Potential Maternal Effects of Elevated Atmospheric CO2 on Development and Disease Severity in a Mediterranean Legume

Global change can greatly affect plant populations both directly by influencing growing conditions and indirectly by maternal effects on development of offspring. More information is needed on transgenerational effects of global change on plants and their interactions with pathogens. The current study assessed potential maternal effects of atmospheric CO2 enrichment on performance and disease susceptibility of first-generation offspring of the Mediterranean legume Onobrychis crista-galli. Mother plants were grown at three CO2 concentrations, and the study focused on their offspring that were raised under common ambient climate and CO2. In addition, progeny were exposed to natural infection by the fungal pathogen powdery mildew. In one out of 3 years, offspring of high-CO2 treatments (440 and 600 ppm) had lower shoot biomass and reproductive output than offspring of low-CO2 treatment (280 ppm). Disease severity in a heavy-infection year was higher in high-CO2 than in low-CO2 offspring. However, some of the findings on maternal effects changed when the population was divided into two functionally diverging plant types distinguishable by flower color (pink, Type P; white, Type W). Disease severity in a heavy-infection year was higher in high-CO2 than in low-CO2 progeny in the more disease-resistant (Type P), but not in the more susceptible plant type (Type W). In a low-infection year, maternal CO2 treatments did not differ in disease severity. Mother plants of Type P exposed to low CO2 produced larger seeds than all other combinations of CO2 and plant type, which might contribute to higher offspring performance. This study showed that elevated CO2 potentially exerts environmental maternal effects on performance of progeny and, notably, also on their susceptibility to natural infection by a pathogen. Maternal effects of global change might differently affect functionally divergent plant types, which could impact population fitness and alter plant communities.

Global change can greatly affect plant populations both directly by influencing growing conditions and indirectly by maternal effects on development of offspring. More information is needed on transgenerational effects of global change on plants and their interactions with pathogens. The current study assessed potential maternal effects of atmospheric CO 2 enrichment on performance and disease susceptibility of first-generation offspring of the Mediterranean legume Onobrychis crista-galli. Mother plants were grown at three CO 2 concentrations, and the study focused on their offspring that were raised under common ambient climate and CO 2 . In addition, progeny were exposed to natural infection by the fungal pathogen powdery mildew. In one out of 3 years, offspring of high-CO 2 treatments (440 and 600 ppm) had lower shoot biomass and reproductive output than offspring of low-CO 2 treatment (280 ppm). Disease severity in a heavy-infection year was higher in high-CO 2 than in low-CO 2 offspring. However, some of the findings on maternal effects changed when the population was divided into two functionally diverging plant types distinguishable by flower color (pink, Type P; white, Type W). Disease severity in a heavy-infection year was higher in high-CO 2 than in low-CO 2 progeny in the more disease-resistant (Type P), but not in the more susceptible plant type (Type W). In a low-infection year, maternal CO 2 treatments did not differ in disease severity. Mother plants of Type P exposed to low CO 2 produced larger seeds than all other combinations of CO 2 and plant type, which might contribute to higher offspring performance. This study showed that elevated CO 2 potentially exerts environmental maternal effects on performance of progeny and, notably, also on their susceptibility to natural infection by a pathogen. Maternal effects of global change might differently affect functionally divergent plant types, which could impact population fitness and alter plant communities. a few species (Farnsworth and Bazzaz, 1995;Andalo et al., 1996;Grünzweig and Körner, 2000;Edwards et al., 2001;Thürig et al., 2003;Hovenden et al., 2008). Subsequent production of shoot and root biomass was ambivalent, and both increased and decreased biomass was observed in high-CO 2 compared with low-CO 2 progenies when grown at ambient CO 2 (Fordham et al., 1997;Huxman et al., 1998;Edwards et al., 2001).
