Geographic patterns of seed trait variation in an invasive species: how much can close populations differ?

Abstract

Seeds play a major role in plant species persistence and expansion, and therefore they are essential when modeling species dynamics. However, homogeneity in seed traits is generally assumed, underestimating intraspecific trait variability across the geographic space, which might bias species success models. The aim of this study was to evaluate the existence and consequences of interpopulation variability in seed traits of the invasive species Carpobrotus edulis at different geographical scales. We measured seed production, morphology, vigour and longevity of nine populations of C. edulis along the Catalan coast (NE Spain) from three differentiated zones with a human presence gradient. Geographic distances between populations were contrasted against individual and multivariate trait distances to explore trait variation along the territory, evaluating the role of bioclimatic variables and human density of the different zones. The analysis revealed high interpopulation variability that was not explained by geographic distance, as regardless of the little distance between some populations (< 0.5 km), significant differences were found in several seed traits. Seed production, germination, and persistence traits showed the strongest spatial variability up to 6000% of percent trait variability between populations, leading to differentiated C. edulis soil seed bank dynamics at small distances, which may demand differentiated strategies for a cost-effective species management. Seed trait variability was influenced by human density but also bioclimatic conditions, suggesting a potential impact of increased anthropogenic pressure and climate shifts. Geographic interpopulation trait variation should be included in ecological models and will be important for assessing species responses to environmental heterogeneity and change.

Keywords: Bioclimatic variables; Carpobrotus edulis; Geographic distance; Longevity; Mediterranean.

Fig. 1

a Relative location of the nine studied populations (filled circles) of C. edulis distributed in three differentiated zones: Maresme (M), Costa Brava (CB) and Cap de Creus (CA). b Results of multidimensional scaling analysis (MDS) evaluating differences in nine seed traits among studied populations. Traits indicated in grey have significant (P < 0.01) contribution population variability. Ellipses represent 95% of confidence intervals. P-values correspond to PERMANOVA results for Zone and Population (nested in Zone) factors

Fig. 2

Trait values (mean ± SE, n = 5) for the nine relevant seed traits in all nine C. edulis populations from three distant zones: Cap de Creus (CC), Costa Brava (CB) and Maresme (M). Different capital letters reflect differences between zones, whereas different lowercase letters reflect significant differences between populations

Fig. 3

a Relative variance decomposition at the population and zone-levels for the different traits. The 33 and 66% thresholds are given by the dashed lines for the nine selected seed traits of C. edulis. b Boxplot of percent trait variation for the nine selected seed traits of C. edulis between diferent populations. c Acronyms, complete trait names, and Mantel Test results for the correlation between geographic variation with trait variation. NS non-significant, R_M mantel statistic

Fig. 4

a Comparative of the zonal variability estimated as the distances to zonal centroid for the different populations at the zones: Cap de Creus (CA), Costa Brava (CB) and Maresme (M) using the nine selected seed traits. Different letters reflect significant differences between zones. b Human density and bioclimatic variables correlations with zonal variability (centroid distance) of the nine seed traits of C. edulis among the different populations. Coloured correlations are significant and blue to red palette represents the R² coefficient following the colour key. Complete bioclimatic variables names can be found in http://www.worldclim.org/bioclim