The published literature on C. sativa VO composition highlights marked regional chemical disparities. However, there remains a notable scarcity of studies in the Americas and Africa, and no records of such research in Oceania. This limited geographical representation can be attributed to a combination of factors. Legal restrictions on C. sativa cultivation and use in many countries continue to pose substantial barriers to research, and regulatory frameworks often differ widely. In several nations, permission to conduct C. sativa research has only recently been granted and is typically subject to stringent conditions. In regions where C. sativa has not historically been incorporated into agricultural systems, such as large parts of Africa and Oceania, there is also a lack of infrastructure, technical capacity, and dedicated funding for studies on this crop. Furthermore, local research agendas often prioritize long-established food or cash crops, which further limits the scope of investigations on C. sativa, particularly in countries where its use remains socially or politically controversial.

In contrast, Asia, and Europe with Italy as a prominent example, stand out as the most represented regions in studies of C. sativa VOs (Fig. 3; reference codes listed in Supplementary Table SS2). The higher number of scientific publications from these regions can be ascribed to more consolidated research frameworks, greater industrial interest in hemp- and Cannabis-derived products, and, in some cases, more permissive legal environments that actively encourage the systematic investigation of the plant’s chemical properties. These factors together create favorable conditions for sustained research, particularly in Europe, where countries such as Italy have emerged as key contributors to the field.

The VOs from cultivated Cannabis in Italy were classified according to the concentration of their major constituents (content > 5%; Supplementary Table SS1). The spatial distribution of chemotypes across Italy (Fig. 5), combined with the global variability in terpenoid classes (Fig. 2) and the relative contribution of monoterpenes and sesquiterpenes in each Italian locality (Fig. 6), shows that monoterpenoids and sesquiterpenoids constitute the dominant components of C. sativa VOs. A pronounced qualitative and quantitative contribution of non-oxygenated sesquiterpenes, particularly β-caryophyllene and α-humulene, was observed, in agreement with previous reports that identify these compounds as chemotaxonomic markers of the species (Jin et al. 2020; Cerrato et al. 2021). Localities such as Rovigo, San Severino Marche, and Ascoli Piceno exhibit unusually high levels of oxygenated sesquiterpenes for C. sativa cultivars, with total contents ranging from approximately 5–60% (Fig. 6). By contrast, VOs obtained from Pakistani Cannabis accessions show profiles enriched in oxygenated monoterpenes (Fig. 6), underscoring the marked phytochemical differentiation between European and Asian germplasm.

Across the Italian dataset, the most abundant individual compound was the bicyclic sesquiterpene β-caryophyllene, which predominated in 33 of the 39 samples analyzed, with concentrations ranging from 4.9% to 50.0%. β-Caryophyllene is widely used as a flavoring agent and has recognized anti-inflammatory, analgesic and antipyretic properties (Cerrato et al. 2021). Other major constituents include the aliphatic monoterpene β-myrcene, the bicyclic monoterpene α-pinene, the sesquiterpene α-humulene, caryophyllene oxide, and terpinolene. α-Humulene (8.7–9.1%) is also notable for its anti-inflammatory activity (Olejar and Park 2022). β-Myrcene (0.2–49.3%), biosynthesized by several plant species, particularly hops (Humulus lupulus L.), is widely employed in synthetic fragrances, pharmaceutical preparations, personal care products, and as a flavoring agent in foods (Adamek et al. 2022). α-Pinene (0.2–29.2%), detected in nearly all VOs, imparts a characteristic pine-resin odor and is used as a flavoring, solvent, and additive in lubricating oils (Adamek et al. 2022).

Profile analysis of VOs has been proposed as a complementary tool to distinguish dioecious from monoecious C. sativa genotypes, and as a chemical marker system to identify fiber-type varieties (Cerrato et al. 2021). Nevertheless, there is still a substantial gap in the literature regarding chemotypes associated with high levels of Δ⁹-THC and cannabigerol (CBG), as well as those enriched in less prevalent cannabinoids. This gap likely reflects regulatory restrictions on cultivars with Δ⁹-THC contents above 0.2%, coupled with the scarcity of studies focused on minor cannabinoids such as tetrahydrocannabivarin (THCV), cannabidivarin (CBDV) and cannabichromene (CBC). Most available work concentrates on genotypes that are legally approved in Europe, including dioecious varieties (e.g., Fibrante, Fibranova, and Carmagnola) and monoecious cultivars (e.g., Carmaleonte and Codimono), as well as hybrids derived from crosses between these lineages, such as 24 K, and Gorilla Glue (Table 1; C. sativa VOs, time-trend analysis).

