Effects of bumblebee visits on the seed set of Pedicularis, Rhinanthus and Melampyrum (

The pollinationecology offive species ofRhinanthoideae (the annual species Rhinanthus serotinus, R. minor and Melampyrumpratense, the biennial Pedicularis sylvatica and the annual and biennial P. palustris) has been investigated. All species contain pollen and nectar and are frequently visited bysternotribically and nototribically pollinating bumblebees. The effectiveness of pollen transfer has been measured by the use of fluorescent powder. In R. serotinus and M.pratense no differences exist in percentages of fluorescent stigmas of flowers nototribically or sternotribically visited by bumblebees. R. minor flowers, visited sternotribically, have very low percentages of fluorescent stigmas. This indicates that the pollen-covered venter cannot touch stigmas enclosed by the galea; the movements of the bumblebees probably caused self-pollination, P. palustris, R. serotinus and M.pratense flowers are very frequently perforatedbynectar-collectingshort-tongued bumblebees. P. sylvatica and R. minor flowers are very rarely perforated. Onthese species nectar is mainly collected bynototribicallypollinating bumblebees. Seed production and dependence upon pollinationby bumblebees (Bombus Latr. spp.) are considered. A range from highdependence upon bumblebee visitsforseed production in P. palustris, to medium dependence in P. sylvatica and R. serotinus and virtual independence in R. minor and M. pratense is established. No species is completely self-sterile. Seed set in caged plants is due to favourable morphology and position offlowers. Close proximity ofthecae and stigma or a downward curving of the pistil under pollen chamber in Melampyrum and Rhinanthus insure seed set in cagedplants. In Pedicularis these characteristics for self-pollination are absent. The importance of bumblebees for thefive Rhinanthoideae and the reciprocal importance of these pollen and nectar providing plants for bumblebees is discussed. The importance of alternative pollination by honeybees, thrips and wind is evaluated.


INTRODUCTION
The present study was undertaken, as part of a general survey, to extend our knowledge of the floral ecology of Rhinanthoideae by detailed analysis of pollinator behaviour in its general ecological context.The pollination mechanisms have been described in a previous paper (Kwak 1977).The zygomorphic, nectari- ferous flowers are mainly visited by bumblebees ( Bombus Latr.spp.), nototribi- cally and sternotribically.This paper deals with the significance of bumblebee visits for the seed set of five hemiparasites in the northern Netherlands: Pedicularis palustris L., P. sylvatica M.
M. KWAK Honey bees, collecting pollen and nectar on alpine Rhinanthus species, are mentioned by Fossel (1974).Their pollination efficiency is unknown.
In the literature the ability to self-pollinate in the five species depends on the view of the authors.Several mechanisms of selfing are described: the close proximity of anthers and stigma, the curving of the style so that the stigma is under the anthers at the endof the flowering period, and the decrease of the pressure with which the pollen chamber is kept closed at the end of the flowering period  Faegri & van der Pul 1966).The presence of hairs on the thecae is mentioned in relation to the ability of self-pollination and of the prevention of lateral pollen release during bumblebee visits (Knuth 1899; Heukels 1910; Schoenigem 1922).No comparative figures are available of crossing and selfing.In this context the remark by Smith (1963) that Melampyrum pratense had a good seed set in the greenhouse, probably in absence of bum- blebees, is most pertinent.Other Melampyrum species also exhibit autogamy (Cantlon et al. 1963; Hartl 1974).Alpine Rhinanthus and Pedicularis species are able to self-pollinate, according to Muller (1.c.) and Knuth (1.c.).Self- pollination in buds of R. minor in the Faroes is mentioned by Hagerup (1951).
Pedicularis species in America produced very few or no seeds when caged (Sprague 1962; Macior 1970, 1973, 1974, 1975, 1977).Maclnnes (1972) men- tions arctic Pedicularis populations consisting of plants which self-pollinate and are self-fertile and plants which neither self-pollinate nor are self-fertile.

