HORTSCIENCE 44(7):2031–2034. 2009. North America (Bowen et al., 2002; Garrett, 2007; Jog and Delong, 2005; USDA Forest Service, 2007) and Europe (Essl, 2005). Identification of Sterile, Noninvasive Eliminating fertile Japanese spirea cultivars, at least in areas where this species has dem- Cultivars of Japanese Spirea onstrated the potential to naturalize, may be 1 of particular importance early in the invasion Rhiannon L. Wilson and William A. Hoch process when multiple introductions, genetic Montana State University, Department of Plant Sciences and Plant Pathology, recombination, and local selection are taking 324 Leon Johnson Hall, Bozeman, MT 59717 place, potentially contributing to the evolu- tion of highly invasive populations. Additional index words. Spiraea japonica, Spiraea ·bumalda, sterility, introduced species, In this study, 20 commonly available DNA content cultivars of S. japonica were evaluated for Abstract. Japanese spirea (Spiraea japonica L. f.), a popular landscape shrub, has shown fertility by measuring the germination of the potential to become an invasive weed in both North America and Europe. Twenty pollen and open-pollinated seed from plants commonly available S. japonica cultivars were evaluated for fertility using pollen and grown in a replicated field plot. In addition, seed germination. Clones were grown in a randomized, replicated field plot, and seed collected from commercial nurseries additional seed samples were obtained from commercial nurseries and hand-pollinations were tested, and plants demonstrating steril- in the greenhouse. Three sterile cultivars were identified: ‘Crispa’, ‘Dart’s Red’, and ity in the test plot were pollinated under ‘Neon Flash’. These cultivars demonstrated poor anther dehiscence and very low mean controlled conditions in the greenhouse. The pollen germination, 2.7%, 3.0%, and 1.3%, respectively, which often produced abnormal DNA content of all clones was determined pollen tubes. None of these three cultivars produced viable seed in the field plot, at using flow cytometry. commercial nurseries, or when hand-pollinated in the greenhouse, whereas seed germination from fertile clones ranged from 91.5% to 100%. The other 17 cultivars Materials and Methods tested, which should be treated as entirely fertile for the purposes of invasive plant management, were ‘Albiflora’, ‘Anthony Waterer’, ‘Candlelight’, ‘Dakota Goldcharm’, Plant material. Eighteen cultivars of S. var. alpina ‘Daphne’, ‘Flaming Mound’, ‘Flowering Choice’, ‘Froebelii’, ‘Golden japonica were maintained at the Montana Princess’, ‘Goldflame’, ‘Goldmound’, ‘Gumball’, ‘Lemon Princess’, ‘Little Princess’, State University Horticulture Farm, Bozeman, ‘Magic Carpet’, ‘Norman’, and ‘Shibori’. Measurements of DNA content indicated that MT, in a randomized complete block design all tested clones are diploid; therefore, the observed sterility was not related to poly- with three replicated blocks. All plants in the ploidy. The identification of these three sterile cultivars can help reduce the use of fertile field plot were obtained from Bailey Nurser- varieties in areas where Japanese spirea has shown the potential to become invasive. ies (St. Paul, MN). Two clones that were not available when the field plot was planted, ‘Candlelight’ and ‘Lemon Princess’, were acquired from Forest Farm (Williams, OR) The escape and establishment of exotic its adaptability, ease of culture, and desirable and three replicates of each were maintained plants from ornamental landscapes is causing ornamental characteristics (Dirr, 1998). Sales in the greenhouse for analysis of pollen and significant ecosystem and environmental dam- of Spiraea species and cultivars, the majority seed germination and DNA content. Three age throughout much of the United States. of which are S. japonica, rank second only cultivars that demonstrated sterility in our These nonnative plants have affected natural to roses in total sales of deciduous ornamen- test plot, ‘Crispa’, ‘Dart’s Red’, and ‘Neon ecosystems in a variety of ways, including tal plants in the United States [U.S. Depart- Flash’, were further tested through hand- a reduction in