Recent investigations demonstrate that post-harvest handling and processing exert a major influence on the chemical composition of VOs from C. sativa and C. sativa subsp. indica. Although extraction protocols can affect the total yield of volatiles, the qualitative profile (i.e., which compounds are present) appears to be more sensitive to pre-harvest factors such as genotype, pedoclimatic conditions and harvest timing (Moreira et al. 2021). The drying of inflorescences, for example, has been associated with VOs enriched in sesquiterpenes (Kant et al. 2015). Comparative analyses of VOs from different geographical regions reveal pronounced variations in total sesquiterpene content: samples from Amstetten (72.8%) and Burgenland (32.5%) in Austria, Zürich (69.7%), and Tenniken (60.7%) in Switzerland, Slovenia (44.7%), the Netherlands (56.5%), and several Italian regions – Pisa (95.7%), Fiuminata (66.1%), Po Valley (64.4%), Pavia (40.1%), and Lucca (58.5%) – exemplify this variability. While sesquiterpene levels vary considerably between regions, extraction methods themselves have a comparatively modest impact on qualitative composition, indicating that growing conditions and chemotype exert stronger control over VO profiles (Torkamaneh and Jones 2022).

Drying is a critical step that affects both the quality and quantity of volatile constituents. Beyond their medicinal value, volatiles recovered from C. sativa biomass and residues have potential applications in pharmaceutical, food, cosmetic, and agrochemical industries, reinforcing the need to re-evaluate industrial hemp as an environmentally sustainable crop (Adamek et al. 2022). During drying, water loss and associated changes in cell integrity can alter the localization and stability of volatiles, resulting in compositional shifts (Campbell et al. 2019). For instance, in Mentha aquatica L., drying at 40 °C for 48 h increased menthol and menthone contents, whereas drying at 60 °C for the same duration led to their depletion (Adhikary et al. 2021). In general, higher drying temperatures tend to enhance the relative abundance of sesquiterpenes, although the magnitude and direction of these effects are compound- and species-dependent (Campbell et al. 2019).

Different drying techniques, including natural drying in sun or shade, hot-air oven drying, vacuum drying, microwave drying, and freeze-drying, transfer energy at distinct rates and over varying exposure times, thereby triggering different, often irreversible, chemical, and biological reactions. These processes also induce structural, physical, and mechanical changes in plant tissues. Consequently, it is essential to evaluate how each drying method affects final VOs quality (Thorpe 2006). In addition to intrinsic factors such as plant variety, pedoclimatic conditions and harvest stage, extrinsic factors including light exposure, oxygen availability, ambient humidity, and storage temperature can profoundly influence compound stability (Chandra et al. 2020). The deliberate choice of monoecious or dioecious varieties, together with careful control of field humidity and harvesting schedules, has been shown to improve VOs yield and concentration, increasing trichome density and terpene production (Cerrato et al. 2021).

In the context of post-harvest processing, extraction by solid-phase microextraction (SPME) has been reported to increase the monoterpene fraction from 51.7% to 95.7% in certain C. sativa samples (Jin et al. 2020; Nemati et al. 2021; Danziger and Bernstein 2021a). Grinding and drying steps also significantly affect VOs composition (Ramos 2019). Zheljazkov et al. (Zheljazkov et al. 2020) showed that grinding the plant material prior to distillation more than doubled the monoterpene content, with most of these compounds being recovered in the early stages of extraction. Conversely, the best conditions for preserving sesquiterpenes such as β-caryophyllene, α-humulene, α-bisabolol, and E-α-bergamotene were obtained when the material was not ground and distillation continued for 80 min, yielding an approximately 80% increase in total sesquiterpenes. This protocol is therefore particularly suitable for maximizing the utilisation of C. sativa residues. Similarly, Sellami et al. (Sellami et al. 2011) demonstrated that carefully controlled drying regimes can favor the retention of specific terpenes. Optimizing these processing parameters is thus crucial to ensure high and reproducible VO quality.