Maternal effects on plant-pathogen interactions have rarely been shown in any system. In a tropical tree, the frequency of damping-off disease of seedlings depended on seed dispersal and seedling density (Augspurger, 1983), which can be influenced by the provision of dispersal traits by the mother plant (Donohue and Schmitt, 1998). Genetic studies in corn revealed unspecified maternal effects on resistance to Fusarium seedling blight and ear rot (Lunsford et al., 1976;Gendloff et al., 1986). In Arabidopsis thaliana, an elicitor of plant defense (a treatment applied to the mother plant only) induced genetic changes (enhanced somatic homologous recombination) that were passed down to the progeny, suggesting an epigenetic mode of inheritance (Molinier et al., 2006). To the best of my knowledge, maternal effects on offspring susceptibility to pathogens have not been shown in response to global-change drivers so far.
For a number of species, significant CO 2 × genotype interactions for biomass and reproductive output were detected, while most species did not show such interactions (Lau et al., 2007). Maternal effects of CO 2 enrichment were rarely studied on different genotypes within a species or on different plant types within a population. Maternal CO 2 concentration interacted significantly with genotype for germination success in offspring of wild populations of A. thaliana that originated from different environments (Andalo et al., 1996). Some genotypes germinated at lower rates when seeds were produced under high maternal CO 2 concentration, while for other genotypes no differences among maternal CO 2 treatments were observed.
The objective of the current study was to assess potential maternal effects of elevated CO 2 on performance of offspring of the Mediterranean legume Onobrychis crista-galli (L.) Lam. (cock's comb sainfoin) under ambient climate and CO 2 conditions. Mother plants were grown in an original experiment with species-rich assemblages under three different CO 2 concentrations that spanned a range between pre-industrial and future levels (280, 440, and 600 ppm CO 2 ). O. crista-galli was the most responsive species to CO 2 enrichment in the original experiment, with an increase of 30-150% in number and mass of fruits and seeds per individual at elevated CO 2 (Grünzweig and Körner, 2001a). Specific aims of the study were (1) assessing maternal effects of elevated CO 2 on seedling emergence, growth, and development of progenies with and without consideration of functionally diverging plants types within the same population of O. crista-galli; (2) analyzing susceptibility of progenies to the fungal disease powdery-mildew under natural infection.

MaterIals and Methods
Onobrychis crista-galli is a common annual legume in the deserts, shrublands, and grasslands of the Middle East and northern Africa. It is one of the larger annual legumes in the northern Negev of Israel, and has a low to intermediate degree of nodulation with symbiotic dinitrogen fixing bacteria under natural conditions (Hely and Ofer, 1972;Atallah et al., 2008). The species was characterized as being relatively mesic according to an analysis of reproductive traits along an aridity gradient (Ehrman and Cocks, 1996).
Seeds of O. crista-galli were collected at the Lehavim Long-Term Ecological Research (LTER) site in the semi-arid northern Negev, Israel (400 m a.s.l., 31º21′N, 34º51′E; Mediterranean climate, with mild winters and 300 mm precipitation, hot summers, and no rain) in 1996-1997. Mother plants were grown on native light lithosol as part of a species-rich community in model ecosystems (large containers of 100 cm × 70 cm surface area and 35 cm depth) subjected to three CO 2 treatments (280 ppm, pre-industrial CO2; 440 ppm, CO2 concentration expected by the year ∼2025 according to the IPCC SRES scenario A1B; 600 ppm CO2, expected by the years 2060-2075IPCC, 2007). Model ecosystems were placed in growth chambers, and were subjected to a dynamic climate simulation over the entire growing season of 5 months. Climate simulations included weekly adjustment of light intensity, temperature, precipitation, and relative air humidity to the average natural conditions at the Lehavim LTER site (for more details see Grünzweig and Körner, 2001b). CO 2 treatments were replicated by three model ecosystems in each of the three growth chambers (one chamber per CO 2 treatment). Chamber effects were minimized by a random weekly reallocation of CO 2 treatments to chambers (Grünzweig and Körner, 2001b). Pseudoreplication could be prevented by weekly rearranging position and orientation of model ecosystems within chambers. Reallocation of chambers and repositioning of model ecosystems were performed 23 times in total during the experiment. Mother plants did not show any disease symptoms. Following seed maturation at the end of the growing season 1997, seeds were collected, stored for 28 months at room temperature and frozen thereafter at −20°C for preservation.
