Carbon and oxygen isoscapes for geographical traceability of italian extra-virgin olive oils

The authentication and verification of the geographical origin of food commodities are important topics in the food sector. Several authors have successfully exploited stable isotope techniques in the characterization, authenticity and traceability of olive oils (Angerosa et al., 1999; Iacumin et al. 2009; Camin et al. 2010a; Camin et al; 2010b, Portarena et al. 2014; Portarena et al., 2015). Nevertheless, no previous studies on extra virgin olive oil (EVOO) traceability have accounted for the spatial component as an independent variable in determination of isotopic signatures. Methods for verifying the geographical origin of food based on a geospatial modelling approach have been developed recently (West et al., 2010; van der Veer, 2013) and, to date only a few studies have been published (West et al., 2007). Previously West and collaborators merged the terms "isotopes" and "landscapes" into "isoscapes", thus defining the geospatial predictive power of stable isotopes, such as those of C, H, O and N involved in biogeochemical processes (West et al., 2010). Such processes produce isotopic fractionations that depend on the geographical location and spatial transport, and determine the spatial variability in the isotopic composition of materials (Bowen et al., 2009). More recently, van der Veer (2013) extended the geospatial model concept to multiple sets of suitable geographical markers (e.g., isotopes of heavy elements, organic compounds, trace elements) in order to better confine the area of possible origin. The general assumption in geospatial modeling for food provenance assessment is that the commodity of interest comes from one or more confined production areas, thus reflecting the peculiar isotopic composition of the provenance. These assumptions are met by some products, such as wine, honey, meat or EVOO, which are all characterized by the intrinsic identity of the areas of origin. We followed a geospatial approach, combining stable isotope analysis with the use of GIS (Geographical Information System) technology and spatial analysis to develop geospatial models for the isotopic composition of Italian EVOOs. This study shows the spatial variability in ?13C and ?18O of 387 samples of EVOO collected in nine Italian regions, from 2009 to 2011, previously analysed by Portarena et al. (2014). EVOOs' ?13C and ?18O values were related to GIS layers of source water ?18O and climate data (mean monthly temperature and precipitation, altitude, xerothermic index) to evaluate the impact of the most significant large-scale drivers for the isotopic composition of Italian EVOOs. We used a hybrid procedure based on the spatial relationship between ?18O and ?13C values and the geo-climatic variables (ancillary variables) of the production areas. Ordinary Least Squares (OLS) analysis allowed us to quantify the positive correlation between EVOO ?18O values and ?18O of long-term average of annual precipitation. We found a positive correlation among ?18O of EVOOs and annual mean temperature, mean temperature of the warmest months, mean precipitations of the spring quarter and the xerothermic index. No significant correlation exists between ?13C values and the ancillary variables considered in 2009 EVOOs. Values of ?13C for 2010 and 2011 samples show positive correlation with annual mean temperature, mean temperature of the warmest months, and the xerothermic index. In the same years, negative correlations were observed with annual mean precipitation, and mean precipitation of both the spring and summer quarters. We derived ?18O and ?13C annual isoscapes of Italian EVOOs by joining regression techniques and geostatistical interpolation (Kriging). A direct estimate of the model variance is also provided. The annual prediction maps of EVOO ?18O lar