Introduction

Understanding why plant species that are often inconspicuous in their native range become successful invaders after introduction to non-native ranges has long been the focus of invasion ecology. Species that successfully naturalize and dominate plant communities in invaded ecosystems are thought to have broad ecological amplitudes or physiological tolerances to multiple environmental stresses (Castillo et al. 2014; Bajwa et al. 2016). Intraspecific variation in functional traits between native and introduced populations has long been thought to underlie invasiveness, rapid colonization, and spread (Rejmánek et al. 2013). Acclimation via phenotypic trait plasticity and population genetic differentiation are two mechanistic pathways that may confer fitness advantages under environmental variation (Richards et al. 2006; Drenovsky et al. 2012). Furthermore, variation of functional traits within plant populations plays a key role in sha** the ecological impacts of genetic diversity and serves as a crucial predictor of plant performance (Hughes 2014). Therefore, evaluation of divergence in phenology and functional traits of plants between native and introduced ranges provides an opportunity to better understand contemporary evolution of species in changing environments. Climate change and biological invasions may act as synergistic processes provoking negative ecological and economic impacts (Hoover et al. 2012; Giejsztowt et al. 2020; Vilà et al. 2021). Improved knowledge of the role of evolutionary processes on functional trait-based responses of invasive plants is crucial for making decisions about management of introduced species in a scenario of globally changing environments (Kelly and Goulden 2008; Nicotra et al. 2010). Knowledge of trait-based responses of invasive species supporting population-level adaptation to global climate change is a prerequisite for predicting future invasions (Kirwan et al 2009) and changes in species’ range distributions (Valladares et al. 2014).

Understanding how changing environmental conditions and genetic factors contribute to the success of invasive plant species in their introduced range can be explored through common garden experiments (CGE) in which multiple native and non-native genotypes are grown under identical environmental conditions to determine the genetic component of the phenotypic variation recorded in field conditions (Colautti et al. 2009). CGE have revealed phenotypic plasticity, genetic differentiation, and rapid evolution of key functional traits in introduced plant species that have become invasive along broad environmental gradients (e.g. Castillo et al. 2005, 2014; Leger and Rice 2007; Molina-Montenegro et al. 2013, 2018). In addition, the combination of CGE and genetic analyses may offer a comprehensive understanding of invasion processes in changing environments. For example, this approach may inform us about the effects of genetic bottlenecks and admixture on the establishment success and spread of invasive species (Bajwa et al. 2016). Rapid evolution in invasive species' responses to environmental changes may inform us about the potential of biodiversity to confront global change, particularly climate change (Moran and Alexander 2014). Invaders often exhibit a high degree of phenotypic variability but it is not clear if this is a product of high population genetic diversity or high phenotypic plasticity (Hovick et al. 2023).

Our study system included Iris pseudacorus L. (yellow-flag iris) populations within tidal wetland plant communities in two estuaries under Mediterranean climate. Study populations in the Guadalquivir River Estuary (Southwest Iberian Peninsula) were within the native geographic range of I. pseudacorus, while introduced invasive populations were from California’s San Francisco Bay-Delta Estuary. Gaskin et al. (2016) recorded high levels of gene flow and genetic diversity within and among populations of I. pseudacorus introduced in the Pacific Northwest region of the United States, supporting the importance of sexual reproduction as the primary reproductive mode for this species. Previous research in our study system has shown that I. pseudacorus coexists with a high degree of plant community diversity in its native range, whereas it is associated with reduced species richness, evenness and diversity in its introduced range (Gallego-Tévar et al. 2022). In addition, an intercontinental observational field study found invasive I. pseudacorus in the introduced range had greater capacity than native plants to adjust to environmental stresses induced by rising sea level such as increased salinity and inundation levels (Grewell et al. 2023). While genetic differentiation could play a major role underlying these observed intraspecific differences, the role of evolutionary processes was unknown. In this study, we evaluated intraspecific phenotypic variation of I. pseudacorus within and among native and introduced study populations through analysis of 25 functional traits using a CGE under benign freshwater conditions. Recorded functional traits were relevant to estuarine salinity and inundation stress responses in estuarine field conditions and included growth, biomass production, morphological, biochemical and physiological response traits. Next, we compared these results to an evaluation of plant traits from populations in the field study sites, and key variable environmental conditions within each study population. In addition, we performed genetic analyses of field-collected samples of I. pseudacorus in our study populations. Analyzing the phenotypic variation of many plant functional traits between and within populations of I. pseudacorus in contrasted and common environments, together with the assessment of genetic diversity, allowed us to reveal mechanisms enabling this species to withstand changing environmental conditions along estuarine gradients. Our overall objective was to assess whether changes in functional plant trait were dependent on genetic differentiation or phenotypic plasticity. We hypothesized that (1) Iris pseudacorus would show high genetic diversity, representing a broad range of phenotypic differences between geographic ranges; (2) native populations would show higher phenotypic variation among populations than introduced populations since we expect them to exhibit greater genetic diversity related to their longer opportunities for evolution and adaptation over time, and also due to the absence of population bottlenecks common in introduced populations; and (3) genetic and phenotypic differences would increase with geographic distance due to isolation-by-distance processes.

