Post-dispersal predation of Taraxacum officinale (dandelion) seed

Authors


A. Honek (tel. +420 233 022269; fax +420 233 311591; e-mail honek@vurv.cz).

Summary

  • 1The importance of predation in determining the fate of post-dispersal dandelion (Taraxacum officinale) seed was investigated. Flowering, seed dispersal, seedling establishment, seed predation and seed predator abundance were recorded in 2002 and 2003, at two sites. Number of flowers were counted in 1-m2 plots, wind-borne seeds were collected in water traps, invertebrate seed predation was estimated from the rate of removal of dandelion seeds exposed on the ground and invertebrate activity density was determined by using pitfall traps. The censuses were made at 2- to 3-day intervals.
  • 2Seed dispersal occurred 10 days after flowering. Although some seeds were blown away, 3.7–24.2 × 103 seeds m−2 fell to the ground. Four weeks after the peak in seed dispersal 0.7–3.1% of these seeds germinated. Three weeks later only 11–13% of the dispersed seed remained on the ground and most of these were damaged, the remainder presumably having been removed by predators.
  • 3Predation of exposed seeds was low before seed dispersal but increased after its onset, in parallel with increases in the number of seeds present on the ground and in the activity density of adults of a seed-consuming carabid, Amara montivaga.
  • 4In cafeteria experiments in which the seeds of 28 perennial and annual herbs were provided A. montivaga consumed the most dandelion seeds, followed by nine other Amara species. In no-choice experiments, under field conditions, A. montivaga consumed six seeds day−1.
  • 5Post-dispersal predation, mainly due to aggregation of a single ground beetle species, was more important than that which occurred prior to dispersal. Although predators destroyed c. 97% of the seeds, the effect on dandelion population biology is likely to be small.
  • 6Post-dispersal seed predation may nevertheless be important in other species, as aggregates of large invertebrate predators can consume large quantities of seed.

Introduction

After dispersal, some seeds of annual and perennial herbs enter the soil or germinate, but mortality, generally due to granivory, is often high (Crawley 1997). Small seeds scattered on the ground are mainly eaten by invertebrates, with ground beetles being the most important consumers in temperate zones (Westerman et al. 2003), along with other insects (Carmona et al. 1999). Carabid seed predation has been studied intensively (reviewed for example by Kromp 1999 and Tooley & Brust 2002), demonstrating that carabids both prefer the seed of particular species of plants (Goldschmidt & Toft 1997; Jorgensen & Toft 1997a,b; Tooley et al. 1999; Hurst & Doberski 2003) and are selective feeders in the field (Honek & Jarosik 2000; Honek & Martinkova 2001; Honek et al. 2003).

The abundant composite perennial, dandelion, Taraxacum officinale G.H. Weber ex Wiggers, colonizes a range of grassy areas including gardens, pastures, roadsides and ruderal areas (Stewart-Wade et al. 2002). Its abundant seeds can be dispersed over considerable distances, germinate under a variety of conditions (Stewart-Wade et al. 2002) and seedlings establish and survive amongst other plants (Isselstein & Hofmann 1996). However, the proportion of seedlings that become established is low (c. 1%) compared with the number of seeds produced (von Hofsten 1954). Although some of the seeds enter the soil (Champness & Morris 1948), most disappear from the ground, partly due to predation by both vertebrates (Wilson 2001) and invertebrates. Carabid larvae and adults are important consumers of these seeds (Bertrandi & Zetto Brandmayr 1991; Jorgensen & Toft 1997a; Hartke et al. 1998), with predation by Amara spp., Harpalus spp., Ophonus spp. and Pseudoophonus spp. recorded in an arable field (Honek et al. 2003). Although the intensity of seed predation in this experiment varied among sites and with overall carabid abundance, predation in patches of dandelion, where most of the seeds fall, remains to be investigated.

We therefore conducted a 2-year study of the post-dispersal fate of T. officinale seeds, at two sites where we had also recorded pre-dispersal predation (Honek & Martinkova 2005). The aims were to establish the proportions of post-dispersal seeds that were removed, germinated and remained, to determine the rate of seed predation and to identify the main seed predators. We can then compare the incidence and relative importance of pre- and post-dispersal seed predation under natural conditions.

