Kin recognition, not competitive interactions, predicts root allocation in young Cakile edentula seedling pairs

Transcription

1 Research Kin recognition, not competitive interactions, predicts root allocation in young Cakile edentula seedling pairs Mudra V. Bhatt, Aditi Khandelwal and Susan A. Dudley McMaster University, LS-225, 1280 Main Street West, Hamilton, ON L8S4K1, Canada Summary Author for correspondence: Susan A. Dudley Tel: ext Received: 24 August 2010 Accepted: 12 October 2010 (2011) 189: doi: /j x Key words: Cakile edentula, competition, kin recognition, phenotypic plasticity, root allocation, seed dimorphism, self nonself recognition. Recent studies have demonstrated sibling vs stranger differences in group root allocation in plants, suggesting that plants have the potential for kin discrimination in competition. However, morphology differences could potentially be generated by competition-based mechanisms. Here, we tested these hypotheses for the sibling vs stranger differences in root allocation in Cakile edentula. Seeds were planted in pairs of either kin (siblings) or strangers, from all combinations of eight families, to give eight kin (sibling) and 28 stranger pair identities. Because the species has a seed dimorphism, the 10 replicates of each pair identity included both seed types. Root allocation, size inequality between seedlings in a pair, and competitive ability were derived from measures of biomass and height. Cakile edentula seedlings demonstrated the same kin recognition response previously observed in juvenile plants, with lower root allocation in kin pairs than stranger pairs. The seed dimorphism was not associated with root allocation. The two competitive mechanisms, genetic differences in competitive ability and increased size inequality in stranger groups, did not explain the root allocation differences in these seedlings. Kin recognition offered the most probable explanation for the differences in root allocation between sibling and stranger pairs. Introduction Juvenile plants of Cakile edentula sharing a pot demonstrate an apparent kin recognition response, with lower root allocation in groups of siblings than in groups of strangers (Dudley & File, 2007). Because roots are entangled in the shared pots and large sample sizes are needed, the phenotype was measured for groups of plants instead of individuals, with the assumption that equal representation of families within each treatment provides sufficient control for genetic differences. The interpretation of plasticity to relatedness as kin recognition is supported by further studies that used a similar methodology to demonstrate kin recognition in shared pots in Impatiens pallida (Murphy & Dudley, 2009) and Chenopodium album (S. A. Dudley et al., unpublished data), and a study in Arabidopsis thaliana showing plasticity in individual seedlings to root exudates that depended on the relatedness of the exudate source (Biedrzycki et al., 2010). However, even at the juvenile stage, accumulated competitive interactions within the groups could change the group phenotype. Thus, a possible alternative explanation for the sibling and stranger differences in root allocation is that they result from competitive interactions that depend on differences in phenotypic variation between sibling and stranger groups (Klemens, 2008; Masclaux et al., 2010). The kin recognition hypothesis draws from the growing body of research on plant identity recognition. In many species, roots respond differently to roots of the same physiological individual than to roots of other plants, even when the nonself plants are clones (Mahall & Callaway, 1991, 1992, 1996; Maina et al., 2002; Falik et al., 2003; Holzapfel & Alpert, 2003; Gruntman & Novoplansky, 2004). These responses are presumed to prevent competition between different parts of the same plant. Sagebrush (Artemisia tridentata) shows recognition of self nonself that is not rootrelated: plants exposed to volatiles from wounded self plants experience less damage than those exposed to wounded nonself plants (Karban & Shiojiri, 2009). Species-specific responses to belowground neighbours have now been demonstrated in root growth and metabolites (Semchenko et al., 2007; Broz et al., 2010). Pollen characters respond to the genotype of belowground competitors (Lankinen, 2008). Other research compares the performance of groups of siblings with that of groups of strangers to test contrasting Ó 2010 Trust (2011) 189:

