Breeding Systems in New Zealand Species of Epilobium L.

Transactions and Proceedings of the Royal Society of New Zealand, Volume 87, 1959, Page 189

Breeding Systems in New Zealand Species of Epilobium L. By W. B. Brockie [Received by the Editor, October 1, 1958.] Abstract The flower parts of nearly every species of New Zealand Epilobium have been visually examined and their relative positions observed, especially those of anthers and stigma. The difference in time between stigma and anther ripening, between flower opening and anther dehiscence, and the various responses to environmental conditions were noted. It is concluded from these observations that all the species are probably very preponderantly self-fertilised, with the exception of E. chloraefolium var. kaikourense and E. chionanthum. Garden. experiments in hybridising the various New Zealand species of Epilobium afforded an opportunity to examine factors controlling their pollination. The observations were confined to the progeny of only one or a few randomly collected plants of each species with the exception of E. chloraefolium var. kaikourense. Where more than one local form of a species was examined appreciable differences in their floral parts were sometimes noted. As similar diversity probably exists in most of the species, the following notes are intended to apply only to the local form of the species listed below. E. alsinoides A. Cunn., Gouland Downs, W.6.* W., white flowers; R, rose pink; P, purple. Numeral indicates approximate average diameter of flower in mm. Old Man Range, W.7. billardierianum Ser., North of Porirua Harbour, P.8. brevipes Hook. f. Mt. Terako, W.15. chionanthum Haussk., Swampy Hill, Dunedin, W.20. chloraefolium Haussk., Kakanui Mts., W.15. Arthur's Pass, W.15. var. kaikourense Cockayne, Mt. Fyffe, R.25. (Usually W.) cinereum A. Rich., Mason River, P.7. Molesworth Station, P.8. cockaynianum Petrie, Mt. Egmont, W.12. confertifolium Hook. f., Campbell Is., R.7. Auckland Is., R.7. crassum Hook. f., Mt. Torlesse, R.10. erectum Petrie, Dunedin, P.10 Lake Lyndon, P.10. erubescens Haussk., Arthur's Pass, W.12. Mt. Wakefield, W.12. glabellum Forst., Pencarrow Head, W.10. gracilipes T. Kirk, Gouland Downs, W.9. Castle Hill, W.11. hectori Haussk., Molesworth Station, W.10. insulare Haussk., near Pencarrow Head, W.8. linnaeoides Hook. f., Mt. Peel, N. W.Nelson, W.8. macropus Hook., Takatimu Mts., W.12. melanocaulon Hook., Temple River, W.10. microphyllum A. Rich., Hope River, W.7. nerterioides A. Cunn., Lake Tekapo, W.4. nummularifolium R. Cunn., Wellington, W.5. pallidiflorum Soland., North of Porirua Harbour, W.25. pedunculare A. Cunn. var. brunnescens Cockayne, Gouland Downs, W.7. var. viride Cockayne, South Makara Stream, W.7. pernitens Cockayne and Allan, Tararua Mts., W.13. Olivine Range, W.8. pictum Petrie, Kakanui Mts., W.6.

