Distribution of bulbil- and seed-producing plants of Poa alpina (Poaceae) and their growth and reproduction in common gardens suggest adaptation to different elevations†
The authors thank S. Ellenberger, U. Halter, R. Steiger, H. Steiner, and J. Steiner for help in the field, the Gemeinde Haldenstein and the CCES BioChange project of the ETH Zürich for enabling the use of the field sites at Mt. Calanda, T. van den Bergh for climate data from the Furka region, and two anonymous reviewers whose comments helped to improve the manuscript. This study has been supported financially by the Swiss National Science Foundation, project no. 3100AO-116785 to J. Stöcklin.
Abstract
• Premise of the study: The European Alps harbor a spatially heterogeneous environment. Plants can be adapted genetically to this heterogeneity but may also respond to it by phenotypic plasticity. We expected the important fodder grass Poa alpina to be adapted to elevation either genetically or plastically.
• Methods: We investigated in three elevational common gardens whether growth and reproductive allocation of plants reproducing either by seeds or bulbils suggest adaptation to their elevation of origin and to what extent they can respond plastically to different elevations. Additionally, we analyzed genetic diversity using microsatellites and tested whether seeds are of sexual origin.
• Key results: In the field, bulbil-producing plants occurred more often at higher elevations, whereas seed-producing plants occurred more often at lower elevations, but bulbil-producing plants were generally less vigorous in the common gardens. The response of plants to elevational transplantation was highly plastic, and vigor was always best at the highest location. The small genetic differences were not clinally related to elevation of origin, underlining the importance of phenotypic plasticity. Reproductive allocation was, however, independent of elevational treatments. Seed-producing plants had higher genetic diversity than the bulbil-producing plants even though we found that seed-producing plants were facultative apomicts mostly reproducing asexually.
• Conclusions: Bulbil-producing P. alpina, showing a fitness cost at lower elevations compared with seed-producing plants, seem better adapted to higher elevations. By means of its two reproductive modes and the capacity to adjust plastically, P. alpina is able to occupy a broad ecological niche across a large elevational range.
Alpine ecosystems are characterized by steep environmental gradients over short distances (31). The compression of climatic life zones over short geographic distances results in a strong spatial heterogeneity and natural fragmentation of the alpine landscape. The variety of microhabitats leads to a high biodiversity of alpine plants, but also to spatial isolation among populations and restricted gene flow (59, 22). At higher elevations, plants also need to cope with lower atmospheric pressure, lower temperatures, and hence a shorter vegetation period and restricted time for reproduction, with more snow and weather-related extreme events (9; 10). To survive in such heterogeneous and harsh conditions, plants can either adapt to the local conditions genetically or respond to the environmental variability by phenotypic plasticity (60; 8).
Local adaptation leads to an increased fitness of a population in its particular habitat (28), while phenotypic plasticity enables a single genotype to express different phenotypes depending on the environmental conditions (11; 53; 44). Phenotypic plasticity allows organisms to react rapidly to a changing environment, which is especially important for plants since they cannot move notably and have to deal with changing ambient conditions (53; 60). Phenotypic plasticity is also expected to occur more frequently at higher elevations as it enables fast adaptation to a temporally highly variable environment, ensuring the survival of alpine plants (59; 45). However, plasticity is not always advantageous, but can also be nonadaptive, with neutral or negative effects to fitness (23). Additionally, there are costs and limitations in the production and maintenance of plasticity (64; 4), so genetically fixed traits may be favored under certain, usually temporally stable, circumstances (5). For predicting how plants will react to future climate change, it is important to know whether and how much a species can respond plastically to a changing environment. If plastic adjustments are limited, plants would either have to rapidly adapt to the new conditions genetically or migrate to track the conditions they are currently adapted to (29). If climate change is very quick, the fitness of locally adapted plants could decline rapidly and local extinctions become likely, especially for isolated, inbred and apomictic species (18; 61; 29).
In the more severe conditions of the Alps with shorter growing seasons, asexual reproduction becomes more important than sexual reproduction and an increase of clonal plant species at higher elevations has therefore been observed (9; 10; 30; 59; 57). Apomixis, the asexual reproduction through seeds (12; 3; 38; 40), offers the possibility of reproduction by seeds in plants that are either unable to produce viable products via meiosis or live in territories in which various factors exclude the possibility of sexual reproduction (e.g., harsh climatic conditions at high elevations; 12).
Poa alpina L., one of the most common and widespread fodder grasses of the European Alps (13), occurs with two reproductive modes. Plants in this species are either seminiferous or pseudoviviparous (37). While it is known that bulbils are formed vegetatively by transforming flower meristem into a vegetative shoot, it is not clear to what extent seeds are formed sexually. Apomictic seed formation seems to be a possibility, especially if populations evolved their reproduction from a bulbil- to a seed-producing mode (35, 36; 55). Interestingly, the occurrence of seed-producing plants decreases and that of bulbil-producing plants increases with elevation (33; 20), which is in line with the hypothesis of an adaptive advantage of clonal reproduction in the harsher conditions at higher elevations (47).
