HORTSCIENCE 51(4):325–329. 2016. The two genera from the Melastomata- ceae family that were chosen for this project are Dissotis and . Dissotis is Interspecific and Intergeneric native to Africa and Tibouchina to as well as Central and Hybridization in Dissotis (Almeda and Chuang, 1992; Renner and Meyer, 2001; Todzia, 1999). Tibouchina and Tibouchina contains species of importance in the orna- mental horticulture trade, such as Tibouchina Susan M. Hawkins ‘Athens Blue’, whereas Dissotis sp. have not Department of Horticulture, University of Georgia, 1109 Experiment Street, been widely used as ornamentals. However, Georgia Station, Griffin, GA 30223-1797 some Dissotis sp., such as D. rotundifolia, have excellent ornamental qualities, such as John M. Ruter1 soft, fuzzy , or attractive pink or purple Department of Horticulture, University of Georgia, 327 Hoke Smith flowers, as well as being easy to cultivate and propagate. Hybridizing Dissotis and Tibou- Building, 1225 Lumpkin Street, Athens, GA 30602 china could create unique combinations of Carol D. Robacker growth form and flower color and produce novel for the ornamental horticul- Department of Horticulture, University of Georgia, 1109 Experiment Street, ture trade. The overall goal of this project was Georgia Station, Griffin, GA 30223-1797 to investigate the feasibility of hybridization Additional index words. , Dissotis rotundifolia, Dissotis princeps, Dissotis within and between various species of Dis- sotis and Tibouchina to create interspecific canescens, , Tibouchina fothergillae 3 pilosa, germination, floral and intergeneric hybrids of these genera. morphology, buzz pollination Abstract. Four species of Dissotis and three species of Tibouchina, two genera of the Materials and Methods Melastomataceae family, were crossed in an attempt to create interspecific and in- tergeneric hybrids. Intergeneric crosses set seed at a rate of 18.1% and interspecific Several species of Dissotis and Tibou- crosses had a 32.3% rate of seed set. Germination was extremely poor, with only four china were placed in a greenhouse in Athens, crosses having germinated seed. Crosses produced 31 seedlings. Three of the seedlings GA in Feb. 2012 (Table 1). Stock were were from intergeneric crosses between Dissotis canescens and Tibouchina lepidota. each clones of one genotype of each species. Interspecific crosses produced 25 seedlings from crosses between Dissotis princeps and Cuttings of each species were taken, treated Dissotis rotundifolia and three seedlings from crosses between D. canescens and D. with a 5-s dip of potassium salt of indole-3- princeps. The prognosis for conventional breeding for species in Dissotis and Tibouchina butyric acid (K-IBA) at 1000 mg·L–1, and is poor due to low seed set, poor germination, and slow growth of progeny. stuck in Fafard Germination Mix containing processed pine bark (40%), Canadian sphag- num peat, perlite, and vermiculite (Sun Grow The Melastomataceae family comprises variously reported by Solt and Wurdack Horticulture, Agawam, MA), and placed un- between 185 and 190 genera containing 5000 as having a chromosome count of n =15 der intermittent mist at 8 s every 10 min and species (Almeda and Chuang, 1992). Mela- and by Almeda as having a chromosome 50% shade. After 6 weeks, cuttings were stomataceae species are found in Asia, count of n = 10 (Almeda, 1997; Solt and potted up in Fafard 3B Mix consisting of Africa, North and South America, and Aus- Wurdack, 1980). Dissotis rotundifolia is re- Canadian sphagnum peat, processed pine tralia; South America contains the majority ported to be a diploid with n =15(Soltand bark, perlite, and vermiculite in 2.8-L (15.2 cm of species (Michelangeli et al., 2013; Renner Wurdack, 1980). Dissotis canescens, for- diameter) trade containers. and Meyer, 2001). Some species, such as merly treated as Heterotis canescens,has Dissotis sp. started blooming in July 2012. D. rotundifolia (Sm.) Triana, have natural- been reported to be n = 17 and is possibly Tibouchina sp. started blooming in Aug. ized in places with tropical climates such as polyploid (Almeda, 1997). 2012. Interspecific and intergeneric crosses Indonesia, , Puerto Rico, and Hawaii Most Melastomataceae species have por- were made as species came into bloom (Liogier and Martorell, 1982; Renner and icidal anthers and exhibit herkogamy (Renner, beginning in July 2012 and ending in Mar. Meyer, 2001). 1989). Herkogamy and poricidal anthers pro- 2013. Flowers were emasculated to prevent Species in the Melastomataceae vary mote outcrossing in Melastomataceae. The self-pollination. Emasculations were carried out widely in ploidy and chromosome count. poricidal anthers require manipulation for the before anthers had completely unfolded when Tibouchina chromosome numbers are based pollen to disperse. Pollen disperses when possible. Pollen was extracted from flowers of on x =9,rangingfromn =9ton =63 the pollinator vibrates the anthers with its male parents by vibrating anthers with a tuning (Almeda, 1997; Almeda and Chuang, 1992). flight muscles, a process called buzz pollina- fork (key of E). Pollen was captured in glass Many species of Tibouchina are polyploid. tion (Luo et al., 2008; Renner, 1989). To containers and applied to the stigmas of flowers Although tetraploidy is the most common obtain pollen for use in hybridization, of the female parents using either a Q-tip or polyploid level, ploidy level varies (Almeda, breeders mimic buzz pollination by using a small paintbrush. A total of 1036 crosses were 1997). has a chromo- a tuning fork in the key of E to get pollen to made, 588 crosses were interspecific and 448 some count of n = 18 and is tetraploid (Solt dehisce (Renner, 1989). crosses were intergeneric. and Wurdack, 1980). Tibouchina lepidota is n = 62 or 2n = 122 and is thought to be 14-ploid (Almeda, 1997). Dissotis has been Table 1. Melastomataceae species used as parents in breeding program. Species Chromosome count Ploidy Dissotis canescens n = 17 Undetermined Received for publication 10 Nov. 2015. Accepted Dissotis debilis Unknown Unknown for publication 23 Feb. 2016. Dissotis princeps Unknown Unknown Dissotis rotundifolia n =15 2x This paper is part of a thesis submitted by Susan M. · Hawkins as part of the fulfillment of a Master’s Tibouchina fothergillae pilosa Unknown Unknown Tibouchina granulosa ‘Gibraltar’ n =18 4x Degree.   1Corresponding author. E-mail: [email protected]. Tibouchina lepidota 2n = 122 or n = 62 Possibly 14-ploid