Materials and methods

study area

The study was undertaken in 2002 and 2003, at two sites situated within a 0.18-km2 area surrounding the Research Institute of Crop Production at Prague-Ruzyne, Czech Republic (50°06′ N, 14°15′ E), which are the same as sites 1 and 2 in our study of pre-dispersal predation (Honek & Martinkova 2005). Site 1 (120-m2 area) was situated in the sward of an abandoned pear orchard, partly shaded by trees and, to the east, abutted an experimental field. The grass was cut twice a year, outside the period when experiments were being undertaken. No control site was established in the adjacent experimental field because of its frequent management and cultivation. Site 2 was in an intensively managed apple orchard, consisting of east–west rows of apple trees with short trunks, separated by 3-m spaces that were alternately sward or superficially cultivated bare ground. An experimental section (Site 2A, 90 m2) was established in a grassy area. A control section in the adjacent cultivated area (Site 2B, 120 m2) was used to establish carabid abundance in an area without dandelions. Dandelion plants grew abundantly in the grassy section, whereas in the cultivated section there was a naturally established stand of weedy annuals (mainly Stellaria media, Capsella bursa-pastoris, Descurainia sophia, Erigeron canadensis) and perennials (Artemisia vulgaris). There was no sward cutting or ground cultivation during the experimental season.

At both sites dandelion flowered only in spring and no seed was produced in summer and autumn.

dandelion flowering and seed dispersal

At each site, five 1 × 1 m plots were marked out before the start of the experiment in areas supporting dense growth of dandelions. Flowering capitula on these plots were counted at 3- or 4-day intervals. Water traps were used to estimate seed production. As the prevailing winds were from the west, one water trap was placed 0.5 m in from the eastern edge of each plot. The area of the water surface in each trap was 30 × 43 cm (2002), or 22 × 35 cm (2003). Trapped seeds were counted at 3- or 4-day intervals throughout seed dispersal.

seeds on the ground

In 2003 the number of dandelion seedlings and seeds remaining on the ground were determined. In this extremely dry year, 19 May (precipitation 16.0 mm), 5 June (4.8 mm) and 8 June (3.6 mm) were the only days following seed dispersal on which it rained, allowing dandelion recruitment. Freshly established seedlings (up to the third leaf stage) were counted in 50 randomly placed plots (15 × 15 cm, marked by a wooden label) at each site on 11 June and again on 19 June. Because the rainy period was followed by drought the counts were pooled and the sum was considered to be the total number of seedlings established on the plots. On 6 and 7 July the vegetation in 20 randomly situated 15 × 15 cm plots at each site was carefully removed and the seeds remaining on the ground collected by means of an aspirator. The counts were made only once because they were time consuming and destructive and the fate of the few germinable seeds present on the ground after 7 July was not therefore established.

seed quality

The collected seeds were divided into undamaged seeds and those that had been partly eaten or had cracked testas. The undamaged seeds were stored in the laboratory at 25 °C, 40% r.h. and then germinated (27 February 2004). Seed from each site was mixed and then divided into groups of 40. Each batch of 40 seeds was put on a dense, slow-filtering filter paper (Filtrak®, Bärenstein, Germany) moistened with 5 mL tap water in a 10-cm diameter glass Petri dish, and kept at constant 25 °C and a 4 hours light : 20 hours dark photoperiod. Germination was monitored every 2 days until no germination had occurred in any dish for 4 consecutive days.

