Forests can experience negative feedbacks in the growth of tree populations but positive feedbacks within the two dominant mycorrhizal types of trees: ectomycorrhizal (EM) and arbuscular mycorrhizal (AM). Positive feedbacks within mycorrhizal types may provide communities with resistance to climate change. We tested whether each mycorrhizal type led to positive feedbacks on seedling survival, while statistically controlling for the effect of congeneric trees in ambient versus rainfall reduced conditions. We explored two potential drivers; the variation in soil fungal community structure and soil chemistry. Seedlings benefited from growing in stands dominated by their own mycorrhizal type, and simultaneously, tree seedlings performed worse in the presence of adult trees of their own genus, but only in rainfall reduced conditions. We found that the composition of the EM fungal community differed between plots dominated by EM versus AM trees. These results indicated that mycorrhizal types may create positive feedbacks in dry conditions that should be considered when predicting future states.

Methods

Site description and plot selection

We assessed the effects of mycorrhizal type and throughfall reduction on tree seedling success in a field experiment in central Illinois. We quantified feedbacks of mycorrhizal type and congenerics by comparing tree seedling success and performance across 12 forest plots that varied in the basal area of EM trees and percent basal area of trees congeneric to our transplanted seedlings. The study site was in Allerton Park and Retreat, a 1500-acre natural area (https://allerton.illinois.edu/). Within a 64-acre area old-growth upland forest, 12 plots with similar topography were selected that represented a gradient of % EM canopy tree basal area (see Fig. S1 and Table S1). Species compositions of plots were determined by recording the diameter at breast height (DBH) and species identity of each tree over 10 cm DBH in a circle of 30 m diameter. Tree species were designated as EM or AM mycorrhizal status (See Table S2). The % basal area of canopy EM tree species was determined by dividing the summed basal area of EM tree stems (Quercus, Carya, and Tilia individuals) by total basal area of all trees.            

Experimental Design

The effects of mycorrhizal guild feedbacks, congeneric feedbacks, and throughfall treatments on tree seedling success were determined using a split-plot design. Within each of the 12 plots we established three sets of paired control and rain out subplots, totaling six subplots per plot. Canopy openness (%) was calculated using Gap Light Analyzer imaging software from hemispherical photographs at each of the subplots (Frazer et al. 1999).. In each subplot, a mix of 11 AM and EM tree seedlings were planted 40 cm apart. Bare-root seedlings were sourced from two nurseries. Tree seedlings from Engel’s Nursery in Fennville, MI were planted May 15-19, 2017, and all tree seedlings were one year old except for Diospyros virginiana seedlings, which were one to two years old. All tree seedlings from Mason State Nursery in Topeka, Il were planted May 25-27, 2017 and were one year old.

To reduce natural precipitation, we constructed three throughfall reduction shelters and three mock shelters at each site.  A 1.2m×1.5m sheet of clear polycarbonate plastic was attached to the top of four PVC legs, approximately 1.3m in length. For the ambient treatment, mock shelters had 80 holes (d=1.27 cm) drilled into each plastic sheet (10.2% of area removed) to allow precipitation through to tree seedlings.  The throughfall reduction shelters were left unaltered and were positioned at an angle to allow for drain off. Soil moisture (m³/m³ water content) and soil temperature on one paired subplot in each of three main plots were measured using a TEROS 12 Moisture sensor. The ZL6 Cloud data logger logged data every 12 hours from June 1 to October 16, 2019. Throughfall reductions shelters reduced average soil moisture by 10.4% + 0.08 SE in the 2019 growing season.                     

The average precipitation for the growing season (June-Oct.) for the 30 years prior to the start of the study (1986-2016) was 449 mm at this site (PRISM 2004). In 2017, the precipitation for the growing season was 321 mm, indicating drought conditions for the first year of the study. In both the 2018 and 2019 growing season, precipitation was slightly above normal (546 and 532 mm, respectively), with higher precipitation in the early growing season in 2018 and the later growing season in 2019.

Each tree seedling was assessed for survival and performance measurements across three years (2017-2019) in early growing season and late growing season. We assessed tree seedling survival and measured stem diameter and height for living plants. At the end of the third growing season (2019), surviving plants were harvested, collecting all aboveground tissue and as much belowground tissue as possible within the subplot perimeters. Roots were immediately assessment of mycorrhizal colonization and the remainder dried for biomass measurements.

            AM fungal colonization on surviving AM seedlings was measured on cleared and stained roots using the grid line intersect method at 40X magnification (Giovannetti & Mosse 1980). Ectomycorrhizal colonization was assessed under a dissecting microscope using methods in Brundrett et al. 1996. Since most seedlings died before the end of the experiment, any destructive sampling of the surviving seedlings was highly censored. For additional details, see Supplemental Methods.

            We collected soil cores (10 cm depth) from each of the 72 subplots (three ambient and three throughfall reduced for each of 12 sites) in summer 2018. Fungal communities were characterized in these soil samples by sequencing the ITS2 region of the fungal rRNA gene, as amplified by the ITS3-KYO2 (Toju et al. 2012) and ITS4 (White et al. 1990) primer pair. Amplicons were sequenced via paired end Illumina Miseq with 300 cycles. Sequences were clustered into exact Amplicon Sequence Variants using the DADA2 program (Callahan et al. 2016) as implemented in the QIIME2 pipeline and identified to the lowest confident taxonomic level using the naïve Bayesian classifier using the UNITE database for ITS reads (Nilsson et al. 2019). We used the FUNGUILD database to assign ASVs (ASVs equivalent to 100% identity OTUs) to functional groups (Nguyen et al. 2016). For additional details, see Supplemental Methods.

We took a soil sample from each pair of subplots at the 12 plots, totaling 36 samples for soil analyses at University of Wisconsin Soil & Forage Analysis Lab (Marshfield, WI). Soil pH was measured using a 1:1 soil to water dilution, concentrations of P and K were extracted using a Bray method, and % organic matter was measured using the Loss-On-Ignition method. Total N and C were assessed using the dry combustion method.