Distribution, Ecology and Invasion Potential of Sorghum halepense (Poaceae) in the Insubric Region (Northern Italy and Southern Switzerland)

1 Introduction

Within the southern slope of the Alps, the Insubric region spans over northern Italy (Lombardy and Piedmont) and southern Switzerland (Canton Ticino) and is characterised by a particularly mild and wet climate [1,2] which supports remarkable native plant diversity and facilitates the establishment of alien species in the lowlands [3,4]. Over the last few decades, several new alien plant species have been identified (e.g., [5,6,7]), and already established ones have considerably expanded their range throughout the region [8,9]. Sorghum halepense (L.) Pers. (Poaceae), Johnson grass, is a notable example of the latter case. It is a plant well adapted to a humid subtropical climate [10]. Indeed, its recent rapid spread in the Insubric region has caused concern among the competent authorities, especially in the Swiss part of the region. This species is considered invasive in an increasing number of countries worldwide [10], causing significant economic losses in agricultural systems [11]. Therefore, a deeper understanding of the species’ distribution, ecology, and invasion potential is crucial for effective management policies. This is especially true in the Insubric region, where the species is in the early phases of invasion.

Sorghum halepense is a perennial C4 rhizomatous geophytic grass that can reach over 2 m in height. The most characteristic traits are the presence of a wide panicle, long and narrow leaves with a prominent white midrib, and relatively large creeping rhizomes [12]. Despite its uncertain origin, S. halepense is considered native to the area spanning from northeast Africa to the Indian subcontinent [13] and, according to POWO [14], actually across all northern Africa to Mainland Southeast Asia. Several biological features, such as rapid growth, secretion of allelopathic substances, and efficient vegetative and sexual reproduction have contributed to the successful establishment of this grass throughout the world outside its native area over the last two centuries, after being spread mainly as a forage plant or through contaminated seed stock and soil (e.g., [11,15]). Indeed, it currently occurs on all continents except Antarctica [11,14] but prefers subtropical and warm temperate climates [15,16]. In Europe, S. halepense has been recorded for over five centuries in some southern and warmer regions, such as in Italy [17,18], but has been spreading northward since the end of the 1800s [10,11,19]. In the last three decades, the species has invaded new areas in Central Europe (e.g., in Austria, Hungary, Slovakia and Slovenia [19,20,21,22,23]) and occupied new types of habitats by apparently expanding its ecological niche, probably promoted by climate change, monoculture crops and the misuse of herbicides [11,23,24,25].

For instance, S. halepense is currently considered a threat to agriculture in over 50 countries [11], considerably damaging the world’s main crop production and causing huge economic losses [19,23,26,27,28,29]. Only recently, some authors have pointed out that S. halepense can reduce the biodiversity of native species in the United States [30], mainly through size-asymmetric competition [31], and may thus pose a threat to natural systems, especially grasslands [29]. In Europe, the species has not yet been thoroughly investigated, and the few relevant studies available (e.g., [11,19,22,23]) primarily focused on agriculture impacts and control methods. Although S. halepense is not on the EU’s list of invasive alien species of union concern [32] nor on any lists by the European and Mediterranean Plant Protection Organization [33], it is increasingly recognized as a problem at the national level. Indeed, it has been included in Switzerland’s List of potentially invasive species [34] and on the Black Lists of some northern regions of Italy, such as Piedmont ([35]; not in Lombardy, see [36]), and Portugal as well (see law decree 92/2019 of the Portugal republic on Invasive Aliens). In the remaining European countries, where it is present, it is mainly treated as an agricultural weed rather than a regulated invasive species.

The current spread of S. halepense in Central Europe and predictions about its future distribution (e.g., [24]) explain concerns about the species in the Insubric region, where climatic conditions appear to be increasingly favourable for the species and no scientific research has been published yet. Therefore, in this study, we aimed at: (1) reconstructing the invasion history and current distribution of S. halepense based on herbarium specimens, databases, literature, and occurrence records, (2) investigating aspects of its ecology, in particular its growth ability, in the main habitats that it occupies; and, finally, (3) assessing and discussing its invasion potential in the Insubric region, especially in natural and semi-natural habitats. Our goal was to provide information on S. halepense outside croplands to eventually support national risk assessments and subsequent listings and decisions on regulation, control and management of the species.

2 Methods

2.1 Distributional Analysis

The present study focuses on the Insubric region (see Supplementary Material S1), i.e., the region that connects the southern Central Alps with the Po Valley and is characterised by a sequence of mountain chains, several glacial Prealpine lakes, moraine hills, and plains (Fig. 1a) and the so-called ‘Insubric climate’ [1,2].

