Spatial interactions between the hemiparasitic angiosperm Rhinanthus minor and its host are species-specific


  • A. M. KEITH,

    1. School of Biological Sciences (Plant and Soil Science), University of Aberdeen, Cruickshank Building, St Machar Drive, Aberdeen AB24 3UU, and
    2. Centre for Ecology and Hydrology – Banchory, Hill of Brathens, Glassel, Banchory, Kincardineshire AB31 4BW, UK
    Search for more papers by this author
  • D. D. CAMERON,

    1. School of Biological Sciences (Plant and Soil Science), University of Aberdeen, Cruickshank Building, St Machar Drive, Aberdeen AB24 3UU, and
    Search for more papers by this author
  • W. E. SEEL

    Corresponding author
    1. School of Biological Sciences (Plant and Soil Science), University of Aberdeen, Cruickshank Building, St Machar Drive, Aberdeen AB24 3UU, and
    Search for more papers by this author

†Author to whom correspondence should be addressed. E-mail:


  • 1Effects of organisms that obtain resources from others (e.g. herbivores) can depend on the location of resource removal with respect to their ‘prey’. We investigated whether such an effect can be seen in hemiparasitic plant–host plant interactions.
  • 2We conducted rhizotron studies of the interactions between Rhinanthus minor and individuals of two common host species, Festuca rubra and Festuca ovina. Parasites were grown at two distances from the host, and growth characteristics of hosts and parasites measured over time.
  • 3Parasites close to their host suffered reduced survival as a result of shading. Without shading these parasites attached earlier, produced more haustoria, grew larger and had a greater fecundity than those planted further away. This effect was significant for parasites infecting the fast-growing F. rubra, but not for parasites infecting the less vigorous F. ovina. Effect of parasite position on host biomass was significant only for F. rubra, which showed a greater reduction when infected by more proximal parasites.
  • 4Differences in the response of hosts to infection, and parasite growth, suggest the effect of the position of R. minor is host-species specific.


Parasitic angiosperms can have dramatic effects on their hosts, reducing growth to a fraction of that seen in uninfected individuals (Press, Graves & Stewart 1990; Frost et al. 1997; Matthies 1998; Davies & Graves 2000; Hwangbo 2000). Sometimes, however, the effect can be minimal (Moore et al. 1995; Labrousse et al. 2001; Koskela et al. 2002). Variation in the observed effect is likely to be a product of the relative ‘virulence’ of the parasite and susceptibility of the host (Lane et al. 1996; Goldwasser et al. 2000). Factors such as parasite density, timing of parasite attachment (Graves 1995; Gurney, Press & Scholes 1999), genetics (Gowda, Riopel & Timko 1999) and physiochemistry (Fate & Lynn 1996) have frequently been implicated in determining the outcome of the host–parasite interaction. The effect of spatial relations between host and parasite has been overlooked. This is surprising as it has long been appreciated that spatial relations between partners is of key importance in other biotic interactions. For example, during plant–plant competition the distance between individuals can influence the timing and extent of growth suppression of the subordinate (Casper, Cahill & Jackson 2000); in plant–herbivore interactions the location of damage has a great influence on the subsequent growth and competitive ability of the targeted individual (Gold & Caldwell 1989; Marquis 1992; Meyer 1993); the same is true for many parasitic fungi–plant interactions (Clay 1991).

We looked at the effect of spatial relationships between Rhinanthus minor and some of its host species. Rhinanthus minor is a northern-temperate facultative annual hemiparasite most abundant in grassland habitats. Gibson & Watkinson (1989) found that most haustorial connections to host roots are formed within a 10 cm radius of an individual R. minor stem. As the root systems of many plants spread >20 cm, parasite haustoria may be, at least partly, localized on the periphery of the host root system, at its core, or between these two extremes. At the core of the root system attached haustoria are expected to have access to better developed xylem and to a greater proportion of the host root system. Resource removal at the core may therefore be of more benefit to parasite growth, and more detrimental to the host, than at the periphery. However it is also likely that greater energy expenditure is required by the parasite in order to penetrate better developed roots. Peripheral host roots are, in contrast, much easier to colonize, but contain fewer xylem vessels. The question thus arises as to how the interaction between a host and a parasite is influenced by their relative locations.

