• Changes in the defence compounds pyrrolizidine alkaloids (PAs) in roots and shoots of Senecio jacobaea are reported in response to nutrient addition in order to investigate whether changes in concentration are adaptive.
• PA concentrations were examined in leaves and roots of 40 vegetative ragwort plants, subjected to four nutrient treatments in a climate chamber study. Roots from 10 plants were subdivided into main root cortex, main root vascular cylinder, lateral roots and root tips and analysed for PA concentrations.
• Increasing nutrients lead to a significant reduction in total PA concentration of both roots and shoots. All individual PAs except jacobine decreased in concentration. The total amount of PA produced in the whole plant was not influenced by nutrient supply. Root tips contained a three times lower concentration than the main and lateral roots. The concentrations in the main root cortex were five times higher than concentrations in the vascular cylinder.
• Changes in biomass rather than changes in production rates can explain alterations in PA concentration of S. jacobaea in response to nutrients.
The maintenance of active secondary metabolite concentrations at a certain level may require the production of defence compounds to be very flexible. This should occur independently of the variation in environmental factors such as climate, light, humidity and nutrients. There are also inducible defence compounds, the levels of which are raised when necessary (Karban & Baldwin, 1997). However, our study focuses on the flexibility of a constitutive defence system in response to nutrient supplies. It is generally accepted that increasing nutrient supplies leads to an increased shoot : root ratio. If secondary metabolites production takes place throughout the whole plant, no consequences would be expected from increasing nutrient supply. However, when the production of secondary metabolites is coupled with root growth only, higher nutrient supplies would lead to lower concentrations in shoot and root. We sought to determine whether a plant species is capable of adapting defence levels to increased nutrient supply through different allocation patterns, increased production or both. A plant species was selected which produces secondary metabolites in the roots, allocates them over the whole plant and occurs under a wide range of nutrient conditions. Senecio spp. grow in poor sandy dune soils and at road sites, but are also found as noxious weeds in pastures, where they cause acute and chronic liver damage when ingested by cattle (Mattocks, 1986). The use of Senecio spp. in traditional human medicine could give rise to carcinogenic effects (Steenkamp et al., 2001). These effects are due to pyrrolizidine alkaloids (PAs), which are regarded as part of the defence mechanism of the plant against herbivores (Hartmann & Witte, 1995; Van Dam et al., 1995a). Recently, we showed that PAs influence soil borne fungi (Hol & Van Veen, 2002). PA production in Senecio spp. takes place in the root (Hartmann & Dierich, 1998) and is related to root growth (Hartmann & Toppel, 1987; Sander & Hartmann, 1989). Therefore it is important to understand the response of PA production and transformation in relation to growth biomass and variation in nutrient supplies.
Differences in the distribution of secondary metabolites over a plant can be considerable. Five-fold differences between different plant parts were observed in PA content of S. vulgaris (Hartmann & Zimmer, 1986) and Cynoglossum officinale (Van Dam et al., 1995b). Most data reported on this subject concern detailed studies of above-ground plant parts and it is unknown whether the same variation in secondary metabolite concentration will be found below-ground. This may have considerable effect on the defence of the root tissue against insects, fungi, nematodes and bacteria, which is essential for survival to the flowering stage. The older, basal root parts may also be used for storage of PAs. Optimal defence theory predicts that the level of defence in plant parts is related to the importance of that plant part for fitness (Rhoades, 1979). We therefor expect high concentration in the root tips, which are important for root growth.
Materials and Methods
Pyrrolizidine alkaloids (PAs) in Senecio spp. are present throughout development and under all circumstances. Highest concentrations were found in the flowerheads (Hartmann & Zimmer, 1986) of the two annual Senecio species S. vernalis and S. vulgaris. Vrieling et al. (1993) demonstrated that 50–100% of the total variation in total PA concentration was due to genetic differentiation under controlled growth conditions. Artificially damaged S. jacobaea plants did not increase PA concentrations as one might expect, but instead a significant decrease was found within 6–12 h after damage (Van Dam et al., 1993). PAs are transported via the phloem to the shoot (Hartmann et al., 1989), where they are transformed into other PAs (Hartmann & Dierich, 1998). Turnover of PAs is negligible (Sander & Hartmann, 1989; Hartmann & Dierich, 1998). It is unknown to what extent the allocation of PAs is dependent on the relative biomass above and below-ground.