seed removal in the field

Dandelion seeds were put out in the field on circular tin trays of 2.5 cm diameter, with a 25-mm nail attached to the convex side to prevent horizontal movement. The trays were filled with white air-hardening plasticine (Jovi®, Barcelona, Spain), because slugs graze soft plasticine. Thirty randomly orientated dandelion seeds were pressed to about half of their transverse diameter into the surface of the plasticine in each tray. The trays were positioned so that the surface of the plasticine was level with the ground and exposed to predation. Two trays were positioned 5 cm apart and covered with an 18 × 18 × 9 cm wire mesh cage (mesh diameter 9 mm, wire 1 mm thick), whose sides were inserted into the soil to a depth of 4 cm. The roof of each cage consisted of an 18 × 18 cm piece of plastic wrapped in aluminium foil, which protected the trays from rain and direct sunlight. These cages prevented vertebrates from reaching the seeds. A cage was placed 0.5 m from each of the five 1 × 1 m plots at each site where dandelion flowering was recorded. The number of seeds was counted at 3- or 4-day intervals and the trays were replaced before all the seeds disappeared. Details of how the seeds were exposed in the field are illustrated in Honek et al. (2003).

carabid activity density

In 2003, the activity density of adult carabids in the field was determined using pitfall traps (Adis 1979). The pitfall traps were plastic cups, 7 cm in diameter (orifice area 38.5 cm2) and 8 cm deep. The cups were inserted into the soil, with the rim at the soil surface, and screened from rain and direct sunshine by a plastic dish wrapped in aluminium foil. This screen was placed c. 5 cm above the trap and supported by short wooden rods. No bait or preservative was used. A few lumps of soil of 0.3–1.0 cm diameter scattered in a 1–1.5 cm deep layer at the bottom of the cups provided shelter and humidity for the trapped arthropods. The traps were emptied at 2- or 3-day (weekends) intervals. The adult carabid beetles were determined to species and released. Five traps were placed on sites 1 and 2A, each within 50 cm of a cage with dandelion seeds, and 30 on the control, dandelion-free, site 2B. All carabid beetles captured at sites 1 and 2A were identified to species, but only the Amara species at 2B, because of the vast numbers of beetles caught at this site.

carabid preference for dandelion seed

The preference of adults of 28 species of carabid for dandelion seed was established by means of a ‘cafeteria’ experiment using seed of 28 species of herbaceous annuals and perennials (details available from the authors): this paper is only concerned with the fate of the dandelion seeds. Adults of each species of carabid were collected in the field, at the time of its maximum activity. Beetles were stored for 5 days at 5 °C to produce a uniform level of hunger before 10 adults were placed in a glass Petri dish (25 cm in diameter, 5 cm high) containing a 2 cm deep layer of sieved soil. The soil used in the seed consumption and preference experiments was dug from > 0.5 m depth to avoid contamination by seeds from the soil bank. Soil in each Petri dish was moistened with 100 mL of tap water and the beetles were provided with drinking water in the form of moist cotton wool. Feeding trays filled with soft plasticine were prepared for each seed species as in the removal experiment and presented in two concentric rings, the outer consisting of 19 and the inner of nine trays, so that all trays were separated by 0.4 cm within rings and the rings by c. 2.5 cm of soil. An experiment period of 5 days was sufficient to reveal carabid preferences, because of high consumption of the preferred seeds. The numbers of seeds remaining were recorded daily and trays were replaced before all the seeds were consumed. Throughout the experiment all kinds of seeds were available ad libitum. The beetles thus had free choice of seed species but their behaviour was not continuously monitored. There were five replicates (Petri dishes) for each carabid species, except for Calathus ambiguus, Harpalus atratus and Parophonus maculicornis, which were rare and replicated only once. The Petri dishes were placed in a laboratory (25–27 °C) and screened from direct sunlight. The average seed consumption (number of seeds individual−1 day−1) and specific consumption (number of seed mg beetle dry mass−1 day−1) were calculated. Average dry body mass of the carabid species was calculated from body length (Hurka 1996) using the formula of Jarosik (1989).