To further explore spatial and chemotypic patterns, multivariate analyses were performed on the Italian dataset compiled from published VO profiles of C. sativa (Bertoli et al. 2010; Da Porto et al. 2014; Fiorini et al. 2019; Pieracci et al. 2021; Smeriglio et al. 2020; Benelli et al. 2018; Mazzara et al. 2022, 2023; Juliano et al. 2024; Nissen et al. 2010; Barbalace et al. 2023; Ascrizzi et al. 2024; Spinozzi et al. 2025); (see Supplementary Table SS1 for the full list).

PCA (Fig. 7) summarized the chemical variability of VOs obtained from Italian Cannabis samples collected across distinct regions. The first two principal components explained 26.52% (PC1) and 11.15% (PC2) of the total variance, respectively, accounting for 37.67% of the overall information captured by the model. In the biplot representation (Fig. 7A, and Supplementary Table SS4), the 95% confidence ellipse delimited the global dispersion of sample scores, showing that most samples remained concentrated around the central multivariate domain, whereas a restricted subset exhibited directional displacement associated with chemically distinctive signatures. Although the first two components explain a moderate fraction of the total variance, the ordination reveals substantial yet structured compositional heterogeneity across Italian localities, supporting the existence of recurrent chemotypic profiles rather than purely stochastic fluctuations in terpene composition.

PC1 represented the major axis of chemical differentiation and primarily described a gradient separating monoterpene-enriched profiles from compositions dominated by sesquiterpene derivatives and oxidation-associated markers within the caryophyllene/humulene block. On the positive side of PC1, the highest loading vectors included oxygenated/oxidized sesquiterpenes such as isoaromadendrene epoxide, humulene oxide II, β-eudesmol, cubenol, and bisabolol-related signatures, together with hydroxylated caryophyllene-type derivatives (notably 14-hydroxy-9-epi-(E)-caryophyllene-type compounds).

Caryophyllene oxide also displayed a relevant positive contribution along PC1 and an opposing (negative) component along PC2, consistent with its role as a marker of oxidative drift and chemotypic specialization along the oxidation axis (Bertoli et al. 2010; Pieracci et al. 2021; Smeriglio et al. 2020; Ascrizzi et al. 2024). In contrast, the negative side of PC1 was supported mainly by hydrocarbon monoterpenes, especially α-pinene, β-pinene, myrcene, limonene, trans-β-ocimene, and terpinolene, reflecting “fresh” monoterpene-rich chemistries commonly reported in several Italian case studies (Bertoli et al. 2010; Fiorini et al. 2019; Benelli et al. 2018; Mazzara et al. 2022; Barbalace et al. 2023; Ascrizzi et al. 2024). Importantly, within this negative PC1 domain, sesquiterpene hydrocarbons also contributed substantially, with E-β-caryophyllene and α-humulene showing strong weights and marking a second, structurally coherent block associated with sesquiterpene hydrocarbon dominance. Accordingly, PC1 can be interpreted as the principal compositional trajectory spanning (1) monoterpene-rich oils and (2) sesquiterpene-dominated oils shifting towards oxygenated/oxidized derivatives, reflecting both biosynthetic emphasis and potential post-harvest oxidative conversion dynamics.

PC2 described a secondary gradient refining the discrimination among samples already positioned along PC1. Positive PC2 values were preferentially associated with compounds such as myrcene, α-pinene, selina-3,7(11)-diene, guaiol, and 10-epi-γ-eudesmol, indicating that this axis integrates both monoterpene abundance (particularly myrcene/α-pinene) and selective contributions of oxygenated sesquiterpenes (e.g., guaiol- and eudesmol-related alcohols). Conversely, negative PC2 scores were strongly aligned with α-humulene and E-β-caryophyllene, and additionally with oxidation-associated markers exhibiting negative PC2 contributions (e.g., caryophyllene oxide and caryophylla-4(14),8(15)-dien-5-ol). This arrangement indicates that PC2 operates as a refinement axis discriminating among sesquiterpene hydrocarbon-rich samples and those combining monoterpene dominance with targeted sesquiterpene oxygenation. In practical terms, PC2 enhances within-group resolution by capturing the balance between (1) caryophyllene/humulene hydrocarbon abundance and (2) mixed monoterpene–oxygenated sesquiterpene signatures, a pattern compatible with modulation by genotype, pedoclimatic conditions, agronomic management, and/or post-harvest processing (Fiorini et al. 2019; Benelli et al. 2018; Mazzara et al. 2022; Barbalace et al. 2023).