A high PA producing genotype was selected from a cross between high and low PA plants of Senecio jacobaea (L.). Plants were propagated using tissue culture. At the start of the experiment, 40 plants were weighed and potted in 1.3 l sterilised dune sand. The experiment was carried out over a 16-h photoperiod and day/night temperatures of 20°C/15°C. Relative humidity was 70%. Every plant received the same amount of water three times a week. One week after planting nutrients were given as Steiner nutrient solution (Steiner, 1968) (macronutrients: N 167 mg l−1, P 31 mg l−1, K 282 mg l−1, S 111 mg l−1, Ca 180 mg l−1, Mg 49 mg l−1). The control treatment of 10 plants did not receive nutrients (N0) but the same volume of water, the other treatment groups (N1, N2, N3), also consisting of 10 plant each, received 20 ml Steiner once, twice or three times a week. N1 received twice and N2 once 20 ml water.
Plants were harvested after 61 d and divided into shoot, main roots and minor roots. The main roots were defined as the upper 5 cm under the hypocotyl, without lateral roots. The remainder of the root system with the younger parts was named minor roots. All material was dried at 50°C for 72 h and weighed. The PA and nitrogen contents of all plants were determined as described below.
This experiment analysed PA distribution in the roots of 10 vegetative plants. Three plants (FI1, FI2 and FI3) were collected from a field site, Meijendel, The Netherlands and seven plants from three genotypes (CE1-1, CE1-2, CE2-1, CE2-2, CE3-1, CE3-2, CE3-3, where CE1, CE2 and CE3 indicated different genotypes, originated from Meijendel) were grown in climate chambers in dune sand with 60 ml Steiner solution per week for 3 months (16-h photoperiod, 20°C/15°C (day/night), rh = 70%). Roots were divided into main and lateral roots and root tips. The main roots were defined as described for experiment I. Fresh material of the main roots was easily separated into cortex and vascular cylinder by hand. Root tips were defined as the last cm of the roots. All plant material was dried for 72 h at 50°C and PAs were measured, as described below.
PA extraction and identification
Total PA concentration in the shoots and different root parts was determined in 10 mg dried and ground plant material, according to a modified version (de Boer, 1999) of the acid-base extraction method of Hartmann & Zimmer (1986). PA content was determined on a GC (HP 6890 30 m × 0.25 µm, HP-1) with heliotrine (Latoxan, France) as an internal standard. Conditions of GC analysis were: injector 250°C, temperature programme 0-22-5-250, split mode 1–30, carrier gas N2 0.9 ml min−1, pressure 56 kPa; detector FID. PAs were identified by comparing retention times with reference samples.
Differences in biomass, PA concentrations and nitrogen concentrations between treatments were tested with a oneway ANOVA if possible. Post-hoc multiple comparisons between all treatments were made with Bonferroni tests. Because not all data met the assumptions of ANOVA, individual PA concentrations were tested with Kruskal–Wallis and treatments were compared with a multiple comparisons test according to Siegel (1956). The allocation of PAs to the shoot was calculated as the fraction of PAs present in the shoot of the total amount of PAs in the whole plant. Correlation between nutrient supply and PA allocation was tested with Spearman's rank correlation. Correlation between PA content and root biomass was compared with theoretical correlations based on Monte Carlo simulations. We compared the empirical relationship with the theoretical correlation based on random chosen concentrations (Partel et al., 1996). Five thousand times a random concentration between 0 and 1 mg g−1 was chosen for each shoot and root and each time the correlation coefficient between total PA production and root biomass was calculated. The correlation is significant when greater than 95% of the simulated correlations. All tests were performed with SPSS 8.0 (SPSS Inc, 1998).
As expected, total plant biomass increased significantly with increasing nutrient-supply (n = 39 ρ: 0.597 P: 0.000). Shoot weight increased 4-fold, but the root biomass was not significantly different between the nutrient treatments. Consequently, the shoot : root ratio increased with increasing nutrient-supply rate from 0.5 (N0) to 1.3 (N4). Nitrogen concentration in roots was much less than in shoots. In the shoots, the N concentration in the N0 was significantly less than in N1–N3.
