Compatibility relationships in distylous bluets: Houstonia serpyllifolia and H. longifolia (Rubiaceae).
Populations of distylous species typically consist of plants with two distinctive floral forms, or morphs, which differ in the positions of their stigmas and anthers. The flowers of distylous species are often tubular, and placement of stigmas and anthers in the two morphs is reciprocal, such that the stigmas of long-styled plants, or "pins," and the anthers of short-styled plants, or "thrums," are held near the top of the corolla tube, whereas thrum stigmas and pin anthers are positioned midway down the corolla tube. Other morphological differences are often associated with the dimorphism in style and anther height, including differences in the size and number of pollen grains and in the length of stigmatic papillae. Differences in floral morphology are usually associated with an incompatibility system, with plants of each morph setting seed only when pollinated by the other morph, although some species apparently lack an incompatibility system (Ganders, 1979; Dulberger, 1992).
In this study we investigated compatibility relationships between floral morphs of two distylous species, Houstonia serpyllifolia and H. longifolia (sensu Terrell, 1991; Rubiaceae), in preparation for further work on the effects of compatible and incompatible neighbor abundances on fruit-set (Beliveau, 1996). We expected these species to show the strong heteromorphic incompatibility typical of distylous species, which would allow us to classify by simple inspection of their flowers whether a plant's neighbors are compatible or incompatible. Previous studies on distylous Houstonia caerulea (Ornduff, 1977; Wyatt and Hellwig, 1979) and Hedyotis nigricans (Bit Bahadur, 1968; Levin, 1974) had shown that the floral dimorphism was accompanied by a strong heteromorphic self-incompatibility system in those species. The genus Houstonia includes an array of homostylous, distylous, short-styled (presumably self-compatible) and even cleistogamous species (Lewis, 1962), so it seemed important to test for heteromorphic self-incompatibility in our species.
The subjects of this study, Houstonia serpyllifolia and H. longifolia, are small-flowered, distylous, perennial herbs. Otherwise, the two bear relatively little resemblance to each other and are placed in different subgenera within Houstonia. Our taxonomy follows Terrell (1991), but many authors place species of Houstonia into the large tropical genus Hedyotis (e.g., Fosberg, 1954; Lewis, 1962; Ornduff, 1977). Houstonia serpyllifolia Michaux, in subgenus Houstonia, has delicate procumbent stems that root at the nodes. Plants form carpets of small (to 1 cm in diam), blue, salverform flowers that are held approximately 10 cm above the ground. Plants occur in cool moist areas, often along stream banks or seepages, and are generally restricted to elevations above 1000 m in the southern and central Appalachian Mountains (from Georgia to Pennsylvania). In floral morphology, the species resembles H. caerulea L., a closely-related distylous perennial whose reproductive biology has been studied by Ornduff (1977, 1980), Wyatt and Hellwig (1979) and Grimaldi (1988). Houstonia serpyllifolia flowers from early April to mid-June in Highlands, North Carolina, where our study was conducted.
Houstonia longifolia Gaertner (var. longifolia of Terrell, 1991; = var. glabra of Terrell, 1959) belongs to subgenus Chamisme. It is an erect perennial with stems about 15 cm tall, linear to lanceolate leaves and small ([approximately]0.5 cm wide), light purple, funnel-form flowers that are less showy than those of H. serpyllifolia. It occurs mostly on granitic rock outcrops, in poor soil and in disturbed areas from South Carolina to West Virginia. Houstonia longifolia exhibits limited clonal growth, with plants ranging from 1 to 100 reproductive stems, but large plants are rare. It flowers from early June through July in the Highlands area. Another distylous species of the same subgenus, H. purpurea L., often occurs near populations of H. longifolia. Houstonia purpurea is similar to H. longifolia, but has wider, lanceolate to ovate leaves and slightly more robust flowers. It occurs in deciduous woods and forest edges, often in disturbed soil and flowers earlier than H. longifolia, although the flowering periods of the two species overlap for 2-3 wk in June. Putative hybrids between H. longifolia and H. purpurea have been reported (Lewis and Terrell, 1962). Several plants of H. purpurea co-occurred in the population of H. longifolia that we studied. Therefore, we performed some pollinations to assess the compatibility of interspecific crosses between the two species.