Materials and methods

Common garden experiment

We collected seeds from randomly selected flowering stems in 7–8 Iris pseudacorus patches per population in four populations spaced across an estuarine salinity and inundation gradient in the native range (Guadalquivir River Estuary, Andalusia, Spain) (Fig. S1A 1–4), and four populations along a comparable estuarine gradient in the invaded range (San Francisco Bay-Delta Estuary, California, USA) (Fig. S1C1-4) in summer 2019. Seeds were stored in dry, dark conditions at 5 °C. Seeds were germinated while floating in freshwater, then 48 experimental plants (n = 6 plants per population × 4 populations × 2 ranges) were sown individually in plastic pots (20 cm diameter × 18 cm height) filled with peat and sand substrate (50/50), randomly arranged in a common garden and grown for 12 months in a glasshouse at the University of Seville. The pot bases were kept permanently flooded to a height of 2 cm with 20% Hoagland solution. Common controlled environmental conditions included ambient sunlight, substrate redox potential of 210 ± 7 mV, electrical conductivity 0.6 ± 0.1 mS cm−1, air temperature 25 ± 2 ºC and relative air humidity 60 ± 3%.

After 12 months of pre-reproductive growth, we evaluated 25 growth, biomass allocation, morphological and biochemical plant traits (Supplementary Data File, Plant traits tab). Traits included were live and senescent leaf biomass (g), above-ground biomass (AGB; including live and senescent leaf biomass) (g), crown, rhizome and root biomass (g), below-ground biomass (BGB; including crown, rhizome and root biomass) (g), root mass ratio (RMR), leaf mass ratio (LMR), leaf width and length from 5 randomly chosen adult leaves per plant (cm), specific leaf area (SLA) (cm2 g−1), total leaf area per plant (cm2), number of leaf fans per plant, number of live and senescent leaves per plant and per leaf fan, apical leaf elongation ratio (cm day−1), rhizome total non-structural carbohydrates (TNC) concentration, foliar C and N concentration (%) and foliar C:N ratio. To record dry weight (DW) biomass, every plant was divided into live and senescent leaves, crown, rhizomes and roots. Substrate that adhered to below-ground biomass fractions was carefully washed away. DW was obtained after drying samples in a forced-air stove at 60 ºC for 48 h. SLA was calculated by dividing the leaf area by the dry mass from 3 randomly chosen adult leaves per plant (Garnier et al. 2001). Total leaf area was calculated as the product between SLA and live leaf biomass. Apical leaf elongation was measured by marking 3 apical leaves per plant with permanent sealant at their lower contact with adult leaves and measuring the distance from the sealant to the original marking point 2 days later. Total C and N concentration of introduced and native plants leaf tissue were analyzed using a Perkin Elmer 2400 CHNS/O analyser (Perkin Elmer, Waltham, MA, USA) and a LECO TruSpec Micro CHN/CHNS/O analyser (LECO Corporation, Geleen, the Netherlands), respectively. Rhizome samples were dried and ground to pass through a 40-mesh sieve. TNC concentration in rhizome samples was analyzed with a colorimetric assay of reducing sugars following ethanol extraction (Chow and Landhäusser 2004) and enzymatic digestion of the starch residue (Quentin et al. 2015).

