[75]

STERILITY IN WILD ROSES AND IN SOME SPECIES HYBRIDS
EILEEN WHITEHEAD ERLANSON¹
University of Michigan,
Ann Arbor, Michigan
Received January 7, 1930

¹ Paper from the Department of Botany of the UNIVERSITY OF MICHIGAN, No. 326, representing work carried on under a National Research Fellowship in the Biological Sciences.

TABLE OF CONTENTS

Most botanists agree that there occur in nature many spontaneous hybrids between related rose species (BOULENGER 1929), and it has been found that wild individuals often exhibit a considerable amount of aborted pollen. HUSKINS (1929) pointed out that neither meiotic irregularities nor pollen sterility can be used as a certain criterion for hybridity, because the balanced conditions necessary for regularity during meiosis may be upset by other agencies besides hybridization. Nevertheless it cannot be denied that these phenomena are as JEFFERY (1929) states "outstanding features of the reduction division of hybrids."

HURST (1925, 1928) bases the experimental testing of his theory of differential polyploidy in Rosa upon the contention that any two diploid species from one of his five fundamental species groups will give "fully fertile" offspring, while any two from two different fundamental species will give sterile hybrids. It was therefore of interest to discover the actual state of fertility and its variation among our wild roses, as well as to make some crosses using parents with known pollen sterility.

The percentage of shrivelled grains in pollen from thirty-seven individuals, collected in 1927, has been given by the writer (ERLANSON 1929).

These samples were stored over calcium chloride and examined dry. Since [76] then pollen from three hundred and sixty individuals collected in 1928 and 1929 has been examined. It was first measured dry and then treated with aceto-carmine, which causes the grains with contents to swell and stain, and greatly facilitates counting. Pollen from the same individual has been examined for empty grains both dry and after treatment, and similar results were obtained by both methods. It was found necessary to count about 400 grains in each sample in order to get a reliable result, since, when the amount of sterile pollen is high, there is apt to be a clumping of the shrivelled grains in some parts of the field under a 3 mm. lens. The plants from which pollen samples were taken are all growing in the Botanical Garden of the UNIVERSITY OF MICHIGAN.

VARIATION IN STERILITY WITHIN THE SPECIES

The pollen sterility in eight groups of North American wild roses, mean sterility and variation among a given number of individuals selected at random are shown in table 1. The percentage of pollen grains without contents is given.

TABLE 1
Mean pollen sterility in wild rose species groups.

In partially sterile pollen many of the grains though apparently perfect morphologically are not able to function in fertilization, as has been found in grapes (DORSEY 1914), in wheat (SAX 1922, WATKINS 1924) and in Nicotiana (EAST 1921). It is therefore probable that the total pollen sterility is actually higher than shown by my figures, since no tests of pollen germination were made. More samples of pollen have been examined from R. blanda Ait. than from any other species but only the first twenty-eight have been included in table 1, in order to make the comparison between this and the other species groups more equal. In figure 1 the variation in pollen sterility in five groups of roses has been plotted in classes differing by 10 percent of empty grains, and a larger number of [77] individuals are included in the graph for R. blanda. These graphs and table 1 bring out strikingly the small amount of empty pollen present in the hexaploid group of R. acicularis Lindl. and the diploid R. palustris Marsh. as compared with the other species. Although only a few counts have been made for R. palustris, it is the least variable of the eastern diploid species and these records probably give a true picture of its pollen conditions. The two large and widespread diploid groups of R. blanda and R. Woodsii Lindl., and all the tetraploid roses examined, show an average of one-fifth to one quarter of empty pollen grains. The range in variation among individuals from each group is large, some having almost perfect pollen morphologically. Variation in pollen sterility is found in all cultures of any size and throughout the range of the various species. In the [78] European species R. pimpinellifolia L., R. cinnamomea L., and R. arvensis Huds., all of which have balanced chromosome complements, SCHWERT-SCHLAGER (1910) found that the percentage of shrivelled pollen grains reached 20 to 25 percent in some individuals.

FIGURE 1.--Variation in pollen sterility in five groups of wild roses.

HARRISON and BLACKBURN (1927) gave the percentage of good pollen found in the various species groups of the section Caninae. They do not record the number of plants examined and, since the greatest range in variation given for any species is only 20 percent, it is probable that only a few samples were analysed for each. My figures bring out very strikingly the fact that in all the diploid and tetraploid groups on this continent (with the exception of R. palustris) there seems to be a segregation in the factors causing sterility, giving some plants with almost perfect pollen, morphologically, and others with as much as 50 percent aborted grains. In the group of R. californica C. et S. only four plants have been examined because these roses are scarcely hardy at Ann Arbor. NICOLAS (1927) has found that different plants of the same species show a great difference in fertility (fruit production) and attributed it to soil and habitat variation. That it may be due to inherent variation in sterility is shown by the fluctuations found in cultures grown under uniform conditions at Ann Arbor.

