Abteilung Systemanalyse, Integrated Assessment und Modellierung

Traditional hydrological theories are based on the assumption that soil is key in determining water's fate in the hydrological cycle. According to these theories, soil hydraulic properties determine water movement in both saturated and unsaturated zones, described by matrix flow formulas such as the Darcy-Richards equations. They also determine plant-available moisture and thereby control transpiration. Here we argue that these theories are founded on a wrong assumption. Instead, we advocate the reverse: the terrestrial ecosystem manipulates the soil to satisfy specific water management strategies, which are primarily controlled by the ecosystem's reaction to climatic drivers and by prescribed boundary conditions such as topography and lithology. According to this assumption, soil hydraulic properties are an effect rather than a cause of water movement. We further argue that the integrated hydrological behaviour of an ecosystem can be inferred from considerations about ecosystem survival and growth without relying on internal-process descriptions. An important and favourable consequence of this climate- and ecosystem-driven approach is that it provides a physical justification for catchment models that do not rely on soil information and on the complexity associated with the description of soil water dynamics. Another consequence is that modelling water movement in the soil, if required, can benefit from the constraints that are imposed by the embedding ecosystem. Here we illustrate our ecosystem perspective of hydrological processes and the arguments that support it. We suggest that advancing our understanding of ecosystem water management strategies is key to building more realistic hydrological theories and catchment models that are predictive in the context of environmental change.

Data imbalance is a common problem in machine learning that can have a critical effect on the performance of a model. Various solutions exist but their impact on the convergence of the learning dynamics is not understood. Here, we elucidate the significant negative impact of data imbalance on learning, showing that the learning curves for minority and majority classes follow sub-optimal trajectories when training with a gradient-based optimizer. This slowdown is related to the imbalance ratio and can be traced back to a competition between the optimization of different classes. Our main contribution is the analysis of the convergence of full-batch (GD) and stochastic gradient descent (SGD), and of variants that renormalize the contribution of each per-class gradient. We find that GD is not guaranteed to decrease the loss for each class but that this problem can be addressed by performing a per-class normalization of the gradient. With SGD, class imbalance has an additional effect on the direction of the gradients: the minority class suffers from a higher directional noise, which reduces the effectiveness of the per-class gradient normalization. Our findings not only allow us to understand the potential and limitations of strategies involving the per-class gradients, but also the reason for the effectiveness of previously used solutions for class imbalance such as oversampling.

We review the main methods used to study spin glasses. In the first part, we focus on methods for fully connected models and systems defined on a tree. These include the replica method, the Thouless-Anderson-Palmer formalism, the cavity method, and the dynamical mean-field theory. In the second part, we deal with the description of low-dimensional systems, mostly in three spatial dimensions, which are mostly studied through numerical simulations. We conclude by mentioning some of the main open problems in the field.

In partially migratory species, individuals either migrate at some point(s) in life or reside within their natal habitat throughout life. For salmonid fish, migration creates opportunities for feeding and growth, but it is also associated with increased mortality risk. Such trade-offs likely differ between the sexes, since reproductive output is more closely tied to body size in females than males. However, testing hypotheses on sex-specific migratory behaviour in would-be first-time migratory salmonids is difficult, since sexes are generally morphologically indistinguishable prior to maturation. Previous studies have evaluated the influence of sex on migration based on dissection of migratory juveniles or the sex ratio of returning adults. However, both approaches are potentially biased by differential survival during migration. Here, we utilise advances in minimally invasive genetic sex-determination methods for salmonids to investigate sex-specific, spring out-migration propensity in potamodromous brown trout (Salmo trutta) in a pre-Alpine, central European lake. We show that there are marked differences in migratory behaviour between males and females, with small (~10 cm) females being approximately twice as likely to migrate out of their natal river in spring compared to similarly sized males, which generally migrate for the first time at larger sizes (in similar proportions to larger females). This study highlights how novel genetic sex-determination techniques can provide insight into the sex- and size-specific life-history trade-offs that shape migration propensity. Moving forward, these techniques should become useful tools for ecologists and fisheries managers.

The Russia–Ukraine conflict reduced global wheat supplies, yet the food security implications vary across countries. We identify a 39% decrease in Ukrainian wheat exports in 2022 resulting in >70% import losses in some of the countries most vulnerable to these disruptions, with substantial impacts felt in Egypt, Oman, Saudi Arabia, Libya, Mauritania, Yemen and Lebanon. Differential impacts are a function of access to capital and international trade, suggesting the need for policy measures to defuse the impending food crisis.