This seed stock was used for experiments during the growing seasons of 2005, 2006, and 2007, i.e., each year seeds were randomly chosen from the same stock to study first-generation offspring performance. Each year, seeds from the three original CO 2 treatments were slightly scarified with sandpaper, sown into soil collected at the Lehavim LTER site, and offspring seedlings were grown under ambient CO 2 conditions (∼380 ppm). Plants were grown in trays with conical compartments of 13 cm depth and 420 cm 3 volume in 2005 and of 15 cm depth and 320 cm 3 volume in 2006 and 2007 (one seedling per compartment; Quickpot, HerkuPlast-Kubern, Ering/Inn, Germany). Trays were placed in a net-house facility at the Faculty of Agriculture, Food and Environment, and subjected to ambient conditions, except for precipitation. Daily average temperature during the main growing season varied between 12.5°C in January and 18.0°C in April, with average monthly temperatures varying among years by up to 1.5°C. The net-house decreased photosynthetic photon flux density (PPFD) by about 10% relative to total incoming PPFD. To adjust precipitation to the semi-arid conditions at the Lehavim LTER site, rainfall was intercepted by a transparent polyethylene sheet that was set up only during major rain events. Water was added to trays at 4-days intervals in 2005 and at 6-day intervals in 2006 and 2007. These intervals allowed transient surface dry-out of the soil as common under field conditions (Grünzweig and Körner, 2001b). Plants were sown in January Seed size was recorded as air-dry mass for each individual seed after seed moisture was adjusted in a growth chamber at 22-24°C and 40-60% relative humidity for 24 h. Seedling emergence was determined at the first aboveground appearance of the cotyledons. Anthesis was defined as opening of the first flower of an individual plant. Biomass was not analyzed in 2007 because plants were harvested only after desiccation and seed maturation to record offspring seed size.
Powdery-mildew naturally infected seedlings in all 3 years. The pathogen was identified as Erysiphe martii Lév, which widely infects Onobrychis plants of different species in Israel (Rayss, 1940) and elsewhere (Mühle and Frauenstein, 1970;Karaboz and Öner, 1982;Amano, 1986). In a preliminary test in 2005, disease severity was estimated for each leaf of each plant by rating cover of fungal mat on an increasing scale from 0 to 3 (from no cover to full cover by fungal mat, respectively) at one point in time. Disease severity was more profoundly assessed in 2006 and 2007 by assigning a severity value to each leaf at various measuring dates according to the following scale: 0 (0% leaf cover by fungal mat), 0.05 (1-5%), 0.2 (6-20%), 0.5 (21-50%), 1.0 (51-100%). For both ways of rating fungal mat cover, the values of all leaves of a plant were added up and divided by the number of leaves to express disease severity at the plant level on a scale ranging from 0 to 1. Area under diseaseprogress curves (AUDPC) was calculated over the period between disease onset (i = 1) and the last record of disease progress (i = n; Figure 2), as follows (Shaner and Finney, 1977): where DS i is disease severity at the end of period i and L i is the length of period i in days. Disease onset was defined here as the day when symptoms became visible on leaves.
Data were analyzed for the following two cases: Case 1 assumed a homogeneous population without diverging plant types. This is the common case for most studies where no phenotypic or genetic information exists to allow splitting up populations into different plant types. Case 2 shows the current situation where distinct plant types can be defined in the population, here by flower color. Plant performance and disease severity were analyzed by one-way mixedmodel ANOVA for Case 1 and two-way mixed-model ANOVA for Case 2. The one-way ANOVA for Case 1 included maternal CO 2 as fixed factor and model ecosystem nested within maternal CO 2 as random factor. The random factor was a consequence of sampling seeds from the three model ecosystems per maternal CO 2 treatment in the original CO 2 -enrichment experiment. To test potential interactions of maternal CO 2 with plant type in Case 2, a two-way ANOVA was performed that included maternal CO 2 , plant type and their interactions as fixed factors, and model ecosystem nested within maternal CO 2 and the interaction of model ecosystem and plant type as random factors. Homogeneity of variance was tested by Bartlett's test, and data transformation (Box-Cox) was carried out where necessary. In disease-progress studies, time was added as an additional fixed factor to the analysis. Multiple comparisons among levels of fixed factors were performed by the Tukey-Kramer honestly significant difference (HSD) test. ANCOVAs with seed size as covariate showed the same results as the above mentioned each year, and were harvested close to peak season (but after fruit set) in 2005 and 2006, and at the end of the season upon plant dehydration in 2007.