METHODS AND MATERIALS
Field studies were conducted from May until October in 1974-1978 in popu- lations of Pedicularis palustris, P. sylvatica, Rhinanthus serotinus, R. minor and Melampyrum pratense in the northern Netherlands including the West Frisian island of Schiermonnikoog.A R. minor population in the southern Netherlands was also visited.Self-pollination potentials for each species were determined by comparing fruit development on plants caged to exclude pollinating insects with fruit development on uncaged plants in the immediate vicinity of the caged ones.
Lace covered (meshes 1 x 1 mm) cages 60 x 60 x 60 cm were placed over whole plants, or bags were placed over individual stems.Except for M. pratense, insect exclosures were used both in the field and in the experimental garden.One of the R.
minorpopulations, grown in the experimental garden, was of French origin; seeds had been collected in a dune population near Portbail (Normandy).
T o analyze the possibility of autodeposition of pollen or self-sterility, stigmas of caged flowers of P. palustris and P. sylvatica were dipped in ripe pollen from the same flower.The effectiveness of pollen transfer by bumblebees was studied by touching the pollen sites on the bodies of the visiting bumblebees with a brush containing a fluorescent powder ("Daylight" Fluorescent Heffner and Co.)These importance of alternative pollination by honeybees, thrips and wind, will be evaluated.
bumblebees continued their foraging trips.Visited flowers were collected and the styles removed and examined with a UV lamp for the presence of fluorescent powder.Pollen germination and pollen tube growth were observed by UV fluorescence microscopy (Linskens & Esser 1957; Martin 1959) using a 2% solution of water-soluble aniline blue dye in 20% K 3 P0 4 (Wilms 1974).The possibility of honeybee pollination was examined.Two hives were placed at the edge of a dense stand of Rhinanthus.Pollen loads were collected by using a pollen trap and examined by light microscopy.Hives were also present at a distance of about 30 m from the experimental garden.
The possibility of wind dispersal of pollen was examined in a very dense stand of flowering R. serotinus when weather conditions were favourable.Test slides, coated with silicon oil on one side to catch air-borne pollen grains, were placed at different levels above the ground: 15, 25, 35, 45 and 55 cm, with the coated side facing the wind.Test slides were left for 12 hours, from 23.00 till 11.00 hours and from 11.00 till 23.00 hours.They were examined microscopically for Rhinanthus pollen.Pedicularis palustris occurs in wet, grassy places, dune slacks, damp reed land and meadows.The biennial is most common.Annual plants, known in the Nether- lands only on the West-Frisian coast, differ in several respects from the biennials (Ter Borg & Koeman-Kwak 1973, and  August-September on the biennial and annual plants, respectively {table 2).P.
palustris flowers were often perforated for nectar; sometimes 100% of the flowers exhibited holes.On the annual plants bumblebee males were observed securing nectar.In open-pollinated plants 63-100% of the flowers produced fruits, with 6.6-15.1 seeds per flower.Seed production varied in different years. 1975was a good seed production year, and 1974 a poor year {table 3).Seed set in caged plants was very low (fig.2).The two methods of insect exclosures, cages or bags, gave different results.Fruiting was 12.3% in cages and 27.2% in individual bags.In individual bags flowers may have been pressed together possibly resulting in artificially high fruiting.Fruits developed within the insect exclosures may also have been the result of incidental success by thysanopteran inhabitants.The low fruiting in P. palustris plants under cages indicated the unimportance of Thysanoptera.The lack of autodeposition possibly combined with self-sterility was responsible for low fruiting under cages.Partial self-sterility may occur: 59.4% of the seeds developed under insect exclosure were empty.Only 0.7 viable seed per flower (total 1.7 seeds per flower) were produced.Caged flowers, self- pollinated by dipping their stigmas through their own ripe pollen, set seed (maximum 10 seeds per capsule, see also table I).The sequence in ripening of the anthers can influence self-pollination positively.
In Pedicularis the thecae of the longest stamens, nearest to the stigma, ripen late (fig.6a).The escape of pollen grains from the pollen chamber is restricted by the compressed galea.The position of the stigma in relation to the thecae is not favourable for selfing (figs.I and 6).