diversity through the displace- ment of Agriculture (USDA), 1998]. pollinations in the greenhouse. For these ment of native species, reduced soil stability Unfortunately, Japanese spirea has also pollinations, container-grown plants of and water quality, interference with plant suc- shown the potential to become an invasive ‘Dart’s Red’ were obtained from Means cession dynamics and wildlife habitat and weed, naturalizing in much of the eastern Nursery (Scappoose, OR) and ‘Crispa’ and foraging, and altered fire regimes and nutrient United States (USDA, 2009). ‘Neon Flash’ from Bailey Nurseries (St. Paul, cycling (Dukes and Mooney, 2004; Mack A number of state governments have MN). All pollen and seed samples were et al., 2000; Raizada et al., 2008). Economic banned, or are in the process of banning, the collected during 2008. costs associated with invasive plants total import and sale of invasive ornamentals Pollen germination. Pollen germination more than $120 billion annually (Pimentel (Mehrhoff et al., 2003). An alternative to was measured for each of the three replicate et al., 2005). Many of these exotic species, banning the sale of Japanese spirea would plants for all cultivars in the field test plot particularly among woody plants, originated be the use of sterile, noninvasive clones. The between 15 July and 8 Aug. 2008. Pollen as ornamental introductions. Reichard (1997) availability of sterile cultivars would precipi- from ‘Candlelight’ and ‘Lemon Princess’ determined that 85% of invasive woody tate the elimination of fertile varieties from was obtained from each of three replicates plants were originally introduced for land- commerce in areas where it has shown the in the greenhouse. At anthesis, pollen from scape purposes. potential to become invasive. one inflorescence was sprinkled onto pollen Japanese spirea, Spiraea japonica L. f. Invasions of exotic species are often germination media by gently tapping the (synonym S. ·bumalda) has become an ex- marked by a lag time between initial coloni- inflorescence directly above the surface of tremely popular landscape plant as a result of zation and the onset of rapid population the media. Pollen germination media con- growth and range expansion (Sakai et al., sisted of 1.6 mM H3BO3, 1.3 mM Ca(NO3)2, 2001). There is compelling evidence that 0.8 mM MgSO4Á7H20, 0.6 mM MnSO4Á1H20, many, if not most, successful invasions are 1mM KNO3, 15% sucrose, and 1% agar. Received for publication 8 June 2009. Accepted for preceded by multiple introduction events that After 8 h of incubation at 23 C, the percent- publication 21 July 2009. provide allelic variation that drives adaptive age of germinated pollen grains was deter- We thank Bailey Nurseries and Means Nursery for evolution (Culley and Hardiman, 2007; mined from 100 total grains on randomly donation of plant material and Cross Nurseries Inc., Durka et al., 2005; Lavergne and Molofsky, selected sections of the petri dish observed at Johnson’s Nursery Inc., Lawyer Nursery Inc., McKay Nursery, and Sampson Nursery for seed. 2007; Pe´rez de la Vega et al., 1991; Prentis 100· magnification using an Olympus mi- We also thank Dr. J. Spencer Johnston at Texas et al., 2008; Sakai et al., 2001). Available croscope (Olympus America Inc., Center A&M University for his help with flow cytometry. evidence suggests that many populations of Valley, PA). A pollen grain was considered 1To whom reprint requests should be addressed; Japanese spirea are in the initial establish- to have germinated if a pollen tube was e-mail [email protected]. ment and/or lag phases of invasion both in visible and had a length that was equal to or