Inspection of sample scores (Fig. 7B) revealed partial geographic segregation, with Rovigo representing the clearest outlying behavior. Several Rovigo samples were markedly displaced towards strongly positive PC1 values, including extreme scores (e.g., 75-Rovigo, 14-Rovigo, 54-Rovigo, 17-Rovigo, 41-Rovigo), consistent with their association with oxygenated/oxidized markers and hydroxylated caryophyllene-related derivatives contributing positively to PC1 (Pieracci et al. 2021; Smeriglio et al. 2020; Ascrizzi et al. 2024). This separation supports the interpretation that a subset of Rovigo oils is chemically specialized and/or affected by enhanced oxidative conversion of sesquiterpene hydrocarbons. By contrast, most other localities remained concentrated at negative PC1 values, exhibiting extensive overlap among regions and a more compact distribution, in line with profiles dominated by hydrocarbon monoterpenes and/or sesquiterpene hydrocarbons such as E-β-caryophyllene and α-humulene. Along PC2, samples exhibiting higher positive values (e.g., 24-Ascoli Piceno and 115-Naples) aligned more strongly with the myrcene/α-pinene/guaiol/eudesmol-associated vectors, whereas markedly negative PC2 scores (e.g., 40-Foggia and several Fiuminata individuals) aligned with the α-humulene/E-β-caryophyllene direction and with oxidation-associated markers displaying negative PC2 contributions.

Collectively, the PCA indicates that volatile-oil variability in Italian Cannabis is primarily organized along an oxidation-associated transition axis (PC1) opposing monoterpene-rich signatures to oxygenated/oxidized sesquiterpene derivatives, while a second axis (PC2) structures additional differences in the relative prominence of sesquiterpene hydrocarbons versus mixed profiles enriched in myrcene and selected oxygenated sesquiterpenes. This multivariate architecture is consistent with regional chemovariation and/or differential modulation of terpene metabolism by environmental and agronomic drivers, with the strongest separation being observed for part of the Rovigo set along the positive PC1 domain.

In chemical biogeography, the existence of an indicator of a “chemical geotype” (chemical phenotypic plasticity arising from environmental conditions at a local scale) is often invoked to support interpretation (Ramos et al. 2023). However, the Italian PCA shows that regional labels do not yield a single, fully segregated chemotype; instead, they reflect a scenario of high chemical phenoplasticity, highlighting the potential of local environmental conditions to shape a geotype. Rovigo samples illustrate this pattern clearly: while a subset shifts strongly toward positive PC1 (e.g., 14-, 41-, 54-, and 74-Rovigo), other Rovigo entries remain embedded within the main multiregional cluster.

The loadings indicate that PC1 is driven primarily by oxygenated sesquiterpenes and by caryophyllene/humulene derivatives, suggesting that the “Rovigo-shifted” subset expresses a more oxidized/oxygenated signature. This within-locality dispersion supports a model in which chemotypes are shaped by both genetic background and environmentally modulated expression (phenotypic plasticity), as well as by anthropogenic influences, potentially compounded by methodological differences among studies (extraction and post-harvest handling), rather than being determined solely by geographic origin.

HCA (Fig. 8) corroborated the PCA patterning and resolved four major chemotypic clusters, reinforcing the existence of discrete recurrent volatile-oil profiles. In contrast to earlier classifications proposing a higher number of clusters, the circular dendrogram supports a more parsimonious chemotypic structure, defined by four dominant compositional regimes (color-coded in Fig. 8): (1) a sesquiterpene hydrocarbon profile dominated by E-β-caryophyllene and α-humulene (blue); (2) a monoterpene-rich profile characterized by myrcene and α-pinene, with relevant contributions of limonene and β-pinene (purple); (3) an oxidized/oxygenated sesquiterpene profile dominated by caryophyllene oxide and caryophylla-4(14),8(15)-dien-5-ol (orange); and (4) a terpinolene-dominant monoterpene profile (green). Thus, considering the major constituents driving sample separation, three structural blocks remain mechanistically decisive hydrocarbon monoterpenes, sesquiterpene hydrocarbons, and oxygenated/oxidized sesquiterpenes, yet the updated clustering indicates that monoterpene-rich profiles subdivide into two chemically meaningful subtypes: a myrcene/α-pinene profile versus a terpinolene-dominant profile.