In addition to maternal CO 2 , the inclusion of plant type in the experimental design was tested. The following two plant types were detected in the offspring population: a pink-flowering type and a white-flowering type, designated "Type P," and "Type W," respectively ( Figure 1). Both plant types belonged to the variety O. crista-galli ssp. crista-galli var. crista-galli (L.) Lam. (Heyn, 1962). In addition to petal color, plant types proved to differ in various characteristics, such as seed size, time to seedling emergence, time to anthesis, disease severity, biomass, and fecundity (see Results). Plant type as a source of variation was not considered in the original CO 2 -enrichment experiment, although both types were present in all model ecosystems. Seeds produced by mother plants at the end of the original experiment were collected in bulk from each model ecosystem, irrespective of plant type (Grünzweig and Körner, 2001b). Therefore, identity of seeds regarding plant type was unknown in the current study, and assignment of plant type to maternal CO 2 concentration was random. In all cases, each plant-type covered at least one-third of all plants per experimental year and maternal CO 2 treatment ( Table 1).  in high-CO 2 progenies (maternal CO 2 concentrations of 440 and 600 ppm) than in low-CO 2 progenies (maternal CO 2 concentration of 280 ppm) at peak season in 2005 (Tables 2 and 3). High-CO 2 offspring also produced 15% less fruits than low-CO 2 offspring. In 2006, when plants were growing in smaller compartments than in 2005 (see Materials and Methods), CO 2 progenies did not differ in shoot biomass. Maternal CO 2 did not significantly affect phenology (emergence, anthesis) and reproductive output (fruit and seed production) in 2006 and 2007 (Table 3).
Detailed measurements of disease progress and severity were conducted in 2006 and 2007, a heavy-infection year with disease incidence of 100% and a low-infection year with disease incidence of 81%, respectively. Twelve days after disease onset in 2006 (59 days after sowing, DAS), disease severity was 40% higher in high-CO 2 compared with low-CO 2 progenies (Figure 2, insert; Table 4). Disease severity increased with time, but differences between maternal CO 2 treatments were maintained 69 DAS. The AUDPC over 65 days of disease progress was 21% higher in high-CO 2 than in low-CO 2 progenies (Figure 3; Table 4). In 2007, disease severity was low and was not affected by maternal CO 2 . Preliminary onetime recording of disease severity in the heavy-infection year 2005 showed no impact of maternal CO 2 (Table 4). However, when a set of individuals of unknown plant type (no record of flower color) was added to the Case 1 analysis, disease severity was 24% higher in high-CO 2 than in low-CO 2 progenies.

case 2 (dIstInguIshIng between plant types)
Case 2 presents results of the current situation where diverging plant types are distinguished within the population. Maternal CO 2 as analyzed under the assumptions of Case 2 affected fruit production, and interacted with plant types on seed size and disease severity, but had no effect on offspring biomass and phenology. Fruit production of high-CO 2 offspring was lower (marginally significant) than fruit production of low-CO 2 offspring across both plant types in 2005 (Table 3). Maternal CO 2 had no statistically significant impact on plant performance in 2006 and 2007. Seeds of the pinkflowering plant type (Type P) that matured under high-CO 2 (440 and 600 ppm) and all seeds of the white-flowering type (Type W) were smaller by 20% on average than Type P seeds produced under low CO 2 (280 ppm; Tables 2 and 3).
