We collected four random samples of water hyacinth and associated invertebrates per site using a custom-built ¼-m2 quadrat with a 200-µm mesh pouch . For each sampling event, we scooped the sampler into the water column beneath the water hyacinth mat then brought the sampler upward, through the mat. This approach allowed us to sample the water column, the epiphytic invertebrates found on the water hyacinth roots, and those on above-water plant structures. In the field, we rinsed invertebrates from the interior of the mesh as well as from the plant material. We bagged the rinsed water hyacinth, preserved the invertebrates in 70% ethanol, and transported them to the laboratory for processing.In the laboratory, we dried water hyacinth material at 60°C , and we recorded dry plant biomass. In our sub-sampling protocol for invertebrates, we sorted the invertebrates by size class by rinsing them through stacked sieves . We completely counted all invertebrates ≥1mm. We transferred invertebrates <1mm to a pan and evenly distributed them across a numbered grid. Using a random number generator, commercial hydroponic systems we selected 10 grid cells, using a pipette to extract the contents of those cells. We counted and identified invertebrates to genus, then extrapolated total sample numbers from sub-sample counts.
To assess differences in invertebrate abundance, we natural-log transformed the data to achieve normality, and compared invertebrate abundance per gram plant biomass before and after treatment, as well as treatment versus control locations using mixed effects generalized linear models , with bootstrapped confidence intervals . We chose to analyze the invertebrate abundance per gram plant biomass because other studies have shown that increased macrophyte biomass can support higher densities of invertebrates . This allowed us to control for differences in water hyacinth biomass between sites and sampling dates to detect differences from herbicide treatment. We used the same method of GLMMs to test whether plant biomass changed before and after treatment, and at control versus treated sites. To see whether invertebrates per ¼-m2 quadrat of water surface area were affected by the combination of herbicide and change in plant biomass, we compared invertebrates per sample using GLMMs as well. To determine whether particular taxa were affected differentially, we used GLMMs to perform univariate comparisons of taxa-specific abundance before and after treatment, in addition to control and treatment locations. We also divided the sampled invertebrates into categories, including zooplankton, gill-breathing arthropods, air-breathing arthropods, and mollusks; and used GLMMs to assess any differences in trait- and size-groups. All p-values were judged significant based on Tukey–Kramer honestly significant difference test for multiple post-hoc comparisons.
To assess differences in invertebrate communities, we tested for overall differences in community composition using permutational multivariate analysis of variance . We performed non-metric, multi-dimensional scaling ordination to visually display differences in invertebrate community diversity. We then used GLMMs to compare the following: average richness and average percent dominant taxa . To further evaluate potential differences in invertebrate communities, we compared diversity indices among times and treatment, including Jaccard Index and Sorenson’s Coefficient of Community Similarity .On average, there was 22.0g more plant biomass per sample at the treated sites than at the control sites . The effect of time alone was small and not significant, with only 6.9g less plant biomass on average in the early-season samples relative to the late season samples . The modeled interaction between time and treatment showed that there was 22.3g more biomass per sample at treatment locations in the late-season samples . Plant biomass from treated sites included fragmented, dead, and dying material as well as the remaining portions of living plants. Fragmented pieces of dead and dying water hyacinth accumulated in layers as a result of wind and tidal action, which may partially account for this surprising finding.On average, we observed increased plant biomass per sample at both the control and the treatment locations in the late-season samples. The increase in biomass at treatment locations likely occurred because some plant material persists in growing between decomposing treated strips. The growing plant material condenses the dying material, which results in vertical layers of dead, dying, and living water hyacinth. Additionally, as treated water hyacinth’s structural elements degrade, wind and tidal action can crowd plant fragments together in the mats, making them denser in fibrous material.
Given that annual aquatic macrophytes typically experience a seasonal peak in plant biomass in the mid- to late-summer and that our after treatment samples were taken between June and August, it is not particularly surprising that the after treatment samples had higher biomass. Subsequent assessments of water hyacinth coverage in the Delta have shown variable trends between years. For example, using Landsat imaging and the Water Hyacinth Mapper Tool, scientists from NASA–Ames reported a 32% decline in 2017 annual peak acreage covered with floating aquatic vegetation compared with 2015 values . However, provisional 2019 data from the same source indicates an increase in water hyacinth acreage compared with values at the same time in 2018 . Longer-term studies are needed to assess overall WHCP effectiveness. In general, we observed more invertebrates per gram plant biomass after the treatment date, though these results were not significant. This increase can likely be attributed to several cumulative influences. For example, as discussed above, more dense plant mats can provide increased surface area for epiphytic invertebrate colonization. Also, several studies indicate strong seasonal differences in aquatic invertebrate abundance , with higher abundance occurring during the summer dry season in Mediterranean climate regions like the Delta. Therefore, it is likely that the increase in plant biomass as well as invertebrate density is a function of season and plant/invertebrate phenology. This increase in invertebrate abundance may help provide food resources for native fishes, particularly given the high abundance of highly nutritious amphipods and insects found in our samples . Since food limitation is considered a factor in many native fish declines in the Delta , finding slightly higher invertebrate abundance in the late-season samples is an encouraging sign for adaptive management efforts that balance ecosystem productivity with water hyacinth control.We did not observe a difference in richness or percent dominant taxa between control and treatment sites before versus after the treatment date. If there were treatment effects, we would expect to also see differences in the members of the invertebrate community—i.e., taxa that are more tolerant of low