The historical distribution of S. halepense was investigated using information associated with herbarium specimens (both general and historical collections) from Northern Italian and Swiss herbaria and literature data either published or retrieved from web sources (Supplementary Material S2) (e.g., [12,37,38,39,40,41,42]). Only herbarium specimens collected in Switzerland, Lombardy and Piedmont were considered, excluding those lacking useful information, duplicate specimens, and specimens collected in the same year at the same place. If not available and whenever possible, the site and/or the year of collection was inferred based on reliable information in the literature about the operational area of the collectors or their personal herbarium collections to which the sample belongs, and by comparison with data from surrounding areas. Datasets were organized (by collection year, collection site, collector, herbarium, and any additional information), georeferenced with Google Earth Pro (v.7.3.4, 2023; https://www.google.com/earth/about/versions/#earth-pro), and then mapped with QGIS software (v.3.22.3 ‘Białowieża’, 2022; https://qgis.org) using base maps from OpenStreetMap (© OpenStreetMap contributors, 2022) licensed under the Open Data Commons Open Database License (ODbL) and the Creative Commons Attribution-ShareAlike 2.0 License (see www.openstreetmap.org/copyright for details). To graphically visualize the spread dynamics of S. halepense and compare the distribution data over time, records were classified into five temporal categories based on consecutive periods in which 20% of herbarium specimens were collected (following the methods by Ciappetta et al. [43] for Ambrosia artemisiifolia). Finally, to infer the collection rate and discuss specimen collections in the context of S. halepense spread, the cumulative number of specimens collected over the years was plotted on a graph using MS Excel.

The current geographical distribution of S. halepense in the Insubric region was reconstructed using occurrence records from Canton Ticino, Lombardy, and Piedmont, from a wide range of sources, including national and regional floristic databases, field work during this study, herbaria, natural history museums, universities, and local experts (see full list and data extraction criteria in Supplementary Material S3). The full dataset from the three regions was then mapped using QGIS (v.3.40.5-Bratislava, 2025; https://qgis.org).

Figure 1: Distributional analysis of Sorghum halepense using herbarium data. (a) Spatial distribution of S. halepense herbarium specimens in Canton Ticino, Piedmont, and Lombardy, classified into five temporal categories based on years in which 20% of the specimens were collected, following the method of Ciappetta et al. [43]. Darker dots represent older specimens. The boundaries of the Insubric region (shaded area) are indicative (see Supplementary Material S1). (b) Cumulative number of S. halepense herbarium specimens collected in Canton Ticino, Lombardy, and Piedmont from 1840 to 2011 (year of the most recent specimen). The first specimen from the Insubric region was collected in 1883.

2.2 Ecological Analysis

The ecological analysis of S. halepense focused on Canton Ticino, the north-western part of the Insubric region. To select appropriate sites, habitats in which the species grows were preliminarily selected by superposing the georeferenced occurrences on a background map of aerial photos and activating a layer of the map of natural environments available on the Swiss online geographic mapping service Swisstopo (https://map.geo.admin.ch/). Subsequently, field surveys allowed selection of four types of habitats occupied by S. halepense from the lowlands to the hill/mountain area: (1) hygrophilous habitats (canal banks and lakeshores), (2) cultivated areas, (3) meadows/roadsides, (4) meadows (less representative type of habitat, only present in the hill/mountain area). Finally, three sampling sites were chosen for each lowland habitat type, whereas only one was found in the meadows in the hill/mountain area. Nine sites were located in Canton Ticino, as far as possible from each other, and one was in Italy at the southern limit of the Insubric region (Pozzuolo Martesana; see Supplementary Material S4).

Vegetation surveys were performed in 44 plots (each 5 × 5 m). At each site, five plots were randomly selected, except for two sites (Pianezzo and Tenero), where only one and three plots, respectively, could be selected because of very small habitat patches. In each plot, the vegetation survey consisted of a list of vascular plant species and a visual estimate of the cover (total and single species). Species were determined using identification keys of floras [12,40,44] and digital identification tools [40]. Precise coordinates and elevation were recorded for each plot using FlorApp, and the species list was compiled using the ‘vegetation survey module’.

For the morphological trait and soil analyses, in each sampling plot, ten S. halepense specimens and an average of two soil samples were collected. Plant height (cm), spike length (cm) and major leaf length (cm) were measured directly in the field after collection. Leaf area (cm2) was calculated for the second leaf under the spike with Easy Leaf Area Free application [45]. Plant and leaf dry weight were measured with a digital electronic analytical scale (0.0001 g precision) after drying them in an oven at 65°C for 58 h. Finally, the specific leaf area (SLA) was calculated by dividing the leaf area by the leaf dry weight. Soil samples were prepared and analysed following the protocol described in Gentili et al. [46]. Thirteen parameters were determined: apparent texture (quantity of sand, silt, and clay), pH, carbonates (Ca), organic matter and organic carbon (C), total nitrogen (N), carbon/nitrogen ratio (CN), cation exchange capacity (CSC) and three exchangeable cations (Ca, Mg, K), base saturation rate, and assimilable phosphorus [47,48].

For statistical analyses, vegetation survey data were used to infer the mean alpha-diversity (species richness) and beta-diversity (Whittaker index) using the Past software (v.4.02; [49]), and then to analyse differences of ecological factors and morphological traits among sites. In particular, the variance of the three ecological variables (alpha-diversity, beta-diversity and S. halepense cover) was analysed with a one-way ANOVA and linear models (lm), whereas that of morphological traits (plant height, spike length, leaf length, dry weight, leaf area, leaf weight and SLA) used linear mixed models (lmm) and site replicates as random factor. ANOVA was performed after applying a Shapiro-Wilk normality test and was followed by a Tukey post hoc test for multiple mean comparisons, and a Bonferroni correction to strengthen the statistical evidence. When necessary, to meet normality assumption, data were log transformed. The analyses were performed using the software R [50].

