Freshwater wetlands for flood control: How manipulating the hydroperiod affects plant and invertebrate communities.

Thoughtfully managed hydroperiods in natural and artificial wetlands could potentially provide a combination of desirable flood control services and high ecological functions. To explore how managed freshwater wetlands typical of the Houston, Texas area would respond to different hydrological regimes that might occur if wetlands were drained in anticipation of a heavy rain that did not materialize, we conducted a mesocosm experiment with six flooding depths and seven drought durations, followed by seven months of recovery. We found that the speed in which mesocosms dried out was a function of initial water depth, with mesocosms initially set with greater water depths (30 cm) taking ~ 38 days to dry out versus zero days for wetlands that were completely drained. Individual plant species (14 species planted; 8 species common at the end of the recovery period) were affected by drought length, flooding depth, or their interaction, although details of these responses varied among the species. The composition of the plant community at the end of the drought period was strongly affected by drought length, and the effect of the drought length treatment persisted through seven months of post-drought recovery, with the 80- and 160-day drought treatments diverging most strongly from shorter drought treatments. Above- and below-ground biomass of plants was not affected by the treatments, but above-ground dead biomass (litter) decreased with increasing drought length. Densities of mosquito larvae, snails and tadpoles were temporally variable, and were affected more during the treatment period and early in recovery than after a disturbance event late in recovery. Our results indicate that managed wetlands in southeast Texas would be quite resilient to dry periods of up to 40 days in duration, especially if water was not completely drained at the beginning of the drought. In addition, many species would persist in managed wetlands even with droughts of up to 160 days. This indicates considerable potential for managing the hydroperiods of artificial detention ponds by retaining water longer to increase ecological function, with little to no loss of flood control services, and for managing the hydroperiods of natural wetlands by draining them in advance of anticipated rains to increase flood control services, with little to no loss of ecological function.

Long-term data reveal that grazer density mediates climatic stress in salt marshes.

Understanding how climate and local stressors interact is paramount for predicting future ecosystem structure. The effects of multiple stressors are often examined in small-scale and short-term field experiments, limiting understanding of the spatial and temporal generality of the findings. Using a 22-year observational dataset of plant and grazer abundance in a southeastern US salt marsh, we analyzed how changes in drought and grazer density combined to affect plant biomass. We found: (1) increased drought severity and higher snail density both correlated with lower plant biomass; (2) drought and snail effects interacted additively; and, (3) snail effects had a threshold, with additive top-down effects only occurring when snails were present at high densities. These results suggest that the emergence of multiple stressor effects can be density dependent, and they validate short-term experimental evidence that consumers can augment environmental stress. These findings have important implications for predicting future ecosystem structure and managing natural ecosystems.

Variance reflects resilience to disturbance along a stress gradient: Experimental evidence from coastal marshes.

Quantifying ecosystem resilience to disturbance is important for understanding the effects of disturbances on ecosystems, especially in an era of rapid global change. However, there are few studies that have used standardized experimental disturbances to compare resilience patterns across abiotic gradients in real-world ecosystems. Theoretical studies have suggested that increased return times are associated with increasing variance during recovery from disturbance. However, this notion has rarely been explicitly tested in field, in part due to the challenges involved in obtaining long-term experimental data. In this study, we examined resilience to disturbance of 12 coastal marsh sites (five low-salinity and seven polyhaline [=salt] marshes) along a salinity gradient in Georgia, USA. We found that recovery times after experimental disturbance ranged from 7 to >127 months, and differed among response variables (vegetation height, cover and composition). Recovery rates decreased along the stress gradient of increasing salinity, presumably due to stress reducing plant vigor, but only when low-salinity and polyhaline sites were analyzed separately, indicating a strong role for traits of dominant plant species. The coefficient of variation of vegetation cover and height in control plots did not vary with salinity. In disturbed plots, however, the coefficient of variation (CV) was consistently elevated during the recovery period and increased with salinity. Moreover, higher CV values during recovery were correlated with slower recovery rates. Our results deepen our understanding of resilience to disturbance in natural ecosystems, and point to novel ways that variance can be used either to infer recent disturbance, or, if measured in areas with a known disturbance history, to predict recovery patterns.

A hurricane alters the relationship between mangrove cover and marine subsidies.

