Turdus spp. were the only frugivorous birds feeding on S. aucuparia pomes present at the time of seed collection. Although Turdus iliacus L. (Redwings) were the dominant species, Turdus merula L. (Blackbird), Turdus pilaris L. (fieldfare), and Turdus philomelos Brehm (Song Thrush) were also present. All are migratory birds between ≈21 and 27 cm in length, foraging mainly on fruit in autumn and winter (Cramp 1988). Turdus spp. are classified as dispersers rather than seed predators (Herrera 1984). Meal sizes for T. merula (mean weight ≈100 g) are about 10 fruits with diameters 8–10 mm (≈7 g). A meal is taken in about 1 min, with an interval of about 20 min between meals (Snow & Snow 1986). We were unable to distinguish droppings from the different species.
Seedling Emergence and Growth
The seeds and pomes used in the growth study were collected 13–14 October 1995 on Turøy (60°27′ N, 04°55′ E) in western Norway. Approximately 150 droppings containing S. aucuparia seeds were collected from slopes of naked rock near the only deciduous spinney on the island. Twenty clusters of pomes were picked arbitrarily from the first 10 S. aucuparia trees found. All seeds extracted from droppings, and a corresponding number of seeds from pomes (nearly all), were sown.
The seeds were extracted by hand from pomes and droppings, and sown as described below, allowing us to study the response of seed gut passage on seedling emergence and seedling growth under six different germination conditions. As seeds aggregate in varying numbers in bird droppings, we sowed seeds singly (120 pots) or in groups of two (72 pots), three (72 pots), and four (72 pots). In addition, single seeds were sown in arbitrarily selected bird droppings (120 pots) and in pomes (120 pots). When sown in pomes, seeds were placed in arbitrarily selected cores which were then reinserted in the pomes (again arbitrarily selected). Ingested seeds would, of course, never be reinserted in pomes under natural conditions, but as the main objective was to study the effect of gut passage on seedling growth, and as seeds sometimes germinate in S. aucuparia pomes, we included this option in the present study. We believe that due to rainy and snowy winter conditions, seeds sown directly in soil best simulate natural Norwegian conditions. To examine if the opening of pomes influenced seed emergence, single unopened pomes (60 pots) were also sown.
The seed groups were all sown in Jiffy pots (made of compressed peat) on 23–25 October 1995. A commercial potting soil was used (Huminal plantejord, Norsk Hydro). When sowing, the pots were nearly filled with soil (approximately 5 mm from the brim), the seed(s) were added (in a pile when more than one seed), and additional soil was added giving a sowing depth of ≈5 mm.
To break the embryo dormancy (Devillez et al. 1980; Flemion 1929; Zentsch 1968) under natural conditions, all pots were dug down to turf level on a germination site in an open field at Homme (at the southernmost tip of Norway, 58°03′ N, 07°17′ E) on 27 October 1995. To offer some protection from black frost they were covered with a thin cloth and Horse Chestnut (Aesculus hippocastanum L.) leaves. On 26 April 1996, cloth and leaves were removed and the emerging seedlings were tallied daily. Day of emergence is defined as the day when a seedling was first visible.
Exactly 28 days after emergence, each growing seedling was moved from the germination site and placed in the middle of a 30 × 30 cm square of turned turf. To avoid interaction between the growing seedlings they were placed 1 m from neighbouring seedlings. Vegetation in this field consists mainly of Trifolium pratense L., Trifolium repens L. and the grasses Phleum pratense L., Festuca pratensis Huds. and Poa pratensis L. The field site was watered regularly twice a week during the growing season, and the grass around the seedlings mown once a week. During seedling emergence, the maximum temperature increased steadily from 7·4 °C on 27 April to 10·6 °C on 12 May. During the same period, cloud cover thinned out gradually. Mean temperature and precipitation in the summer of 1996 were 14 °C and 43 mm, respectively (data from the nearby Lindesnes Weather Station, ≈15 km, courtesy of the Norwegian Meteorological Institute, DNMI).
