far enough from the edge to remain in shade. An equal number of seedlings were planted by three researchers using dibble bars to reduce the potential influence of different planting techniques (Pinto et al., 2011a).
The soil at this site is classified as “very paragravelly sandy loam” to two inches in depth, “very paragravelly sandy clay loam” from 2 to 5 in. in depth and “very gravelly sandy clay loam” from 5 to 13 in. of depth (SSS, 2023), with low nitrate-nitrogen levels, possibly due to soil run-off after wildfire (Ebel et al., 2018). In high-severity burn areas, carbon was measured as 10.37 ± 3.77 % and nitrogen as 0.80 ± 0.35 %, with pH values 5.01 ± 0.83 in ponderosa pine sites in 2011, three months after the Las Conchas fire (Weber et al., 2014). The climate is characterized by a bimodal precipitation distribution, with 41 % of the 496 mm mean annual precipitation falling as summer rain and 59 % falling as winter snow (Guiterman et al., 2018). July is historically the warmest month, with a mean temperature of 28 °C, and January is the coldest month, with a mean temperature of -1.6 °C, as recorded by the Los Alamos weather station roughly 10 km northeast of the study site (recorded between 1911 and 1988, Bowen, 1990). Historically, summer precipitation is highest across July (80.7 mm), August (99.8 mm), and September (41 mm), with the onset of the North American monsoon (Bowen, 1990). Planting during summer is increasingly becoming common practice in the region because planting can be timed with the onset of the monsoon, whereas fall planting carries the risk of subsequent winter snow drought.
Over the three-year period of the experiment, 2019 was the most climatically similar to historical maximum, mean, and minimum air temperatures (mean difference; 0.26 °C / -0.7 °C/ -0.45 °C, respectively) and was considerably wetter in July, the hottest month (101.8 mm rainfall) than the historical rainfall average (80.7 mm). In contrast, 2020 was hotter than historical means with a particularly hot and dry June, with maximum, mean, and minimum air temperatures reaching 4.7 °C, 3.7 °C, and 4.9 °C, higher than mean historical records, respectively. Coupled with 100 mm less rainfall than historical averages for the year, 2020 was likely climatically unfavorable for planted seedlings. Following this relatively hot and dry year, 2021 was uncharacteristically cool and wet until the end of the experiment, with maximum, mean, and minimum air temperatures cooler than historical averages (-7.58 °C / -1°C / -0.82 °C) and an average of 65 mm more rainfall per month than the historical mean (Supplemental Materials Fig. 2).
Three weeks prior to planting (June 24–27), we amended soils with pecan biochar. The biochar had a mean pH of 9.44 (n = 3), electrical conductivity of 909.4 μS cm-1, nitrogen concentration of 0.59 %, carbon concentration of 84.1 %, and a bulk density of 0.29 g cm-3. This high electrical conductivity, carbon-to-nitrogen ratio, and low bulk density can aid in water retention and increase soil pore space (Novak et al., 2009; Busscher et al., 2010, Richard et al., 2018). When the biochar was applied at the study site, we excavated soil at each seedling planting location in a 20 cm diameter circle down to a 15 cm depth, or ∼ 4.7 L of soil. The biochar was applied at a rate of 4 % by mass to the soil when it was excavated, then backfilled into excavated holes and patted flat, so each seedling was treated with 188 g of biochar, equivalent to 80 metric tons/ha if biochar was applied evenly throughout the study site. This application rate was based on the results of Omondi and colleagues’ study (2016) and Edeh and colleagues’ meta-analysis (2020), the results of which found increased water capacity in coarse-textured soils at biochar applications of 30–70 MT ha-1. We chose to exceed this upper limit as seedling mortality had been high in previous planting efforts (Marsh et al., 2022b).