The Caninae all exhibit unpaired chromosomes at diakinesis and there is a large proportion of empty grains in the pollen of most of them (HARRISON and BLACKBURN 1927). In the cultivated Rubi LONGLEY (1927) finds indications of an association existing between even chromosome numbers and the production of fertile pollen. Table 1 shows that the few wild triploid rose individuals discovered thus far have almost completely abortive pollen. In the classic hybrid Primula floribunda. X P. verticillata the sterile diploid F1 plants have normal and regular maturation divisions, the contents of the microspores later disintegrating (DIGBY 1912, NEWTON and PELLEW 1929). A similar condition was reported in cultivated grapes by DORSET (1914). An example of this type of pollen sterility associated with regular meiosis was described by the writer in a tetraploid Oregon rose related to R. Durandii Crépin which had 54.6 percent of the spores empty (ERLANSON 1930). I have come across two other tetraploid individuals and one diploid in which all the chromosomes usually pair at diakinesis, and which have over 2/3 of the mature pollen grains empty. The pollen of the tetraploid plants has been examined for more than one season, and some yearly fluctuation in sterility is shown, but the amount of empty grains is always of the same order. These amounts are listed as percentages in table 2.

[79]

R. relicta Erlanson is a dwarf from Illinois related to the tetraploid R. arkansana Porter. It sets very little fruit and shows some irregularities during meiosis (ERLANSON 1929). The plant N117, 17, found in a diploid culture from R. blanda var. glandulosa Schuette, shows fourteen pairs at diakinesis, produces few flowers and has not set any fruit. It will be discussed in a future paper. A plant of the irregular tetraploid species R. mollis Sm. (Section Caninae) had only 67 percent of empty grains in 1929. The highly sterile specimen of R. Fendleri Crépin was sent from the Medicine Bow Mountains, Wyoming. It is not very vigorous but usually bears some fruit.

TABLE 2
Percentage of empty grains in pollen of three plants with all chromosomes pairing and with high sterility.

TABLE 3
Variation in pollen sterility among the offspring of individual wild roses.

The variation in the amount of empty pollen grains in different individuals in cultures from wild diploids and wild tetraploids is shown in table 3. In each case the plants in a culture were the offspring of one wild seed parent. Column two gives the year in which the pollen was collected. These figures indicate a high degree of heterozygosis.

[80]

Table 4 contains the percentages of empty grains found in some wild diploid species of which there are only a few individuals at the Michigan Botanical Garden. In view of the fact that external conditions have been found to cause pollen sterility, it is interesting to observe the small amount of empty pollen in the specimens of R. pisocarpa A. Gray, since these are barely hardy in Michigan, being frozen back to the ground every two or three winters.

TABLE 4
Percentage of empty grains in pollen of miscellaneous diploids in 1929.

Rosa microphylla Roxb. is tender at Ann Arbor and rarely flowers. A plant which flowered in 1929 showed 29.3 percent of empty pollen grains. This plant is one of a very uniform culture and the species is not closely related to any other rose. In the pollen studies of CLEMENTS and PENLAND (1925-1926) they found that very few species in nature produce over 95 percent of good pollen, and they suggested three classes according to the percentage of sterile pollen, namely, with 50 percent and over of sterile pollen, with 20 to 50 percent, and under 20 percent. My pollen examinations show that the proportion of empty pollen grains in any group or species of Rosa may vary from about 5 percent to 25 percent. This amount of sterility may not be of much significance in relation to the detection of natural hybrids, but rather due to physiological and ecological conditions. This matter will be referred to again in connection with sterility in known hybrid roses.

DIPLOID AND TETRAPLOID R. SUBSERRULATA

A diploid plant of R. subserrulata Rydberg (No. 9528), discovered in 1927 (ERLANSON 1929) had 24.8 percent empty microspores in 1929. This plant was sent from Oklahoma by Mr. RALPH SHREVE. It has always been very weakly in vegetative growth and all parts, including the flowers, are smaller than in the normal tetraploid race. It has been severely attacked by anthracnose the past two seasons. It may have originated as a [81] haploid from a normal tetraploid, or it may have been produced from a triploid hybrid by the union of two gametes possessing only seven chromosomes. Another plant of R. subserrulata (No. 7697) sent by Mr. SHREVE from Arkansas, is a normal vigorous tetraploid plant. It had 7.8 percent of empty pollen grains in 1929. The abnormal chromosome number of the diploid R. subserrulata is not revealed by any unusual amount of sterile pollen, but rather by the depauparate habit.

STERILITY IN INTERSPECIFIC HYBRIDS

Beginning in 1925 many crossings have been made between species growing in the rose collection at Ann Arbor. The difficulty with which rose seeds germinate is well known. If the seeds germinate promptly the seedlings of most species reach maturity in their third season. It is not surprising, therefore, that there were not many F1 plants mature in 1929. Table 5 shows the variation in the amount of aborted pollen in the members of some F1 cultures, two of which had diploid parents and four tetraploid. In each instance the parents belonged to closely related species.

I. R. Schuettiana Erlanson X R. acicularioides Schuette

Both these species are possibly descended from R. blanda by natural hybridization, with R. palustris to give R. Schuettiana and with R. acicularis to give R. acicularioides. The particular plants used in this cross both show under 14 percent of empty pollen grains. In nine F1 plants only one had more than 14 percent and the mean sterility for the hybrid plants was less than the mean sterility for the two parents. The parent species are seemingly completely compatible. The results from this cross suggest that the production of an amount of pollen up to about 20 percent may be an innate characteristic in plants which may be considered as "fertile" wild rose individuals. The F1 hybrids are fairly uniform in appearance and resemble R. blanda. They are extremely fioriferous and the inflorescence tends to be compound. The stems, new shoots and foliage are richer in anthocyanin pigmentation than in typical R. blanda. A good crop of hips and achenes is obtained by open field pollination, but very few seeds are produced after artificial self-pollination. I have often found indications of self-sterility in R. blanda and in other wild roses, another factor adding to the difficulties of genetical work in this genus.