As global change alters the composition and productivity of ecosystems, the importance of subsidies from one habitat to another may change. We experimentally manipulated black mangrove (Avicennia germinans) cover in 10 large plots and over a 5-year period (2014-2019) quantifying the effects of mangrove cover on subsidies of floating organic material (wrack) into coastal wetlands. As mangrove cover increased from 0% to 100%, wrack cover and thickness decreased by ~60%, the distance that wrack penetrated into the plots decreased by ~70%, and the percentage of the wrack trapped in the first 6 m of the plot tripled. These patterns observed during 4 "normal" years disappeared in a fifth year following Hurricane Harvey (2017), when large quantities of wrack were pushed far into the interior of all the plots, regardless of mangrove cover. Prior to the storm, the abundance of animals collected in grab samples increased with wrack biomass. Wrack composition did not affect animal abundance or composition. Experimental outplants of two types of wrack (red algae and seagrass) revealed that animal abundance and species composition varied between the fringe and interior of the plots, and between microhabitats dominated by salt marsh versus mangrove vegetation. The importance of wrack to overall carbon stocks varied as a function of autochthonous productivity: wrack inputs (per m2 ) based on survey data were greater than aboveground plant biomass in the plots (42 × 24 m) dominated by salt-marsh vegetation, but decreased to 5% of the total aboveground biomass in plots dominated by mangroves. Our results illustrate that increasing mangrove cover decreases the relative importance of marine subsidies into the intertidal at the plot level, but concentrates subsidies at the front edge of the mangrove stand. Storms, however, may temporarily override mangrove attenuation of wrack inputs. Our results highlight the importance of understanding how changes in plant species composition due to global change will impact marine subsidies and exchanges among ecosystems, and foster a broader understanding of the functional interdependence of adjacent habitats within coastal ecosystems.

Drivers of litter mass loss and faunal composition of detritus patches change over time.

Decomposition of vegetal detritus is one of the most fundamental ecosystem processes. In complex landscapes, the fate of litter of terrestrial plants may depend on whether it ends up decomposing in terrestrial or aquatic conditions. However, (1) to what extent decomposition rates are controlled by environmental conditions or by detritus type, and (2) how important the composition of the detritivorous fauna is in mediating decomposition in different habitats, remain as unanswered questions. We incubated two contrasting detritus types in three distinct habitat types in Coastal Georgia, USA, to test the hypotheses that (1) the litter fauna composition depends on the habitat and the litter type available, and (2) litter mass loss (as a proxy for decomposition) depends on environmental conditions (habitat) and the litter type. We found that the abundance of most taxa of the litter fauna depends primarily on habitat. Litter type became a stronger driver for some taxa over time, but the overall faunal composition was only weakly affected by litter type. Decomposition also depends strongly on habitat, with up to ca. 80% of the initial detrital mass lost over 25 months in the marsh and forest habitats, but less than 50% lost in the creek bank habitat. Mass loss rates of oak versus pine litter differed initially but converged within habitat types within 12 months. We conclude that, although the habitat type is the principle driver of the community composition of the litter fauna, litter type is a significant driver of litter mass loss in the early stages of the decomposition process. With time, however, litter types become more and more similar, and habitat becomes the dominating factor in determining decomposition of older litter. Thus, the major driver of litter mass loss changes over time from being the litter type in the early stages to the habitat (environmental conditions) in later stages.

Directional movement of consumer fronts associated with creek heads in salt marshes.

Consumers often deplete local resources and aggregate along edges of remaining resources, forming "consumer fronts." We examined the factors that promote Sesarma reticulatum crab aggregations at saltmarsh creek heads to explain the directional but slow movement of these fronts. We also created artificial creek heads to test the hypothesis that hydrological conditions at creek heads create superior habitat for crabs. Soil temperatures were ˜11-12% cooler, hydrogen sulfide concentrations lower (0.0 vs. ˜0.58 mg/L), and dissolved oxygen concentrations twofold higher at the creek head versus the marsh platform. In the artificial creek-head experiment, altering hydrological conditions led to lower dissolved sulfide levels, higher dissolved oxygen levels, and increased densities of crab burrows and Sesarma crabs. Moreover, the elevation of the soil surface declined rapidly at artificial creek heads versus controls, suggesting that crabs were increasing erosion. Our results suggest that abiotic conditions for crabs are better at the leading edge of the creek head than the trailing edge, explaining the directional movement of the front. Moreover, the speed at which the front propagates appears to be limited by the rate at which the creekhead erodes, rather than by crab mobility. The directional and slow movement of Sesarma fronts compared to consumer fronts of other invertebrates appears to result from the inextricable link between Sesarma and marsh geomorphology, whereas other consumer fronts are associated mostly with food resources.