polyclonum Haussk., Temple River, W.5. porphyrium Simpson, Olivine Range, W.8. pubens A. Rich., near Opotiki, W.9. var., Mt. Hikurangi, East Cape, R.20. purpuratum Hook. f., head of Matukituki River (West Branch), W.9. pycnostachyum Haussk., Mt. Torlesse, W.14 Mt. Terako, R.14. rostratum Cheesem., Ohau River, N. Otago, R.15. var. pubens Petrie, Molesworth Station, W.6. rotundifolium Forst., Gouland Downs, W.9. tasmanicum Haussk., Mt. Torlesse, W.9. tenuipes Hook. f., Old Man Range, W.7. vernicosum Cheesem., Mt. Peel, N. W. Nelson, R.15. sp. with glandular hairs on leaves, shingle slides, Mt. Terako, W.12. Flower Structure In the New Zealand species the flower structure follows much the same pattern of variation as that occurring among members of the genus in other parts of the world. Size of Flowers The flowers range in size from a diameter of about 4 mm in E. nerterioides to 25 mm in E. chloraefolium var. kaikourense and E. pallidiflorum. Those of 21 of the species and varieties listed measure less than 10 mm diameter; 15 of them are between 10 and 15 mm diameter; 4 are over 15 mm diameter. Flower Colour Thirty have white flowers, 7 are rose coloured and 3 purple. Calyx Limb The 4-partite calyx limb reaches almost as high as the corolla in one species, E. cinereum and, in the other extreme, in one species, E. chloraefolium (Arthur's Pass), it reaches only half way up the corolla. Relative Position of Anthers and Stigma. In all except one species the introrse anthers of at least the four longer of the eight stamens at their dehiscence are level with some part of the stigma, usually the lower portion. After the commencement of their dehiscence the filaments lengthen a little, and in some species, notably E. rotundifolium, some of the anthers stand well above the stigma. The exception is E. chloraefolium var. kaikourense in which all the anthers are always far below the stigma level. In E. chionanthum the four longer anthers at the commencement of their dehiscence are slightly below the receptive part of the stigma but a day later they may overtop it. I have been unable to detect any special advantage deriving from the four short stamens except the negative one that overcrowding of the anthers of the four longer stamens is avoided. In certain circumstances this permits the latter and, in some species, also those of the short stamens when they have dehisced to affix themselves to the stigma in a neat, closed ring. Pollen The mature pollen grains remain attached in tetrads. By tapping them on a glass slide with a penknife the tetrads can be broken into their four pollen grains. Normally they are transferred unbroken to the stigma. Each grain usually bears a delicate, viscous and often curled filament, in length usually two to four times the diameter of the grain. Commonly four of these filaments may be seen to project from a tetrad. They function as grappling hooks and very frequently hold the tetrads together in masses or strung like beads.

Stigma The stigmas of the New Zealand species are either clavate, cylindrical or capitate. No stigmatic lobing as occurs in several of the European, African and North American species is apparent except in one species, E. chionanthum (flowers 20 mm diam.), in which the capitate stigma is shallowly but conspicuously 4-lobed, a form of Epilobium stigma which has not previously been observed in New Zealand species. Difference in Time Between Stigma and Anther Ripening Flowers of all the species and varieties listed above have been artificially pollinated with resulting fertilisation immediately after the removal of their undehisced anthers. This clearly shows that in these species the stigmas are receptive before the anthers dehisce—i.e., the flowers are protogynous. This is contrary to the remarks of G. M. Thomson (1880) who stated, “I have not been able to notice any appreciable difference in time between the maturing of the stamens and pistil”. The time lag between stigma ripening and anther dehiscence varies considerably among the different species but, being affected by weather changes, it is not of constant duration in any one of them. On bright warm days this time lag in all species is shorter than when colder conditions prevail, and the situation in this respect may be further complicated when, after the stigma has become fully receptive, a sudden lowering of air temperature delays anther dehiscence. The difference in time between stigma and anther ripening is very short in some species and it was always found necessary at Otari when using these as the maternal parents in hybridising to emasculate the flowers at the unopened bud stage. These species are E. cinereum (Mason River), E. billardierianum, E. hectori, E. polyclonon, E. microphyllum and E. rostratum var. pubens. All the rest of the species were sometimes also successfully crossed at the unopened flower bud stage, although emasculation at that stage was not altogether necessary. The longest interval noted between the beginning of stigma ripening (as evidenced by the development of papillae) and anther dehiscence was 75 hours in the case of E. chloraefolium var. kaikourense during a period of warm but mostly cloudy weather. A few flower buds with the tips of their petals just showing above the tightly clasping calyx lobes were opened at 8 a.m. on February 1. The tops of the stigmas, which are flattened at this stage, appeared to be receptive. One of these flowers was emasculated and the stigma was smeared with pollen from E. brevipes, which was known to cross readily with this species, and a small measure of fertilisation was achieved as was proven by the resulting hybrid progeny. Four similar flower buds were marked, and these did not open to expose their fully receptive stigmas until between 8 and 9 a.m. on February 4. Four hours later partial dehiscence of a few anthers in each of the four flowers was noted and most of the anthers were fully ripened, exposing all their pollen at 5 p.m., but a few did not dehisce until next day. Flower Opening in Relation to Anther Dehiscence The difference in time between flower opening and anther dehiscence could be an important factor in cross-pollination. It varies considerably among the different species, and it is also variable in any particular species, being affected by weather changes. But a fairly satisfactory assessment of the relative value of the factor, as between the species, is obtained from the state of the flowers at a time when the optimum condition for cross pollination prevails. Probably this is during a day bright enough to fully open the flowers but fairly cool and humid so that another ripening is delayed. Older flowers may provide ripe pollen at this time.