seed consumption bya. montivaga

Consumption of dandelion seeds by Amara montivaga Sturm was studied in a no-choice experiment, between 27 May and 8 June 2003. Adults were collected in the field and stored for 5 days at 5 °C, before being placed individually in cylindrical glass tubes, 10 cm in diameter and 10 cm high, covered with glass lids. Each tube contained 2 cm of sieved soil moistened with 15 mL of tap water and a piece of moist cotton wool as a source of drinking water and a feeding tray with 30 dandelion seeds (as above). The number of seeds was recorded every day and the trays replaced before the seed supply was exhausted. Twelve experimental tubes were placed in a laboratory (26–28 °C, indirect sunshine, natural photoperiod) and 12 in the field at ground level, shaded from direct sunshine and rain by a wooden structure. During the field experiment, the minimum temperature (mean ± SE) at ground level was 12.3 ± 0.6 °C and the maximum was 29.8 ± 1.2 °C.

data processing

Arithmetic means (± SE) were calculated for the number of capitula m−2 (n = five plots per site), seeds m−2 day−1 (n = five water traps per site), seeds removed tray−1 day−1 (n = 10 trays per site) and carabid adults trap−1 day−1 (n = five traps per site). Because the average flowering time of a capitulum was < 3 days the total number of capitula produced m−2 over the flowering period was calculated as the sum of numbers of capitula on each census date (A. Honek and Z. Martinkova, unpublished data). Similarly, the average number of seeds m−2 produced over the season was the sum of the seeds caught m−2 day−1 in the water traps.

To compare flowering and seed production at sites 1 and 2A, the data for years and sites were synchronized so that the start of seed production was set as day 0, with days before this assigned a negative and days after a positive number. There was annual and between-site variation in numbers of seeds that dispersed m−2 day−1, seeds removed tray−1 day−1 and carabid adults trap−1 day−1. To compare the seasonal changes at sites 1 and 2A the data were standardized so that the highest value for a particular site and year was 1 and other values were expressed as proportions.

The data were fitted by linear (y = a0 + a1x) or second order polynomial (y = a0 + a1x + a2x2) models. The latter model was used when the quadratic term significantly increased the explained variance.

The similarity in the composition of the Amara populations between sites was compared using the Renkonnen index

Re = min ∑ (pip, pjp)

where pip and pjp are the proportions of species p in samples i and j. Re increases from 0, when there is no resemblance, to 1 when samples are identical (Balogh 1958).

Results

flowering and seed production

Depending on year and site flowering started on 21–28 April and continued for 18–39 days (Fig. 1a shows site 1 in 2003, but the patterns for other sites and dates are similar). Most capitula flowered within 1 week of the peak, which occurred 9–12 days after the onset of flowering. Seed dispersal started 9–14 days after the start of flowering and continued for 23–40 days, with the maximum 7–14 days after the start. Numbers of capitula m−2 that flowered over the whole period and of seeds collected m−2 were higher in 2003 than 2002 (Table 1).

Figure 1.

Time of flowering and seed dispersal of dandelions, and of seed predation and carabid activity density, at site 1, 2003. (a) Number (+ SE) of flowering capitula m−2 and number of seeds that dispersed m−2 day−1. (b) Number (+ SE) of dandelion seeds removed tray−1 day−1. (c) Number (– SE) of A. montivaga (AMON) and number (+ SE) of all seed-eating carabid species trap−1 day−1 (TOTAL).

Table 1.  The number of flowers, seed production, seedlings present 4 weeks after seed dispersal, number of ungerminated seeds on the ground 7 weeks after seed dispersal and number of missing (probably eaten) seeds m−2 (estimated). The figures in brackets indicate percentages assuming all the seeds dispersed (= 100%)
 Flowers (number m−2)Seed production (number m−2)Emergence seedlings m−2 (%)Ungerminated seeds m−2 (%)Missing seeds m−2 (%)
Site 1
 2002132 ± 26.1  3668 ± 731.1   
 2003177 ± 26.012 501 ± 1902.7393 ± 42.2 (3.1)1661 ± 209.7 (13.3)[10 447] (83.6)
Site 2A
 2002177 ± 14.9  6223 ± 1238.1   
 2003324 ± 25.724 166 ± 1658.7157 ± 23.3 (0.7)2684 ± 286.9 (11.1)[21 325] (88.2)

seeds on the ground

Eighty-one to 94% of the observed seedlings were present at the first count and the rest 8 days later. The number of seedlings (Table 1) represented 3.1% of the seeds dispersed on site 1 but only 0.7% on site 2A where partial cover of dense turf prevented emergence. One month later, 11.1% and 13.3% of the seeds were still present on the ground (Table 1), although 85% and 93% of these were damaged at sites 1 and 2A, respectively. At least a part of this damage could be attributed to pre-dispersal seed predators. Of the undamaged seeds, 68.4 ± 3.2% germinated. Seven weeks after dispersal only 1.1% and 0.6% of dispersed germinable seeds remained on the ground and 83.6 and 88.8% had disappeared without trace.

