American Journal of Botany 90(9): 1328±1332. 2003. FLORAL LONGEVITY AND REPRODUCTIVE ASSURANCE: SEASONAL PATTERNS AND AN EXPERIMENTAL TEST WITH KALMIA LATIFOLIA (ERICACEAE)1 BEVERLY J. RATHCKE2 Department of Ecology and Evolutionary Biology, University of Michigan, Ann Arbor, Michigan 48109-1048 USA Floral longevity is assumed to re¯ect a balance between the bene®t of increased pollination success and the cost of ¯ower main- tenance. Flowers of Kalmia latifolia (Ericaceae), mountain laurel, have a long duration and can remain viable up to 21 d if unpollinated. I experimentally tested whether this long duration increases pollination success by clipping stigmas to reduce functional ¯oral longevity to 3±4 d. Clipping stigmas decreased fruit set from 65% to only 10%. Flowers with natural life spans were not pollination-limited, demonstrating that long ¯oral duration ensured female reproductive success. The long ¯oral duration of K. latifolia was unique in this site (the Great Swamp, Rhode Island, USA). Co¯owering shrubs in summer had a mean ¯oral life span of 3.4 d. Spring-¯owering species had signi®cantly longer mean ¯oral durations (7.2 d). These duration differences may re¯ect seasonal variation in pollinator availability. However, K. latifolia ¯owers in summer, when its bumble bee pollinators are abundant but it is a poor competitor for bees because its ¯owers produce little nectar. The long ¯oral duration allows K. latifolia to outlast co¯owering competitors and attract suf®cient pollinators. I hypothesize that the long ¯oral duration of K. latifolia functions as a mechanism for competitive avoidance and reproductive assurance. Key words: competition for pollinators; competitive avoidance; Ericaceae; ¯oral longevity; ¯owering phenology; fruit set; Great Swamp, Rhode Island; induced ¯oral senescence; Kalmia latifolia; pollination limitation; reproductive assurance. In pollination studies, much attention has focused upon ¯o- have demonstrated that longer durations of ¯oral life increase ral traits that assure successful pollination by increasing the fruit or seed set of ¯owers in the ®eld. attraction or rewards for animal pollinators (e.g., Waser, 1983; Here I present an experimental test of the pollination bene®t Bell, 1985; Real and Rathcke, 1991; Caruso, 2000) or by in- of long ¯oral duration for mountain laurel, Kalmia latifolia L. creasing the effectiveness of pollen transfer (e.g., Campbell et (Ericaceae). Flowers can remain viable for 2±3 wk if unpol- al., 1996). More recently, it has been recognized that ¯oral linated, and they rapidly senesce after pollination (Rathcke, longevity (the length of time a ¯ower remains open and func- 1988a; Nagy et al., 1999). I reduced the functional longevity tional) is a trait that could ensure successful pollination in of ¯owers by clipping stigmas and then compared fruit sets habitats where pollinators are sparse or uncertain (Primack, (fruits initiated per number of ¯owers) of experimental and 1985; Ashman and Schoen, 1994, 1996; Khadari et al., 1995). control ¯owers. I also did procedure controls, and I tested For example, plants growing in alpine habitats, which typically whether control (naturally pollinated) ¯owers were pollination have few, unpredictable pollinators, have longer ¯oral dura- limited by augmenting ¯owers with cross-pollen. I compare tions than plants at lower elevations with more abundant, pre- the ¯oral longevity of K. latifolia with other shrub species dictable pollinators (Arroyo et al., 1981; Stratton, 1989; Bing- growing in the same habitat and for different ¯owering seasons ham and Orthner, 1998; Blionis and Vokou, 2001; Blionis et (spring and summer). I discuss possible reasons for the main- al., 2001). Ashman and Schoen (1994, 1996) have proposed a tenance of long ¯oral duration in this species. mathematical model in which the duration of ¯oral life is de- termined by a balance between female ®tness accrual rates MATERIALS AND METHODS (pollen receipt), male ®tness accrual rates (pollen dissemina- This study was done in the Great Swamp, Rhode Island, USA (418299 N, tion), and the cost of ¯ower maintenance (see also Charnov, 728329 W) (for a description of the site, see Rathcke [1988a, b]). Kalmia 1996). By measuring pollen deposition and removal and esti- latifolia, mountain laurel, is a tall evergreen shrub that grows in the understory mating maintenance costs, Ashman and Schoen (1994) found of forests and on mountains in eastern North America. Shrubs have tubular, a signi®cant positive correlation between ¯oral longevities ob- pink-white ¯owers displayed in in¯orescences (corymbs) of 10±300 ¯owers, served in the ®eld and predicted ¯oral longevities for 11 plant and individual shrubs may have thousands of ¯owers open at the same time. species. However, they note that there was not a close quan- Bumble bees were the only ¯ower visitors and pollinators of K. latifolia (Rath- titative ®t and that environmental factors, including pollina- cke, 1988a, b). In this low elevation habitat, ¯owers do not self-pollinate tion, may in¯uence ¯oral longevity in the ®eld. Also, they did (Rathcke, 1988a), although ¯owers can self-pollinate at the end of ¯ower life not measure effects on fruit or seed set. No experimental tests in a montane habitat in Virginia (Rathcke and Real, 1993; Levri, 1998; Nagy et al., 1999). Fruits are dry capsules with hundreds of tiny seeds. 