The largest chemotypic block recovered by HCA (blue branches, Fig. 8) corresponds to a sesquiterpene hydrocarbon-dominated chemotype, defined by high mean contributions of E-β-caryophyllene and α-humulene. This cluster reflects oils with a clear sesquiterpene hydrocarbon backbone, typically accompanied by secondary monoterpene contributions (e.g., α-pinene and myrcene) but without monoterpenes becoming dominant. From a biosynthetic standpoint, this profile is consistent with strong flux through the farnesyl diphosphate pathway and preferential accumulation of caryophyllene/humulene congeners, a pattern repeatedly reported as a stable hallmark in multiple Italian localities and cultivation contexts (Da Porto et al. 2014; Fiorini et al. 2019, 2020; Benelli et al. 2018; Mazzara et al. 2022, 2023; Aguzzi et al. 2023; Ovidi et al. 2022; Juliano et al. 2024; Vinciguerra et al. 2024; Nissen et al. 2010; Micalizzi et al. 2021b).

In practical terms, this chemotype occupies the negative-to-intermediate PC1 domain while projecting towards negative PC2 values, reflecting the alignment of α-humulene and E-β-caryophyllene with the negative direction of PC2. This cluster may represent a baseline “caryophyllene–humulene” signature that is relatively resilient to monoterpene fluctuations, potentially reflecting genetic stability and/or cultivation conditions favoring sesquiterpene hydrocarbon production.

A second major cluster (purple branches) corresponds to a monoterpene-rich chemotype dominated by myrcene and α-pinene, with consistent secondary enrichment in limonene (≈ 8.4%) and β-pinene. Chemically, this profile reflects a pronounced monoterpene imprint (“resinous/pine-like” with citrus accents), where hydrocarbon monoterpenes overwhelmingly structure the aromatic identity and the sesquiterpene fraction becomes quantitatively subordinate.

This pattern aligns with negative PC1 values and positive PC2 contributions, coherently reflecting the loading vectors of myrcene, α-pinene, and limonene. Within this monoterpene-rich cluster, an internal subgroup is evident for Sassari (Namdar et al. 2018; Georgieva and Kosev 2018), which tends to separate due to extremely high myrcene levels combined with comparatively lower α-pinene. By contrast, the remaining samples in the purple cluster exhibit a more balanced α-pinene/myrcene relationship (approximately a 1:1 ratio), supporting the interpretation of a geographically influenced substructure within the monoterpene domain (Juliano et al. 2024; Vinciguerra et al. 2024; Menghini et al. 2021; Pellegrini et al. 2021). This Sassari-associated deviation is compatible with insular Mediterranean conditions and/or cultivar-specific effects, reinforcing that monoterpene dominance can occur through alternative “myrcene-driven” or “balanced α-pinene/myrcene” approximately within the same broad chemotype.

The third cluster (orange branches) defines a highly specialized oxidized/oxygenated sesquiterpene chemotype, strongly isolated by the dominance of caryophyllene oxide and caryophylla-4(14),8(15)-dien-5-ol, with the volatile signature being essentially concentrated into these two constituents. In the dendrogram, this profile is most clearly represented by a restricted Rovigo subset (notably 4-Rovigo and 12-Rovigo), indicating that the most extreme oxidation-dominated phenotype corresponds to a narrowly defined chemotypic enclave rather than encompassing the entirety of the Rovigo material. This configuration is nevertheless congruent with the strong positive PC1 loadings of oxygenated/oxidized sesquiterpenes and hydroxylated caryophyllene derivatives (Pieracci et al. 2021; Smeriglio et al. 2020; Ascrizzi et al. 2024), and with the displacement of part of the Rovigo set towards positive PC1 values in the PCA.

From a mechanistic standpoint, this profile can be interpreted through two non-mutually exclusive hypotheses. First, it may represent genetically distinct material with enhanced intrinsic capacity for sesquiterpene oxidation and accumulation of oxygenated derivatives. Second, it may reflect pronounced post-harvest oxidative conversion of sesquiterpene hydrocarbons into oxygenated congeners, driven by storage conditions, drying parameters, time, oxygen exposure, and temperature. Notably, the overwhelming predominance of caryophyllene oxide strongly supports an oxidative route, consistent with the well-documented susceptibility of sesquiterpene hydrocarbons to conversion into epoxides and alcohol derivatives under oxidative environments (Bertoli et al. 2010; Pieracci et al. 2021; Smeriglio et al. 2020; Ascrizzi et al. 2024). Regardless of the driving mechanism, the orange cluster represents the most chemically distinctive phenotype recovered in the Italian dataset, functioning as a robust marker group with high discriminatory power.

