Characterizing wine terroir using strontium isotope ratios: a review

ABSTRACT This paper presents a detailed review of the use of 87Sr/86Sr isotope systematics for wine provenance studies. The method is based on the principle that the Sr isotope ratio in wine reflects that of the labile fraction of the vineyard soil from which the wine is produced. The review encompasses 87Sr/86Sr data from wine samples published between 1993 and 2021 from terroirs in 22 different countries. The analytical procedures and techniques adopted by the different authors and the range of isotope ratios obtained in the different studies are discussed and evaluated. This study provides a bibliometric analysis of the 87Sr/86Sr isotope approach for wine authentication at different scales. Although limitations are evident when implemented at large (global) scales, we demonstrate that the 87Sr/86Sr isotope tracing technique remains a powerful and reliable tool for determining the geographical origin of wine when combined with detailed knowledge of the geological and soil characteristics of the substrata. For example, this combination of data allows the wines grown in the volcanic soils of Central and Southern Italy to be unambiguously fingerprinted. We present a detailed protocol for the application of the Sr isotope technique to wine authentication.


Introduction
In the last 25-30 years, augmented consumer interest in the origin of the foodstuffs and the multidisciplinarity of the research groups have contributed to an increasing number of agri-food traceability studies.Determining the geographical origin of a wine is of particular interest as it is one of the most important factors that contribute to its commercial value in the world market and to the recognition of a specific terroir.The term vitivinicultural 'terroir', according to the Organisation Internationale de la Vigne et du Vin (International Organization of Vine and Wine; OIV) definition, refers to an area in which collective knowledge of the interactions between the identifiable physical and biological environment and applied vitivinicultural practices develops, providing distinctive characteristics for the products originating from this area.It includes specific soil, topography, climate, landscape characteristics and biodiversity features (Resolution OIV/VITI 333/ 2010).During growth, plants absorb the so-called bioavailable elements from the soil.These inherited elements provide information about the geological substratum on which the plants grew and can transfer this information to the food and beverages (such as wine) that are produced from the plants [1][2][3][4].This soilwine connection creates 'chemical fingerprints' that can be used to trace their origin [5,6].
Radiogenic Sr isotopes ( 87 Sr/ 86 Sr systematics) are increasingly being used for traceability studies [7,8] because: i) Sr isotopes do not fractionate during the vegetative life of the plants [9], and ii) 87 Sr/ 86 Sr ratios present rather large variations owing to the wide range of Rb/Sr ratios in soils of different ages and mineral compositions [2,3,[10][11][12].
Strontium is a lithophile element that naturally occurs as Sr 2+ , which allows substitution for Ca 2+ in a variety of rock-forming minerals such as feldspars, phyllosilicates, gypsum, apatite and, more importantly, calcite and dolomite [13].Strontium has four naturally occurring stable isotopes, with natural abundances ranging from 0.55-0.58% for 84 Sr, from 9.75-9.99% for 86 Sr, from 6.94-7.14% for 87 Sr, and from 82.3-82.8% for 88 Sr [14].Among these, 87 Sr is the only stable isotope that is also radiogenic.The radioactive decay of 87 Rb (half-life 48.8 × 10 9 a) [15] to the radiogenic 87 Sr increases the natural relative abundance of 87 Sr. Higher initial concentrations of Rb and/or older ages of rocks therefore contribute to a higher abundance of 87 Sr and to an increase in the 87 Sr/ 86 Sr ratio.The 87 Sr/ 86 Sr ratio of every geological formation is thus a function of the age of the rock and its initial Rb/Sr ratio.Consequently, 87 Sr/ 86 Sr is widely used as a tracer of strontium in rocks and minerals [16], providing valuable information on both age and geochemical origin [3,17].The 87 Sr/ 86 Sr values of geochemical reservoirs and rocks generally vary between 0.702, the typical value for the depleted, low Rb/Sr mantle, and values as high as 0.943 in old continental crust rocks [18].
The application of the 87 Sr/ 86 Sr technique has proven to be a valuable tool for tracking the origin of different agricultural products [9] including wines [2,4,7,8,[10][11][12].Studies of wines from around the world have investigated the influence of winemaking procedures, different grape varieties, vintages, soils, and geological substratum on the 87 Sr/ 86 Sr ratios of the wines [7,8,22,26,27,34,35,37,45].While Almeida and Vasconcelos [22] showed that the total Sr concentration increases during the vinification process, all studies concluded that this does not affect the corresponding 87 Sr/ 86 Sr ratio, i.e. it is not modified from grape juices to wine (both red and white wine) in all cases considered [4,22,31,33].
This work presents a review of the published Sr isotope studies used to discriminate the geographical origin of wines from around the world.The effects of specific analytical procedures and techniques adopted by the various authors are discussed in order to compare the published results from a wide variety of studies.