HORTSCIENCE VOL. 44(7) DECEMBER 2009 2031 greater than the diameter of the pollen grain inside 100 · 15-mm petri dishes and wrapped those observed in pollen from triploid plants (Rajora and Zsuffa, 1986). with parafilm (Pechiney Plastic Packaging, such as stunted growth, twisting, and swelling Greenhouse pollinations. Three cultivars Menasha, WI). Petri dishes were placed on (Park et al., 2002). Anther dehiscence was also that demonstrated sterility in our test plot, a laboratory bench 1.2 m beneath a 400-W abnormal in these clones with the majority of ‘Crispa’, ‘Dart’s Red’, and ‘Neon Flash’, were high-pressure sodium vapor lamp (average anthers opening only partially or not at all. further evaluated through hand-pollinations in photon flux of 80 mmolÁm–2Ás–1) operating on This indicates that the pollen of these four the greenhouse. Because self-incompatibility an 18-h photoperiod. For all tests, seed germi- cultivars is mostly, if not entirely, nonviable. of Spiraea has not been characterized, and nation was recorded every 3 d until germina- Pollen from all other cultivars germinated our work from a previous study indicates that tion was complete (up to 8 weeks). at significantly higher rates and produced S. japonica is primarily an outcrossing spe- Flow cytometry. To release nuclei, fresh, normal pollen tubes. Pollen collected from cies (W. Hoch, unpublished data), each hand- newly expanded leaves of spirea samples the two greenhouse-grown cultivars, Candle pollination was performed using pollen from [30 mg fresh weight (FW)] and the DNA Light and Lemon Princess, displayed sig- two different clones. One inflorescence on standard (50 mg FW) were chopped to- nificantly higher germination percentages each of three replicates of the three cultivars gether with a double-edged razor blade in (69.3% and 67.0%, respectively) than the was pollinated with a mixture of pollen from 1 mL ice-cold chopping buffer for 60 s. other clones. ‘Albiflora’ and ‘Gumball’ plants maintained Care was taken not to include spirea leaves Seed germination of the 20 cultivars in the greenhouse. These two clones were with noticeable levels of anthocyanins, be- revealed two markedly different groups: also reciprocally crossed as a positive con- cause the presence of anthocyanins has been those with a high mean percentage of germi- trol. Pollen was checked on germination found to inhibit propidium iodide DNA nation (greater than 91%) and those that did media before pollination to assure a high fluorescence (Bennett et al., 2008). The chop- not yield any viable seed (Table 1). As a result percentage of viable pollen, and florets were ping buffer consisted of 10 mM MgSO4,50mM of the length of the growing season at our pollinated at least twice over a period of KCl, 5 mM Hepes, 6.5 mM dl-dithiothreitol, test plot in 2008 (118 d; U.S. Department of several days to increase the probability of 2% polyvinylpyrrolidone 40, and 0.7% Tween Interior, 2009), 12 clones did not produce stigmatic receptivity during pollinations. 20. Sorghum bicolor Pioneer 8695 (Pioneer enough mature seed for a full analysis of Seed heads were collected 2 to 3 months Hi-Bred International, Johnston, IA), which germination. For these cultivars, available after pollination, dried, and the contents re- has a DNA content of 1.74 pg/2C (Johnston seed from the field plot were tested for moved under an Olympus dissecting scope et al., 1999), was used as the standard in all germination, and additional seed samples (Olympus America Inc.). samples. The suspension was filtered through were obtained from commercial nurseries to Seed germination. Mature seed heads a30-mm nylon mesh filter (Spectrum Labora- complete the analysis. Three cultivars, from each of the three replicate plants per tories, Rancho Dominguez, CA) and stored on Crispa, Dart’s Red, and Neon Flash, did not clone in the test plot were collected and dried ice until staining. After nuclei extraction, produce viable seed in the field plot, at at room temperature to release the seed. Seed propidium iodide and RNase (RNase Oneä commercial nurseries, or when hand-pollinated samples of several cultivars were obtained Ribonuclease; Promega Corp., Madison, WI) in the greenhouse. In all instances, the dried from commercial nurseries either because were added to each sample to a final concen- dehiscent follicles of these three clones con- additional samples were desired to verify tration of 50 mgÁmL–1 and 20 mgÁmL–1,re- tained small, shriveled seed, sometimes with our findings of sterility in the test plot spectively, and briefly vortexed. Samples were small amounts of abnormal endosperm. For (‘Crispa’, ‘Dart’s Red’, and ‘Neon