To understand the relationships among plots, vegetation, and environmental variables (ecological and soil variables, plus latitude, longitude and elevation), a Canonical Correspondence Analysis (CCA) was performed using CANOCO software (v.4.5; [51]). A correlation analysis was performed to examine the Spearman index and exclude the most correlated variables (Spearman index > 0.75), namely, those that provide the same information describing in the same way the distribution of data [52]. To avoid multicollinearity among variables, variance inflation factor (VIF) was assessed through a specific function of the CANOCO software. In this way, 11 variables (elevation, latitude, species richness, Whittaker index, S. halepense cover, sand, pH, total N, C/N, cation exchange capacity, calcium, magnesium and phosphorus) were retained for subsequent analyses.

2.3 Assessments of Invasive Potential and Impacts

The invasive potential results from a combination of the spread potential and the negative impacts (impact potential) caused by a species (e.g., [53]). To assess the invasive potential of S. halepense its biological traits, spread potential, and published impacts were compared to the criteria listed in the Catalogue of criteria by InfoFlora [54], used to compile the Black List and Watch List of alien species of Switzerland [55] and, more recently, to support compilation of the new national list of invasive and potentially invasive alien plant species [32]. Criteria about the spread potential consider sexual and vegetative reproduction rate of the assessed species, persistence of a seedbank, natural and human-mediated dispersal, etc. Those about the impact potential consider impacts on human and animal health, native species, ecosystems (structure, water and nutrient cycles, etc.), infrastructure, and economy (e.g., agricultural systems, tourism, etc.). In this type of risk assessment, a percentage score representing three levels (high, medium, low) is assigned to each criterion based on the respective information available for the assessed species. These percentages are used to support the discussion of the level (high, medium, low) of the invasive potential.

Impacts of S. halepense on native environment were assessed with the Environmental Impact Classification of Alien Taxa (EICAT; [56,57]) following the protocol by the International Union for Conservation of Nature [58]. An impact score is assigned using a five-category system (massive, major, moderate, minor, and minimal concern), depending on the taxonomic level that is negatively affected by the alien species (individual, species, community) and whether damages are reversible or irreversible. These five categories are also used by the Socio-Economic Impact Classification of Alien Taxa (SEICAT; [59]) to score the impact of alien species, referring to their ability to interfere with human activities and with economic wellbeing. The magnitude of the impact of S. halepense was scored based on EICAT/SEICAT-relevant information.

The relevant information on S. halepense for assessment was gathered by searching the scientific literature (Google Scholar, Scopus, WorldCat, using the species name as a keyword) and using the results of this study. Assignment of percentage scores and EICAT and SEICAT categories was performed by the first author, and results were used to support the discussion of the species’ spread and impact potential and draw conclusions on its invasion potential in the Insubric region.

3 Results

3.1 Distributional Analysis

A total of 237 S. halepense specimens from 22 herbaria (out of a total of 33 contacted) were examined, but only 99 (about 42%) were useful for inferring the historical distribution and spread history of the species in the Insubric region in general, and in Canton Ticino, Piedmont and Lombardy more specifically (Supplementary Material S2).

The five temporal categories with each 20% of collected herbarium specimens were: 1840–1897, 1898–1922, 1923–1950, 1951–1982, and 1983–2011 (Fig. 1a). The oldest specimens (not included in the categorization due to the lack of precise collection dates) date back to the late 1700s and early 1800s and came from ruderal areas, cultivated fields, and river and canal banks in major provinces of the Po Valley (Mantua, Turin, Pavia), which are off the Insubric region considered in this study. The first specimen with a precise date (accuracy within one year) is from 1840 and was collected in the hills of Turin (Italy). Approximately a century later, the species began to be collected in more northerly localities (Lecco, Como, Canton Ticino), within the foothill area of the Insubric region, near larger urban centres and lakeshores. The first specimens from this area date back to 1883 and were collected in Galbiate (LC), Lombardy. In general, the oldest specimens from the Insubric region come mainly from major urban centres, while more recent ones also come from suburban areas and roadside meadows. The cumulative number of herbarium specimens collected over time remains nearly constant from 1840 to 1900, increases slightly until the early 2000s, and then shows a sharp decline up to 2011, when the last specimen was collected in Milan (Fig. 1b).