Plasticity and selection drive hump-shaped latitudinal patterns of flowering phenology in an invasive intertidal plant.

Patterns of flowering phenology can affect the success of plant invasions, especially when introduced species spread across a wide range of latitude into different climatic conditions. We combined a 4-yr field survey and a 3-yr common garden experiment with the invasive grass Spartina alterniflora that is now widespread along the coast of China to document the latitudinal pattern of flowering phenology, determine if phenology was related to climate or oceanographic variables, and determine whether phenology patterns were fixed versus plastic. In the field, first flowering day displayed a hump-shaped relationship with latitude, with low- and high-latitude plants flowering 100 d and 10 d earlier than plants at middle latitudes, respectively. Peak flowering day showed a similar hump-shaped relationship with latitude, with the interval between first and peak flowering day decreasing with increasing latitude. First flowering day had a hump-shaped relationship with annual growing degree days but a linear positive relationship with tidal range. In the common garden, first flowering day decreased linearly with increasing latitude of origin, as did peak flowering day, and the interval between first and peak flowering day increased with increasing latitude. First flowering day in the common garden had weak or no relationships with abiotic variables at the sites of origin. In both the field and common garden, first flowering day was later in site years for which plants were taller. These results indicate a high degree of plasticity in flowering phenology, with plants flowering later in the field at sites with intermediate temperatures and high tide ranges. Common garden results indicate some selection for earlier flowering at sites with low temperatures, consistent with a shorter growing season. Consistent with life-history theory, plants flowered later under conditions favoring vigorous growth. Earlier flowering and smaller size of plants at high and low latitudes suggests that S. alterniflora has already occupied much of the geographic range favorable for it on the East Coast of Asia.

Effects of mangrove cover on coastal erosion during a hurricane in Texas, USA.

We tested the hypothesis that mangroves provide better coastal protection than salt marsh vegetation using 10 1,008-m2 plots in which we manipulated mangrove cover from 0 to 100%. Hurricane Harvey passed over the plots in 2017. Data from erosion stakes indicated up to 26 cm of vertical and 970 cm of horizontal erosion over 70 months in the plot with 0% mangrove cover, but relatively little erosion in other plots. The hurricane did not increase erosion, and erosion decreased after the hurricane passed. Data from drone images indicated 196 m2 of erosion in the 0% mangrove plot, relatively little erosion in other plots, and little ongoing erosion after the hurricane. Transects through the plots indicated that the levee (near the front of the plot) and the bank (the front edge of the plot) retreated up to 9 m as a continuous function of decreasing mangrove cover. Soil strength was greater in areas vegetated with mangroves than in areas vegetated by marsh plants, or nonvegetated areas, and increased as a function of plot-level mangrove cover. Mangroves prevented erosion better than marsh plants did, but this service was nonlinear, with low mangrove cover providing most of the benefits.

Variation in synchrony of production among species, sites, and intertidal zones in coastal marshes.

Spatially synchronous population dynamics are important to ecosystem functioning and have several potential causes. By looking at synchrony in plant productivity over 18 yr across two elevations in three types of coastal marsh habitat dominated by different clonal plant species in Georgia, USA, we were able to explore the importance of plant species and different habitat conditions to synchrony. Synchrony was highest when comparing within a plant species and within a marsh zone, and decreased across species, with increasing distance, and with increasing elevational differences. Abiotic conditions that were measured at individual sites (water column temperature and salinity) also showed high synchrony among sites, and in one case (salinity) decreased with increasing distance among sites. The Moran effect (synchronous abiotic conditions among sites) is the most plausible explanation for our findings. Decreased synchrony between creekbank and mid-marsh zones, and among habitat types (tidal fresh, brackish, and salt marsh) was likely due in part to different exposure to abiotic conditions and in part to variation in sensitivity of dominant plant species to these abiotic conditions. We found no evidence for asynchrony among species, sites or zones, indicating that one habitat type or zone will not compensate for poor production in another during years with low productivity; however, tidal fresh, brackish. and salt marsh sites were also not highly synchronous with each other, which will moderate productivity variation among years at the landscape level due to the portfolio effect. We identified the creekbank zone as more sensitive than the mid-marsh to abiotic variation and therefore as a priority for monitoring and management.

Crazy ants craving calcium: macronutrients and micronutrients can limit and stress an invaded grassland brown food web.

