The Missing Metric: An Evaluation of Fungal Importance in Wetland Assessments
To preserve wetland ecosystem function, federal and state agencies have developed assessment procedures to better manage remaining wetland areas. Currently, wetland assessments do not consider microorganisms when determining wetland quality. This is notable, because fungi are often the primary decomposers of organic material and thus important players in nutrient cycling. The objective of this study is to quantify how wetland quality, as measured using the Ohio Rapid Assessment Method (ORAM), relates to fungal community composition. We sampled soils from six depressional emergent marshes in Ohio belonging to each of the three ORAM quality categories, assessed soil physicochemical properties, and recovered fungal DNA. We then determined if wetland quality as expressed by the ORAM reflects soil health. Our results indicate that ORAM scoring methodology significantly explains differences in fungal community composition between wetlands. We also found that soil physicochemical properties not currently included in the ORAM are strong drivers of fungal community composition, particularly bulk density, pH, soil organic matter, and soil moisture. Overall, our results suggest fungal community composition reflects wetland quality as assessed by the ORAM, and that the ORAM and potentially other wetland assessments could better capture the soil environment by including easily measured soil physicochemical properties.
KeywordsExact Sequence Variant (ESV) Fungi Marsh Ohio Wetland Quality
To facilitate the regulation and restoration of wetlands, federal and state agencies established assessment procedures to monitor wetland quality (Fennessy et al. 2007). The “quality” of wetlands is largely determined by metrics that characterize their condition or the degree of anthropogenic disturbance (Mack 2001a). High quality, high functioning wetlands are generally considered to be pristine, having experienced little anthropogenic disturbance, and are capable of carrying out ecosystem services (Miller and Gunsalus 1999; Mack 2001a; Berglund and McEldowney 2008; Hruby 2014). In contrast, low quality wetlands are heavily degraded and as a result, are less functional and provide fewer ecosystem services than high quality wetlands.
To evaluate wetland quality, the United States Environmental Protection Agency recognizes three levels of assessment that vary in intensity and requisite expertise (Fennessy et al. 2007). Level 1 assessments are cursory evaluations that take little time and do not require a visit to the site in question. Level 3 assessments are in-depth field investigations that focus on one aspect of a wetland such as flora or fauna. Level 2, or rapid assessments, require 24 h or less to complete and incorporate several descriptive metrics that are tied to wetland functions.
Common rapid assessment metrics include wetland size, width of upland buffers, hydrology, and plant community composition (Miller and Gunsalus 1999; Mack 2001a; Berglund and McEldowney 2008; Hruby 2014). Wetland size or area can determine a wetland’s ability to serve as habitat for a range of wetland flora and fauna (Brown and Dinsmore 1986; Babbitt 2005). Consequently, larger wetlands score higher in wetland assessments although smaller wetlands may serve as habitat for unique species (Semlitsch and Bodie 1998; Mack 2001a). Upland buffers protect the functional integrity of the wetlands they separate from human disturbances and are typically scored using average buffer width, with wide buffer zones (≥ 50 m) providing greater nutrient filtration and support of flora and fauna (Castelle et al. 1994; Gilliam 1994; Burke and Gibbons 1995; Mack 2001a; Semlitsch and Bodie 2003; Stapanian et al. 2016). Wetland hydrology, assessed by water depth, hydroperiod, water sources, and hydrologic regime alterations (e.g. drainage tiles, ditches, dams), regulates soil anoxia, which in turn controls nutrient removal and carbon sequestration (Miller and Gunsalus 1999; Mack 2001a; Fisher and Acreman 2004; Kimmel et al. 2008; Kayranli et al. 2009; Hruby 2014). Finally, wetland vegetation, which is scored based on the number and interspersion of different vegetation classes (e.g. aquatic, emergent, scrub-shrub), plant diversity, and the abundance of native and invasive plants, plays an important role in nutrient cycling (Miller and Gunsalus 1999; Mack 2001a; Fisher and Acreman 2004; Kimmel et al. 2008; Kayranli et al. 2009; Hruby 2014). Increases in wetland plant functional diversity, for example, often lead to reduced methane emissions (Bouchard et al. 2007).
Despite its importance in wetland ecosystem function, few wetland assessments consider the soil environment. Assessments that currently include soil based metrics assess soil on a coarse scale considering only recent soil disturbances or by cursory assessments of soil type (e.g. organic or mineral) as opposed to any quantifiable soil parameter such as microorganism community structure, bulk density (BD), soil organic matter (SOM), or available soil nutrients (Mack 2001a; Hruby 2014). A need for incorporating soil based metrics into assessments was recently highlighted when it was shown that wetlands with similar quality scores, determined using the Ohio Rapid Assessment Method (ORAM), did not group together when soil physicochemical properties were considered (Rokosch et al. 2009). Additionally, certain soil metrics were identified to be promising metrics to include in assessments due to trends they exhibited with swamp quality scores. For instance, BD was negatively related to swamp quality and SOM was positively related to swamp quality, with higher quality swamps having less compact soils and more SOM than low quality swamps (Rokosch et al. 2009).