Pedicularis sylvatica -Lousewort
Pedicularis sylvatica is a plant of moist heath, marshes and moorland.It occurs in the Netherlands in rather small populations in remnants of heath and along small paths across heath.
The flowers were visited by nototribically and sternotribically pollinating bumblebees (table 2).Perforations were rarely observed.Flowers still in bud with unfolded lower lip and unripe pollen were already visited by pollen collecting sternotribically visiting bumblebees.These bumblebees spent a lot of time on the buds.They did not receive pollen grains because the anthers were; not ripe, but at this stage the stigma is receptive since pollen tubes could be distinguished, growing just below the stigmatic surface.The period of maturity of only the short anthers was very limited.Nototribic B. pascuorum workers were observed when allanthers or the two long ones contained ripe pollen grains.In this phase many pollen tubes in the style had reached the ovary.B. pascuorum continued collecting nectar in flowers whose pistils were already brown and dried (fig.3).On P. sylvatica the sternotribically visiting bumblebees collected pollen, while the nototribically visiting bumblebees collected pollen and nectar.The last group was not seen collecting nectar on buds.
Caged flowers produced moderate (36 %) numbers of fruits and seeds per flower (1.5-3.2).These were probably the result of self-pollination (table 4).In caged flowers, self-pollinated by dipping stigmas through their own ripe pollen, 82.7% developed fruits.This percentage is similar to that in open-pollinated flowers.The position of the P. sylvatica flower was more favourable for self-pollination than that of P. palustris.The stigma was more or less under the galea and under the thecae ofthe long anthers (Jig.1).b).One can wonder whether self-pollination did occur in the fieldsince theflowers were visited frequently, and no pollen grains were left in their anthers.

Rhinanthus serotinus -Greater Yellow Rattle
Populations of Rhinanthus serotinus with different flowering times and mor- phology are present in various habitas such as grasslands, dunes and roadsides.
Early flowering ecotypes (aestivals) are characterized by few branches and a low number of flowers.Autumnals, on the contrary, are richly branched with nearly every branch bearing flowers.Intermediately flowering plants are morphologically intermediate (Ter Borg 1972, Hartl 1974).
Table 2 shows that all ecotypes were pollinated both by nototribically and sternotribically pollinating bumblebees.The frequency of nototribically pollinating bumblebees increased during the season.Perforations in calyces and corollas of R. serotinus were often observed, up to 100%.
Pollen analyses of corbicular loads of honeybees returning from their foraging trips showed the absence of Rhinanthus pollen.Direct observations on foraging honeybees also showed the nearly complete neglect of Rhinanthus.They visited Taraxacum spec., Trifolium repens L. and Lychnis flos-cuculi L.
Analysis of test slides measuring possible wind pollination showed very many pollen grains but no Rhinanthus grains indicating the absence of wind pollination.
Wind, however, could play a role in self-pollination by shaking the flower.
Sternotribically visiting bumblebees, with fluorescent powder on their venter, were at least as good vectors as nototribically visiting bumblebees (table 5).
In open-pollinated plants fruit production was almost 100%, with 5.9-10.8seeds per flower.In caged plants fruitand seed production was decreased.Plants of early blooming populations produced, if caged, more seeds per flower than plants POLLINATION ECOLOGY OF FIVE HEMIPARASITIC RHINANTHOIDEAE 11   of later blooming populations (table 6).Minimumand maximum values of seed production per plant were always higher in open-pollinated plants than in caged ones.In caged plants a mechanism insuring self-pollination was observed.During anthesis the style curved downwards {figs.I, 4, 5), and the stigma moved into a more suitable position for pollen deposition.The lower lip, closing the galea, then bent downwards so that a free fall of pollen grains out of the galea was possible; meanwhile the pressure with which the pollen chamber was closed, decreased.The thecae were more or less kept together by curly white hairs on their surface.Pollen grains were retained among the hairs (figs.Rhinanthus minor grows in grasslands, roadsides and dunes.Plants with blue-and white-toothedflowers may occur in separate populations or mixed ones.R. minor flowers were visited nototribically and sternotribically by bumblebees.Though plants of a certain population were in full bloom in May, bumblebees preferred to visit Symphytum officinale L., Lamium album L. and later Trifolium pratense L. instead of R. minor.In meadows where R. minor grew together with R. serotinus, the latter was preferred (K wak 1978).Nectar was collected mainly by nototribi- cally visiting bumblebees.Perforations in calyces and corollas of R. minor were rarely observed.Sternotribically visiting bumblebees did not transfer fluorescent powder at the same rate as did nototribically visiting bumblebees ( table 5).The consequence is that pollen grains were removed from the thecae without cross- pollination.