2 1136 Research hypotheses: kin selection predicts that siblings will have higher fitness than strangers because siblings cooperate with each other (Hamilton, 1964), while the niche partitioning hypothesis predicts that strangers will therefore have higher fitness than siblings because they differ more in niche use than siblings and so compete less (Young, 1981). The most notable feature of these performance comparisons in plants is the variation in results, with positive, negative, and no fitness benefits to growing with siblings compared with strangers (reviewed in Milla et al., 2009). In A. thaliana, accessions vary in fitness, plant biomass and competitive ability, but do not show fitness benefits of growing with either siblings or strangers (Masclaux et al., 2010). Unlike most such studies, that by Masclaux et al. (2010) controlled genetic variation by creating stranger groups consisting of only two families. Consequently, the results imply an alternative hypothesis to kin selection or niche partitioning; the hypothesis that in larger groups a few families that are highly competitive could disproportionately dominate in competition. This competitive dominance would then depress the overall performance of stranger groups compared with sibling groups, in which weaker competitors contend among themselves (Masclaux et al., 2010). Among-family variation in competitive ability and in root allocation could also potentially bias measures of kin recognition. If the highly competitive families produce most of the biomass in the pot, and those highly competitive families differ in root allocation from less competitive families, then within-pot natural selection for greater competitiveness could affect root allocation in stranger pots (Fig. 1). Measuring root allocation through an analysis of covariance approach (Dudley & File, 2007) removes any associations between size and allocation at the group level. However, size inequality within groups could potentially affect the root allocation results (Klemens, 2008). Groups with the same total plant mass could contain four plants of intermediate size (low size inequality), or one large plant and three small plants (high size inequality). The group phenotype for root allocation (Dudley & File, 2007) is weighted towards the largest plant in the pot (Klemens, 2008). Stranger groups, because of their greater genetic and therefore phenotypic variation, are expected to have greater size inequality than sibling groups. If root allocation is related to size or developmental stage such that larger plants differ in root allocation from smaller plants, differences in root allocation between sibling and stranger groups would result (Fig. 1). These hypotheses provide plausible alternative explanations for sibling vs stranger differences in root allocation, and are consistent with the direction of the sibling vs stranger differences varying among species. As these mechanisms are somewhat related, rather than mutually exclusive, a combined mechanism is also plausible. Taken together, they provide a reasonable summary of competition-based mechanisms to explain away the plant kin recognition result. However, there has yet been no empirical test for them. Here, we used single seed descent lines of C. edentula descended from the field-collected seed families used by Dudley & File (2007) to determine whether kin recognition or competitive interactions best explain sibling vs stranger differences in root allocation. The pair-wise family design (Masclaux et al., 2010) has several advantages for testing the role of competitive interactions in root allocation. Growing pairs of seedlings together, in all possible combinations of families, provides a robust test for the prediction of kin recognition that root allocation will depend only on whether members of the pair are from the same family or from different families. Analysis of variance for the effects of target plant family and competitor family and their interactions on target plant biomass tests several hypotheses (Masclaux et al., 2010). Superior competitive ability is demonstrated by a significant main effect of Fig. 1 Alternative hypotheses for the increased root allocation in groups of strangers compared with siblings in Cakile edentula (Dudley & File, 2007). Top path: increased allocation results from kin recognition, that is, a competitive response to strangers but not siblings (Dudley & File, 2007). Middle path: increased root allocation results if stranger groups have greater size inequality, and if larger plants have more allocation to roots than smaller plants (Klemens, 2008). Bottom path: increased root allocation results if a few genotypes with higher competitive ability dominate stranger groups, and if the genotypes with higher competitive ability have greater root allocation (Masclaux et al., 2010). (2011) 189: Ó 2010 Trust

