Biocultural diversity of common walnut (Juglans regia L.) and sweet chestnut (Castanea sativa Mill.) across Eurasia

Abstract A biocultural diversity approach integrates plant biology and germplasm dispersal processes with human cultural diversity. An increasing number of studies have identified cultural factors and ethnolinguistic barriers as the main drivers of the genetic diversity in crop plants. Little is known about how anthropogenic processes have affected the evolution of tree crops over the entire time scale of their interaction with humans. In Asia and the Mediterranean, common walnut (Juglans regia L.) and sweet chestnut (Castanea sativa Mill.) have been economically and culturally important crops for millennia; there, in ancient times, they were invested with symbolic and religious significance. In this study, we detected a partial geographic congruence between the ethno‐linguistic repartition of human communities, the distribution of major cognitive sets of word‐related terms, and the inferred genetic clusters of common walnut and sweet chestnut populations across Eurasia. Our data indicated that isolation by distance processes, landscape heterogeneity and cultural boundaries might have promoted simultaneously human language diversification and walnut/chestnut differentiation across the same geographic macro‐regions. Hotspots of common walnut and sweet chestnut genetic diversity were associated with areas of linguistic enrichment in the Himalayas, Trans‐Caucasus, and Pyrenees Mountains, where common walnuts and sweet chestnuts had sustained ties to human culture since the Early Bronze Age. Our multidisciplinary approach supported the indirect and direct role of humans in shaping walnut and chestnut diversity across Eurasia from the EBA (e.g., Persian Empire and Greek–Roman colonization) until the first evidence of active selection and clonal propagation by grafting of both species. Our findings highlighted the benefit of an efficient integration of the relevant cultural factors in the classical genome (G) × environmental (E) model and the urgency of a systematic application of the biocultural diversity concept in the reconstruction of the evolutionary history of tree species.

shaping the genetic resources in crop species invested with esthetic, ethnobotanic and religious significance. Examples include cereals in the Yunnan Province of China (Xu et al., 2014), cassava in Gabon (Delêtre, McKey, & Hodkinson, 2011), sorghum in Kenya (Labeyrie et al., 2014), and maize in Mexico (Orozco-Ramírez, Ross-Ibarra, Santacruz-Varela, & Brush, 2016). These researchers investigated farming communities at local-national level, but did not provide a comparative and broader overview of the plant biocultural diversity.
A few pioneering studies shed light on the cultural forces affecting plant diversity and seed dispersal routes at a continental scale.
A strong geographic coincidence between ethnolinguistic boundaries, used as a proxy for human cultural differences, and the population genetic structure of plants has been detected for sorghum and pearl millet in Africa (Naino Jika et al., 2017;Westengen et al., 2014), banana and sweet potato in Oceania (Perrier et al., 2011;Roullier, Benoit, McKey, & Lebot, 2013), and baobab in Australia (Rangan et al., 2015). In these studies, the integration of ethnobotanical evidence with genetic and linguistic data (e.g., movement of word-re- Both sweet chestnut and common walnut are economically important, monoecious, dichogamous, long-lived, perennial trees cultivated worldwide for high-quality wood, edible nuts, and several secondary products. Pollination is described as anemophilous for both species but also entomophilous for chestnut. For millennia, both species have been endowed with symbolic and religious significance by societies in Asia and the Mediterranean which fully incorporated them into their cultures (Conedera, Krebs, Tinner, Pradella, & Torriani, 2004;Vahdati, 2014). In the last two decades, the history of common walnut and sweet chestnut has emerged as a complex interaction of biogeographic and human forces ( Figure S1).
Genetic and fossil evidence demonstrated that the actual distribution of common walnut in Eurasia resulted from the combined effects of expansion/contraction from multiple refugia, ranging from Central Asia, through the Caucasus to the Balkans and Western Europe, after the Last Glacial Maximum .
Despite this information, the onset of common walnut and sweet chestnut arboriculture is generally attested from 2,750-1,900 BP in Europe, coincident with the beginning of Greek and Roman dominance (Conedera et al., 2004). The Romans introduced both species across North-Central Europe (Fig. S1), although no clear evidence of systematic planting exists. Nevertheless, Pollegioni et al. (2015) revealed that humans harvested and traded walnut along "corridors" such as the Silk Roads and the Persian Royal Road during the same time-window, and what appeared to be native walnut stands were actually the result, at least in part, of ancient human efforts to modify the Asian landscape. The first evidence of active selection and clonal propagation by grafting was attested in Europe only from 15th to 18th centuries AD for sweet chestnut (Pereira-Lorenzo et al., 2019) and in the last century for common walnut (Dehghan, Vahdati, Rezaee, & Hassani, 2009). We may therefore assume that long-standing human contact and exclusive seed-mediated propagation through the centuries have affected the genetic structure of common walnut and sweet chestnut natural populations across Eurasia until at least the Medieval period.
In this study, the analysis of two large and unique collections of J. regia  and C. sativa (Mattioni et al., 2017) populations gave us the opportunity to provide a comprehensive and comparative view of the biocultural diversity of common walnut and sweet chestnut across Eurasia. Our objective was the integration of archeological, linguistic, and genetic data to address the role of landscape and ethno-linguistic boundaries, used as a proxy of cultural similarities between human communities, on limiting and/or facilitating the gene flow of walnut and chestnut K E Y W O R D S anthropogenic processes, common walnut, human linguistic diversity, population genetics, sweet chestnut germplasm across Eurasia. We also sought to provide insight into the indirect and/or direct human-mediated expansion of both species during historical eras such as the Aegean-Anatolian EBA, the Persian Empire (starting from the Achaemenid phase, 2,450-2,280 YBP), and Greek-Roman colonization (≤2,550 BP). In particular, we aimed to infer (a) geographic coincidences between genetic boundaries among tree populations and languages repartition of human communities, (b) spatial congruence between the genetic richness of common walnut and sweet chestnut populations and linguistic human diversity in terms of associated word-terms within language families of the native range, and (c) geographic overlaps between cognate sets of associated word-terms and tree population genetic clusters.