A total of 2914 occurrence records from Canton Ticino, Piedmont and Lombardy (Fig. 2a) were used to infer the current distribution of S. halepense in the Insubric region (~1400 occurrences; Fig. 2a, black dots). The species is widely distributed across the entire Po Valley, and, within the Insubric region, in anthropogenic habitats surrounding Maggiore, Lugano and Como lakes, Mesolcina Valley, as well as the lowlands of the main valleys, and the Brescian and Bergamasque Pre-Alps, throughout the Camonica Valley. In this region, it is widespread from the lowland level up to 1378 m a.s.l (in Bergamasque Pre-Alps), with 98.2% of the occurrences below 700 m a.s.l. in the foothill zone. The northernmost occurrence is in Ponto Valentino (Acquarossa municipality, 615 m a.s.l.) in Canton Ticino. According to the information provided on occurrences, the species most frequently colonises anthropogenic habitats, including ruderal areas, cultivated fields, roadsides, irrigation canals and ditches, and wetlands. Based on field observations of this study, the species also occurs in semi-natural areas, such as unfertilized meadows (e.g., Agra, Bellinzona along the river bank), also in sub-montane areas (e.g., above Brissago, 810 m a.s.l., where the species reaches the highest elevation in Canton Ticino; Brè sopra Lugano, 790 m a.s.l.; Mergoscia, 720 m a.s.l.), though always in proximity to infrastructures (roads or buildings). The cumulative number of occurrence records (Fig. 2b) increases gradually from the mid-1990s until 2003, followed by a sharp rise between 2004 and 2009. After a period of slower growth from 2010 to 2019, the number increased steeply again from 2020 to the present (while herbarium records end in 2011).

Figure 2: Distributional analysis of Sorghum halepense using occurrence data. (a) Spatial distribution of S. halepense in Canton Ticino, Lombardy and part of Piedmont based on all occurrences used in this study. (b) Cumulative number of S. halepense occurrences in Canton Ticino, Lombardy and part of Piedmont from 1990 to 2024.

3.2 Ecological Analysis

The three main biotic factors investigated such as alpha diversity (richness), the beta diversity (Whittaker index) and Sorghum halepense cover exhibited significant differences across sites. Specifically, Alpha diversity consisted of a total of 123 species identified through vegetation surveys, with 5 and 21 species in the poorest and richest plots, respectively, and a mean of 12 species per plot. Site-level differences were statistically significant (ANOVA; F8 = 14.883; p < 0.001) (Fig. 3a). Mean species richness was lowest in Cadenazzo (6 species) and highest in Stabio (18 species). The species most frequently found in association with S. halepense were Arrhenantherum elatius (Poaceae; 23 plots, 52% of all plots), Calystegia sepium (Convolvulaceae; 21 plots, 48%), and Setaria pumila (Poaceae; 18 plots, 41%). Difference in Beta-diversity among sites was statistically significant (F8 = 12.941; p < 0.001) (Fig. 3b); it was lowest in Balerna (Whittaker index = 0.33 mean per plot) and highest in Pozzuolo (0.59), with a mean of 0.44 per site. Differences in Sorghum halepense cover among sites were statistically significant (F8 = 2.2724; p < 0.05) (Fig. 3c); it was lowest in Stabio (27% mean plot) and highest in Balerna (about 59%), with a mean cover of 47.6% per site.

Figure 3: Bar charts showing the log of mean and standard deviation of species richness (a), Whittaker index (b) and S. halepense cover (c) in the surveyed sites. Different letters indicate significant differences among sites according to the post hoc test of Tukey (p < 0.05; ANOVA).

The same parameters showed no statistically significant differences among the three habitat types where the species grows: cultivated, hygrophilous and meadows s.l. (i.e., meadow + roadsides) areas (species richness: F2,6 = 0.25527; p = 0.7827; Whittaker index: F2,6 = 0.27213; p = 0.7707; S. halepense cover: F2,6 = 1.0712; p = 0.4001).

In the morphological analyses of S. halepense, plant height ranged from 140 to 292 cm (mean 195 ± 28) and differed statistically among sites (F9,35 = 2.796; p < 0.05) (Fig. 4a). Based on mean values per site, the shortest and tallest specimens were found in Agra (177 ± 19 cm) and Pozzuolo (230 ± 33 cm), respectively, based on mean values per site. Inflorescence length ranged from 13 to 65 cm (mean 28.5 ± 6.5 cm), with significant differences among sites (F9,35 = 17.183; p < 0.0001) (Fig. 4b); both extremes were observed in Riazzino. Based on mean values per site, the shortest inflorescences were in Agra (22 ± 2.5 cm) and the longest in Cadenazzo (38 ± 6.5 cm). Plant dry weight ranged from 3.4 to 106.63 g (mean 14.46 ± 10.53) and differed statistically among sites (F9,35 = 4.0864; p < 0.01) (Fig. 4c), with the lightest plants found in Agra (mean 8.85 ± 3.43 g), and the heaviest in Cadenazzo (24.42 ± 18.77 g). Leaf length ranged from 32 to 92 cm (mean 65 ± 11), with no significant differences among sites (F9,35 = 1.5779; p = 0.1604) (Fig. 4d). The shortest leaves were in Pianezzo (mean 56.5 ± 6 cm), and the longest in Balerna (70 ± 9.5 cm). Leaf area ranged from 2.50 to 90 cm2 (mean 22.5 ± 15 cm2), and differed significantly among sites (F9,35 = 3.14712; p < 0.01) (Fig. 4e), with the lowest values in Agra (mean 11.58 ± 7.46 cm2) and the highest in Giubiasco (30.44 ± 13.09 cm2). Leaf dry weight ranged from 0.011 to 0.974 g (mean 0.144 ± 0.109), with no significant differences among sites (F9,35 = 1.367831; p = 0.2398) (Fig. 4f). The lightest leaves were found in Agra (mean 0.074 ± 0.048 g) and the heaviest in Cadenazzo (0.186 ± 0.089 g). Specific leaf area (SLA) ranged from 12.32 to 1077.59 cm2 g−1 (mean 164.83 ± 55.5), and differed significantly among sites (F9,35 = 5.201; p < 0.001) (Fig. 4g). The lowest SLA was in Balerna (mean 138.01 ± 29.18 cm2 g−1) and the highest in Stabio (202.59 ± 133.96 cm2 g−1), which also included the specimens with the lowest and highest values, respectively.

