Invasive plants often successfully occupy large areas encompassing broad environmental gradients in their invaded range, yet how invader dominance and effects on ecological communities vary across the landscape has rarely been explored. Furthermore, while the impacts of invasion on plant communities are well studied, it is not well understood whether responses of aboveground (plant) and belowground (microbial) communities are coupled.

Here we test patterns in Phragmites australis (common reed) invasion in a field survey of eight sites situated across a salinity gradient, ranging from freshwater to saline marsh, in Southeast Louisiana. At each site, we surveyed plant composition and used metabarcoding methods to assess soil fungal and bacterial composition in plots within the dense Phragmites stand, in a transition zone of ~50:50 Phragmites:native plants, and in native-only areas. We hypothesized that Phragmites’ abundance and impact on above and belowground communities would vary across the salinity gradient and that the responses of above and belowground communities to invasion would be coupled.

We found weak evidence that invasion varied across the gradient: Phragmites stem densities increased slightly with salinity, and Phragmites increased aboveground litter accumulation more in fresh and saline areas compared to brackish. We found stronger evidence that plant and microbial responses to invasion varied with salinity. Phragmites strongly reduced native plant density across the gradient, with slightly greater reductions in fresh and saline areas. Plant species richness displayed consistent decreases with invasion across the salinity gradient; however, fungal and bacterial richness increased sharply with invasion only in brackish sites. Furthermore, the effect of Phragmites on plant and microbial community composition became stronger as salinity increased. Plants and microbes exhibited coupled responses to invasion in the magnitude of compositional shifts brought on by Phragmites, but Phragmites’ effects on richness were not coupled.

Synthesis. Overall, the variability in Phragmites impacts across the gradient, particularly soil microbial impacts, suggests that it may be difficult to generalize invader effects from single-site or single-ecosystem studies. However, above and belowground communities show some coupled responses to Phragmites; thus understanding plant community responses to invasion gives insight into impacts occurring belowground.

Eight field sites across SE Louisiana were surveyed, ranging from freshwater to saline marsh. We classified sites based on dominant vegetation. Freshwater marshes were Barataria Preserve and Turtle Cove Research Station. Intermediate/Brackish mashes were Pearl River WMA, Fontainebleau State Park, Big Branch NWR, and Bayou Sauvage NWR. Saline marshes were two sites at the Louisiana Universities Marine Consortium, LUMCON.

In 2017 at each site, 21 permanent 1 m² plots were established in three transects based on vegetation type: Phragmites stand, transition, and native, each with 7 plots. The Phragmites stand transect plots were located entirely within an area where Phragmites was highly dominant. The native plots ran parallel to, but outside of, the Phragmites stand and contained only native plants representing the native marsh community. The transition plots ran along the edge of the Phragmites stand, capturing the interface of the native community and the invading Phragmites front. Plots were spaced approximately 10 m apart. 

Plant species composition was sampled from September to November 2017 by performing stem counts of all plant species rooted in each 1 m² plot. For bunch grasses, each ramet (culm) was counted separately. Live biomass and litter mass per m² were estimated for each plot by harvesting from a 20 × 20 cm area, drying at 60°C for 48 hrs, and weighing.

Soil samples for microbial analysis were collected in October and November 2017. Soils were collected with a sterilized soil corer (5 cm diameter, approximately 10 cm depth), homogenized in a plastic bag, and a subsample collected in a sample tube. Samples were placed directly into a liquid nitrogen container in the field and transferred to a -80°C freezer upon returning to the lab.

DNA was extracted from the soil samples with a Qiagen DNeasy PowerSoil Kit. We used a dual-index, two step PCR approach. We first amplified the fungal ITS region (primers ITS1f/ITS2) and the bacterial 16S region (primers 515F/806R). PCR was done in duplicate and duplicates were pooled. We then performed a second PCR to attach barcodes and Illumina adaptors. Samples were purified and standardized with a SequalPrep kit (Invitrogen Inc.), and pooled into ITS and 16S libraries. Libraries were sequenced on two lanes of an Illumina Miseq v3 (300bp PE) by Duke Sequencing Core, Duke University, NC.

Sequence data were processed using the amplicon sequence variants (ASV) method in QIIME2 (Bolyen et al., 2019) and DADA2 (Callahan et al., 2016). We first trimmed reads where the median quality score fell below ~30, then quality-filtered the reads (no N’s, max expected errors 2, truncated at quality score 2, minimum read length 50) and denoised the data and joined paired reads using DADA2. We assigned taxonomy using a pre-trained Naïve Bayes classifier. The classifier was trained on the UNITE 8.2 database (Abarenkov et al., 2020) for ITS and Greengenes 13.8 (DeSantis et al., 2006) for 16S. Prior to analyses, both data sets were rarefied to an even sampling depth (fungi to 5334 reads and bacteria rarefied to 9316 reads per sample).

We used FUNGuild (Nguyen et al., 2016) to classify fungal taxa into functional guilds. We focused on taxa classified as “plant pathogen” and “arbuscular mycorrhizae” as there were sufficient taxa classified as such for these two guilds. For each of these two guilds, we summed the relative abundance of taxa classified as plant pathogens and AMF to yield total pathogen and AMF relative abundance per sample.

Phragmites haplotyping was performed at each site, and we found that the Phragmites at our freshwater and brackish sites was haplotype I, and the Phragmites at our saline sites was haplotype M1.

Some ITS and 16S samples did not sequence well and thus are NAs in this dataset.

One 16S sample had an extraordinarily high number of reads and a very high richness and Chao1. Thus we added a separate column omitting this sample from the column (the value is NA).

The Plant Pathogen relative abundance also contained two outliers, thus we added a separate column omitting these samples from the column (the value is NA).