| Genetic database for common walnut and sweet chestnut populations
This study makes use of two recently published datasets for displaying the genetic diversity of J. regia and C. sativa in their respective native range. The first study included 40 Asian autochthonous common walnut populations sampled from China, Kyrgyzstan, Uzbekistan, Tajikistan, Pakistan, Iran, Turkey, and Georgia and 51 European walnut populations sampled from Greece, Romania, Moldova, Hungary, Slovakia, Spain, France, and Italy, growing in thirteen mountain systems for a total of 91 populations and 2,008 genotypes. The second dataset comprised 73 sweet chestnut populations for a total of 1,608 wild chestnut trees sampled in 11 European (Spain, Portugal, France, England, Italy, Slovakia, Hungary, Romania, Bulgaria, Greece, and Russia) and 3 Asian (Turkey, Georgia, and Azerbaijan) countries (Table S1). All the sampling sites refer to natural or naturalized areas, excluding orchard or recent forest plantations. Both collections were genotyped using unlinked nuclear, neutral microsatellite (SSR) markers (fourteen and six loci in common walnut and sweet chestnut, respectively; Mattioni et al., 2017;Pollegioni et al., 2017).
Eight geographic areas were classified as sites with bilingual speakers. Uyghur is classified as an Eastern-Turkic language currently spoken by 11 million people mainly living in the Xinjiang Province of North-Western China. The urban areas of Xinjiang have recently faced major changes in their demographic and linguistic landscapes. Since the bilingual education policy was introduced in 2002, Chinese-Mandarin language has been rapidly institutionalized (Smith Finley & Zang, 2015). The former multilingual pluralism of this region has been progressively replaced in favor of monolingual model.
Similarly, standard Tibetan, along with Mandarin Chinese, is the official language spoken in the Tibet Region of South-Western China (Gordon, 2005). Bakhmal is located in the Jizakh province of Central

