Frost appears to be an important abiotic factor limiting fruit production, as late spring frosts can kill a significant proportion of the flowers and fruits (Dumas & Maillette, 1987; Rapp et al., 1993), which occurred in Baie Saint-Ludger, Qc, in 1998 (pers. observ.). The availability of nutrients may also limit yield as fertilization of bogs with superphosphate or complete fertilizers increased yields several fold (Rapp et al., 1993). Herbivory and parasites may also limit the reproductive success (see Ågren, 1987, 1989), as well as insufficient pollination, especially if the plant is principally entomophilous since female flowers are reported to offer little reward (no pollen and small amounts of nectar) to insect visitors (Ågren et al., 1986).

There is an abundant and diversified diurnal insect fauna known to frequent cloudberry flowers. In Fennoscandia, Hippa et al. (1976, 1981a, 1981b, 1981c) differentiated two groups of insects found on cloudberry flowers: flower stayers and flower visitors. Flower stayers, such as thrips and staphylinids, were very poor cloudberry pollen-carriers while the best pollen-carriers were considered to be the diurnal medium- and large-sized flower visitors such as hymenopterans (Apidae and Formicidae) and dipterans (Syrphidae, Muscidae, Empididae, Fanniidae, Coelopidae, and Heleomyzidae) (Hippa et al., 1981c). However, these observations only provide indirect evidence that insects serve as pollinators as anemogamy may also play a role, especially in windswept coastal areas (Rapp et al., 1993).

We therefore conducted a study to examine cloudberry pollination and the three objectives of this study were to experimentally determine:

To address the first two objectives, we crossed the exclusion of pollinators and supplemental hand-pollination treatments. Each of the four combinations of treatments was assigned randomly to ten 1.2 m x 1.2 m (2.44 m2) quadrats, for a total of 40 quadrats. We marked all female flowers within quadrats with a numbered vinyl tag attached around the base of the shoot. Only quadrats with more than three female flowers were kept for the analysis. We conducted this experiment twice, in 1998 and in 1999.

To address the third objective, we added ten insect exclosure quadrats in 1999. Flowers in five exclosures were available to pollinators only during the day (official photophase from about 04:00 to 20:30), and the other five only at night (official scotophase from about 20:30 to 04:00). This was accomplished by removing the exclusion cages for the designated pollination period each day. We compared the reproductive success of these two groups with that of the permanently open and permanently closed quadrats.

To hand-pollinate flowers, we collected pollen by brushing dehiscent anthers of male flowers with a small paintbrush (damp in 1998, dry in 1999), and checked for the presence of pollen with a 10x magnifying glass before brushing the stigmas of the female flower several times. For each female flower, we repeated the operation with the pollen of at least two males. Different individuals hand-pollinated flowers in the two years of the study.

Insect exclosure cages (1.2 m long x 1.2 m wide x 0.4 m high) were made of a wooden frame and vinyl insect netting (size of mesh = 1.4 mm). The surface area of openings in the insect netting (1.96 mm2) was 2700 times larger than the planar surface of a cloudberry pollen grain (7.1 x 10-4 mm2 ) based on an approximate diameter of 30 µm (Eide, 1981). Hence, openings in the netting were large enough to allow pollen to pass. Furthermore, there were male plants within the enclosure cages that could serve as a source of pollen.

To reduce any possible beneficial or deleterious effect of enclosing plants on fruit development we minimised enclosure time. We set up quadrats and exclosures on 3-5 June 1998 and 1 June 1999, with marking and hand-pollination occurring a week later (10-12 June 1998, 6-16 June 1999). We removed exclosures two weeks after set-up, after the last female flowers had shut. We harvested the fruits prior to complete maturation (5-6 July 1998, 7-8 July 1999) as locals would harvest mature fruits, despite notices that these were research plots.

Variables. We used three variables as indices of the reproductive success of cloudberry flowers per quadrat:

Statistics. We tested the effect of hand-pollination, insect exclosures, of the year, and of all possible interactions among these factors, on fruit set with multiways analyses of variance (ANOVA). We used the general linear model (GLM) procedure in SAS (SAS Institute, 1996) to conduct the ANOVAs because of the unequal cell sizes. The unbalanced design resulted from the fact that we did not include quadrats with less than three flowers in the analyses. We used the same analyses for seed set and fresh weight but the data for the two years were analysed separately due to difference in hand pollination efficacy and the fact that fruits were collected before they were fully mature. We tested the effect of the regime of day or night exclosure on fruit-set in 1999 with a one-way ANOVA.

We transformed proportions of the fruit-set per quadrat with the arcsine transformation of Freeman and Tukey (1950, cited in Zar, 1999):

(1)

where p’ is the transformed proportion and X/n = p is the actual proportion. We transformed proportions of the mean seed-set per quadrat with the common arcsine transformation (Zar, 1999): . Standard errors (s.e.) were calculated using the transformed data and then reconverted to proportions with . We checked normality in the distribution of residuals with the Kolmogorov-Smirnov’s test and homoscedasticity with Levene’s test. We used three multiple comparisons tests: the Tukey’s studentized range test, the Bonferonni t-tests, and the Student-Newman-Keuls (SNK) test. We set the level of significance to P < 0.05 for all statistical tests.