Here we propose strengthening current methods by incorporating soil microorganisms that regulate valued wetland functions such as nutrient cycling and the breakdown of harmful chemicals (Gutknecht et al. 2006; Faulwetter et al. 2009). Specifically, including measurements of fungi could greatly improve current assessment methods because fungi perform vital ecosystem functions as plant symbionts, saprobes, and pathogens (Blaney and Kotanen 2001; Thormann 2006; Smith and Read 2008; Neori and Agami 2017). Fungi that act as saprobes are the primary decomposers in both upland and wetland ecosystems and include white, brown, and soft rot fungi which degrade lignin, cellulose and hemicellulose contributing to the build-up of partially decomposed SOM in wetlands (Hibbett and Donoghue 2001; Thormann 2006). Mycorrhizal fungi, which exchange immobile soil nutrients such as phosphorous (P) and nitrogen (N) for carbon (C) from their plant hosts, comprise the largest group of fungal symbionts in wetlands and include arbuscular mycorrhizal (AM) and ectomycorrhizal (EM) fungi which associate with herbaceous and woody plants respectively (Smith and Read 2008; Neori and Agami 2017). Finally, fungi that act as plant pathogens in both upland and wetland ecosystems can structure plant communities due to differences in virulence, tolerance, and susceptibility in host identity and can be important in determining the establishment of invasive plants (Mitchell and Power 2003; Inderjit and van der Putten 2010; Rúa et al. 2011). By understanding the fungal functional guilds present in wetlands of differing quality, the potential functional capacity of the fungal community at each wetland type can be better understood allowing for improved land management.
The forces that determine the structure and function of belowground fungal communities include both biotic and abiotic properties, such as vegetation and soil physicochemical properties. For instance, the richness of saprotrophic and mutualistic fungi typically increases with increasing plant richness, possibly as a product of increased plant richness generating a more heterogeneous soil environment allowing for greater resource partitioning in below-ground communities (Wardle 2006; Hiiesalu et al. 2014; Hiiesalu et al. 2017). Also, soil properties such as soil nutrient availability (particularly N and P), SOM, and pH are important determinants of fungal community richness and composition, with differences in EM fungi and saprotroph richness explained by the amount of SOM, C:N ratios, and pH (Erlandson et al. 2016; Glassman et al. 2017; Hiiesalu et al. 2017).
The primary objective of this study is to quantify how the quality of freshwater depressional marshes, as measured using the ORAM, relates to fungal community composition as a first step in determining if current assessment methods need to be adjusted to account for microbial communities as proxies for healthy ecosystems. The ORAM, which is comprised of six metrics: (1) wetland size, (2) upland buffer width and surrounding land use, (3) hydrology, (4) substrate disturbance and habitat development, (5) special wetland status (e.g. bogs, fens, etc.), and (6) wetland vegetation, is used in Ohio to rapidly assess wetland quality. The ORAM scores wetlands on a scale of 0 to 100 based on the six metrics and categorizes them into one of three quality categories: Category 1, Category 2, and Category 3 (Mack 2001a). Category 1 wetlands are low quality (highest disturbance and lowest functional value) and are afforded little regulatory protection by the Ohio Environmental Protection Agency (OH EPA). Category 3 wetlands are high quality (lowest disturbance and highest functional value) and are a priority for protection by the OH EPA.
To achieve this objective, we recovered soil fungal communities from six natural marshes within the state of Ohio that represent the three ORAM quality categories. Using the Illumina MiSeq platform to sequence soil DNA, we determined the diversity and composition of the fungal community at each marsh. We used this data to address the central hypothesis that marshes with higher quality ratings will have more diverse fungal communities that differ in composition from lower quality marshes. We additionally identified the role of soil physicochemical properties and vegetation in structuring fungal communities, with particular emphasis on soil properties which are seldom included in wetland assessments. Together we combine these data as an important first step in determining if current wetland assessment frameworks would benefit from the addition of soil based metrics.
