Summary
Pollen extracts from live locally‐common grassland species in eastern Ontario, Canada were tested for in vitro allelopathic effect on the pollen germination of 17 sympatric target species in 1987. The flowering phenologies of the test and target species were recorded to determine if any phenologically‐divergent species showed evidence of in vitro pollen‐allelopathic interactions. Phleum pratense L. demonstrated pollen‐allelopathic effects on all targets except Linaria vulgaris Hill. P. pratense also exhibited significant phenological divergence with seven of the target species. Pollen extracts from Agrostis stolonifera L., Melilotus alba Desr., and Vicia cracca L. inhibited pollen germination in some of the target species, but the results suggest that these effects were pH‐mediated. The pollen extract of Erigeron annuus (L.) Pers. did not cause inhibition of pollen germination in any of the target species. Further tests with pollen extract of P. pratense and another 23 sympatric target species were performed in 1988. Pollen germination was significantly inhibited in all but two of these species: A. stolonifera and P. pratense. Hence the pollen of P. pratense was not auto‐allelopathic. The breeding system, relatively tall growth habit and relatively large quantity of pollen produced all support the in vitro evidence of P. pratense as a pollen‐allelopathic species.
Methodology
The study site was a 1.2 ha hay-field located in Frontenac County, Ontario, Canada (44° 17' N, 76° 34' W). The field was last sown in 1977 with a seed mixture consisting of Trifolium pratense L. and Trifolium repens L. and abandoned in 1978 (H. McDonald, pers. comm.). The site was chosen to provide a sufficiently large sample size of several species to test for pollen-allelopathic eflects.
Five 'test species' were chosen from among the 58 species found within the field. The initial selection of the test species was based on criteria inferred from Sukhada & Jayachandra (1980): each test species was relatively abundant and had published evidence of root allelopathy, at least within the genus (see Rice, 1984). The test species were: Agrostis stolonifera L. (Poaceae), Erigeron annuus (L.) Pers. (Asteraceae), Melilotus alba Desr. (Fabaceae), Phleum pratense L. (Poaceae) and Vicia cracca L. (Fabaceae). A total of 40 species was used as 'target species' to determine their susceptibility to pollen exudate from selected test species (Table 1). P. pratense was tested for pollen-allelopathic effect on all target species. The target species included all test species, except the inviable E. annuus.
One hundred permanent 1 m x 1 m quadrats were randomly positioned within the field. The initial phenological survey was carried out on June 3, 1987 and surveys were repeated at approximately weekly intervals until Sept. 30, 1987, for a total of 18 surveys. Flowering frequency for each species in each survey was recorded as the number of quadrats where at least one individual of the species was in flower, divided by the number of quadrats within which that species was present. The definition of 'in flower' depended upon whether the species was a test species or a target species. A test species was said to be 'in flower' if the stamens were extruded and the anthers were dehiscent. The targets were 'in flower' if the styles and stigmata were not withered. The surveys were not repeated in 1988 as the focus of this investigation was no longer on the divergence in flowering phenologies.
The pollen from the five test species matured on various dates in early July, 1987. Maturity of pollen was confirmed by use of a magnifying lens to examine individuals for extrusion of the stamens and dehiscence of the anthers. At this time several flowers from the test species were collected from outside the quadrats, stored in open paper hags and transported immediately to the laboratory. The extraction procedure was based on that of Sukhada and Jayachandra (1980). The dehiscent anthers were tapped to release approximately 100 mg of pollen into a sterilized Erlenmeyer flask containing 10 ml of distilled water. The flask was then sealed and magnetically stirred (gently) for 1 h. Random samples(five replicates of 100 pollen grains each) were taken to assess the degree of disruption of the pollen grains due to this procedure. The flask was then sealed and refrigerated at 9 °C for 72 h. After refrigeration, the mixture was centrifuged and poured into a sterilized Erlenmeyer flask through Whatman No. 2 filter paper. The pH of the supernatant was then recorded. The extracts were stored under room conditions in a sealed flask and examined weekly for any changes in pH and any bacterial or fungal contamination.
In order to test as many species as possible, this study used an in vitro pollen tube germination assay similar to that used by Sukhada & Jayachandra (1980). Four chemicals that are known to promote pollen tube germination in most species were used (see Johri & Vasil, 1961 ; Brewhaker & Kwack, 1963 ; Stanley, 1971; Stanley & Linskens, 1974; Bilderback, 1981). To account for possible inhibition of pollen germination by any of these chemicals, five solutions were created : one complete with calcium,potassium, sucrose and boron and the others each lacking a different chemical (Table 2). Since the provision of optimal pH for each species was not feasible, a sodium acetate/acetic acid buffer was used to maintain a pH of 5.8. It was anticipated that this would allow the pollen of most species to germinate (Stanley & Linskens, 1974). Lactose (100 ppm in the control series and 200 ppm in the pollen-allelopathic test series) was added to reduce the possibility of bacterial or fungal contamination (Bishop, 1949).
The concentrations in the pollen-allelopathic test media were twice that of the control media to compensate for dilution occurring with the 1:1 addition of pollen extract (Table 2).
The flasks containing the solutions were sealed and stored at room temperature (15-25 °C). Thereafter, they were inspected daily for pH imbalance and bacterial contamination and changed every two weeks.
Flowers with mature pollen were collected from target species as previously described. To minimize loss of viability, the time period between pollen collection and initiation of in vitro germination was a maximum of 25 minutes. Five cavity slides, each containing 0.5 ml of a different control medium (numbered one to five) (Table 2), were set up for each species collected on a particular day. Pollen from a given species was added to each series of five slides by tapping the anthers. The pollen was then separated gently with a fine probe to avoid local pollen clustering within the slides. The cavities were covered with slightly-bent coverslips and coated with petroleum jelly to minimize contamination but still permit aeration. The slides were left for 24 h on a laboratory bench, under room conditions (15-25 °C and at a relative humidity of 50-70%). The extra flower heads were left in open bags at room temperature for future pollen harvest until pollen viability was lost.
After 24 h these slides were examined under a microscope at a magnification factor of 200 and the percentage of pollen germination within each of five randomly-chosen replicate fields-of-view was recorded for each slide. In cases where pollen germinated, pollen from the stored flowers was placed in newly-prepared slides and the above procedure was repeated daily until viability was lost. After the initial tests using freshly-harvested pollen (d 1),percentage germination records were kept only for tests using three day-old (d 3) and six day-old (d 6) pollen (if applicable). All other tests bad only the presence or absence of germination noted, thereby determining if that specific species/media combination should be continued.
The 17 species that flowered after the 1987 phenological peak of the five test species and produced a sufficient quantity of pollen were tested for susceptibility to the pollen extracts (Table 4). In 1987, only the later-fiowering species could be used as target species because the pollen extracts from the test species had to be collected first. The pollen of each of the targets was added to 0.5 ml of each of the test media (numbered six through ten) (Table 2) and 0.5 ml of one of the test extracts in cavity slides. This was repeated for each test extract giving a total of 25 pollen-allelopatbic test slides, in addition to the five control slides (with media one to five). The tests then proceeded, as for the control series, until loss of viability occurred.
In 1988, in vitro allelopathy tests were run using extract from P. pratense pollen that was harvested in July, 1987 and all target species which flowered before or concurrently with P. pratense. These tests used only freshly-harvested (d 1) pollen and germination media one and six, which contained all four chemicals (Table 2).