| Reconstructing walnut and chestnut protowords, inherited terms, loanwords, and open compound words
Linguistic terms for common walnut (Table S2) and sweet chestnut (Table S3) were collected from published sources and from the Language of the World Etymological Database (LWED) (http:// starl ing.rinet.ru). If the etymological reconstruction was available, the proto-word for walnut and chestnut forms, conventionally denoted with an asterisk (*) at the start of the word, was reported in the Dené-Sino-Caucasian (Basque, Proto_Burushaski, Proto-North Caucasian, and Sino-Tibetan), and Afro-Asiatic and Eurasiatic (Dravidian, Kartvelian, Altaic, Indo-European, and Uralic) Super-Phylum. Despite the lack of written records, the etymological analysis of later attested words provided the opportunity to trace back through successive intermediate steps to the common ancestral word form in the ancestral reconstructed proto-language. We considered linguistic terms as "inherited" when they were inherited from a proto-language through nodes of phylogenic tree of more recent descendant languages following a vertical transmission. We classified linguistic terms as "loanwords" when words were borrowed from different language families and adopted in other languages by horizontal transmission. The identification of loanwords was further corroborated by archeological and historical data. We referred to a "cognate set" when the words have a common etymological origin sharing the same proto-word. However, the cognates progressively changed their form and sometimes meaning over the course of the time (called false friends), but in most cases they have similar sounds (Campbell, 2013). In this study, we included already established cognate sets for walnut and chestnut word forms as proposed by The Languages of the World Etymological Database, part of the Tower of Babel project (LWED). Finally, we included open compound forms denoting walnut or chestnut made up of two words written separately but providing a unique meaning.

| Genetic structure of tree populations and language diversity in human communities
To explore the genetic relationships among 91 common walnut populations and 73 sweet chestnut populations in the native range (d GEN ), we computed pairwise genetic differentiation with Jost's coefficient (Jost, 2008) using GenAlEx version 6.5 (Peakall & Smouse, 2012).
Linguistic distances among human communities living in the sampling areas were calculated as simple dissimilarity indexes ranging from 0 to 4 according to the d LAN matrix method described by Excoffier, Harding, Sokal, Pellegrini, and Sanchez-Mazas (1991) and Belle and Barbujani (2007 To estimate the genetic differentiation at both levels, among populations and among ethnolinguistic regions, we grouped common walnut and sweet chestnut genotypes according to their occurrence in the language phylum areas. Hierarchical analysis of molecular variance (AMOVA) was performed as implemented in Arlequin version 3.11 software (Excoffier, Laval, & Schneider, 2005), and statistical significance of Wright's F-statistic estimators was tested using a nonparametric approach with 1,000 permutations. Furthermore, the spatial congruence between the genetic relationships among J. regia or C. sativa populations and the linguistic-phylum patterns of human communities in the sampled sites was shown using a multivariate graph approach (EDENETWORKS v2.16, Kivela, Arnaud-Haond, & Samarki, 2015). We constructed a minimum-spanning tree plotting all populations (nodes) in a network graph with connections (edges) between all nodes. In the resulting graph, each edge was weighted according to its pairwise genetic distance and n populations were represented by n nodes with color equivalent to the language phylum spoken by human communities. Nodes were connected by the minimum number of edges necessary to minimize the overall genetic differentiation. Partial Mantel tests of the genetic differentiation among common walnut and sweet chestnut populations (d GEN ) versus human linguistic diversity (d LAN ) with geographic distance as a covariate (d GEO ) was used to test whether any statistical significance inferred by the AMOVA was a result of isolation by distance (IBD) (Smouse, Long, & Sokal, 1986). The p-value for the z-score of the Mantel association parameter was inferred using 5,000 permutations as implemented in ZT software (Bonnet & Van der Peer, 2002 (Guillot & Rousset, 2013), the influence of geographic distances and human linguistic diversity on d GEN was also evaluated with a multiple regression on distance matrices approach using MRM function implemented in the R-ecodist package (Goslee & Urban, 2007). The significance of regression coefficients and model r 2 was estimated using 5,000 permutations.

| Genetic richness of tree populations and word form-related diversity in human communities
The level of genetic diversity was estimated for each common walnut and sweet chestnut population by computing the allelic richness (Rs) parameter using the rarefaction method as implemented in the HP-Rare (Kalinowski, 2004).

| Genetic structure of tree populations and cognitive sets referred to walnut and chestnut terms
The geographic coincidences between genetic structure of walnut and chestnut populations and the distribution of the major cognitive sets referred to their respective word forms were examined. As reported in Pollegioni et al. (2017) and Mattioni et al. (2017), a fully Bayesian clustering approach as implemented in STRUCTURE 2.3.3 (Pritchard, Stephens, & Donnelly, 2000) was conducted to detect the most likely number of tree populations. After determining the most probable number of clusters (K), we derived two synthetic maps representing the genetic structure of common walnut and sweet chestnut in Eurasia. In addition, the absolute number of migrants exchanged between the inferred genetic clusters per generation (2Nm) was calculated using Arlequin version 3.11 software. The spatial distribution of walnut/ chestnut word-terms and their proto-words was compared to the inferred genetic clusters of J. regia and C. sativa populations across Eurasia. Chi-squared tests were conducted to compute statistical differences in the distributions of the major cognate sets that referred to walnut and chestnut linguistic terms among the inferred genetic clusters.