1 Manuscript received 22 November 2002; revision accepted 18 April 2003. Reduction of ¯oral durationÐFor the experiment, I tagged four budded The author thanks Lee Kass, Carol Landry, Hitashi Sato, Rachel Simpson, in¯orescences on each of 14 shrubs. For each shrub, two in¯orescences were and two anonymous reviewers, who provided many helpful suggestions on the manuscript; Irene Stuckey, who provided a home for me near the Great assigned to controls and two were assigned to experimental treatments. For Swamp so I could do this research; and William Crepet, who provided me each assigned in¯orescence, new ¯owers (in anthesis) were tagged with color- with an of®ce and facilities in the L. H. Bailey Hortorium during my sab- coded yarn every 2 d throughout the ¯owering season (12±29 June). To reduce batical at Cornell University. the functional ¯oral longevity for female function, I clipped the stigmas of 2 E-mail: [email protected]. all ¯owers after they had been receptive (glistening and sticky) for 3±4 d. I 1328 September 2003] RATHCKEÐFLORAL LONGEVITY OF KALMIA LATIFOLIA 1329 TABLE 1. Percentage fruit sets of Kalmia latifolia ¯owers with natural TABLE 2. Percentage fruit sets of Kalmia latifolia ¯owers in controls ¯oral longevities and with reduced functional longevities (3±4 d) and in two experimental treatments: (1) with stigmas pollinated and created by clipping stigmas (means 6 1 SD). Fruit set 5 100 (fruit/ subsequently clipped and (2) with stigmas clipped and styles im- ¯owers), N 5 number of plants (number of ¯owers). Difference mediately pollinated (means 6 1 SD). Fruit set 5 100 (fruit/¯ow- between control and treatment was tested using Friedman's non- ers), N 5 number of plants (number of ¯owers). Difference be- parametric t test with plant as the sample unit. The asterisk denotes tween control and treatment was tested using Friedman's nonpara- a signi®cant difference; probability level is shown. metric t test with plant as the sample unit. NS indicates no signif- icant difference. Treatment N Fruit set (%) P Fruit Natural longevity 14 (612) 65 6 17.7* ,0.0001 Treatment N set (%) Signi®cance Reduced longevity 14 (568) 10 6 8.3 Control 14 (612) 65 6 17.7 NS Pollinated, then clipped 9 (224) 61 6 16.2 Control 14 (612) 65 6 17.7 not tested chose 3±4 d because most other shrubs that co¯owered with K. latifolia in Clipped, then pollinated 9 (149) 0 June had ¯oral longevities of 3±4 d (Rathcke, 1988b). Stigmas were clipped with scissors just below the sticky surface, and the scissors were sterilized with ethanol after each use. Experimental ¯owers and control ¯owers were same family (Ericaceae), the same comparisons were made using only eri- exposed to natural pollination, and subsequent fruit sets were measured and caceous shrubs. compared. I also did a procedural control treatment to ascertain whether stigma clip- Statistical analysesÐDifferences in fruit sets among different treatments ping harmed ¯owers or normal fruit development. I augmented other ¯owers were tested with nonparametric Friedman's paired-sample sign tests because with cross-pollen and clipped their stigmas the following day. To determine percentage data are not distributed normally and because individual shrubs whether stigma clipping would preclude subsequent pollination and fertiliza- can differ in fruit set for other reasons and can be considered blocks. In this tion as assumed, I clipped stigmas of virgin ¯owers and immediately added test, the ranking in one block (plant) is independent of the ranking in another pollen to the remaining styles. Fruit sets were compared between these treat- block (Sokal and Rohlf, 1981). Tests are based on plants (blocks), not ¯owers, ments and controls. although the numbers of ¯owers are shown. Differences in ¯oral longevities of different species were tested with Student's t tests because the data met Fruit setÐFruit set was calculated as the percentage of ¯owers that initi- the assumption of homogeneity of variances based on Bartlett's tests. Statis- ated fruits. Percentage fruit set was calculated for each in¯orescence, and a tical analyses were done using SYSTAT, version 5.01 (Systat Inc., Evanston, mean was calculated for each individual shrub for each treatment. Fruit ini- Illinois, USA). tiation after a month of development was assumed to indicate that ¯owers had been successfully pollinated and fertilized. Previous studies showed that RESULTS little fruit abortion occurs after initiation, so these values also closely re¯ect ®nal fruit set (Rathcke, 1988a). Seeds were not measured. Floral longevity experimentÐFlowers with reduced func- tional ¯oral longevity had only 10% fruit set, whereas control Pollination limitationÐTo test for pollination limitation of fruit set of ¯ow- ¯owers with natural longevities had 65% fruit set (Table 1).