Seedlings were cultivated at the New Mexico State University John T. Harrington Forestry Research Center as one growing season-old container stock, grown in greenhouses in 7 × 14 “cone-tainer” trays. Each seedling was initially grown in SC10R “cone-tainers” with a diameter of 1.5”, a depth of 8”, and a volume of 164 ml or 10 cu. in. Seed sourcing followed USFS seed transfer guidelines, in that the nearest available seed sources of each planted species were selected and approval was given by the regional USFS geneticist. Ponderosa pine and Douglas-fir seed were from populations in the Jemez Mountains, and southwestern white pine seed was sourced from the Lincoln National Forest. At the time of planting, the mean ponderosa pine seedling diameter was 3.3 ± 0.69 mm, and height was 17.9 ± 3.1 cm. For southwestern white pine, mean seedling diameter was 3.3 mm ± 0.57 mm, and height was 15.5 ± 2.49 cm. For Douglas-fir seedling, mean stem diameter was 2.4 ± 0.49 mm and height was 20.5 ± 3.21 cm. Each seedling was tagged and protected with a Vexar tube to reduce the risk of herbivory and remained covered for the duration of the experiment. At the beginning (April) and end (October) of each growing season for three years, 2019, 2020, and 2021. We monitored seedling survival as the presence of green needles and measured growth. We measured height using a tape measure from the ground to the topmost woody part of the plant (in cm) and basal diameter (in mm) using calipers. Because of herbivory and partial stem death of some individuals, the change in some growth metrics was negative (n = 150).
For some seedlings (n = 162) in the shrub treatments, we measured shrub height above the seedlings by creating a canopy height model using data from an Unpiloted Aerial System (UAS) 3-dimensional structure-from-motion workflow that had been collected as part of Krofcheck et al. (2019). These data were collected for a portion of the experimental planting area between July 2–18 in 2018 using a custom‐built hexacopter equipped with a Sony a6000 RGB camera (https://www.sony.com/) and a 19 mm prime lens. The camera shutter activated with an EMLID Reach global navigation satellite system (GNSS) receiver on the UAS, so each picture taken simultaneously recorded location data relative to a second EMLID Reach GNSS receiver, positioned as a base station and running concurrently during the UAS operation. Flight planning occurred in Mission Planner (version 1.3 [https://ardupilot.org]) and all flights were within visual line of sight and at 80–100 m above ground level, resulting in a ground sample distance of 1.4–3.5 cm per pixel. A front image overlap of 85 % and side image overlap of 80 % were designated for the flight plan. The RAW format images were converted to 16‐bit linear TIFF files in Python 3.6 and imported into Agisoft Metashape (version 1.7.0, Agisoft 2017) for structure-from-motion processing.
Imagery was post-processed using RTKLib and nearby CORS stations, resulting in a total geolocation accuracy of +/- 5 cm. In-situ measurement accuracy of the imagery was characterized using distributed targets of known dimensions and determined an RMSE in image-derived measurement uncertainty of +/- 3 cm in the × and y dimensions, for targets on level ground. We then followed a procedure similar to Cunliffe et al. (2016), to convert imagery into raster layers of a digital surface model (DSM) and digital elevation model (DEM) by manually removing vegetation points from the DSM point cloud. Using tools in QGIS (version 3.14, QGIS Development Team, 2020), we then measured the amount of depth of shrub ‘canopy’ above each seedling and the distance from seedling to shrub edge.
We computed survival curves using non-parametric Kaplan-Meier methods, which estimate the true survival function of populations by using a tabulation of the number of seedlings at risk, and the number of deaths at each measurement point. Seedling survival probability was estimated for each species and treatment using the ‘survival’ R package (Therneau et al., 2013), with results plotted using the ‘survminer’ R package (Kassambara et al., 2017) to provide a survival probability for each species and treatment through time. We evaluated the effects of different treatments on seedling growth by comparing height and stem width growth between groups using ANOVAs, followed by the post-hoc Tukey’s honest significant difference method, a multiple comparison test of means allowing the comparison between more than two groups (Abdi and Williams, 2010). We visualized growth results by plotting cumulative diameter and height growth through time with Local Polynomial Regression (loess) curves. For 162 seedlings planted in shrub treatments, we explored the effects of distance from shrub edge (we treated the shrub edge as ‘zero’ and values increased toward the shrub center), and the amount of shrub above each seedling on height and diameter growth at the time of the last measurement using simple linear regression models. For linear models, each seedling was treated as an individual data point, grouped by species.