Field data

Two newly expanded, non-diseased and non-damaged leaves were collected from each of 20 randomly selected I. pseudacorus plants spaced > 5 m apart in each study population following Gaskin et al. (2016). GPS coordinates were recorded for each sample. Leaves were blotted dry and immediately placed in plastic tubes or bags partially filled with silica beads, labeled, and stored at room temperature prior to genetic analysis.

Detail of plant traits and environmental conditions in the study populations were reported previously by Grewell et al. (2023). Plant traits (numbers of live and senescent leaves per fan, apical leaf elongation, leaf width and length, SLA, rhizome TNC, foliar C and N concentrations and foliar C:N ratio) were recorded as reported above for plants located at the same field locations where the seeds were collected from for the CGE. For environmental conditions, eleven factors (soil bulk density (g cm−3), organic matter content (%), total C and N concentrations (mg g−1), pH, electrical conductivity (EC) as a measure of salinity, total cover of plant species other than I. pseudacorus (%), intertidal elevation (m), hydrologic distance (km), inundation period (%) and maximum inundation depth (m)) were documented (Supplementary Data File, Environment tab). Many of these environmental factors, such as organic matter content, salinity and inundation, are simultaneously changing in tidal wetlands in the present scenario of climate change and sea level rise (Luo et al. 2019).

Genetic analyses

Genomic DNA was extracted from approximately 20 mg of silica-dried leaf material using a modified cetyltrimethylammonium bromide (CTAB) method (Hillis et al. 1996). Amplified fragment length polymorphism (AFLP) method followed Vos et al. (1995) was performed with modifications as in Gaskin and Kazmer (2009) and Gaskin et al. (2016) using two polymorphic primer pairs (MseI + CAT/EcoRI + ACC and MseI + CAC/EcoRI + ACT). We generated all AFLP data on an Applied Biosystems (ABI, Foster City, CA, USA) 3130 Genetic Analyser, and omitted any individuals that did not produce a typical electropherogram pattern (i.e., noise > 20 relative fluorescence units or failure to produce sufficient number of peaks). AFLP peaks (124 loci) were read in GeneMapper v6 (ABI) and corrected manually for presence or absence as 1/0 data in a spreadsheet (Supplementary Data File, AFLP tab).

Data analyses

Deviations were calculated as standard error of the mean (SEM). A significance level (α) of 0.05 was applied for every analysis. Plant traits were classified into three functional groups: (1) growth and biomass production and allocation (AGB, BGB, live leaf biomass, senescent live biomass, crown biomass, root biomass, rhizome biomass, RMR, LMR and apical leaf elongation rate), (2) morphological traits (SLA, number of fans, total number of live leaves, number of live leaves per fan, total number of senescent leaves, number of senescent leaves per fan, total number of leaves per fan, leave width and leave length, and total leaf area per plant), and (3) biochemical traits (rhizome TNC, leaf C concentration and leaf N concentration, leaf C:N ratio). Data series were tested for homoscedasticity using the Levene’s test, and for normality using the Shapiro–Wilk test. Three plant traits (number of live leaves per fan, total number of senescent leaves and number of senescent leaves per fan) were transformed using the function 1/x, and one trait (rhizome TNC) was transformed using the function ln(x) to meet the assumption of homogeneity of variances for parametric tests. To protect analyses from type I error, we used the protected analysis of variance (ANOVA) protocol (Scheiner 2001): The means of the dependent variables of each trait group were compared using multivariate analysis of variance (MANOVA) and Pillai's Trace to evaluate the significance of the factor distribution range (native and introduced). Redundant, highly correlated variables (r > 0.95) were identified prior to MANOVA analysis and omitted from the statistical models: AGB and BGB were omitted since they were highly correlated to live leaf and root biomass, respectively, the total number of leaves and the number of fans were omitted because they were highly correlated to the total number of live leaves, and foliar C:N ratio was omitted because it was highly correlated with foliar N concentration. Once multivariate significance was confirmed via MANOVA, the main univariate differences between distribution ranges were evaluated for each functional plant trait with General Lineal Models (LMs) and Bonferroni-Dunn's test as a post hoc analysis, using the seed mass as a covariate in the model (Ng and Cribbie 2017). These analyses were carried out using lme4 package of the R-software (Bates et al. 2015).