removal of seeds

The rate of removal of seeds exposed on the ground varied during the season (Fig. 1b), increasing from low levels once seed dispersal had started (Fig. 2). At different sites and years, peak removal was observed 9–21 days after the start of seed dispersal, after which seed consumption gradually decreased. Average removal rates varied between years and sites, and daily rates of removal varied greatly, probably because of variation in weather.

Figure 2.

The consumption of experimentally exposed seeds before and after the beginning of seed dispersal (day 0). Common plot of synchronized and standardized data for both sites (S1 and S2A) and years. The regression for days −5–20: a0 = 0.126, a1 = 0.032, R2 = 0.460, P < 0.001.

carabid activity density

In 2003 adult carabids were abundant on both experimental sites, with seed-eating species making up 52.3% and 94.4% of all the carabid species captured on sites 1 and 2A, respectively. Among the seed eaters, Amara spp. comprised 91.9% and 65.4%. The dominant species at the experimental sites was A. montivaga, which made up 92.9% and 55.6% of all the Amara spp. at site 1 and site 2A but at the control site (2B), which supported a growth of herbaceous species other than dandelions, it represented only 2.2% (Fig. 3). Amara montivaga appeared to be the main dandelion seed predator: it was attracted by dispersed seeds and was responsible for the increase in the rate of seed removal after the start of seed dispersal. Overall, A. montivaga activity density increased after seed dispersal (Fig. 4), although less so than seed removal (Fig. 2), probably because of the effect of weather on the size of pitfall catches.

Figure 3.

The percentage species composition of the total catch of Amara spp. in 2003, at sites 1 (n = 188), 2A (n = 199) and 2B (n = 2834). AMON =A. montivaga; AAEN =A. aenea; AFAM =A. familiaris; ASIM =A. similata; AOVA =A. ovata; OTHERS = rare species A. aulica, A. bifrons and A. littorea.

Figure 4.

The activity density (number of individuals day−1 trap−1) of A. montivaga before and after the beginning of seed dispersal (day 0). The synchronized and standardized data of 2003 are arranged as in Fig. 2. The regression for days −10 to +20: a0 = 0.214, a1 = 0.0189, R2 = 0.307, P < 0.005.

consumption of dandelion seeds by carabid beetles

Under standard conditions, A. montivaga consumed the most dandelion seed (3.7 seeds day−1, or 0.46 seeds mg dry mass−1 day−1; Table 2), followed, in terms of consumption per unit body mass, by nine other Amara species. These species, and others, often preferred dandelion to other offered seeds, but consumption was much lower than that of A. montivaga. Some large carabid species, Amara eurynota (Panzer), A. convexiuscula (Marsham), Zabrus tenebrioides (Goeze) and Anisodactylus signatus (Panzer), may eat large numbers of dandelion seeds. In a no-choice experiment, A. montivaga consumed 8.9 ± 0.6 seeds day−1 in the laboratory, and 5.8 ± 0.6 seeds day−1 in the field where the average temperature (20.8 ± 0.7 °C) was lower. When only dandelion seeds were offered, consumption was two to three times higher than in the multichoice experiment.