Insect activity was sufficient to fully pollinate female flowers in both years of the study as supplementary hand-pollination in open habitats did not significantly increase any of the parameters of reproductive success (Tables 2, 3, 4). This was not due to an ineffective technique as hand-pollination significantly increased fruit production in exclosures (Tables 2, 3, 4).

There is a highly significant difference in the efficacy of noctural and diurnal pollinators, as measured by fruit set (one-way ANOVA: F = 45.90, dfmodel = 3, dferror = 26, P = 0.0001) (Fig. 1). While nocturnal pollinators significantly increased fruit set over the controls where insects were always excluded, they were not as effective as diurnal pollinators (Fig. 1). The activity of diurnal pollinators, such as bumblebees and syrphids, was sufficient to fully pollinate cloudberry flowers, for fruit-set was as high in quadrats open only during the day as in those always open (Fig. 1).

It is clear from the comparison of fruit-set in quadrats open only at night and those always closed (Fig. 1) that nocturnal pollinators do exist in the habitat. The fact that we found unidentified lepidopteran larvae (Tortricidae) feeding on leaves and fruits in both years make different Lepidoptera potential candidates. However, their relative contribution to cloudberry reproductive success was minor compared to diurnal pollinators (Fig. 1). In other words, in 1999 diurnal visitors were sufficient for pollination, and nocturnal visitors could be seen as complementary pollinators. Nonetheless, if the short duration of exclosure opening at night (7.5 h) compared to the duration of opening at day (16.5 h) were taken into account, nocturnal insect visitors could prove to be as efficient pollinators as diurnal visitors. In years where overall insect numbers are low in the habitat of the cloudberry, the activity of noctural pollinators could significantly contribute to the sexual reproduction of the plant. Furthermore, noctural pollinators are believed to have a lower frequency of flower visitations than diurnal ones during foraging and to transfer pollen over greater distances (Herrera, 1987). If true, then this could be important as it might increase the genetic heterozygosity of a plant with a high degree of clonal propagation.

The next step will be to determine experimentally the effectiveness and efficiences of the most important groups of diurnal and nocturnal insects visiting cloudberry flowers.

Ågren, J., 1989. Seed size and number in Rubus chamaemorus : between-habitat variation, and effects of defoliation and supplemental pollination. Journal of Ecology 77: 1080-1092.

Ågren, J., T. Elmqvist & A. Tunlid, 1986. Pollination by deceit, floral sex ratios and seed set in dioecious Rubus chamaemorus L. Oecologia 70: 332-338.

Dumas, P. & L. Maillette, 1987. Rapport des sexes, effort et succès de reproduction chez Rubus chamaemorus , plante herbacée vivace dioïque de distribution subarctique. Canadian Journal of Botany 65: 2628-2639.

Eide, F., 1981. On the pollen morphology of Rubus chamaemorus L. (Rosaceae). Grana 20: 25-27.

Herrera, C. M., 1987. Components of pollinator «quality»: comparative analysis of a diverse insect assemblage. Oikos 50: 79-90.

Hippa, H. & S. Koponen, 1976. Preliminary studies on flower visitors to and potential pollinators of the cloudberry ( Rubus chamaemorus L.) in subarctic Lapland. Annales Agriculturae Fennicae 15: 56-65.

Hippa, H., S. Koponen & O. Osmonen, 1981a. Flower visitors to the cloudberry ( Rubus chamaemorus L.) in northern Fennoscandia. Reports Kevo Subarctic Research Station 17: 44-54.

Hippa, H., S. Koponen & O. Osmonen, 1981b. Diurnal activity of flower visitors to the cloudberry ( Rubus chamaemorus L.). Reports Kevo Subarctic Research Station 17: 55-57.

Hippa, H., S. Koponen & O. Osmonen, 1981c. Pollen transport and pollinating efficiency of flower visitors to the cloudberry ( Rubus chamaemorus L.) in northern Fennoscandia. Reports Kevo Subarctic Research Station 17: 58-66.

Rapp, K., S. K. Næss & J. Swartz, 1993. Commercialization of the cloudberry ( Rubus chamaemorus L.) in Norway. In: J. Janick & J. E. Simon (eds), Proceedings of the second National Symposium: New Crops. Exploration, Research, and Commercialization, October 1991. John Wiley & Sons, Indianapolis, Indiana, pp. 524-526.

SAS Institute, 1996. Cary, North Carolina.

Zar, J. H., 1999. Biostatistical Analysis. Prentice Hall, Upper Saddle River, New Jersey.

* P < 0.05 ** P < 0.01 *** P < 0.001 n.s. non-significant

* P < 0.05 ** P < 0.01 *** P < 0.001 n.s. non-significant