Sampling strategies
In order to accurately identify the provenance of agri-food products, systematic sampling is an essential prerequisite.Among the available wine studies, Sr isotope analysis were carried out on four different matrices (Figure 2): bottled wine (including wine made from controlled microvinification, see below), must, and soil.Analyses of red wines represented the largest part of the dataset, totalling 77 % of the data set versus 23 % for white wines.Bottled wine samples (either purchased by the authors or made available by wineries) accounted for 93 % of the total data.Because the vinification process may alter the original 87 Sr/ 86 Sr of the grape juice, mainly through the addition of additives used to filter or attenuate the taste of wine [31], a limited number of authors (3 %) analyzed wine produced by microvinification [4,7,11,12] or must samples 4 %, [10,32,33].Microvinification (vinification carried out at small scale) is used for experimental wine studies where a limited amount of grape is pressed and fermented in small vats.The advantage of microvinification in Sr isotope studies, is that it avoids potential contamination resulting from industrial winemaking (e.g.white wine treatment with refinement agents such as bentonite) [7].Microvinification also limits the possible variations of the 87 Sr/ 86 Sr ratios due to the heterogeneities in the associated soils, since the grapes used for the microvinification are sampled in one single vineyard.Large wineries may use grapes from different vineyards that have different geological substrata.From a total of 842 wine 87 Sr/ 86 Sr analyses, only 377 include corresponding data on soil samples [4,7,8,[10][11][12]24,29,45].Most of the soil samples (283) correspond to the analysis of the labile fraction, whereas a much smaller number (94) also includes data for the corresponding bulk soil samples (Figure 2) [4,8,[10][11][12]33].The labile fraction of the soil refers to easily soluble components of the soil that are absorbed by the plant roots and are therefore considered representative of the solutions and nutrients that derive from the weathering and leaching of minerals in the soil [12,46,47].Consequently, the labile fraction represents the most likely reservoir in the soil that can provide easily soluble Sr to the plant [2,4,12].

Sample preparation
The different procedures followed by the authors for sample preparation (wines and soil samples) are summarized in the Suppl.Table A1.Briefly, biological and geological samples are first acid-digested, then Sr ions are separated and purified from the other ions by chromatographic methods (i.e. through ion exchange resins) in a clean laboratory before their 87 Sr/ 86 Sr ratios can be measured [48].

Mass spectrometry instrumentation
Measuring the 87 Sr/ 86 Sr ratios for a geological or biological sample with a targeted precision of 0.01 % requires the use of a high-precision mass spectrometer [2,20].Reaching this precision depends on the ionization capacity of the instrument used, as well as on the mass-dependent isotope fractionation that occurs over the course of the analysis.Different types of mass spectrometer instruments can be employed to measure Sr isotope ratios.Among the 843 wine and 377 soil samples, 510 were analyzed using a multi-collector inductively coupled plasma mass spectrometer (MC-ICP-MS), 432 on a thermal-ionization mass spectrometer (TIMS), 267 on a Quadrupole Mass Spectrometer (Q-ICP-MS) and 11 on a sector field multi-collector inductively coupled plasma mass spectrometer (SF-MC-ICP-MS) (Figure 2).The main difference between the instruments is how the sample is introduced into the instrument and how the sample is ionized.Other differences are related to detection limits, mass resolution, mass interferences, associated costs, ease of use, availability of a validated methodology and, most importantly, the accuracy of the data.The ICP-MS has found favour when the number of samples and presumed contrast in isotope ratios are large, however, the current precision and accuracy of ICP-MS analyses for isotope abundance ratios are inferior those of the TIMS and SF-ICP-MS [2].The precision and accuracy of the isotope measurements is usually evaluated by the repeated analysis of a certified international standard such as the NIST SRM 987 strontium carbonate (SrCO 3 ) with a certified 87 Sr/ 86 Sr value of 0.710245 ± 0.000011 [49].
Figure 3A compares the 87 Sr/ 86 Sr values reported for the NIST SRM-987 standard from the reviewed literature.Results show that while the accuracy for TIMS, MC-ICP-MS and SF-MC-ICP-MS varies from to 0.00001-0.0001%, that of Q-ICP-MS varies from 0.001-0.1 %.Recently, Cellier et al. [43] developed an 'in house' reference Champagne wine (ChRM) produced by blending 6 bottles (750 ml each) produced during the same year.This standard yields an average 87 Sr/ 86 Sr of 0.708146 ± 0.000036 (n = 36).

Results
3.1.Influence of the mass spectrometry instrumentation on the determination of 87 Sr/ 86 Sr of wine and soil samples Figure 3A shows that TIMS, MC-ICP-MS and SF-MC-ICP-MS instruments provide the highest precision and overall higher quality data for 87 Sr/ 86 Sr measurements compared to data obtained by Q-ICP-MS.Although TIMS provides a more stable ionization/ isotope fractionation platform, resulting in more precise Sr isotopes analysis than MC-ICP-MS, the higher ionization efficiency (almost 100 %), higher sensitivity and rapid sample throughput make the MC-ICP-MS an attractive alternative to the TIMS method [50].In contrast, Q-ICP-MS instruments are relatively low cost, robust and easy to use, but they have a lower resolution and accuracy compared to the TIMS and MC-ICP-MS instruments.The large variability of the Q-ICP-MS data could represent a severe obstacle when trying to differentiate small regional Sr isotope variations in wine isotope studies or to make comparisons between wine and soils.For example, the difference between 87 Sr/ 86 Sr in wine and corresponding soil samples obtained from different instruments (Figure 3B) suggests that the dispersion may be partly related to the accuracy of the analytical instrument employed, because higher deviations are observed for data obtained by Q-ICP-MS.Other factors contributing to the isotope variations in the wines and soils are discussed in the following sections.