3 Research 1137 competitor family, superior tolerance to competition is demonstrated by a significant main effect of target plant family, and kin vs stranger effects on biomass are demonstrated by significant interactions between target plant and competitor plant families. The absolute aboveground mass and height differences between the seedlings in a pair provide measures of size inequality to test for its relation with root allocation (Klemens, 2008). In this study, we grew pairs of seedlings in small pots. For eight families, 10 replicates were produced for each possible combination of families, including each family paired with itself. Because C. edentula has a seed dimorphism that affects dispersal and the expected competitive environment, the 10 replicates included both seed types. We measured the dry masses of cotyledons, leaves, and stems for each seedling and the combined root mass for the seedlings sharing a pot. We performed two harvests to study early and late effects of neighbour interaction in C. edentula seedlings. We asked the following questions: Did root allocation depend on relatedness, seed type, or time of harvest? Did aboveground size differences depend on relatedness? Was root allocation correlated with aboveground size difference? Were there among-family differences in competitiveness and root allocation, and was there evidence for family family interactions? Materials and Methods Plant material Cakile edentula (Bigelow) Hook. ssp. edentula var. lacustris Fernald (Great Lakes sea rocket), is an annual beach plant found in high-light, low-water, low-nutrient environments. This species has a seed dimorphism (Rodman, 1974) which creates diversity in the social environment of seedlings. Each flower develops into a fruit with two segments, each with a single seed, which probably result from the same pollination event (Rodman, 1974). The lower or proximal fruit segments are strongly attached to the maternal plant, while the upper or distal fruit segments are loosely attached to the proximal segment. Seeds that remain close to the maternal plant emerge in high-density kin clumps (Rodman, 1974; Keddy, 1982), while those that are dispersed farther away, usually by wind or water, may emerge in high-density groups of strangers or be solitary (Payne & Maun, 1981; Donohue, 1998). Distal fruit segments contain larger seeds and have better floating ability than proximal fruit segments. As proximal seeds are more likely to germinate in high-density kin clumps, they are more likely to experience kin selection than distal seeds. Cakile edentula has been shown to recognize kin (Dudley & File, 2007) and to have higher fitness in kin groups than in stranger groups (Donohue, 2003), but these studies did not measure seedtype effects. The seeds used were taken from parents grown in the glasshouse that had been allowed to naturally self-pollinate. The parents were field-collected as seeds from maternal sibships originally collected in October 2005, from one population at Confederation Park in Hamilton, Ontario, Canada (Dudley & File, 2007). All families derived from single seed descent. Experimental protocol For the seed mass and allocation study, 20 seeds per family, evenly divided between proximal and distal seeds, from nine different families were imbibed after the pericarp was removed. After 18 h, seed coats were removed and the cotyledons and radicle were separated. The wet masses of the cotyledons (g) and the radicle (g) were recorded separately for each seed. For the seedling competition study, on 6 July 2009, the pericarp was removed and the seeds were scarified with sterilized coarse sand and then soaked in distilled water in individual wells for 18 h. Because the seed coats inhibit germination, they were removed from all seeds before planting to synchronize emergence. On 7 July 2009, two seeds were sown per 9-cm pot, c. 1 inch (2.54 cm) apart and 0.5 inches (1.27 cm) deep in a mixture of three parts coarse sand and one part Turface (Profile Products LLC, Buffalo Grove, IL, USA). Most seedlings emerged on days 5 and 6 after planting. The experimental design included all possible pairings of the eight families (36 combinations, with eight possible kin pairs and 28 stranger pairs). Every family pair combination was replicated five times for distal distal pairs and five times for proximal proximal pairs. Consequently, there were more stranger pots than kin pots. The resulting 360 pots were arranged randomly on a single bench in the glasshouse. Density was high (267 seedlings per m 2 ). To control for edge effects and local environmental effects, pots were randomly rearranged weekly. Plants received fluorescent and incandescent lighting in addition to natural sunlight. The plants were watered