HORTSCIENCE VOL. 51(4) APRIL 2016 325 Fruits were harvested when ripe. Since all Waltham, MA). Pictures were taken with examination of the seedlings from all crosses species produced capsular fruit, fruits were an Olympus D70 microscope camera and showed that seedlings resembled female determined to be ripe when they were light to Olympus DP Controller software. parents in growth habit, shape, and leaf medium brown and hard. Seed was extracted Pollen tube data were taken on the number size. Although several of the D. princeps · from the ripe fruits and examined under of pollinated styles in which pollen tubes D. rotundifolia putative hybrids eventu- a dissecting microscope to assess probable reached the end of the style. Data were taken ally bloomed, all plants except one had viability. Seed that was shriveled and flat- on number of crosses producing fruit with seed tened was deemed to be nonviable. Seed of probable viability and number of crosses determined to be of probable viability was germinating. Percentages of seed-producing sown in 0.56-L (10.2 cm diameter) pots on crosses and germinating crosses were calcu- top of Fafard Germination Mix under fluo- lated from this data. Number of seedlings rescent lighting at a set temperature of 21 C. resulting from these crosses was counted. Two different methods of germination were used. The first method, used from Sept. Results to Nov. 2012, was a humidity tent. Seeds were sown on top of moist substrate. The pots Pollen tubes were able to grow to the end were covered with clear plastic and substrate of the style in every cross tested (Table 2; was misted periodically to maintain a high Figs. 1 and 2). Intergeneric crosses and level of humidity. Beginning in Jan. 2013, interspecific crosses showed the same pat- harvested seeds were sown on top of substrate tern, indicating that prezygotic barriers to and placed under mist in a greenhouse using fertilization due to pollen and pistil incom- natural lighting only. patibility did not exist in these crosses. Seedlings that germinated were trans- Seed set from both interspecific and in- planted into separate 0.56-L (10.2 cm di- tergeneric crosses was low and showed great ameter) pots containing Fafard Germination variability among parents. For interspecific Mix and placed in a greenhouse under 50% crosses, D. princeps and D. rotundifolia had shade. Once seedlings had grown to at least the highest rate of seed set at 42.7% and 5.08 to 7.62 cm tall, they were transplanted 42.9%, respectively (Table 3). Species with into 2.8-L (15.2 cm diameter) trade con- the lowest rate of seed set were Tibouchina tainers filled with a pine bark substrate with fothergillae ·pilosa at 0.0%, and T. lepidota micronutrients and removed from shade. at 4.3% (Table 3). Seed set among parents in Seed set from initial crosses was low, so intergeneric crosses was also highly variable. pollen germination and pollen tube growth Dissotis canescens had the highest rate of through the styles were evaluated to deter- seed set at 31.5% (Table 4). Tibouchina mine whether barriers to fertilization existed fothergillae ·pilosa and T. lepidota again between the species used as parents in the had the lowest rate of seed set at 3.5% and study. Interspecific and intergeneric crosses 4.0%, respectively (Table 4). Fig. 1. (A) Pollen germinating on end of stigma and (Table 2) were performed during Jan. and The first method of propagation, germi- beginning of pollen tubes in Dissotis debilis · Feb. 2013. The number of repetitions of each nating seed under high humidity, produced Tibouchina lepidota cross. (B) Pollen tubes cross was dependent on the number of no seed germination after 90 d. Germina- growing to end of style in D. debilis · T. flowers available. Some species were less tion was obtained only from placing seeds lepidota cross. floriferous in the winter, such as D. princeps, under intermittent mist in the greenhouse. T. granulosa ‘Gibraltar’, and T. lepidota, so Germinationtimerangedfrom2to9weeks. fewer crosses were performed using those Crossing results for seeds germinated on the species. Styles were harvested 24 h after mist bench is presented in Tables 5 and 6. pollination and placed in ethanol:acetic acid Germination rates for the few crosses that (1:2 w/v) for 1 to 24 h. They were transferred did germinate were still very low. to 65% ethanol for 20 min, autoclaved in Seeds from only three crosses germi- 0.8 mol·L–1 NaOH at 120 C for 20 min, nated, one of the crosses was intergeneric and stained with 0.1% aniline blue in 0.1 M and two were interspecific (Table 7). Dissotis K3PO4 (Bo et al., 2009; Ledesma and canescens by D. princeps produced only Sugiyama, 2005). The styles were kept in three seedlings, whereas D. princeps · the aniline blue for 4 to 24 h, then mounted D. rotundifolia produced 25 seedlings. Three on slides and examined under an Olympus seedlings were obtained from the cross of BX51 fluorescent microscope (Olympus Corp., D. canescens · T. lepidota. Morphological