In the present study, we had three primary goals. First, we monitored the occurrence of seed-producing and bulbil-producing P. alpina in the field along an elevational gradient at two locations, expecting a decline in seed-producing and an increase in bulbil-producing plants with increasing elevation. Second, in a molecular study, we asked whether populations of either seed- or bulbil-producing plants differ in their genetic diversity and whether seeds of seminiferous P. alpina are formed sexually or partially apomictically according to an old but so far never rigorously tested hypothesis of 35. We hypothesized that, even with partial apomixis in seed-producing plants, their genetic diversity would be higher compared to genetic diversity in bulbil-producing plants. Third, by transplanting P. alpina to common gardens at three different elevations, we tested whether populations from different elevations differed in their growth and reproduction genetically (56) or responded plastically to growing conditions at different elevations (60).
MATERIALS AND METHODS
Study species
Poa alpina L. (Poaceae) is a common plant species in arctic and alpine areas (48). In the European Alps, it occurs predominantly from 1400 to 2500 m a.s.l. (occasionally from 200 to 3600 m a.s.l.), mostly in nutrient-rich meadows and pastures, but it also grows on pioneer sites like scree slopes and snowbeds (54; 13; 63). In the upper montane and alpine zone of the European Alps, it is one of the most important fodder grasses because of a high content of proteins and fatty acids (54; 58; 13). Poa alpina is polyploid with highly variable, often aneuploid chromosome numbers, ranging from 2n = 22 to >60 (35, 37; 17; 33).
Poa alpina occurs with two reproductive modes: plants are either seminiferous, i.e., seed-producing, or pseudoviviparous, i.e., bulbil-producing (37; 42). According to 35, seeds of seminiferous plants are produced either sexually or apomictically. The pseudoviviparous plants reproduce vegetatively by forming bulbils in the panicles that grow into little plantlets on the parent plant and can root and establish quickly (37; 43). The reproductive mode is largely genetically determined, while environmental effects are of minor importance for the type of reproduction (51). 20 found that the frequency of bulbil-producing plants is increasing with elevation.
Occurrence of seed- and bulbil-producing plants
Study site
The occurrence of seed- and bulbil-producing plants in the field was studied along the pass road on the Furka, western Swiss Alps, and along the hiking trail Filisur-Curtins-Muchetta on Mount Muchetta, eastern Swiss Alps, both along an elevational gradient from 1650 to 2600 m a.s.l.
Experimental design
To determine the ratio between seed-producing and bulbil-producing plants, the reproductive mode of P. alpina individuals was checked stepwise every 50 altitudinal meters along the elevational gradient. Plants were checked for their reproductive mode in plots of 1 m2, until at least 100 adult plants with erect culms were counted (between three and fourteen 1 m2-plots were needed).
Data analysis
For each elevational step, the percentage of seed-producing and bulbil-producing P. alpina was calculated. If there were less than 100 P. alpina plants found at a particular elevation, which was the case at 1650 m a.s.l. with 84 plants, 1850 m with 74 plants, and at 2300 m with 3 plants on the Furka, and at 2450 m with 46 plants on Mt. Muchetta, the percentage was calculated with the available plants. For the data analysis in a regression model using the program R ver. 2.10.1 (49), data sets from the Furka and the Muchetta were merged, and a logistic regression based on a binomial distribution including the factors elevation and study site was performed. Here, the data from 2300 m elevation on the Furka, where only three plants were found, was excluded from the analysis.
Molecular analyses
Sample collection
For the genetic diversity analysis, leaf samples were taken from 60 plants in total, 30 seed-producing and 30 bulbil-producing ones, whose seeds and bulbils originated from the Furka Pass. We used 10 samples of seed-producing plants each of three different elevations at 1630, 1850, and 2100 m a.s.l., and 10 samples of bulbil-producing plants each of three different elevations at 2100, 2350, and 2600 m a.s.l. The samples, at least 2 cm long, healthy parts of leaves, were dried with silica gel immediately after being separated from the plants. Additionally, we sampled two siblings of four sampled plants (two seed-producing ones from 1630 and 2100 m a.s.l. and two bulbil-producing ones from 2100 and 2600 m a.s.l.) for a preliminary test of genetic similarity among offspring originating from the same seed and bulbil family.
Because we found that not only offspring (siblings) of bulbil-producing plants were genetically identical (clones), but the majority of offspring (siblings) of seed-producing plants was also genetically identical, we decided to test for the presence of apomixis in seed-producing plants. For this analysis of offspring from seed-producing plants, we used plant material from the Furka. Eight seed-producing individuals with already ripe seeds were collected in paper bags, one each at 1850, 1950, 2000, 2050, 2150, 2200, 2250, and 2400 m a.s.l. Leaf material was dried with silica gel the day after sampling. From the seeds, nine offspring per mother plant were raised in potting soil under controlled conditions in a greenhouse in Basel. The plants were regularly watered and were grown until the leaves were big enough for DNA extraction, for which leaves were sampled and dried with silica gel.