Table 2.  Dry body mass and seed consumption by 28 species of carabid in the multichoice experiment. Specific seed consumption = number of seeds eaten mg dry mass−1 day−1. Rank = the order of preference of the particular carabid species for seed of dandelion presented together with seed of 27 other herbaceous species (rank 1 = dandelion is the most consumed species). Individual daily consumption = number of seeds eaten day−1
SpeciesMass (mg)Seed consumption
Specific (n mg−1 d−1 × 102)RankIndividual (n d−1)
Amara montivaga (Sturm) 7.946.2 ± 2.59  23.7 ± 0.20
Amara bifrons (Gyllenhal) 3.943.3 ± 4.64 11.7 ± 0.18
Amara aenea (DeGeer) 6.333.6 ± 3.53 22.1 ± 0.22
Amara similata (Gyllenhal) 9.328.8 ± 1.03 42.7 ± 0.10
Amara ovata (F)10.126.1 ± 0.87 12.6 ± 0.09
Amara sabulosa (Audient-Serville) 4.825.7 ± 6.97 11.2 ± 0.34
Amara littorea (C.G. Thomson) 6.721.0 ± 5.36 61.4 ± 0.36
Amara anthobia (A. Villa et G.B. Villa) 4.120.4 ± 4.32 50.8 ± 0.18
Amara eurynota (Panzer)14.819.7 ± 0.20 12.9 ± 0.03
Amara apricaria (Paykull) 6.017.7 ± 2.08 11.1 ± 0.13
Harpalus signaticornis (Duftschmid) 5.016.0 ± 4.14 30.8 ± 0.21
Amara ingenua (Duftschmid)12.015.2 ± 3.64 11.8 ± 0.44
Amara convexiuscula (Marsham)18.915.0 ± 0.42 12.8 ± 0.08
Parophonus maculicornis (Duftschmid) 4.513.4 50.6
Harpalus luteicornis (Duftschmid)  6.312.9 ± 2.18 20.8 ± 0.14
Anisodactylus signatus (Panzer)24.110.7 ± 0.93 12.6 ± 0.22
Harpalus affinis (Schrank)14.110.0 ± 2.07 31.4 ± 0.29
Amara familiaris (Duftschmid) 4.1 9.5 ± 1.17 70.4 ± 0.05
Zabrus tenebrioides (Goeze))36.3 7.8 ± 1.95 12.8 ± 0.71
Amara consularis (Duftschmid) 8.2 7.5 ± 3.19 10.6 ± 0.26
Calathus ambiguus (Paykull)14.14.1 10.6
Acupalpus meridianus (L) 0.9 2.7 ± 1.59 80.0 ± 0.01
Pseudoophonus rufipes (DeGeer)31.3 2.4 ± 0.48 60.8 ± 0.15
Amara aulica (Panzer)24.1 1.5 ± 0.39 40.4 ± 0.09
Harpalus atratus (Latreille)12.01.3180.2
Ophonus azureus (F) 6.9 0.4 ± 0.31110.0 ± 0.02
Calathus fuscipes (Goeze)20.2 0.1 ± 0.03130.0 ± 0.01
Trechus quadristriatus (Schrank) 1.1 0.0 ± 0.0 0.0

Discussion

seed rain

The calculated densities of seeds falling to the ground m−2, ascertained by water traps, were similar to those recorded earlier (e.g. von Hofsten 1954) but lower than those calculated by multiplying the abundance of capitula by their seed contents (Roberts 1936). Dividing the numbers of seeds on the ground by the total number of capitula produced over the season revealed that each capitulum contributed 28–75 seeds on average, i.e. much less than the average dandelion seed set, which is 186 ± 64 seeds (Honek & Martinkova 2002). Although this suggests that 60–85% of the seeds might have been dispersed from the dandelion patch where they were produced, local aggregation close to the parent had an important effect on the behaviour and activity of seed predators.

carabid predation

The rate of seed consumption increased after seed dispersal. This may be explained by the concurrent increase in activity density as carabids aggregated at sites where seeds were abundant. However, because predator abundance and seed consumption are poorly correlated (cf. Figs 2 and 4), it is also possible that predators learn to eat dandelion seeds when their availability increases (a functional response III, Begon et al. 1990), although behavioural studies would be needed to test this hypothesis.