Garden plants of R. Schuettiana have been grown from the type clump only, but we have several plants of R. acicularioides from different stations which show great diversity in pollen sterility, one plant having 43 percent of empty grains. Therefore it cannot be concluded that the cross [82] R. Schuettiana X R. acicularioides would give fertile hybrids invariably. When HURST (1925, 1928, 1929) proposed breeding as a test for relationship in wild species, he cannot have been aware of the great differences in fertility among plants which belong morphologically to the same species.

TABLE 5
Pollen sterility of F1 hybrids compared with that of parents.

[83]

II. R. subblanda Rydberg X R. blanda var. Hermanni Erlanson

The plants used in this cross both belong within the collective species R. blanda, R. subblanda differing from the type of R. blanda chiefly in its glabrous foliage. Both parents had exceptionally "good" pollen, yet among nineteen F1 individuals the amount of empty pollen grains varied from 2 percent to 46 percent in 1929. The mean sterility of the hybrids was 17.7 percent greater than the mean sterility of the parents. If the results were used to test whether the parents belong to the same "fundamental diploid species" of HURST, or not, it would be difficult to decide. Certainly the hybrids are not completely sterile, but if one by chance had only those with over a third of the pollen abortive the classification would be still more puzzling, since HURST does not allow for the possibility of "semi-sterile" diploid hybrids. The R. subblanda parent happened to have very little aborted pollen, but this condition is not characteristic for all individuals of the species. Another plant, from Missouri, had 57 percent of the microspores lacking contents, also in 1929.

The achenes of one of these F1 plants which had 7.5 percent empty pollen grains, were examined. In four hips a total of 83 large achenes and 34 aborted ovaries were found. Only 71 of the achenes contained embryos.

The F1 plants in this culture show segregation in colour of stigmas, filaments and petals, and with regard to pubescent and glabrous foliage. Therefore it is evident that in spite of the high pollen fertility of the parents they were not homozygous for these characters.

The conclusion to be drawn is that the processes of mutation, crossing over and so forth, have proceeded in different directions in these two related forms, rendering them partially incompatible when interbred. BOULENGER (1929) found a series of plants which he suspected of being hybrids between the two diploid species R. arvensis and R. sempervirens. He sent pollen samples from the same sets of plants to two men who obtained widely different results as to the percentage of bad pollen. However, by both investigators the suspected hybrids were found to have more sterile pollen than either of the parents. Both these species are placed by HURST in his fundamental species AA, and should give "fully fertile" hybrids.

III. Tetraploid hybrids

The plants used in the first three crosses listed under tetraploids in table 5 are all suspected of hybridizing in nature. BEST (1887) at one time proposed to make R. virginiana Mill. and R. carolina L. and the related forms all varieties of one species, but this suggestion has not been generally [84] followed because of the distinct habit of R. virginiana as compared with the others.

Two F1 plants from R. virginiana X R. carolina had a mean pollen sterility 8 percent less than that of the two parents.

The tetraploid forms R. Lyoni and R. carolina are closely related, R. Lyoni being a strongly pubescent and vigorous form. The five F1 plants obtained from crossing these roses all show fairly good pollen, the mean sterility of the hybrids being only 3.7 percent greater than the sterility of the seed parent. The sterility of the pollen parent was not ascertained.

Two tetraploid hybrids obtained from R. virginiana X R. obovata show unusually low amounts of abortive pollen grains, only half as much as that shown by the seed parent. The sterility of the pollen parent in this cross is also unknown. R. obovata Raf. is very doubtfully distinct from R. carolina L.; it was described as R. humilis grandiflora by Baker (R. humilis Marsh = R. carolina L. 1753 not 1762).

The tetraploid hybrids I and III in table 5 produce abundant pollen and a good crop of fruit. This is an indication that R. virginiana is closely related to R. carolina and its varieties and it may be said, in these instances, to be completely compatible with them in hybridization, judging from the fact that none of the hybrid plants has over 20 percent of empty pollen grains, nor a higher sterility than the parent plant of R. virginiana used in the experiments. In HURST'S cytological classification (1928, 1929) he classes R. humilis grandiflora Baker (R. obovata Raf.) as AADD and R. virginiana as CCDD. Each capital letter designates a set of seven homologous chromosomes. Following this scheme the hybrids from R. virginiana X R. obovata would be ACDD, unbalanced tetraploid plants with seven pairs and fourteen univalent chromosomes at first meiotic metaphase. These plants have not been examined cytologically, but one would expect them to have a greater amount of poor pollen than the balanced tetraploid parents.

A plant of R. suffulta Greene was crossed with R. carolina in 1925 and one F1 plant was raised, which flowered first in 1928. This plant is of particular interest because it resembles the form R. rudiuscula Greene, a highly variable species found in the central United States along the Mississippi Valley, in the region where the ranges of R. carolina and R. suffulta overlap. The hybrid plant has only a quarter of the pollen grains empty and a wild plant sent from Missouri by Mr. B. F. BUSH had 23.5 percent aborted pollen in 1929.