Disease severity was 73% higher in high-CO 2 progenies than in low-CO 2 progenies of plant type P 59 DAS in 2006 (marginally significant interaction; Figure 2; Table 4). Considering the entire period of infection by the pathogen (AUDPC), high-CO 2 progenies of Type P were 30% more diseased than low-CO 2 progenies of the same plant type (Figure 3; Table 4). Maternal CO 2 concentration had no impact on disease progress and severity in offspring of Type W, and did not affect disease onset in any plant type. In 2007, maternal CO 2 interacted with plant type for disease severity 63 DAS (marginally significant).
Plant types differed in phenology, biomass, reproductive output, and disease severity. Seedlings of Type P emerged 1 day earlier on average in 2006 and 1.5 days earlier in 2007 compared with seedlings of Type W (Tables 2 and 3). In addition, Type P plants flowered 5 days earlier than plants of Type W in 2006, but not in 2007. Type P plants had 33 and 56% larger shoot biomass than Type W plants in 2005 and 2006, respectively (Tables 2 and 3). Seed size ANOVAs, and were not presented here (the covariate was statistically non-significant in almost all analyses). Disease incidence (proportion of diseased plants) was analyzed by a Chi-square test.
Sample size (number of individuals per model ecosystem) varied greatly according to seed availability. For Case 1 (plant type not included as a factor), sample size ranged from 27 to 36 for determination of seed size in mother plants, from 3 to 12 for variables  Case 1 simulates the common scenario where no diverging plant types are known. When plant types were disregarded maternal CO 2 had a significant impact on offspring biomass and fruit production and on disease severity. Shoot biomass was lower by 17% on average Inserts show disease-progress affected by maternal CO 2 only (Case 1). Disease severity was determined for each leaf and was subsequently added up for all leaves of a plant and expressed on a scale ranging between 0 and 1 (see Materials and Methods). Mean ± 1 SE, n = 3 model ecosystems. Non-identical letters indicate statistically significant differences among combinations of maternal CO 2 treatments and plant type (main panels) or among maternal CO 2 concentrations (inserts) as analyzed by the Tukey-Kramer HSD test (P ≤ 0.05 within mixed models, see Table 4). Sixty-three days after sowing in 2007, disease severity was significantly higher in 440-ppm progenies of plant-type W compared with all Type P progenies (Tukey-Kramer HSD test, P ≤ 0.05).  (40) 644 (39) 664 (62) 585 (87) 484 (26) 484 (11)   2006 338 (17) 318 (46) 311 (22) 384 (6) 403 (41) 391 (28) 241 (32) 269 (25) 243 (   CO 2 can impair reproductive output and susceptibility to diseases in offspring. The effect of elevated CO 2 still needs to be tested when both mother plants and offspring are grown at elevated CO 2 . However, the consequences of maternal CO 2 for progeny may be independent of the CO 2 concentration during progeny development, as shown for a majority of temperate legume and grass species (Fordham et al., 1997;Steinger et al., 2000;Edwards et al., 2001;Lau et al., 2008).

Maternal effects and theIr ModIfIcatIon by plant types and growIng condItIons
Considering Case 1, elevated CO 2 imposed on mother plants reduced offspring shoot biomass and reproductive output (number of fruits) compared with low CO 2 in 2005. In addition, severity of the fungal disease powdery mildew was higher in high-CO 2 than in low-CO 2 offspring in the heavy-infection year 2006, and the same results was obtained in 2005, if all individuals were considered. However, maternal effects of elevated CO 2 in Case 1 were often modified when plant types were distinguished in Case 2. The effect of maternal CO 2 on offspring biomass in 2005 was not significant anymore when the population was divided into two plant types. The seemingly maternal effect in the normal case when no functionally diverging plant types are obvious (Case 1) turned out to be a plant-type effect in Case 2. Changes in statistically significant main effects from maternal CO 2 in Case 1 to plant type in Case 2 could be caused by a different distribution of the two plant types among maternal CO 2 treatments. However, this was not the case in 2005 where plant types were distributed at almost identical proportions among maternal CO 2 treatments ( Table 1). In contrast to offspring biomass, lower reproductive output in high-CO 2 compared with low-CO 2 offspring in 2005 was significant in both Case 1 and Case 2, with no obvious plant-type effect in the latter case. Fruit production is of particular importance for dispersal of O. crista-galli, since the indehiscent, spiny fruit is the dispersal unit of this species. Less fruits in high-CO 2 progeny might result in less dispersal and lower colonization potential compared with low-CO 2 progeny. A maternal CO 2 effect on disease severity in Case 2 was recorded only in Type P plants. High maternal CO 2 increased disease severity above the level obtained for low maternal CO 2 in the more resistant plant-type P, but not in the more susceptible plant-type W. Therefore, high-CO 2 reduces the relative advantage of Type P over Type W and might impair its fitness through maternal effects.