Principal Component Analyses (PCA) were performed to analyze the relationships between plant traits in each distribution range and to reduce plant variables, analyzing the correlation matrix with 25 maximum iterations for convergence without rotation to extract independent PCA factors with eigenvalues > 1. These analyses were conducted using the software SPSS 15.0 (SPSS Inc., Chicago, USA).

Inter-population trait variation index (%) was calculated for those plant traits recorded both in the CGE and the field in each distribution range of I. pseudacorus as the relation of the difference between the maximum (Xmax) and the minimum (Xmin) values of a given plant trait divided by its maximum (1) (Valladares et al. 2006).

$${\text{Inter - population trait variation}}\,\left( {\text{\% }} \right)\,{ = }\frac{{\left( {{\text{X}}_{{{\text{max}}}} - {\text{X}}_{{{\text{min}}}} } \right) \times {100}}}{{{\text{X}}_{{{\text{max}}}} }}$$
(1)

This index is satisfactory for comparison of plasticity sensu lato focused on the responses to variable environments since it allows traits with different variation ranges and/or expressed in different units to be compared (Valladares et al. 2005, 2006).

Mean intra-population trait variation index (%) was calculated for those plant traits recorded both in the CGE and the field in each distribution range as the arithmetic mean of the relation between the difference between the maximum (xmax) and the minimum (xmin) values divided by the maximum of a given plant trait for each of the four studied populations at each distribution range (2) (Castillo et al. 2018).

$${\text{Intra - population}}\,{\text{trait}}\,{\text{variation}}\, = \frac{{\mathop \sum \nolimits_{{{\text{i}} = 1}}^{4} \frac{{\left( {{\text{X}}_{{{\text{max}}}} - {\text{X}}_{{{\text{min}}}} } \right) \times 100}}{{{\text{X}}_{{{\text{max}}}} }}}}{4}$$
(2)

Relative intra-specific trait variation index (RITV) (%) in common garden conditions and in field conditions was determined for every plant trait after subtracting the intra-population from the inter-population trait variation (3).

$${\text{RITV}}\, = \,{\text{inter - population}}\,{\text{trait}}\,{\text{variation}} - {\text{intra - population}}\,{\text{trait}}\,{\text{variation}}$$
(3)

Then, we calculated the average of RITV values of every plant trait for the native and introduced ranges, named relative intra-specific phenotypic variation (RIPV). The calculation of RIPV followed that of the phenotypic plasticity index by Gallego-Tévar et al. (2018). Since environmental conditions were the same for every plant in the CGE, RIPV in these conditions was only related to the origin of the plants from different distribution ranges (native and introduced), reflecting genetic differences in expressing the variation of plant traits. In contrast, environmental conditions changed between populations in the field, so RIPV was integrating genetic differences together with phenotypic plasticity in response to different environments. The differences in inter- and intra-population trait variation indexes and RIPV between CGE and field conditions, geographical distribution ranges and their interaction were analyzed using LMs. Linear correlation (Pearson coefficient, r) and linear regression analyses were used to characterize the relationships between plant traits recorded in CGE and environmental factors recorded in the field in the same locations where seeds for the CGE were collected.