[85]

The amounts of empty pollen grains in four miscellaneous species hybrids not made by myself, which are in the collection of the UNIVERSITY OF MICHIGAN, are shown in table 6, together with the classification of the parents involved according to HURST'S scheme.

HURST places R. palustris Marsh. among the tetraploid species. Our plants are diploid and are probably his diploid "R. carolina" which he designates as a DD species (HURST 1928, 1929). The plants involved in each of the hybrids listed in table 6 happen to belong to HURST'S fundamental diploid species CC and DD. The hybrids are therefore all CD, and should be "completely sterile." The hybrid R. foliolosa X R. rugosa alone fulfills this condition. It was found to be completely sterile in 1928.

TABLE 6
Pollen sterility in miscellaneous diploid hybrids (Hort.).

It was scarcely hardy in Michigan and has been sent to California for further observation. In 1928 a plant appeared, in an isolated bed among bushes of R. blanda and R. subblanda, which had distinctly rugose foliage. It is thought to have originated from a chance pollination of R. blanda by R. rugosa Thunb. This plant bears good fruit but shows more abortive pollen than either of the plants of R. Tetonkaha Hansen. It produces, however, a larger amount of pollen than the latter.

R. Tetonkaha was produced by HANSEN by crossing a hybrid R. rugosa with a native wild rose in South Dakota, probably R. blanda. Our plants closely resemble the illustration of R. Warleyensis Willmott, a plant which was judged by Miss WILLMOTT (1914) to be R. blanda X rugosa, except that extra petals are frequent in the flowers of R. Tetonkaha. The low percentage of sterile pollen in the two plants of R. Tetonkaha (table 6) is somewhat misleading, because these plants usually produce only a small amount of microspores in each anther. One of these individuals was examined cytologically and was found to have seven pairs at diakinesis and to [86] exhibit very little polyspory (ERLANSON 1929). The plants set good fruit. Under HURST'S scheme there should be no pairing of chromosomes at meiosis in these hybrids and they should be sterile. Judging by external morphology one would not place R. blanda and R. rugosa in the same collective species. The production of supernumerary petals and the deficiency of mature pollen would be indicative of the hybrid nature of these plants of R. Tetonkaha if their origin were unknown. The tetraploid F1 hybrids listed in table 5 do not exhibit these characteristics.

Seed of the hybrid R. palustris X rugosa were obtained from the BOTANICAL GARDEN at Amsterdam, Holland, and our plants are perhaps F2 hybrids. Several of the plants of this culture resemble strongly the rugosa parent. These have never bloomed. Two plants which have many characteristics in common with the palustris parent show very little sterile pollen and bear several hips containing plump achenes each year. They seem to exemplify segregation in F2 of a partially sterile hybrid into sterile and more fertile types, as reported by KRISTOFFERSON (1926) to occur in interspecific Malva hybrids.

A hexaploid F1 plant of R. Engelmanni X R. acicularis var. Sayiana flowered for the first time in 1929 and had 6.6 percent of the pollen grains aborted. The female parent had 13.3 percent of empty grains in 1927 and the male parent 5.3 percent in 1929. R. Engelmanni S. Wats. is closely related to R. acicularis var. sayiana Erlanson, although HURST classifies the former as BBDDEE and the latter as CCDDEE.

VARIATION IN POLLEN STERILITY IN THE SAME INDIVIDUAL IN ONE SEASON

CRÉPIN (1889) reported that in his studies of pollen sterility in wild roses he found that the amount varied from bud to bud on the same bush as well as from year to year. My own observations show that the sterility in two samples from the same bush taken in one season usually varies very little. Some examples are listed in table 7. The large variation in sterility in R. relicta is anomalous, this type being one of the balanced tetraploids with high sterility referred to above. It is evident that the amount of lagging-at reduction division in this plant differs from bud to bud. The average variation in percentage of sterility for the ten diploid individuals listed in table 7 is 4.3, which is very close to the average difference in sterility from one season to the next for twenty diploid individuals and for eight tetraploids listed in table 8.

[87]

TABLE 7
Variation in the percentage of empty grains in pollen samples taken from the same individual in the same season.

VARIATION IN POLLEN STERILITY IN THE SAME INDIVIDUAL FROM SEASON TO SEASON

Increase in pollen sterility has been induced in several genera by exposing the plants to abnormal temperatures. STOW (1926-1927) found that high temperatures would cause pollen sterility in Solanum tuberosum. An increase in pollen sterility was obtained at high temperatures by SAKAMURA and STOW (1926) in Gagea lutea, and by HEILBORN (1928) in cultivated apples. Cold caused an increase in the number of aborted pollen grains in Datura (BLAKESLEE and CARTLEDGE 1927). Sterility induced by gall mites in Lycium halimifolium was reported by KOSTOFF and KENDALL (1929).