Flash’) or kept on ice in the dark and analyzed by flow these three cultivars, the entire sample from because the length of the growing season at our cytometry within 1 to 2 h. The ploidy level of each source was processed through the seed test plot in 2008 precluded the maturation of S. japonica cultivars was based on the known germination protocol, but in all instances, no enough seed for a full analysis (‘Albiflora’, tetraploid (2n =4x =36)S. japonica var. germination occurred. ‘Anthony Waterer’, ‘Dakota Goldcharm’, var. fortunei (Zhao-Yang et al., 2002). Pollinations in the greenhouse produced alpina ‘Daphne’, ‘Flaming Mound’, ‘Flower- The mean fluorescence of nuclei was viable seed in the fertile cultivars Candle ing Choice’, ‘Goldflame’, ‘Goldmound’, ‘Lit- quantified using a BD FACSCalibur Flow Light and Lemon Princess (Table 1) as well tle Princess’, ‘Magic Carpet’, ‘Norman’, and Cytometer (Becton, Dickinson and Com- as in the reciprocal crosses between the two ‘Shibori’). Nursery samples, which consisted pany, San Jose, CA.) equipped with an air- cultivars, Albiflora and Gumball, used as of open-pollinated seed heads from many cooled argon ion laser tuned at 488 nm and male parents for all hand-pollinations (data plants per clone pooled into a common con- 15 mW. Fluorescence was collected by a pho- not shown). For all cultivars, the majority of tainer, were dried on arrival. Samples were tomultiplier screened by a long-pass filter. seed germination occurred during the first obtained from the following nurseries: Bailey The results were acquired using the BD 14 d of each test, whereas sporadic germina- Nurseries (Yamhill, OR), Cross Nurseries Inc. CellQuest Pro software, Version 5.2 (Becton, tion continued for up to 8 weeks for some (Lakeville, MN), Johnson’s Nursery Inc. Dickinson and Company, San Jose, CA). clones. Prolonged germination times were (Menomonee Falls, WI), Lawyer Nursery Statistical analysis. Pollen and seed ger- particularly evident in seed from the field Inc. (Plains, MT), McKay Nursery (Waterloo, mination data were analyzed using the Tukey- plot and greenhouse. WI), and Sampson Nursery Inc. (Godwin, Kramer multiple comparisons test with SAS Flow cytometric analysis found the DNA NC). A total of three replicates of 100 seeds for Windows Version 9.1 (SAS Institute Inc., content of the 20 cultivars ranged from from the field plot and nurseries were tested for Cary, NC) using an experimentwise error rate 0.52 to 0.61 pg/2C (Table 2) with a mean of each cultivar. When adequate seed were avail- of a = 0.05. 0.56 pg/2C. Zhao-Yang et al. (2002) karyo- able from the field plot, 100 seeds from each of typed the eight varieties comprising the three replicates of each cultivar were tested. Results S. japonica complex and found seven to be When adequate seed were not available from diploid with the one tetraploid being the field plot, seed from commercial nurseries Cultivars with the highest and lowest mean S. japonica var. fortunei (2n =4x = 36). In was tested to bring the total number of repli- percent pollen germination were in two dis- our study, S. japonica var. fortunei was found cates for each clone to three. In that case, when tinct groups (67.0% to 69.3% and 1.3% to to have a DNA content of 1.14 pg/2C, which possible, tests were performed on seed from 4.0%, respectively) (Table 1), whereas most is very close to double the mean DNA content more than one nursery. For plants demonstrat- clones fell into overlapping groups with pollen of the tested cultivars. This indicates that all ing sterility in the field plot, three replicates of germination between 36.3% and 57.3%. tested clones are diploid. 100 seeds from commercial nurseries were Among the four cultivars with the lowest tested in addition to the three replicates from germination percentage (‘Crispa’, ‘Dart’s Discussion the field plot. For each germination test, 100 Red’, ‘Flaming Mound’, and ‘Neon Flash’), filled seed were placed onto moistened #42 germinating pollen frequently developed This study identified three sterile cultivars filter paper (Whatman Inc., Florham Park, NJ) tubes with abnormal morphology similar to of S. japonica that can be used as landscape
