In a previous study, 32 soil fungi isolates (isolated from the same soil environment) were tested for their response to recurring heat events (Szymczak et al., 2020). Fungi were isolated from 0 to 10 cm deep soil collected in grassland in Mallnow Lebus, Brandenburg Germany (Andrade-Linares et al., 2016b).

Based on the previous performance of the isolates in monoculture, we selected a subset of six isolates with contrasting responses and combined them as pairs or three species growing on Petri dishes containing potato dextrose agar medium (PDA). We then subjected them to the same recurring heat events as previously used in monocultures (Szymczak et al., 2020).

We selected three isolates that previously did not show priming ability: RLCS06 (Chaetominum anustispirale), RLCS07 (Amphisphaeriaceae sp.), and RLCS16 (Pleurotus sapidus), and three that showed priming ability: RLCS02 (Mortierella elongate), RLCS05 (Fusarium sp.), and RLCS32 (Fusarium oxysporum) (more details about the isolates in Supplementary Figure 1 and Szymczak et al., 2020). For ease of following the results, we refer to the different isolates with a code that combines their priming response (np, for no priming, and p, for priming) with the strain identity information (isolate numbers). Then, we refer to the isolates that previously did not show priming as np06, npo07, and np16 and the isolates that showed priming as p02, p05, and p32. All the fungal isolates were kept active in PDA at 20°C until used in the experiments.

We ran a test of the isolates growing with themselves in pairs or triplets at control temperature (C, constant temperature of 22°C) and the colonies looked very symmetrical and presented very similar growth patterns (Supplementary Figure 2). Consequently, to reduce experimental units, we did not apply the temperature treatment combinations on isolates growing with themselves, but only in the combinations of different isolates, as described in the following text.

First, the six selected isolates were co-cultivated in pairs in Petri dishes containing PDA and received different heat treatments. The heat treatments were previously established by simulating hypothetical heat waves near the soil surface and using temperatures and duration tested in previous experiments, as described in Szymczak et al. (2020): control (C, constant temperature of 22°C), mild perturbation (M, 35°C for 2 h), severe perturbation (S, 45°C during 1 h), and sequence of the two perturbations (MS, 35°C during 2 h, followed by 45°C during 1 h). The heat treatments were applied in a 2 × 2 fully factorial design, composed of mild perturbation (35°C per 2 h, yes or no) followed by a severe perturbation (45°C per 1 h, yes or no) (Szymczak et al., 2020).

For the experiment set up, the fungal isolates were first grown on PDA at 22°C and handled as previously described (Szymczak et al., 2020). Later, plugs were taken from the edge of the active colonies and transferred to new PDA plates according to the fungal combinations. The plugs were inoculated equidistantly from the border of the plate and between each other. After the inoculation of fungal combinations, six replicated plates for each treatment (fungal pairs and temperatures) were placed in replicate incubators according to the treatments (plates were placed in pairs in three different incubators). This means that the treatments were applied to two technical replicates (agar plates) nested within each of the three experimental units (three incubators). For the experiment I, we tested 15 combinations of fungal isolates with four heat treatments (2 × 2 factorial design) applied to six replicated plates (two agar plates placed/nested in three different incubators), in a total of 15 × 4 × 6 = 360 plates.

After the heat treatments, the fungal combinations were grown at 22°C for 2 weeks. Finally, the diameter of the colonies was measured and the growth rate of each strain was calculated for the incubation period, both for mycelial pairs that did and did not contact during the experiment period (Szymczak et al., 2020). Thus, the experiment design treated fungal interactions as an aggregate of interaction at distance, interference upon contact, and competition for space.

Later, the six selected strains described above were inoculated in 20 random combinations of three species in PDA media and submitted to the treatments previously described (the combinations are stated in Figure 3). The experiment had the same design described earlier and followed all the same laboratory steps but co-cultivated three isolates instead of two. For experiment II, we tested 20 combinations of fungal isolates, with four heat treatments and six replicates, in a total of 20 × 4 × 6 = 480 plates. Both experiments summed, 360 plates + 480 plates, resulting in 840 plates.

Analysis followed a previously proposed factorial design considering the factors as mild perturbation (yes or no) and severe perturbation (yes or no) (Hilker et al., 2016; Szymczak et al., 2020). The experimental design included a nested design, where duplicated cultivation plates were nested within three replicated incubators. Accordingly, the factorial design was fitted to a nested design (plates nested to the replicated incubators) in a linear-mixed effects model (lme). The model was fit using the software R (R Core Team, 2013) and the package nlme (Pinheiro and Bates, 2000). Some data did not fit normality (even though we tested a box-cox optimal transformation), thus, the models were analyzed by robust analysis of variance (rANOVA) using the R package Rfit (Hettmansperger and McKean, 2010). The p-values were adjusted by the Benjamini-Hochberg (BH) false discovery rate procedure to reduce the chance of type I error (Benjamini and Hochberg, 1995).

The individual responses of the isolates for the fungal combinations and thermal treatments were analyzed as univariate data (rANOVA), while the reciprocal effect of one isolate on the companion strains was analyzed as multivariate data (bivariate for the pairs and trivariate for the three-species combinations). The bivariate and trivariate data were analyzed by permutational multivariate analysis of variance (perMANOVA) using the R package vegan (Anderson, 2017; Oksanen et al., 2019). We also analyzed the data as a factorial focusing on each isolate and considering in addition to thermal treatments the company strains as a factor, i.e., mild perturbation (yes or no) and severe perturbation (yes or no) and company strain (combinations). This provided general insight into how the focal isolates’ growth rates overall changed when growing in pairs or triplets. For experiment I, we also plotted a chord diagram to indicate when a company strain changed the response of the focal strain compared to the control (according to the rANOVA result).