In three species, E. microphyllum, E. polyclonum and E. rostratum var. pubens the flowers are virtually cleistogamous at Otari. Only a very few partially opened flowers of these species were seen and these had been copiously self-pollinated. But I examined E. microphyllum in the bed of the Mason River, North Canterbury, where the atmosphere is commonly much warmer and drier than at Otari, and there the flowers were fully open and in a few instances the anthers of these open flowers were still undehisced so that cross-pollination could have occurred prior to selfing. E. rostratum var. pubens opens its flowers in its natural habitat at Molesworth Station, Marlborough, where high summer day temperatures with low humidity have often been recorded (L. B. Moore; pers. comm.). Probably warmer and drier conditions than at Otari also frequently occur at Temple River, North Otago, where E. polyclonum was collected, and it appears likely that this species, too, will open its flowers in its native home. E. cinereum (Mason River), E. billiardierianum and E. hectori also appear to need warmer and drier conditions to activate flower opening than they normally get at Otari, where they open their flowers only on exceptionally fine days. The opened flowers of these three species were also always seen to be copiously self-pollinated. A smaller form of E. cinereum from Molesworth Station, Marlborough, does in bright sunshine at Otari open its flowers a little time before anther dehiscence, especially when the plants are in the full flush of early flowering. In the rest of the species, that is, other than those having flowers that, at Otari, are always self-pollinated in the closed bud, anther dehiscence has been noted to occur some time after flower opening. It may be up to seven hours afterwards on a rather dull day in E. chloraefolium var. kaikourense, E. gracilipes, E. pedunculare var. brunnescens and a large flowered variety of E. pubens from Mt. Hikurangi in the East Cape District. Of course, effective cross-pollination can occur at any time after flower opening and until full fertilisation has taken place. Evidence of late cross-fertilisation was shown in an experiment to test the possible occurrence of natural hybridisation in E. cinereum (Mason River) when all the corollas were tightly closed. As previously stated the flower buds of this plant seldom open at Otari. When the flowers are at a well advanced stage the stigma, which by this time is copiously self-pollinated in its lower portion, sometimes exserts its receptive tip a little way above the closed corolla and tightly clasping, enveloping sepals both of which are almost equal in length. Five of these protruding stigma tips were covered with pollen from E. pallidiflorum. (The two species were known to be interfertile.) All the seeds from the five capsules were sown and 110 of the thousand or so seedlings at the cotyledon stage were planted in a box. Of these, one plant was the expected hybrid, all the others were pure E. cinereum. No natural cross-fertilisation other than that resulting from cross-pollination by another biotype or by another species could, of course, occur in this instance. Self-Pollination Apart from the few species having a flower mechanism ensuring a very high percentage of self-fertilisation—i.e., those having flowers largely self-pollinated before they unfurl their petals even on a sunny day, the other species also have a very effective means of self-pollination. This is in the closing of the flowers at night. If the day has been calm and insects have not disturbed the anthers these, in the larger flowered species especially, stand out free from the stigma and laden with ripe pollen. At dusk the corolla closes to a cylindrical shape (in E. pubens var. from Mt. Hikurangi it is often flattened) thus pressing the anthers firmly against the stigma. This happens also in daylight hours when there is a change to dull, rainy weather. On a dull, rainy morning no flower of any species may open, and