Table 2. Interspecific and intergeneric crosses in Dissotis and Tibouchina performed to assess pollen tube germination. Female parent Male parent Repetitions Dissotis debilis D. rotundifolia 11 D. debilis T. lepidota 5 Dissotis princeps D. rotundifolia 2 D. princeps T. granulosa ‘Gibraltar’ 5 D. princeps T. lepidota 8 Dissotis rotundifolia D. debilis 11 D. rotundifolia D. princeps 14 Fig. 2. (A) Pollen germinating on end of stigma and D. rotundifolia T. granulosa ‘Gibraltar’ 14 beginning of pollen tubes in Dissotis rotundi- D. rotundifolia T. lepidota 10 folia · Dissotis debilis cross. (B) Pollen tubes Tibouchina Granulosa ‘Gibraltar’ D. princeps 1 growing to end of style in D. rotundifolia · D. Tibouchina lepidota D. princeps 3 debilis cross.

326 HORTSCIENCE VOL. 51(4) APRIL 2016 Table 3. Crossings performed and seed set results for interspecific Melastomataceae crosses. No. of species No. of cross-pollinated No. of cross-pollinated Percent of cross-pollinated Female parent mating flowers flowers setting seed flowers setting seed Dissotis canescens 3 225 75 33.3 Dissotis debilis 3 93 10 10.8 Dissotis princeps 3 75 32 42.7 Dissotis rotundifolia 3 168 72 42.9 Tibouchina fothergillae ·pilosa 1 4 0 0.0 Tibouchina lepidota 1 23 1 4.3 Total 588 190 32.3