Resemblance of plant traits, genetic diversity and environmental conditions based on distances between field monitoring plots were calculated using Gower’s similarity index with coefficients ranging between 0 and 1 (Gower 1971) using the R-software package vegan (Oksanen et al. 2018) to obtain a pairwise dissimilarity matrixes. Matrixes were then subjected to Mantel’s test for Matrix Correspondence (Mantel 1967) to examine relationships among plant traits, genetic, environmental and geographic distances (hydrological distances). The proportion of genetic diversity among and within populations was calculated by molecular variance analysis (AMOVA). Genetic differentiation among populations (ΦPT), analogous to Wright’s F’ST for dominant data, was obtained following standardization (Meirmans 2006). These analyses were conducted using GenAlEx 6.502 software (Peakall and Smouse 2006).

Results

The geographical distribution range of I. pseudacorus (native vs introduced) had a significant effect on biomass production, allocation and growth, morphological and biochemical traits in the common garden experiment (MANOVA, P < 0.05; Table S1). Results of the LMs for all measured plant traits are summarized in Table S2. Herein we report significant results. Native plants showed lower values of SLA (− 34%), total number of live leaves (− 25%) and rhizome TNC (− 63%) than introduced plants from the introduced range. In contrast, the total number of leaves per fan was 34% higher in native than introduced plants, and leaf length tended also to be higher (+ 15%) in native plants in the common garden experiment (Table S2).

Six factors, which explained 87% of the variance, were obtained from the PCA for native I. pseudacorus plants. The first factor (PC1, explaining 36% of the variance) was positively correlated with AGB, crown and live leaf biomass, the numbers of total and live leaves, leaf area and rhizome TNC, and negatively with SLA and the number of senescent leaves per fan. The second factor (PC2, explaining 21% of the variance) was positively correlated with BGB, root biomass, the number of live and total leaves per fan and leaf width, and negatively with the number of fans (Table S3). Six factors, which explained 88% of the variance, were also obtained from the PCA for introduced plants. The first factor (PC1, explaining 37% of the variance) was positively correlated with AGB, BGB, root biomass, rhizome, crown and live leaves, leaf width, length and area, and foliar C:N ratio, and negatively with foliar N concentration. The second factor (PC2, explaining 18% of the variance) was positively correlated with the total number of leaves per fan, and negatively with the total number of leaves and fans per plant. PC3 (explaining 15% of the variance) was positively correlated with the number of senescent leaves per fan and per plant, and negatively with RMR (Table S3).

Inter- and intra-population variation indexes did not change between native and introduced plants, nor for CGE/field conditions, and their interaction (Fig. 1; Table 1). Moreover, inter- and intra-population variation indexes were correlated between each other in and between CGE and field conditions (Table S4). Geographic range was a significant factor explaining relative intraspecific trait variation (Table 1). RIPV was the same for CGE and field conditions, and its interaction with geographic distribution range, but it was 27% higher for native than introduced plants (Fig. 1; Table 1).

Fig. 1
figure 1

Inter- and intra-population trait variability indexes (black and white columns, respectively) and relative intra-specific trait variation index (grey columns) for plant traits of Iris pseudacorus recorded in a common garden experiment (CGE) for native plants grown from seeds collected in the native range (Guadalquivir River Estuary, Andalusia, Spain; black columns) and for introduced plants in the introduced range (San Francisco Bay–Delta Estuary, California,USA; white columns), and from trait evaluations of populations in the field. Values are means ± SEM (n = 10)

Full size image
Table 1 General Linear Models with growing conditions (common garden (CG) and field), distribution range (native and introduced) and their interaction as fixed factors for inter- and intra-trait variation indexes and relative intra-specific trait variation index of Iris pseudacorus
Full size table