External Sr isotope sources
A prerequisite before applying the 87 Sr/ 86 Sr approach to authenticate agri-food products such as wine is to understand how external Sr isotope sources (natural or anthropogenic) may affect the original strontium concentrations (and Sr isotope ratios) from the soil and bedrock to the wine [10,22,31,33].Common sources of external strontium include the use of organic and mineral fertilizers on the soil, inorganic pesticides on the plants, as well as environmental pollution [51].Fertilizers frequently contain calcium, and thus Sr, that will ultimately modify the 87 Sr/ 86 Sr of soils and plants, and corresponding wines [10].The  [49], reported as a grey line (0.710245 ± 0.000011).Results were separated based on the type of analytical instrument used.B) Difference between 87 Sr/ 86 Sr values for wine and corresponding soil samples from the reviewed literature according to the type of analytical instrument used [4,8,24,27,45].
average Sr content of the Upper Continental Crust (UCC) is at 320 ± 46 ppm [52].In soils the typical concentration of plant-available Sr ranges from 0.2-20 ppm [53].Hoogewerff et al. [54] modelled the 87 Sr/ 86 Sr of bioavailable Sr at the scale of European soils and calculated an average value of 0.7107 ± 0.0058 (ranging from 0.7038-0.7596).Old rocks (>100 Ma) generally yield Sr isotope ratios >0.710 whereas rocks formed more recently (<1-10 Ma) generally display values <0.704.These can be compared to the maximum Sr concentration of ∼1 % for Sr-rich superphosphate fertilizers [55].In parallel, Vitoria et al. [56] showed that fertilizers are characterized by a large range of 87 Sr/ 86 Sr, from 0.70335-0.71522(median of 0.70823).However, Aguzzoni et al. [57], using the examples of oat plants and apple trees, showed that fertilization might slightly alter the soil's 87 Sr/ 86 Sr ratio and subsequently that of the plants.The authors concluded that this does not alter the potential for Sr isotopes to geographically trace agricultural commodities, but recommend that young plants should not be used for traceability studies.
Petrini et al. [10] reported an 87 Sr/ 86 Sr value of 0.70769 for a single sample of organic fungicide, but this is not sufficient to verify whether their application may modify the final Sr isotope ratio of a wine.Further studies are needed as the Sr isotope ratios of fungicides are expected to vary, depending on the country or region of manufacture and as the produced shift in the isotope ratio will depend on the amount of additive that is used.
Discriminating natural and external sources of Sr in soils and wines is difficult, due to variations in the type of soil of each vineyard, their intrinsic heterogeneities, and in the agricultural and winemaking practices employed by each winery.The strontium isotope ratios of soils vary primarily as a function of geology but may also be affected by climatic factors such as precipitation [58].The addition of strontium dissolved in rainwater or atmospheric particles may modify the original isotope signature of the soil, and those of the plants and agricultural products [59].Studies of 87 Sr/ 86 Sr in the soilplant system, combined with Sr isotope analyses of sources in an ecosystem contribute to the local knowledge of wine terroirs.Strontium isotope ratios in rainwater, for example, have been reported for Portugal (0.708-0.711) [60], Spain (0.709-0.710) [61], France (0.708-0.711) [62]; and China (0.701-0.710) [63,64] showing values very similar to the isotope ranges found in wines.Given that the average concentrations of Sr in crustal rocks and soils are so much higher (370 and 240 ppm) in comparison to rainwater (<0.1 ppm) [65], the potential impact of rain on the Strontium isotope ratios of the wine [2] is considered to be minimal.
The general winemaking procedure itself includes the collection of the grapes, maceration, extraction, aging, and a final refinement phase.Kaya et al. [66] investigated the variations of 87 Sr/ 86 Sr in red wine during wood aging.A red wine of Castelão grape variety from Portugal (Palmela Appellation) was aged for 3 months with French oak staves in stainless steel vats (34,000 L).Results obtained from 32 wine samples (without pre-stabilization treatment), including three industrial scale replicates, showed that wood aging and time (0-90 days) did not alter the strontium isotope ratio.
During the final refinement phase, bentonite [(Na,Ca) 0,3 (Al,Mg)₂Si₄O₁₀(OH)₂•nH₂O; Montmorillonite] is widely used to clarify and stabilize insoluble matter suspended in the wine prior to bottling.The addition of bentonite during winemaking could increase the calcium, and concomitantly the strontium content of wines, and potentially altering their original 87 Sr/ 86 Sr ratios [21,41].The alkali and alkaline earth metal content of bentonites can vary; in particular, there are sodium bentonites and some richer in calcium.These latter calcium-rich bentonites could contribute strontium, given the association with calcium, to the treated wine.Usually only white wines are stabilized by bentonite treatment.Thus, the interpretation of strontium isotope ratios in white wines should be done with the potential effects of bentonites in mind.Bentonite and yeast samples were studied by Tescione et al. [31], showing 87 Sr/ 86 Sr values of 0.70891 and 0.70821, and Sr contents of 59.3 and 11.5 ppm, respectively.The authors concluded that 87 Sr/ 86 Sr remains constant during the vinification process, even after aging and refinement.This indicates the amount of bentonite added to wine is not sufficient to affect the isotopic signature of the final product even if this additive contains a considerable amount of Sr.Given that this conclusion is based on a single study, further investigations are needed to demonstrate whether the different types of bentonite refinement may or may not affect the 87 Sr/ 86 Sr of wines.