Table 4. Crossings performed and seed set results for intergeneric Melastomataceae crosses. No. of species No. of cross-pollinated No. of cross-pollinated Percent of cross-pollinated Female parent mating flowers flowers setting seed flowers setting seed Dissotis canescens 2 130 41 31.5 Dissotis debilis 2 32 3 9.4 Dissotis princeps 3 70 7 10.0 Dissotis rotundifolia 2 106 26 24.5 Tibouchina fothergillae ·pilosa 3 85 3 3.5 Tibouchina lepidota 3 25 1 4.0 Total 448 81 18.1

Table 5. Summary of results for interspecific Melastomataceae crosses—mist bench propagation only. No. of species No. of cross-pollinated No. of cross-pollinated No. of cross-pollinated flowers Female parent mating flowers flowers setting seed producing germinated seed Dissotis canescens 3 164 14 2 Dissotis debilis 3830 0 Dissotis princeps 365221 Dissotis rotundifolia 3 133 37 0 Tibouchina fothergillae ·pilosa 14 0 0 Tibouchina lepidota 1231 0 Total 472 74 3

Table 6. Summary of results for intergeneric Melastomataceae crosses—mist bench propagation only. No. of species No. of cross-pollinated No. of cross-pollinated No. of cross-pollinated flowers Female parent mating flowers flowers setting seed producing germinated seed Dissotis canescens 2934 1 Dissotis debilis 2290 0 Dissotis princeps 3707 0 Dissotis rotundifolia 2 105 25 0 Tibouchina fothergillae ·pilosa 3831 0 Tibouchina lepidota 3240 0 Total 404 37 1

Table 7. Number of seedlings produced for Melastomataceae crosses with germinating seed showing number of seedlings produced. Female parent Male parent No. of crosses Crosses with seed Crosses with germinating seed No. of seedlings Dissotis canescens D. princeps 68 26 2 3 D. canescens Tibouchina lepidota 46 12 1 3 Dissotis princeps Dissotis rotundifolia 61 29 1 25 lavender-colored flowers like the female Discussion We visually examined the seed to de- parent. One D. princeps · D. rotundifolia termine probable viability and deemed seed putative hybrid produced some pink flow- Pollen tubes grew through the styles of all that was flat instead of rounded to be non- ers as well as lavender flowers, but did not crosses tested, indicating that pollen–pistil viable. Wiens et al. (1987) found that flattish, survive. One seedling of the D. canescens · incompatibility did not cause low fertility. collapsed seed was a sign of aborted embryos T. lepidota cross bloomed and the flower Low rates of seed set could have been partly in Epilobium angustifolium L.; it is probable was like that of the female parent in size, due to lack of fertilization or embryo abor- that flat seed in this study also contained shape, number, and color. Although tion. Incompatibilities between parents in an aborted embryos. Prezygotic barriers to fer- seedlings strongly resembled the female interspecific or intergeneric cross may lead to tilization appeared to be low based on pollen parent, they were weak and did not grow poor seed set. In crosses between Chimonan- tube germination tests. Postzygotic barriers vigorously. Seedling performance was thus praecox and Chimonanthus nitens, even could have been caused by parent species poorer than that of either parent, suggesting though successful fertilization occurred in having different chromosome numbers, as that they could be hybrids. Seedling mor- 30.4% of pollinations, 100% of embryos the species with known chromosome counts tality was high. Of the 31 seedlings pro- aborted by 25 d after pollination due to the had widely varying numbers of chromo- duced, none survived to Oct. 2015. Detailed low number of viable embryos that devel- somes. For instance, differing endosperm data and results for crosses are presented in oped and the complete failure of the endo- balance numbers in parental species or the Table 8. sperm to develop (Wang et al., 2014). incompatibility of parental genomes could