Descriptions of the six marshes included in the study. Note that there was good agreement between our ORAM scores and scores from a previous OH EPA assessment which evaluated four of the six marshes
40.250o N, 83.800o W
40.585o N, 83.224o W
39.863o N, 82.620o W
39.834o N, 82.725o W
39.583o N, 83.001o W
41.649o N, 80.884o W
Surrounding Land Use
Abandoned Farmland/ Wildlife Area
Walkill Silt Loam
Milford Silt Clay Loam
Pewamo Silt Clay Loam
Caneadea-Canadice Silt Loam
Maximum Water Depth
Dominant Plant Communities
S1: Phalaris arundinacea
S2: Leersia oryzoides
S1: Phalaris arundinacea
S2: Leersia oryzoides / Eleocharis spp.
S1: Typha latifolia
S2: Scirpus fluviatilis
S3: Polygonum amphibium /Urtica dioica
S1: Polygonum amphibium
S2: Wolffia spp. / Lemna spp.
S1: Typha latifolia
S2: Nuphar advena
S3: Sparganium eurycarpum / Cephalanthus occidentalis / Typha latifolia
S1: Nuphar advena
S2: Juncus effusus /Dulichium arundinaceum
To account for within-marsh variation due to environmental gradients, we established soil sampling stations within each marsh following a stratified random design. Using data generated from the vegetation surveys, we divided each site into two or three strata delineated by dominant plant communities using the hclust function from the Cluster package (Maechler et al., 2017; Table 1). From each stratum, we randomly selected five of the surveyed quadrats for a total of 10 or 15 soil sampling stations (quadrats) per marsh and a total of 70 sampling stations in the study.
Soil Core Sampling
To measure soil physicochemical properties, we collected soil cores in July, August, and September of 2017 using PVC soil corers (11.5 cm depth × 10 cm diameter). From each sampling station, we collected two soil cores from opposing corners of the quadrat for a total of 140 cores. Cores were placed on ice; transported back to the lab for immediate determination of soil moisture and bulk density; and then stored at 4 °C for three months until further processing.
For extraction of DNA from soil, we collected an additional soil core from the center of each quadrat using a PVC corer (11.5 cm depth × 10 cm diameter) for a total of 70 cores across all six marshes. We subsampled each DNA core in the field by homogenizing the soil and packing two 1.5 mL centrifuge tubes with approximately 1 g of soil from the homogenized core. Soil subsamples were flash frozen in liquid nitrogen, placed on dry ice, and transported back to the lab for immediate DNA extraction (see below). The remainder of each soil subsample was stored at -80 °C.
Soil Physicochemical Properties
In the lab, we measured BD and gravimetric water content (soil moisture) by drying 20 g of field moist soil (105 °C, 24 h). We measured soil pH using a modified version of the protocol described by Tan (2005). Using a 1:4 ratio of field moist soil to water, we stirred the soil slurry at 80 rpm for 15 min on an orbital shaker (Lab-Line Instruments Inc., Melrose Park, IL) and measured the pH of the slurry using a Beckman Coulter Φ360 pH/Temp/mV meter (Brea, CA).
The remaining soil was air dried for four weeks, ground, passed through a 2 mm sieve (No. 10), and pooled per sampling station. A portion of the ground soil was submitted to Brookside Laboratories (New Bremen, OH) for measurement of SOM (loss on ignition 360 °C), Mehlic III extractable P, NO3-N, and NH4-N. The remaining air-dried soil was passed through a 0.212 mm sieve (No. 70) and submitted to Washington State University Stable Isotope Core Laboratory (Pullman, WA) for C:N analyses using an elemental analyzer (ECS 4010, Costech Analytical, Valencia, CA).
Soil Fungal DNA Extraction
We extracted DNA in triplicate for each sampling station with 25 μL DI water acting as a negative control. Within 24 h of the initial soil core sampling, we extracted DNA from 0.25 g of soil with the DNeasy PowerSoil kit (Qiagen, Carlsbad, CA) using the default protocol. DNA extracts were stored at -80 °C for four months, then pooled per sampling station and purified using a modified protocol of the DNeasy PowerClean Cleanup kit (Qiagen, Carlsbad, CA). Specifically, after cleaning 150 μL of pooled extract by washing with solution CB, we added 500 μL of undiluted ethanol to pooled DNA extracts in MB spin columns (Qiagen, Carlsbad, CA). Following incubation for 5 min at room temperature, the MB spin columns were centrifuged for 30 s at 10,000×g. We discarded the flow through and repeated the previous step. The MB spin columns were then centrifuged for 4 min at 10,000×g to remove any residual ethanol before proceeding with the remainder of the protocol.
We diluted cleaned DNA extracts to 5 ng/μL DNA by adding Molecular Biology Grade Water to a volume of 90 μL (IBI Scientific, Dubuque, IA) and submitted samples to the Ohio State University Molecular and Cellular Imaging Center (MCIC; Wooster, OH) for library preparation and Illumina sequencing using the MiSeq platform.