Bulk and labile fractions soil and vintage control
The chemistry of a soil is determined by rock weathering and the abundance of organic acids [58].During pedogenesis, Sr and other readily soluble nutrients are leached from disaggregated primary and secondary minerals by a mixture of water and organic acids to form the so-called labile fraction [12].The nutrients in this labile fraction are more bioavailable to (and easily absorbed by) the roots of the grapevines.The Sr isotope ratios of the labile fraction differs from those of the bulk soil or bulk rock because the Sr isotope ratio of the labile fraction is dominated by the more water-soluble components in the soil (e.g.carbonates vs silicates) [4,[10][11][12].Thus, the 87 Sr/ 86 Sr ratios of the labile fractions show greater concordance with those of the corresponding wines/musts (Figure 4) compared to the 87 Sr/ 86 Sr ratios of the bulk soil samples [8,[10][11][12].
Studies also demonstrated that 87 Sr/ 86 Sr values remain constant over multiple years.Braschi et al. [11] monitored the 87 Sr/ 86 Sr values of wine samples from four consecutive vintage years (2010, 2011, 2012 and 2013) and found that the 87 Sr/ 86 Sr values of the soil labile fraction and wine samples remained relatively constant and concordant during this time period.These results indicate that the 87 Sr/ 86 Sr values of wine are independent of the vine cycle and thus can be used as a reliable geochemical proxy to authenticate the geographical origin and terroir of any given wine.

South America
The data from South America mostly correspond to red wines from Argentina (Malbec, Cabernet Sauvignon and Syrah vines) from the regions of Mendoza, with 87 Sr/ 86 Sr values in the range of 0.70710-0.70730,San Juan, with values of 0.70810-0.70840,and Cordoba, with 87 Sr/ 86 Sr values in the range of 0.70920-0.70930[24].These regions are underlain by different geological formations: Ternary conglomerates and sandstones (Mendoza), Ternary clastic, alluvial and eolian deposits (San Juan) and Paleozoic magmatic rocks (Cordoba).The wines were obtained directly from producers who applied good manufacturing practices and traceability systems.Wines were produced using grapes from the same vineyards where soil samples were collected to ensure a link between the wines, grapes and vineyards.The 87 Sr/ 86 Sr ratios of the soil samples from each region yield values of 0.70720 (Mendoza), 0.70810 (San Juan) and 0.71040 (Cordoba), which correlate well with the ratios obtained for the wines.A single Argentina sparkling wine sample (no details given) showed an 87 Sr/ 86 Sr of 0.70912 [42].A red wine sample from Chile, coming from a vineyard located on volcanic deposits/basaltic soil (no detailed information available) was analyzed by Barbaste et al. [20], yielding a value of 0.70471.Additionally, Cellier [42] measured a single sparkling wine sample from Chile, showing an 87 Sr/ 86 Sr of 0.70605.Brazilian sparkling wines are more radiogenic (i.e. higher 87 Sr/ 86 Sr ratios) than Argentinian and Chilean ones, with values ranging between 0.71463 and 0.71782 [42].

North America
Two detailed studies were carried out in Quebec (Canada) on microvinified red wines and soil samples [4,12], aiming at isotopically characterizing vineyards located in two different geologic areas: the St. Lawrence Platform and the Appalachian province.While wines from the St. Lawrence Platform vineyards cover an 87 Sr/ 86 Sr range of 0.70988-0.71138,those from the Appalachian province range from 0.71108-0.71546.Both articles also included a systematic study of the soilplant system.For the St. Lawrence Platform, 87 Sr/ 86 Sr values of 0.71211 and 0.70986 were obtained for soil samples developed on bedrock consisting of grey shale, sandstone, siltstone, and limestone deposits, 0.71203, 0.71538 and 0.71036 for those having a substratum of mudrock, slate, dolostone and sandstone deposits, and 0.71012 for soils from limestone, shale, dolostone and sandstone deposits.On the other hand, soil samples from the Appalachian province yielded the following 87 Sr/ 86 Sr values: 0.71335 (mudrock, sandstone, conglomerate, and limestone deposits in the substratum), 0.71174 and 0.71363 (phyllite, schist, slate, sandstone, quartzite, dolostone and conglomerate deposits), 0.71012 (limestone, shale, dolostone and sandstone deposits), and 0.71546 (mudrock, conglomerate, sandstone, and limestone deposits).This demonstrated a strong correlation between wine and soil samples, in terms of both isotope ranges and variability of the datasets.To our knowledge, a single red wine 87 Sr/ 86 Sr datum is available for the United States, referring to a Californian vineyard [20].The authors report an 87 Sr/ 86 Sr of 0.70688 for a wine from a vineyard rooted on a basaltic soil.Two additional Californian sparkling wines were reported by Cellier, [42] showing Sr isotope ratios ranging from 0.70657-0.70687.