Additionally, Case 2 revealed CO 2 effects on seed size that were not observed in Case 1. The larger size of seeds on Type P plants from low CO 2 could have contributed to some of the growth and disease-resistance effects observed on offspring of this combination of CO 2 and plant type.

Interannual varIatIon In Maternal effects
Maternal effects of elevated CO 2 on plant performance were recorded in 2005, but not in the following 2 years. Since seeds were randomly selected from the same stock for each experimental year, it can be assumed that interannual variation in maternal effects were related to environmental factors, not to a biased seed source. On the one hand, interannual variation in climatic conditions might influence maternal effects (Gendloff et al., 1986). On the other hand, growing conditions, such as larger growth average than seeds of Type W. Root biomass and root/shoot ratio were not significantly affected by maternal CO 2 and plant type (data not shown).
Type P plants were less affected by the disease than Type W plants in all three experimental years (Figures 2 and 3

dIscussIon
This study on the Mediterranean legume O. crista-galli showed that atmospheric CO 2 -enrichment potentially exerts maternal effects on the next generation's performance and, most notably, also on its susceptibility to natural infection by a fungal pathogen. However, maternal effects on susceptibility to diseases in offspring might differ among plant types within a population. Maternal effects of global change on disease severity in plants have not been shown so far, but they could affect the species' performance and fitness under future conditions. Environmental maternal effects have evolutionary consequences when offspring fitness is influenced by the maternal environment (Rossiter, 1996;Galloway and Etterson, 2007;Wolf and Wade, 2009). This study presents some evidence that rising atmospheric Mean ± 1 SE, n = 3 model ecosystems. Non-identical letters indicate statistically significant differences among maternal CO 2 concentrations (left panels) or combinations of maternal CO 2 treatments and plant type (right panels) as analyzed by the Tukey-Kramer HSD test (P ≤ 0.05 within mixed models, see Table 4).
compartments which resulted in larger plants in 2005 than in 2006 and 2007 might also influence maternal effects. Regarding disease severity, degree of infection seemed to affect interactions between maternal CO 2 and plant type. High maternal CO 2 increased disease severity in the more resistant plant type in a heavy-infection year, but had not in a low-infection year. The year 2005 was also characterized by heavy infection by the disease, but measurements of disease severity in that year were rather preliminary, and potential differences in severity among maternal CO 2 treatments went undetected.
The study investigated natural infection of seedlings which, on the one hand, is of relevance for pathogenesis under real field conditions. On the other hand, homogenous infection of plants had to be assumed to allow for proper analyses of findings. Studies with controlled infection could provide further information on maternal effects on susceptibility to fungal and other diseases.

potentIal MechanIsMs of Maternal co 2 effects
Maternal effects are a mechanism for phenotypic adaptation to environmental variation (Donohue and Schmitt, 1998;Mousseau and Fox, 1998). For example, life history of progenies in a forest understory was largely determined by the environment encountered by mother plants, and maternal effects in this context represented adaptive plasticity that was transmitted to offspring (Galloway and Etterson, 2007). Several potential mechanisms for environmental maternal effects could be discussed regarding the results obtained in this study. Mother plants of Type P that developed at low CO 2 could have transmitted traits related to growth and reproduction to the next generation. However, those plant could not transmit traits for disease susceptibility, since the disease was not evident in the original CO 2 experiment. Type P seeds that matured on mother plants at high-CO 2 were smaller than seeds that matured at low CO 2 , which could result from a trade-off between total seed production and the investment in each seed at high vs. low CO 2 . The smaller high-CO 2 seeds can give rise to reduced seedling vigor on the one hand, and less investment in defense mechanisms against infection by pathogens on the other hand (Herms and Mattson, 1992). Smaller nitrogen stores, as obvious in high-CO 2 O. crista-galli seeds across plant types (Grünzweig and Dumbur, in press), might provide less nitrogen-based defense compounds in seeds and subsequently in young offspring (Glen et al., 1990;Agrawal, 2002). Other mechanisms might involve inherited epigenetic changes affecting interactions between offspring and the abiotic or biotic environment, including pathogens (Bossdorf et al., 2008). Such epigenetic changes might include yet unspecified mechanisms related to plant defense in offspring (Molinier et al., 2006).
