Climate change can strongly alter soil microbial functioning via
plant–microbe interactions, often with important consequences for ecosystem
carbon and nutrient cycling. Given the high degree of intraspecific trait
variability in plants, it has been hypothesized that genetic shifts within
plant species yield a large potential to control the response of
plant–microbe interactions to climate change. Here we examined if sea-level
rise and plant genotype interact to affect soil microbial communities in an
experimental coastal wetland system, using two known genotypes of the
dominant salt-marsh grass
Climate change strongly affects soil microbial decomposition, with important consequences for global carbon (C) and nutrient cycles (Davidson and Janssens, 2006; Dijkstra et al., 2010). Plant–microbe interactions in the rhizosphere are particularly susceptible to various climate change factors (Philippot et al., 2013; Pugnaire et al., 2019; Wieder, 2014). It is therefore not sufficient to only study the direct effects of abiotic climate change drivers on soil microbial communities and resulting changes in ecosystem functioning. Plant-mediated indirect effects of climate change on soil microbial communities also need to be examined (Bardgett et al., 2008; Van der Putten et al., 2013). Prior work on a wide range of ecosystems indicated that changes in plant productivity and community composition control soil microbial functioning in response to climate change, often with marked effects on ecosystem C as well as greenhouse-gas and nutrient dynamics (Fuchslueger et al., 2014; Mueller et al., 2020; Stagg et al., 2018; Ward et al., 2013).
Climate change is known to affect the intraspecific genetic structure within plant populations (Bustos-Korts et al., 2018; Crutsinger et al., 2006; Jump and Peñuelas, 2005), and it has been hypothesized that these intraspecific genetic shifts can translate into important changes in soil microbial functioning (Fischer et al., 2014; terHorst and Zee, 2016; Van Nuland et al., 2016; Ware et al., 2019). This hypothesis is based on studies demonstrating differences in soil microbial community structure or activity in soils of different plant genotypes (Madritch and Lindroth, 2011; Pérez-Izquierdo et al., 2019; Schweitzer et al., 2008; Seliskar et al., 2002; Zogg et al., 2018). Furthermore, genotype effects on soil C and nitrogen (N) stocks as well as N transformations have been observed to be variable across multiple common garden sites (Pregitzer et al., 2013). However, experimental evidence for the interaction effects of plant genotype and climate change factors on soil microbial C cycling is virtually absent.
Plant-mediated climate change effects on soil microbial functioning are expected to be particularly pronounced in wetlands, because here plants not only control the microbial substrate (i.e., electron donor) supply, but they also regulate the availability of electron acceptors by providing oxygen to an otherwise reducing rhizosphere (Kirwan and Megonigal, 2013; Wolf et al., 2007). At the same time, wetland soil microbial functioning plays a disproportionately large role in the global climate system (Freeman et al., 2001; Megonigal et al., 2003). In recent years, climate change research in tidal wetlands and other so-called blue carbon ecosystems has gained increasing attention by the scientific community (Kirwan et al., 2013, 2014; Spivak et al., 2019). These ecosystems are among the most effective long-term C sinks of the biosphere (Chmura et al., 2003; McLeod et al., 2011), but the impacts of accelerated rates of sea-level rise (SLR) destabilize tidal wetlands worldwide (Kirwan and Megonigal, 2013).
SLR affects the flooding frequency of tidal wetlands and represents the overriding climate change factor impacting tidal wetlands (Kirwan and Megonigal, 2013). Its effects on ecosystem functioning are largely plant-mediated and extremely variable, ranging from strong positive effects on soil C sequestration to ecosystem destabilization and ultimately loss (Rogers et al., 2019). SLR and the resulting flooding frequency alter plant primary production and microbial decomposition, the two primary factors controlling C sequestration in coastal marine ecosystems (Kirwan and Megonigal, 2013). Primary production often follows a unimodal (i.e., optimum) response to SLR, although interspecific variability is high (Kirwan and Guntenspergen, 2012; Morris et al., 2013). The microbial decomposition response to SLR is less understood. The prevailing notion is that decomposition rates are inversely related to flooding. However, recent studies demonstrated that the responses of decomposition and primary production to SLR are coupled (Janousek et al., 2017; Mueller et al., 2016; Stagg et al., 2017). For instance, Mueller et al. (2016) have demonstrated soil microbial activity is not directly affected by SLR and its control on soil oxygen availability but indirectly by the aboveground-biomass response to flooding frequency, which determines the input of both oxygen and labile substrates to soil microbial communities.
