Water availability is one of the main limiting factors that control ecosystem functions and productivity in semiarid regions. Vegetation of these regions usually presents a patchy distribution where sparse plant cover is interspersed over a bare soil. During the few rainfall events, runoff is generated in non-vegetated areas and redistributed towards vegetation, which act as surface obstruction for water, sediments and nutrients. Thus, non-vegetated areas are more susceptible to water erosion processes. Non-vegetated areas from semiarid ecosystems around the world, are often covered by Biological Soil Crusts (BSCs). BSCs result from an intimate association between soil particles and cyanobacteria, algae, microfungi, lichens and bryophytes. These communities live within, or immediately on top of, the uppermost millimeters of soil, influencing soil surface properties involved in infiltration, runoff generation and water erosion. Several papers have demonstrated that BSCs are one of the most important soil stabilizing factors in drylands. There are, however, contradictory results on the role that BSCs play in regulating soil water fluxes. Some studies point BSCs as runoff sources that may increase downslope erosion or on the contrary may represent an additional supply of water for downslope vegetation allowing its survival. The impact of this additional runoff should be evaluated at less detailed scales than the patch and to analyze all interactions in terms of water, sediments and nutrients between areas covered by BSCs and vegetated patches in order to establish the real effects of BSCs on both runoff and erosion. Also, to correctly predict the impact of future climate changes or antropic disturbances on hydrological behavior and water erosion in systems dominated by BSCs their effects should be included on spatially distributed runoff and erosion models. Until now, the influence of BSCs on these processes has been addressed almost exclusively at patch scale, despite the fact some authors have pointed the need of upscaling their effects, and even more their influence on runoff generation and water erosion was never considered in spatially implicit medelling. The goal of this thesis is to determine BSC effects on runoff and water erosion from plot to catchment scale in a typical semiarid ecosystem.
To achieve this objective, first direct and indirect effects of BSCs at patch scale must be clearly defined under natural rainfall conditions to solve the controversy about BSCs effects on runoff generation. To know the direct and indirect relationships among soil surface characteristics, BSC cover and type, topography, rainfall characteristics (duration, amount and intensity) and runoff, structural equation models (SEM) were applied. Our results reveal the critical importance of BSCs on runoff and water erosion. Both processes in biologically crusted areas are directly controlled by crust type and cover. BSCs also modified some soil surface properties involved in runoff generation and water erosion, such as microtopography, surface stability or water repellency. The final interaction of both, direct and indirect BSCs effects, determine the hydrological behavior of these surfaces under natural rainfall conditions. Moreover, the final effect of BSCs on runoff generation is strongly driven by rainfall properties, which determined the set of complex interactions among BSCs, type and developmental stage and soil surface properties: on one hand, during low intensity rains, BSC-induced microtopography increases the amount of surface micro-depressions, which act as temporal water sinks, reducing the connectivity among source areas, delaying runoff initiation and reducing runoff rates; on the other hand, during intense rainfall events, BSCs type and water repellency are the main factors determining runoff generation.
When the effects of BSCs are analyzed at coarser scales, including all interactions among BSCs and vegetated areas on a whole catchment, our results reveal the importance of the interactions between areas with BSCs and areas with vegetation on runoff generation and water erosion. We show the capacity of vegetated areas to retain runoff waters generated by upslope biologically crusted areas as an important driver for the hydrological and erosional response at catchment scale. However, the capability of vegetated areas to trap and retain water and sediments is limited and can be exceeded during high magnitude events, increasing catchment connectivity, as well as runoff and water erosion at the catchment outlet. Even during high-magnitude events, when the runoff generated in BSC areas reaches the channel network, the local protection provided by BSCs also affects downslope areas and the catchment response. These results confirm that BSCs must be included in runoff and soil erosion models to obtain reliable predictions of the spatial pattern of runoff and water erosion in catchments with abundant BSCs.
In order to correctly introduce the effects of BSCs in these models, it is necessary to have an accurate spatial characterization of BSCs. It is shown that a spectral mixture analysis is required for the precise characterization of the complex spatial distribution of BSCs, due to the intrinsic spatial heterogeneity of semiarid ecosystems and to the spectral similarities among BSCs, dry vegetation and bare soil. Due to the methodological and practical application problems of spectral mixture analysis when it is applied to spectrally complex areas or when some surface elements only appear in specific areas of the image, we needed to develop a novel methodology for BSCs classification and quantification (lichen and cyanobacteria-dominated CBS), based on hyperspectral images. Support vector machine classification was applied for spectral and ecological classification of homogenous areas to solve the mentioned problems inherent to spatial heterogeneity. Inmediately afterwards, spectral mixture analysis (SMA) was applied to each SVM class to quantify the proportion of each type of surface cover within each pixel. Relative abundance images obtained with this methodology achieve a relatively high accuracy for different types of BSCs, and have demonstrated to be an adequate source of spatially distributed information, to correctly characterize surface properties in biologically crusted drylands systems. Moreover, to have the spatial distribution of type and abundance of BSCs allows to increase the accuracy of modeled runoff and erosion. Thus, when BSCs effects are not included in the LISEM model, an important increase in modeled water erosion was observed in areas where BSCs was not considered.