Temperatures in Michigan are notoriously variable. The first flower buds of the early-flowering octoploid and hexaploid roses in the group of R. acicularis were killed by frosts late in May, 1929, and some buds in this group and also in R. Woodsii had their petals destroyed by the late cold spell. Early in June there is often a hot dry period, which may be followed by cold rains before the heat of July with its drying winds. Since temperature has been found to affect the amount of pollen sterility it was of interest to find out to what extent the latter fluctuated in different years in the same rose plant. Should the amount of sterile pollen vary greatly from one year to the next it would be obviously useless as a

[88]

TABLE 8
Percentage of empty pollen grains in the same individual rose for two different seasons.

[89]

criterion for hybridity. The percentage of empty grains in pollen samples from thirty-six different individuals for two, and sometimes three, seasons is given in table 8. In the last column the difference between the highest and lowest percentage is given for each individual. Among the dipbid plants considered there are nine in which there was a greater amount of aborted pollen in 1929 than in 1928 and eight in which there was less. All these plants come into flower about the same time, which indicates that cold at least has a negligible effect upon reduction division in the pollen mother cells of these roses. The mean variation in the percentage of aborted pollen for two seasons in twenty diploid individuals is only 3.8 percent.

Among the ten tetraploids listed in table 8, two show an abnormally large variation in the amount of aborted pollen in two consecutive seasons. These are R. relicta Erlanson, which was referred to in this respect above, and the other is a plant of R. polyanthema Lunnell from Missouri. The sample of pollen from the latter had been discarded before it was realized that the amount of sterility was so much greater in 1929². There is a possibility that the bud from which the pollen was taken in 1929 was parasitised by weevils or otherwise diseased. If these two plants be excepted the mean difference in sterility for eight tetraploid plants in two seasons is 3.9 percent.

The difference in the percentage sterility in hexaploids from year to year is necessarily small since they exhibit so little. Similarly, triploid plants produce so few morphologically good pollen grains that the variation in sterility shown by them in different years is slight.

OVULE STERILITY

As stated above, it has been found by several workers that morphologically perfect pollen grains are sometimes unable to function in fertilization. In his work on F1 hybrids from Nicotiana rusticana X N. paniculata, EAST (1921) found that 2-3 percent of the pollen grains appeared to be good, but that less than 0.1 percent were functional, and that about 4 percent of the ovules were functional. He suggests that sterile female gametes are apt to coincide with the normally non-functional megaspores, and, further, that others are able to develop because they have a more favorable environment in the ovule than that to which the microspores are subjected. KRISTOFFERSON (1926) when studying interspecific hybrids in Malva found that the external nature of the pollen alone was an unreliable test [90] for hybridity or sterility. F1 hybrids of Malva crispa X M. neglecta showed apparently 80 percent well-formed pollen, yet this would not function when used to pollinate M. neglecta in a backcross. He therefore examined the number of seeds in four fruits on each plant and calculated the fertility as the ratio between the number of seeds and the number of carpels. In hybrids of Malva neglecta X M. pusilla he found that the F1 fertility varied from 65 percent to 90 percent, and F2 from 55 percent to 95 percent, giving, he thought, a di-hybrid segregation for sterility.

It is true for Rosa as for Fragaria (DARROW 1927) that the amount of seed set varies with the position of the flower. Therefore whenever possible the contents of terminal hips have been examined in determining ovule sterility in wild roses. It follows that the ovule sterility percentages shown in table 9 are about the minimum for each plant.

In order to determine the percentage of functional female gametes the proportion of undeveloped ovaries to the total number of ovaries in four hips was calculated. This is considered to be a measure of ovule fertility and is given in column three of table 9. Many plump achenes contain no embryos. These empty achenes can be separated out by floating in water (CROCKER 1926), those with embryos sinking after they become thoroughly wet. The proportion of the number of empty achenes (parthenocarpic fruits) to the total number of achenes gives some measure of zygotic sterility and is recorded in column two of table 9. It was interesting in this connection to find that all achenes of R. palustris float, a faculty which is no doubt of value to a plant which grows beside water or in bogs. The zygotic and ovule sterility is given in table 9 for fourteen wild roses selected at random. In the last column the difference between percentage ovule sterility and percentage pollen sterility is given for each individual.

The unexpected result of these few measurements is that ovule sterility appears to vary inversely with pollen sterility to some extent. The ovule sterility is larger in six plants with less than 20 percent of aborted pollen grains. The other seven native plants have over 20 percent of sterile pollen and in six of these the ovule sterility is less than the pollen sterility. In the two plants showing over 20 percent more ovule than pollen sterility, this discrepancy may be due in part to self-sterility. One of these plants is an F1 of R. subblanda X R. blanda and the other is a vigorous hexaploid bush of R. muriculata Greene. There is only one plant of R. muriculata in our collection, and it produces many flowers but very few hips. In 1929 one bud on this plant was pollinated with R. acicularis Lindl. var. Bourgeauiana Crépin, and when ripe the hip contained a hundred plump achenes and no aborted ovules; sixty-one contained embryos. Some plants of [91] R.blanda have set more fruit in hips pollinated with R. acicularis than in hips which were self-pollinated.

TABLE 9
Percentages of ovule, zygotic and pollen sterility compared.

 *Figures given by HARRISON and BLACKBURN 1927.