daily. Beginning 7 d following planting, seedlings were fertilized once a week with 200 ppm NPK. Density, irradiance and nutrients were kept constant for every pot. Half of the pots were harvested at 20 d of age (early harvest). For the early harvest, hypocotyl height (cm), plant height (cm), number of cotyledons and number of leaves were measured. Pots where either one or both seedlings died were not measured. Leaves, cotyledons, stems and roots were separated and dried in an oven at 30 C overnight. The dry masses of leaves (g), cotyledons (g) and stem (g) were recorded for each seedling. A combined root mass (g) was measured for all roots found in one pot that were produced by both seedlings. Ó 2010 Trust (2011) 189:

4 1138 Research Because of an error in the watering protocol, the late harvest plants experienced a drought event and died, on approximately 14 August 2009 (late harvest). For these plants, we measured only the combined aboveground mass and root mass, except for pots with fungal growth. Statistical analysis The data were primarily analysed using the General Linear Model in the SAS statistical software (version 9.2 for Windows (English); SAS Institute, Cary, NC, USA). Analyses of variance (ANOVA) and covariance (ANCOVA) were carried out using the PROC GLM command. Residual analysis was performed to check whether the residuals met the assumptions of the ANOVA, and, if necessary, natural logarithms (log e ) were taken of raw data so that residuals were homoscedastic and normally distributed. To test whether size differed between distal and proximal seed types, an analysis of variance (ANOVA) was performed for log e seed total mass, log e cotyledon mass and log e radicle mass with seed type and family as independent variables. Least square means (lsmeans) was measured for each dependent variable with the LSMEANS option in PROC GLM. To determine if the radicle : cotyledon ratio differed between distal and proximal seed types, ANOVA was performed on the quotient of radicle mass divided by cotyledon mass as the dependent variable, with seed type and family as independent variables. For the competition experiment, effects of treatments on root allocation were measured in an analysis of covariance (ANCOVA) with log e combined root mass as the dependent variable and log e combined aboveground mass as the covariate (PROC GLM). Because root mass could not be separated, all root allocation results are for a pair of seedlings, not for individuals. We dropped four observations with extremely low mass as outliers, which did not affect the major results (see Supporting Information Table S1 for final number of observations per family and treatment). All possible interactions of the treatments seed type (proximal or distal), relatedness (kin or stranger) and harvest (early or late) were estimated. Root allocation was estimated for a treatment combination as the least square mean (lsmean) from the ANCOVA (LSMEANS option, PROC GLM). Despite the death by drought for plants in the second harvest, root allocation did not differ between the harvests, nor were there any significant treatment interactions with harvest, giving no rationale for dropping the second harvest data. We used only the first harvest data to estimate the effects of competitive interactions on root allocation because individual seedling measurements were not available for the second harvest. To measure competitive ability, an ANOVA was performed with individual seedling aboveground mass as the dependent variable and seed type, focal plant family, competitor family and their interactions as treatments. In this analysis observations were not independent because each seedling was used as both focal plant and a competitor plant. Therefore, we carried out the ANOVA procedure in R, using the Companion to Applied Regression (car) library (John Fox, McMaster University, Hamilton, Canada) to obtain Type II sums of squares. Statistical significance was obtained from runs of a Montecarlo analysis, using Python to randomize the pot identity while keeping the data from pairs of seedlings in the same pot together, and then calling the R ANOVA procedure. We calculated two measures of size inequality: the absolute values for height differences and aboveground mass differences between the seedlings in a pair. To measure the effects of relatedness on size inequality, ANOVAs were conducted on log e height difference and log e mass difference as the dependent variables and seed type (proximal or distal) and relatedness (kin or stranger) and their interaction as treatments. An estimate of root allocation for each pair was obtained from a regression of log e root mass on log e combined aboveground