Pollen size, viability and floral lifespan. - Pollen was stained with aniline blue lactophenol to increase visibility and indicate viability (Kearns and Inouye, 1993). Pollen diameters were measured for 100 grains from 10 individuals of each morph of Houstonia serpyllifolia in May 1990, for 25 grains from each of 6 plants of each morph of H. longifolia in July 1991 and for 24 grains from each of 3 plants of each morph of H. purpurea in July 1991. For all species, 100 grains from each of the individual plants were scored for viability, indicated by dark staining with the aniline blue dye.
To estimate floral lifespan, flowers of Houstonia serpyllifolia were tagged as they opened and followed until the corolla abscised. Floral lifespan was followed in the greenhouse in December 1990 and in a natural population at Whiteside Mountain, near Highlands, in June 1991. Mean floral lifespan was not quantified for H. longifolia and H. purpurea, but corollas typically abscised
after the second day of flowering for plants of H. longifolia.
Compatibility relationships. - Hand-pollinations were performed in 1991 to assess intermorph and intramorph compatibility of Houstonia serpyllifolia and H. longifolia. For H. serpyllifolia, 216 intermorph crosses, 153 intramorph crosses and 154 unpollinated controls were distributed among flowers on 129 small plants (71 pins and 58 thrums) grown from seeds harvested from the population at Whiteside Mountain. The plants were grown outdoors in 5 cm x 6 cm plastic cells in flats at the Highlands Biological Station. All flats were covered with bridal veil (mesh size = 10 threads/cm) to exclude insects. Pollinations were accomplished by removing an anther from a flower on another plant with forceps and brushing it against the stigma until the yellow pollen was apparent on the stigma. Each flower was marked with a jewelry tag that indicated the type of cross and the identity of the pollen source. The corollas of thrum flowers had to be split to the level of the stigma to allow pollination. Likewise, the corollas of pin flowers often had to be split to obtain anthers for pollination. Fruit-set in early compatible crosses was high, indicating that splitting the corolla did not hinder performance (pers. obs.). Intermorph (i.e., legitimate: pin x thrum or thrum x pin) crosses and xenogamous intramorph (i.e., illegitimate: pin x pin or thrum x thrum) crosses were made, and a similar number of flowers were left unpollinated as controls. Most crosses were done on flowers that appeared fully receptive and 1-5 d old. Some crosses were performed on buds and old flowers with fading corollas to determine their behavior, and flower age was noted in these cases. Treatments were distributed approximately equally among plants, in proportion to flower availability. Thus, plants that produced several flowers received all treatments. Fruit capsules were scored as "developed," "large aborted" (ovaries which had swollen large enough to have held one or more seeds, but which contained no viable seeds), "small aborted" (ovaries which showed some swelling, but which were not large enough to have contained even one viable seed) or "undeveloped." Seed-set, the number of seeds per fruit, was scored for developed capsules.
We performed crosses of Houstonia longifolia in July 1991 in a natural population near the intersection of Bowery and Horse Cove Roads in Highlands. Hand-pollinations were done on three plants of each morph, which had been previously bagged with bridal veil. Crosses were made within and between morphs, and control flowers were left unpollinated. Each treatment was performed on all plants, with the various classes of pollination being distributed approximately equally among the plants. Treatments were performed on separate stems or branches, which were labeled with jewelry tags. On the days when pollinations were done, each treated flower was marked with enamel paint on its sepals. Untreated flowers were left on the stems but not painted. Fruits were scored as developed or undeveloped, and seed-set was scored for developed capsules.
Interspecific compatibility. - The presence of a co-flowering congeneric species, Houstonia purpurea, growing interspersed in the H. longifolia population, raised the question of interspecific compatibility. Unfortunately, relatively few interspecific crosses could be made because H. purpurea was well past peak flowering when pollinations were done. In the area where the plants used for pollinations grew, only one plant of H. purpurea, a pin, had enough buds remaining to justify bagging for crosses. Pollen for interspecific crosses using H. purpurea thrums as the paternal parent was collected from unbagged plants.
For both Houstonia serpyllifolia and H. longifolia, we used the GLIMMIX macro of the Mixed procedure to analyze the effects of plant morph and pollination treatment on fruitset (Littell et al., 1996). Type-1 error arising from multiple treatment comparisons was corrected by using a Tukey adjustment (SAS Institute Inc., 1996). We compared seed-set between the two morphs using ANOVA, and comparisons of seed-set among pollination treatments were made using the Student-Newman-Keuls multiple range test (proc GLM; SAS Institute Inc., 1989).
