The influence of aridity on plant intraspecific chemical diversity supports adaptive differentiation and convergent evolution

Abstract

Plants synthesize a broad array of specialized chemical compounds that mediate their interactions with the surrounding environment. Some of this chemical diversity is functional and subject to natural selection, but the factors underlying chemical evolution at the intraspecific level remain largely unknown. Here, we combined chemical, environmental and genetic data to investigate the effect of aridity on the expression of chemotypes in the invasive shrub Senecio pterophorus. We studied the variation in pyrrolizidine alkaloids (PAs), a group of specialized metabolites widespread across the families Boraginaceae, Asteraceae and Fabaceae, from native populations spanning a cline of aridity and from three cross-continental introductions, under natural and common garden conditions. We examined whether the relationship between chemistry and aridity was compatible with a process of adaptive differentiation using a method that partitions the variance and covariance by controlling for the population neutral genetic structure. We found a consistent shift from retrorsine-like to seneciphylline-like compounds under increasing aridity in both natural and controlled conditions in coherence with the biosynthetic pathways. This pattern was independent of the neutral genetic structure and occurred along the environmental gradient in the native range and in a convergent manner in all nonnative regions, which suggests adaptive differentiation in response to aridity. Our findings show that the diversity of PAs in S. pterophorus has been partially shaped by aridity. Investigating how abiotic factors influence chemical evolution is key to elucidating the plant responses in future climate scenarios and the cascading effects on other trophic levels.

Keywords: Adaptive differentiation; Senecio pterophorus; aridity; evolutionary ecology; invasive species; neutral markers; plant–climate interactions; pyrrolizidine alkaloids; specialized metabolites.

Fig. 1

Structures of the main pyrrolizidine alkaloids (PAs) in Senecio pterophorus from the native and three invasive regions, and their biosynthetic relationships as known from the literature. Abundant PAs are shown in bold. Minor compounds also derived from senecionine (jacobine, jaconine, riddelliine, and unknown retronecine PAs) are omitted. Homospermidine, the first specific intermediate of PA biosynthesis, originates two necine base precursors leading to different pathways: isoretronecanol, which initiates the biosynthesis of 1,2‐saturated PA, such as rosmarinine, platyphylline, and their isomers (Hartmann & Ober ; Langel et al. 2011), and trachelanthamidine, which generates senecionine, the backbone structure for most 1,2‐unsaturated PA. Senecionine is diversified into a bouquet of retronecine PAs in one‐step or two‐step reactions, including retrorsine, seneciphylline, and their isomers (Hartmann & Dierich ; Hartmann & Ober 2000). The black circles show the two divergent steps in PA biosynthesis. The chromatographic and mass spectral properties of all PAs are in Table S2.

Fig. 2

(a) Relative concentrations of PAs in Senecio pterophorus from foliage sampled in the field, averaged within chemotype (mean ± SE). SC = senecionine; IG = integerrimine; SP = seneciphylline; ST = spartioidine; HSP = hydroxyseneciphylline; ASP = acetylseneciphylline; AST = acetylspartioidine; SV = senecivernine; RT = retrorsine; US = usaramine; PLI = platyphylline; RO = rosmarinine; ROII = rosmarinine‐II; ROIII = rosmarinine‐III; ROIV = rosmarinine‐IV. Chemotypes were named by the most abundant PA. (b) A simplified diagram of the main biosynthetic pathways (see Fig. 1) for each chemotype. (c) Geographic distribution of chemotypes in the native range (populations S06–S19), the expanded range in the Western Cape of South Africa (populations S01–S05), and two cross‐continental introductions: Australia (populations A01–A12) and Europe (populations C01–C12). Sector graphs display chemotype percentage in a population.

Fig. 3

(a) Score graph of the PCA for the relative abundance of leaf pyrrolizidine alkaloids (PAs) in 421 individuals sampled at the native range in eastern South Africa, the expanded range in western South Africa, and two invasive ranges (Australia and Europe). (b) Loadings of individual PAs: SC = senecionine; IG = integerrimine; SP = seneciphylline; ST = spartioidine; HSP = hydroxyseneciphylline; ASP = acetylseneciphylline; AST = acetylspartioidine; SV = senecivernine; RT = retrorsine; US = usaramine; PLI = platyphylline; RO = rosmarinine; ROII = rosmarinine‐II; ROIII = rosmarinine‐III; ROIV = rosmarinine‐IV. (c) Relationship between aridity (P/PET; precipitation over potential evapotranspiration) and the first two principal components of the PCA. The depicted line corresponds to a significant correlation as obtained by a linear model. (d) Relationship between the intensity of herbivory and the first two principal components of the PCA.

Fig. 4

Relation between aridity index and relative abundance of leaf PAs from the native range in South Africa. The aridity index is the ratio of summer precipitation to potential evapotranspiration (P/PET), with lower P/PET values indicating higher drought. Each dot represents a population average. The lines show significant correlations from mixed linear models (Table S3).

Fig. 5

Correlations between chemical, genetic and climate distances, including all native and non‐native populations. Each dot represents the distance for a population pair. The chemical distance was calculated using (a and b) all compounds, (c and d) the RO group, (e and f) the RT group, and (g and h) SP group measured in individuals growing in a common garden. Neutral genetic distances were obtained by neutral markers, and climatic distances were calculated using the Euclidean distance in P/PET (precipitation over potential evapotranspiration) from the original locations. The correlation coefficient (r) was determined using a Mantel test. Level of significance **P < 0.01.

Fig. 6

Variance and covariance of the chemical composition in relation to aridity (estimated by the P/PET, where lower P/PET indicates higher aridity) independent of neutral genetic variation. Analyses include individuals from the native and three nonnative regions grown in common garden (a and b) and in natural conditions (c and d). The left column shows correlations across chemical groups and aridity independent of the neutral genetic structure. The right column shows the amount of variance in each chemical group independent of the neutral genetic structure, depicted in colour as a sector graph, along with the correlations among chemical groups that are explained by aridity (line). Only correlations whose absolute value >0.1 are shown, with positive correlations as green solid lines and negative correlations as red dashed lines. PAs are clustered by their biosynthetic relationship: RO group (rosmarinine and isomers), SP group (seneciphylline and related compounds) and RT Group (retrorsine and related compounds).