in these circumstances, if it is not too cold, some of the anthers may dehisce when they are in direct contact with the stigma. It has often been noted that the stamen filaments in such flowers are bent with their pressure of growth against their firmly affixed anthers. During a period of cold, rainy weather lasting for three days or so when the flowers of all species remained closed anther dehiscence appeared to be completely suppressed. In E. pictum it was noted after this cold period that the more advanced flowers did not open and their anthers, none of which had dehisced, were discoloured and decayed. Discussion It may be that the amount of cross-pollination in the species varies proportionately with the length of time, under optimum conditions, between flower opening and anther dehiscence. However, as the incidence of cross-pollinating insect visitations to the flowers of the different species in their natural habitats is not known, and it may be that in some species having a short time lag between flower opening and anther ripening these visitations are more frequent than in other species with a longer time lag, it would be unsafe to place too much reliance on any one factor. By means of its viscous threads a pollen tetrad fixed to the body of an insect could probably often drag several other tetrads with it. Doubtless this assists cross pollination. On the other hand, because some of the anthers at least in all the species, with the exception of one variety, are level with some part of the stigma, it is probable that the larger visiting insects very frequently push the pollen laden anthers against the stigma thus effecting self-pollination. Cross-pollinating insects do not appear to be greatly attracted to Epilobium flowers which, although producing nectar, are unscented. At Otari an area of about 150 square yards was filled with various F2 samples, most of which were in flower together in December-January, but even at the height of the floral display never more than a very few butterflies, honey bees, hover-flies and other insects of similar size were seen to visit the flowers at any one time. This was in marked contrast to the large number of such insects visiting the flowers of Hebe growing nearby. The only small insect commonly seen in Epilobium flowers was a species of thrips, but each insect usually remained within a single flower for quite a long time, and they shelter in the flowers overnight. They often collected pollen on their bodies, and it was noted that when one of these insects remained stationary on a petal it was briskly engaged in freeing itself from clinging pollen. In these circumstances it seemed probable that pollen carried by these insects would far more often be transferred to the stigma of the flower in which the insect was feeding than to that of another flower. Practically nothing is known about insect visitations to flowers of New Zealand Epilobium species in their natural habitats. I saw thrips in a few flowers of E. microphyllum and E. melanocaulon in one of the upper tributaries of the Mason River (North Canterbury). However, Mr. E. S. Gourlay, entomologist at Cawthron Institute, informs me that no New Zealand native thrips are known so, pre-historically, this group of insects played no part as a pollination factor in this country. Because of the viscous threads tending to hold the pollen tetrads together they are not readily detached from the anthers by wind. Even after a dry, windy day a large amount of pollen has often been seen adhering to the anthers. Nevertheless detachment of a considerable amount of pollen is by wind agency, but this appears to result mainly from the anthers being themselves blown against the petals and stigma, or battered by the petals into direct contact with the stigma. Probably only a very small amount of cross-pollination is effected by wind-borne pollen.

The closing of the flowers at night not only effects a large measure of self-fertilisation, it is also a mechanism preventing cross-fertilisation by night insect agency. There appears to be no doubt that flower closing at night, their occasional closing during the day, and sometimes their failure to open before anther dehiscence accounts for the result of the experiment made by G. M. Thomson (1880). “I have grown E. nummularifolium and E. pubens, and carefully isolated them under glasses when about to flower, so that all access of insects or of wind was prevented, and they have produced a vast number of capsules and seeds”. Drawings I made to portray the position of the uppermost anthers in their relation to the stigma when they commence dehiscence show that self-pollination caused by flower closing must be very common to all the species excluding E. chloraefolium var. kaikourense the anthers of which, as has been stated earlier, are always far below the stigma level. Even so, in this variety self-pollination in some considerable amount appears to take place, and especially in closed, older flowers in windy, rainy weather when they are wet and drooping. In E. chionanthum, the only New Zealand species known to have a 4-lobed stigma, the anthers at the commencement of their dehiscence are, as was previously stated, slightly below the receptive part of the stigma. At this stage it appears that self-pollination would be unusual. Later, when the anthers are fully ripened and the stamen filaments have elongated, the stigma lobes are curved back and fully receptive and self-pollination could easily take place. Probably these two large flowered plants, E. chloraefolium var. kaikourense and E. chionanthum, are cross-pollinated by insects in much larger measure than any of the other species. These observations appear to indicate that in all the New Zealand species of Epilobium, with two probable exceptions, the amount of gene exchange within natural populations is very small. Acknowledgment I wish to express my very grateful thanks to Dr. E. J. Godley, Director of Botany Division, Department of Scientific and Industrial Research, for his kindly guidance in the arrangement of this paper. Literature Cited Thomson, G. M., 1880. On the Fertilisation of New Zealand Flowering Plants, Trans. N.Z. Inst., 13: 241–288. W. B. Brockie, Curator, Otari Open-Air Native Plant Museum, Wellington.