| Genetic differentiation of common walnut and sweet chestnut populations across language family areas
We observed a statistically significant positive trend between ge- Furthermore, the genetic diversity of both species showed a significant isolation by distance pattern (IBD) in Eurasia (Table 1). The pairwise linearized genetic differentiation values and the natural logarithm of geographic distances (straight-line distances in km) among sampling sites were in fact significantly correlated in common walnut (r = 0.737, p = .0002) and sweet chestnut (r = 0.569, p = .0002). Simple Mantel tests revealed that human linguistic diversity was also positively correlated with straight-line geographic distances among common walnut (r = 0.739, p = .0002) and sweet chestnut (r = 0.567, p = .0002) sampling sites. Thus, the observed relationship between d GEN and d LAN matrices might have occurred in J. regia (r = 0.636, p = .0002) and C. sativa (r = 0.546, p = .0002) as a result of a common spatial component (Table 1). However, the partial correlation between human linguistic distances and genetic diversity remained significant but low even after the effect of d GEO matrix was held constant in both species (walnut: r = 0.200, p = .0002; chestnut: r = 0.377, p = .0002). The MRM analysis indicated that the effects of geographic distance and human linguistic diversity on genetic tree divergence were significantly positive among common walnut populations (standardized partial regression coefficient: Table 1). The MRM model showed that geographic and language distance together explained 51.60% (p = .0002) and 39.7% (p = .0002) of common walnut and sweet chestnut genetic differentiation, respectively (Table 1).
The hierarchical AMOVA revealed that the majority of the molecular variance was partitioned within J. regia (79.77%) and C. sativa (76.65%) individuals, while 11.75% and 7.93% was due to differences among language phylum areas, respectively (p < .0001). The remaining molecular variance, 8.49% for common walnut and 15.42% for sweet chestnut, was distributed among populations within language phylum groups. A multivariate population graph displayed a nonrandom pattern of association between the language phyla and the genetic differentiation inferred in common walnut and sweet chestnut populations across Eurasia ( Figure 1). In particular, all tree TA B L E 1 Correlation between genetic distances among common walnut/chestnut populations (d GEN ) and human linguistic distances (d LAN )

Common walnut
Sweet chestnut

| Word-form richness and the genetic diversity of walnut and chestnut populations across Eurasia
We assigned a "word-form richness" to each sampling site by collecting more than 70 and 52 word-terms related to common walnut (Table S2) and sweet chestnut (Table S3), respectively, across Eurasia.
We evaluated the level of tree genetic diversity by computing the allelic richness (Rs) parameter for each common walnut and sweet chestnut population based on SSR markers. By comparing these datasets, we detected statistically significant differences between genetic allelic richness (Rs) within J. regia and C. sativa populations, and the associated number of the related inherited terms, loan- words, and open compound words (Kruskal-Wallis tests: p < .05).
The nonparametric post hoc Dunn's tests revealed that geographic sites with low Rs values were often associated with singletons (single word-terms) in both species ( Figure S3).
In J. regia, three small ancient cognate classes of the Dene-Sino-Caucasian super-phylum, *HwV́rƛ ̣ V, *ḳV̆rḳV, and *ṭVɫV, were scattered only in few common walnut populations (20%) of the inferred cluster 1 (Table 4) Figure 6). The basic meaning of the PIE form *derw-"tree," derived from the Proto-Euro-Asiatic F I G U R E 6 Geographic coincident between (a) the population genetic structure of chestnut inferred across Eurasia and (b) the distribution of the four major cognate sets (*derw-, *kastAno-, *blwt, *kVl) referred to chestnut. Inverse Distance Weighted (IDW) interpolations of the estimated mean population membership values (cluster analysis: Qi) were modified from Mattioni et al. (2017). Gray arrows indicated migration rates per generation. Word forms connected to proto-words but changing meaning over the time are reported as empty dots. The earliest date of appearance of the attested chestnut linguistic terms was provided for each cognitive set TA B L E 2 The six major cognate sets (*HwV́rƛ ̣ V, *ḳV̆rḳV, *ṭVɫV, *a-/an-gōza, *KVrV, *ŋuńV-) referred to walnut terms across Eurasia