mass, with root allocation for an observation as the sum of its residual and the overall mean. Pearson correlations were estimated among the pair traits of aboveground mass difference, height difference and root allocation (PROC CORR). To determine whether relatedness or size inequality predicted root allocation, ANCOVAs for log e root mass were carried out with log e aboveground mass and a size inequality measure (ln aboveground mass difference mass or log e height difference) as covariates and seed type (proximal or distal) and relatedness (kin or stranger) as treatments. To explore family family interactions, an ANCOVA was conducted over both harvests with log e root mass as the dependent variable, log e aboveground mass as the covariate (PROC GLM) and pair identity (e.g. AA, AB, etc.) as the treatment. A priori contrasts were applied to the pair identity variation to test for variance among stranger pairs, variance among kin pairs, and differences between kin and stranger pairs (CONTRAST statements, PROC GLM). Lsmeans for root allocation were obtained for each pair identity (LSMEANS option, PROC GLM). Results Differences between proximal and distal seeds The wet mass of the embryo components from the imbibed seeds depended on the seed type. Distal seeds had greater total embryo mass (distal = 10.8 ± 0.2 mg, mean ± SE; proximal = 5.5 ± 0.9 mg; F 1,342 = 823.5; P < ), cotyledon mass (distal =7.4 ± 0.1 mg; proximal = 3.5 ± 0.7 mg; F 1,342 = ; P < ) and radicle mass (distal = 3.3 ± 0.5 mg; proximal = 2.0 ± 0.3 mg; F 1,342 = 507.3; P < ). Radicle to cotyledon ratio was higher for proximal seeds (distal = ± 0.009; proximal = ± g; F 1,342 = 114.8; P < ). (2011) 189: Ó 2010 Trust

5 Research 1139 Table 1 Root allocation analysis for pairs of Cakile edentula seedlings grown in the glasshouse, from an analysis of covariance with the natural logarithm of combined root mass as the dependent variable and natural logarithm of combined aboveground mass as the covariate, including both harvests (Fig. 2) Source Kin recognition As expected, the root mass and aboveground mass were positively correlated (Table 1, Fig. 2). In both harvest 1 and harvest 2, stranger pairs averaged a higher root mass for their aboveground mass than kin pairs (Table 1, Fig. 2), indicating that root allocation was higher for stranger seedling pairs than kin pairs. There were no effects of harvest or seed type on root allocation (Table 1, Fig. 2). Competitive mechanisms Aboveground biomass of focal plants at harvest 1 was strongly affected by focal plant family, seed type, and their interaction. There was no direct effect of competitor family, df Log e (root mass) F-ratio P-value Log e (aboveground mass) Seed type (S) Relatedness (R) Harvest (H) S R S H R H S R H Error df is 298. n = 307. Seed type is either distal or proximal; relatedness is kin or stranger; harvest is early or late. Overall model significance is P < Table 2 Analysis of variance for effects of seed type, focal genotype, and competitor genotype and their interactions on aboveground mass for Cakile edentula seedlings grown in pairs for harvest 1 Source df Aboveground mass F-ratio P-value Focal genotype (F) Competitor genotype (C) Seed type (S) F C F S C S F C S Error df is 168. n = 269. Each member of the pair is considered as both a focal plant and a competitor plant in this analysis. Seed type is either distal or proximal. Overall model significance is P < nor an overall focal family competitor family interaction (Table 2), indicating little significant variation in competitive ability. However, the focal family competitor family seed type interaction was significant, with a greater focal family competitor family interaction in the proximal seeds. The average aboveground dry biomass for individual seedlings was greater in distal than proximal seeds (distal = 16.4 ± 0.5 mg; proximal = 14.1 ± 0.5 mg), but the significant seed type family interaction indicated that families varied in how the seed type affected biomass (Table 2). The effects of size inequality differed between aboveground mass and height. Relatedness did not affect the aboveground mass difference between seedlings in a pair (Table 3, Fig. 3a), but aboveground mass difference was correlated with root allocation (r =0.16,P <0.05;n =156). Stranger pairs did exhibit greater height differences than kin pairs (Table 3, Fig. 3b), but height difference was not correlated with root allocation (r = 0.11, P < 0.16; n = 150). In the multivariate approach, when relatedness and a measure of size inequality were included in the ANCOVA model, relatedness remained highly significant, and size inequality measures had no explanatory value (Table 3). Family family