Asia
Epova et al. [39,67] studied bottled wine samples purchased from Chinese wine stores.Among them, some red wine samples were identified as a certified Chinese production ( 87 Sr/ 86 Sr ranging from 0.70880-0.71036),while the others were labelled as 'Bordeaux' (i.e. from France) and had no information on the region of production.These samples gave 87 Sr/ 86 Sr ratios varying from 0.70523-0.71266.A single Chinese sparkling wine was analyzed by Cellier, [42], yielding an 87 Sr/ 86 Sr of 0.71152, while an Indian sparkling wine sample analyzed by the same author yielded a value of 0.70956.

Oceania
The available 87 Sr/ 86 Sr data for wines from Australia (no information available about the region of production) consist of one red wine sample with an 87 Sr/ 86 Sr value of 0.70963 [20] and eight sparkling wines ranging between 0.70688 and 0.71549 [68].While these values spread over a wide range, the author did not give any detail on the geology of the vineyards these sparkling wines originated from.A single sparkling wine from New Zealand was measured by Cellier [42] showing an 87 Sr/ 86 Sr of 0.70782.

Africa
Coupled winesoil studies were carried out by Vorster et al. [45] in four different South African vineyards (red and white varieties).The 87 Sr/ 86 Sr measured in wines from the Robertson region ranges from 0.71130-0.71540,higher than those from the Stellenbosch region (0.70700-0.71100)and moderately overlapping with those from the Swartland (0.70750-0.71410)and Walker Bay (0.70780-0.71310)regions.However, a poor correlation was observed between the Sr isotope ratio of the wines and those of the bulk soil samples of the corresponding vineyards: 0.71280-0.71550(Robertson), 0.70820-0.71590(Stellenbosch), 0.70480-0.71500(Swartland) and 0.70920-0.71380(Walker Bay).The lack of soilwine correlation may be due to using the Sr isotope ratio of the bulk soil fraction and/or the high errors associated with 87 Sr/ 86 Sr analysis obtained from the Q-ICP-MS instrument employed in the study (Figure 4b).Studies conducted on South African sparkling wines yield 87 Sr/ 86 Sr values ranging from 0.71046-0.71315[42].

