Drylands cover about 41% of the Earth land’s surface (Prãvãlie, 2016) and host 38% of global population (Maestre et al., 2021). Due to intensive agriculture and overgrazing, about 10–20% of drylands are already degraded (FAO, 2019), and this percentage is expected to rise due to population growth and climate change (Reynolds et al., 2007). Land degradation processes in drylands have implications both for ecosystem health and humans, as they reduce water supply and soil fertility, increase soil losses and reduce biodiversity (Ravi et al., 2010; Zhang et al., 2021). Land degradation in drylands can culminate in desertification, i.e., changes in soil properties which reduce its capacity to retain nutrients and water, causing the loss of productivity and serious environmental and socio-economic problems that specially affect people living in developing countries (FAO, 2019).

After the triggering activity of the degradation process has finished, natural recovery of the ecosystem may take a long time (hundreds of years, Weber et al., 2016) or, sometimes, the ecosystem is not capable of recovering by itself (Svejcar and Kildisheva, 2017). Thus, active restoration is needed to facilitate the process, especially in drylands where water scarcity limits plant growth and establishment (Hulvey et al., 2017; Ayangbenro and Babalola, 2021). Dryland restoration is recognized as an urgent need [Food and Agriculture Organization of the United Nations (FAO), 2017; European Environment Agency, 2019], but traditional restoration techniques, such as the introduction of native vascular plants, mostly trees (FAO, 2015) and the increase of water input or nutrient concentration, are often expensive and unsuccessful in drylands (Valdecantos et al., 2014). Thus, there are still numerous limitations to successfully restore degraded drylands (James et al., 2013) and novel efficient methodologies are needed. In this sense, restoration and rehabilitation strategies that are based on natural processes and cycles (nature-based solutions) are receiving a growing attention as they are resource efficient and adapted to systems (Nesshöver et al., 2017). For example, it is well known the substantial impact that spatial organization of bare and vegetated surfaces can have on the retention or loss of resources (water and nutrients) in dryland ecosystems (Boer and Puigdefábregas, 2005; Tang et al., 2021). Thus the installation of resource sinks during the design of the restoration activity can be an effective way to increase restoration success in drylands (Berghuis et al., 2020).

Natural drylands are heterogenous landscapes composed of sparse and patchy vegetation occupying the most favorable positions within the landscape, and biocrusts frequently colonizing the apparently bare open spaces between plants. These are a consortium of different organisms that include cyanobacteria, algae, lichens, mosses, bacteria and fungi, living in intimate association with soil particles, and occupying the first centimeters of soil surface (Weber et al., 2016). Biocrusts possess specific adaptations that provide them with the ability to withstand the high solar radiation and the scarcity of water that characterize dryland regions all over the world (Bowker et al., 2002; Bowker, 2007). By covering the soil surface, biocrusts adhere fine soil particles, stabilizing the soil surface and protecting it from wind and water erosion (Chamizo et al., 2012b; Belnap et al., 2014; Fattahi et al., 2020). They contribute to carbon and nitrogen fixation (Housman et al., 2006), increasing the fertility of dryland soils (Zhang, 2005; Chamizo et al., 2012a). Through their influence on roughness, albedo and porosity, they affect water processes, such as infiltration, runoff, evaporation and soil moisture (Chamizo et al., 2016; Eldridge et al., 2020). Biocrusts are also the support of abundant and diverse microbial and microfauna communities (Darby et al., 2010; Xiao and Veste, 2017). Because of all these functions on soils, they affect the establishment, survival and productivity of vegetation (Havrilla et al., 2019). At the same time, vegetation patches create microenvironments (shade, more humid conditions and less extreme temperatures) that affect biocrust growth and succession (Maestre and Cortina, 2002; Bowker et al., 2005; Bowker, 2007).

Biocrusts are also linked to vegetation survival and productivity in drylands through complex water and nutrient redistribution interactions (Belnap et al., 2005; Rodríguez-Caballero et al., 2018). When a rain event occurs, runoff is generated in the open areas between plants, mostly covered by biocrusts (which, compared to vegetation, act as runoff sources), and redistributed toward adjacent vegetation, which frequently acts as surface obstruction for most of the runoff water and dissolved sediments and nutrients (Belnap et al., 2005; Cantón et al., 2011). This water surplus promotes vegetation productivity and biomass (Rodríguez-Caballero et al., 2018) and increases vegetation capacity to retain more water and nutrients in the next rain events (Belnap et al., 2005; Ludwig et al., 2005). The result of biocrust-vegetation feedback processes is a spatial mosaic in which the coupling between biocrusts (source areas) and vegetation (sinks) determines the spatial arrangement of plants within the landscape (Cantón et al., 2004; Rodríguez-Caballero et al., 2019), in such a way that plants are redistributed to minimize the loss of water, nutrients and sediments in the ecosystem (Rodríguez-Caballero et al., 2015) and to maximize vegetation productivity and coverage (Rodríguez-Caballero et al., 2018).

