Nitrogen fixation and other biogeochemically important features of Atacama Desert giant horsetail plant microbiomes inferred from metagenomic contig analysis.

BACKGROUND AND AIMS: Canyon stream beds in the hyperarid Atacama Desert surprisingly harbour magnificent groves of endemic giant horsetail wetland plants, Equisetum xylochaetum. Our previous metagenomic study of eukaryotes closely associated with this plant indicated that the microbiome included prokaryotes that might likewise influence host success and environment. We explored this possibility by using the metagenomic sequence to characterize prokaryote taxa and functional genes present in the microbiome of E. xylochaetum sampled from remote sites differing in the degree of anthropogenic disturbance. We focused on biogeochemical functions known to be important in wetland ecosystems. METHODS: To ensure that analyses were conducted on microbes most closely associated with plants, we extracted DNA from well-washed plant organs whose microbial biofilms were revealed with scanning electron microscopy. To assess the benefits of longer sequences for taxonomic and gene classifications, results of analyses performed using contigs were compared with those obtained with unassembled reads. We employed methods widely used to estimate genomic coverage of single taxa for genomic analysis to infer relative abundances of taxa and functional genes. KEY RESULTS: Key functional bacterial genera (e.g. Hydrogenophaga, Sulfuritalea and Rhodoferax) inferred from taxonomic and functional gene analysis of contigs - but not unassembled reads - to occur on surfaces of (or within) plants at relatively high abundance (>50× genomic coverage) indicated roles in nitrogen, sulfur and other mineral cycling processes. Comparison between sites revealed impacts on biogeochemical functions, e.g. reduced levels of the nifH gene marker under disturbance. Vanadium nitrogenases were more important than molybdenum nitrogenases, indicated by both functional genes and taxa such as Rhodomicrobium and Phaeospirillum inferred from contigs but not unassembled reads. CONCLUSIONS: Our contig-based metagenomic analyses revealed that microbes performing key wetland biogeochemical functions occur as tightly adherent biofilms on the plant body, not just in water or sediments, and that disturbance reduces such functions, providing arguments for conservation efforts.

Formation of tussocks by sedges: effects of hydroperiod and nutrients.

Tussock formation is a global phenomenon that enhances microtopography and increases biodiversity by adding structure to ecological communities, but little is known about tussock development in relation to environmental factors. To further efforts to restore wetland microtopography and associated functions, we investigated Carex stricta tussock size in relation to elevation (a proxy for water depth) at a range of sites in southern Wisconsin, USA, and tested the effect of five hydroperiods and N+P addition (15 g N/m2 + 0.37 g P/m2) on tussock formation during a three-year mesocosm experiment. Wet meadows dominated by C. stricta averaged 4.9 tussocks/m2, with a mean volume of 1160 cm3 and height of 15 cm. Within sites, taller tussocks occurred at lower elevations, suggesting a structural adaptation to anoxic conditions. In our mesocosm experiment, C. stricta accelerated tussock formation when inundated, and it increased overall productivity with N + P addition. Within two growing seasons, continuous inundation (+18 cm) in the mesocosms led to tussocks that were nearly as tall as in our field survey (mean height in mesocosms, 10 +/- 1.3 cm; maximum, 17 cm). Plants grown with constant low water (-18 cm) only formed short mounds (mean height = 2 +/- 0.4 cm). After three growing seasons, the volume of the largest tussocks (3274 +/- 376 cm3, grown with +18 cm water depth and N + P addition) was 12 times that of the smallest (275 +/- 38 cm3, grown with -18 cm water depth and no N + P). Though tussock composition varied among hydroperiods, tussocks were predominantly organic (74-94% of dry mass) and composed of leaf bases (46-59%), fine roots (10-31%), and duff (5-13%). Only the plants subjected to high water levels produced the vertically oriented rhizomes and ascending shoot bases that were prevalent in field-collected tussocks. Under continuous or periodic inundation, tussocks achieved similar heights and accumulated similar levels of organic matter (range: 163-394 g C/m2), and we conclude that these hydroperiods can accelerate tussock formation. Thus, C. stricta has high utility for restoring wetland microtopography and associated functions, including carbon accumulation.

Diversity-function relationships changed in a long-term restoration experiment.