interactions Root allocation showed significant among-pair variation (Table 4, Fig. S1). A priori contrasts demonstrated that root allocation differed between kin and strangers, but there were no significant differences among stranger pairs, or among kin pairs (Table 4). Fig. 2 Scatter plot of the natural logarithms of root mass vs aboveground mass for pairs of glasshouse-grown Cakile edentula seedlings. Kin pairs were siblings from the same family, and stranger pairs were derived from two different families. Harvest 1 was 20 d after planting and harvest 2 was c. 40 d after planting. Discussion In this study, we asked whether pairs of C. edentula seedlings would show the same kin recognition response observed in groups of four juvenile plants (Dudley & File, 2007). Again, we found kin recognition in competition, Ó 2010 Trust (2011) 189:

6 1140 Research Table 3 Analyses of variance for pairs of Cakile edentula seedlings from harvest 1 to determine if seed type and seedling relatedness affect two measures of size inequality, (a) mass difference and (b) height difference; and the corresponding analyses of covariance as in Table 2, but with a size inequality measure, either (a) mass difference or (b) height difference as another independent variable Source df Log e (mass difference) Log e (root mass) F-ratio P-value F-ratio P-value (a) Seed type (S) Relatedness (R) S R Log e (aboveground mass) Mass difference Log e (height difference) F-ratio P-value (b) Seed type (S) Relatedness (R) S R Log e (aboveground mass) Height difference Mass difference is the difference between the aboveground masses of larger and smaller seedlings in the same pot, and height is the difference between the main stem heights of larger and smaller seedlings in the same pot. For mass difference in analysis of variance, n = 156, error df is 152, and overall model significance is P < For the log e (root mass), mass difference, in analysis of covariance, n = 156, error df is 144, and overall model significance is P < For height difference in analysis of variance, n = 150, error df is 146, and overall model significance is P < For the log e (root mass), height difference, in analysis of covariance, n = 150, error df is 144, and overall model significance is P < Seed type is either distal or proximal; relatedness is kin or stranger. with lower root allocation in siblings than strangers (Dudley & File, 2007). Because we used a pair-wise experimental design (Masclaux et al., 2010), we could further test whether competition-based mechanisms explained the difference in root allocation between siblings and strangers. However, we found relatively little evidence for any competition-based mechanisms other than kin recognition, and no evidence that such competition-based mechanisms affected root allocation. The only significant predictor of root allocation was whether the individuals in the pot were kin or strangers. In the present study there was no difference between the seed types in plasticity of root allocation to kin vs strangers. Though the proximal and distal seeds differed markedly in size and allocation, few differences between seed types remained at the seedling stage. The cotyledon mass was higher and the radicle to cotyledon mass ratio was lower in distal seeds, suggesting that distal seeds had increased maternal investment in food reserves. Increased provisions may be adaptive as they can increase successful germination of distal seeds under unfavourable conditions (Zhang, 1993, 1994). Cakile edentula seedlings from larger seeds produce leaves and branches more rapidly and reproduce earlier than seedlings from smaller seeds (Zhang, 1996). This gives larger distal seedlings a competitive advantage when growing amongst strangers. However, distal and proximal seedlings did not differ in root allocation or in response to relatedness. Thus, despite the expected differences in dispersal, distal and proximal seed types were not predisposed towards competing with strangers or cooperating with kin. We tested two overlapping hypotheses for how sibling vs stranger differences in root allocation could arise without kin recognition. Both are two-part mechanisms, postulating first that there are more size differences among plants in stranger groups than those in sibling groups, and secondly that larger plants differ in root allocation from smaller plants (Fig. 1). In one mechanism (Masclaux et al., 2010), (a) (b) Fig. 3 Scatter plot of pair root allocation vs two size inequality measures for harvest 1: (a) aboveground mass difference and (b) height difference (kin, closed circles; stranger, open circles). Pair root allocation is estimated from a covariate correction for root allocation using a regression of log e root mass on log e aboveground mass, with root allocation for an observation as the sum of its residual and the overall mean. (2011) 189: Ó 2010 Trust