Eight weeks after the individual day of seedling emergence, the growth variables (length of stem, leaf number, and length and width of longest leaf) were recorded for the first time. These measurements were repeated every 4 weeks through the whole of the first growth season. The seedlings’ stem length was measured from ground level to apical shoot using a folding ruler. The number of fully expanded leaves was counted, and length and width of the longest leaf measured with a slide caliper. The length was defined as length from the rachis at the petiole base of the first pair of (innermost) leaflets to the tip of the apical leaflet. The width was defined as length from tip to tip of the same first pair of leaflets. All growth variables were measured to ±0·5 mm. Where there were more than one seedling per pot, the seedlings with the longest stems after 8 weeks were measured, irrespective of how many seedlings were growing in the pot. (An unequal number of growing seedlings in pots where more than one seed was sown is assumed to increase the group variance, and increased variance yields a more conservative statistical test result.) Only pots where the initial number of seedlings survived the whole season were used in calculations. Seeds sown in pomes were excluded from the growth experiment due to few emerging seedlings.
As seedlings from defecated seeds grew larger than control seedlings, we decided to examine the effect of ingestion on seed mass. During 19–22 October 1996, two T. merula were held in captivity, starved for 12 h overnight, then fed only S. aucuparia pomes. To prevent the birds from feeding selectively on larger pomes, they were fed limited amounts and not given a fresh supply until all pomes were eaten. Water was freely available.
The pomes were collected (on Turøy) from 30 arbitrarily selected S. aucuparia trees on 17 and 19 October 1996 (again the first trees we could find). Two clusters of pomes were arbitrarily picked from each tree, one for each treatment group (randomly assigned), so that half the pomes were given to the birds while the seeds from the remainder were extracted by hand. The resulting bird droppings were allowed to dry, then crumbled, and the seeds removed by hand. The seeds from both treatments were briefly (<60 s) submerged in water to wash off remainders of dropping or fruit flesh adhering to the seeds. Floating seeds, believed to be incapable of germination, were removed from the sample. All seeds were allowed to dry at room temperature for 2 weeks. Both samples consisted of a little over 500 seeds. A random subsample of 250 was drawn from each treatment and the seeds were weighed individually to an accuracy of ±0·1 mg. To examine if water absorption caused the differences observed, 50 new seeds, again selected at random from both treatments, were weighed and placed in an electric airing cupboard at 50 °C until constant weight was achieved. They were then left to adjust to room humidity and weighed again. The mean mass of the 50 dried seeds, as a percentage of original weight, was calculated for both treatments and used to individually correct the weight of the original two sets of 250 seeds. These converted values were then tested statistically.
Several different statistical methods, all with critical P values set to 0·05, were used due to the various experimental designs and the nature of the experiments.
A Student's t-test was used to test for differences in mean seed mass between ingested and control seeds. When testing for differences in percentage seedling emergence, the χ2 goodness of fit procedure was followed according to Zar (1996), using the Yates correction for continuity. According to Fox (1993), failure–time analysis should be used when testing for differences in rate of seedling emergence. The Wilcoxon life-table method was applied as it is most sensitive to differences early in the time series when testing for differences in homogeneity of the seedling survivorship functions (SAS Institute Inc. 1989).
Repeated measures ancova (SAS Institute Inc. 1989) was used to compare the seedlings’ growth performance of the single growth parameters length of stem, number of leaves, and length and width of longest leaf. Day of emergence was treated as a covariate to exclude random differences in weather conditions. (For an introduction to repeated measures, consult von Ende 2001.) For an overall comparison of the seedlings’ growth performance, combining probabilities were conducted according to Sokal & Rohlf (1995). The doubly multivariate repeated-measures manova design (SAS Institute Inc. 1989) is unsuitable for our data due to the decrease in power of the test with increasing number of dependent variables in combination with low sample size (von Ende 2001).
Homogeneity of variance was tested using the Box M multivariate test procedure. Critical P value was set to 0·01. Log transformation (in measures of length) and square root transformation (in counts) were used in cases where the Box M-tests indicated unequal variances (Sokal & Rohlf 1995).