Relationships between plant traits in the CGE and field environmental conditions

Senescent leaf biomass increased and leaf elongation rate decreased for both native and introduced plants of I. pseudacorus established in the CGE from seeds coming from field locations with higher sediment EC (Fig. 2A, B). This increase in senescent leaf biomass with sediment EC was due to the high values recorded for the native population located closest to the Guadalquivir River confluence with the Gulf of Cadiz, Atlantic Ocean. Native and introduced plants showing higher SLA in the CGE came from seeds collected at locations with lower sediment pH and absolute plant cover in the community (Fig. 2C, D). The number of total leaves per fan in the CGE was higher for plants coming from habitats with higher sediment pH. Leaf length in the CGE was higher for plants sourced from habitat with higher absolute cover in the plant community, which was especially high in some native study populations (Fig. 2E). Plants showing higher rhizome TNC in the CGE were sourced from habitats in lower intertidal elevations in the introduced geographic range that were exposed to higher maximum inundation depths in the field (Fig. 2F; Table S5).

Fig. 2
figure 2

Relationships between plant traits of Iris pseudacorus recorded in a common garden experiment (CGE) and environmental factors recorded in the field at native (circles) and introduced (triangle) locations where seeds for the CGE were collected

Full size image

Genetic analyses and relationships with plant traits and environmental factors

Maximum genetic distance between populations was 30% higher in the native than in the introduced range of I. pseudacorus. Genetic distance increased together with phenotypic distance in the field and with hydrologic distance in both native and introduced ranges. Phenotypic distance in the field also increased together with hydrologic distance in both distribution ranges. In addition, genetic distance increased together with phenotypic distance in the CGE in the introduced but not in the native range (Fig. 3). Intra-population genetic variance was greater than inter-population genetic variance in both native (89% and 11%, respectively) and introduced ranges (92% and 8%, respectively).

Fig. 3
figure 3

Results of the Mantel’s test between genetic, phenotypic and hydrological distances in field and common garden (CG) conditions for native (A, C, E) and introduced (B, D, F) range of Iris pseudacorus. Regression equations: A, y = 1.887x + 0.2435 (R2 = 0.085, P = 0.010); B, y = 1.5203x + 0.2420 (R2 = 0.027, P = 0.020); D, y = 1.2375x + 0.2762 (R2 = 0.0175, P = 0.030); E, y = 0.0004x + 0.0072 (R2 = 0.314, P = 0.010); F, y = 0.0003x + 0.0097 (R.2 = 0.3908, P = 0.010)

Full size image

Discussion

Our results show that I. pseudacorus presents genetic differentiation between native and introduced populations for four key functional plant traits: SLA, rhizome TNC and total live leaves per plant and per fan, since the inter-geographic differences recorded in the field were kept in common garden conditions. In addition, the number of senescent leaves per fan tended to be higher for native than introduced plants in both field and common garden conditions. These differences clearly configure two geographically segregated phenotypes of I. pseudacorus. Moreover, we identified the relationships between the recorded inter-range genetic differentiations with main environmental drivers of plant stress in tidal wetlands.