The plants of R. suffulta Greene and R. rudiuscula Greene show about equal amounts of ovule and pollen sterility; but in a specimen of the related R. Bushii Rydberg there appears to have been an elimination of sterile megaspores. R. relicta exhibits a consistently high sterility in pollen, eggs and zygotes. The three plants of R. blanda examined all show a. low percentage of aborted ovules together with about 50 percent of empty microspores. Plant 3502/A produced six hips from six castrated buds in 1927 (ERLANSON 1929); buds castrated in 1929 gave no hips. Successful fruit production in R. blanda after castration has been observed in a few other bushes, as well as in some members of culture 3753 to which plant 9/N8 belongs. Therefore there is a possibility that the high ovule fertility in these plants of R. blanda, with semi-sterile pollen, may be partially due to facultative apomixis, in conjunction with the elimination of sterile megaspores. There are numbers of R. blanda cultures in our collection and the flowers should have been supplied with plenty of functional pollen. The two plants of culture 3502 are the only examples found so far in which the sum of the abortive ovules and empty achenes is less than the percentage of sterile pollen. The section Caninae exhibits much apomictical [92] seed production. I have appended to table 9 the zygotic and ovule sterility found in R. rubrifolia, a member of this section. The pollen sterility of this individual is unknown, but HARRISON and BLACKBURN (1927) give it as 50 to 70 percent for the species. Only 12 percent of the ovules were aborted.

GENERAL DISCUSSION

It is apparent from research as to the causes of pollen and ovule sterility in plants that sterility in itself is a symptom which may be due to one or more of several causal agents. Hybridization often, though not invariably, results in a disturbance of the regularity of reduction division, and non-conjunction, lagging of chromosomes during meiosis, and polyspory appear. The same phenomena have been induced by the influence of heat and cold (STOW 1926-1927, SAKAMURA and STOW 1926, BLAKESLEE and CARTLEDGE 1927), by insect infection (KOSTOFF and KENDALL 1929), and by X-ray treatment. GOODSPEED (1929) obtained meiotic irregularities and pollen sterility, similar to those found in hybrids, by subjecting either the parental germ-cells or the pollen used to X-rays.

Triploidy, whenever it appears, causes a high degree of sterility in both male and female gametes. This is true whether the triploid plants are trivalent in synapsis and have arisen from the fusion of a diploid and a tetraploid gamete in the same species as in Datura (BLAKESLEE and CARTLEDGE 1927), Lycopersicum (LESLIE 1928), Crepis (NAVASHIN 1929) and Zea (McCLINTOCK 1929); or whether only two sets of homologous chromosomes are present and the triploid has arisen from hybridization between different species as in Triticum (SAX 1922, THOMPSON 1926), Rubus (CRANE and DARLINGTON 1927), Nicotiana (GOODSPEED and CLAUSEN 1927, CHRISTOFF 1928) and Rosa (ERLANSON 1929). An abnormal chromosome constitution increases sterility, whether the abnormality be the presence of one or more extra chromosomes (BLAKESLEE and CARTLEDGE 1927, LESLIE 1928, McCLINTOCK 1929) or a deficiency (HUSKINS 1927, 1928).

There are also genes responsible for sterility, as the "bad pollen inducers" in Datura (BLAKESLEE and CARTLEDGE 1927). These factors segregate in F2. SALAMAN (1910) found that in the potato male sterility is a dominant Mendelian character. A curious inheritance of semi-sterility in descendants from a cross between two species of Stizolobium was described by BELLING (1914). The F1 plants had 50 percent of both male and female gametes non-viable. Half of the F2 progeny resembled the F1 hybrids in the semi-sterility of their gametes, the rest had perfect pollen and good ovules and gave fully fertile offspring. The pollen in wild forms of [93] Mentzelia intermediate between M. multiflora and M. nuda was found by CLEMENTS and PENLAND (1926-1927) to vary from 5 percent to 95 percent sterile. This shows that segregation of sterility types occurs in nature. The wide variation in pollen sterility found by the writer in most diploid and tetraploid North American wild roses suggests that genetic factors are at least partially responsible.

The percentages of defective pollen grains found in Vitis by DORSEY (1914) reveal a considerable range in variation for Vitis vulpina (from o percent to 45 percent), though the sterility is under 5 percent for both male and female vines. Cultivated varieties exhibited an average of about 23 percent of defective pollen, whether they were of hybrid origin or not. DORSEY concludes that in this genus pollen sterility is not necessarily caused by hybridization, but may be a step towards functional dicliny.

The effect of polyploidy upon sterility is not always the same. DARLINGTON (1928) pointed out that there seems to be an inverse correlation between the sterility of a diploid and the fertility of a tetraploid to which it gives rise. JØRGENSON (1928) by inducing somatic doubling in diploids obtained tetraploid Solana which were less fertile than the diploids. The pollen of tetraploid plants of Datura is less fertile than that of diploids (BLAKESLEE and CARTLEDGE 1927). LONGLEY (1927) found that tetraploid Rubus forms, both wild and cultivated, had a greater percentage of pollen sterility than the diploid forms, while the higher polyploids were less sterile. Many instances are now known of the production of a fertile tetraploid from a sterile diploid hybrid and the literature has been reviewed often. MEURMAN (1929) found that in Prunus laurocerasus the basic chromosome number for the genus was present about 22 times, and that in such a high polyploid tetrad formation and viability of gametes are unaffected by the lagging of some of the chromosomes. The pollen in diploid, tetraploid and hexaploid species of Triticum is equally fertile, but pentaploid hybrids are less sterile than triploid hybrids (SAX 1922). Similarly LONGLEY (1927) gives figures showing that pentaploid forms in Rubus have slightly less poor pollen than the triploids.