No direct transfer of powdery-mildew propagules from the mother plant to seeds is to be expected, since the disease was not apparent on mother plants and the pathogen is not seed-borne. This also excludes the possibility of a selection effect for more disease resistance during development of mother plants. Although atmospheric CO 2 enrichment can impose selective pressure on annual C 3 plants (Ward and Kelly, 2004), such an effect was not obvious on growth and reproductive traits. Offspring of the 440-ppm treatment were not superior to offspring of the other treatments, despite the fact that this CO 2 concentration was closest to ambient CO 2 of ∼380 ppm.

plant types
In many cases, diverging plant types are not obvious when heterogeneous natural populations are randomly sampled. In this study, plant types within this population of O. crista-galli could be distinguished by differences in flower color (Figure 1). It turned out that the two plant types diverged in all measured variables of plant performance and disease severity, at least in some years. Both plant types were part of the same population, and, thus, represent population-level functional diversity. Without a clear phenotypic marker, many studies are necessarily affected by unknown underlying interference by the variability among plant types, unless detailed genetic analyses are carried out.
The modifying effect of dividing the population of O. cristagalli into plant types on maternal CO 2 effects derives mainly from the fact that the two plant types diverge in most variables of plant performance and susceptibility to the disease. Mostly, Type P plants were more vigorous than Type W plants, and disease severity in Type P plants was consistently less pronounced than in Type W plants.
This study also provides a potential explanation for the coexistence of the more disease-resistant Type P and the more susceptible Type W at the same site. In both 2006 and 2007, emergence and initial plant growth was accelerated in Type P compared with Type W seedlings. However, a greater number of leaves produced by Type W plants during mid-season of 2007 indicated that vigor of the latter plant type was at least as high as vigor of Type P plants in nearly disease-free years. This might have evolutionary consequences. In a year with a high disease rate (2006), Type P produced more fruits than Type W, but in a year with low disease rates (2007), Type W produced more fruits (marginally significant). The opposed relative success in fruit production between plant types in contrasting years should result in alternate predominance in dispersal success. This fact may also lead to higher mean fitness of the overall population in that location. However, rising atmospheric CO 2 might reduce the relative advantage of plant-type P in heavy-infection years as obvious in 2005, which could negatively affect dispersal success in the long term.
conclusIon Elevated CO 2 applied to mother plants potentially affects offspring performance and susceptibility to a fungal pathogen. However, maternal effects differed among years, and, notably, were modified when the plant population was divided into functionally diverging plant types. Environmental maternal effects are a considerable source of variation in offspring trait-expression, and their consequences for population fitness and for interactions with pathogens could be of high significance for global-change impacts on plant communities. acknowledgMents I am grateful to Jaacov Katan, Robert Kenneth, and Maggie Levy for help with the phytopathological analyses and assistance with the identification of the pathogen, to Shai Morin for help with statistical analyses, and to Baruch Rubin for providing space in the net-house facility. Many thanks to Eyal Fridman, Jaacov Katan, Jaime Kigel, Maggie Levy, and Shmuel Wolf for helpful discussions and comments to an earlier version of the manuscript. The technical assistance of Hadas Sibony, Rita Dumbur, Ahuva Nir, Gil Yogev, Ziv Kleinman, Aia Oz, and Shirly Dafni is greatly acknowledged.