Considering the low plant community-level diversity of many wetland types, such as
salt marshes and ombrotrophic peatlands
(Wanner et
al., 2014; Warner and Asada, 2006), and the strong plant control of microbial C
cycling in wetland soils, it is possible that
intraspecific variation and adaptive capacity function as important yet
largely overlooked mediators of wetland–climate feedbacks. Here, we study the
interaction effect of flooding frequency and plant genotype on soil
microbial community structure and functioning, using the dominant
tidal wetland grass
Conceptual diagram illustrating the hypothesis that effects of
sea-level rise on soil microbial functioning are mediated by intraspecific
genetic variation in plants. Two genotypes of the dominant tidal wetland
grass
The experiment was conducted from July to October 2017 (12 weeks) at the
Institute of Plant Science and Microbiology (IPM), Universität Hamburg,
Germany. We used platforms positioned at three elevations in a 12 m
Plants were collected in April 2015 from
Soil sampling took place in October 2017 after 12 weeks of exposure to
different flooding treatments and plant genotypes. Plant biomass and litter
were removed prior to sampling. From each mesocosm, one soil sample was
taken as a 5 cm diameter and 5 cm deep core using a volumetric steel ring.
Subsamples of 20 g were homogenized and stored frozen until used for
microbial enzyme assays and DNA extraction. The residual sample was passed
through a 2.5 mm sieve, air-dried at 65
Potential exo-enzyme activity (EEA) of ß-glucosidase, cellobiosidase,
leucine-aminopeptidase, and chitinase was determined in fluorometric assays
following Mueller et al. (2017). Briefly,
We assessed the decomposition of standardized plant litter in the
rhizosphere to evaluate if genotype effects on soil microbial exo-enzyme
activity translate into altered organic matter turnover and thus into
ecosystem functioning (Ochoa-Hueso
et al., 2020). The decomposition rate constant (
Soil DNA was extracted from
We used two-way ANOVA or two-way PERMANOVA to analyze the data of our
two-factorial design (two genotypes and three flooding frequencies). Normal
distribution of residuals was assessed visually prior to ANOVA testing. Due
to the fully balanced study design, potential moderate deviations from
homogeneity of variance between groups were considered unimportant for both
ANOVA and PERMANOVA testing
(Anderson,
2017; Box, 1954; McGuinness, 2002). Along with ANOVA tests, we used
Cochran's C test with
Two-way ANOVA was conducted to test for effects of flooding frequency, plant
genotype, and their interaction on EEAs,
Exo-enzyme activities (nmol g DW
Enzyme activities were only affected by flooding frequency in soils planted with the intolerant genotype, whereas none of the four EEAs were affected in soils planted with the tolerant genotype (Table 1). In soils with the intolerant genotype, all four EEAs showed a unimodal response to flooding: they were always highest at the intermediate (i.e., weekly) flooding frequency and always lowest at the highest (i.e., daily) flooding frequency, whereas no consistent pattern was found in soils of the tolerant genotype (Table 1, Fig. 2). Overall, the effect size of flooding frequency (i.e., the difference between highest and lowest mean activity of the three flooding treatments) was 1.7–4.7 times greater in the intolerant vs. tolerant genotype (Fig. 3).
C-acquisition enzymes (ß-glucosidase and cellobiosidase, sensu Sinsabaugh et
al., 2009) showed different responses than N-acquisition enzymes
(leucine-aminopeptidase and chitinase, sensu Sinsabaugh et al.
Analyzing the EEA data in relation to the activity under unplanted conditions revealed contrasting plant effects between genotypes (Fig. 2). Specifically, at our highest flooding frequency, the activity change in relation to the unplanted condition was negative in the intolerant genotype but positive in the tolerant genotype (Fig. 2). This contrasting pattern in the direction of plant effects was generally found for all enzymes assayed, but it was significant in the N enzymes only (Fig. 2). The absolute values of enzyme activities under unplanted conditions are presented in the Supplement. None of the four enzymes assayed showed a significant response to changes in flooding frequency under unplanted conditions (Fig. S2, Table S1).