In the American roses of the Cinnamomeae the mean pollen sterility is about the same in related diploid and tetraploid species. There are, however, some tetraploid plants which show a very high degree of sterility. These latter may have originated from an unreduced egg in a diploid, fertilized by pollen from a tetraploid. A tetraploid Rubus obtained in this way in the F1 progeny of Rubus rusticanus inermis X R. thyrsiger was highly fertile (CRANE and DARLINGTON 1927), but it is conceivable that sometimes maturation in such a tetraploid would be irregular.

[94]

In considering female fertility in plants we are forced usually to deal with zygotic fertility, since it is often impossible to obtain large numbers of megaspores or even of embryo-sacs for statistical studies. Female fertility is affected by the factors enumerated above as affecting pollen sterility, and further by self-sterility, by apomixis and by purely environmental conditions affecting the nourishment of the fruit, such as unsuitable soil (SCHUSTER 1926). Many sterile megaspores are no doubt eliminated after meiosis among those that normally develop no further. If the pollen is highly sterile some normal embryo-sacs may fail to develop owing to lack of sufficient functional microspores, as found in wheat hybrids by WATKINS (1923),

Self-sterility, another important factor in causing reduced seed production, is particularly widespread in the family Rosaceae (SUTTON 1918, CRANE 1925). It is undoubtedly present to some degree in some wild roses. This again is a manifestation of many complex genetic and physiological factors (EAST and MANGELSDORF 1925).

Since some roses are facultatively apomictical, a plant may be judged more fertile than is actually the case, if seed production be taken as a criterion.

CONCLUSION

GOODSPEED (1929) has recently pointed out that the successful production of fertile gametes in plants depends upon nicely balanced physiological growth factors arranged in a particular manner in the living organism. This balance can be upset by a number of agents, environmental, physiological, or hereditary. Hybridity is only one of these agents and its effect is conditioned by the compatibility of the parents.

HURST'S contention (HURST 1923, 1928) that either fertile or sterile offspring will result from hybridization between diploid rose forms is not borne out by the analysis of the pollen in the few F1 hybrids thus far obtained by the writer. There is a wide variation in the amount of pollen sterility in most cultures of rose species. Sterile pollen up to 25 percent appears to be characteristic of relatively fertile plants in any species and maybe due to conditions affecting growth and nutrition of the pollen mother cells.

It is probable that a wild rose plant with over 70 percent of the pollen grains aborted, and which sets little or no fruit, is an F1 hybrid. Partially sterile plants and even those showing a relatively high production of good pollen may also be hybrids. There is only one way by which we may come to know much about the amount of hybridization among roses in nature, [95] and that is to grow large progenies of seedlings from single wild seed parents, and to make interspecific crosses artificially. 1 do not believe that it is possible to predict what the result will be in any species cross owing to the high degree of heterozygosis and consequent variation within each species. Dependence upon the degree of sterility alone as sufficient evidence for or against hybridity is unwarranted by the facts that have been assembled.