2032 HORTSCIENCE VOL. 44(7) DECEMBER 2009 Table 1. Mean pollen and seed germination for twenty Spiraea japonica cultivars in 2008. commonly fell below 20% (U.S. Department Mean pollen Mean seed germination (%) of Interior, 2009) and was below 12% on 6 d Cultivar germinationz (%) Field ploty Nurseryx Greenhousew Clone total during the period of pollen sampling, which Albiflora 44.3 bcv NSu 99.7 (B, M) — 99.7 a may have affected the duration of pollen Anthony Waterer 36.7 c NS 97.7 (J,S) — 97.7 ab viability during this study. Candlelight 69.3 a — — 91.5 91.5 b The long period of seed germination (up Crispa 2.7 d 0.0 0.0 (L) 0.0 0.0 c to 8 weeks) for some samples could be the Dakota Goldcharm 47.7 bc 96.0 (1) 100.0 (B, M) — 98.7 a result of this seed becoming excessively dry. var. alpina Daphne 41.3 bc 99.0 (1) 96.5 (B, L) — 97.3 ab Although references indicate that no stratifi- Darts Red 3.0 d 0.0 0.0 (C) 0.0 0.0 c cation is required for S. japonica seed (Dirr Flaming Mound 4.0 d NS 100.0 (M) — 100.0 a and Heuser, 1987; USDA Forest Service, Flowering Choice 43.7 bc NS 100.0 (B) — 100.0 a Froebelii 49.3 bc 95.3 — — 95.3 ab 2008), Dirr and Heuser (1987) suggest that Golden Princess 57.3 ab 94.0 — — 94.0 ab if Spiraea seed are allowed to dry, germina- Goldflame 41.7 bc NS 100.0 (B) — 100.0 a tion may be improved by 1- to 2-month cold Goldmound 48.7 bc 100.0 (2) 100.0 (B) — 100.0 a stratification. Calmes and Zasada (1982) Gumball 40.7 c 95.7 — — 95.7 ab found 30-d cold stratification improved the Lemon Princess 67.0 a — — 98.0 98.0 ab germination of Spiraea stevenii. Thus, it is Little Princess 36.3 c 100.0 (1) 100.0 (B, M) — 100.0 a possible that a short period of cold stratifica- Magic Carpet 50.3 bc NS 100.0 (B, M) — 100.0 a tion may have resulted in more uniform seed Neon Flash 1.3 d 0.0 0.0 (L, S) 0.0 0.0 c germination among these samples. Norman 50.0 bc 96.0 (2) 100.0 (M) — 97.3 ab The variation in DNA content among the Shibori 40.7 c NS 100.0 (M, S) — 100.0 a cultivars is possibly the result of differences zData are from three replicates of 100. Pollen from ‘Candlelight’ and ‘Lemon Princess’ are from greenhouse plants; all other pollen collected were from the field plot in Bozeman, MT between 15 July and in chromosome morphology within the 8 Aug. 2008. S. japonica complex. Zhao-Yang et al. yData are from three replicates of 100 unless number of replicates is indicated in parentheses as a result of (2002) found considerable karyomorpholog- inadequate availability of mature seed from the field plot in Bozeman, MT. ical variation in chromosome size and orga- xFor clones that produced inadequate amounts of mature seed before the end of the growing season or nization among the eight taxa comprising the demonstrated sterility in the field plot, additional samples were obtained from commercial nurseries that S. japonica complex. With a wide diversity of are indicated in parentheses: B = Bailey Nursery; C = Cross Nurseries; J = Johnson’s Nursery; L = Lawyer origins, the cultivars in this study (Dirr, 1998) Nursery; M = McKay Nursery; S = Sampson Nursery. Data are from three replicates of 100 for clones that likely include at least some of the chromo- demonstrated sterility or did not produce mature seed in the field plot and for the remainder tests of 100 somal variation observed among the S. ja- were performed to bring the total number of replicates from the field plot and nurseries to three. wData are from three replicates of 100. Clones that demonstrated sterility in the field plot along with ponica taxa. ‘Candlelight’ and ‘Lemon Princess’, which were not in the field plot, were hand-pollinated in the Polyploidy, in particular triploidy, is a greenhouse. common cause of sterility in ornamental vMeans within columns followed by the same letter are not significantly different at a = 0.05 by the Tukey- selections (Datta, 1960; Godo et al., 2004; Kramer multiple comparisons test. Kondo et al., 1993; Saito et al., 2003; Tanaka uNS = No seed matured before the end of the growing season. et al., 2005; Wybe and Peterse, 1999). These sterile plants probably result from chance plants with