Figure 4: Bar charts showing the log of mean and standard deviation of morphological traits of S. halepense samples in the surveyed sites. (a) Plant height. (b) Length of inflorescence (spike length). (c) Plant dry weight. (d) Leaf length. (e) Leaf area. (f) Leaf dry weight. (g) SLA. Different letters indicate significant differences among sites according to the post hoc test of Tukey (p < 0.05; ANOVA).

Differences in the same morphological traits among habitat types (Fig. 5a–g) were statistically significant for inflorescence length (F3,41 = 5.201, p < 0.001) (Fig. 5b), plant dry weight (F3,41 = 4.0567, p < 0.05) (Fig. 5c), leaf area (F3,41 = 2.29057, p = 0.0925) (Fig. 5d) and SLA (F3,41 = 2.871, p < 0.05) (Fig. 5e). In general, cultivated and hygrophilous habitats showed the highest values for all morphological traits, except for SLA.

Figure 5: Bar charts showing the log of mean and standard deviation of morphological traits of S. halepense samples in the different habitats. (a) Plant height. (b) Length of inflorescence (spike length). (c) Plant dry weight. (d) Leaf length. (e) Leaf area. (f) Leaf dry weight. (g) SLA. Different letters indicate significant differences among sites according to the post hoc test of Tukey (p < 0.05; ANOVA).

For the CCA, the correlation among the 11 selected environmental variables and the distribution of surveyed species and plots is shown in Fig. 6. The four canonical axes cumulatively explained 53.7% of the total variance, with the first and second axes accounting for 30.6%. Species and plots mostly clustered in the first and third quadrants and were significantly distributed along the first and second axes (Table 1). The environmental variables that best explained variance (p < 0.001) were: assimilable phosphorus (λ = 0.39), elevation (λ = 0.32), species richness, calcium, and longitude (Table 2). All sites except Stabio and some plots of Tenero were negatively correlated with the phosphorus and calcium gradients. The altitudinal gradient was negatively correlated with Pozzuolo, Grancia, and Giubiasco, while pH showed a positive correlation with these same sites and a negative one with Agra, Balerna, and Tenero. Regarding habitat types, all hygrophilous sites fell in the first quadrant, negatively correlated with phosphorus, calcium, and elevation gradients, and positively with the Whittaker index and pH. Meadows/roadsides and cultivated sites (except for the Stabio, positioned on the opposite side) were arranged between the graph centre and the fourth quadrant, negatively correlated with phosphorus, calcium, pH, and Whittaker index, and positively with the altitudinal gradient.

Figure 6: Scatter plot depicting the result of CCA. The distribution of species (crosses) and surveys (colored squares) is shown in relation to the different environmental variables considered (red vectors).

3.3 Assessments of Invasive Potential and Impacts

A total of 51 publications concerning the spread potential and the negative impacts of S. halepense included relevant information allowing assignment of percentage scores to the assessment’s criteria and to support the discussion of the species’ invasive potential (Table 3). For the impact assessment, 11 and 17 publications provided relevant information, allowing classification into an EICAT and SEICAT category, respectively (Table 3).

4 Discussion

Our study is the first to reconstruct the invasion history of S. halepense across the Insubric region, where its presence was first documented by herbarium specimens from 1883. Since then, the species has spread throughout the region and along the lowlands of the Prealpine valleys of the southern Alps, progressively expanding northward and colonising increasingly higher elevations. This pattern is consistent with the rapid spread of the species documented elsewhere in Central Europe, favoured by current climate warming (e.g., [19,21,23]). We discuss its invasive potential and impacts across the Insubric region and elsewhere in Italy and Switzerland.

4.1 Introduction, Spread History and Current Distribution of Sorghum Halepense

In Italy, S. halepense was reported for the first time in 1561 [37], although it is likely to have occurred earlier, as it is considered an archaeophyte (i.e., introduced before the discovery of America) by several authors [17,18]. According to our analyses of herbarium specimens, the spread of S. halepense in the Insubric region began in the late 1800s. This is when the species reached the region during its northward range expansion from populations occurring further south in the Po Valley, where it was already widespread at least a century earlier. In other areas near the Alps and the Po Valley, such as in Trentino-Alto-Adige and around Lake Garda, the species also began to expand northward in the mid-19th century, reaching latitudes further north than Bolzano in the last decade [39,94]. The second half of the 19th century is also when S. halepense started to appear in other Central European countries, such as Austria, Slovakia, Slovenia, and Hungary [10,11,19,23].