The central tenet of biodiversity-ecosystem function (BEF) theory, that species richness increases function, could motivate restoration practitioners to incorporate a greater number of species into their projects. But it is not yet clear how well BEF theory predicts outcomes of restoration, because it has been developed through tests involving short-run and tightly controlled (e.g., weeded) experiments. Thus, we resampled our 1997 BEF experiment in a restored salt marsh to test for long-term effects of species richness (plantings with 1, 3, and 6 species per 2 x 2 m plot), with multiple ecosystem functions as response variables. Over 11 years, 1- and 6-species assemblages converged on intermediate richness (mean = 3.9 species/ 0.25-m2 plot), and composition changed nonrandomly throughout the site. While three species became rare, the two most productive species became co-dominant. The two dominants controlled and increased shoot biomass, which appeared to decrease species richness. Diversity-function relationships became less positive over 11 years and differed significantly with (a) the species-richness metric (planted vs. measured), and (b) the indicator of function (shoot biomass, height, and canopy layering). The loss of positive relationships between species richness and function in our restored site began soon after we stopped weeding and continued with increasing dominance by productive species. Where species-rich plantings are unlikely to ensure long-term restoration of functions, as in our salt marsh, we recommend dual efforts to establish (1) dominant species that provide high levels of target functions, and (2) subordinate species, which might provide additional functions under current or future conditions.

A unifying approach for evaluating the condition of wetland plant communities and identifying related stressors.

Assessment of vegetation is an important part of evaluating wetland condition, but it is complicated by the variety of plant communities that are naturally present in freshwater wetlands. We present an approach to evaluate wetland condition consisting of: (1) a stratified random sample representing the entire range of anthropogenic stress, (2) field data representing a range of water depths within the wetlands sampled, (3) nonmetric multidimensional scaling (MDS) to determine a biological condition gradient across the wetlands sampled, (4) hierarchical clustering to interpret the condition results relative to recognizable plant communities, (5) classification and regression tree (CART) analysis to relate biological condition to natural and anthropogenic environmental drivers, and (6) mapping the results to display their geographic distribution. We applied this approach to plant species data collected at 90 wetlands of the U.S. Great Lakes coast that support a variety of plant communities, reflecting the diverse physical environment and anthropogenic stressors present within the region. Hierarchical cluster analysis yielded eight plant communities at a minimum similarity of 25%. Wetlands that clustered botanically were often geographically clustered as well, even though location was not an input variable in the analysis. The eight vegetation clusters corresponded well with the MDS configuration of the data, in which the first axis was strongly related (R2 = 0.787, P < 0.001) with floristic quality index (FQI) and the second axis was related to the Great Lake of occurrence. CART models using FQI and the first MDS axis as the response variables explained 75% and 82% of the variance in the data, resulting in 6-7 terminal groups spanning the condition gradient. Initial CART splits divided the region based on growing degree-days and cumulative anthropogenic stress; only after making these broad divisions were wetlands distinguished by more local characteristics. Agricultural and urban development variables were important correlates of wetland biological condition, generating optimal or surrogate splits at every split node of the MDS CART model. Our findings provide a means of using vegetation to evaluate a range of wetland condition across a broad and diverse geographic region.

Topographic heterogeneity influences fish use of an experimentally restored tidal marsh.

Ecological theory predicts that incorporating habitat heterogeneity into restoration sites should enhance diversity and key functions, yet research is limited on how topographic heterogeneity affects higher trophic levels. Our large (8-ha) southern California restoration experiment tested effects of tidal creek networks and pools on trophic structure of salt marsh habitat and high-tide use by two regionally dominant fish species, California killifish (Fundulus parvipinnis) and longjaw mudsucker (Gillichthys mirabilis). We expected tidal creeks to function as "conduits" that would enhance connectivity between subtidal and intertidal habitat and pools to serve as microhabitat "oases" for fishes. Pools did provide abundant invertebrate prey and were a preferred microhabitat for F. parvipinnis, even when the entire marsh was inundated (catch rates were 61% higher in pools). However, G. mirabilis showed no preference for pools. At a larger scale, effects of tidal creek networks were also mixed. Areas containing creeks had 12% higher catch rates of G. mirabilis, but lower catch rates and feeding rates of F. parvipinnis. Collectively, the results indicate that restoring multiple forms of heterogeneity is required to provide opportunities for multiple target consumers.

Partitioning vegetation response to anthropogenic stress to develop multi-taxa wetland indicators.

Emergent plants can be suitable indicators of anthropogenic stress in coastal wetlands if their responses to natural environmental variation can be parsed from their responses to human activities in and around wetlands. We used hierarchical partitioning to evaluate the independent influence of geomorphology, geography, and anthropogenic stress on common wetland plants of the U.S. Great Lakes coast and developed multi-taxa models indicating wetland condition. A seven-taxon model predicted condition relative to watershed-derived anthropogenic stress, and a four-taxon model predicted condition relative to within-wetland anthropogenic stressors that modified hydrology. The Great Lake on which the wetlands occurred explained an average of about half the variation in species cover, and subdividing the data by lake allowed us to remove that source of variation. We developed lake-specific multi-taxa models for all of the Great Lakes except Lake Ontario, which had no plant species with significant independent effects of anthropogenic stress. Plant responses were both positive (increasing cover with stress) and negative (decreasing cover with stress), and plant taxa incorporated into the lake-specific models differed by Great Lake. The resulting models require information on only a few taxa, rather than all plant species within a wetland, making them easier to implement than existing indicators.