The PA concentrations in different plant parts decreased with increasing nutrient supply (Table 1). Highest concentrations were found in the main roots of plants N0 and N1. The concentration in the shoots was less than in the main roots. There were no significant differences between the minor roots and shoots except for N0. Total PA content of the whole plant and the different organs was not different between the treatments, yet, there was a significant, positive correlation between nutrient supply and PA allocation to the shoot (n = 39, ρ: 0.615, P: 0.000). There is a strong positive correlation between total PA and root biomass (ρ: 0.857 P < 0.05) (Fig. 1) if all treatments are combined. Since root biomass was not different between treatments, this seems to indicate a constant production of PAs independent of the nutrient concentration. However, PA production calculated as total PA in the whole plant divided by root biomass was lower in N3 than in N1 and N2, although it was not different from N0. The decrease in total PA content with increasing nutrients holds for most PAs in the shoot (Table 2). In the shoot senecionine, seneciphylline and erucifoline concentration decreased, while the jacobine concentration was not affected by nutrient treatments. In roots the same PAs are found as in the shoot, but less evenly distributed. Erucifoline is found only in trace amounts, whereas senecionine is the single dominant alkaloid (60–80%). Although present at low concentrations, jacobine is the only alkaloid that actually increases in concentration with increasing nutrient supply.
Table 1. The effect of nutrients on mean (± SE) biomass, pyrrolizidine alkaloid (PA) concentration, total PA and relative allocation of PAs to the shoot in Senecio jacobaea
N0: control: 0 mg Steiner wk−1; N1: 20 ml Steiner wk−1; N2: 40 ml Steiner wk−1; N3: 60 ml Steiner wk−1. *Kruskal–Wallis test.
Table 2. The effect of nutrient supply on average (± SE) PA concentrations in the shoot and roots of Senecio jacobaea
N0: control: 0 mg Steiner wk−1; N1: 20 ml Steiner wk−1; N2: 40 ml Steiner wk−1; N3: 60 ml Steiner wk−1; tr: trace amounts different letters indicate significant differences between treatments.
1.30 ± 0.16a
1.04 ± 0.11ab
0.66 ± 0.05bc
0.34 ± 0.06c
(mg g−1 d. wt)
1.67 ± 0.17a
1.40 ± 0.07ab
1.07 ± 0.08bc
0.68 ± 0.08c
0.31 ± 0.07ab
0.41 ± 0.05a
0.26 ± 0.05ab
0.16 ± 0.03b
1.72 ± 0.20a
1.67 ± 0.21a
1.81 ± 0.17a
1.54 ± 0.17a
1.10 ± 0.10a
0.80 ± 0.09ab
0.73 ± 0.07b
0.48 ± 0.05b
5.72 ± 0.38a
5.70 ± 0.39a
3.91 ± 0.28b
2.87 ± 0.26b
(mg g−1 d. wt)
0.56 ± 0.07ab
0.59 ± 0.05a
0.45 ± 0.04ab
0.37 ± 0.07b
0.48 ± 0.06b
0.69 ± 0.04a
0.55 ± 0.4ab
0.44 ± 0.06b
0.10 ± 0.00b
0.33 ± 0.10ab
0.68 ± 0.14a
0.77 ± 0.16a
3.92 ± 0.26a
4.14 ± 0.30a
3.03 ± 0.28ab
1.91 ± 0.20b
(mg g−1 d. wt)
0.36 ± 0.06a
0.30 ± 0.05a
0.23 ± 0.04a
0.21 ± 0.04a
0.29 ± 0.03b
0.44 ± 0.03a
0.38 ± 0.04ab
0.22 ± 0.05b
0.10 ± 0.00a
0.17 ± 0.05a
0.26 ± 0.05a
0.25 ± 0.06a
Differences were observed in PA concentration in roots (Fig. 2). Lateral roots and the exterior of main roots contained high concentrations. PA concentrations in root tips and in the vascular cylinder of main roots were low, at the minimum detection level. Root tips contained about three times less PA than main and lateral roots. Concentrations in the main root cortex were on average five times higher than concentrations in the vascular cylinder.
Decreasing PA concentrations in roots and shoots with increasing nutrient-supply appear to result from a diluting effect. Increasing nutrient supplies increased plant biomass, while PA production remained constant. Other studies on the response of plants to nutrient addition found significant changes in concentration of secondary metabolites (Iason & Hester, 1993; Höft et al., 1996; Baricevic et al., 1999; Salmore & Hunter, 2001). However, statistical analyses suggest that changes or differences in concentration of allelochemicals can sometimes be explained by changes in tissue biomass only (Koricheva, 1999). Thus, changes in concentration may not result from changes in production, but from dilution or concentration due to enhanced or reduced growth in the whole plant or plant parts.