Library Preparation and Sequencing
The MCIC amplified the ITS1 locus using ITS1F and ITS2 PCR primers with an added heterogeneity spacer to compensate for the low nucleotide diversity of the amplicon (Smith and Peay 2014). Adapters, containing Nextera indices, were ligated to sequences during PCR for sample indexing. Samples were amplified in two rounds, the first to amplify the DNA and attach a portion of the Illumina adapter sequence and the second to complete the adapter sequence. The following steps were carried out on the Eppendorf epMotion5075 automated liquid handler (Hauppauge, NY). In the first round (PCR 1), 25 ng of genomic DNA was amplified using the following conditions: initial denaturation at 96 °C for 3 min, 25 cycles of 96 °C for 30 s for denaturation, 55 °C for 30 s for annealing, and 72 °C for 30 s for elongation. The second round of PCR was conducted using 3 μL of clean PCR 1 product. PCR conditions for round 2 were the same as PCR 1 except 8 cycles were performed rather than 25 cycles. After each round of PCR, samples were cleaned using the Agencourt AMPure XP beads (Beckman Coulter Life Sciences). Purified amplicon libraries were quantified and pooled by plate at equimolar ratios before sequencing. The final pools were purified using the Pippin Prep size selection system (Sage Science; Beverly, MA) to discard primer dimers.
Amplicon libraries were sequenced using the Illumina MiSeq sequencing platform at a final concentration of 14.3 pM. PhiX was mixed in the pool of amplicon libraries for the sequencing run (expected at 20%). The run was clustered to a density of 681 ± k/mm2 and the libraries were sequenced using a 300PE MiSeq sequencing kit with the standard Illumina sequencing primers. Image analysis, base calling, and quality assessment were performed on the MiSeq instrument. The resulting sequences were demultiplexed and adapters were removed.
Resulting sequences were processed using the bioinformatics pipeline Quantitative Insights into Microbial Ecology 2 (QIIME2; http://qiime2.org; Caporaso et al. 2010). We removed heterogeneity spacers from sequences using the cutadapt trim-paired function of the cutadapt QIIME2 plugin (Martin 2011). We used the dada2 denoise-paired function of the DADA2 plugin to denoise, remove chimeras, merge paired end reads, and identify exact sequence variants (ESV), a proxy for species (Callahan et al. 2016). Using the feature-table group function of the feature-table plugin, we pooled the ESVs by strata. We assigned taxonomy using the feature-classifier classify-sklearn function with a 0.70 confidence threshold (Pedregosa et al. 2011) from the feature-classifier plugin (Bokulich et al. 2018). We trained the RDP classifier with the UNITE v. 7.2 database ( https://doi.org/10.15156/BIO/587481; Kōljalg et al. 2013) using the feature-classifier fit-classifier-naïve-bayes function of the feature-classifier plugin. ESVs assigned to at least a fungal phylum were retained. We then rarefied the resulting ESV table and representative sequences to the lowest number of sequences observed from all strata using the feature-table rarefy function (Fig. S1) from the feature-table plugin to standardize statistical analyses.
Fungal Functional Guilds
We assigned ESVs to a functional guild using the online version of FUNGuild with a confidence cut-off of “Possible” (http://funguild.org, accessed 17 June, 2019; Nguyen et al. 2016a). For any ESV not assigned to a functional guild but classified to the family, genus, or species level, we assigned the most probable functional guild or guilds based on the literature. We then calculated the relative abundance of each functional guild at the stratum level. For ESVs assigned to multiple functional guilds, we equally allocated sequences to each of the assigned functional guilds so that each possible function of the ESV was equally weighted and patterns could be more easily discerned. We considered both “narrow” functional guilds (e.g. AM fungi, white rot, plant pathogen, etc.) and “broad” functional guilds (e.g. saprotroph, pathogen, endophyte, etc.).
We performed all statistical analyses in R version 3.4.4 (R Core Team 2018). We fit all linear mixed effects models using the lme function from the nlme package (Pinheiro et al. 2018). To meet the assumptions of normality and homoscedasticity the following transformations were performed: soft rot, saprobe, and pathogen richness were log transformed; and white rot, EM, and AM richness were square root transformed. Ref. "R Core Team 2018" is cited in the body but its bibliographic information is missing. Kindly provide its bibliographic information in the list.R Core Team (2018) R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL http://www.R-project.org/.