Introduced I. pseudacorus plants showed higher SLA and rhizome TNC than native plants in the field (Grewell et al. 2023) and these differences were also observed in the CGE. These results point to genetic differentiation, rather than phenotypic plasticity, as the mechanism explaining the recorded differences in select plant traits between geographical distribution ranges (Castillo et al. 2014, 2016; Grewell et al. 2016). This genetic differentiation was supported by a positive relationship between genetic distance in the field and phenotypic distance in the CGE for introduced plants, but not for native plants. The recorded high values of SLA and rhizome TNC for introduced plants suggest that resources may have been allocated to competition-related traits rather than herbivore defense traits (Grotkopp and Rejmánek 2007). Introduced plants with high rhizome TNC content have high levels of below-ground reserves that can support re-sprouting following winter dormancy. This trait contributes to early I. pseudacorus regrowth in the plant community, pre-empting marsh space before seasonal reemergence and growth of other dormant plant species. Furthermore, a high content of below-ground reserves enables introduced plants to withstand periods of deep inundation and also to rapidly allocate a large energy pool to sexual reproduction (Gallego-Tévar et al. 2019). Supporting this premise, introduced I. pseudacorus plants produce heavier seeds and more capsules per flowering stem than plants in the native range (Grewell et al. 2023). These results were supported by rhizome TNC in the CGE especially increasing in introduced plants sourced from populations occupying relatively lower marsh intertidal elevations where they were exposed to higher maximum inundation depths. Furthermore, SLA in the CGE was the lowest for native plants sourced from populations growing in fine-textured soils (high bulk density) with relatively low organic matter content (< 6%). Thicker leaves (low SLA) have been reported as a typical plant trait from low soil resource availability related to low organic matter content, especially under grazing since a greater investment in cell wall material can increase resistance against trampling (Laliberte et al. 2012). We have observed high herbivore pressure by livestock on native populations of I. pseudacorus, whereas vertebrate grazing has not been observed at our study sites in the naturalized range (Grewell et al. 2023). Previous work has also reported that livestock will eat young leaves of I. pseudocorus (Sutherland 1990). At the same time, SLA was the lowest for native plants sourced from habitats with very high (> 150%) absolute plant cover, and leaf length was the highest for those native and introduced plants showing the lowest SLA in the CGE. This may be interpreted as a way to maximize light interception in environments with low nutrient availability under high canopy cover of vegetation (Poorter and Nagel 2000). Producing more leaf area and thicker leaves given microsite resource limits could be achieved through clonal integration of resources; for example, clonal integration allows concentrating roots in high-nutrient soil patches and ramets in low light may benefit from connection to ramets in high light conditions (He et al. 2011). In this context, the relatively high differences in below-ground carbon reserves and foliar thickness recorded in native plants in the CGE coincided with differences in apical leaf elongation rate, AGB, root biomass and the number of leaf fans, whereas leaf length was the trait driving LMR in introduced plants.

Introduced plants presented fewer leaves per fan, with smaller differences within and between populations, and more leaves per plant because they had higher leaf fan densities than native plants in the CGE. The high number of leaves per plant may explain the higher iris cover recorded in introduced populations (79%) compared to native (74%) populations in the field (Grewell et al. 2023). High leaf fan densities would increase the competitive ability of introduced plants in comparison with native plants (Cheplick 1997; Yang et al. 2021).

The number of senescent leaves per fan was higher for native than introduced plants in the field (Grewell et al. 2023) and this trend was also in the CGE. In addition, senescent leaf biomass in the CGE increased together with sediment salinity (higher EC) in the field, a response of I. pseudacorus to salinity regimes confirmed in a greenhouse experiment (Grewell et al. 2021). Native populations coming from higher salinity habitats in the field had the highest senescent leaf biomass in the CGE, with smaller variations within and between populations than introduced populations. Salt tolerance may increase due to ion isolation in senescent leaf tissue (Reddy et al. 2017). The adaptation to salt stress in I. pseudacorus was also supported by the fact that leaf elongation rate in the CGE was lower for both native and introduced plants coming from saltier locations. This relationship likely reflects a transgenerational trade-off between salinity tolerance and growth (Berens et al. 2019), which may also be retained in freshwater conditions (Liu et al. 2019).

Genetic differentiation in introduced plants may result from different processes such as demographic bottlenecks, genotype filtering and/or mutations (Colautti et al. 2017). Vigorous outcrossing-based reproductive traits, as those recorded for introduced plants of I. pseudacorus in the field (Grewell et al. 2023), may counteract the detrimental effects of genetic bottlenecks often associated with species introductions, leading to high levels of genetic diversity (Gaskin et al. 2016). We recorded high genetic diversity coinciding with a wide range of phenotypic differences between geographic ranges, which supported our first hypothesis. Our results point to ongoing local adaptation processes both in native and introduced I. pseudacorus populations. Our results pointed to longer divergence processes among native than introduced populations, which was reflected in higher genetic diversity in the field and higher (+ 27%) phenotypic variation in the CGE for native compared to introduced plants. These results supported our second hypothesis which anticipated greater variability in native populations due to their longer evolutionary and adaptation history, and the lack of bottleneck events. Additionally, they shed light on the discussion about phenotypic variation in introduced and native populations in relation to the roles of phenotypic plasticity, and genetic and epigenetic variation, in invasions success (Fleming and Dibble 2015; Banerjee et al. 2019).