Practical proxies for tidal marsh ecosystem services: application to injury and restoration.

Tidal marshes are valued, protected and restored in recognition of their ecosystem services: (1) high productivity and habitat provision supporting the food web leading to fish and wildlife, (2) buffer against storm wave damage, (3) shoreline stabilization, (4) flood water storage, (5) water quality maintenance, (6) biodiversity preservation, (7) carbon storage and (8) socio-economic benefits. Under US law, federal and state governments have joint responsibility for facilitating restoration to compensate quantitatively for ecosystem services lost because of oil spills and other contaminant releases on tidal marshes. This responsibility is now met by choosing and employing metrics (proxies) for the suite of ecosystem services to quantify injury and scale restoration accordingly. Most injury assessments in tidal marshes are triggered by oil spills and are limited to: (1) documenting areas covered by heavy, moderate and light oiling; (2) estimating immediate above-ground production loss (based on stem density and height) of the dominant vascular plants within each oiling intensity category and (3) sampling sediments for chemical analyses and depth of contamination, followed by sediment toxicity assays if sediment contamination is high and likely to persist. The percentage of immediate loss of ecosystem services is then estimated along with the recovery trajectory. Here, we review potential metrics that might refine or replace present metrics for marsh injury assessment. Stratifying plant sampling by the more productive marsh edge versus the less accessible interior would improve resolution of injury and provide greater confidence that restoration is truly compensatory. Using microphytobenthos abundance, cotton-strip decomposition bioassays and other biogeochemical indicators, or sum of production across consumer trophic levels fails as a stand-alone substitute metric. Below-ground plant biomass holds promise as a potential proxy for resiliency but requires further testing. Under some conditions, like chronic contamination by organic pollutants that affect animals but not vascular plants, benthic infaunal density, toxicity testing, and tissue contamination, growth, reproduction and mortality of marsh vertebrates deserve inclusion in the assessment protocol. Additional metrics are sometimes justified to assay microphytobenthos, use by nekton, food and habitat for reptiles, birds and mammals, or support of plant diversity. Empirical research on recovery trajectories in previously injured marshes could reduce the largest source of uncertainty in quantifying cumulative service losses.

Canopy Architecture of Natural and Planted Cordgrass Marshes: Selecting Habitat Evaluation Criteria.

In order to set standards for restoration or mitigation "success" of salt marsh habitat designed for an endangered southern California bird, I explored the bird's nesting requirements, compared constructed habitats that lack the bird with natural habitats that support it, and selected habitat assessment criteria that best distinguish suitable and unsuitable nesting habitats. Selected attributes of canopy architecture are recommended for assessing the suitability of intertidal cordgrass (Spartina foliosa) marshes for the endangered Light-footed Clapper Rail (Rallus longirostris levipes) of southern California. Nests are built at °145 cm above Mean Lower Low Water (MLLW = 0 tidal datum), with nest rims at °165-170 cm MLLW; however, extreme high water is 232 cm MLLW during the March-July nesting season, and tidal inundation is a major cause of nest failure. Cordgrass that is >60 cm tall appears necessary for birds to weave a canopy that allows the nest to float upward, but not away, as the tide rises. Cordgrass height distributions and density data describe attributes of canopy architecture that assess Clapper Rail habitat value better than previously used measures (i.e., cover, biomass, mean height, maximum height). Height histograms differ for planted marshes (which do not support Clapper Rails) and natural marshes. The constructed marshes have few plants over 60 cm, while most stems in natural marshes exceed 60 cm. In natural marshes, cordgrass heights increase with freshwater flooding and nitrogen enrichment. Reference data from natural marshes that are used by Clapper Rails indicate that the standard for "suitable habitat" should be a density of at least 100 stems/m2 with at least 90 stems/m2 >60 cm, of which at least 30 stems/m2 are >90 cm in height. High interannual and spatial variability indicates the need for several (e.g., 20) years of data for assessment purposes and a large data base for reference wetlands.

Nitrogen Assessments in a Constructed and a Natural Salt Marsh of San Diego Bay.