Self-pollination during the visits of sternotribic bumblebees is very likely because ofthe very close proximity of thecae and stigma (fig.be).A certain amount of cross-pollination is established by nototribically visiting bumblebees, mainly B. pascuorum workers in populations flowering in June.96-100% of open- pollinated flowers produced fruits and 9.0-13.3seeds per flower; 89-94% of caged flowers produced fruits and 7.1-10.3seeds per flower (table 4).Minimum and maximum numbers in seed production per plant were always higher in open- pollinated plants than in caged ones.Maximum fruit development per plant was in all cases 100 %.By the very close proximity of thecae and stigma and a curving backwards of the stigma inside the galea, pollen grains dropped from the thecae can be deposited on the stigma.Hairs on the thecae were present like in R. serotinus (fig.6e).pollination within one flower.Although they did not touch the stigmas, as for instance pollen thieves in Cassia do (Thorp & Estes 1975), the term pollen thiefis not completely applicable.In species with a protruding stigma (R. serotinus) or with a rather accessible stigma by the broad galea (M.pratense) transfer of fluorescent powder by bumblebees was clear.
Long-tongued bumblebees generally visit more flowers per minute than shorttongued ones (Stapel 1933; Benedek et al. 1973).In the species studied, in which two methods of collecting were strongly related to tongue length, the same rule appeared to exist.The speed of the nototribic visitors, however, did not balance the higher frequency of sternotribic pollinating events due to the greater number of sternotribically visiting bumblebees (Kwak 1978).
The specializations in pollen gatherers and nectar gatherers occurred probably more often in short-tongued bumblebee species than in long-tongued species.In the Rhinanthoideaespecies studied, the methods of pollen and nectar collecting by short-tongued bumblebees were very distinct because the flowers were perforated for nectar.On some dates 100% of the flowers in R. serotinus.P. palustris and M.
pratense were perforated.R. minor and P. sylvatica both contained nectar; the impression exists that this nectar was mainly collected nototribically by B.   pascuorum (see also Kwak 1978).In P. sylvatica it was very difficult for a perforating bumblebee to assume an appropriate position.Only the corolla tube outside the calyx was perforated.Because of the long calyx (10-13 mm) it would still be difficult to reach the nectar with a 6.7 mm tongue.The negative effect of perforation behaviour on seed set could be expected to be limited as long as the pistil is not bitten by perforating bumblebees.The latter was observed in Melam- pyrum arvense (unpublished data).Soper (1952) mentionedthe possibility of pollen transfer in a self-fertile flower ( Vicia faba) by movements of the perforating bee.It is comprehensible that Bilinski (1970) found no decrease in seed set in highly perforated M./?/-ate«.sepopulations.HEiNRiCH&RAVEN( 1972) postulated, on the basis of energetics, that under conditions of ample nectar secretion longtongued bumblebees visited more flowers whenless nectar remained per flower after robbing than without robbing.If this is true, the unprofitable effect of perforating bumblebees on the seed set of a certain plant species depends on the visits of other insect species and on the self-pollination and self-fertilization potentials of the plant species.