oxygen, or more detritivores. However, our lack of evidence for such community differences, and the observed similarities in Jaccard Index and Soreson’s Coefficient of Community Similarity, lead us to conclude that any treatment effects are not great enough to significantly alter the invertebrate community richness or dominance measures. The lack of hull separation in NMDS ordination for control and treatment locations, and the lack of significant PERMANOVA results for control versus treatment, further strengthens our finding that the aquatic invertebrate communities sampled in water hyacinth are largely similar between control and treatment locations. However, non-significant phenological differences exist in community composition before versus after the treatment date, particularly for invertebrate abundance.Of the most common and abundant taxa, ostracods are the only taxa for which abundance was significantly greater after the treatment date. Their increased relative abundance is likely a seasonally driven phenomenon . Planorbella spp., Zygoptera nymphs, and N. bruchi are functionally important taxa that were generally rare relative to many other observed taxa. As for ostracods, we observed significant increases in abundance for these taxa after the treatment date that were likely normal seasonal increases. Zygoptera have been reported to play an important role as forage for fish in other systems , and both living and decaying water hyacinth provided habitat for them. Spiders had significantly greater abundance after treatment at the treatment locations only. They are voracious predators of emerging aquatic insects, and, along with damselflies, cannabis racking systems may also be a predator of the N. bruchi weevils that currently serve as biological control agents of water hyacinth.The significant increase in zooplankton abundance after treatment was likely driven by the large relative increase in ostracods found later in the season, as described above. Mollusk populations may have also increased in the late-season as a result of phenological changes, or as a result of potentially increased biofilms as a by-product of plant decomposition. We did not observe significant differences in abundance for gill-breathers or airbreathers .
Through our case study, we assessed whether current control efforts for invasive water hyacinth in the Delta affect invertebrate communities that use water hyacinth as habitat. The results of this study provide valuable insight into the little-examined transition phase between the herbicide application and open water phases of water hyacinth management. We concluded that current management efforts for invasive water hyacinth using glyphosate do not affect invertebrate abundance or diversity during a month-long, post-treatment period of decay. Our results demonstrated that zooplankton and macroinvertebrate communities clearly continue to use herbicide-treated water hyacinth mats as habitat even when the vegetation is dead or dying. The observed late-season increase in damselflies is ecologically meaningful because they serve a dual role in aquatic food webs—that of predator and prey—and may be particularly attractive to foraging Delta fish species later in the season. More generally, the decaying water hyacinth plants may provide a resource pulse or temporary subsidy for fish, because as the plants begin to sink, and root structures degrade, the epiphytic invertebrates disperse and are much easier to prey upon when they lack the protective cover of submerged roots . Future research is needed to determine the degree to which fish are foraging in and/or near water hyacinth mats, what happens with invertebrate dispersal after herbicide-induced plant decomposition, and if/how fish foraging behavior changes as a result. If we had collected the late-season samples much later than 4 weeks post-treatment, it is likely that we would have encountered different circumstances. For example, herbicide-treated vegetation continues to decompose and eventually sinks, leaving open water in its place—and pelagic areas without macrophytes are comparatively depauperate in macroinvertebrate abundance . However, since the CDBW’s herbicide applications are spaced temporally and geographically throughout the growing season, weed mats senesce and sink in a mosaic fashion, which may allow invertebrates to move between habitat patches. Additionally, if we had selected sites across a salinity gradient or across changing hydrodynamic circumstances, we might have encountered different findings—in terms of initial invertebrate community composition, behavior of weed mats during the study, and community responses to herbicides. Finally, it is important to note that this study took place during the last year of a 5-year drought. In a non-drought year, flows from winter storms would have likely washed away more water hyacinth weed mats, resulting in lower spring and summer weed biomass. Because higher macrophyte biomass is generally correlated with higher invertebrate abundance, it is possible that we would have observed relatively lower invertebrate abundance in a non-drought year. However, we do not have reason to believe that the overall patterns we observed would necessarily be different in a non-drought year. Future research with longer time horizons would also be beneficial in informing the CDBW’s current adaptive management efforts in the Delta. For example, it would be advantageous to determine how invertebrate communities might respond to herbicide treatment as macrophyte community composition changes over time , how longer-term changes in hydrologic regimes might affect the outcomes of herbicide treatment and the composition of invertebrate communities that use the target weed as habitat, and how macrophyte mat-dwelling invertebrates may become available to fishes. In the absence of the abundant littoral habitats that were once characteristic of the Delta ecosystem, invertebrates that are crucial links in Delta food webs employ the habitats that are available—which includes water hyacinth because of its sheer abundance and Delta-wide distribution. However, water hyacinth mats are not necessarily optimal habitat for invertebrates. Nonetheless, aquatic invertebrates persist in water hyacinth beds in the physically altered and biologically invaded Delta. Given the urgent socio-economic and ecological need to manage water hyacinth invasions around the world, those leading such activities should carefully conduct control efforts in the context of adaptive management, and integrate water hyacinth control efforts with habitat-restoration efforts. From a long-term perspective, if water hyacinth were eradicated without extensive restoration of native macrophytes Delta-wide, there would likely be consequences for invertebrates that use water hyacinth as habitat and also potentially serve as forage for local fishes. Many potential scenarios might result from water hyacinth eradication in the absence of other ecosystem restoration activities that are beyond the scope of this paper.