Increased trade in those years likely played a key role in the spread of the species. For instance, one of the most reported pathways of introduction of S. halepense in Europe is as a contaminant of agricultural products, such as seed or fodder [11,23,95]. In Canton Ticino, it was introduced unintentionally as well [79], but apparently also deliberately as a fodder species. Indeed, an article published in 1894 in an agricultural magazine (i.e., Agricoltore Ticinese) encouraged the cultivation of the species in the canton, describing it as a hardy and vigorous fodder plant, and the editors even provided the seeds [96]. However, we found no additional historical information about its use as a fodder plant elsewhere in the Insubric region (while it certainly was in the United States [16]), and in several countries of Oceania and South America [10]. Agricultural practices may have contributed to the spread of the species, but unintentional transport seems to be the main cause. This is supported by the fact that, by the end of the 19th century, connections were already in place between towns near the Po Valley and those in Canton Ticino. Furthermore, in the last decades, an increasing number of field specialists have alerted authorities about a rapid spread of S. halepense in the Insubric region [79], which seems to be confirmed by the dataset analysed in this study.

A lag-phase (or stabilisation phase, i.e., between the time of introduction and the start of exponential spread; [53,97]) of approximately 100 years for many invasive species [98], including S. halepense, was found in Central European countries such as Austria and Slovakia [19,23]. Introduced there by the end of the 19th century and once considered rare, S. halepense is now reported in the phase of rapid spread, especially in cultivated fields. In regions with milder climates and earlier introductions, such as northern Italy and Hungary, the species appears to be at the beginning of the saturation phase, as already described by other authors [23,53,97] and according to many personal observations of the authors in the Po Valley during 2021. In the Insubric region, our results suggest that the species is still in the phase of rapid spread, implying that further expansion is expected before reaching the saturation phase. This is also supported by the apparent benefit that S. halepense derives from ongoing anthropogenic climate warming, as average annual temperature is a well-documented limiting factor for its development in Central Europe [23,99]. According to the most optimistic climatic scenario, by 2050, all major agricultural areas in Europe, as well as urban areas and grasslands all over the world, will be climatically suitable for S. halepense [24,25]. This trend is already partially observable in Austria, Germany, and Slovenia [20,22,23], and likely plays a key role in explaining its spread also in the Insubric region, where the species is currently found at the northern limit of its climatic range south of the Alps.

Land use change and certain agricultural practices also contribute to the spread of invasive species [25,99]. For instance, in Hungary [100] and Austria [22], the distribution area of S. halepense has expanded following the intensification of maize cultivation. Agricultural machinery significantly contributes to the fragmentation and dispersal of rhizomes during tillage [71], and to the dispersal of seeds by combine harvesters, which may scatter them up to 26 m forward and 12 m backward from the infestation point [60]. Additionally, the use of certain herbicides, an undiversified choice of them, and incorrect application have, in some cases, led to resistance, allowing for the selection of late-emerging ecotypes [100,101,102]. New ecotypes can also emerge naturally due to species’ high phenotypic plasticity and adaptability, as observed in Canada and Russia, where several populations developed deeper rhizomes and increased tolerance to cold temperatures [15].

4.2 Ecological Aspects of S. halepense in Invaded Habitats

In the Insubric region, S. halepense grows in the same habitats where it is considered an invasive alien elsewhere in the world: ruderal areas, cultivated fields, vineyards, canal margins, and wetlands (especially artificial ones), as well as meadows along roadsides and railway embankments [19,26]. In Switzerland, the latter two habitats are often classified within the plant alliance Arrhenatherion, indicative of low-elevation, mown grassland environments [102]. These are considered semi-natural habitats, as they are generally mown only two to three times per year and can support a relatively high floristic diversity of notable biological value [102,103,104]. Even in the Insubric region, such habitats can host plant species of conservation interest [105]. Thus, to our knowledge, the present study may be the first in Europe to show that S. halepense is able to colonise semi-natural habitats. Furthermore, in Switzerland, areas characterised by the Arrhenatherion alliance are undergoing rapid decline in locations subject to intensive agricultural management [102]. Therefore, the dominance of S. halepense may represent an additional threat to this habitat, where the species has already reached an average cover of 53% in the invaded sites in our surveys. However, its presence remains clearly dependent on the proximity of transport infrastructure, which facilitates continuous dispersal of seeds and propagules, particularly at higher elevations.

The ability of S. halepense to colonise a wide range of habitats may reflect a high degree of phenotypic plasticity, as previously highlighted by Atwater et al. [106]. Although our results showed significant differences in some morphological traits among sites and in the meadows/roadside habitat, the overall consistency of traits across other environments may indicate that the species maintains performance across contrasting conditions, which is consistent with a generalist ecological strategy. However, the lack of significant differences in other morphological parameters among habitats supports the species’ adaptability to a wide range of environments, with the species’ robust rhizomes ensuring drought resistance, functioning also as water and nutrient storage organs [107].