Pollen size, viability and floral lifespan. - Haustonia serpyllifolia expresses the dimorphism of pollen size often found in distylous plants: thrum pollen is significantly larger (F = 826.3, P [less than] 0.0001), although the distributions of pollen sizes produced by pins and thrums overlap (pin pollen diameter = 7.7 [+ or -] 0.4 [[micro]meter] (mean [+ or -] SD), thrum pollen diameter = 9.2 [+ or -] 0.7 [[micro]meter]). Differences in stigmatic papillae appear to complement the dimorphism in pollen size, with the stigmatic papillae of pins appearing longer (pers. obs.). Pollen viability was high, with [greater than]95% of grains staining darkly with aniline blue dye. In natural populations, mean floral lifetime (from opening to corolla abscission) was about a week (mean = 7.8 d for flowers that received little pollination). In the greenhouse, however, corollas of unpollinated flowers could remain attached for more than 14 d.
Thrum pollen is also significantly larger in both Houstonia longifolia (F = 393.6, P [less than] 0.0001) and H. purpurea (F = 228.5, P [less than] 0.0001), with the distributions of pollen sizes produced by pins and thrums overlapping (H. longifolia: pin pollen diameter = 34.1 [+ or -] 3.0 [[micro]meter], thrum pollen diameter = 41.5 [+ or -] 3.4 [[micro]meter]; H. purpurea: pin pollen diameter = 37.5 [+ or -] 2.7 [[micro]meter], thrum pollen diameter = 45.7 [+ or -] 3.7 [[micro]meter]). More than 95% of pollen grains from both morphs of both species stained darkly with aniline blue dye, indicating high pollen viability. Flowers of H. longifolia typically lasted 1-2 d, and styles were sometimes brittle and ready to abscise on the second day.
Compatibility relationships. - Both Houstonia serpyllifolia and H. longifolia displayed fairly typical dimorphic incompatibility systems. For both species, intermorph crosses set significantly more fruits and more seeds per fruit than intramorph crosses or unpollinated controls (Tables 1, 2). Fruit-set of intramorph crosses was significantly higher than that of unpollinated controls, but seed-set did not differ significantly between the two groups. In intermorph crosses, the two morphs did not differ significantly in fruit-set for either species (t = 0.27, P = 0.7880 for H. serpyllifolia; t = 1.90, P = 0.0607 for H. longifolia). Seed-set of capsules from intermorph crosses also failed to differ significantly between the two morphs of both species (F = 2.67, P = 0.1061 for H. serpyllifolia; F = 1.48, P = 0.2273 for H. longifolia). Compatibility did not appear to be affected by the age of H. serpyllifolia flowers; both buds and flowers nearing corolla abscission set fruits following intermorph pollination (9 of 9 buds and 7 of 9 old flowers set fruit) but were unlikely to set fruits following intramorph pollination (1 of 4 buds and 0 of 6 old flowers set fruit).
Morphs of the two species differed slightly in their response to intramorph (illegitimate) crosses, but the differences did not prove to be significant. For Houstonia serpyllifolia, pins set more fruits and more seeds per fruit than thrums following intramorph crosses, suggesting that pins might be slightly more intramorph-fertile (Table 1). The differences between morphs in intramorph crosses were not significant, however, for either fruit-set (t = 1.72, P = 0.0881) or seed-set (F = 3.25, P = 0.0850). For H. longifolia, thrums set fruits at twice the rate of pins, but pins set more seeds per fruit. Again, the differences did not [TABULAR DATA FOR TABLE 1 OMITTED] [TABULAR DATA FOR TABLE 2 OMITTED] prove significant for either fruit-set (t = -1.32, P = 0.1909) or seed-set (F = 4.65, P = 0.0540).
There were several partially developed or "aborted" capsules formed in the Houstonia serpyllifolia hand-pollinations, especially in the intramorph and unpollinated treatments. Pins produced a higher percentage of large but seedless capsules, whereas thrums had more capsules that were only slightly swollen. The differences between morphs were not significant, however, for either type of "aborted" fruit. Partial development or abortion of capsules was less common in hand-pollinations of H. longifolia than in H. serpyllifolia. Swollen capsules without apparent seeds were only produced by two unpollinated thrum flowers. The causes of partial development or abortion of capsules cannot be determined from this study.
Unpollinated control flowers of Houstonia serpyllifolia produced an unexpectedly high percentage of fruits, with pins showing 14% fruit-set and thrums showing 7%. These proportions conflict with earlier data (see Discussion), and we suspect that they represent a background level of pollination by thrips, which inhabited the flowers. Fruit-set in the unpollinated controls of H. longifolia was very low. None of the 68 pin and 15 thrum flowers that were marked with paint produced fruits. A further 260 pin and 119 thrum flowers were left unpollinated on stems that received pollination treatments; of these only two pins (0.8%) produced fruits, while four thrums (3.4%) produced fruits.
