Multiple constraints cause positive and negative feedbacks limiting grassland soil CO2 efflux under CO2 enrichment.

Terrestrial ecosystems are increasingly enriched with resources such as atmospheric CO2 that limit ecosystem processes. The consequences for ecosystem carbon cycling depend on the feedbacks from other limiting resources and plant community change, which remain poorly understood for soil CO2 efflux, JCO2, a primary carbon flux from the biosphere to the atmosphere. We applied a unique CO2 enrichment gradient (250 to 500 µL L-1) for eight years to grassland plant communities on soils from different landscape positions. We identified the trajectory of JCO2 responses and feedbacks from other resources, plant diversity [effective species richness, exp(H)], and community change (plant species turnover). We found linear increases in JCO2 on an alluvial sandy loam and a lowland clay soil, and an asymptotic increase on an upland silty clay soil. Structural equation modeling identified CO2 as the dominant limitation on JCO2 on the clay soil. In contrast with theory predicting limitation from a single limiting factor, the linear JCO2 response on the sandy loam was reinforced by positive feedbacks from aboveground net primary productivity and exp(H), while the asymptotic JCO2 response on the silty clay arose from a net negative feedback among exp(H), species turnover, and soil water potential. These findings support a multiple resource limitation view of the effects of global change drivers on grassland ecosystem carbon cycling and highlight a crucial role for positive or negative feedbacks between limiting resources and plant community structure. Incorporating these feedbacks will improve models of terrestrial carbon sequestration and ecosystem services.

Expert-augmented machine learning.

Machine learning is proving invaluable across disciplines. However, its success is often limited by the quality and quantity of available data, while its adoption is limited by the level of trust afforded by given models. Human vs. machine performance is commonly compared empirically to decide whether a certain task should be performed by a computer or an expert. In reality, the optimal learning strategy may involve combining the complementary strengths of humans and machines. Here, we present expert-augmented machine learning (EAML), an automated method that guides the extraction of expert knowledge and its integration into machine-learned models. We used a large dataset of intensive-care patient data to derive 126 decision rules that predict hospital mortality. Using an online platform, we asked 15 clinicians to assess the relative risk of the subpopulation defined by each rule compared to the total sample. We compared the clinician-assessed risk to the empirical risk and found that, while clinicians agreed with the data in most cases, there were notable exceptions where they overestimated or underestimated the true risk. Studying the rules with greatest disagreement, we identified problems with the training data, including one miscoded variable and one hidden confounder. Filtering the rules based on the extent of disagreement between clinician-assessed risk and empirical risk, we improved performance on out-of-sample data and were able to train with less data. EAML provides a platform for automated creation of problem-specific priors, which help build robust and dependable machine-learning models in critical applications.

CO2 enrichment and soil type additively regulate grassland productivity.

Atmospheric CO2 enrichment usually increases the aboveground net primary productivity (ANPP) of grassland vegetation, but the magnitude of the ANPP-CO2 response differs among ecosystems. Soil properties affect ANPP via multiple mechanisms and vary over topographic to geographic gradients, but have received little attention as potential modifiers of the ANPP-CO2 response. We assessed the effects of three soil types, sandy loam, silty clay and clay, on the ANPP response of perennial C3 /C4 grassland communities to a subambient to elevated CO2 gradient over 10 yr in Texas, USA. We predicted an interactive, rather than additive, effect of CO2 and soil type on ANPP. Contrary to prediction, CO2 and soil additively influenced grassland ANPP. Increasing CO2 by 250 µl l-1 increased ANPP by 170 g m-2 across soil types. Increased clay content from 10% to 50% among soils reduced ANPP by 50 g m-2 . CO2 enrichment increased ANPP via a predominant direct effect, accompanied by a smaller indirect effect mediated by a successional shift to increased dominance of the C4 tallgrass Sorghastrum nutans. Our results indicate a large, positive influence of CO2 enrichment on grassland productivity that resulted from the direct physiological benefits of CO2 augmented by species succession, and was expressed similarly across soils of differing physical properties.

Flowering in grassland predicted by CO2 and resource effects on species aboveground biomass.