Results of CCA showed a negative correlation between species richness and the cover of S. halepense, suggesting that size-based competition is one of the most effective strategies through which the species competes with and suppresses co-occurring species [36]. Although Rout et al. [86] identified S. halepense cover as a major factor influencing species richness in invaded sites, in our study this parameter was not statistically significant.

The negative correlation between sites hosting individuals with greater biomass and size and the altitudinal gradient confirms the thermophilic nature of S. halepense. Conversely, the negative correlation between the species’ occurrence and the presence of nutrients such as phosphorus and calcium is not clearly explainable. Rout and Chrzanowski [86] observed that soils long invaded by S. halepense tend to show higher nutrient concentrations, likely due to the species’ ability to produce certain chemical elements through association with specific bacteria, as well as allelopathic substances that alter the soil geochemical properties. This finding suggests that S. halepense could potentially gain a competitive advantage and thrive even in originally nutrient-poor soils.

4.3 Invasive Potential: Spread and Impact

Sorghum halepense is considered a noxious invasive alien species in many countries around the globe [10,11], but is not yet blacklisted in Europe and in the Insubric region, where its potential spread and impacts may currently be underestimated. We now review and discuss the ecological characteristics that contribute to the species’ rapid spread and the negative impacts that it can cause, following the order of the InfoFlora Criteria Catalogue [54]).

Spread potential is estimated based on four criteria. First, sexual reproduction is high in S. halepense: the species is typically self-pollinating [61], producing an average of 28,000 [62] and up to 100,000 seeds per plant [63]. These remain dormant in the first year [60,64] and viable for up to 7 years in dry locations [65], with viability decreasing to 50% after 5 years [66,67], or none after 2.5 years if buried less than 20 cm deep [68] or in managed soils [69]. Local spread occurs predominantly through horizontal vegetative growth [15,65,69], thanks to its thick and deep rhizomes and roots, 40 to 90 m long (per individual plant) and bearing numerous new potential shoots [62]. Individuals developing from rhizomes are more competitive, emerge earlier, grow faster, and have a higher fresh weight than seed-derived individuals [28]. Rhizomes tolerate desiccation as well as frosty days and sprout with temperatures as low as 15°C (maximum between 23 and 30°C; e.g., [15,72]). In addition, seed dispersal ranges from within a radius of a few meters by gravity [15] to considerable distances carried by water, wind, or animals (e.g., [65]), with measured water transport of at least 400 m [73]. Seeds are transported on the animals’ fur thanks to sticky appendages or manage to pass unaltered through the digestive tract of most herbivores [61,65].

Human activities strongly contribute to the long-distance dispersal of S. halepense, with agriculture among the most significant due to past cultivation as a forage crop (e.g., [10,19]). After tillage and threshing operations, invaded fields agricultural machinery can transport seeds and rhizomes tens of meters away [75]. Seeds are also dispersed along major roads unintentionally by vehicles (in truck loads or attached to tyres or interior mats) and by turbulence from vehicle and railway traffic [76,77]. An estimate of the spreading speed of the species in the region of interest is the last criterion to assess spread potential. In our case, based on our distributional dataset, we estimated that S. halepense increased its invasion front by expanding about 30 km northwards in a dozen years in Canton Ticino, which is considered to be fast. In sum, taken all together, information about the spread criteria indicates that the spread potential of S. halepense is to be considered primarily high (75%) and to a lesser extent medium (25%), because of the relatively high temperature (>10°C) required for successful germination [36], which currently impedes invasion of cooler areas up north or at higher elevations.

Impact assessment considers a multitude of negative impacts documented for S. halepense. Impacts on human health are considered ‘low’ according to the Catalogue of criteria [54]; even if its pollen was shown to prompt allergic reactions [80,81,82], in Canton Ticino this seems not to be a problem [79]. Furthermore, substances secreted by the plant are of concern only if ingested by livestock. In case of damage and stress (drought, frost, injury to tissue cells) the plant produces hydrocyanic acid from dhurrin, which is toxic to livestock [65,67,83,84]. During August 2022, the Piedmont region (N-Italy) experienced drought and tropical temperature conditions that resulted in the deaths of sixteen cows, intoxicated after ingesting S. halepense [85].