Further, the spread, size and growth rate of root systems varies between plant species, and the 50 or more species that are known to act as hosts for R. minor (Gibson & Watkinson 1989) represent a broad range of different rooting types. Inherent differences in host rooting characteristics may result in differential impacts of parasitism. Those species with spreading and fast-growing roots could be more likely to encounter a parasite than those with roots remaining in a relatively restricted area. Conversely, species with spreading roots may end up with only a small proportion of their total root system parasitized, as they are able to ‘escape’ parasite-rich patches of ground.

The influence of R. minor on the species around it may thus be a function not only of the inherent susceptibility of the species, but also of the position of the parasite. We tested the following hypotheses.

  • 1Hosts in the immediate vicinity of a parasite will suffer a greater reduction in biomass than those further away.
  • 2Parasites growing nearer a host will have a greater biomass and number of reproductive structures than those growing further away.
  • 3Parasites will have a smaller impact on host species with a spreading root system than on those with a laterally constrained root system.

Materials and methods


Plants were grown in rhizotrons, 40 × 40 × 1 cm, that had one transparent side to allow measurements of root characteristics (Fig. 1). This was covered in light-excluding polythene to prevent root greening. The rhizotrons contained a 50 : 50 mixture of sharp, washed, sand and John Innes No. 2 potting compost. Host plants were grown in the centre of a rhizotron with an R. minor plant either 5 or 10 cm from the centre. The grass species Festuca rubra and Festuca ovina were chosen as hosts because they are closely related and differ in their rooting patterns. F. rubra has a rhizomatous, spreading root system, whereas that of the clump-forming F. ovina is more laterally constrained. These species are both common hosts for R. minor in the field. R. minor seed (Emorsgate Seeds, Kings Lynn, UK) was sown (in Petri dishes) weekly from the beginning of July 2002 to ensure germination would occur when needed. R. minor seeds were surface sterilized with 2% v/v sodium hypochlorite solution, sown on damp filter paper and stored at 4 °C to induce germination.

Figure 1.

Diagram of an individual rhizotron.

pre-attachment parasite survival

Because there might be adverse effects of dense vegetation on the survival of unattached Rhinanthus (Vanhulst, Shipley & Theriault 1987), we examined the effect of distance from host on preattachment parasite survival. Host seeds (Emorsgate Seeds) were sown in trays on 18 July 2002 and repotted into sand 3 weeks later. They were supplied with 20 ml 40% strength Long Ashton's nutrient solution per day until transplant into the centre of the rhizotrons after 2 weeks. The hosts were then grown for a further 2 weeks in the rhizotrons before infection to allow development of the different root systems. Because the parasite germinated sporadically, rhizotrons were infected sequentially. A total of 26 germinated R. minor seeds were planted between 10 September and 11 October These were used to infect five replicates of F. rubra and three of F. ovina at both 5 and 10 cm spacing between host and parasite (one parasite per host plant). Day-length was maintained at 16 h under supplementary lighting and temperature was maintained between 17 and 22 °C. Measurements of environmental variables were taken at a range of distances from the hosts in the rhizotrons. Temperature and humidity were recorded at 5 and 10 cm (RHT+ probe, Skye Instruments, Llandrindod Wells, UK) and photon flux density (PFD) was measured at seedling level, 0, 5, 10 and 15 cm from the hosts using a light meter (SKP200, Skye Instruments). On 31 October parasite seedling survivorship was recorded for the 5 and 10 cm distances. The remaining parasites were removed, oven-dried for 3 days at 80 °C and weighed. None of the parasites attached to the hosts during this study.