The presence of pollen tubes in pistils of unopened flowers of P. sylvatica demonstrated the early visists of sternotribically visiting bumblebees on the one hand and the necessity of pollen for the bumblebee colony on the other.Since B. pascuorum workers visitedolder flowers, they could serve as pollinators on flowers which had not been visited in bud stage.
Cross-pollination by wind in Rhinanthus was not observed.Wind might play a role in shaking and moving the plant resulting in escape of pollen which can be deposited on stigmas ifflower morphology is favourable.If the stigma is protruding this is less likely by the loss of pollen grains passing through the air.
Fossel (1974) mentioned Rhinanthus pollen in honey from Alp meadows.The proportion of Rhinanthus nectar was possibly underestimated by the behaviour of honeybees as secondary thieves, using perforations made by bumblebees.Ho- neybees with corbicular loads of pale-yellow Rhinanthus pollen were also obser- ved.It came from R. major (= R. serotinus), R. hirsutus ( = R. alectorolophus) and R. angustifolius (= R. aristatus).It is striking that we did not observe frequent visits of honeybees in the field or in the experimental garden despite the fact that bee hives were present within 30 meters or less.On R. alectorolophus and R.
aristatus in the Alps I have also observed honeybees visiting flower buds of which only one theca was ripe and where nectar was present.The number of bumblebees was very low if compared with numbers observed in the Netherlands.Perhaps the nearly complete absence of bumblebees permitted honeybees to visit typical bumblebee flowers (unpublished data).
A range in seed production of open-pollinated plants existed.In P. palustris the realized seed production varied from 17-40% of the capacity, in P. sylvatica 16-29%, in R. serotinus 33-60%, in R. minor 50-74% and in M. pratense 60-75% of the capacity {table 4).Withinthe five Rhinanthoideaestudied, a range from high dependence on to independence from bumblebee visits for seed set was observed.
The seed set ratio of caged and open-pollinated plants varied from 0.003 to 0.11 in P. palustris, a species very dependent on bumblebee visits; from 0.25 to 0.59 in P.
sylvatica ; from 0.28 to 0.65 in R. serotinus, two species rather insect-dependent; and from 0.81 to 0.97 in R. minor and 0.93 in M. pratense, two species nearly independent of bumblebee pollination.Species, able to self-pollinate do not have such great differences in seed production in open pollinated flowers as the species that are dependent on bumblebee visits for their seed set.In M. pratense and in R.
minor, both with a close proximity of stigma and anthers, the percentages of realized seed production of caged plants are the highest, followed by R. serotinus with a downward curving of the pistil and finally by Pedicularis.In Pedicularisno mechanism of self-pollination by curving of the pistil was observed.Both Pedi- cularis species are at least partially self-fertile.
Some bagged plants in P. palustris produced more seeds than others.This was due to some external circumstances, as, for instance, removing of the bag to photograph unpollinated flowers.Seed production of this plant was 2.5 seeds per flower and that of the control 1.7.The differencein seed production in plants caged and individually bagged was influenced by movements of the plants and the narrowness of the bags resulting in autodeposition.The observation by Mac-  Innes (1972) that some caged plants of a Pedicularis population had a higher seed production than others could not be confirmed.Germination capacity of seeds of self-pollinated plants was lower than that of seeds of cross-pollinated plants (Koeman-Kwak 1973).In caged P.palustris plants the stylecan elongate resulting in a still greater distance between thecae and stigma and in an unfavourable position.The position ofthe flower and the sequence of ripening of the thecae have also an effect on self-pollination.The drawings of P. oederi, P. lapponica and of P.
sudetica (Kaigorodova 1976) showed that the position of the flower influenced self-pollination.In P. sylvatica the position of the flower makes autodeposition more likely.

Table 3 .
Fruit-and seed production in Pedicularis palustris (no value available).

Table 4 .
6c, d).Fruit-and seed production in ranges of data observed in different populations and years, n = number of populations observed and/or years of observation.Maximum number of seeds per capsule, as given in table I, used as capacity.

Table 6 .
Fruit and seed production in R. serotinus, grown in the experimental garden, except