Europe
The majority of the available 87 Sr/ 86 Sr data for European wines correspond to Italian wines, totalling 471 analyses from 6 different regions.The distribution of the data between the Italian regions is as follows (from north to south): 31 for Veneto (6 % of the total), 246 for Emilia Romagna (52 %), 122 for Tuscany (26 %), 42 for Latium (10 %), 17 for Campania (4 %) and 13 for Basilicata (2 %).More than a half of these (263) are coupled with data from the corresponding soils: 30 for Veneto, 154 for Emilia Romagna, 78 for Tuscany and 1 for Campania.Data for each of these regions of production are summarized below.
In the Veneto region, Petrini et al. [10] investigated the applicability of the Sr isotope approach to the authentication of the geographic origin of Prosecco wine (white variety).The must samples have an 87 Sr/ 86 Sr range of 0.70706-0.71266.The corresponding labile soil samples, developed on a sedimentary substratum of Pleistocene -Holocene age, cover a more restricted range from 0.70772-0.71097.Additionally, a single 87 Sr/ 86 Sr value of 0.70872 is available for a Merlot di Piave (red variety) vineyard in the same region [2].
The available 87 Sr/ 86 Sr data for the Emilia Romagna region are mostly for wines from Lambrusco di Sorbara, Lambrusco di Mantova, Lambrusco di Grasparossa and Lambrusco di Salamino vineyards (red variety).Their 87 Sr/ 86 Sr values range between 0.70823 and 0.71179 [44].Lambrusco vineyards are rooted on soils developed on Middle Cenozoic terrigenous deposits, whose 87 Sr/ 86 Sr are in the range of 0.70773 and 0.71209 [30,69].This again demonstrates a fairly good winesoil correlation despite the intrinsic heterogeneity in the 87 Sr/ 86 Sr ratios of the pedological (and geological) substratum.Indeed, this includes a large variety of sediments of different compositions, grain sizes and origins, as well as older granite debris responsible for the most radiogenic Sr isotope ratios [2].Valpolicella red wine samples were also analyzed in the Emilia Romagna region by Horn et al. [2], yielding 87 Sr/ 86 Sr values between 0.70889 and 0.70900.
The 87 Sr/ 86 Sr data for red wines from the Tuscany region include Chianti, Sangiovese and Ansonaco vineyards, reported in [11,26,38].
These studies analyzed wine samples and compared their 87 Sr/ 86 Sr values with those of the corresponding soil labile fraction in order to evaluate the potential impact of additives on Sr isotope ratios.Chianti wines displayed a range of 87 Sr/ 86 Sr between 0.70813 and 0.71068 [11,26,38], coming from grapes grown on soils that originate from quartzofeldspathic, limestone and marlstone deposits.The range of 87 Sr/ 86 Sr in these soils is between 0.70804 and 0.70903 (labile fraction), well within the isotope range of the corresponding wine samples.Additional Sr isotope data are available for white and red wines from the Giglio Island (Ansonaco and Sangiovese wines), with 87 Sr/ 86 Sr ranging from 0.70909-0.71209.Giglio Island has a geological substratum consisting of young quartzofeldspathic igneous rocks (granite) with an 87 Sr/ 86 Sr range between 0.71322 and 0.72071 [26].The authors suggested that the lack of correlation between Sr isotope values in the corresponding wine and soil samples from this small vineyard is due to the use of additives during the production of the white wine.
In the Latium region, analyzed wine samples correspond to the Frascati, Cesanese di Affile, Cesanese di Piglio and Cesanense di Olevano Romano vineyards, with an overall 87 Sr/ 86 Sr range of 0.70897-0.71058[8,26].Most of the analyses are derived from the Cesanese vineyards located on a substratum consisting of pyroclastic rocks of the Colli Albani volcanic field ( 87 Sr/ 86 Sr = 0.70951-0.71067)[23,70].On the other hand, the vineyards labelled as 'hill' (in flat areas) by the authors are characterized by a substratum consisting of Mesozoic to Tertiary sedimentary rocks such as limestone, marlstone and sandstone ( 87 Sr/ 86 Sr = 0.70885-0.71785).
Tescione et al. [31,32] analyzed white wines (Greco, Ciliegiolo, Merlot, Sauvignon and Trebbiano) from vineyards located on the volcanic units of the Vulsini Volcanic District.The Sr isotope ratio of wines ranged from 0.70883-0.71021[31,33].In a separate publication, the authors characterized the bulk soil fraction ( 87 Sr/ 86 Sr from 0.70937-0.71026)and its labile fraction ( 87 Sr/ 86 Sr from 0.70888-0.71009),and noted a better Sr isotope correlation between the wine and the labile fraction of the soil [32].
Marchionni et al. [8,26] showed there was good concordance between 87 Sr/ 86 Sr in wines and their geological substrata, especially for wines rooted on volcanic soils ( 87 Sr/ 86 Sr from 0.70994-0.71138).The authors focused on two vineyards, measuring the 87 Sr/ 86 Sr ratios in wine, must, rock and soil samples.Soil samples were collected at various depths and analyzed after sequential leaching, in order to assess the Sr isotope ratios of the bioavailable fractions.The wines yielded lower 87 Sr/ 86 Sr ratios than the bulk soil/bedrock, but were similar to the leached soil solution.Moreover, the latter was shown to progressively approach the isotope ratios of must/wine samples with decreasing depth.Frascati white wine samples were studied by Horn et al. [2], who reported 87 Sr/ 86 Sr values between 0.70835 and 0.70943.
In the Campania region, the Aglianico Campano and Piedirosso vineyards are underlain by rocks from the Roccamonfina, Somma -Vesuvius and Campi Flegrei volcanic districts, as well as by sedimentary rocks of the Campanian plain (including the Benevento, Irpinia and Lettere areas).The Aglianico Campano wines yield 87 Sr/ 86 Sr values between 0.70822 and 0.70865 and the Piedirosso wines range from 0.70772-0.70839[7,26].A single 87 Sr/ 86 Sr value of 0.70884 is reported from the Campanian plain that consists of Tertiary terrigenous arenite, clay deposits with subordinate evaporite and limestone layers [26].Volcanic rocks from the aforementioned districts display a range of 87 Sr/ 86 Sr between 0.70650 and 0.70860 [23,70] and references therein, consistent with the values measured in wine samples.Mercurio et al. [7] conducted a detailed study coupling the analysis of Piedirosso wine samples, soil samples collected at various depths, as well as samples from other biological matrices (grapes, leaves, branches) in the volcanic area of Campi Flegrei.The authors showed an overall agreement between all the biological samples (0.70776-0.70844,including the wine samples), and the soil samples from the different horizons (0.70788-0.70813).Also in this case, Sr isotope ratios in the soil perfectly match the range of the Campi Flegrei volcanic rocks [7].
Vineyards investigated in the Basilicata region are rooted on pyroclastic deposits from the Mount Vulture volcano that have 87 Sr/ 86 Sr values between 0.70522 and 0.70705 [23,70].Studies conducted in the Aglianico vineyards (red variety) yield 87 Sr/ 86 Sr values of 0.70679-0.70817[26]i.e.partially overlapping those of the corresponding geological substrata but yielding slightly more radiogenic values.While the authors did not provide any explanation for that partial discrepancy, the fact that some of the sampled vineyards are rooted on a sedimentary substratum could explain the more radiogenic 87 Sr/ 86 Sr values.
A total of five Portuguese red wine regions have been characterized using Sr isotopes [20][21][22]27,41]: Madeira (0.70660), Dao (0.73700), Obitos (0.71400), Douro Valley (0.71400-0.71600)and Palmela (0.71100).Vineyards located on Madeira Island have a substratum consisting of basaltic rocks, which explains their low 87 Sr/ 86 Sr ratios [20], whereas the highest values for wines from Dao correspond to a substratum consisting of granitic rocks [27].Bulk soils were also analyzed in the Douro Valley, showing a range of 0.71360 and 0.71800 [41], that is slightly higher compared to the corresponding wines.Sampling of the 'provenance soil', rather than the labile fraction, can explain the discrepancy between the 87 Sr/ 86 Sr values of the soil and wine samples.Horn et al. [2] studied red wines from the Amselfeld region of Kosovo (southeastern Europe), which yielded 87 Sr/ 86 Sr values from 0.70981-0.71033.Wines from Romania yielded highly variable 87 Sr/ 86 Sr values of 0.70230 (Murfatlar area) 0.71700 (Dealu Bujorului area) and 0.76810 (Ștefănești-Argeș area; [37]).This variability may reflect lower accuracy related to the use of a Q-ICP-MS instrument.For example, Dehelean & Voica [25] studied bottled wines from Romania using Q-ICP-MS and reported 87 Sr/ 86 Sr ratios between 0.7500 and 2.0000.These values clearly exceed the natural range of Sr isotope ratios reported in geological matrices and are not discussed any further.