parasite growth and the impact on hosts with different rooting patterns

The rhizotrons and host plants described above were used again following parasite removal on 31 October to examine the effect of parasite position and host rooting patterns on the host–parasite interaction. The lateral spread of visible roots was measured using a ruler to a depth of 10 cm at intervals of 2 cm before reinfection with new parasites. Our observations on preattachment survival of parasites showed that seedling mortality increased as a result of shading by the host (see Results). Host grasses were therefore trimmed to a height of 3 cm before reinfection. New parasites were added to rhizotrons on 5 November 2002. Trimming was repeated every 2 weeks after reinfection, and dry weights measured. Repeated trimming allowed a comparison of host growth before and after parasite attachment. A final harvest on 10 January 2003 gave at least 5 weeks’ host growth following parasite attachment. The complete host was harvested and separated into root and above-ground components. Parasite attachment was recorded 3, 5 and 7 weeks after planting seedlings. Parasite heights were measured between 13 December 2002 and 6 January 2003. Numbers of flowers, unopened buds and branches were recorded before harvesting, along with numbers of haustoria. The above-ground biomass of R. minor was harvested and separated into reproductive structures, main leaves, main stem, branches and branch leaves. At harvest on 10 January 2003, R. minor roots were removed from host roots. Dry weight of all biomass from the final harvest was measured.

data analysis

Host root spread data were analysed using an unpaired t-test; temperature and humidity data were analysed using a one-way anova. The light data could not be normalized and so were analysed using a Kruskal–Wallis test. Parasite characteristics and the host and parasite harvest data were analysed using two-way GLM anova. Log10 and Box-Cox transformation were carried out where necessary. Where data have been transformed, original means are shown. All analyses were carried out using minitab ver. 13.


pre-attachment parasite survival

50 days after planting the first germinated R. minor seed in the rhizotrons, there was 77% mortality in parasites planted 5 cm from their hosts. In contrast, no seedlings planted at 10 cm died in this period (Fig. 2). There was a significant difference in PFD between the 5 and 10 cm distances (Kruskal–Wallis test, H = 78·23, n = 36, P < 0·0001). Significant differences in temperature and humidity were also noted between the 5 and 10 cm distances. Mean temperature increased significantly from 21·8 °C at 5 cm to 23·1 °C at 10 cm (anova, F1,14 = 4·81, P = 0·046). Mean relative humidity decreased significantly from 41·7% at 5 cm to 34·6% at 10 cm (anova, F1,14 = 15·83, P = 0·001).

Figure 2.

Survivorship of R. minor seedlings and photon flux density taken at seedling height, at 5 and 10 cm distances from the host grass. Error bars, ± 1 SE; n = 8. NM, not measured.

parasite attachment

There were no clear differences in attachment caused by host species identity during the experiment (Fig. 3). In contrast, the effect of parasite position (planting distance) appeared to be highly important. After 23 days 85% of F. rubra with R. minor at 5 cm had parasites attached, compared with only 33% of F. rubra with R. minor at 10 cm. Similarly, 71% of F. ovina with R. minor at 5 cm had parasites attached, compared with only 28% of F. ovina with R. minor at 10 cm (Fig. 3a). After 36 days the differences in attachment success between the two distances were still apparent (Fig. 3b). Attachment at the 10 cm distance matched that at 5 cm by 49 days after planting parasite seedlings, when ≈84% of all R. minor plants were attached to hosts (Fig. 3c).

Figure 3.

(a–c) Parasite attachment with host species Festuca ovina (white bars) and Festuca rubra (grey bars), and parasite planted at distances of 5 cm (unhatched) or 10 cm (hatched). Attachment calculated (a) 23; (b) 36; (c) 49 days after infection. Percentages calculated from data pooled from five to six rhizotrons.