Taking into account the above-mentioned key roles of biocrusts in dryland ecosystem functioning and their prevalent presence, biocrust restoration after a disturbance has occurred, can be crucial to accomplish the structural and functional recovery of degraded ecosystems (Zhao et al., 2016). Natural biocrust recovery can be slow depending on factors such as climatic and edaphic conditions, and severity and timing of disturbance (Weber and Hill, 2016), as well as biocrust characteristics (Xiao et al., 2019).

Recent papers point to lower biocrust recovery times, ranging from a few years for early successional biocrusts (cyanobacteria) to 10–30 years for later successional biocrusts (lichens and mosses) (Kidron et al., 2020; Phillips et al., 2022). However, even though, initial soil colonization by biocrusts can be greatly retarded or impeded in unstable soils, subjected to high erosion. Soil stabilization and biocrust recovery can be prompted by active restoration through soil inoculation with biocrust organisms. As pioneer colonizers of barren soils and promoters of biocrust succession, cyanobacteria are one of the most interesting organisms to be used as inoculants. Cyanobacteria are able to fix the atmospheric carbon and nitrogen, and stabilize soils surfaces (Mager and Thomas, 2011), improving the habitat for later-successional biocrust organisms such as lichens and mosses (Lázaro et al., 2008; Zhang et al., 2018). Eventually, these more favorable soil conditions will facilitate the establishment and growth of plant species. Recent studies have shown positive effects of soil cyanobacteria inoculation in the reduction of soil erosion and improvement of soil properties (Park et al., 2017; Sadeghi et al., 2017; Chamizo et al., 2018; Muñoz-Rojas et al., 2018; Román et al., 2018; Antoninka et al., 2020; Kheirfam et al., 2020). However, there are no previous studies that evaluate the synergic effects of soil inoculation with cyanobacteria combined with a spatial distribution of native vascular vegetation based on the resource use balance. The positive runoff-runon feedback processes between biocrusts and vegetation is one of the most promising nature-based solutions for dryland restoration (Cortina et al., 2011; Rodríguez-Caballero et al., 2019). An adequate design of biocrusts-vegetation spatial distribution can be key for plants to optimize water availability in these environments, where 90% of rainfall events are less than 10 mm (Martínez-Mena et al., 1998; Puigdefábregas, 2005), thus contributing to vegetation and ecosystem recovery. This would allow designing in advance a restoration action that helps to overcome the difficulties faced during field application for both the cyanobacteria inoculum (Román et al., 2021a) and plants, such as the inoculum loss and the enhancement of erosion rates and further reduction of plant cover (Cantón et al., 2021). For this purpose, the influence of biocrust should be included in runoff and erosion simulation models (see Rodríguez-Caballero et al., 2015) to simulate the hydro-geomorphological response of the target area and identify the best design based on the most efficient use of resources.

The main goal of this work is to present a new methodology to support future dryland restoration activities by selecting the most appropriate spatial configuration of plants and biocrusts that maximizes water and nutrient capture by plants, while reducing erosion. To do this, we will first evaluate different restoration scenarios, through spatially distributed modeling, to identify the most optimal spatial configuration of plants that maximizes runoff-water capture by vegetation. Second, we will compare results obtained in the proposed scenario with another scenario including also active biocrust restoration by means of cyanobacteria inoculation, to evaluate its effect on soil erosion.

In dryland ecosystems, most runoff and sediments are generated during the sporadic intense rains, which are responsible for water and sediment redistribution along the landscape (Cantón et al., 2011; García-Ruiz et al., 2015). We simulated runoff and erosion in a degraded hillslope during several rain events occurring between July 2014 and December 2015 using the LISEM model. This is challenging due to the difficulty of simulating the hydrological and erosive response in semi-arid areas where ground cover and soil properties are highly heterogeneous and this variability affects infiltration, runoff and sediment yields (Cantón et al., 2002). Despite the lack of detailed information of total erosion and spatial distribution runoff and erosion records to calibrate the model, our model was able to predict the threshold for runoff generation of 3 mm observed in a nearby area with similar hydrological properties (Rodríguez-Caballero et al., 2015) and the real erosion measured on the slope during the study period, with an error of just 17.8%, thus demonstrating its potential as a tool for planning of restoration actions.

