Testing for microbe-mediated adaptation
Reciprocal transplant experiments can be used to understand how microbes affect patterns of plant adaptation and to distinguish between microbe-mediated local adaptation and microbe-mediated adaptive plasticity. Reciprocal transplants can be employed exactly as in the traditional design to study local adaptation and plasticity (Clausenet al. 1948) with the added element of manipulating microbes (Fig. 2).
In classical reciprocal transplant experiments, seeds from one population are transplanted both into a different habitat and replanted into their natal habitat. Populations are often chosen because one or more environmental variable is demonstrably different between populations (e.g. serpentine vs. non-serpentine soils, Wrightet al. 2006 or presence or absence of a competing species, Lau 2006). A reciprocal transplant to test for microbe-mediated adaptation would involve transplanting seeds into different habitats and providing those plants with local or foreign microbes. Although it can be difficult or impossible to completely exclude microorganisms, sterilized control plants should be included to definitively attribute effects to microbes (e.g. , if microbe-mediated local adaptation is strong, then patterns of local adaptation will be much weaker or not evident when microbes are absent).
A statistical model to understand interactive contributions of plant and microbes to plant local adaptation includes some measure of plant fitness (e.g. germination,
survival, fecundity, or ideally an integrated fitness metric encompassing all life history stages [ASTER models] Geyer et al. 2007; Shaw et al. 2008) as a response variable and habitat type, plant source, microbial source, and sterilization as fixed effects (Table 1). While this design potentially results in a four-way interaction, with large enough sample sizes and patience, the complexity of this and other interactions in the models can be understood, especially through hypothesis testing with planned comparisons. The figures presented here assume that patterns of adaptation are largely attributable to habitat-specific microbes (Fig. 1B, C, D), however in the text we also discuss the importance of significant sterilization terms.
A three-way interaction between plant source, microbial source, and habitat (GpxGmxE) would indicate microbe-mediated local adaptation, if home genotypes have higher fitness with their home microbes in their home habitat (Fig. 3A). Whereas, an interaction between plant source and microbial source (GpxGm) may indicate that plant genotypes do best with their own microbes, regardless of habitat and are locally adapted to their natal microbes (Fig. 3B). For example, Petipaset al. (2020) found that germination of Hypericum perforatum sourced from limestone barrens was highest when transplanted with microbes from limestone barrens into limestone habitats, a pattern consistent with microbe-mediated local adaptation (GpxGmxE). However, survival of H. perforatum from limestone barrens was highest when transplanted with microbes from limestone barrens regardless of transplant site, a pattern suggesting that plants are locally adapted to their natal microbial communities (Gp x Gm), rather than indicating a pattern of microbe-mediated local adaptation.
An interaction between habitat and microbial source may be indicative of microbe-mediated adaptive plasticity. In this case, local microbes are responsible for locally adapted plant phenotypes; however, plant source populations may be equally adept at utilizing local microbes (i.e. they exhibit plasticity in microbe facing traits, allowing them to attract, retain, regulate relevant local microbes; Fig. 3C). An example of a trait underlying these adaptive patterns is microbe-mediated flowering time, where plants only achieve the optimal flowering time phenotype in association with local microbes (Fig. 3D).
Significant sterilization terms can also indicate microbe-mediated local adaptation and microbe-mediated adaptive plasticity. For example, where hosts strongly control microbiome composition (e.g. Gehring et al. 2017), microbe-mediated local adaptation would be evident as a Gp x E x sterilization effect (rather than GpxGmxE), where plant local adaptation is only observed in the presence of microbes, but the microbial effects are primarily driven by plant genotype specific recruitment from a common soil source. This may still be considered microbe-mediated local adaptation because the relative importance of different microbes varies across context and the observed plant adaptation is still due to interactions with relevant microbes. However, in this case, the plant nearly entirely controls the composition of those microbes. Similarly, sterilization could indicate microbe-mediated adaptive plasticity when the environment elicits plastic shifts in plant phenotypes that in turn influence microbial communities in ways that feedback to affect plant fitness. Because the sterilization effect could also simply indicate the benefits or costs of association with microbial communities, inferring microbe-mediated adaptive plasticity from a significant sterilization effect is challenging and would require: 1) showing that shifts in plant phenotypes are associated with changes to the microbial community, and 2) these changes to the microbial community are associated with shifts in plant traits and/or fitness. The potential for this type of interaction is exemplified by host-mediated microbiome engineering approaches that use plant phenotypes to select adaptive microbiomes (Swenson et al. 2000; Panke-Buisse et al. 2015, 2017). In one example, wheat was cyclically exposed to drought and in each cycle the microbiomes associated with the most drought tolerant seedlings were moved to the next cycle. After six cycles, a plant phenotype mediated drought microbiome was selected that deferred drought symptoms in wheat seedlings by five days (Jochum et al. 2019).
Additionally, microbe-mediated adaptation necessarily involves a complicated experimental design and multiple players each with the capacity of becoming locally adapted and exhibiting trait plasticity. Consequently, there are many patterns that can emerge from these interactions, which requires careful experimental planning (including of experimental tests) and careful interpretation of the resulting data. Finally, although we have discussed microbe-mediated local adaptation and microbe-mediated adaptive plasticity largely independently, they are not mutually exclusive.
The same caveats and considerations that apply to typical reciprocal transplant experiments also apply to microbe-mediated adaptation transplant experiments, including the relevant spatial scale, which aspects of fitness to measure, and how long the experiment should be monitored (Cheplick 2015). The latter is especially important given that contamination in the field is likely to erode microbial treatments over time. For tests of microbe-mediated adaptation, additional care should be taken to determine relevant microbes, proper controls, and the feasibility of different microbial transplant designs (Box 3, Fig. 4).