host growth

No differences in lateral spread of host root systems were observed: lateral spread was ≈11 cm at 2 cm depth, and 23 cm at 10 cm depth. The biomass of F. rubra was significantly greater than that of F. ovina at all six harvests (P < 0·0001, df = 1, 29). Therefore a one-way anova was carried out at each harvest to determine any significant within-species treatment differences. The biomass of F. ovina at each harvest was unaffected by treatment (Fig. 4a). In contrast, infected F. rubra exhibited a significant parasite-induced reduction in biomass 55 days after infection (anova, F2,16 = 5·46, P = 0·016), although the effect of distance between host and parasite was not significant at that time (Fig. 4b). The same pattern of results was seen in harvests after 63 days (anova, F2,16 = 7·11, P = 0·006) (Fig. 4a,b). Sixty-six days after infection, significant differences were present between all treatments for F. rubra– both the parasite and its position had an effect (anova, F2,16 = 7·67, P = 0·005) (Fig. 4b). Final cumulative leaf and pseudostem weight were significantly reduced in infected hosts of both species, but the effect of parasite position was significant only for F. rubra (Tukey's multiple comparison test, P < 0·05; Table 1). Cumulative leaf weight of infected host plants was 70–75% of that in the absence of R. minor for all infected F. ovina, and for F. rubra with parasites at 10 cm. For F. rubra with R. minor at 5 cm, however, this figure was reduced to only 60% of uninfected controls, indicating a greater relative effect of the parasite. The same pattern was evident for total dry weight of host plants. Final root dry weight was unaffected by treatment in either host species, and the interaction between host species and treatment was not significant for any measure of productivity (Table 1).

Figure 4.

Biomass accumulation of individuals of (a) Festuca ovina and (b) Festuca rubra, uninfected (▴) and infected by Rhinanthus minor at 5 cm (•) or 10 cm (○) from the host. Error bars, ± 1 SE; n = 5–7. A, significant differences between uninfected and 5 cm treatments; B, significant differences between 5 and 10 cm treatments (anova, P < 0·05). There was no significant interaction between host species and distance from parasite (two-way anova, P < 0·05).

Table 1.  Final growth measurements of grass hosts Festuca rubra and Festuca ovina in the absence of Rhinanthus minor and with R. minor at 5 or 10 cm (treatment)
MeasurementF. rubraF. ovinaTwo-way anova
Control5 cm10 cmControl5 cm10 cmHost speciesTreatmentInteraction
  1. Values represent means ± 1 SE; n = 5–7 replicates. Numbers within rows sharing the same letter are not significantly different (anovaP < 0·05). Probability values derived from two-way anova are shown by symbols: ** = P < 0·01; * = P < 0·05; ns = not significant.

Final cumulative leaf DW (g)1·58 ± 0·14bc0·93 ± 0·09b1·17 ± 0·11b0·59 ± 0·07a0·44 ± 0·04a0·42 ± 0·1a***ns
Pseudostem to 3 cm above ground level, DW (g)1·65 ± 0·31bc0·75 ± 0·1b1·03 ± 0·11b0·32 ± 0·05a0·26 ± 0·03a0·25 ± 0·04a***ns
Root DW (g)1·55 ± 0·3b0·84 ± 0·2b1·07 ± 0·14b0·21 ± 0·03a0·18 ± 0·04a0·15 ± 0·02a**nsns

parasite growth

Parasites planted 5 cm from their hosts were significantly taller than those at 10 cm at every measurement point from 38 to 63 days after infection (anova, P < 0·05) (Fig. 5). Parasite height was also significantly affected by host species 55 and 63 days after infection (anova, P < 0·05), being greater with F. rubra than with F. ovina (Fig. 5). Likewise, the number of parasite branches, total stem plus branch weight, total branch leaf weight, total main stem leaf weight and total weight were all significantly affected by both planting distance and host species (anova, P < 0·05 to P < 0·001) (Table 2). In all cases the values were greatest in parasites growing 5 cm from F. rubra. There were no significant differences due to host species in the number of parasite flowers, number of unopened buds or total weight of reproductive structures (Table 2). Planting distance, however, had a highly significant effect on the number of flowers (anova, F1,20 = 38·85, P < 0·001) and total weight of reproductive structures (anova, F1,20 = 18·26, P < 0·001), with those parasites closest to their hosts producing the most. There were no significant interactions between host species and distance for any parasite growth measurement (Table 2).

Figure 5.

Cumulative height of individual Rhinanthus minor plants growing 5 cm (closed symbols) and 10 cm (open symbols) from hosts Festuca ovina (circles) and Festuca rubra (triangles). Error bars, ± 1 SE; n = 3–7. A, significant effect of distance to host; B, significant effect of host species. There was no significant interaction (two-way anova, P < 0·05).