Using per stratum raw ESV abundance data, we calculated Shannon diversity indices, observed richness, and Chao1 richness using the diversity and the specnumber functions from the vegan package (Oksanen et al. 2018) and chao1 function from the fossil package (Vavrek 2011). We used linear mixed effects models to test if ORAM score significantly affects Shannon diversity, observed richness, Chao1 richness, and the richness of broad and narrow functional guilds with site as a random effect. Linear mixed effects models were also used to quantify the relationships between ORAM score and individual soil variables (e.g. BD, pH, soil moisture, Mehlich III extractable P, NO3−, NH4+, N, C, and C:N) with site as a random effect.
We then calculated relative ESV abundances per stratum by dividing the number of sequences per ESV in a stratum by the total number of sequences in the stratum. We visually assessed the ability of the ORAM to group wetlands based on fungal communities and fungal functional guilds using relative ESV abundances with non-metric multidimensional scaling (NMDS) using the metaMDS function from vegan (Oksanen et al. 2018). We used NMDS ordination for these assessments because it allowed use of non-Euclidean distances (e.g. Bray-Curtis), which are preferred for comparison based on community composition (Faith et al. 1987). To identify indicator taxa, we used the multipatt function of the indicspecies package at the site level (De Caceres and Jansen 2016). To visually assess the ability of the ORAM to group marshes based on soil physicochemical properties we performed a principal component analysis (PCA) using the prcomp and fviz_pca_biplot functions from the factoextra package (Kassambara and Mundt 2017). We used the Euclidean-based PCA ordination for this assessment following common methodology for environmental data (Janžekovič and Novak 2012). Ref. "De Caceres and Jansen 2016" is cited in the body but its bibliographic information is missing. Kindly provide its bibliographic information in the list.De Caceres, M, & Jansen, F (2016) Indicspecies: relationship between species and groups of sites. R package version, 1(6).
We then calculated weighted relative ESV abundances, weighted relative vegetation cover, and weighted averages of soil physicochemical properties per site using strata weights (i.e. number of surveyed vegetation quadrats belonging to a stratum divided by the total number of surveyed quadrats in a marsh). We tested if ORAM and component metric scores significantly explained differences in fungal community composition, fungal functional guild composition, and soil physicochemical properties using permutational multivariate analysis of variance (PERMANOVA) using the adonis function from vegan (Oksanen et al. 2018) with 720 permutations (maximum possible).
We assessed the impacts of soil properties (averaged per stratum) and plant communities (relative cover per stratum) on (1) fungal communities and (2) functional guilds by canonical correspondence analyses (CCA) using the cca function from vegan (Oksanen et al. 2018). CCA is a constrained ordination technique that assumes a unimodal relationship between the set of response variables (e.g. ESV abundance) and the set of predictor variables (e.g. soil properties). We then tested for statistical significance with the anova.cca function from vegan.
We recovered 6,115,790 sequences from all sampling sites and retained 248,712 sequences after quality filtering and chimera removal with DADA2. Sequence data are available at NCBI Sequence Read Archive: PRJNA525991 and filed under accession numbers SAMN11081749 - SAMN1108197. Of the sequences that passed quality filtering, 222,241 sequences were identified as fungi with 135,696 sequences being assigned to a fungal phylum using the UNITE database. We rarefied strata to a sampling depth of 2861 sequences per strata yielding 40,054 sequences across 759 unique ESVs (Fig. S1).
Fungal Community Composition
Ascomycota was the most abundant phylum and comprised 47% of the total fungal community, followed by Basidiomycota which comprised 45% of the community (Fig. S2). The Blastocladiomycota, Chytridiomycota, Entomophthoromycota, Entorrhizomycota, Glomeromycota, Kickxellomycota, Monoblepharomycota, Mortierllomycota, Mucoromycota, Rozellomycota together comprised the remaining 8% of the fungal community.
Fungal Functional Guilds
A total of 441 ESVs out of 759 ESVs were assigned to a functional guild by FUNGuild; the classifiable ESVs ranged from 51% at Category 3B to 88% at Category 1A. At all six marshes, saprotroph was the most abundant broad functional guild comprising approximately 68% of the total classifiable ESVs, with undefined saprotrophs being the most abundant narrow functional guild comprising approximately 39% of the classifiable community (Fig. 1b).
Marsh Quality and Fungal Communities
Soil Physicochemical Properties and ORAM Score
Soil Physicochemical Properties and Fungal Communities
Vegetation Effects on Fungal Community Composition
Wetland assessments use easily evaluated properties to score the condition and by proxy, the functional capacity of wetlands (Miller and Gunsalus 1999; Mack 2001a; Berglund and McEldowney 2008; Hruby 2014). Current assessment procedures do not consider soil microorganisms despite their importance in regulating and carrying out valuable wetland functions (Gutknecht et al. 2006; Faulwetter et al. 2009). As an important first step in determining the need and feasibility of adjusting current wetland assessment frameworks to consider microbial communities, we assessed the relationship between the ORAM, and fungal community composition and diversity. Our data suggests that current assessment methods are capable of distinguishing fungal communities of high quality marshes from low quality marshes; however, current assessments of wetland quality do not distinguish fungal functional guild composition. We further provide evidence that soil physicochemical properties are important determinants of marsh quality and have a role in structuring fungal communities and functional guild composition in marshes, alongside vegetation.
