Matching data on habitat richness, species information was available for the whole region with the same spatial resolution . While the grid is the same used by the Spanish Inventory of Terrestrial Species , the BDD is a distinct data source with more detailed and up-to-date information for the study area. The most recent Spanish land cover/land use map was then used to determine the percentage cover of agricultural land per 10 × 10 km grid cell, information used in the modelling exercise to control for the share of farmland within the landscape . SIOSE allows an accurate estimation of land cover due to a small minimum mapping unit and because it includes an estimation of the share of a given cover within each polygon, even if it is not the dominant cover within the polygon. In this research, we analysed the relationship between farming systems and biodiversity at the landscape level, using farm-level data on farmers’ practices and available biodiversity data. Overall, our analysis resulted in the identification of seven different FS distributed across Galicia: three cattle-based farming systems, one dominated by annual crops, one where permanent crops prevailed, and two non-specialized FS . Based on farm-level data from the IACS/ LPIS dataset, which includes data on farmers’ management reported under CAP payments, we characterized each individual FS in what concerns their spatially-explicit distribution and respective farming practices, in line with previous research performed in other socialecological contexts . Cattle-based FS were found to prevail in Galicia , followed by Non-specialized FS . While the prevalence of livestock –based farming in Galicia was previously reported , our study discriminated three Cattle-based FS based on a decreasing gradient of intensity from the most intensive in the west, intermediate in the central area, to less intensive farms located in eastern Galicia . Such gradient of decreasing intensity, depicted by decreasing levels of livestock density and shares of forage crops, mobile grow system and increasing shares of pastures, seem to relate to an increasing number of biophysical constraints to agricultural practices, namely increasing altitude, slopes and remoteness .
The biophysical characteristics, along with farmers’ decisions mainly driven by agricultural policies, have been acknowledged as major drivers of FS occurrence . Without a relevant expression in the region, FS dominated by annual and permanent crops were found to have restricted distributions. FS 4, dominated by annual crops, was found in the deep soils and mild slopes of the region of A Limia , whereas FS 5, dominated by permanent crops , was found on productive lowlands located close to the sea or in the valleys of rivers, under Protected Designation of Origin . Finally, two non-specialized FS were found widespread across the whole study area and depicting low-intensity small farms owning a low number of livestock and producing mainly forage . The occurrence of small family farm holdings within complex mosaics and parcels was previously reported for Galicia and related to the failure of land consolidation programmes in a region characterized by the occurrence of biophysical constraints limiting agricultural land use . Farming systems approaches based on spatially-explicit farm-level data allow identifying areas sharing a set of similar agricultural management practices e.g. Ribeiro et al. ; Santos et al. . An analysis of FS at the landscape scale allowed pinpointing seven types of landscapes mostly coincident with the seven referred FS, and thus, as expected, exhibiting similar characteristics and distribution patterns across the region, with the exception of LT 2, characterized by a mosaic where FS 2 and FS 3 co-occur. This fact stems from changing the scale of analysis from farm to landscape level, which allowed capturing transition landscape types, specifically reflecting the co-occurrence of intermediate to extensive cattle-based FS along the western-eastern gradient of decreasing intensity. Nevertheless, the coherence between results obtained at the regional and landscape scales highlight the suitability of IACS and LPIS datasets to support the assessment and monitoring of the impacts of farming practices on the environment at multiple scales of decision .
The relationship between the occurrence of species and habitats, including those of conservation concern, and specific farming systems, has been previously advocated . An increasing number of habitats was found associated to a gradient of decreasing intensity from LT1 to LT3 . Higher values of total habitats richness were also recorded within LTs dominated by non-specialized and permanent crops . Still, no significant differences between LTs were found when modelling habitat richness across LTs , except for landscapes dominated by non-specialized farms and dominated by annual crop farms , for which lower values of total and priority habitats richness were observed, respectively. In both cases, lower richness of habitats was coincident with landscapes where the area occupied by farming is higher. Our results are partially supported by Rotch´es-Ribalta et al. that recently reported increasing richness of semi-natural habitats with decreasing levels of farming intensity when analyzing two agricultural landscapes in Ireland. In a European study, García-Feced et al. also reported higher abundance of semi-natural habitats in Northern Spain in marginal farmlands in mountain areas under less intensive farming practices. Landscapes dominated by low-intensity farming systems have been shown to support higher levels of protected species diversity when compared to farmlands under intensive management . Overall, increasing richness of protected species was observed for birds, reptiles, and mammals along the western-eastern gradient of decreasing agricultural intensity, in line with the results achieved for habitat diversity . Our results are in line with previous research. Maskell et al. , observed higher diversity of birds of conservation concern in areas exhibiting higher habitat diversity. Gentili et al. found a decrease of small mammals’ diversity with increasing agricultural intensification and resulting loss of landscape naturalness and complexity in farmlands in northeastern Italy. Biaggini et al. reported low diversity of reptile assemblages in landscapes dominated by intensive agriculture located in Central Italy. Contrastingly, trends for increasing amphibian richness were observed across a gradient of increasing intensity, from landscapes dominated by extensive cattle farms in the east to intensive cattle farms in the west . Still, such patterns relate to global patterns of their distribution in the territory , where higher amphibian richness is observed in western coastal areas and nearby mountains and are related to higher levels of rainfall and lower thermal amplitudes. Relevant differences were detected when modelling richness across cattle dominated LTs, showing higher bird, mammal, flora and habitat richness in landscapes dominated by extensive cattle farms . Bird richness was also found to be significantly higher in landscapes where annual and permanent crops and non-specialized farms prevail when comparing with cattle-based FS . Significant differences were also observed within cattle-based LTs, with the higher richness associated to extensive farming.
