Ecological Modelling 251 (2013) 73–84 Contents lists available at SciVerse ScienceDirect Ecological Modelling jo ur n al homep ag e: www.elsevier.com/locate/ecolmodel Grazing management or physiography? Factors controlling vegetation recovery in Mediterranean grasslands Carlos P. Carmona a,∗ , Achim Röder b , Francisco M. Azcárate a , Bego ˜ na Peco a a Terrestrial Ecology Group (TEG), Departamento de Ecología. Universidad Autónoma de Madrid, C/Darwin, 2, Edificio de Biología, 28049 Madrid, Spain b Remote Sensing Department, FB VI Geography/Geosciences, University of Trier, Campus II, D-54286 Trier, Germany a r t i c l e i n f o Article history: Received 29 June 2012 Received in revised form 3 November 2012 Accepted 16 December 2012 Available online 16 January 2013 Keywords: Cost surfaces Boosted Regression Trees Grazing gradients Rangeland management Mediterranean Spectral Mixture Analysis a b s t r a c t Grazing intensification and abandonment are increasing the risk of degradation of Mediterranean grass- lands. The development of techniques for monitoring grazing effects on herbaceous vegetation is an essential need for the management of these rangelands. However, the high variability of these systems make physiographical and management effects hard to disentangle and quantify. We present a method- ology to support rangeland management and assess grazing effects on environmentally heterogeneous areas, and provide an example of its application in a Mediterranean rangeland in central Spain. We eval- uated the difference in photosynthetically active vegetation cover between spring and summer using Spectral Mixture Analysis of very high spatial resolution (2.4 m) Quickbird images. To analyze this dif- ference, we developed Boosted Regression Trees models using grazing management (accumulated cost distance to points of livestock concentration and water points) and physiographical variables (slope, wetness, proximity to the closest tree and orientation). Results show that the main factor determining changes in vegetation cover is habitat type. The magnitude of this change was maximized at interme- diate grazing pressures for humid habitats, suggesting the existence of an optimal level of grazing in this zones, while in dry habitats differences in vegetation cover increased consistently along with graz- ing pressure. Our models provided a valuable insight into how different variables and its interactions affect the observed recovery capacity of vegetation. Moreover, by identifying areas in which grazing- induced land degradation could be taken place, our methodology can be used as a powerful tool in the management of highly heterogeneous rangelands. © 2012 Elsevier B.V. All rights reserved. 1. Introduction Extensive grazing is a fundamental factor in the shaping and maintenance of Mediterranean grasslands. Depending on pro- ductivity, grazing history and intensity, grazing can both increase or decrease grassland diversity and can be a significant agent of land degradation (Cingolani et al., 2005; Thompson et al., 2009; Kleinebecker et al., 2011; Carmona et al., 2012). In Mediterranean rangelands, soil fertility, water retention capacity and vegetation cover are enhanced at moderate levels of grazing (Peco et al., 2006, 2012). Nevertheless, when grazing intensity is too high, trampling and defoliation can reduce vegetation cover and increase the level of soil compaction, increasing the risk of soil erosion and productivity loss (Thornes, 2007). Mediterranean environments are experiencing an intensification process in the more pro- ductive areas, associated with the abandonment of others and ∗ Corresponding author. Tel.: +34 91 4972780; fax: +34 91 4978001. E-mail addresses: carlos.perez@uam.es, carlosperezcarmona@yahoo.es (C.P. Carmona). environmental degradation (Stoate et al., 2009). Extensive grazing management practices are shifting to more intensive regimes, in which large-scale free range grazing is preferred over traditional shepherding, sheep are being substituted by cattle, and many extensive farms and transhumance practices are being abandoned (Peco et al., 2001). In this context, the development of techniques for monitoring the effects of grazing activities on vegetation is essential for rangeland management. Assessment of the effects of grazing on vegetation and soil has traditionally been performed either by comparing farms with different grazing pressures (e.g. Peco et al., 2006, 2012) or by exam- ining grazing gradients around water points. In the latter approach rangeland condition is assessed over broad spatial and tempo- ral scales by relating vegetation cover with animal distribution patterns around water sources (e.g. Blanco et al., 2008). Remote sensing data fulfill the requisites of repetitiveness, objectivity and consistency and are observable over large areas (Röder and Hill, 2009), being a very important tool for monitoring grazing effects. The main assumption of the grazing gradient technique, developed to be used in arid Australian rangelands, is that grazing effects decrease as distance from water increases. These effects are usually 0304-3800/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.ecolmodel.2012.12.005