The simulations during the calibrating period show how the presence of spontaneous vegetation on the study hillslope effectively reduces the runoff generation and water erosion (Figure 3), as previously obtained by Chen et al. (2018). Vegetation acts as a runoff sink, infiltrating not only the water that falls directly in the form of precipitation, but also the runoff generated upstream (Ludwig et al., 2005; Puigdefábregas, 2005). This effect can be enhanced by the capacity of vegetation to trap wind-borne sediments, which may result in changes in soil properties such as soil coarsening that may also increase soil infiltration (Gonzales et al., 2018). Besides, vegetation increases water storage capacity and organic carbon and nutrient inputs, promotes biological activity and accelerates plant litter decomposition and nutrient turnover rates, thus resulting in the formation of “islands of fertility” that greatly differ from the nearby bare ground areas (Puigdefábregas, 2005).

At the hillslope scale, runoff interception by vegetation produces a decrease in hydrological connectivity, which ultimately reduces the loss of water, soil and nutrients out of the system, increasing ecosystem productivity (Ludwig et al., 2005). However, the hydrological response of the slope is not controlled only by the responses of the different covers (vegetation and bare soil), but it is also affected by the interaction between them, which ultimately depends on their spatial distribution (Peters et al., 2004). Due to the importance of spatial arrangement of source and sink areas, there are differences in the runoff and erosion response of the experimental hillslope under different vegetation spatial distributions tested during the simulation phase (Tables 4, 6). Nevertheless, considering that all these spatial distributions try to maximize water uptake by the vegetation and reduce the hydrological connectivity, differences among the tested spatial configurations are not very marked. As shown in Tables 5, 7, among all the configurations analyzed the one that maximized infiltration and greater reduced erosion was the scenario 5 (Figures 4, 5) with vegetation located in the areas with the greatest water accumulation (higher WI). This is because plants in this configuration are located in the areas where topography favors the runoff redistribution from bare areas and its storage (WI increases as slope gradient decreases) into the vegetated patches, in a similar way as observed in natural drylands where vegetation occupies the most favorable positions within the landscape (Bonachela et al., 2015; Rodríguez-Caballero et al., 2019). This spatial distribution and associated source-sink interactions between open areas and vegetation increases water availability for plants (Ludwig et al., 2005; Magliano et al., 2015). As water scarcity is the main limiting factor for vegetation survival in traditional dryland restoration (Valdecantos et al., 2014), this extra runoff water contribution is expected to favor vegetation growth and may minimize the death of the reintroduced plants during the first years of restoration, as was already observed during field water capture experiments in the same study site (García-Ávalos et al., 2018). Moreover, water redistribution processes promote pulses of plant growth and increase vegetation potential to capture water and nutrients during future rainy events, as we can see when we compare the simulations with 25 cm plants with the one with 50 cm plants (Tables 5, 7). However, the capacity of vegetation to capture and store the extra contribution of water by runoff is limited (Rodríguez-Caballero et al., 2014), especially in the initial stages of the restoration, when the plant is small. Thus, under high intense and high magnitude rain events, infiltration capacity of vegetation could be exceeded, and this may explain the similar infiltration rates obtained in the high and low intensity runoff simulations representing initial stages of plant development (Table 5). This phenomenon gives rise to a threshold in the hydrological and erosive response of the hillslope, which will increase with vegetation development, and above which water and associated sediments and nutrients are no longer retained by vegetation, getting these resources out of the system. Indeed, a rainfall threshold of 20 mm was found in a nearby area with similar soil types and vegetation (Rodríguez-Caballero et al., 2014), but also in more mesic environments (Nadal-Romero et al., 2018).