LITERATURE CITED

• BELLING, J., 1914 The mode of inheritance of semi-sterility in the offspring of certain hybrid plants. Z. indukt. Abstamm.-u. VererbLehre. 12: 303-342.
• BEST, G. N., 1887 Remarks on the group Carolinae of the genus Rosa. Bull. Torrey Bot. Cl. 14: 253-256.
• BLAKESLEE, A. F., and CARTLEDGE, J. L., 1927 Sterility of pollen in Datura. Mem. Hort. Soc. N.Y. 3: 305-312.
• BOULENGER, G. A., 1929 Sur les hybrides des roses de l'Europe Centrale et Occidentale. Revue des Questions scientifiques. 15: 251-266.
• CHRISTOFF, M., 1928 Cytological studies in the genus Nicotiana. Genetics 13: 233-277.
• CLEMENTS, F. E., and PENLAND, C. W., 1926-1927 Hybridization in Nature. Carnegie Year Book 26: 310-311.
• CRANE, M. B., 1925 Sell-sterility and cross-incompatabiity in plums and cherries. J. Genet. 15: 301-322.
• CRANE, M. B., and DARLINGTON, C. D., 1927 The origin of new forms in Rubus. Genetica 9: 241-276.
• CRÉPIN, F., 1889 Recherches sur l'état de developpement des grains de pollen dans diverse espèces du genre Rosa. Bull. Soc. Bot. Belg. 28: Pt. 2. 114-125.
• CROCKER, W., 1926 After-ripening and germination of rose seeds. Am. Rose Annual 11: 34-40.
• DARLINGTON, C. D., 1928 Studies in Prunus, land II. J. Genet. 19: 213-256.
• DARROW, G. M., 1927 Sterility and fertility in the strawberry. J. Agric. Res. 34: 393-411.
• DIGBY, L., 1912 The cytology of Primula kewensis and other related Primula hybrids. Ann. Bot. Oxford 26: 357-388.
• DORSEY, M. J., 1914 Pollen development in the grape with special reference to sterility. Minnesota Agric. Expt. Sta. Bull. 144.
• EAST, E. M., 1921 A study of pollen sterility in certain hybrids. Genetics 6: 311-365.
• EAST, E. M., and MANGELSDORF, A, J., 1926 Studies on self-sterility. VII. Heredity and selective pollen tube growth. Genetics 11: 466-481.
• ERLANSON, E. W., 1929 Cytological conditions and evidences for hybridity in North American wild roses. Bot. Gas. 87: 443-506.
• 1930 A group of tetraploid roses in central Oregon. Bot. Gas. (in press).
• GOODSPEED, T. H., 1929 Cytological and other features of variant plants produced from X-rayed sex-cells of Nicotiana Tabacum Bot. Gaz. 87: 563-582.
• GOODSPEED, T. H., and CLAUSEN, R. E., 1927 Interspecific hybridization in Nicotiana. VI. Cytological features of sylvestris-Tabacum hybrids. Univ. Calif. Pub. Bot. 11: 127-140.
• HARRISON, J. W. H., and BLACKBURN, K. B., 1927 The course of pollen formation in certain roses, with some deductions therefrom. Men. Hort. Soc. N. V. 3: 23-32.
• HEILBORN, O., 1928 Zytologische Studien über Pollensterilitat von Appelsorten. Svensk bot. Tidskr. 22: 185-199.
• HURST, C. C., 1925 Chromosomes and characters in Rosa and their significance in the origin of species. Experiments in Genetics pp. 534-550. Cambridge: Cambridge Univ. Press.
• ——— 1928 Differential polyploidy in the genus Rosa L. Verh. des V internat. Kongr. f. Vererbungswissenschaft, Berlin.
• ——— 1927. Supplementband II, Z. indukt. Abstamm.- u. VererbLehre. 866-906.
• ——— 1929 The Genetics of the Rose. Rose Annual 23: 37-64.
• HUSKINS, C. L., 1927 On the genetics and cytology of fatuoid or false wild oats. J. Genet. 18: 315-364.
• ——— 1928 On the cytology of speltoid wheats in relation to their origin and genetic behaviour. J. Genet. 20: 103-122.
• ——— 1929 Criteria of hybridity. Science 69: 399-400.
• JEFFREY, E. C., 1929 Recent discussions of the reduction division in Drosophila melanogaster. Science 70: 579-580.
• JØRGENSEN, C. A., 1928 The experimental formation of heteroploid plants in the genus Solanum. J. Genet. 19: 133-211.
• KOSTOFF, DONTCHO, and KENDALL, JAMES, 1929 Irregular meiosis in Lycium halimifolium Mill. produced by gall mites (Eriophyes). J. Genet. 21: 113-115.
• KRISTOFFERSON, K. B., 1926 Species crossing in Malva. Hereditas 7: 233-354.
• LESLIE, JAMIES W., 1928 A cytological and genetical study of the progenies of triploid tomatoes. Genetics 13: 1-43.
• LONGLE, A. E., 1927 Relationship of polyploidy to pollen sterility in the genera Rubus and Fragaria. Mem. Hort. Soc. N, Y. 3: 15-17.
• McCLINTOCK, B., 1929 A cytological and genetical study of triploid maize. Genetics 14: 180-222.
• MEURMAN, O., 1929 Prunus laurocerasus a species showing high polyploidy. J. Genet. 21: 85-94.
• NAVASHIN, M., 1929 Studies on polyploidy. I. Cytological investigations on triploidy in Crepis. Univ. Calif. Pub. Agric. Sci. 2: 377-400.
• NEWTON, W. C. F., and PELLEW, C., 1929 Primula kewensis and its derivatives. J. Genet. 20: 405-467.
• NICOLAS, J. H., 1927 Sterility encountered in rose breeding. Mem. Hort. Soc. N. Y. 3: 55-57.
• SAKAMURA, T., and STOW, I., 1926 Ueber die experimental veranlasste Entstehung von keimfähigen Pollenkörnern mit abweichenden Chromosomenzahlen. Jap. J. Bot. 3: 111-137.
• SALAMAN, R. N., 1910 Male sterility in potatoes, a dominant Mendelian character, with remarks on the shape of pollen in wild and domestic varieties. J. Linn. Soc. Bot. 39: 301-312.
• SAX, K., 1922 Sterility in wheat hybrids. II. Chromosome behavior in partially sterile hybrids. Genetics 7: 513-552.
• SCHUSTER, C. E., 1926 Field production of Ettersburg 121 strawberry. Circ. Oregon Agric. Expt. Sta. 67.
• SCHWERT-SCHLAGER, J., 1910 Die Rosen des südlichen und mittleren Frankenjura. München.
• STOW, I., 1926-1927 A cytological study on pollen sterility in Solanun tuberosum. Jap. J. Bot. 3: 217-238.
• SUTTON, 1., 1918 Report on tests of self-sterility in plums, cherries and apples at John Innes Horticultural Institution. J. Genet. 7: 281-300.
• THOMPSON, W. P., 1926 Cytological behavior in triploid wheat hybrids J. Genet. 17: 43-48.
• WATKINS, A. E., 1924 Genetic and cytological studies in wheat. I. J. Genet. 14: 129-171.
• WILLMOTT, E. A., 1914 The genus Rosa. 2 Vols. London.