Exo-enzyme activity (EEA) of C-acquisition enzymes (
Maximum change in exo-enzyme activity (EEA) induced by the
flooding treatment in soils planted with flooding-intolerant vs. flooding-tolerant
genotypes of
The initial belowground litter decomposition rate,
Significant relationships between plant biomass parameters (taken from
Reents et al., 2021), soil EEAs, and litter-breakdown parameters have been
observed (Table 2). C enzymes were not significantly related to any plant
biomass parameter, reflecting the missing plant genotype effect on microbial
C-enzyme activities, whereas N-enzyme activities were significantly
positively related to plant aboveground biomass (Table 2). Relationships
between plant biomass parameters and litter-breakdown parameters (
Correlations between plant biomass parameters and soil microbial
activity parameters using a Bonferroni correction for multiple comparisons
(
C activity: sum of C-acquisition enzyme activities (ß-glucosidase
Initial decomposition rate constant (
Flooding frequency (two-way PERMANOVA,
NMDS plot showing prokaryotic (bacterial and archaeal) community
composition in soils planted with intolerant and tolerant plant genotypes of
The present study provides experimental evidence of genotype–environment
interaction effects on soil microbial enzyme activity (Figs. 2, 3) and
belowground litter breakdown (Fig. 4), two key processes controlling
ecosystem C and nutrient cycling. Specifically, plant genotype determined
the presence or absence of flooding-frequency effects on microbial enzyme
activities and litter breakdown. This result yields important implications
for our understanding of soil–climate feedbacks in the coastal zone, because
it shows that plant-genotype controls can mask or enhance the effects of SLR
on soil microbial processes. Our data furthermore suggest genotype–SLR
interaction effects on the soil microbial community structure (Fig. 5).
This finding is in agreement with a recent observational study on
genotype–environment interactions in terrestrial ecosystems, suggesting that
climate-driven reduction of genetic variation in
While the majority of studies on genotype–environment interactions are concerned with plant responses to temperature or latitudinal climate gradients in terrestrial ecosystems (Bauerle et al., 2007; Curasi et al., 2019; Taylor et al., 2019; Walker et al., 2019; Ware et al., 2019), the present work is focused on SLR, the overriding climate change factor in coastal ecosystems, such as tidal wetlands (Kirwan and Megonigal, 2013). The effects of SLR on soil microbial activity can be tightly controlled by the plant response to changes in flooding frequency, as demonstrated by recent studies showing strong positive correlations between aboveground biomass and soil litter decomposition (Janousek et al., 2017), cellulose decomposition (i.e., tensile strength loss; Jones et al., 2018), or recalcitrant soil organic matter decomposition (Mueller et al., 2016). The importance of plant processes in controlling soil microbial functioning in response to changing flooding frequency is reflected in the findings of the present study: in the absence of plants, flooding frequency affected neither soil microbial enzyme activities nor the soil microbial community structure (Figs. S2 and S5). In the presence of plants, however, flooding frequency and genotypic variation in plant biomass exerted significant effects on soil microbial activity and community structure. Most notably, microbial enzyme activities only responded to changes in flooding frequency when aboveground biomass responded. Aboveground and belowground biomass across flooding treatments was unchanged in the tolerant genotype, whereas the intolerant genotype showed a strong reduction of aboveground biomass at our highest flooding treatment (Reents et al., 2021). Consequently, only the flooding-sensitive intolerant genotype showed changes in soil microbial activity, whereas the tolerant genotype was able to maintain microbial enzyme activities at a constant level across the flooding gradient (Table 1; Fig. 2).