Fungal communities in all six marshes were dominated by saprotrophs (Fig. 1b), highlighting the importance of fungi as regulators of SOM accumulation and C storage in marshes (Thormann 2006). Across quality categories, white rot fungi were the most abundant specific guild of saprotrophic fungi recovered while soft rot fungi were also highly represented (Fig. 1b). The high abundance of white and soft rot fungi indicates an abundance of recalcitrant C sources such as lignin, cellulose, and hemicellulose which they degrade (Hibbett and Donoghue 2001). Brown rot fungi were not specifically recovered, although they may exist as a part of the undefined saprotroph category but this may reflect the lack of conifers with which brown rot fungi are tightly associated (Hibbett and Donoghue 2001).
In contrast to saprotrophs, our study recovered a relatively low abundance of mycorrhizal ESVs (Fig. 1b). This lack of mycorrhizal fungi may be a product of the time of year of the study which took place during the summer when the vegetation present at each site was already mature. Levels of AM colonization in wetlands are seasonal, most likely tied to plant phenology, with colonization levels being the greatest during the spring and the lowest during late summer (Bohrer et al. 2004). Had we sampled in spring or early summer during periods of new vegetative growth, we may have recovered more mycorrhizal ESVs and fewer saprotroph ESVs. Also, the lack of EM fungi in this study could be attributed to the dominance of emergent herbaceous plants at each marsh (Table 1). EM fungi more commonly associate with woody plant species, and thus had this study been conducted in wetlands dominated by trees or shrubs, such as members of the Salicaceae, we could expect to observe more EM fungi (Smith and Read 2008; Erlandson et al. 2018). Ref. "Bohrer et al. 2004" is cited in the body but its bibliographic information is missing. Kindly provide its bibliographic information in the list.Bohrer, KE, Friese, CF, & Amon, JP (2004) Seasonal dynamics of arbuscular mycorrhizal fungi in differing wetland habitats. Mycorrhiza, 14(5): 329-337.
Our central hypothesis that the fungal communities would differ based on quality and be more diverse in higher quality marshes was only partially supported. While fungal diversity and functional guild composition were not impacted by marsh quality (Figs. 2a and S3), fungal richness significantly declined with quality and fungal composition was significantly explained by marsh quality (Figs. 2b,c and 3). The ability of the ORAM to describe differences in fungal community composition is likely tied to the impacts of anthropogenic disturbances on soil physicochemical properties. For instance, wetland soil bacterial communities were observed to differ based on land use type (i.e. farmed, restored, undisturbed reference, etc.) as a product of differences in soil chemistry, particularly soil pH and N (Hartman et al. 2008). In our study, there were particularly strong relationships between fungal community composition, substrate disturbance (Metric 4), and soil physicochemical properties. Soil properties also significantly distinguished the fungal functional guild communities (Fig. S4), which were not separable by ORAM score (Fig. S3).
Differences in soil physicochemical properties between the marshes in our study were significantly explained by ORAM score (Fig. 4); however in swamps, the ORAM was not able to describe differences in soil physicochemical properties (Rokosch et al. 2009), indicating additional research is required to determine the ability of wetland assessments to describe the soil environment across wetland types. One possible way to improve the relationship between wetland assessments and the soil environment is to include easily measured soil physicochemical properties that respond consistently across multiple wetland types to disturbance gradients, and also serve as important regulators of microbial communities. For instance, SOM, BD, and soil moisture exhibited trends with wetland quality in marshes (P = 0.089, P = 0.066, and P = 0.070) and in swamps (Rokosch et al. 2009), and were also strong drivers of fungal community composition and functional guild composition (Figs. 5 and S4). The potential utility of these soil measures in wetland assessments is not only highlighted by their importance in determining the structure of soil fungal communities, but also because similar trends with wetland quality were observed in different wetland types. This suggests that SOM, BD, and soil moisture, three easily measured soil properties, would serve as useful quantitative measures to incorporate into wetland quality assessments.