Table 2.  Growth measurements of Rhinanthus minor with the hosts Festuca rubra and Festuca ovina at 5 or 10 cm distance from host
MeasurementF. rubraF. ovinaTwo-way anova
5 cm10 cm5 cm10 cmHost speciesDistanceInteraction
  1. Values represent means ± 1 SE; n = 5–7 replicates. Values with different letters within each row are significantly different at P = 0·05 (Tukey's test). Probability values derived from two-way anova are shown by symbols (** = P < 0·01; * = P < 0·05; ns = not significant).

Final stem height (cm)29·0 ± 2·17b 16·3 ± 4·36a 18·6 ± 2·25ab11·6 ± 3·50a**ns
Number of primary branches10·3 ± 1·20b  7·6 ± 0·86ab  5·3 ± 2·63ab 1·2 ± 1·20a**ns
Number of flowers 8·7 ± 0·99b  1·3 ± 0·99a  5·7 ± 0·84b 1·4 ± 0·75ans**ns
Number of unopened buds 4·7 ± 1·61a  3·6 ± 0·61a  1·1 ± 0·83a 1·4 ± 0·87ansnsns
DW of stem + branches (mg) 251 ± 39·1b 89·7 ± 25·7a 94·3 ± 20·9a30·6 ± 9·04a****ns
DW of main leaves (mg) 147 ± 14·3b 86·7 ± 10·9a 83·2 ± 17·3a39·6 ± 7·66a****ns
DW of branch leaves (mg) 120 ± 29·7b 38·1 ± 11·7a 28·7 ± 17·5a 1·4 ± 1·4a**ns
DW of reproductive structures (mg) 185 ± 38·9b 63·6 ± 24·2ab 99·5 ± 16·9b27·8 ± 7·13ans**ns
Total DW (mg) 703 ± 106b278·1 ± 70·1a305·7 ± 63a99·4 ± 21·8a****ns
No. of haustoria (per plant)58·3 ± 8·4b 35·7 ± 4·5ab 22·5 ± 6·8a 9·4 ± 2·2a***ns

Both host species identity and planting distance caused significant differences in the number of haustoria formed by the parasite (anova, P < 0·001 and P < 0·05, respectively) (Table 2). More haustoria were formed on F. rubra, and most were formed by parasites planted at 5 rather 10 cm from the host. Regression between the number of haustoria and final parasite weight showed a highly significant positive relationship with a reasonably strong association (r2 = 0·52, F = 24·0, P < 0·0001) (Fig. 6a). A similar relationship was shown between the number of haustoria and total leaf weight of the parasite (r2 = 0·65, F = 41·0, P < 0·0001) (data not shown). Regression between host root weight and number of haustoria also revealed a significant positive relationship, but the association was weaker (r2 = 0·34, F = 11·5, P = 0·003) (Fig. 6b). There was a tendency for parasites closer to the centre of their host to form more haustoria per unit host root (Fig. 7).

Figure 6.

Regression between (a) number of haustoria and final parasite weight and (b) host root weight and number of visible haustoria. Each point represents an individual parasite or host–parasite combination.

Figure 7.

Number of haustoria per mg host root tissue when Rhinanthus minor is planted at 5 cm (unhatched bars) or 10 cm (hatched bars), infecting the hosts Festuca ovina (white) and Festuca rubra (grey). Error bars, ± 1 SE; n = 6. Bars sharing the same letter are not significantly different (anova, P < 0·05).


is there a relationship between parasite positionrelative to hostand magnitude of effect on the host?

A clear relationship existed between the position of R. minor and its impact on F. rubra, supporting the hypothesis that hosts in the immediate vicinity of a parasite will suffer a greater reduction in biomass than those further away. However, this relationship (or at least its magnitude) may be host species-specific, as there was no significant difference between the growth suppression of F. ovina with the parasite at 5 or 10 cm. Parasites attached to their hosts more quickly if planted close to the core of the root system – but this happened with both F. ovina and F. rubra, so does not explain the difference between the two species. Detailed mapping of the haustoria, along with an examination of their efficacy in solute transport, would be required to explain fully why F. rubra is affected by parasite position and F. ovina is not.

does proximity to host benefit the parasite?