The fourth cluster (green branches) corresponds to a terpinolene-dominant monoterpene chemotype, characterized by high terpinolene and substantial myrcene, accompanied by secondary E-β-caryophyllene and trans-β-ocimene. This cluster differs from the broader monoterpene-rich purple chemotype by the clear “terpinolene peak” structuring the profile, defining a chemically coherent subtype within monoterpene dominance. This group is geographically represented by a restricted subset of samples, including Pisa, Amandola, and Ascoli Piceno, in agreement with the partial segregation observed in PCA and supporting earlier regional associations for the Tuscany/Marche belt (Bertoli et al. 2010; Mazzara et al. 2022; Barbalace et al. 2023; Ascrizzi et al. 2024). Moreover, an internal subpattern is evident within the green cluster: the individual 83-Pisa exhibits unusually elevated trans-β-ocimene, whereas the remaining green members (25-Ascoli Piceno and 57-Amandola) show higher limonene and greater contribution of E-β-caryophyllene. This indicates that, even within the terpinolene-dominant regime, secondary monoterpenes and the sesquiterpene backbone modulate the final aromatic implementation, suggesting either cultivar effects or fine-scale environmental drivers acting on monoterpene branching pathways.

Taken together, the integrated PCA–HCA framework supports a chemically meaningful and internally consistent classification of Italian Cannabis VOs into four major chemotypes: (1) a sesquiterpene hydrocarbon E-β-caryophyllene/α-humulene type (blue), (2) a monoterpene-rich myrcene/α-pinene type with limonene/β-pinene contributions (purple), (3) an oxidized/oxygenated sesquiterpene type dominated by caryophyllene oxide and caryophylla-4(14),8(15)-dien-5-ol (orange), and (4) a terpinolene-dominant monoterpene type (green). Importantly, this updated structure refines previous interpretations by separating monoterpene-rich oils into two distinct regimes (myrcene/α-pinene versus terpinolene-driven), while preserving the biological and practical relevance of the sesquiterpene hydrocarbon and oxidized sesquiterpene extremes. Such multivariate convergence strongly indicates that regional Italian volatile-oil diversity is organized around a limited set of recurrent terpene design principles, plausibly shaped by genotype-by-environment interaction, cultivation conditions, and the chemical lability of sesquiterpene hydrocarbons toward oxidation during post-harvest handling (Bertoli et al. 2010; Fiorini et al. 2019; Benelli et al. 2018; Mazzara et al. 2022; Barbalace et al. 2023; Ascrizzi et al. 2024).

Pearson’s correlation analysis between geography and VO composition (Fig. 9) provides further support for these chemotypic patterns and was calculated on the same integrated dataset (Bertoli et al. 2010; Da Porto et al. 2014; Fiorini et al. 2019, 2020; Pieracci et al. 2021; Smeriglio et al. 2020; Benelli et al. 2018; Mazzara et al. 2022, 2023; Ascrizzi et al. 2024; Spinozzi et al. 2025). Elevation is negatively correlated with latitude (r = − 0.34, p < 0.05) and positively with longitude (r = 0.42, p < 0.05), reflecting a spatial gradient in which higher-altitude sites are located predominantly in central Apennine regions, whereas low-lying areas correspond to northern plains such as the Po Valley and Rovigo. Elevation shows positive correlations with classical monoterpenes (α-pinene, myrcene; r = 0.26–0.31, p < 0.05) and with non-oxygenated sesquiterpenes (E-β-caryophyllene, α-humulene; r up to 0.35, p < 0.05), but negative correlations with several oxygenated sesquiterpenes, including caryophyllene oxide and related epoxides and alcohols (r = − 0.28 to − 0.45, p < 0.05). Conversely, latitude is negatively associated with monoterpenes (α-pinene, myrcene, limonene; r = − 0.23 to − 0.24, p < 0.05) and positively correlated with oxygenated sesquiterpenes such as caryophyllene oxide, humulene oxide II, β-eudesmol, and cubenol (r = 0.26–0.42, p < 0.05). Thus, higher-altitude Apennine sites tend to accumulate monoterpenes and non-oxygenated sesquiterpenes, whereas low-altitude northern localities are enriched in oxidized sesquiterpenes.