Nitrogen and phosphorus are thought to be the most important limiting nutrients in most terrestrial ecosystems, but little is known about how other elements may limit the abundance of arthropods. We utilized a fully factorial fertilization experiment that manipulated macronutrients (N and P, together) and micronutrients (calcium, sodium, potassium, separately), in large 30 × 30 m plots and sampled litter arthropods via pitfall trapping to determine the nutrients that limit this group. An invasive ant, Nylanderia fulva, numerically dominated the community and increased in abundance 13% in plots fertilized by Ca. Detritivores were not limited by any nutrient combination, but macronutrients increased predator abundance 43%. We also found that some combinations of macronutrients and micronutrients had toxic or stressful effects on the arthropod community: detritivores decreased in abundance 23% with the combination of macronutrients, Ca, and K, and 22% with macronutrients and K; and N. fulva decreased in abundance 24% in plots fertilized by K and 45% in plots fertilized by the combination of Na and K. Our work supports growing evidence that micronutrients, especially Ca and K, may be important in structuring grassland arthropod communities, and suggests that micronutrients may affect whether or not invasive ants reach numerical dominance.

Weak latitudinal gradients in insect herbivory for dominant rangeland grasses of North America.

Patterns of insect herbivory may follow predictable geographical gradients, with greater herbivory at low latitudes. However, biogeographic studies of insect herbivory often do not account for multiple abiotic factors (e.g., precipitation and soil nutrients) that could underlie gradients. We tested for latitudinal clines in insect herbivory as well as climatic, edaphic, and trait-based drivers of herbivory. We quantified herbivory on five dominant grass species over 23 sites across the Great Plains, USA. We examined the importance of climate, edaphic factors, and traits as correlates of herbivory. Herbivory increased at low latitudes when all grass species were analyzed together and for two grass species individually, while two other grasses trended in this direction. Higher precipitation was related to more herbivory for two species but less herbivory for a different species, while higher specific root length was related to more herbivory for one species and less herbivory for a different species. Taken together, results highlight that climate and trait-based correlates of herbivory can be highly contextual and species-specific. Patterns of insect herbivory on dominant grasses support the hypothesis that herbivory increases toward lower latitudes, though weakly, and indicates that climate change may have species-specific effects on plant-herbivore interactions.

Contrasting plant adaptation strategies to latitude in the native and invasive range of Spartina alterniflora.

Biological invasions offer model systems of contemporary evolution. We examined trait differences and evolution across geographic clines among continents of the intertidal grass Spartina alterniflora within its invasive and native ranges. We sampled vegetative and reproductive traits in the field at 20 sites over 20° latitude in China (invasive range) and 28 sites over 17° in the US (native range). We grew both Chinese and US plants in a glasshouse common garden for 3 yr. Chinese plants were c. 15% taller, c. 10% denser, and set up to four times more seed than US plants in both the field and common garden. The common garden experiments showed a striking genetic cline of seven-fold greater seed set at higher latitudes in the introduced but not the native range. By contrast, there was a slight genetic cline in some vegetative traits in the native but not the introduced range. Our results are consistent with others showing that introduced plants can evolve rapidly in the new range. S. alterniflora has evolved different trait clines in the native and introduced ranges, showing the importance of phenotypic plasticity and genetic control of change during the invasion process.

Variation in microplastics composition at small spatial and temporal scales in a tidal flat of the Yangtze Estuary, China.

Microplastics are small, degrade slowly, and easily persist in the water column because they are close to neutrally buoyant. Understanding the distribution of microplastics is fundamental to evaluating the ecological risks that they cause and to identifying ways to control microplastics pollution. Most of the existing research on the distribution of microplastics in the coastal zone has focused on large spatial and temporal scales. To build on past work, we investigated variation in microplastics in a tidal flat of the Yangtze Estuary on small spatial (sediment depth, mudflat vs. vegetation zone) and temporal (fortnightly and semidiurnal) scales. Microplastics were more abundant in surface (0-2 cm) sediments during neap versus spring tide cycles, likely indicating increased deposition during periods with calm waters and increased suspension when water was more turbulent, but did not vary at greater depths in the sediment. Individual microplastics particles were also larger during neap versus spring tide periods. In contrast to the variation between spring and neap tide periods, we found no variation in the abundance of microplastics on the semidiurnal scale. Microplastics were also more abundant in the transect in the vegetation than at slightly lower elevations in the adjacent mudflat. Across all samples, the abundance of microplastics was negatively correlated with the strength of hydrological processes such as submergence time and flow velocity. Our results showed that sampling of microplastics in the intertidal environment needs to consider variation among spring and neap tide cycles, and also among different intertidal habitats that may differ only slightly in elevation. We encourage coupling sampling with direct measures of hydrological processes so that variation in microplastics abundance and size can be rigorously linked to hydrological processes.