Interspecific compatibility. - Interspecific intermorph crosses, involving plants of Houstonia longifolia pollinated by H. purpurea, did not differ significantly from intraspecific intermorph crosses in fruit-set or seed-set per fruit, but they were significantly lower in average number of seeds set per pollination. All three plants of each morph developed healthy looking fruits from most of the interspecific intermorph pollinations. The seeds from interspecific crosses, however, tended to be small or extremely slow to mature; most were still soft and light brown when they were collected late in October, a month after seeds from legitimate intraspecific crosses had matured. Most were probably inviable, although a few large seeds were noted. Interspecific intramorph crosses were not significantly different from intraspecific intramorph crosses in any category (percent fruit-set or average number of seeds per fruit or per pollination).
A few pollinations were also done on one plant of Houstonia purpurea, a pin that had several buds remaining when the H. longifolia pollinations were begun. On this plant, 5 of 6 intermorph crosses with other H. purpurea plants developed fruits and seeds, whereas 1 of 4 intramorph crosses produced a fruit. In interspecific crosses with H. longifolia, 6 of 8 intermorph crosses produced small capsules with no apparent seeds and 1 of 2 intramorph crosses produced a capsule with a single small seed. Overall, the pattern for this plant was similar to that seen in H. longifolia, although the plant filled fewer seeds from interspecific pollinations.
Hand-pollinations confirmed that Houstonia serpyllifolia and H. longifolia possess a heteromorphic genetic incompatibility system typical of distylous species. For both species, intermorph (legitimate) crosses produced significantly more fruits and more seeds per fruit than intramorph (illegitimate) crosses or unpollinated controls. Intramorph crosses produced more fruits, but no more seeds per fruit than unpollinated controls. The two morphs of each species did not differ significantly in the number of seeds set in fruits from intermorph crosses.
In both species some fruits were formed following intramorph pollination, but these fruits usually contained few seeds. Although differences in fruit-set and seed-set between intra-morph crosses and unpollinated controls, and differences in seed-set between morphs, were not significant for either species, we note the following trends for comparison with other studies of species in the genus. For Houstonia serpyllifolia, pins set more fruits and more seeds per fruit than thrums following intramorph crosses or no pollination, suggesting that pins might be slightly more intramorph- or self-fertile. For H. longifolia, thrums set a higher percentage of fruits, but pins set more seeds per fruit following intramorph pollination.
The success of intramorph and unpollinated flowers of Houstonia serpyllifolia was higher than might be expected for a strong heteromorphic incompatibility system (Ganders, 1979). We suspect that pollination rates of both these treatments are overestimated in this study and would be closer to zero if thrips had not inhabited the flowers of our study plants. Thrips probably transferred small amounts of compatible pollen by moving between nearby flowers, particularly when pins and thrums were grown in close proximity. In several cases, neighboring pin and thrum plants of H. serpyllifolia flowered simultaneously. If suspect data from just one pair of adjacent pin and thrum plants are removed from the analysis, 17 fewer capsules would result from intramorph pollination or unpollinated treatments, and the percentages of developed fruits would decrease (pin x pin from 27 to 19%, thrum x thrum from 15 to 13%, unpollinated pin from 14 to 9% and unpollinated thrum from 7 to 4%) and the average number of seeds set per capsule and per pollination would decrease. Ornduff (1980) also noted thrips in the flowers of H. caerulea and thought it possible that they effected some pollination. Thrips are not often considered to be important as pollinators, although their efficacy has rarely been tested. Studies have, however, shown them to be important to the pollination of several species (Baker and Cruden, 1991).
The view that fruit-set in intramorph crosses and unpollinated flowers was inflated by thrip activity in the above crosses is supported by data from an earlier set of pollinations performed on plants of Houstonia serpyllifolia that flowered early' in the greenhouse during the previous winter. The flat that flowered contained only thrums, so data on pins are lacking. The absence of pin flowers, however, gave thrips no opportunity to transfer pollen between the morphs. In that set of pollinations only 1 of 73 intramorph crosses (1.4%) and 1 of 211 unpollinated flowers (0.5%) developed capsules. Both of these capsules were small and contained no viable seeds. Pollen from pins which flowered outside the greenhouse was used to perform intermorph crosses on 16 flowers in the flat, and 15 (93.8%) of these pollinations produced capsules. Moreover, small plants that were open to insect pollination, but separated from sources of compatible pollen by at least 100 m, showed lower levels of fruit-set (pins 8%, thrums 5%; Beliveau, 1996) than those reported in Table 1.