In support of the notion that the soil microbial activity response to
increasing flooding frequency follows the response of plant aboveground
processes, we found a significant relationship between aboveground biomass
and microbial N-acquisition activity (aminopeptidase
To evaluate if genotype effects on soil microbial communities translate into
altered organic matter turnover and thus ecosystem functioning, we assessed
the decomposition of standardized plant litter in the rhizosphere. The
parameters
Although significant correlations between microbial activity and plant-biomass parameters were found, these are insufficient to clearly identify functional-trait differences between genotypes that control soil microbial functioning. Plants can control soil microbial activity and ultimately the decomposition of different soil organic matter pools via at least three non-exclusive mechanisms: (1) supplying oxygen to an otherwise anoxic soil system via root oxygen loss (Wolf et al., 2007), (2) competing with microbial communities for nutrients (Kuzyakov and Xu, 2013), and (3) supplying labile microbial substrates via rhizodeposition (Jones et al., 2004; Kuzyakov, 2002). Root oxygen loss (mechanism 1) is only relevant in oxygen-deficient soils, like those found in coastal marshes. This suggests that it might be the most important mechanism, but strong genotype effects on belowground litter decomposition were also present in our well-aerated monthly-flooding treatment (Fig. 4b). Therefore, root oxygen loss is unlikely to represent the primary and sole driver of the observed genotype effects. Differences in nutrient demand between genotypes (mechanism 2) are supported by the clear differences in aboveground biomass production (Reents et al., 2021) and soil microbial N-acquisition activities (Fig. 2). However, these differences in biomass production and microbial N-acquisition were restricted to our highest flooding frequency and cannot explain the changes in belowground litter decomposition we observed under lower flooding frequencies. We therefore hypothesize that genotypic differences in root exudation patterns (mechanism 3) could have played an important role in the studied system. Root exudates are a key component of the plant control on soil decomposition processes in terrestrial soils, and their quantity and quality are not necessarily related to plant biomass parameters (Henneron et al., 2020; Jones et al., 2004; Koelbener et al., 2010). Furthermore, differences in root-exudation patterns between genotypes are known to alter microbial community structures in terrestrial ecosystems (Micallef et al., 2009). For wetlands, however, the current understanding of root-exudate effects on soil decomposition dynamics is insufficient to explore this hypothesis more thoroughly without additional research (Dinter et al., 2019; Mueller et al., 2016). Taken together, our findings highlight the need for further investigations into rhizosphere-trait variability, plant–soil interactions, and the mechanisms of rhizosphere priming effects in wetland ecosystems.
We previously demonstrated realistic plant-productivity responses to variations in flooding frequency simulated by the tidal-tank facility at Hamburg University (Reents et al., 2021). Therefore, we argue that the present investigation on plant–soil interactions can also provide relevant mechanistic insight into flooding effects on tidal wetland functioning. However, owing to the artificial nature of the simulated tidal wetland system, absolute effect sizes reported here need to be considered with caution. For the same reason, we refrain from providing a detailed interpretation of changes in single microbial taxa. One important caveat in this context is the restriction of our study to a single soil type. Because plant–microbe interactions in the rhizosphere can reflect provenance (e.g., Di Lonardo et al., 2018), future investigations will need to assess the generality of our findings using different combinations of plant genotype and soil type, including the native home soils from the locations at which the plants are sampled. We furthermore recommend repeating this experiment in situ, e.g., in the form of reciprocal transplantations, in order to improve the quantitative understanding of plant-genotype-mediated SLR effects on soil microbial functioning.
Larger variability in microbial enzyme activities and litter decomposition in soils planted with the intolerant plant genotype support our general hypothesis that effects of changing abiotic conditions on soil microbial functioning depend on plant intraspecific genetic variation. Our findings suggest that intraspecific variation in wetland plants could represent an important factor determining the response of soil microbial communities and soil C turnover to climate change. If our findings apply more generally to coastal wetland ecosystems, they could yield important implications for experimental climate change research and models of soil C accumulation, because they show that plant-genotype controls can mask or enhance the effects of changing abiotic conditions on soil microbial processes. Future research will need to put more emphasis on the intraspecific variability in plant functional traits as well as climate-change-driven intraspecific genetic shifts in wetland plant communities.
All data presented in this paper are available upon reasonable request. Raw sequencing data are available at the European Nucleotide Archive (ENA) under BioProject accession number PRJEB38150 and sample accession numbers ERS4541081–ERS4541134.
The supplement related to this article is available online at:
HT, SR, SN, KJ, and PM designed and set up the experiment. HT conducted enzyme and decomposition assays and analyzed the resulting data. SL, PM, and FH planned the molecular microbial work. PM conducted the molecular microbial lab work. FH carried out the bioinformatics and analyzed the molecular data. HT and PM wrote the original draft with input from all co-authors.
The authors declare that they have no conflict of interest.
Publisher’s note: Copernicus Publications remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
We thank Chris Smit and his colleagues from Groningen University for the provision of plants; Max Beiße, Marion Klötzl, and Maren Winnacker at Hamburg University for assisting in the experimental phase; and Anke Saborowski at GFZ for her assistance with lab work.
Hao Tang received financial support from the China Scholarship Council (grant no. CSC201606910043). Peter Mueller was supported by the DAAD (German academic exchange service) PRIME fellowship program funded through the German Federal Ministry of Education and Research (BMBF).
This paper was edited by Michael Bahn and reviewed by two anonymous referees.