Quantifying how changing mangrove cover affects ecosystem carbon storage in coastal wetlands.

Despite overall global declines, mangroves are expanding into and within many subtropical wetlands, leading to heterogeneous cover of marsh-mangrove coastal vegetation communities near the poleward edge of mangroves' ranges. Coastal wetlands are globally important carbon sinks, yet the effects of shifts in mangrove cover on organic-carbon (OC) storage remains uncertain. We experimentally maintained black mangrove (Avicennia germinans) or marsh vegetation in patches (n = 1,120, 3 × 3 m) along a gradient in mangrove cover (0-100%) within coastal wetland plots (n = 10, 24 × 42 m) and measured changes in OC stocks and fluxes. Within patches, above and belowground biomass (OC) was 1,630% and 61% greater for mangroves than for recolonized marshes, and soil OC was 30% greater beneath mangrove than marsh vegetation. At the plot scale, above and belowground biomass increased linearly with mangrove cover but soil OC was highly variable and unrelated to mangrove cover. Root ingrowth was not different in mangrove or marsh patches, nor did it change with mangrove cover. After 11 months, surface OC accretion was negatively related to plot-scale mangrove cover following a high-wrack deposition period. However, after 22 months, accretion was 54% higher in mangrove patches, and there was no relationship to plot-scale mangrove cover. Marsh (Batis maritima) leaf and root litter had 1,000% and 35% faster breakdown rates (k) than mangrove (A. germinans) leaf and root litter. Soil temperatures beneath mangroves were 1.4°C lower, decreasing aboveground k of fast- (cellulose) and slow-decomposing (wood) standard substrates. Wood k in shallow soil (0-15 cm) was higher in mangrove than marsh patches, but vegetation identity did not impact k in deeper soil (15-30 cm). We found that mangrove cover enhanced OC storage by increasing biomass, creating more recalcitrant organic matter and reducing k on the soil surface by altering microclimate, despite increasing wood k belowground and decreasing allochthonous OC subsidies. Our results illustrate the importance of mangroves in maintaining coastal OC storage, but also indicate that the impacts of vegetation change on OC storage may vary based on ecosystem conditions, organic-matter sources, and the relative spatiotemporal scales of mangrove vegetation change.

A comparison of coastal habitat restoration projects in China and the United States.

We compared coastal restoration projects in a developing country, China, and a developed country, the United States of America, both of which are facing loss and degradation of coastal habitats at similar latitudes, for the period of 1992-2014. To document the scale of coastal habitat restoration projects in the two countries, we identified 914 coastal restoration projects with an accumulated area of 300,521 acres in China, with most of our information coming from scientific papers, and 1,620 coastal restoration projects with an accumulated area of 243,064 acres in the USA, with most of our information coming from public databases. In both countries, about half the projects were in wetland habitats, but China had a greater proportion of projects in submerged habitats (43% versus 28% in the USA) and the USA a greater proportion in coastal upland habitats (21% versus 9% in China). The number of new projects steadily increased over time in China, but dropped after 2006 in the USA, although the total cost of new projects continued to increase. The number of projects in China and the total cost of projects in the USA were correlated with national GDP. Restoration projects in China used fewer techniques, had fewer partners, and took longer to complete than projects in the USA. Information about projects was incomplete, especially in China, and both countries could do more to make information publically available. We know more about project construction than project outcomes, and it is unclear whether projects are achieving their goals or whether the techniques used are optimal.

Chronic but not acute saltwater intrusion leads to large release of inorganic N in a tidal freshwater marsh.

Sea level rise is expected to increase inundation and saltwater intrusion into many tidal freshwater marshes and forests. Saltwater intrusion may be long-term, as with rising seas, or episodic, as with low river flow or storm surge. We applied continuous (press) and episodic (pulse) treatments of dilute seawater to replicate 2.5 × 2.5 m field plots for three years and measured soil attributes, including soil porewater, oxidation-reduction potential, soil carbon (C), and nitrogen (N) to investigate the effects of continuous and episodic saltwater intrusion and increased inundation on tidal freshwater marsh elemental cycling and soil processes. Continuous additions of dilute seawater resulted in increased porewater chloride, sulfate, sulfide, ammonium, and nitrate concentrations. Plots that received press additions also had lower soil oxidation-reduction potentials beginning in the second year. Episodic additions of dilute seawater during typical low flow conditions (Sept.-Oct.) resulted in transient increases in porewater chloride and sulfate that returned to baseline conditions once dosing ceased. Freshwater additions did not affect porewater inorganic N or soil C or N. Persistent saltwater intrusion in freshwater marshes alters the N cycle by releasing ammonium-N from sorption sites, increasing nitrification and severely reducing N storage in macrophyte biomass. Chronic saltwater intrusion, as is expected with rising seas, is likely to shift tidal freshwater marshes from a sink to a source of N whereas intermittent intrusion from drought may have no long term effect on N cycling.