Discussion
4.1.Discriminating wine terroirs by their 87 Sr/ 86 Sr Figures 5 and 6 show the variations in the 87 Sr/ 86 Sr ratios of wines from around the world.Results show that 87 Sr/ 86 Sr discriminates North American, Canadian and South American wines.For example, Californian vineyards are located over less radiogenic volcanic terrains than the Canadian sedimentary vineyards.Likewise, in South America, Brazilian wines are more radiogenic than Argentinian and Chilean ones due to their older geological substrata.Countries from the other continents present overlapping 87 Sr/ 86 Sr values.Barbaste et al. [20] compared 87 Sr/ 86 Sr ratios of wines from vineyards rooted on volcanic (Chile, Madeira-Portugal, California-U.S.A.), granitic (Douro-Portugal) and sedimentary bedrock (St.Emillion-France).Wines from volcanic (basaltic) rocks show lower 87 Sr/ 86 Sr values compared to wines from granite vineyards in Portugal.French wines, coming from sedimentary vineyards, showed intermediate 87 Sr/ 86 Sr values.The data highlight that there is a large overlap of 87 Sr/ 86 Sr values for wine samples from different countries, meaning that Sr isotope ratios cannot fingerprint provenance areas on a global scale.For example, the 87 Sr/ 86 Sr range measured in Bordeaux wines (France) overlaps with that for wines from U.S.A., Italy, China, and Argentina (Figure 5).Comparing 87 Sr/ 86 Sr ratios based on the lithology on which the vine grew instead of the producing country may improve isotope fingerprinting, but unfortunately most studies do not address the geology in detail.Still, Figure 7 compares the 87 Sr/ 86 Sr ratios of wines rooted on volcanic rocks (Figure 7).The low Rb/Sr ratio of the mantle source for most mantle-derived volcanic rocks (basalts, andesites) results in low 87 Sr/ 86 Sr ratios.Figure 7 shows that volcanic wines from Chile have low 87 Sr/ 86 Sr and can be thus distinguished from those of California (U.S.A), Madeira Island (Portugal) and Italian wines.On the other hand, American and Portuguese wines overlap the   [8,26], Mercurio et al., [7], Braschi et al., [11], Tescione et al., [31,33].Wines from Italy are listed from north to south (LT: Latium; CP: Campania; BS: Basilicata).
Italian wines, underlining that the 87 Sr/ 86 Sr ratios of the wine will reflect the underlying volcanic substrata.This comparison also confirms that the 87 Sr/ 86 Sr approach to distinguish wine-producing areas has limitations on a large scale.
However, the Italian wine samples exhibit a rather narrow range of 87 Sr/ 86 Sr values that generally decrease from North to South [7,8,11,26].This trend mimics the Sr isotope trend observed in rocks [70] and further supports the conclusion that Sr isotope ratios in wine and underlying rocks are correlated.On such basis, it seems reasonable to conclude that wines from the Italian volcanic vineyards can be efficiently discriminated at a small to medium scale of investigation, thus possibly representing a useful tool for detecting fraud in wine appellation.
In Figure 8 we plotted the difference between the Sr isotope ratios of the bulk and labile soil fractions (bulk fraction minus labile fraction) against the ratios obtained for the bulk soil fraction.The plot reveals two groups of points.The larger group, forming a positive correlation, consists of soil samples from Canada [4] and Italy [10] that are underlain by sediments or sedimentary rocks.Quartz-feldspar-rich soils with high Rb contents will produce more radiogenic strontium isotope ratios at the upper right of the diagram, and thus a larger bulk minus labile value.In contrast, carbonate-rich (high Sr) soils yield lower strontium isotope ratios at the bottom left of the diagram and smaller  bulk minus labile values.Soils that yield negative bulk minus labile values occur when the strontium isotope ratios of the bulk soil fraction are less radiogenic than the labile fraction.This likely reflects the dissolution of secondary carbonates (with higher 87 Sr/ 86 Sr ratios) in the soil.A second grouping of points is defined by bulk and labile soil fractions from volcanic rocks from southern Tuscany in Italy [31].This grouping suggests the possibility that volcanic soils (and wines) produce a different trend due to the generally lower Sr isotope ratios in these soils compared to most sedimentary soils.The relatively low bulk minus labile soil fraction values reflect the Sr-rich nature of the soils developed over maficintermediate volcanic rocks.Further studies are needed to investigate the possibility of using diagrams such as Figure 8 to discriminate between sedimentary soils (wines) and volcanic soils (wines).

