Defense of pyrethrum flowers: repelling herbivores and recruiting carnivores by producing aphid alarm pheromone

Abstract

(E)-β-Farnesene (EβF) is the predominant constituent of the alarm pheromone of most aphid pest species. Moreover, natural enemies of aphids use EβF to locate their aphid prey. Some plant species emit EβF, potentially as a defense against aphids, but field demonstrations are lacking. Here, we present field and laboratory studies of flower defense showing that ladybird beetles are predominantly attracted to young stage-2 pyrethrum flowers that emitted the highest and purest levels of EβF. By contrast, aphids were repelled by EβF emitted by S2 pyrethrum flowers. Although peach aphids can adapt to pyrethrum plants in the laboratory, aphids were not recorded in the field. Pyrethrum's (E)-β-farnesene synthase (EbFS) gene is strongly expressed in inner cortex tissue surrounding the vascular system of the aphid-preferred flower receptacle and peduncle, leading to elongated cells filled with EβF. Aphids that probe these tissues during settlement encounter and ingest plant EβF, as evidenced by the release in honeydew. These EβF concentrations in honeydew induce aphid alarm responses, suggesting an extra layer of this defense. Collectively, our data elucidate a defensive mimicry in pyrethrum flowers: the developmentally regulated and tissue-specific EβF accumulation and emission both prevents attack by aphids and recruits aphid predators as bodyguards.

Keywords: (E)-β-farnesene synthase; (E)-β-farnesene; aphid honeydew; cortex-specific expression; false alarm.

Figure 1

Insect distribution on field pyrethrum plants. (a) Relative abundance of different insect species visually recorded on pyrethrum plants during weeks 1–3 (week 1 represents early flowering stages S1–3; week 2 represents stages S2–4, and week 3 stages S4–5). Thrips were scored separately (Fig. 1b). Data are represented as mean ± SEM. The total number of individuals per week is 532, 393 and 2967 for weeks 1, 2 and 3, respectively. (b) Distribution of coccinellid beetles (mainly Coccinella septempunctata) and thrips (mainly Frankliniella occidentalis) across the different tissues and flower stages. Scored plant parts for beetles were leaves (L), peduncles (P) and flower stages 1–5 (S1–S5) in two fields scored in weeks 1 and 2. Beetle densities were 76 and 19 per 100 plants in weeks 1 and 2, respectively. Beetle density in week 3 was too low and is not shown here (two per 100 plants). Thrips were scored in week 3 with a highest density of eight thrips per flower on S3 flowers (Yang et al., 2012). Inset: ladybird beetles on an S2 flower head. Data are represented as mean ± SEM. (c) Distribution of the number of coccinellid beetles found per plant. The distribution recorded in week 1 is statistically significant compared to the calculated random distribution of the same number of insects (t‐test: *, P < 0.05; **, P < 0.01). All data in (a)–(c) are based on the assessment of six pools of 100 plants except for thrips (three pools). Data are represented as mean ± SEM.

Figure 2

EβF emission and attraction of ladybird beetles. (a) Images of different pyrethrum flowering stages: Stage 0 (S0, younger bud), Stage 1 (S1, bud), Stage 2 (S2, ray flowers half open), Stage 3 (S3, first row of disk flowers open), Stage 4 (S4, half rows open), Stage 5 (S5, all rows open), Stage 6 (S6, overblown). (b) Response of ladybird beetles (Coccinella septempunctata) to volatile compounds against an air blank. Y‐tube olfactometer choice assays were performed with S2 flowers, different dilutions of pure compounds or a blend of (Z)‐3‐hexen‐1‐ol (HO), (Z)‐3‐hexenyl acetate (HA), 6‐methyl‐5‐hepten‐2‐one (MHO) and (E)‐β‐farnesene (EβF) (χ² test; **, P < 0.01; ***, P < 0.001; ns, not significant). (c) Quantitated EβF emission during plant development (pink bars) and in response to artificial damage (S2–4 stage, inset). On the right axis, peak area ratios of EβF and germacrene D (GD). Data are represented as means ± SE (n = 4–5). (d) Y‐tube olfactometer choices of ladybird beetles in response to pyrethrum volatiles at different developmental stages or upon wounding. At the right, the number of responsive/total beetles (χ² test, *, P < 0.05; **, P < 0.01; ns, not significant).

Figure 3

Localization analysis of TcEbFS gene expression and (E)‐β‐farnesene (EβF) in pyrethrum. (a) GC‐MS single ion traces (m/z 93) (left) of products formed by incubation of recombinant TcEbFS1 or TcEbFS2 protein with farnesyl diphosphate (FPP). Mass spectrum and retention index are compared to an authentic standard of (E)‐β‐farnesene. (b) Gene expression of TcEbFS in different organs and flower developmental stages. Floral parts of the S3 flower stage. Results of qRT‐PCR were normalized for TcGAPDH. Data are represented as means ± SE (n = 3). (c) Left: ladybird beetle searching on a flower bud of pyrethrum. Right: aphid infestation of a bud of shasta daisy (Leucanthemum × superbum). (d) Expression of TcEbFS‐GUS promoter fusions in transgenic chrysanthemum. d1, a whole leaf; d2, a longitudinal section of a leaf vascular bundle (vb); d3 and d4, cross‐sections of a shoot. (e) Pyrethrum sections showing the specific location of terpene products and callose by purple staining with, respectively, NADI and aniline blue reagents. e1, a NADI‐stained cross‐section of flower peduncle; e2, a NADI‐stained cross‐section of upper flower peduncle; e3, a NADI‐stained cross‐section of lower flower peduncle; e4, brightfield image with white arrows indicating stained oil accumulations inside inner cortex cells; e5, fluorescence image of the same section as e4 with red arrows indicating callose of some phloem sieve plates (Se). Lignified walls show autofluorescence. e6, a longitudinal section of double‐stained material (fluorescence image) showing the vascular bundle (vb) with stained callose bands of sieve plates and alongside in elongated parenchyma cells of inner cortex trains of stained EβF oil. Co, cortex; phs, phloem sclerenchyma; ph, phloem; Se, sieve elements; xy, xylem; xys, xylem sclerenchyma.

Figure 4

Flower‐produced and synthetic (E)‐β‐farnesene (EβF) cause alarm responses in Myzus persicae aphids at two levels. (a) Alarm responses to S2 pyrethrum flower odor or EβF standard (nearly 100 ng) after prerearing under different conditions. Data are represented as mean ± SEM (n = 6–8). Arcsin‐transformed data were subjected to ANOVA followed by Duncan's multiple range test (**, P < 0.01; ***, P < 0.001). (b) Alarm response at different exposure times to artificial honeydew (25% sucrose in water) and hexane solutions with or without 10 ng μl⁻¹ EβF. Data and statistics are the same as in (a). (c) Representative total ion chromatograms of hexane extracts of pooled aphid honeydew droplets and S1 flower buds and peduncles relative to an EβF standard. Aphid honeydew‐1 is a negative sample (consisting of 50 droplets in 50 μl hexane). Aphid honeydew‐3 is a positive sample (consisting of 50 droplets in 50 μl hexane) with EβF estimated at 0.12 ng μl⁻¹ (Table 2). EβF standard was 0.1 ng μl⁻¹. Internal standard (IS) was carvone.