Continuing enrichment of atmospheric CO2 may change plant community composition, in part by altering the availability of other limiting resources including soil water, nutrients, or light. The combined effects of CO2 enrichment and altered resource availability on species flowering remain poorly understood. We quantified flowering culm and ramet production and biomass allocation to flowering culms/ramets for 10 years in C4 -dominated grassland communities on contrasting soils along a CO2 concentration gradient spanning pre-industrial to expected mid-21st century levels (250-500 µl/L). CO2 enrichment explained up to 77% of the variation in flowering culm count across soils for three of the five species, and was correlated with flowering culm count on at least one soil for four of five species. In contrast, allocation to flowering culms was only weakly correlated with CO2 enrichment for two species. Flowering culm counts were strongly correlated with species aboveground biomass (AGB; R2  = .34-.74), a measure of species abundance. CO2 enrichment also increased soil moisture and decreased light levels within the canopy but did not affect soil inorganic nitrogen availability. Structural equation models fit across the soils suggested species-specific controls on flowering in two general forms: (1) CO2 effects on flowering culm count mediated by canopy light level and relative species AGB (species AGB/total AGB) or by soil moisture effects on flowering culm count; (2) effects of canopy light level or soil inorganic nitrogen on flowering and/or relative species AGB, but with no significant CO2 effect. Understanding the heterogeneity in species responses to CO2 enrichment in plant communities across soils in edaphically variable landscapes is critical to predict CO2 effects on flowering and other plant fitness components, and species potential to adapt to future environmental changes.

Accelerated development in Johnsongrass seedlings (Sorghum halepense) suppresses the growth of native grasses through size-asymmetric competition.

Invasive plant species often dominate native species in competition, augmenting other potential advantages such as release from natural enemies. Resource pre-emption may be a particularly important mechanism for establishing dominance over competitors of the same functional type. We hypothesized that competitive success of an exotic grass against native grasses is mediated by establishing an early size advantage. We tested this prediction among four perennial C4 warm-season grasses: the exotic weed Johnsongrass (Sorghum halepense), big bluestem (Andropogon gerardii), little bluestem (Schizachyrium scoparius) and switchgrass (Panicum virgatum). We predicted that a) the competitive effect of Johnsongrass on target species would be proportional to their initial biomass difference, b) competitive effect and response would be negatively correlated and c) soil fertility would have little effect on competitive relationships. In a greenhouse, plants of the four species were grown from seed either alone or with one Johnsongrass neighbor at two fertilizer levels and periodically harvested. The first two hypotheses were supported: The seedling biomass of single plants at first harvest (50 days after seeding) ranked the same way as the competitive effect of Johnsongrass on target species: Johnsongrass < big bluestem < little bluestem/switchgrass, while Johnsongrass responded more strongly to competition from Johnsongrass than from native species. At final harvest, native plants growing with Johnsongrass attained between 2-5% of their single-plant non-root biomass, while Johnsongrass growing with native species attained 89% of single-plant non-root biomass. Fertilization enhanced Johnsongrass' competitive effects on native species, but added little to the already severe competitive suppression. Accelerated early growth of Johnsongrass seedlings relative to native seedlings appeared to enable subsequent resource pre-emption. Size-asymmetric competition and resource-pre-emption may be a critical mechanism by which exotic invasive species displace functionally similar native species and alter the functional dynamics of native communities.

Climate modifies response of non-native and native species richness to nutrient enrichment.

Ecosystem eutrophication often increases domination by non-natives and causes displacement of native taxa. However, variation in environmental conditions may affect the outcome of interactions between native and non-native taxa in environments where nutrient supply is elevated. We examined the interactive effects of eutrophication, climate variability and climate average conditions on the success of native and non-native plant species using experimental nutrient manipulations replicated at 32 grassland sites on four continents. We hypothesized that effects of nutrient addition would be greatest where climate was stable and benign, owing to reduced niche partitioning. We found that the abundance of non-native species increased with nutrient addition independent of climate; however, nutrient addition increased non-native species richness and decreased native species richness, with these effects dampened in warmer or wetter sites. Eutrophication also altered the time scale in which grassland invasion responded to climate, decreasing the importance of long-term climate and increasing that of annual climate. Thus, climatic conditions mediate the responses of native and non-native flora to nutrient enrichment. Our results suggest that the negative effect of nutrient addition on native abundance is decoupled from its effect on richness, and reduces the time scale of the links between climate and compositional change.

A CO2 Concentration Gradient Facility for Testing CO2 Enrichment and Soil Effects on Grassland Ecosystem Function.

Continuing increases in atmospheric carbon dioxide concentrations (CA) mandate techniques for examining impacts on terrestrial ecosystems. Most experiments examine only two or a few levels of CA concentration and a single soil type, but if CA can be varied as a gradient from subambient to superambient concentrations on multiple soils, we can discern whether past ecosystem responses may continue linearly in the future and whether responses may vary across the landscape. The Lysimeter Carbon Dioxide Gradient Facility applies a 250 to 500 µl L-1 CA gradient to Blackland prairie plant communities established on lysimeters containing clay, silty clay, and sandy soils. The gradient is created as photosynthesis by vegetation enclosed in in temperature-controlled chambers progressively depletes carbon dioxide from air flowing directionally through the chambers. Maintaining proper air flow rate, adequate photosynthetic capacity, and temperature control are critical to overcome the main limitations of the system, which are declining photosynthetic rates and increased water stress during summer. The facility is an economical alternative to other techniques of CA enrichment, successfully discerns the shape of ecosystem responses to subambient to superambient CA enrichment, and can be adapted to test for interactions of carbon dioxide with other greenhouse gases such as methane or ozone.