Conclusions
This study reviewed published 87 Sr/ 86 Sr datasets to investigate the practicality of the radiogenic Sr isotope technique for wine authentication.Parameters that were considered include those that may influence the 87 Sr/ 86 Sr ratio of wine such as geology, soil, wines obtained from white and red grape varieties and with different vinification technologies, vintage, winemaking, and agricultural practices.These studies suggest that the 87 Sr/ 86 Sr ratio of a wine is strongly linked to the geological heterogeneities of the terroir and is not seriously impacted/modified by the winemaking process [8,10,30,[32][33][34], vintage [7,10,25,42,63] and grape variety [11,36].Further studies are needed to evaluate how the extensive use of fungicides/fertilizers may affect the 87 Sr/ 86 Sr ratio of a wines and soils.Results show that the Sr isotope approach can be a reliable tool to characterize the region of production of a wine at a local scale (i.e.regional, national) but is more limited when trying to implement it at a larger scale.Nevertheless, we demonstrated that the 87 Sr/ 86 Sr isotope technique is a reliable tool for wine authentication and could be adopted by international wine regulatory organizations as a new traceability application.This authentication approach requires a very scrupulous approach to sample collection, instrumentation capable of yielding precise analysis (TIMS, MC-ICP-MS or SF-MC-ICP-MS) and a strong understanding of the geological substrata underlying the studied vineyards.This includes knowledge of the various viticultural and winemaking procedures because the use of fungicides or fining agents may alter the Sr isotope ratios of the soils and wines (e.g.[8,10,32]).This review identifies the best practices in Sr isotope investigations of the provenance of wines, which is summarized in a detailed protocol, reported in Figure 9, and that we are proposing as a reference standard procedure to be followed in future studies.This will not only produce more robust and reliable results, but will also allow meaningful comparisons between different studies, leading to a better understanding of the role of the numerous variables that are involved.

Figure 1 .
Figure 1.World map showing the distribution of available 87 Sr/ 86 Sr data for wine samples.

Figure 3 .
Figure 3. A)87 Sr/ 86 Sr values for the SRM-987 standard measured in the reviewed literature papers, compared with the certified value of Thirlwall[49], reported as a grey line (0.710245 ± 0.000011).Results were separated based on the type of analytical instrument used.B) Difference between 87 Sr/ 86 Sr values for wine and corresponding soil samples from the reviewed literature according to the type of analytical instrument used[4,8,24,27,45].

Figure 4 .
Figure 4. 87 Sr/ 86 Sr values in wines, and corresponding labile fraction and bulk soil samples collected in Canada (n = 9) and Italy (n = 11).Note that the bulk soil samples are always more radiogenic, and that the labile soil fraction samples show a better isotope agreement with the wine samples.

Figure 5 .
Figure 5. Overview of global 87 Sr/ 86 Sr values for a) wine, b) soil labile fraction and c) bulk soil samples based on the country of provenance, showing the number of analysed samples (on the right).Data sources provided in the Electronic Appendix 2.

Figure 6 .
Figure 6.Box -Whisker-plot.Mean value and standard deviation of the 87 Sr/ 86 Sr wine results in the world.Data sources provided in the Electronic Appendix 2.

Figure 8 .
Figure 8.A plot of the bulk soil 87 Sr/ 86 Sr ratios verus the difference between the bulk soil 87 Sr/ 86 Sr minus labile soil 87 Sr/ 86 Sr fractions ratios.In the diagram, soils underlain by sediments or sedimentary rocks form the larger correlation, while soils underlain by volcanic rocks form the smaller grouping.See text for details.Data for[4,10,31].

Figure 9 .
Figure 9. Proposed protocol for implementing the Sr isotope provenance technique to wine certification.