It is unclear whether the same effect may be expected in flowering plants. Brown & Molyneux (1996) did not find an effect of nutrient deficiency on PA concentrations. Both flowerhead tissue biomass and total amount of PAs increased with nutrient supply, while concentrations remained constant. However, they only measured PAs in flowers and therefore it is not certain whether this reflects a change in total PA production in the plant, or only a change in allocation to the flowerheads in order to compensate for the dilution effect. In an experiment where resource availability to individual plants was controlled, Vrieling & Van Wijk (1994a) found no effect of N or P limitation on PA concentration.
Although root biomass was not significantly influenced by nutrient supply, concentrations of PAs in the roots decreased as nutrients increased (Table 1). This might be the consequence of a larger shoot resulting in a larger PA sink in the above-ground parts. The correlation between root biomass and PA production was positive and stronger than the simulated correlation. This is what is expected, since PAs in Senecio spp. are produced in the roots (Hartmann & Toppel, 1987). However, this reduces flexibility in PA production. If PA production per root is not flexible, and independent of resources like N, then changes in biomass may have a large impact on PA concentrations and thus on the defence of the plant.
Often the negative correlation between plant biomass and secondary metabolite concentration is regarded as evidence for a trade-off. The effect of nutrients on concentrations of secondary metabolites can be explained by two mechanisms. Nutrients, in particular N, are integral components of secondary compounds. Nutrient addition may therefore increase the concentration of PAs. Alternatively, when nutrients are not limiting for the production of PAs, they control to a large degree plant biomass production and secondary metabolites may be diluted when nutrients are supplied at larger rates, which is reflected in a lower concentration.
The resource availability hypothesis (Coley et al., 1985) predicts that under conditions of slow growth, plants should invest more in defence against herbivores. Although large and more nutritious plants may attract more herbivores, there is no need for more defence as long as the growth is faster than the removal of mass by herbivory. This would imply that under conditions of increased plant growth, due to increased availability of nutrients, dilution of secondary plant metabolites is acceptable for a plant explaining the present observations.
Concentrations of individual PAs consistently decreased with one exception. Jacobine remained constant in the shoot and increased in the roots. This might suggest that jacobine is of more importance for the defence of the plant than the other PAs are under the prevalent conditions. Jacobine is unusually toxic to guinea pigs (Cavia porcellus), which are generally resistant to PAs (Chung & Buhler, 1995). However, jacobine concentrations in the roots were very low in comparison to the other PAs and therefor would have to be very toxic. We could not find any study in which effects of jacobine on herbivorous insects or pathogens were compared with the effects of other PAs. By contrast, Vrieling & de Boer (1999) found that the jacobine and erucifoline chemotype of S. jacobaea are equally attractive for preference and performance of the specialist herbivore Tyria jacobaeae.
PA concentrations in root tips were on average lower than in the other root parts. High concentrations were expected in root tips, which are extremely important for the plant with regard to water and nutrient uptake and at the same time vulnerable to attack by herbivores and pathogens. Optimal defence theory (Rhoades, 1979) argues that the plant should protect the most valuable parts best, implying high concentrations of defence compounds for instance in the flowers, seeds, young leaves and below-ground in fine roots.
Sander & Hartmann (1989) found the root apex to be the site of enhanced biosynthesis of PAs, although production was found in all root parts. In the present experiment PAs could hardly be detected at these production sites, which may indicate a fast transport to storage sites, such as the cortex of the main roots. This is opposite to observations by Hartmann & Toppel (1987), who found significant differences in total PA concentration of root parts in root cultures of the annual S. vulgaris. The fine roots contained twice as much PA as the coarse roots. In root cultures there is no transport to above-ground plant parts and also re-allocation in the root cultures may be different from natural circumstances. In this study we used whole plants from the vegetative stage of the monocarpic perennial S. jacobaea. Mature roots often contained higher concentrations of alkaloids than the root tips, like hyoscyamine accumulation in Atropa belladonna (Falk & Doran, 1996). Also in lupines epidermal cells are preferred sites of storage for quinolizidine alkaloids (Wink et al., 1984). The more mature parts of roots are important for regrowth. Defoliation will not kill Senecio spp. (Obeso & Grubb, 1994; Vrieling et al., 1996), but Longitarsus jacobaeae larvae, which feed on the roots can be deleterious to S. jacobaea (Windig, 1993). Although L. jacobaeae is a specialist herbivore, Vrieling & Wijk (1994b) found a negative correlation between PA concentration and herbivory by both adults and larvae. Thus an important part of the root system seems to have a defence in the form of relative high concentrations of PAs, but at present it is unknown against which herbivores or pathogens.
We thank Wiecher Smant for nutrient analysis and T. de Jong, P. G. L. Klinkhamer and E. van der Meijden for valuable comments.