Another soil physicochemical property with the potential to serve as a useful measure in wetland assessments is soil pH. Soil pH had a significant negative relationship with ORAM score (P = 0.042), and was important in determining fungal community structure. This fits in line with the growing number of studies reporting soil pH as an important determinant in both the structure and richness of fungal communities (Tedersoo et al. 2014; Hiiesalu et al. 2017; Erlandson et al. 2018). It should be noted, however, that soil pH was more important in distinguishing communities within a quality category rather than between quality categories, particularly the two Category 3 marshes (Fig. 5) and that soil pH was not impacted by swamp quality in Ohio (Rokosch et al. 2009). This suggests that further research may be required to determine if soil pH is a useful quantitative measure to include in wetland assessments for multiple wetland types or for only freshwater depressional marshes, such as those investigated in this study.
Vegetation significantly explained differences in fungal community composition (Fig. 6) but was not significant for fungal functional guild composition (Fig. S5). The weak explanatory power of vegetation for fungal functional guild is likely due to the dominance of saprotrophs which are less dependent upon living vegetation, and a lack of mycorrhizal ESVs which tend to be more tightly associated with their plant hosts (Nguyen et al. 2016b; Fig. 1b). Additionally, we did determine that vegetation quality, measured both by Metric 6 and FQAI score, explained differences in fungal community composition. Specifically, we observed that plant species with higher C of C scores are tightly associated with the fungal communities of high quality marshes. While Metric 6 is capable of explaining differences in fungal community composition, the metric could be made more quantitative through the inclusion of FQAI scores. Although this would require assessors to have plant identification expertise and may increase the requisite experience needed to complete the assessment, it would reduce scorer subjectivity and strengthen the ability of assessments to capture wetland function.
Due to the pivotal role of soil fungi in the regulation of key wetland functions (Thormann 2006; Neori and Agami 2017), we suggest that additional research be conducted to examine the ability of the ORAM and other wetland assessment methods to capture differences in these valuable wetland components. Specifically, we propose additional research be done to identify soil physicochemical properties that respond consistently, across wetland types, to disturbance gradients and are important drivers of microbial communities. Other studies have similarly observed a strong link between soil physicochemical properties and fungal community composition (Erlandson et al. 2016; Glassman et al. 2017; Hiiesalu et al. 2017), highlighting the potential utility of these properties as additional quantitative measures for wetland quality. The identification and inclusion of quantitative measures of soil into assessments will serve only to bolster the ability of wetland quality assessments to serve as proxies for wetland function (Miller and Gunsalus 1999; Mack 2001a; Berglund and McEldowney 2008; Hruby 2014).
We would like to thank Scott Kelsey, Shante Eisele, Michaela Woods, Adam Reed, Molly Simonis, Ashlea Kennedy, Kayty Wintermute, Rosa Tweed, Kayla Watters, and Drew Sheaffer for their help in the field and the lab. We would also like to thank the OH EPA for providing data on previously scored wetlands; the Columbus Audubon Society, Ohio Department of Natural Resources, The Nature Conservancy in Ohio, and all the private landowners who granted us access to the marshes; and the anonymous reviewers, whose comments and suggestions helped improve this manuscript. This project was supported through funding from the Wright State University Department of Biological Sciences.
- Andreas BK, Mack JJ, McCormac JS (2004) Floristic quality assessment index (FQAI) for vascular plants and mosses for the state of Ohio, Division of Surface Water Ohio Environmental Protection Agency. Wetland Ecology Group, ColumbusGoogle Scholar
- Berglund J, McEldowney R (2008) MDT montana wetland assessment method. Montana Department of Transportation, HelenaGoogle Scholar
- Bohrer, KE, Friese, CF, & Amon, JP (2004) Seasonal dynamics of arbuscular mycorrhizal fungi in differing wetland habitats. Mycorrhiza, 14(5): 329-337.Google Scholar