Adults of the specialized consumer, Amara montivaga, appeared on the experimental sites at the time of maximum dandelion flowering (Fig. 1) and before the onset of seed dispersal (Fig. 4). This is consistent with observations of A. montivaga feeding on dandelion flowers (P. Saska, personal observation). The activity density of A. montivaga also increased with the numbers of dandelion seeds dispersing, suggesting that aggregation of this species in dandelion patches explained the increased predation. In contrast, meadows that lack dandelions are nearly free of seed consumers and have low seed consumption rates (Honek et al. 2003). Aggregation of A. montivaga influenced the composition of carabid communities. The similarity between Amara communities on the closely adjacent sites, 2A and 2B, was low (Re = 0.205) because these sites supported different communities of herbaceous plants, one of which did not support A. montivaga. In contrast, the similarity between the grassy sites with dandelions, sites 1 and 2A, was high (Re = 0.871). The preference of A. montivaga for dandelion seeds was confirmed by the laboratory experiment in which it was the top consumer of dandelion seeds of the 28 carabid species tested.

The larvae and adults of Amara montivaga have similar food specificities. Saska & Jarosik (2001) failed to rear larvae on diets of Capsella bursa-pastoris seeds, although development can be completed when fed T. officinale seeds (P. Saska, unpublished data). Multichoice experiments (A. Honek, Z. Martinkova and P. Saska, unpublished data) reveal that despite the high consumption rate of dandelion seed (Table 2) Crepis biennis is even more favoured. Taraxacum officinale and the closely related C. biennis often share the same habitat and their seeds may serve as alternative food. At both site 1 and site 2A, flowering and seed maturation of C. biennis followed that of dandelion in June and July: it is likely that individuals of A. montivaga colonize dandelion patches during the short period when seeds are present after seed dispersal and developing larvae might then move to feed on C. biennis.

None of the other species that the multichoice tests revealed as important consumers of dandelion seed (A. eurynota, A. convexiuscula, Z. tenebrioides, A. signatus, P. rufipes) were trapped in significant numbers at sites 1 and 2. They are most active in summer and autumn and may be predators of dandelion seed at sites where there is a second period of flowering in September and October.

dandelion population biology

An estimate of the effect of seed predation on dandelion population biology can be obtained by combining these results with those from our study of pre-dispersal predation (Honek & Martinkova 2005). In the period ± 1 week around the peak in seed dispersal, pre-dispersal predation at sites 1 and 2 was very low and accounted for only 2.2 ± 1.2% of seed production, whereas post-dispersal predation accounted for 83.6–88.8% of the dispersed seeds, leaving mostly damaged seeds. In total therefore 84.0–89.1% of the seeds were destroyed by predators. Because c. 90% of the seeds remaining on the ground were damaged and non-viable, the proportion of seeds surviving at the end of the experiment plus the proportion that germinated was no greater than 2–4%. However, even the small proportion of seeds that germinated resulted in a recruitment of 160–400 seedlings m−2. Although most of these probably have perished because of competition with grass, on bare ground this density approaches the carrying capacity. If the seedlings were arranged uniformly, the rosettes could reach 26–40 cm in diameter before contacting another plant.

Seed consumption, based on multiplying the removal rates from trays over a period of 20 days from the onset of seed dispersal (0.316 seed day−1 tray−1) by the area of the trays, indicates a predation rate of 643 seeds m−2 day−1. Although the predation rate may be lower when seeds are less abundant, all dandelion seeds may be eaten within several days. However, the presence of viable seeds 6 weeks after seed dispersal indicates that some of the seeds may enter the soil seed bank, even in unfavourable habitats. Their number must be greater in the absence of predators. We conclude that at our sites predation did not significantly affect dandelion recruitment but probably decreased the proportion of seeds entering the soil.

Acknowledgements

We thank Professor A. F. G. Dixon for critical comments on the manuscript, helpful suggestions and improving the English, and Mrs H. Uhlirova and Mrs L. Kreslova for excellent technical assistance. The work was supported by grant no. 521/03/0171 of the Grant Agency of the Czech Republic and grants 0002700601 (ZM) and 0002700603 (AH, PS) of the Ministry of Agriculture of the Czech Republic.

Ancillary