The combinations of two species were scatter plotted to visualize the effects of the heating treatments on the pair species simultaneously, while the responses of the individual isolates were visualized by box plots. For the combinations with three species, the data were visualized in ternary graphs plotted using the R package ggternary (Hamilton and Ferry, 2018), as also as box plots focusing on individual isolates, to allow a general comparison of the two experiments (pairs or triplets).

The six fungal isolates were first tested in pairs (Figure 1). The effects of the co-cultivation in response to the thermal treatments were either negative (the growth of one or both strains decreased after the heating treatments), neutral (no effect on the growth of the isolates), or facilitative (one or both isolates had increased growth rates after the heat events), depending on the pairs.

Compared to the control of each combination, four pairs had a changed interaction after the mild heating, all 15 combinations responded to the strong heating and three responded simultaneously to the MS heating event (Figure 1, perMANOVA p < 0.05). Observing the scatter plots (Figure 1 and Supplementary Figure 3), the moderate heating event had no evident effect on the co-cultivation growth rate of some isolates (e.g., combination np06-np07). On the other hand, some isolates grew better after the heat event, compared to the control without such a heat event. For example, there was a growth rate shift in the combination np06-np16, since the growth rate of np06 was reduced after the strong and mild-strong heating events in comparison to the control, while the growth rate of np16 increased after the strong and mild-strong heating events in comparison to the control (Figures 1, 2, 3.1). Furthermore, some isolates were severely affected by the strong heat event, resulting in a shift of growth rates in some combinations (Figure 1 and Supplementary Figure 2). For example, the isolate p02 co-cultivated with np07 had a growth rate of around 170 mm² day^–1 in the control, but the rate was reduced to less than 100 mm² day^–1 after the strong heat event. Simultaneously to this reduction, the isolate np07 doubled its growth rate from less than 50 mm² day^–1 in the control to 50–150 mm² day^–1 after the heat treatments (Figures 1, 2, 3.2).

When comparing the isolates individually, these changes were more evident and the growth of some fungi in response to heat events depended on the presence of the accompanying fungi (Figure 1, details in Supplementary Figures 3, 4 and Supplementary Table 1). For example, np06 was not affected by the mild, strong, and mild-strong heat events when this isolate was grown in pairs with the strains np07 or p05; nevertheless, some heat effects were observed when this strain was paired with strains p02, np16, or p32 (Figures 1, 3.3, 3.4). In another example, for p02 paired with np16, isolate p02 had a stable growth independent of the heat treatments. Yet, when paired with isolates p05, np07, or p32 (Figures 1, 3.5), isolate p02 had a strong reduction in growth rates after the strong heat event, compared to its growth in the control.

The positive effect of co-culturing was observed in some cases, e.g., p32 growing in the combinations with np06 and p02 (Figures 1, 3.6), had improved growth after the heat events (mild or strong), compared to the control, which did not occur in any other combination (Figure 1, details in Supplementary Figure 3 and Supplementary Table 1). Surprisingly, isolates that had priming ability in the previous study lost their ability to benefit from it when growing in combinations (Supplementary Table 1). Only the isolate np06, which did not have priming ability in the previous study growing alone (Szymczak et al., 2020), showed priming-like behavior in the present study when paired with isolate np16 (Supplementary Figure 4 and Supplementary Table 1).

Plotting the data for each isolate (Figure 2A) and analyzing also the fungal pair combinations as a factor (Figure 2B), showed that all isolates suffered interference of a company strain, while the thermal effect was variable according to each isolate (Figure 2).

The six fungal isolates were then tested in triplets in 20 random combinations (Figures 4, 5 and Supplementary Figure 5). The heat events changed the growth pattern of the isolates for some combinations. For example, in combination 1, p02-np06-p32, the isolate p02 growth rate in the triplet was largely suppressed by the strong and mild-strong heat event compared to the control (Figures 3.7, 4.1). A similar effect was observed in combination 3, np07-np06-p02 (Figures 3.8, 4.3), however, the suppression was less intense as the example before. In some other combinations, the heat event had no effect on the interaction of the isolates. For example, in combination 14, np07-np06-p05, all the isolates had no change in their growth pattern (Figure 4.14).

Surprisingly, the occurrence of the interaction of recurring heating events (mild-strong) was more frequently observed in the triplets than in the pairs (Figure 4), where 10 significant events were observed out of 20 combinations, 50% (perMANOVA p < 0.05). In the pairs (Figure 1), only five significant events were observed out of 15 combinations, 33% (perMANOVA p < 0.05).

When analyzing the data considering the fungal triplet combinations as a factor (Figure 5), the result was slightly different from when the strains grew in pairs (Figure 2). For example, for triplets, the mild treatment was statistically significant across all the isolates, while for pairs it was only for half of the isolates (rANOVA p < 0.05). However, in both, pairs and triplets, it was evident that there was always an effect of the fungal combinations on the focal isolate growth rate (rANOVA p < 0.01), while the thermal effects were variable according to isolates (Figures 2, 5 and Supplementary Figures 4, 5). Among all isolates, p02 was the one that had the most interactions with other isolates. This isolate influenced the heat response of other isolates and it was also influenced by the other isolates (Figures 2, 5 and Supplementary Table 1). This isolate is the one with the highest growth rate (Supplementary Figure 2), and consequently, it has potentially the best advantage in the competition for space. However, our data showed that this isolate potential advantage was challenged by the isolate responses to the different temperature treatments, which was modulated by the pair and triplet combinations.