Results obtained from runoff and erosion simulations (Figures 4, 5) also corroborated the importance of biocrust recovery for water and erosion control (Gao et al., 2020). It is well known the key role that biocrusts play in hydrological and erosion processes in drylands (Rodríguez-Caballero et al., 2013; Chamizo et al., 2016; Eldridge et al., 2020). Biocrusts increase soil aggregation and soil porosity (Menon et al., 2011; Miralles-Mellado et al., 2011; Bowker et al., 2018), surface roughness (Rodríguez-Caballero et al., 2012), water retention capacity (Wu et al., 2013; Chamizo et al., 2016; Eldridge et al., 2020) and soil stability (Chiquoine et al., 2016). Improvement of these soil properties translates into greater infiltration and lower erosion when compared with bare soil areas, which due to the poor soil structure and low organic matter contents of dryland soils, are often prone to physical crusting (Chamizo et al., 2015). Physical crusts are known to reduce hydraulic conductivity and increase runoff and consequent erosion (Nciizah and Wakindiki, 2015). For this reason, biocrust recovery following a disturbance is a key step to improve soil hydraulic properties and above all, to increase soil stability and its resistance to erosive forces by water and wind (Knapen et al., 2007; Zhao et al., 2014; Fick et al., 2019). In this study, we have simulated the effect of biocrust recovery in the interplant soil by artificially promoting biocrust formation through soil inoculation with cyanobacteria. This strategy has provided successful results in biocrust recovery and promotion of vegetation succession in desertified areas from China (Lan et al., 2013). In all the simulations in which biocrust presence was included, there was a decrease in runoff, and especially in water erosion (Tables 4, 6). However, similar to the results found in the simulations where biocrust effect was not considered, there were important differences depending on the properties of the rainy event. During the low intense event, which was close to the infiltration capacity of the recovered biocrust, almost no runoff was generated in the open areas between vegetation, slightly reducing the contribution of water toward downslope vegetation (Table 4). This water is used by biocrust organisms to reactivate their metabolic activity, fixing C and N (Belnap et al., 2005; Aanderud et al., 2018; Miralles et al., 2018; Williams et al., 2018). Fixed nutrients are incorporated into the soil in the form of organic matter and biomass, increasing soil fertility and stability (Ferrenberg et al., 2017; Bao et al., 2019). During rains of similar intensity but greater magnitude, surface saturation of the first centimeters of soil, which is the layer affected by biocrust presence (Chamizo et al., 2012a), is expected (Rodríguez-Caballero et al., 2013). Under this situation, runoff is generated, and the runoff flow will redistribute part of the nutrients fixed by the biocrust along with the water toward the vegetated areas, increasing their productivity and ability to survive (Figure 6). In any case, runoff generation over biocrust-covered soils is more frequent during intense events, in which rainfall intensity exceeds biocrust infiltration capacity, and runoff occurs (Dunkerley, 2011; Rodríguez-Caballero et al., 2013; Zhao et al., 2014). Part of this runoff is reinfiltrated into the vegetation, but most of it runs on downslope, in a similar way to what happens in the scenario without biocrust (Figure 4). However, contrary to the scenario lacking biocrust, when modeling biocrust active restoration, erosion is strongly reduced (Figure 4 and Table 6).

These findings demonstrate the potential of biocrust inoculation for the restoration of degraded arid and semi-arid areas, where vegetation growth is limited by water availability, and ecosystem recovery to the original stage before disturbance may take a long time (Kidron et al., 2008). Under these conditions, soil inoculation with cyanobacteria to restore the biocrust can prevent erosion during the first stages of soil restoration, while maintaining water supply to downslope vegetation when it is located in strategic positions to maximize water capture coming from biocrust areas. To correctly locate the vegetation, we can rely on runoff and erosion simulation models, which efficiently allow evaluating the effect of different spatial configuration of ground covers, helping optimize restoration practices. Although based on simulated and not real measured data, it is important to highlight the value of this type of exercise prior to a restoration since it will allow us to quickly identify the most efficient strategy before its implementation in the field, being a very useful and low-cost tool for land managers. However, more effort and long-term field measurements on restoration experiments are necessary to calibrate and validate this or similar models in a wide range of environmental conditions.

Application of a spatially distributed runoff and erosion model on a degraded hillslope allowed to identify the most optimal spatial configuration of vegetation to reduce runoff and erosion and increase water availability to the introduced plants. Among all the vegetation spatial configurations analyzed, those that locate the plants in the areas that have a greater contribution area and low to moderate slope gradients were predicted as the most efficient in reducing runoff and erosion and increasing water availability to plants. Although the effect of vegetation in the reduction of runoff and erosion was higher for larger plants, the modeling also foresees an appreciable influence of the recently introduced young plants. Moreover, according to our modeling results, active biocrust restoration by cyanobacteria inoculation reduced runoff and especially erosion. Although runoff decrease reduced the water input to vegetation, the higher amount of water that infiltrates in the interplant soil improves moisture conditions and provides a substantial benefit to the activity of soil microbiota and microfauna, and the activation of biogeochemical cycles, which has also indirect beneficial effects to adjacent vegetation.

Overall, all these results highlight the importance of accounting for the eco-hydrological interactions between biocrust-source areas and vegetation-sink areas when planning restoration strategies in drylands and demonstrates the potential of spatially distributed runoff and erosion models as a powerful tool to design restoration actions in drylands in order to increase its success.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

LM acquired data, performed the modeling analyses, and wrote the manuscript. ER-C acquired data, performed the modeling analyses, and improved manuscript editing. SC acquired data and improved manuscript editing. BR-L performed the modeling analyses and improved manuscript editing. YC conceived the idea, designed the experiment, and improved manuscript editing. All authors contributed with constructive comments to the manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.