The ecology of ditches. A modelling perspective

The Netherlands is well-known for its extended networks of drainage ditches, with a total ditch length of about 300.000 km. Their main function is to enable agriculture by draining water. Nonetheless, ditches also have important ecological functions. They serve as ecological corridors and harbor a high biodiversity in which water plants play a crucial role. The last decades, the ecological quality of ditches is at stake. Enhanced nutrient inputs promoted the invasion by dense mats of free-floating plants like duckweed. Underneath these mats the water becomes dark and anoxic, which severely constrains aquatic life. In this thesis I developed new concepts to better understand, predict and combat the dominance by free-floating plants in ditches. The following questions are addressed. Are floating plants a self-stabilizing state - an alternative stable state - which would make it more difficult to combat floating-plant dominance (chapter 2)? Does it make sense to fight floating-plant dominance by reducing nitrogen (N) inputs to the ditches or will it lead to an invasion of floating plants that can fix N2 from the atmosphere (chapter 3)? What about spatial aspects, does the vulnerability of a ditch to floating plants depend on the position of a ditch in a polder, like its distance to the polder outlet (chapter 4)? To answer these questions, I used ecological models that predict the abundance of free-floating plants based on the competition for nutrients and light with other plants such as submerged plants, and where possible validated these models with field data. Starting from the ecosystem model PCDitch, I developed and combined models with different complexity to see how theoretical concepts, developed in minimal models, translate to the ecosystem level. Chapter 5 deals with a method that facilitates this up- and downscaling in model complexity. Are floating plants an alternative stable state? To answer this question I extended mechanistic resource competition theory with a framework (minimal model) describing the competition of floating and submerged plants for light and nutrients. The model predicts that the competitive advantage of floating plants - they have a primacy for light and shade submerged plants, giving rise to asymmetry in competition for light - makes that floating plants always dominate at high supply of light and nutrients. At intermediate nutrient supply, there can be alternative stable states: either the submerged plants or the floating plants dominate depending on who established first. However, based on the traits of common floating plants (duckweed; Lemna ) and submerged plants (waterweed; Elodea ) the model predicts, in line with field data, that floating plants are not an alternative stable state. Furthermore, from a theoretical point of view this study shows that the asymmetry in light competition ensures that common rules from standard competition theory do not apply anymore. Like the R* rule, which states that the species that can persist at the lowest resource levels always wins the competition. Can duckweed-dominance be combatted by reducing N inputs to the ditches? Or does this promote other floating plants like water fern ( Azolla ) that can fix N2 from the atmosphere? Important is the question whether such N2-fixers can provide enough N to prevent N-limitation and keep the system P-limited, which would make steering on N inputs ineffective. To investigate this, I considered the competition between Lemna and Azolla for N, P and light. Both a minimal model, an ecosystem model (PCDitch) and field data reveal that N2-fixation is unlikely to lead to P-limitation. This can be explained by N2-fixers typically requiring higher P concentrations to persist, implying that they cannot keep the P concentration low enough for non-N2-fixers to become P-limited. In combination with field data that hint at constraints on N2-fixation that prevent N2-fixers from becoming abundant at low N availability, this suggests that it certainly pays off to combat floating plant-dominance by reducing N inputs. Is every ditch in a polder equally vulnerable to floating plants? Each ditch in a polder receives water and nutrients from the adjacent land. This leads to a spatial gradient in water flow and associated nutrient loading, from low in the remote polder sites to high in the direction of the polder outlet where the water leaves the polder. I explored if this spatial gradient affects the vulnerability of a ditch to floating plants, by investigating with a simple nutrient model how this gradient affects the nutrient concentration of the ditches and by subsequently predicting the gradient's effect on the ditch ecology by applying the ecosystem model PCDitch spatially, through coupling PCDitch to the 1-D hydrodynamic model SOBEK. Surprisingly, we found that every ditch is equally vulnerable to floating plants, despite the spatial gradient in water flow and nutrient loading. It turned out that the ecological state of each ditch could already be predicted by regarding only the lateral supply of water and nutrients from the adjacent land, and not the supply from upstream ditches. However, these findings are violated when there is spatial heterogeneity in the water and nutrient supply from the adjacent land or in ditch characteristics like depth and sediment type. Then, the chance on floating-plant dominance differs throughout the network and a spatial modelling approach (PCDitch-SOBEK) is required to predict this chance. Developing and combining models of different complexity plays an important role in this thesis. To do so, I used a Database Approach To Modelling (DATM), a recently developed method in which a model is stored in tables in a clear and clean way, which facilitates model development. In addition, with DATM a model can be automatically implemented in a modelling environment of choice. This relieves technical implementation issues and leaves room to focus on ecology rather than technology. I illustrated the use of DATM by implementing and analyzing the ecosystem model PCDitch and its twin model for shallow lakes PCLake in different modelling environments by using DATM. This showed that DATM allows one to use the environment one is familiar with and eases the switch to other environments for complementary analyses, including analysis in a spatial 1-D to 3-D setting. The insights provided by this thesis can help us to improve the ecological quality of ditches. A challenging task, given the fast human-driven environmental changes at both local and global level. To predict and to anticipate the effect of these changes on the ecology, it is essential to understand how the ditch ecosystem functions. The developed and applied methods described in this thesis may be helpful in that. For example, using models of different complexity makes it possible to translate fundamental theory to the ecosystem scale, which is essential to better grasp the behavior of an ecosystem. Furthermore, the in this thesis established coupling between PCDitch and SOBEK breaks new grounds for spatial ecosystem modelling. In combination with the growing amount of remote sensing data from satellites and drones, which allow for the continuous and potentially real-time validation and calibration of spatial ecosystem models, such a spatial approach has the potential to greatly increase our ecological understanding of ditches. These advances facilitate the development of successful management strategies that make our ditch ecosystems future-proof.