Search

Biogeochemical models of vegetation dynamics could potentially be used to complement empirical studies on the effect of plant species richness. A key precondition is the simulation of species coexistence. While community scale models regularly incorporate respective processes, models at the field or landscape scale used for larger scale assessments, require additional model development. However, it is unclear how the particular process description within these models affects simulations of species performance and resulting ecosystem functions. We compare simulations of two grassland models of different complexity for monocultures and two-species mixtures in a grassland experiment in Jena, Germany. By providing an in-depth analysis of the models' process descriptions, we evaluate their ability to simulate the response of different species, their interactions and their joint performance to drought and mowing. Both models simulated similar average above-ground biomass (AGB) but showed different intra-annual variability. Generally, the models had difficulties representing a balanced species composition in multiple species mixtures and competition for space was the main driver of community composition in both models. The resulting communities were dominated by the more competitive species, while the weak competitor was only marginally present in most mixtures independent of drought and mowing. The competitive strength which we derived from the calibrated parameter sets of the species differed between the models and the agreement on which species dominate specific mixtures was mixed. While both models simulated reduced soil water content and above-ground biomass in response to drought, the strength and duration of these responses differed. Despite these differences, simulated species interactions were barely affected, and strong competitors remained dominant. Mowing had opposing effects on the competition for space in the models, which could be attributed to the different representations of plants in the two models. The models selected for the comparison are two representatives for local- and large-scale applications and use widely applied approaches for which our comparison highlighted strengths and weaknesses. To enable the investigated models (and those with similar complexity) to simulate coexistence of multiple species, niche differentiation needs to be improved. This requires a stricter separation of access to different resources and improved representation of different ecological strategies for which community scale models that are able to simulate coexistence may be an inspiration. Our approach may serve as an example for other modellers looking for ways to identify important model processes for further model development in the context of species interaction. • The model comparison showed similar biomass (AGB) with different seasonal patterns. • Competition for space was the main driver of species interaction. • Assessed grassland models have deficiencies in simulating stable coexistence. • To improve niche differentiation, incorporation of below-ground traits is needed. • Approaches from community scale models could inspire models at larger scales. [ABSTRACT FROM AUTHOR]

• Plant diversity and productivity are involved in explaining δ15N plant , δ15N soil, and δ15N soil-to-plant in North Tibet. • Precipitation is critical for driving spatial variability of δ15N plant , δ15N soil and δ15N soil-to-plant in North Tibet. • The δ15N plant and δ15N soil-to-plant indicate growth limitation shifts from P in meadows to N in desert-steppes. Nitrogen (N) cycling is a critical pathway by which producer, consumer, and decomposer interact with each other and with environmental circumstances simultaneously. The natural abundance composition of 15N/14N in plants and soils (termed as δ15N plant and δ15N soil), as well as the difference between them (δ15N soil-to-plant = δ15N plant −δ15N soil), is a useful tool for better understanding ecosystem N cycling. However, the drivers and mechanisms of ecosystem N cycling in alpine grasslands on the Tibetan Plateau are mostly unknown, especially across different grassland types at a regional scale. To fill this knowledge gap, we measured δ15N plant (200 samples of top-dominant species) and δ15N soil (85 samples of top-layer soils, 0–20 cm) at nine sites that represent zonal communities of alpine deserts, steppes, and meadows in North Tibet, and calculated the corresponding δ15N soil-to-plant. Our results showed that δ15N plant, δ15N soil, and δ15N soil-to-plant were significantly different among the three zonal grassland types (analysis of differences with non-parametric Kruskal Test, P < 0.05), with the lowest values in meadows and the highest values in deserts. Regression analyses showed that the δ15N plant , δ15N soil, and δ15N soil-to-plant decreased with the increases of growing season precipitation (GSP) and habitat aridity index (Aridity), soil organic carbon (SOC) and soil total nitrogen (STN), plant species richness, Shannon diversity index, and plant community productivity, whereas increased with the increases of accumulated active temperature (AccT) and soil total phosphorus (STP) across alpine grassland types at the regional scale. Multiple linear models with analysis of covariance (ANCOVA) confirmed GSP to be the most critical driver, which alone explained most variances of δ15N plant (56%), δ15N soil (62%), and δ15N soil-to-plant (35%). However, structural equation modeling performed better than multiple linear modeling in predicting δ15N plant (76% vs. 66%) and worse in predicting δ15N soil (79% vs. 84%) and δ15N soil-to-plant (31% vs. 46%), likely due to the exclusion of collinear predictors and the removal of non-significant influencing paths. Overall, this study has highlighted the importance to uncover the complexity of climate, soil nutrients, and vegetation properties in networking to drive the different components of ecosystem N cycling in alpine grasslands on the Tibetan Plateau. [ABSTRACT FROM AUTHOR]