HORTSCIENCE VOL. 51(4) APRIL 2016 327 Table 8. Detailed data for Melastomataceae crosses. Germination data are for crosses propagated on the mist bench only. No. of No. of crosses % Crosses No. of crosses with % Crosses with Female parent Male parent crosses with seed with seed germinating seed germinating seed Dissotis canescens D. debilis 52 25 48.1 0 0.0 D. canescens D. princeps 68 26 38.2 2 12.5 D. canescens D. rotundifolia 105 24 22.9 0 0.0 D. canescens T. fothergillae ·pilosa 84 29 34.5 0 0.0 D. canescens T. lepidota 46 12 26.1 1 10.0 Dissotis debilis D. canescens 19 1 5.3 0 0.0 D. debilis D. princeps 42 6 14.3 0 0.0 D. debilis D. rotundifolia 32 3 9.4 0 0.0 D. debilis T. fothergillae ·pilosa 16 3 18.8 0 0.0 D. debilis T. lepidota 16 0 0.0 0 0.0 Dissotis princeps D. canescens 10 2 20.0 0 0.0 D. princeps D. debilis 4 1 25.0 0 0.0 D. princeps D. rotundifolia 61 29 47.5 1 14.3 D. princeps T. fothergillae ·pilosa 19 1 5.3 0 0.0 D. princeps T. granulosa ‘Gibraltar’ 6 2 33.3 0 0.0 D. princeps T. lepidota 45 4 8.9 0 0.0 Dissotis rotundifolia D. canescens 46 16 34.8 0 0.0 D. rotundifolia D. debilis 20 10 50.0 0 0.0 D. rotundifolia D. princeps 102 46 45.1 0 0.0 D. rotundifolia T. fothergillae ·pilosa 50 14 28.0 0 0.0 D. rotundifolia T. lepidota 56 12 21.4 0 0.0 Tibouchina fothergillae ·pilosa D. debilis 2 0 0.0 0 0.0 T. fothergillae ·pilosa D. princeps 32 3 9.4 0 0.0 T. fothergillae ·pilosa D. rotundifolia 51 0 0.0 0 0.0 T. fothergillae ·pilosa T. lepidota 4 0 0.0 0 0.0 Tibouchina lepidota D. canescens 1 0 0.0 0 0.0 T. lepidota D. princeps 17 1 5.9 0 0.0 T. lepidota D. rotundifolia 7 0 0.0 0 0.0 T. lepidota T. fothergillae ·pilosa 23 1 4.3 0 0.0

have caused postzygotic barriers (Talluri, Low germination of seed from crosses and mutations through ethyl methanesulfonate 2012). In addition, chromosome elimination subsequent slow growth of the seedlings (EMS) or gamma radiation might create novel in hybrid seed during initial mitosis after indicates that the species used in this study cultivars for the ornamental industry. Creating fertilization may have resulted in the weak- are poor prospects for a conventional breed- interspecific and intergeneric hybrids through ness of the hybrids and contributed to the ing program. Seedlings took several months conventional breeding could take too long to be nonviability of seed (Hancock et al., 2015). from germination to grow more than 5 cm economically or practically feasible for many Use of species with the same numbers of tall. Once seedlings started to become pot- breeding programs. Although hybridization be- chromosomes as parents could result in bound in the 0.56-L (10.2 cm diameter) pots tween species in Dissotis and Tibouchina has successful hybridization. As well, ovule cul- into which they were initially transplanted, the potential to create novel ornamental culti- ture is a possibility for increasing the number the seedlings grew rapidly and commenced vars, practical barriers to creating such hybrids of progeny from future crosses if the failure flowering. Establishing a root system that is remain to be overcome. of endosperm to develop is the cause of substantial in size relative to the aerial parts embryo abortion. of the seedling may be a survival strategy for Literature Cited Even though the Melastomataceae seed these species, which evolved in environments used for germination appeared to be fully with dry and rainy seasons. Low seed set, Almeda, F. 1997. Chromosome numbers and their mature, germination of both interspecific and poor germination, and slow progeny growth evolutionary significance in some neotropical and paleotropical Melastomataceae. BioLlania intergeneric crosses was extremely poor. from the initial crosses were the main factors 6:167–190. Incompatibility between the embryos and in the decision not to continue the study past Almeda, F. and T.I. Chuang. 1992. Chromosome- the endosperm could account for poor ger- the first generation of crosses. numbers and their systematic significance in mination (Wiens et al., 1987). The propaga- Since the progeny of the crosses resem- some Mexican Melastomataceae. Syst. Bot. tion protocol used affected the germination bled the female parents, determination that 17:583–593. rate. No germination resulted when seeds they were truly hybrids was inconclusive. Bo, J., S. Zonggen, S. Jinbo, Y. Da, S. Xianyong, were kept under a humidity tent, although Apomixis has been reported in nine Mela- and L. Hongfei. 2009. Germination and growth this method had worked for other researchers stomataceae genera (dos Santos et al., 2012; of sponge gourd (Luffa cylindrica) pollen tubes (Solt and Wurdack, 1980). Even seed of Mendes Rodrigues and Oliveira, 2012; and FTIR analysis of the pollen tube wall. Sci. Hort. 122:638–644. D. rotundifolia that was obtained from B & T Renner, 1989). Although none of the species dos Santos, A.P.M., C.M. Fracasso, M. Luciene dos World Seeds, Paguignan, France, failed to we used in the study have been reported to be Santos, R. Romero, M. Sazima, and P.E. germinate under the humidity tent, but did apomictic, apomixis cannot be ruled out as Oliveira. 2012. Reproductive biology and spe- germinate once placed on the mist bench. a factor due to the identical appearance of the cies geographical distribution in the Melasto- The mist may have mimicked the rainy progeny and the female parents. The main mataceae: A survey based on New World taxa. season of the regions where the species in factor that indicated the hybrid status of the Ann. Bot. 110:667–679. this study originated, providing the neces- progeny was their extreme weakness. If the Hancock, W., V. Kuraparthy, S. Kernodle, and R. sary amount of moisture to germinate the seedlings had been apomictic, we would have Lewis. 2015. Identification of maternal hap- seeds and wash away any phenolic com- expected them to grow as vigorously as the loids of Nicotiana tabacum aided by transgenic expression of green fluorescent protein: Evi- pounds that could have inhibited ger- female parent. Other factors, such as the loss of dence for chromosome elimination in the mination. Our recommended protocol for paternal chromosomes, could have led to the N. tabacum · N. africana interspecific cross. germination is to sow seed on top of media close resemblance to the female parents. Mol. Breed. 35(9):179. and place under intermittent mist until seed Species in Dissotis and Tibouchina have Ledesma, N. and N. Sugiyama. 2005. Pollen germinates. high ornamental value. Methods such as inducing quality and performance in strawberry plants