| Genetic boundaries among tree populations and languages diversity in human communities
In this study, spatial analysis showed that the inferred barriers to gene flow among tree populations coincided with large differences in human language. Genetic differentiation computed between    There are numerous obvious differences between common walnut and sweet chestnut trees and annual crops (e.g., cereals or legumes) for which parallel demic and cultural expansion is assumed across Eurasia (Fort, 2012). Small seed size, long storage life, short life cycle, and easy cultivation contributed to the adoption of many annual crops. They provided immediate sources of food in large quantities on relatively small areas. These traits promoted the domestication and long-distance dispersal of annual crops throughout the continents from the Early Neolithic (Bar-Yosef, 2017). In sharp contrast, common walnut and sweet chestnut trees are very demanding in terms of soil quality and sensitive to temperature oscillations. Both species produce suborthodox and temperate-recalcitrant seeds, whose nursing is difficult (Bonner, 2008), but both species are relatively precocious, bearing seeds in less than nine years. Adult trees can remain productive for at least 40 years for common walnut and 100 years for sweet chestnut, and their nuts can survive up to two years under simple storage conditions (Bonner, 2008;Vahdati et al., 2019). Thus, short-and long-distance transportation of these seeds mediated by humans cannot be excluded from our scenario. As reported by Bar-Yosef (2017), during the second half of the Holocene people transported seeds, animals, and technologies across Eurasia within few months by rivers or sea and by land with support of draft animals (e.g., donkeys, horses, and camels). Considering all those elements and the wide use of common walnut and sweet chestnut as religious, spiritual, and ritual plants (e.g., seeds exchanged as inherited via gifts at weddings as symbol of fertility), food, and medicine, it is reasonable to evoke the homophily theory as a plausible explanation for part of the large-scale genetic patterns of these two species in Eurasia. Similar to small-scale farming systems (Pautasso et al., 2013), seed exchanges might have occurred between members of the same historical macro-groups as a result of their frequent and preferential interactions (Leclerc & Coppens d'Eeckenbrugge, 2012).
We cannot rule out that these types of seed transactions as a component of human cultural institutions, as well as typical economic transactions such as barter and the simple act of carrying familiar, convenient food on a journey, shaped the genetic structure and diversity of J. regia and C. sativa across Eurasia. Over time, humans and their crops coevolved in response to climate changes, socio-economic pressures, and the cultural significance of crops and seeds.
The importance of these human interactions in the spread of annual crops seems almost intuitive, but it is much less so for tree species that remained undomesticated, even if they were widely cultivated.

| Co-occurrence of word-term richness and genetic diversity of walnut and chestnut across Eurasia
Our analysis revealed higher diversity in terminology was associated with higher allelic richness (Rs) in samples from ( regia (Krebs et al., 2019). In the same macro-regions, marked concentration of localized idiosyncratic terms related to walnut and chestnut were detected, mainly reflecting the enormous phylogenetic diversity in languages recorded in such geographic areas (Gavin et al., 2013).
At the global scale, ecoregions of high biological diversity often coincide with hotspots of indigenous-linguistic diversity (Gorenflo et al., 2012). Studies in biogeography suggested that environmental and spatial heterogeneity as well as socio-cultural factors affected biological processes and language evolution/persistence in every continent (Amano et al., 2014). Vicariance appears to act similarly on human and plant populations, generating inter-and intrademe diversification. High levels of topographic complexity may have generated barriers to the movement of people, increased the potential for geographic isolation, and promoted the divergence of ethnolinguistic groups as a result of drift and adaptation to complex and fragmented environmental habits (Gavin et al., 2013). Similarly, we postulate that the heterogeneous environments of Himalayas, Trans-Caucasus and Pyrenees Mountains promoted common walnut and sweet chestnut diversity as well as the evolution and/or persistence over thousands of years of languages in different language families spoken by a small number of people in isolated enclaves. At least one additional historical process may contribute to local peaks in diversity; the Himalayas and Trans-Caucasia are classified as "accretion zones," where many language macro-families are present and the structural diversity of languages is high and increases over time through immigration (Nichols, 1997 (Lone, Khan, & Buth, 1993;Pokharia, Mani, Spate, Betts, & Srivastava, 2018), where two ancient Burushaski (khakhā́jo and tili), and three Indo-European (Kashmiri doon, Urdu akhrot, and Old Persian gawz, ≤3,600 BP) walnut word-terms were highlighted. There is evidence of the religious importance of common walnut in Kashmir from the 1st millennium BC onward; walnut became the essential element of several rituals in the Shaivism religion of Kashmir (Tikoo, 2013, ~2,800 BP) and pre-Buddist Bön religion in the Tibetan area of Eastern Himalayas (Weckerle, Huber, Yongping, & Weibang, 2005; ~2,515 BP).
Although the human consumption of common walnut and sweet chestnut seeds has been demonstrated in Aceramic (9,450 BP) and Ceramic (8,450 Kvavadze et al., 2015;Ananauri-3-kurgan site, 4,450 BP, Makharadze, 2015). All these studies suggested that common walnut and sweet chestnut cultivation, together with viticulture and pasturing, was fully incorporated in the Trans-Caucasian agricultural landscape of EBA. These early horticultural practices were substan-