7 Research 1141 Table 4 Analysis of covariance with root mass as the dependent variable and pair identity and aboveground mass as independent variables, including both harvests Source genetic variation in competitive ability creates the size difference, and genetic variation in root allocation, if correlated with size, creates the change in root allocation. In the present study, aboveground mass was independent of the family of the competitor, indicating that no one particular family competed more fiercely than others at this early life stage. Aboveground mass did differ among families, with that difference depending on seed type. There was a minor interaction for focal competitor seed type which implied more focal competitor interactions in the smaller proximal seedlings. However, because of the small sample sizes, this needs further investigation. Root allocation did not differ among the kin pairs, indicating a lack of genetic variation for root allocation. Thus, there is no evidence that a competition mechanism involving genetic differences in competitiveness and root allocation could explain sibling vs stranger differences. Consequently, the hypothesis that plasticity to relatedness resulted from uncontrolled genetic variation was not supported in this study. Another competitive interaction hypothesis for root allocation differences between sibling and stranger groups (Klemens, 2008) suggests that greater phenotypic variation in stranger groups, potentially enhanced by asymmetric competition for light, creates the size differences. Then, allometry results in differences in root allocation between larger and smaller plants. We tested the predictions from this mechanism that stranger pairs will have greater size inequality than kin pairs, and that the size inequality of a pair affects its root allocation. Stranger pairs showed a greater height difference than kin pairs, but kin and stranger pairs had similar aboveground mass differences. This pattern may simply reflect differences in underlying genetic variation for these traits, but could potentially indicate differences in competitive processes in sibling and stranger pairs. Because aboveground competition is asymmetric and driven by height differences (Schwinning & Weiner, 1998), height inequality provides a mechanism for developing greater size df Log e (root mass) F-ratio P-value Log e (aboveground mass) Pair identity Preplanned contrast for pair identity Among kin Among strangers Kin vs stranger Pair identity refers to the genotypes of the members of the pair (AA, AB, AC, etc.). Preplanned contrasts were used to separate out the components of the pair identity variation. n = 307, error df = 270. Model significance is P < inequality in stranger groups than siblings over time. But neither height nor mass differences explained the observed root allocation differences in the kin and stranger pairs. Thus, despite some support for the hypothesis that size inequality is greater in stranger groups, these results do not support the hypothesis of Klemens (2008) that bias in root allocation created by size differences within a group explains differences between the kin and stranger treatments. Competition-based mechanisms clearly did not explain root allocation differences between siblings and strangers in this study. However, in these relatively young seedlings, little resource limitation and competitive interaction would be expected. Thus, in larger plants these competitive processes may affect root allocation as resources become limiting and plants shade one another. In fact, the increased height inequality seen in stranger groups offers a potential mechanism for the continued generation of increased size inequality in stranger groups as the plants grow and interact to a greater extent. The impact of genetic diversity on competitive processes could modify initial responses to relatedness of competitors. The results of the present study not only allow rejection of the alternative competition-based hypotheses but also support the predictions from the kin recognition hypothesis of consistent sibling vs stranger differences for seedling pairs. However, more work needs to be done to determine whether these changes do affect belowground competitive ability. Remarkably, the kin recognition response was present in very young seedlings and so, like the stem elongation response to density (Ballare et al., 1990), may allow a plant to anticipate competition and identify the nature of its competitors before any resource depletion