For Houstonia longifolia, unpollinated flowers rarely produced any fruits or seeds, indicating that thrips probably transferred little compatible pollen to flowers. Plants of H. longifolia were bagged individually for hand-pollinations, however, and were not immediately proximate to compatible plants, as plants of H. serpyllifolia were.
Thrips should have affected unpollinated and intramorph-pollinated flowers similarly, so we should be able to correct for their effect on intramorph cross fertility by subtracting the fertility of unpollinated controls. Performing this operation to get a rough correction for thrip effects, we get the following levels of intramorph fruit-set: for Houstonia serpyllifolia: pins 13%, thrums 8%; and for H. longifolia: pins 8%, thrums 15%. These differences indicate a low level of intramorph fertility. Differences in fruit- and seed-set between the morphs in intramorph crosses were not significant for the original data, and doubtless remain so for "thrip-corrected" data. In the closely related H. caerulea, Ornduff (1977) found thrums to be more self- and intramorph-compatible than pins, and Wyatt and Heliwig (1979) detected a weak trend in the same direction. For the more distantly related Hedyotis nigricans (Lamarck) Fosberg, Bir Bahadur (1968) found pins to be more self- (but not intramorph-) compatible, but no plump seeds were produced by either morph following Levin's (1974) self- and intramorph-pollinations. When differences in intramorph compatibility have been seen in distylous species, pins have more often shown greater intramorph compatibility than thrums (Ganders, 1979).
Self-incompatibility systems are often "leaky," and partial self-compatibility is seen in many species that are considered to have self-incompatibility systems (Lloyd and Schoen, 1992; Levin, 1996). Such limited intramorph fertility is not unusual in distylous species (Ganders, 1979; Lloyd and Webb, 1992), but little is known about the genetic basis of self-incompatibility or the site of rejection for most heteromorphic incompatibility systems. Although some studies have been done on the site of rejection for species in the Rubiaceae (Bawa and Beach, 1983; Murray, 1990), we know of no such work on Houstonia or Hedyotis.
Interspecific pollinations between compatible morphs of Houstonia longifolia and closely-related H. purpurea resulted in fruit-set, but most seeds in these fruits did not mature. Apparent hybridization and introgression between H. longifolia and H. purpurea have been reported (Lewis and Terrell, 1962). Fosberg (1954) treated the two taxa as varieties, although he noted little evidence of intergradation between sympatric "var. purpurea" and "var. longifolia" in the Virginia Piedmont. Terrell (1959) reported widespread intergradation between H. purpurea and H. longifolia "var. compacta" in the Piedmont of the Southeast, but he distinguished plants in the Highlands area as H. longifolia "var. glabra." He noted that the latter variety and H. purpurea often grew in close proximity, but he found no evidence of hybridization. Later, however, Terrell (1991) synonymized both "var. compacta" and "war. glabra" with "var. longifolia." If hybridization does occur, it would seem to be most likely in disturbed areas of intermediate habitat, such as the roadside where our plants grew. Although our study plants appeared to fit species descriptions closely, a few plants of intermediate leaf morphology were seen in the area, and it is possible that some hybridization has occurred in this population. If this were the case, it could have influenced the success of our hand-pollinations. Plants behaved consistently in our interspecific crosses, however, and differences among individuals in the degree of compatibility with H. purpurea were not observed.
We conclude that Houstonia serpyllifolia and H. longrifolia possess heteromorphic genetic compatibility systems typical of distylous plants. It appears that the incompatibility system is somewhat leaky, as evidenced by the greater success of intramorph pollinations than of unpollinated controls, but it is likely that most fruits and seeds will be sired by pollen from the other morph, especially when flowers receive the mixed pollen loads likely under open pollination. Therefore, we can assume that most successful matings will involve plants of different morphs, and identify a plant's neighbors as compatible or incompatible based on their floral morph. In further work, we use this information to create experimental populations to investigate the effects of plant (clone) size on male and female reproductive success, and to investigate the effects of plant morph, size, location and compatible and incompatible neighbor abundances on female reproductive success in natural populations (Beliveau, 1996).
Acknowledgments. - We thank J. Hamrick, C. Peterson and two anonymous reviewers for comments on earlier drafts of this manuscript. Funding was provided by grants-in-aid from the Botany Department of the University of Georgia, the Highlands Biological Station, and the W. C. Coker Fellowship Fund.