Soil animal responses to moisture availability are largely scale, not ecosystem dependent: insight from a cross-site study.

Climate change will result in reduced soil water availability in much of the world either due to changes in precipitation or increased temperature and evapotranspiration. How communities of mites and nematodes may respond to changes in moisture availability is not well known, yet these organisms play important roles in decomposition and nutrient cycling processes. We determined how communities of these organisms respond to changes in moisture availability and whether common patterns occur along fine-scale gradients of soil moisture within four individual ecosystem types (mesic, xeric and arid grasslands and a polar desert) located in the western United States and Antarctica, as well as across a cross-ecosystem moisture gradient (CEMG) of all four ecosystems considered together. An elevation transect of three sampling plots was monitored within each ecosystem and soil samples were collected from these plots and from existing experimental precipitation manipulations within each ecosystem once in fall of 2009 and three times each in 2010 and 2011. Mites and nematodes were sorted to trophic groups and analyzed to determine community responses to changes in soil moisture availability. We found that while both mites and nematodes increased with available soil moisture across the CEMG, within individual ecosystems, increases in soil moisture resulted in decreases to nematode communities at all but the arid grassland ecosystem; mites showed no responses at any ecosystem. In addition, we found changes in proportional abundances of mite and nematode trophic groups as soil moisture increased within individual ecosystems, which may result in shifts within soil food webs with important consequences for ecosystem functioning. We suggest that communities of soil animals at local scales may respond predictably to changes in moisture availability regardless of ecosystem type but that additional factors, such as climate variability, vegetation composition, and soil properties may influence this relationship over larger scales.

Precipitation legacies in desert grassland primary production occur through previous-year tiller density.

In arid ecosystems, current-year precipitation often explains only a small proportion of annual aboveground net primary production (ANPP). We hypothesized that lags in the response of ecosystems to changes in water availability explain this low explanatory power, and that lags result from legacies from transitions from dry to wet years or the reverse. We explored five hypotheses regarding the magnitude of legacies, two possible mechanisms, and the differential effect of previous dry or wet years on the legacy magnitude. We used a three-year manipulative experiment with five levels of rainfall in the first two years (-80% and -50% reduced annual precipitation (PPT), ambient, +50% and +80% increased PPT), and reversed treatments in year 3. Legacies of previous two years, which were dry or wet, accounted for a large fraction (20%) of interannual variability in production on year 3. Legacies in ANPP were similar in absolute value for both types of precipitation transitions, and their magnitude was a function of the difference between previous and current-year precipitation. Tiller density accounted for 40% of legacy variability, while nitrogen and carryover water availability showed no effect. Understanding responses to changes in interannual precipitation will assist in assessing ecosystem responses to climate change-induced increases in precipitation variability.

Response of dominant grass and shrub species to water manipulation: an ecophysiological basis for shrub invasion in a Chihuahuan Desert grassland.

Increases in woody vegetation and declines in grasses in arid and semi-arid ecosystems have occurred globally since the 1800s, but the mechanisms driving this major land-cover change remain uncertain and controversial. Working in a shrub-encroached grassland in the northern Chihuahuan Desert where grasses and shrubs typically differ in leaf-level nitrogen allocation, photosynthetic pathway, and root distribution, we asked if differences in leaf-level ecophysiology could help explain shrub proliferation. We predicted that the relative performance of grasses and shrubs would vary with soil moisture due to the different morphological and physiological characteristics of the two life-forms. In a 2-year experiment with ambient, reduced, and enhanced precipitation during the monsoon season, respectively, the encroaching C(3) shrub (honey mesquite Prosopis glandulosa) consistently and substantially outperformed the historically dominant C(4) grass (black grama Bouteloua eriopoda) in terms of photosynthetic rates while also maintaining a more favorable leaf water status. These differences persisted across a wide range of soil moisture conditions, across which mesquite photosynthesis was decoupled from leaf water status and moisture in the upper 50 cm of the soil profile. Mesquite's ability to maintain physiologically active leaves for a greater fraction of the growing season than black grama potentially amplifies and extends the importance of physiological differences. These physiological and phenological differences may help account for grass displacement by shrubs in drylands. Furthermore, the greater sensitivity of the grass to low soil moisture suggests that grasslands may be increasingly susceptible to shrub encroachment in the face of the predicted increases in drought intensity and frequency in the desert of the southwestern USA.