occurs. Acknowledgements We thank Amanda File, Guillermo Murphy, Sinah Lee, Sunny Attarde and Mandip Puri for help in measuring and harvesting plants; Lovaye Kajiura, Jonathan Stone and Robin Cameron for useful discussions; and Arthur Yeas for glasshouse care. We also thank Peter Chu for the Montecarlo program, and three anonymous reviewers for helpful comments on an earlier draft of this paper. This research was supported by an NSERC Discovery Grant to S.A.D. References Ballare CL, Scopel AL, Sanchez RA Far-red radiation reflected from adjacent leaves: an early signal of competition in plant canopies. Science 247: Biedrzycki ML, Jilany TA, Dudley SA, Bais HP Root exudates mediate kin recognition in plants. Communicative and Integrative Biology 3: 1 8. Broz AK, Broeckling CD, De-La-Pena C, Lewis MR, Greene E, Callaway RM, Sumner LW, Vivanco JM Plant neighbor identity Ó 2010 Trust (2011) 189:

8 1142 Research influences plant biochemistry and physiology related to defense. BMC Plant Biology 10: 115. Donohue K Maternal determinants of seed dispersal in Cakile edentula: fruit, plant, and site traits. Ecology 79: Donohue K The influence of neighbor relatedness on multilevel selection in the Great Lakes sea rocket. American Naturalist 162: Dudley SA, File AL Kin recognition in an annual plant. Biology Letters 3: Falik O, Reides P, Gersani M, Novoplansky A Self non-self discrimination in roots. Journal of Ecology 91: Gruntman M, Novoplansky A Physiologically mediated self nonself discrimination in roots. Proceedings of the National Academy of Sciences, USA 101: Hamilton WD The genetical evolution of social behavior, I & II. Journal of Theoretical Biology 7: Holzapfel C, Alpert P Root cooperation in a clonal plant: connected strawberries segregate roots. Oecologia 134: Karban R, Shiojiri K Self-recognition affects plant communication and defense. Ecology Letters 12: Keddy PA Population ecology on an environmental gradient: Cakile edentula on a sand dune. Oecologia 52: Klemens J Kin recognition in plants? Biology Letters 4: Lankinen A Root competition influences pollen competitive ability in Viola tricolor: effects of presence of a competitor beyond resource availability? Journal of Ecology 96: Mahall BE, Callaway RM Root communication among desert shrubs. Proceedings of the National Academy of Sciences, USA 88: Mahall BE, Callaway RM Root communication mechanisms amd intracommunity distributions of two Mojave Desert shrubs. Ecology 73: Mahall BE, Callaway RM Effects of regional origin and genotype in intraspecific root communication in the desert shrub Ambrosia dumosa (Asteraceae). American Journal of Botany 83: Maina GG, Brown JS, Gersani M Intra-plant versus inter-plant root competition in beans: avoidance, resource matching or tragedy of the commons. Plant Ecology 160: Masclaux F, Hammond RL, Meunier J, Gouhier-Darimont C, Keller L, Reymond P Competitive ability not kinship affects growth of Arabidopsis thaliana accessions. 185: Milla R, Forero DM, Escudero A, Iriondo JM Growing with siblings: a common ground for cooperation or for fiercer competition among plants? Proceedings of the Royal Society B-Biological Sciences 276: Murphy GP, Dudley SA Kin recognition: competition and cooperation in Impatiens (Balsaminaceae). American Journal of Botany 96: Payne AM, Maun MA Dispersal and floating ability of dimorphic fruit segments of Cakile edentula var lacustris. Canadian Journal of Botany 59: Rodman JE Systematics and evolution of the genus Cakile (Cruciferae). Contributions of the Gray Herbarium, Harvard University 205: Schwinning S, Weiner J Mechanisms determining the degree of size asymmetry in competition among plants. Oecologia 113: Semchenko M, John EA, Hutchings MJ Effects of physical connection and genetic identity of neighbouring ramets on rootplacement patterns in two clonal species. 176: Young JPW Sib competition can favor sex in two ways. Journal of Theoretical Biology 88: Zhang JH Seed dimorphism in relation to germination and growth of Cakile edentula. Canadian Journal of Botany 71: Zhang JH Early seedling development in relation to seed mass and morph in Cakile edentula. Canadian Journal of Botany 72: Zhang JH Seed mass effects across environments in an annual dune plant. Annals of Botany 77: Supporting Information Additional supporting information may be found in the online version of this article. Fig. S1 Root allocation for each pair identity. Table S1 Sample size for each of the single seed descent families of Cakile edentula for the main treatment combinations Please note: Wiley-Blackwell are not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing material) should be directed to the Central Office. (2011) 189: Ó 2010 Trust