Soil nitrogen pools were lower in a 4-yr-old constructed salt marsh than in an adjacent natural marsh of San Diego Bay. Aboveground biomass and foliar nitrogen content of Spartina foliosa were both lower for the constructed marsh. Soil organic carbon, which was highly positively correlated with total nitrogen, was also lower in the constructed marsh. Rates of nitrogen fixation were higher for the natural marsh in surface soils (1 cm depth) but not in the rhizosphere (10 cm depth). Experimental additions of organic matter increased rates of nitrogen fixation substantially for both the constructed and natural marsh soils, with glucose stimulating greater increases than Spartina foliosa detritus (roots and rhizomes). In comparison with natural marshes studied elsewhere, the San Diego Bay sites have low nitrogen pools and little soil organic carbon. Nitrogen mineralization rates (in situ incubations) were high in both marshes studied. The low nitrogen pools reflect low tidal import and infrequent streamflow influxes and, possibly, high nitrogen demands of vegetation stressed by hypersaline soils.

Food web analysis of southern California coastal wetlands using multiple stable isotopes.

Carbon, nitrogen, and sulfur stable isotopes were used to characterize the food webs (i.e., sources of carbon and trophic status of consumers) in Tijuana Estuary and San Dieguito Lagoon. Producer groups were most clearly differentiated by carbon, then by sulfur, and least clearly by nitrogen isotope measurements. Consumer 15N isotopic enrichment suggested that there are four trophic levels in the Tijuana Estuary food web and three in San Dieguito Lagoon. A significant difference in multiple isotope ratio distributions of fishes between wetlands suggested that the food web of San Dieguito Lagoon is less complex than that of Tijuana Estuary. Associations among sources and consumers indicated that inputs from intertidal macroalgae, marsh microalgae, and Spartina foliosa provide the organic matter that supports invertebrates, fishes, and the light-footed clapper rail (Rallus longirostris levipes). These three producers occupy tidal channels, low salt marsh, and mid salt marsh habitats. The only consumer sampled that appears dependent upon primary productivity from high salt marsh habitat is the sora (Porzana carolina). Two- and three-source mixing models identified Spartina as the major organic matter source for fishes, and macroalgae for invertebrates and the light-footed clapper rail in Tijuana Estuary. In San Dieguito Lagoon, a system lacking Spartina, inputs of macroalgae and microalgae support fishes. Salicornia virginica, S. subterminalis, Monanthochloe littoralis, sewage- derived organic matter, and suspended particulate organic matter were deductively excluded as dominant, direct influences on the food web. The demonstration of a salt marsh-channel linkage in these systems affirms that these habitats should be managed as a single ecosystem and that the restoration of intertidal marshes for endangered birds and other biota is compatible with enhancement of coastal fish populations; heretofore, these have been considered to be competing objectives.

Relationships between canopy complexity and germination microsites for Phalaris arundinacea L.

Microsites that prevent seed germination are critical for slowing the invasion of native plant communities by aggressive, clonal species. A suitable model for study is the clonal grass, Phalaris arundinacea, which reproduces prolifically from seed and is spreading into wetlands across temperate North America. Knowing that light conditions control its seed germination in the laboratory and that light varies with canopy complexity in a Wisconsin fen, we tested multiple attributes of microsites under spatially and temporally dynamic canopies (namely, presence/absence of a matrix species, number of species in the canopy, plus indirect effects of three soil water levels) for their control of germination in microcosms. Our 6-species canopies + the matrix of Glyceria striata had the densest cover and reduced P. arundinacea germination to 1.9%, compared to 7.3% for 1-species canopies + the matrix. After selectively removing canopy components, germination increased to 36.1% for 6-species and 33.0% for 1-species canopies. Comparing canopies with each of the six species, germination declined in relation to increasing leaf width. Given moist soil, P. arundinacea germination microsites are determined by canopy complexity, which affects light penetration, which in turn determines germination rate.

Differential invasion of a wetland grass explained by tests of nutrients and light availability on establishment and clonal growth.