The observed pattern may be related to lower intensity of agricultural management of non –specialized farms , reflected by higher landscape heterogeneity due to natural and/or semi-natural areas and uncultivated patches occurrence embedded in the agricultural matrix e.g. see . The significant association between higher bird richness and LTs dominated by annual crops relates to their specific location: A Limia . A Limia is characterized by the occurrence of continental wetlands and coincides with an area designated for conservation of bird diversity – the SPA ES0000436 ‘A Limia’. Thus, the observed pattern is likely due to the coexistence of farmland and the wetland system, rather than to the occurrence of the farmland itself. In fact, farmlands in the area resulted from the process of land reclamation of wetlands that occurred in the late 1950s . Still, significant remnants of the wetland system exist, allowing the occurrence of very vulnerable steppe birds . Finally, modelling results depicted decreasing richness of mammals and reptiles with increasing shares of farmland area . Such results are in line with previous studies reporting a link between higher shares of farmland area and species diversity through increased crop field sizes in more intensively managed and less heterogeneous landscapes e.g. see Martin et al. . Overall, our results suggest that the natural value of agricultural landscapes in our study area increases from the lowlands in the coast, towards mountain areas along a west-east gradient of decreasing agricultural intensity. Such results diverge from previous research by Olivero et al. and Gonz´ alez-García et al. , in which High Nature Value farmlands were associated with farming areas under higher agricultural management intensity . Such contrasting results seem related to the different approaches used, including the taxonomic groups considered, criteria used to select species and analytical approaches implemented. While an extended list of plants, reptiles, amphibians, birds and mammals of conservation interest showing a relation to agricultural landscapes were considered in this study, Olivero et al. included only species protected by Spanish legislation and species associated to agricultural landscapes, regardless of their conservation interest. Selection of indicator species in this study followed consolidated guidelines in the field of High Nature Value farmlands assessment , providing, in our view, a more accurate description of HNVf systems. Altogether, such differences, including the temporal mismatch of data reflecting farm-level management used, may explain the contrasting results, and highlight the importance of the selection of indicators and the definition of common approaches to assure comparability of assessments and monitoring in space and time. Still, our results are consistent with previous research obtained for the targeted region performed at the EU level e.g. see Paracchini et al. . While our results are promising,mobile vertical rack there is room for improvement and further research.
The mismatch between the spatial and temporal resolutions of farm-level management and species and habitats data hindered a fine-scale assessment of the impact of agricultural management on biodiversity patterns. In fact, the best available biodiversity data, a 10 × 10km presence-only dataset, do not include information on the abundance of species, which would be relevant to further scrutinize patterns of biodiversity across farming systems. The use of habitats and species protected under the different legal regulations at the European, Spanish and Galician level, is an asset for the work, since they allow an evaluation of the impact of agricultural management on the most vulnerable elements of biodiversity. In contrast, while being the best information available for the study area, biodiversity datasets target mainly species under any legal protection , are unbalanced for some taxonomic groups, often dominated by mobile species, thus providing a limited overview of the species richness across the taxonomic groups considered, and likely to impact the observed patterns. As an example, while expected, no significant relationship was found between plant richness and the distribution of different farming systems , fact that may reflect the lower number of plants in the dataset as being linked to the occurrence of agricultural landscapes.Further, while not within the scope of this research, habitat richness patterns could be scrutinized by focusing only on those known to be fully or partly dependent on agricultural practices, and by comparing patterns inside and outside Natura 2000 sites. While based on farm-level data and state-of-the-art approaches, results from our modelling should consider the dominance of specific farming systems across the study area. In fact, cattle-based farming systems and the resulting landscape types include most of the farms and thus prevail in the study area, potentially masking any significant link between other farming systems and biodiversity. Moreover, IACS dataset does not include other relevant information, such as N input, or the share of irrigated area that could be useful to characterize the identified FS from a management intensity perspective. Also, IACS/LPIS are provided after a process of anonymization, assuring that farmers’ identity and the geographical location of each agricultural plot are not disclosed, which limited our ability to account for spatial complexity in our assessment. Finally, while IACS dataset includes all farm holdings under CAP payments, it does not integrate information from other farmers and respective farm holdings, namely family farms, and part-time or retired farmers’, which may be reflected by the patterns observed at the landscape level. Farming systems are a complex component of food systems. They are primarily focused on food production. Farming systems consist of all the subcomponents that are pertinent to farm production, but they have blurred boundaries with other components of the food system and involve a wide range of stakeholders and players .