Unpredictable soil conditions can affect the prevalence of a microbial symbiosis

Introduction

The evolution of cooperation varies with ecological unpredictability (Scott, 2023; Scott et al., 2023b). For example, the prevalence of cooperative breeding in birds is associated with unpredictable environmental conditions (Jetz & Rubenstein, 2011; Griesser et al., 2017). Cooperative breeding is thought to allow organisms to invade unpredictable environments (Cornwallis et al., 2017) or buffer against times when conditions are harsh (Capilla-Lasheras et al., 2021). So far studies on the relationship between ecological unpredictability and cooperation have focused on interactions between members of the same species (Jetz & Rubenstein, 2011; Sheehan et al., 2015; Griesser et al., 2017; Firman et al., 2020). Cooperative associations between different species in a symbiosis, or mutualism, has been suggested to have similar benefits in unpredictable environments (Lekberg & Koide, 2014; Veresoglou et al., 2021; Scott, Queller & Strassmann, 2022a) and may thus be associated with them. However, this association has not been tested.

We investigated whether symbiosis was associated with unpredictable conditions using the microbiome of Dictyostelium discoideum. D. discoideum is a social amoeba that spends part of its lifecycle as a single cell eating bacteria in the soil (Raper, 1937). After exhausting edible bacteria, individual amoebae come together and form a multicellular structure called a fruiting body to disperse resistant spores (Kessin, 2001). Inside some fruiting bodies in the wild, different species of bacteria have been identified (Brock et al., 2018; Sallinger, Robeson & Haselkorn, 2021; Steele et al., 2023). Most of these bacteria appear to be regular soil bacteria that happen to be in the matrix of the spore-containing part of the fruiting body. Many of these bacteria are even edible by D. discoideum (Brock et al., 2018).

A subset of the bacteria that are found in D. discoideum fruiting bodies appear to be prevalent symbionts. The first symbionts to be discovered were three species of facultatively endosymbiotic Paraburkholderia bacteria (Brock et al., 2011; DiSalvo et al., 2015). The life histories of these Paraburkholderia bacteria in their natural soil habitats are unknown, but they can be cultured outside of their hosts in the lab (DiSalvo et al., 2015; Brock et al., 2020) and one species has been shown to be horizontally transferred in the lab (Noh et al., 2024). Two of these Paraburkholderia species, P. hayleyella and P. bonniea, may have a longer history of host association as shown by their reduced genomes, while P. agricolaris may be a newer symbiont (Noh et al., 2022).

All three Paraburkholderia species increase host fitness by allowing hosts to carry other species of edible bacteria along with Paraburkholderia inside the spore-containing part of the fruiting body called a sorus (Khojandi et al., 2019; Brock et al., 2020). Paraburkholderia are often carried inside spores while prey bacteria are carried outside of the spores in the sorus (Khojandi et al., 2019). Carriage allows host amoebae to seed out populations of prey bacteria that hosts can then eat (Brock et al., 2011). However, the ability to carry comes at the cost of reduced spore production when edible bacteria are common (DiSalvo et al., 2015; Scott, Queller & Strassmann, 2022b). The source of this fitness cost for hosts is unknown, though there is some evidence that Paraburkholderia itself harms hosts. For example, the density of Paraburkholderia tends to be associated with lower host spore production (Scott, Queller & Strassmann, 2022a, 2022b; Noh et al., 2024) and Paraburkholderia infection interferes with host immune cells that develop during the multicellular stage (Scott et al., 2023a).

D. discoideum also harbors endosymbiotic bacteria that are obligate: one from the genus Amoebophilus and different haplotypes from the phylum Chlamydiae. These obligate endosymbionts cannot be cultured outside of their hosts. Both Amoebophilus and Chlamydiae have not been found to measurably affect host fitness, even when they occur as coinfections with Paraburkholderia (Haselkorn et al., 2021). We will refer to these obligate endosymbionts as Amoebophilus and Chlamydiae.

Environmental sampling has found that Paraburkholderia prevalence is about 25% of sampled hosts but varies by sampling location (Haselkorn et al., 2019) and over time (DuBose et al., 2022). Obligate endosymbionts are found in about 40% of sampled hosts (Haselkorn et al., 2021). Paraburkholderia and Amoebophilus coinfections are more common than expected due to chance (Haselkorn et al., 2021).

A key source of unpredictability in the soil environment of D. discoideum that has not been studied is precipitation. Precipitation can drastically shift the soil environment because of the complex structure and physical properties of the soil (Or et al., 2007). Such shifts are known to affect the abundance of microbes in the soil (Zeglin et al., 2013). When precipitation is unpredictable, it is likely to impact the availability of soil bacteria for D. discoideum to eat. We hypothesize that hosts that have Paraburkholderia symbionts may be buffered from unpredictable changes in prey abundance because they can carry prey bacteria (Scott, Queller & Strassmann, 2022a).

Other soil characteristics may also be important for the prevalence of symbiosis. pH has already been shown to affect the D. discoideum microbiome (Sallinger, Robeson & Haselkorn, 2021). Temperature can have strong effects on host amoebae that could shape their interactions with symbionts (Shu et al., 2020). Soil nutrients (usually measured by the ratio of carbon to nitrogen in the soil) have not been studied in this symbiosis but are known to affect other symbiotic interactions (Johnson et al., 2010) and to affect soil bacteria (Bahram et al., 2018) that are potential prey of D. discoideum. Here we test how unpredictability and other soil characteristics affect the symbiosis between D. discoideum and its symbionts.