Phalaris arundinacea (Poaceae) is aggressively invading wetlands across North America. We tested the hypotheses that open canopies and increased nutrients facilitate vegetative establishment in the field, using a phytometer (6 rhizome fragments/plot, 24 plots/wetland). In each of three wetlands, phytometers received three levels of an NPK fertilizer or served as controls. Emergence and survival differed among sites (P=0.0005), but not due to NPK addition. P. arundinacea survival was highest in a wet prairie with a late-developing canopy, but limited by prolonged flooding in one sedge meadow and by an early-growing, dense plant canopy in a second. These patterns were explained in greenhouse experiments, where both flooding (P<0.0001) and heavy shade (P=0.0002) decreased P. arundinacea aboveground biomass by up to 73% and 97%, respectively. Rhizome fragment survival was reduced by 30% under flooded conditions and 25% under heavy shade. We then tested the hypothesis that a clonal subsidy facilitates vegetative expansion into heavy shade. Established clones were allowed access to bare soil under four levels of shade and two levels of NPK fertilizer in a two-factor greenhouse experiment. Young ramets attached to parent clones readily grew into heavy shade, and the high nutrient treatment increased aboveground growth (P<0.0001) and distance of ramet spread (P=0.0051) by nearly 50%. Under low nutrient conditions, root biomass increased by 30% (P<0.0001). P. arundinacea's rapid expansion into a variety of wetland types is likely a function of clonal subsidy, morphological plasticity, and nutrient availability: young ramets that emerge under shaded conditions are supported by parental subsidies; where nutrients are plentiful, P. arundinacea can maximize aboveground growth to capture more light; and where nutrients are scarce, it can increase belowground foraging.

Salt marsh productivity with natural and altered tidal circulation.

The effects of altered tidal circulation on southern California salt marshes are investigated by comparing a well-flushed wetland and two modified wetlands which have reduced tidal flow. The Tijuana Estuary had continuous exchange of seawater but relatively low net aerial primary productivity (0.4-1.0 kg m-2yr-1) of vascular plants. Low productivity (0.6 kg m-2yr-1) was also found in the Flood Control Channel of the San Diego River, where tidal exchange was restricted to flow through a riprap dike. High productivity (1.2-2.9 kg m-2yr-1) in Los Penasquitos Lagoon was attributed to the influences of freshwater impounded behind a sand bar which blocked the mouth of the lagoon during much of the study period.It is hypothesized that elimination of tidal flow during the growing season increased primary productivity of vascular plants because freshwater runoff decreased soil salinity and because nutrients were retained within the marsh. However, we predict that sand bar obstruction can decrease productivity if below-average rainfall leads to hypersalinity of closed lagoons. Comprehensive evaluation of the effects of altered tidal circulation requires longterm study and examination of the total ecosystem.

Multiple disturbances accelerate invasion of reed canary grass (Phalaris arundinaceaL.) in a mesocosm study.

Disturbances that intensify with agriculture and/or urban development are thought to promote the spread of invasive plants, such as the clonal perennial reed canary grass (Phalaris arundinaceaL). To test this relationship and interactions among disturbances, we subjected wet prairie assemblages within 1.1 m(2) mesocosms to invasion by Phalaris and addition of nutrients, sediments, and flooding. Species richness decreased with the application of sediments and/or flooding of 4 consecutive weeks or longer. Losses of up to six dominant and subdominant species in these treatments increased light transmission through the plant canopy by as much as 400% over the control. Light availability in July and September was a strong predictor of end-of-season aboveground biomass of Phalaris. Phalaris was also 35% and 195% more productive when nutrients were added at low and high levels, respectively. Multiple factors in combination were usually additive in their effects on invasion, but sediments and nutrients interacted with flood regime to synergistically increase invasion in some cases. A separate experiment likewise revealed a synergistic interaction between added nutrients and simulated grazing. We suggest that multiple factors be mitigated simultaneously to reduce invasion of Phalaris.

Potential for using native plant species in stormwater wetlands.

Spartina pectinata (prairie cordgrass) was grown under five hydroperiods (wet-dry cycles) to determine its potential for use in stormwater wetlands, particularly as an alternative to the highly invasive Phalaris arundinacea (an exotic grass). Rhizomes planted in outdoor microcosms grew vigorously in all treatments, namely, weekly flooding in early summer, weekly flooding in late summer, flooding every three weeks throughout the summer, weekly flooding throughout the summer, and no flooding. Neither the timing nor frequency of 24-hour floods (10-20 cm deep) affected total stem length (grand mean 1003 +/- 188.8 cm per pot, n = 140) or above-ground biomass (46.5 +/- 8.3 g per pot, equivalent to approximately 360 g/m2). However, by late summer, fewer new tillers were found in unflooded microcosms, indicating that vegetative expansion is drought-sensitive. The growth of Spartina plants was further assessed with and without Glyceria striata (a native grass) and Phalaris arundinacea. Glyceria growth was not affected by hydrologic treatment. Glyceria reduced Spartina growth by approximately 11%, suggesting potential as a cover crop that might reduce establishment and growth of Phalaris seedlings. Seeds of Phalaris did not germinate, but branch fragments established where soil was moist from flooding, regardless of the presence of Glyceria. The ability of Spartina to establish vegetatively and grow well under variable water levels leads us to recommend further testing in stormwater wetlands, along with early planting of Glyceria to reduce weed invasions.