328 HORTSCIENCE VOL. 51(4) APRIL 2016 exposed to high-temperature stress. J. Amer. Michelangeli, F.A., P.J.F. Guimaraes, D.S. Penneys, Talluri, R. 2012. Interploidy interspecific hybrid- Soc. Hort. Sci. 130:341–347. F. Almeda, and R. Kriebel. 2013. Phylogenetic ization in Fuchsia. J. Genet. 91(1):71–77. Liogier, A.H. and L.F. Martorell. 1982. Flora of relationships and distribution of New World Todzia, C.A. 1999. Ten new species of Tibouchina Puerto Rico and adjacent islands: A systematic Melastomeae (Melastomataceae). Bot. J. Linn. (Melastomataceae) from Mexico. Brittonia synopsis. 1a ed. Editorial de la Universidad de Soc. 171:38–60. 3:255–279. Puerto Rico, R´ıo Piedras, PR. Renner, S.S. 1989. A Survey of reproductive biology Wang, W., L. Zhou, Y. Huang, Z. Bao, and H. Zhao. Luo, Z., D. Zhang, and S.S. Renner. 2008. Why in Neotropical Melastomataceae and Memecyla- 2014. Reproductive barriers in interspecific hy- two kinds of in buzz-pollinated ceae. Ann. Mo. Bot. Gard. 76:496–518. bridizations among Chimonanthus praecox (L.) flowers? Experimental support for Darwin’s Renner, S.S. and K. Meyer. 2001. Melastomeae link, C. salicifolius SY Hu, and C. nitens Oliver division-of-labour hypothesis. Funct. Ecol. come full circle: Biogeographic reconstruction from pollen–pistil interaction and hybrid embryo 22:794–800. and molecular clock dating. Evolution development. Sci. Hort. 177:85–91. Mendes-Rodrigues, C. and P.E. Oliveira. 2012. 55:1315–1324. Wiens, D., C.L. Calvin, C. Wilson, C. Davern, D. Polyembryony in Melastomataceae from Bra- Solt, M.L. and J.J. Wurdack. 1980. Chromosome Frank, and S. Seavey. 1987. Reproductive suc- zilian : Multiple embryos in a small numbers in the Melastomataceae. Phytologia cess, spontaneous embryo abortion, and genetic world. Plant Biol. 14:845–853. 47:199–220. load in flowering plants. Oecologia 71:501–509.

HORTSCIENCE VOL. 51(4) APRIL 2016 329