| Tracing indirect and direct human-mediated dispersal routes of walnut and chestnut across Eurasia
In this study, we detected a partial spatial congruence between the distribution of language terms and their proto-words with the inferred genetic clusters of common walnut and sweet chestnut populations. The great variety of modern forms can be described in terms of six major cognate sets referred to walnut terms, *HwV́rƛ ̣ V, *ḳV̆rḳV, *ṭVɫV, *a-/an-gōza, *ŋuńV-, and *KVrV and four major cognate sets referred to chestnut terms *qẇăɫV́, *blwt, *dʷirV, and *kastAno-, with distinctive distributions for both species.
Large cognate class sizes with high congruence with genetic clusters were observed for walnut cognates *a-/an-gōza, *ŋuńV-, and *KVrV, and for chestnut cognates, *blwt, and *kastAno-. As formulated by Gamkrelidze and Ivanov (1995), the spread of the descendants of It is also true that we cannot exclude the adoption of *kar among cultures that had the most successful engagement in inter-regional exchanges of seeds across the Near East and the Aegean region from the 6th millennium BP onward. However, we did not find similar congruence with chestnut genetic data and any Indo-European word proto-form.  (Haldane, 1993;Ward, 2003 Even so, sweet chestnut was mainly consumed by the low social classes and was subsidiary to common walnut (Allevato, Saracino, Fici, & Di Pasquale, 2016). Similarly, the second Indo-European proto-form for walnut *khneu-, from the Euro-Asiatic proto-form *ŋuńV, progressively replaced the original derived terms of *kar in the Italic-Celtic-Germanic languages. According to Gamkrelidze and Ivanov (1995), *khneu-inherited words initially did not refer to J. regia through its non-native range in Central Europe but to the nut tree or hazelnut which was common in the Early Holocene.
In connection with Roman campaigns, *khneu-derived terms were translated as "walnut", attesting its key role in the Roman agro-forest management across Europe (Allevato et al., 2016;Bakels & Jacomet, 2003). In the German languages, the term "wal-

| CON CLUS ION
In this study, we detected a partial geographic congruence between ethno-linguistic repartition of human communities, the distribution of major cognitive sets, and the inferred genetic clusters of common walnut and sweet chestnut populations across Eurasia.
Our data indicated that IBD processes, landscape heterogeneity, and cultural boundaries might have promoted both human language diversification and walnut/chestnut differentiation across the same geographic macro-regions. In particular, we found three hotspots of common walnut and sweet chestnut genetic diversity associated to high linguistic-related form richness: Himalayas, Trans-Caucasus, and Pyrenees Mountains. In agreement with genetic and linguistic data, the archeological evidence suggested a long-term interaction between walnuts, chestnuts, and people liv- We are aware that our reconstruction of walnut and chestnut history will be affected by adjustments and revisions based on the availability of linguistic databases and theoretical developments in

CO N FLI C T O F I NTE R E S T S TATE M E NT
The authors declare no conflicts of interest.

ACK N OWLED G M ENTS
The authors warmly thank Marcello Cherubini, Irene Olimpieri, and Virginia Tortolano for their support for laboratory activities.
Mention of a trademark, proprietary product, or vendor does not constitute a guarantee or warranty of the product by the U.S. Dept.
of Agriculture and does not imply its approval to the exclusion of other products or vendors that also may be suitable.