We found mixed support for the hypothesis that parasites growing nearer to a host will have a greater biomass and number of reproductive structures than those growing further away. Being nearer to the host has both costs and benefits for the parasite. A major cost is decreased survivorship of R. minor close to an established host, an effect with several possible causes. Relative humidity increased close to the established host and could have promoted fungal infection of germinated R. minor seeds, reducing survival. Survivorship of R. minor seedlings decreased with PFD. This suggests that R. minor seedlings may be sensitive to shading. Vanhulst et al. (1987) showed that R. minor survivorship was negatively correlated with the surrounding vegetation biomass in grassland, and it was suggested that in established vegetation the decreased survivorship of R. minor is probably a result of more severe competition for light (Vanhulst et al. 1987). Shading sensitivity would not be expected of an effective invader of closed communities, and presumably the benefits of successful root parasitism outweigh the negative effects of above-ground competition. Indeed, once attached, the parasites are much less sensitive to shading (Hwangbo & Seel 2002).

In terms of benefits, proximity to the host appears to have a critical influence on the reproductive fitness of R. minor. In particular, the number of flowers and total weight of reproductive structures are sensitive to distance from the host. Attachment to immediate hosts occurs earlier, therefore these parasites may benefit from a longer period of heterotrophic growth. Svensson et al. (2001) showed this effect for Euphrasia, a hemiparasite closely related to R. minor, where early attaching individuals produced a greater number of seeds. An additional possibility is that, closer to the host, the parasite has better access to a larger proportion of the host's root system and thus to a larger supply of nutrients. Seel, Cooper & Press (1993) found that improved parasite growth was associated with nutrient-rich hosts. Festuca rubra, with its greater number of roots, may be accessing a larger volume of nutrients from the soil, to the particular benefit of those parasites with haustoria at the core of the root system where xylem vessels converge.

is effect of parasite position influenced by host rooting characteristics?

We hypothesized that parasites would have a smaller impact on host species with a spreading root system than those with a laterally constrained root system. Festuca rubra and F. ovina had the same lateral spread in our experiment, although there was a sevenfold difference in root mass. Despite lack of difference in root spread the presence of R. minor affected the two grass species differently, and in the opposite direction to our hypothesis. Host leaf, root and total productivity as a percentage of unparasitized growth was much smaller in F. rubra than in F. ovina. An explanation of the greater suppression of F. rubra may be that there is an increased probability of parasite attachment to plants with large, dense root systems (ter Borg & van Ast 1991). Several previous studies have shown, as we do here, that the number of haustoria is positively correlated with both parasite growth and impact on the host (Tennakoon, Pate & Fineran 1997; Hwangbo 2000). Two to three times the number of haustoria were formed per host on F. rubra compared with F. ovina. It is unlikely that this is the result of greater (or different) chemical signalling from the roots of F. rubra because there was no difference between the numbers of haustoria formed on the two grass species when the data were expressed on a per unit host-root basis (Fig. 7).


As with both parasitic fungi and herbivores, the position of R. minor relative to its host has important consequences for the interaction between the two partners. The proximity to a host affects both the impact of the host on the parasite and, in some species, of the parasite on the host. Being too close to a host may have a cost in terms of increased chance of death before attachment, particularly if the host is ungrazed. The extent to which the apparent benefits of being close to the host are mediated through quicker attachment (and increased period of heterotrophy), compared with a possibly increased volume of host resources accessible through each haustorium, remains unknown. There is evidence that the effect of spacing between host and parasite on the growth of the two partners differs between species, and this may be a product of different host-root architectures. This begs the question of what happens in the field, where multiple host species are simultaneously available to the parasite. It is clear that further work, using multiple host species and a three-dimensional soil environment, is needed to explore further the spatial relationships between root hemiparasites and their hosts.


We would like to thank Janet Woo for technical support and Roger Smith (University of Newcastle) for useful discussion. We acknowledge the financial support provided by NERC (NER/S/A/2001/05959).