Materials and Methods

Results

To test for relationships between soil characteristics and symbiont prevalence, we used presence-absence data of symbionts that were collected from 22 collection trips to 21 locations (Fig. 1A, Table 1) across the eastern United States (Haselkorn et al., 2019, 2021). Because some coinfections are known to be more common than expected (Haselkorn et al., 2021), we first tested all screened hosts for non-random coinfections that may also vary with the soil environment using logistic regression. Paraburkholderia coinfections were not more common than expected (Fig. 1B) even when we tested for coinfections at individual locations using Fisher’s exact tests (Table SI). Generally, Paraburkholderia and Amoebophilus coinfections are more common than expected with P. hayleyella and Amoebophilus coinfections being the most enriched (Fig. 1B). This extends prior findings that focused on a subset of locations (Haselkorn et al., 2021). Amoebophilus and Chlamydiae coinfections are less common than expected across our sampled sites.

To identify associations with symbiont prevalence, we used logistic regression models. To measure precipitation unpredictability, we calculated Colwell’s P (see Table 2 in Colwell, 1974) using monthly precipitation data for each location since 1901 (Fig. 2A). Rainfall was generally unpredictable as Colwell’s P ranged from 0.25 to 0.42. To make sure that more recent unpredictability did not deviate from long-term predictability, we also calculated Colwell’s P for the last 5 years before collection at each location (Fig. 2B). Colwell’s P ranges from 0 to 1, with 0 being unpredictable and one being perfectly predictable (Colwell, 1974). Along with our measures of unpredictability, we collected mean annual precipitation (Fig. 2C), soil pH (Fig. 2D), soil mean annual temperature (Fig. 2E), soil carbon to nitrogen ratio data (Fig. 2F). Correlations between these variables tended to be low with the exception of mean annual temperature being correlated with pH and long-term Colwell’s P (Fig. S1).

We found that the frequencies of the two Paraburkholderia species with reduced genomes, P. hayleyella and P. bonniea, were associated with unpredictable precipitation, but in opposite directions (Figs. 3A and 3B). Other variables measuring mean soil characters were not associated with prevalence unless also included with unpredictable precipitation (Table SI). P. hayleyella prevalence was higher in more unpredictable environments (log-odds = −1.047, se = 0.545; Fig. 3A) while P. bonniea prevalence was higher in more predictable environments (log-odds = 1.181, se = 0.442; Fig. 3B). These opposite responses to the predictability of precipitation remained even when we accounted for the influence of the two largest sampling locations (Fig. S2). Moreover, our data show (histograms in Fig. 3) that P. hayleyella is found where precipitation is relatively unpredictable (0.33.5 on average) while P. bonniea is found where precipitation is more predictable (38.1 on average; Permutation test: p < 0.001). For the other symbiont infections – P. agricolaris, the obligate Amoebophilus and Chlamydiae endosymbionts, P. hayleyella-Amoebophilus coinfections, and even symbiosis overall–prevalence was not associated with unpredictable precipitation or mean soil characteristics (Table SI).

Discussion

Our finding that P. hayleyella prevalence increases in unpredictable conditions supports our hypothesis that symbiosis may buffer hosts during times when conditions are unpredictable. However, our finding for the relatively less common P. bonniea is in the opposite direction of this hypothesis.

One explanation for why unpredictability differently affects P. hayleyella and bonniea prevalence is that these sister species (Brock et al., 2020) compete and are partitioning their niches within hosts based on unpredictable precipitation. Indeed, P. hayleyella and P. bonniae were found on different ends of our measure of the predictability of precipitation. Some additional support for niche partitioning comes from a previous finding that P. hayleyella and P. bonniea differ on which sugars they can metabolize (Brock et al., 2020). One argument against niche partitioning within hosts is that the prevalence of these symbionts may not be high enough for strong competition within hosts. Instead, competition in the soil could drive niche partitioning between these symbionts. In this case, interactions with other members of the D. discoideum microbiome that were not included here may be involved. Another possibility is that life history characteristics of P. hayleyella and P. bonniea affect the ability of these bacteria to survive in soils with different levels of unpredictable precipitation.

In addition to the role of unpredictability, we also identified new associations between different symbionts in D. discoideum hosts. We found that P. hayleyella and Amoebophilus coinfections are more common than expected and Amoebophilus and Chlamydiae coinfections are less common than expected (Fig. 1B). The association between P. hayleyella and Amoebophilus suggests that the abundance of both may be driven by the same environmental conditions. Another possibility is that P. hayleyella and Amoebophilus are mutualists that have increased survival when together in the same host. The rarity of Amoebophilus and Chlamydiae coinfections may indicate competitive exclusion inside D. discoideum hosts. Another explanation is that Amoebophlius or Chlamydiae actively prevent each other’s colonization. Chlamydial endosymbionts have been shown to reduce the success of other endosymbionts in other species of amoebae (König et al., 2019; Arthofer et al., 2022).

Our results provide suggestive evidence of the role of unpredictability driving symbiosis that should be followed up in future studies. However, our study is limited in several ways due to data constraints. First, our samples were not replicated over time, so our results do not capture the variation in Paraburkholderia symbiosis over time. This may be an important factor as other soil sampling studies have found that symbiosis with Paraburkholderia may vary over time in some locations (DuBose et al., 2022). Second, our soil and climate measures do not capture within site heterogeneity that is important in many microbial systems (Nannipieri et al., 2019). For this reason, our study should inspire future fieldwork with better sampling to better understand the drivers of symbiosis in this system.

This study demonstrates that the frequency of a microbial symbiosis can be associated with unpredictable environmental conditions. Unpredictable conditions may be an important driver of cooperation between members of the same species and between different species.

Supplemental Information