BAKER, J. D. AND R. W. CRUDEN. 1991. Thrips-mediated self-pollination of two facultatively xenogamous wetland species. Am. J. Bot., 78:959-963.
BAWA, K. S. AND J. H. BEACH. 1983. Self-incompatibility systems in the Rubiaceae of tropical lowland wet forest. Am. J. Bot., 70:1281-1288.
BELIVEAU, B. D. 1996. Factors affecting reproductive success in distylous bluets: effects of morph, size, location, and neighborhood. M. S. Thesis. University of Georgia, Athens. 203 p.
BIR BAHADUR. 1968. Heterostyly in Hedyotis nigrricans (Lam.) Fosb. J. Genetics, 60:175-177.
DULBERGER, R. 1992. Floral polymorphisms and their functional significance in the heterostylous syndrome, p. 41-84. In: S. C. H. Barrett (ed.). Evolution and function of heterostyly. Springer-Verlag, Berlin.
FOSBERG, F. R. 1954. Notes on plants of the eastern United States. Castanea, 19:25-37.
GANDERS, F. R. 1979. The biology of heterostyly. New Zealand J. Bot., 17:607-635.
GRIMALDI, D. 1988 Bee flies and bluets: Bombylius (Diptera: Bombyliidae) flower-constant on the distylous species, Hedyotis caerulea (Rubiaceae), and the manner of foraging. J. Nat. Hist., 22:1-10.
KEARNS, C. A. AND D. W. INOUYE. 1993. Techniques for pollination biologists. University Press of Colorado, Niwot.
LEVIN, D. A. 1974. Spatial segregation of pins and thrums in populations of Hedyotis nigricans. Evolution, 28:648-655.
-----. 1996. The evolutionary significance of pseudo-self-fertility. Am. Nat., 148:321-332.
LEWIS, W. H. 1962 Phylogenetic study of Hedyotis (Rubiaceae) in North America. Amer. J. Bot., 49:855-865.
----- AND E. E. TERRELL. 1962 Chromosomal races in eastern North American species of Hedyotis (Houstonia). Rhodora, 64:313-323.
LITTELL, R. C., G. A. MILLIKEN, W. W. STROUP AND R. D. WOLFINGER. 1996. SAS System for mixed models. SAS Institute Inc., Cary, NC. 633 p.
LLOYD, D. G. AND D. J. SCHOEN. 1992. Self- and cross-fertilization in plants. 1. Functional dimensions. Int. J. Plant Sci., 153:358-369.
----- and C. J. Webb. 1992. The evolution of heterostyly, p. 151-178. In: S. C. H. Barrett (ed.). Evolution and function of heterostyly. Springer-Verlag, Berlin.
MURRAY, B. G. 1990. Heterostyly and pollen-tube interactions in Luculia gratissima (Rubiaceae). Ann. Bot., 65:691-698.
ORNDUFF, R. 1977. An unusual homostyle in Hedyotis caerulea (Rubiaceae). Pl. Syst. Evol., 127:293-297.
-----. 1980. Heterostyly, population composition, and pollen flow in Hedyotis caerulea. Am. J. Bot., 67:95-103.
SAS INSTITUTE INC. 1989. SAS/STAT user's guide, version 6, fourth edition. SAS Institute Inc., Cary, NC. 1686 p.
-----. 1996. SAS/Stat software: changes and enhancements through release 6.11. SAS Institute Inc., Cary, NC. 1104 p.
TERRELL, E. E. 1959. A revision of the Houstonia purpurea group (Rubiaceae). Rhodora, 61:157-181.
-----. 1991. Overview and annotated list of North American species of Hedyotis, Houstonia, Oldenlandia (Rubiaceae) and related genera. Phytologia, 71:212-243.
WYATT, R. AND R. L. HELLWIG. 1979. Factors determining fruit set in heterostylous bluets, Houstonia caerulea (Rubiaceae). Syst. Bot., 4:103-114.
|Printer friendly Cite/link Email Feedback|
|Author:||Beliveau, Bethany D.; Wyatt, Robert|
|Publication:||The American Midland Naturalist|
|Date:||Apr 1, 1999|
|Previous Article:||The Lazarus effect: rejuvenation of leaf-senescent seedlings in a rare grassland perennial.|
|Next Article:||Vegetation biomass dynamics and patterns of sexual reproduction in a northern mixed-grass prairie.|