minimal risk of escape into unreduced gametes contributed by one or natural areas. These three cultivars produced both parents (Otto and Whitton, 2000) and Table 2. Mean 2C DNA content and ploidy of 20 little, if any, viable pollen and no viable seed are then preferentially selected as a result of Spiraea japonica cultivars and S. japonica var. from any of the sources. The other 17 clones pleiotropic effects of polyploidy, e.g., larger fortunei. demonstrated considerable pollen germina- flowers and fruit, longer bloom time, compact 2C DNA tion and female fertility with the exception growth habit, and reduced fruiting (Anurita z y Cultivar content (pg) Ploidy of ‘Flaming Mound’, which exhibited very and Girjesh, 2007; Liu et al., 2007). However, Albiflora 0.55 ± <0.01 2x low pollen viability despite having a high polyploidy is not a factor in the sterility of the Anthony Waterer 0.55 ± 0.01 2x percentage (100%) of seed germination. The Japanese spirea cultivars identified in this Candlelight 0.56 ± 0.01 2x Crispa 0.61 ± <0.01 2x high degree of seed viability of the fertile study, because all clones were determined to Dakota 0.55 ± 0.02 2x cultivars from both the field plot and com- be diploid based on DNA content. Aneuploidy Goldcharm mercial nurseries indicates that adequate is one possible explanation for the sterility of var. alpina 0.53 ± <0.01 2x cross-pollination existed at all locations. these cultivars, particularly if it involves one Daphne These results strongly support the conclusion or only a few chromosomes, because karyo- Darts Red 0.53 ± <0.01 2x that ‘Crispa’, ‘Dart’s Red’, and ‘Neon Flash’ types were not performed, and the inherent Flaming Mound 0.58 ± <0.01 2x are sterile cultivars. variation in DNA content among the cultivars Flowering Choice 0.57 ± <0.01 2x The significantly higher percentage of prohibited detection of small differences in Froebelii 0.58 ± 0.01 2x pollen germination from the two cultivars chromosome number using flow cytometry. Golden Princess 0.54 ± 0.02 2x Goldflame 0.52 ± <0.01 2x grown in the greenhouse, ‘Candle Light’ and The variation in seed and pollen germi- Goldmound 0.55 ± <0.01 2x ‘Lemon Princess’, is most likely the result of nation data among the fertile clones of this Gumball 0.61 ± <0.01 2x more stable conditions in the greenhouse study should not be interpreted as relative Lemon Princess 0.58 ± 0.01 2x relative to the field plot. High temperatures, differences in invasive potential. As described Little Princess 0.53 ± 0.01 2x ultraviolet-B radiation, and large changes in in the beginning of this article, the establish- Magic Carpet 0.54 ± 0.01 2x relative humidity can shorten pollen viability ment and lag phases of successful plant in- Neon Flash 0.54 ± <0.01 2x (Bots and Mariani, 2005) and are conditions vasions are often marked by multiple Norman 0.55 ± 0.01 2x that would be more pronounced in the field. introduction events that contribute to the evo- Shibori 0.54 ± 0.02 2x Although we could not find data concerning lution of invasive genotypes. It is entirely S.j. var. fortunei 1.14 ± 0.02 4x z the effect of relative humidity on pollen possible that, relative to more prolific clones, Data are from three replicates ± SE. All samples are viability for any rosaceous species, Stone from the field plot in Bozeman MT in 2008, except a cultivar with lower seed and/or pollen ger- for ‘Candlelight’ and ‘Lemon Princess’, which et al. (1995) found that low humidity greatly mination may actually produce offspring that were from greenhouse plants. shortened pollen viability in a number of are better locally adapted and/or supply alleles yPloidy based on known tetraploid (2n = 4x = 36) plant families. During the summer of 2008 at that contribute more to the adaptive evolution Spiraea japonica var. fortunei. our field plot, daytime relative humidity of established S. japonica populations. Thus,
HORTSCIENCE VOL. 44(7) DECEMBER 2009 2033 in the absence of extensive study, there is no Carolina. Duke Univ., Durham, Master’s Diss. evolution in invasive species. Trends Plant Sci. way of knowing how the offspring of any 12 Mar. 2009.
2034 HORTSCIENCE VOL. 44(7) DECEMBER 2009