The internal structure of the correlation matrix also reveals distinct chemical modules. Monoterpenes covary positively (for example α-pinene with β-pinene, and myrcene), defining a coherent monoterpenic block associated with the α-pinene/myrcene chemotype described for several central and southern localities (Fiorini et al. 2019; Benelli et al. 2018; Juliano et al. 2024; Palmieri et al. 2021). Non-oxygenated sesquiterpenes E-β-caryophyllene and α-humulene are strongly correlated (r = 0.56, p < 0.05), reflecting their shared biosynthetic origin and their joint occurrence in many Apennine samples (Mazzara et al. 2022, 2023; Fiorini et al. 2020). Oxygenated sesquiterpenes form an exceptionally cohesive “oxidative” module: caryophyllene oxide, isoaromadendrene epoxide, humulene oxide II, β-eudesmol, β-bisabolene, guaiol, 10-epi-γ-eudesmol, and α-bisabolol are all highly and positively intercorrelated (many pairs with r ≥ 0.8, p < 0.05), mirroring the strongly oxidized profiles documented for Rovigo and related samplesn (Pieracci et al. 2021; Smeriglio et al. 2020). Importantly, this oxidative module is antagonistic to the monoterpenic block, as indicated by significant negative correlations between α-pinene or myrcene and caryophyllene oxide or caryophylla-dienol (r = − 0.36 to − 0.41, p < 0.05). When monoterpenes dominate, oxygenated sesquiterpenes are consistently reduced, and vice versa.

At the regional scale, these gradients translate into clear geographic patterns. Low-altitude northern plains (Rovigo, Po Valley, and part of Friuli Venezia) are characterized by chemotypes rich in oxygenated sesquiterpenes, consistent with low elevation, high latitude, and the strong internal coherence of the oxidative module (Da Porto et al. 2014; Pieracci et al. 2021; Smeriglio et al. 2020; Nissen et al. 2010). In contrast, central Apennine and central–northern regions (Fiuminata, Ascoli Piceno, Amandola, San Severino, mountainous Abruzzo, parts of Tuscany and Umbria) show higher elevations and intermediate latitudes, favoring monoterpenes and non-oxygenated sesquiterpenes and aligning with the α-pinene/myrcene/caryophyllene–humulene chemotypes (blue and yellow clusters in Fig. 8) (Fiorini et al. 2019; Benelli et al. 2018; Zheljazkov et al. 2020; Aguzzi et al. 2023; Palmieri et al. 2021; Barbalace et al. 2023; Spinozzi et al. 2025; Micalizzi et al. 2021b; Di Sotto et al. 2022; Menghini et al. 2021). The central–eastern belt (Pisa, Amandola, part of Ascoli Piceno) hosts the citrus/terpinolene chemotype, occupying an intermediate position between monoterpenic and oxidized sesquiterpenic extremes (Bertoli et al. 2010; Mazzara et al. 2022; Barbalace et al. 2023). Southern and insular regions (Naples, Ragusa, Foggia, Sassari) generally present higher relative levels of monoterpenes and lower proportions of oxidized sesquiterpenes, in agreement with their lower latitudes and more Mediterranean climatic conditions (Ovidi et al. 2022; Juliano et al. 2024; Vinciguerra et al. 2024; Micalizzi et al. 2021b).

Taken together, the mapping of chemotypes (Figs. 5–6), multivariate analyses (Figs. 7–8) and correlation structure (Fig. 9), derived from the integrated analysis of Italian C. sativa VOs reported across more than twenty-five independent studies (Bertoli et al. 2010; Da Porto et al. 2014; Fiorini et al. 2019, 2020; Pieracci et al. 2021; Smeriglio et al. 2020; Benelli et al. 2018; Mazzara et al. 2022, 2023; Juliano et al. 2024; Nissen et al. 2010; Barbalace et al. 2023; Ascrizzi et al. 2024; Spinozzi et al. 2025), and references therein, depict Italian C. sativa inflorescences as a mosaic of volatile profiles organized along two main axes: (i) an environmental gradient (elevation/latitude) that opposes monoterpene- and non-oxygenated sesquiterpene -rich oils to oxygenated sesquiterpene-rich oils, and (ii) an internal chemical axis that structures compounds into tightly covarying modules (α-pinene/myrcene, limonene/terpinolene, caryophyllene/humulene, and oxidized caryophyllene/humulene derivatives). This mosaic reflects the combined influence of genotype, local environment, and post-harvest management and emphasizes the importance of territoriality in shaping the chemodiversity and potential bioactivity of Cannabis VOs.