- Bokulich NA, Kaehler BD, Rideout JR, Dillon M, Bolyen E, Knight R, Huttley GA, Caporaso JG (2018) Optimizing taxonomic classification of marker-gene amplicon sequences with QIIME 2’s q2-feature-classifier plugin. Microbiome 6:90Google Scholar
- De Caceres, M, & Jansen, F (2016) Indicspecies: relationship between species and groups of sites. R package version, 1(6).Google Scholar
- Fennessy MS, Jacobs AD, Kentula ME (2007) An evaluation of rapid methods for assessing the ecological condition of wetlands. Wetlands 27:543–560. https://doi.org/10.1672/0277-5212(2007)27[543:Aeormf]2.0.Co;2 CrossRefGoogle Scholar
- Hruby T (2014) Washington state wetland rating system for western washington: 2014 update. Washington Department of Ecology, OlympiaGoogle Scholar
- Janžekovič F and Novak T (2012) PCA–a powerful method for analyze ecological niches. Principal component analysis–multidisciplinary applications:127–142Google Scholar
- Kassambara A and Mundt F (2017) factoextra: Extract and Visualize the Results of Multivariate Data Analyses. R package version 1.0.5. https://CRAN.R-project.org/package=factoextra
- Mack JJ (2001a) Ohio rapid assessment method for wetlands v. 5.0 user's manual and scoring forms, Division of Surface Water Ohio Environmental Protection Agency. Wetland Ecology Group, ColumbusGoogle Scholar
- Mack JJ (2001b) ORAM v. 5.0 quantitative score calibration, Division of Surface Water Ohio Environmental Protection Agency. Wetland Ecology Group, ColumbusGoogle Scholar
- Mack JJ (2004) Integrated wetland assessment program. Part 4: Vegetation index of biotic integrity (vibi) and tiered aquatic life uses (talus) for ohio wetlands. Ohio EPA technical report wet/2004–4., Division of Surface Water Ohio Environmental Protection Agency, Wetland Ecology Group, Columbus, OhioGoogle Scholar
- Maechler M, Rousseeuw P, Struyf A, Hubert M, Hornik K (2017) cluster: Cluster Analysis Basics and Extensions. R package version 2.0.6Google Scholar
- Magee TK, Dwire KA, Holland CC, Gwin SE, Gibson RG, Honea JE, Shaffer PW, Sifneos JC, Kentula ME (1993) Quality assurance project plan for the Oregon wetlands study. Epa/600/r-93/221. Environmental Research Laboratory U.S. Environmental Protection Agency, CorvallisGoogle Scholar
- Miller RE and Gunsalus BE (1999) Wetland rapid assessment procedure (WRAP), Regulation Department Natural Resource Management Division, South Florida Water Management DistrictGoogle Scholar
- Nguyen NH, Williams LJ, Vincent JB, Stefanski A, Cavender-Bares J, Messier C, Paquette A, Gravel D, Reich PB, Kennedy PG (2016b) Ectomycorrhizal fungal diversity and saprotrophic fungal diversity are linked to different tree community attributes in a field-based tree experiment. Mol Ecol 25:4032–4046. https://doi.org/10.1111/mec.13719 CrossRefPubMedGoogle Scholar
- Oksanen J, Blanchet FG, Friendly M, Kindt R, Legendre P, McGlinn D, Minchin PR, O'Hara RB, Simpson GL, Solymos P, Stevens MHH, Szoecs E and Wagner H (2018) vegan: Community Ecology Package. R package version 2.4–6. https://CRAN.R-project.org/package=vegan
- Pedregosa F, Varoquaux G, Gramfort A, Michel V, Thirion B, Grisel O, Blondel M, Prettenhofer P, Weiss R, Dubourg V (2011) Scikit-learn: Machine learning in Python. J Mach Learn Res 12:2825–2830Google Scholar
- Pinheiro J, Bates D, DebRoy S, Sarkar D, authors E, Heisterkamp S, Van Willigen B and R-core (2018) _nlme:Linear and Nonlinear Mixed Effects Models_. R package version 3.1–131.1, <URL: https://CRAN.R-project.org/package=nlme>
- R Core Team (2018) R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL http://www.R-project.org/.Google Scholar
- Smith S, Read D (2008) Mycorrhizal symbiosis. Academic Press, New YorkGoogle Scholar
- Tan KH (2005) Soil sampling, preparation, and analysis. CRC press, Boca RatonGoogle Scholar
- Tedersoo L, Bahram M, Polme S, Koljalg U, Yorou NS, Wijesundera R, Villarreal Ruiz L, Vasco-Palacios AM, Thu PQ, Suija A, Smith ME, Sharp C, Saluveer E, Saitta A, Rosas M, Riit T, Ratkowsky D, Pritsch K, Poldmaa K, Piepenbring M, Phosri C, Peterson M, Parts K, Partel K, Otsing E, Nouhra E, Njouonkou AL, Nilsson RH, Morgado LN, Mayor J, May TW, Majuakim L, Lodge DJ, Lee SS, Larsson KH, Kohout P, Hosaka K, Hiiesalu I, Henkel TW, Harend H, Guo LD, Greslebin A, Grelet G, Geml J, Gates G, Dunstan W, Dunk C, Drenkhan R, Dearnaley J, De Kesel A, Dang T, Chen X, Buegger F, Brearley FQ, Bonito G, Anslan S, Abell S, Abarenkov K (2014) Fungal biogeography. Global diversity and geography of soil fungi. Science 346:1256688. https://doi.org/10.1126/science.1256688 CrossRefGoogle Scholar
- Vavrek MJ (2011) Fossil: Palaeoecological and palaeogeographical analysis tools. Palaeontol Electron 14:16Google Scholar
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.