Our results suggest key functional traits of introduced I. pseudacorus have undergone rapid evolution, complementing previous CGEs revealing rapid evolution of introduced plant species that have become invasive along environmental gradients (Maron et al. 2004; Leger and Rice 2007; Dlugovich and Parker 2008). The genetic differentiation revealed in introduced populations in San Francisco Bay-Delta Estuary seems to be a product of rapid local adaptation since genetic distance in the field increased together with phenotypic distance in the CGE. This reflects invasion history as I. pseudacorus was detected in tidal wetlands of the inland Sacramento-San Joaquin Delta in the late 1960s (Light et al. 2005) and rapid range expansion downstream into brackish tidal wetlands in the greater San Francisco Bay-Delta Estuary recently occurred counter to expectations that the species was limited to freshwater (Gillard et al. 2021; Gallego-Tévar et al. 2022). Besides inter-range genetic differentiation, native and introduced plants presented higher percentage of intra- than inter-population genetic variability (native: + 78%; introduced: + 84%), pointing to high sexual reproduction rates within populations and high genetic exchange between populations, probably due to buoyant seeds quickly dispersing with tidewater flow in estuaries (Gillard et al. 2021). Interestingly, although absolute values of plant traits changed between field and CGE conditions, the relative amount of phenotypic variation (recorded as RIPV) was the same, independent of environmental conditions for both distribution ranges of I. pseudacorus. These constant values of RIPV in field and CGE conditions were due to the compensation of the changes in some plant traits by opposite changes in other plant traits as a result of allocation trade-offs (Reich et al. 2003), such as the values recorded between RMR and LMR. Additionally, the constant values of RIPV in field and CGE conditions were also related to proportional changes in inter- and intra-population variation. That is, those plant traits showing high inter-population trait variation also showed high intra-population trait variation in both field and in CGE conditions. These results point to high levels of transgenerational epigenetic and/or genetic regulation determining the phenotypic plasticity of different plant traits (Westerband et al. 2021). Consequently, two main phenotypes of I. pseudacorus were identified as a product of the above-mentioned genetic differentiations between geographic ranges. Introduced I. pseudacorus plants, in contrast to native plants, were characterized by rhizomes containing relatively large carbohydrate storage reserves and by high densities of leaf fans that each had relatively few live and senescent thin leaves, but with overall large densities of leaves per plant (Fig. 4).

Fig. 4
figure 4

Schematic illustrations showing the typical phenotype of (A) native Iris pseudacorus in the Guadalquivir River Estuary (Andalusia, Spain) and (B) introduced plants in the San Francisco Bay-Delta Estuary (California, USA)

Full size image

Besides genetic differentiation, some recorded inter-range phenotypic differences between native and introduced I. pseudacorus plants in the field (Grewell et al. 2023) were likely the result of phenotypic plasticity in response to contrasted environmental conditions, since these field differences were not expressed by plants in the CGE. Although environmental conditions appeared to primarily determine some plant traits, such as leaf width and foliar N concentration, a genetic component influencing them cannot be excluded, since optimum growth conditions in the CGE may not necessarily have provided conditions for different genotypes to express distinct phenotypes (Thompson 1991).

Invasive introduced I. pseudacorus plants are actively displacing native vegetation, reducing native plant diversity in the San Francisco Bay-Delta Estuary (Gallego-Tévar et al. 2022). Moreover, these introduced plants produce many viable seeds (Grewell et al. 2023) that have the buoyancy and reserves to disperse and establish in other locations along the Pacific Coast of North America (Gillard et al. 2021). In view of our results and previous supporting work (Grewell et al. 2021; Gillard et al. 2022), a management plan to control and eradicate introduced I. pseudacorus plants in the San Francisco Bay-Delta Estuary should be established urgently since they represent novel genotypes with key functional traits that can support invasiveness through high competitive ability and physiological stress tolerance responses to sea level rise.