Impacts on native species and ecosystems primarily occur through competition and progressive displacement. Although its seeds typically germinate later due to higher temperature requirements [36], the species gains a rapid size advantage in just two to three weeks [87]. Once established, it is able to suppress the growth of native plants by outcompeting them for essential resources (light, water, and nutrients). This competitive interaction can lead to biomass reduction of up to 95% in native species compared to individuals grown without S. halepense [36]. Neighbouring species can also be inhibited by allelopathic substances [16], like dhurrin [88] and sorgoleone [89], which alter soil biochemistry [35,86,90], modify the soil bacterial composition and thus nutrient availability [35,90,91,92]. In addition, roots and rhizomes bear nitrogen-fixing bacteria that also modify the biogeochemical characteristics of soil [86]. However, soil analyses in invaded areas show diverging effects, with either an increased availability of all nutrients except calcium [86] or a reduced availability of nutrients [90]. Results from our analyses appear to be consistent with the latter case, as sites with the most developed individuals and the highest S. halepense cover showed a significantly lower amount of nutrients. Impacts at the species level and on soil biochemistry are reflected at the habitat level. In the United States, species richness of semi-natural habitats is significantly lower when S. halepense is present [35]. About 80% of the prairie areas in the country are potentially colonizable by the species and, where already present, it spreads about half a meter per year, creating transitional environments at first and stably invaded habitats later, with a mean cover of 77% [29,30,31,32,33,34,35]. In our study, S. halepense mean cover was 53% in meadows/roadsides habitats, and species richness was lower in the plots where the species showed the most pronounced functional traits, although no direct comparisons were made with non-invaded sites. Overall, impacts on native species and ecosystems in the Insubric region appear as relevant as highlighted for other invasive species occupying ruderal habitats (i.e., Senecio inaequidens; [108]); however, the studied species is still restricted primarily to anthropogenic habitats and secondly to semi-natural habitats. According to the EICAT [58], such a decline in the number of individuals within native communities corresponds to a ‘Moderate’ (MO) impact category (see Table 3).

Impacts of S. halepense on agricultural systems are the most studied, due to the massive economic damage through crop losses (e.g., [16,26,29]) and crop quality reduction [16]. Corn and soybean (followed by cotton [93]) are the most affected crops, with crop loss ranging between 57 and 88% [28] and between 59 and 88% [27], respectively. In Austria, corn and seed pumpkin crop loss reached an average of 40% in 2013 [23]. In terms of economic loss, 300 million dollars per year were estimated to be lost in soybean production alone in Argentina [19] and about 165 million euros per year were spent on the control of the species in cotton production in Türkiye [109]. In the past, in the United States losses were so high that in some cases farmers completely abandoned the crops [65]. According to the SEICAT [59], severe crop loss leading to the abandonment of the cultivated field corresponds to a ‘Major’ (MR) impact category (see Table 3) for S. halepense. Impacts on crop fields can be exacerbated by the fact that the species can host and be a vector for numerous agricultural pathogens [16], such as insects, nematodes and especially fungi and viruses, which can cause additional damage to crop production (see full list in the Supplementary Material S4). In the Insubric region, the impact on agriculture appears significant, particularly in the Italian regions, where the species has been observed to establish dense monospecific populations along the boundaries of cultivated fields, subsequently spreading within these environments. However, agricultural practices are different between Switzerland and Italy, meaning that risk of invasion and negative impacts may differ as well. In particular, crop rotation is widely implemented in Switzerland and less in Italy, and has shown to be in some cases effective against S. halepense proliferation and to prevent pathogens outbreaks [11].

Finally, by combining spread and impact potential, the invasion potential (risk) of S. halepense can be considered medium-high in the Insubric region, as the species is quite phenotypically plastic, able to potentially affect a wide range of communities by forming dense stands and altering soil properties. To compare with other invasive species, medium-high means that the risk is not as high as fast transformers like the highly invasive liana Pueraria lobata or some invasive aquatic aliens, like Lagarosiphon major [110,111,112]. The potential invasion risk in agriculture appears higher than that on native biodiversity in the Insubric region. Given the projected future expansion of the species’ potential suitable habitats [25], anthropogenic climate warming is expected to facilitate its invasion into currently too cold and thus unsuitable areas of the Insubric region, as already documented elsewhere in Central Europe or N-America [20,22,23,24,113]. Furthermore, phenotypic plasticity may enable the species to spread from anthropogenic to natural habitats. Unfertilized and extensively managed meadows and dry meadows in sub-montane areas both host a higher-value biodiversity that may risk invasion by S. halepense.

4.4 Study Limitations

We acknowledge that our study can have some limitations associated with the limited number of sampled occurrences. Specifically, although we are confident that the analyses included in the results section (i.e., CCA) reflect field observations and actual ecological gradients, we suggest that these results be considered with caution. In addition, the analyses on biodiversity trends would have ideally followed a treatment–control approach comparing habitats with and without the presence of the species. However, we underline that across the habitats considered in this study, the species was always detected, which prevented the implementation of such a design.

5 Conclusions

In conclusion, the eradication of S. halepense in the Insubric region appears unrealistic (partially due to its archaeophyte status in the Italian flora). A strategy should be implemented to minimize the impact by containing its spread and reducing population density. A fundamental goal is prevention: avoiding the spread of S. halepense into uninvaded areas. This means identifying and tackling relevant dispersal mechanisms, which probably will result in preventive measures like avoiding the displacement of contaminated soil, careful cleaning of contaminated machinery and vehicles, and repeated mowing before flowering along roadsides and railways to minimize seed dispersal by turbulence. In any case, we call for urgent action in the Insubric region, as global warming will keep favouring the rapid spread of the species north and upwards.