Supporting Spartina: Interdisciplinary perspective shows Spartina as a distinct solid genus.

In 2014, a DNA-based phylogenetic study confirming the paraphyly of the grass subtribe Sporobolinae proposed the creation of a large monophyletic genus Sporobolus, including (among others) species previously included in the genera Spartina, Calamovilfa, and Sporobolus. Spartina species have contributed substantially (and continue contributing) to our knowledge in multiple disciplines, including ecology, evolutionary biology, molecular biology, biogeography, experimental ecology, biological invasions, environmental management, restoration ecology, history, economics, and sociology. There is no rationale so compelling to subsume the name Spartina as a subgenus that could rival the striking, global iconic history and use of the name Spartina for over 200 yr. We do not agree with the subjective arguments underlying the proposal to change Spartina to Sporobolus. We understand the importance of both the objective phylogenetic insights and of the subjective formalized nomenclature and hope that by opening this debate we will encourage positive feedback that will strengthen taxonomic decisions with an interdisciplinary perspective. We consider that the strongly distinct, monophyletic clade Spartina should simply and efficiently be treated as the genus Spartina.

Biotic homogenization of wetland nematode communities by exotic Spartina alterniflora in China.

Introduced species may homogenize biotic communities. Whether this homogenization can erase latitudinal patterns of species diversity and composition has not been well studied. We examined this by comparing nematode and microbial communities in stands of native Phragmites australis and exotic Spartina alterniflora in coastal wetlands across 18° of latitude in China. We found clear latitudinal clines in nematode diversity and functional composition, and in microbial composition, for soils collected from native P. australis. These latitudinal patterns were weak or absent for soils collected from nearby stands of the exotic S. alterniflora. Climatic and edaphic variables varied across latitude in similar ways in both community types. In P. australis there were strong correlations between community structure and environmental variables, whereas in S. alterniflora these correlations were weak. These results suggest that the invasion of S. alterniflora into the Chinese coastal wetlands has caused profound biotic homogenization of soil communities across latitude. We speculate that the variation in P. australis nematode and microbial communities across latitude is primarily driven by geographic variation in plant traits, but that such variation in plant traits is largely lacking for the recently introduced exotic S. alterniflora. These results indicate that widespread exotic species can homogenize nematode communities at large spatial scales.

Effects of grasshoppers on prairies: Herbivore composition matters more than richness in three grassland ecosystems.

Understanding how biodiversity affects ecosystem processes is a key question in ecology. Previous research has found that increasing plant diversity often enhances many ecosystem processes, but less is known about the role of consumer diversity to ecosystem processes, especially in terrestrial ecosystems. Furthermore, we do not know how general biodiversity responses are among ecosystem types. We examined the role of insect herbivore (Orthoptera) diversity on plant production using parallel field experiments in three grassland ecosystems (mixed grass prairie, tallgrass prairie and coastal tallgrass prairie) to determine whether the effects of grasshopper diversity were consistent among sites. Using mesocosms, we manipulated orthopteran species richness (0, 1, 2, 3 or 4 species), functional richness (number of functional feeding groups present; 0, 1 or 2 functional groups) and functional composition (composition of functional groups present; mixed-feeders only, grass-feeders only, both mixed-feeders and grass-feeders). Diversity treatments were maintained throughout the experiment by replacing dead individuals. Plant biomass was destructively sampled at the end of the experiment. We found no effect of species richness or functional richness on plant biomass. However, herbivore functional composition was important, and effects were qualitatively similar across sites: The presence of only grass-feeding species reduced plant biomass more than either mixed-feeding species alone or both groups together. Orthopterans had consistent effects across a range of abiotic conditions, as well as different plant community and orthopteran community compositions. Our results suggest that functional composition of insect herbivores affects plant communities in grasslands more than herbivore species richness or functional richness, and this pattern was robust among grassland types.