Genetic analyses
Extraction of genomic DNA of silica-dried leaf material and measurement of DNA concentration was performed as described by 46. Amplification of microsatellite alleles of five loci (CA1D4, GAC1, GA1C3, CA1F4, CAB12) was done as in the pilot study by 34. Some modifications were applied, i.e., PCR was performed with self-dissolving illustra puReTaq Ready-To-Go PCR Beads (GE Healthcare, Buckinghamshire, UK). The bead system allowed highly reproducible fingerprints (discussed later). Three substances were added to the PCR beads, i.e., 10–50 ng of genomic DNA, 25 pmol of unlabeled forward and reverse primer and ddH20. Besides these substances, the PCR mixture (25 µL) finally contained 10 mmol/L Tris-HCl buffer, 200 µmol/L dNTPs, 1.5 mmol/L MgCl2, 50 mmol/L KCl, and ca. 2.5 U polymerase (all latter substances provided by the manufacturer). PCRs were always run in the same machine (Eppendorf Mastercycler gradient, Eppendorf, Hamburg, Germany), with the same cycling conditions as in 34. Annealing temperature also followed the latter study, except for locus CA1F4 in which primer annealing was set to 58°C. Horizontal Spreadex gel electrophoresis was used (see 2; 1), following ethidium bromide staining to visualize bands. Clear and distinct microsatellite alleles (i.e., bands), or absence of such bands, were scored manually. The locus-specific size range of microsatellite amplicons was similar to the pilot study by 34. No PCR dropouts occurred; i.e., all individuals yielded readable banding profiles.
Data analyses
For the genetic diversity analysis, treatment of polyploid microsatellite data of Poa alpina has been described by 51, p. 1252). Following that study, banding patterns of loci and all 60 samples were one/zero coded (alleles present or absent) and uploaded in the program GenAlEx ver. 6.2 (41). GenAlEx was used for finding identical multilocus genotypes (i.e., clones) and to perform an analysis of molecular variance (AMOVA) of seed- and bubil-producing plants. We used R ver. 2.10.1 (49) to measure the genetic diversity as number of allelic bands per individual and to perform a Welch's two-sample t test to see whether the genetic diversity between seed-producing and bulbil-producing plants differed.
For the test for apomixis, the banding pattern of offspring from seed-producing plants was visually compared to that of their mothers. If the offspring was identical to the mother, it was assumed to have been produced by apomixis. If a band was missing in the offspring compared to the mother, it was not possible to say whether it had been produced by self-fertilization or by outcrossing, but there must have been a viable meiosis. If an additional band appeared in the offspring compared to the mother, we could be sure that sexual reproduction had occurred by outcrossing; i.e., nonmaternal pollen was used for seed production.
Common garden experiment
Location of sample sites and common gardens
Plants for the common garden experiment originated from five sites along the elevational gradient on the Furka, western Swiss Alps, from Realp along the pass road up to Blauberg. The five sampling elevations of populations at 1630, 1850, 2100, 2350 and 2600 m a.s.l. will henceforth be called “elevations of origin”. Along this elevational gradient, average temperature during summer 2010 (21 June–22 September) was 11.9°C at 1490 m a.s.l., 10.3°C at 1960 m a.s.l., and 6.8°C at 2440 m a.s.l., and average precipitation during summer was 3.4 mm/day at 600 m a.s.l., 3.7 mm/day at 1200 m a.s.l., and 4.1 mm/day at 1800 m a.s.l.
The common gardens were located on the southeast-facing slope of Mt. Calanda, eastern Swiss Alps. These treatment sites, located at 600, 1235, and 1850 m a.s.l., will henceforth be called “elevations of experiment” and will sometimes be referred to as lowest, medium, and highest elevation. Average temperature during summer 2010 was 19.3°C at 600 m a.s.l., 16.9°C at 1200 m a.s.l., and 15.3°C at 1800 m a.s.l., and average precipitation during summer was 4.1 mm/day at 600 m a.s.l., 2.2 mm/day at 1200 m a.s.l., and 2.5 mm/day at 1800 m a.s.l.
Experimental design
Either seeds or bulbils of 60 individuals (genotypes) of P. alpina were sampled in summer 2009 along the elevational gradient on the Furka. At 1630, 1850, and 2100 m a.s.l., seeds of 10 seminiferous individuals were collected; at 2100, 2350, and 2600 m a.s.l., bulbils of 10 pseudoviviparous individuals were collected. At 2100 m a.s.l., seed-producing and bulbil-producing plants occurred in sufficient number to be sampled both (Fig. 1). Plants were collected randomly with interdistances of 4–5 m to minimize the risk of sampling the same genotype more than once.
Experimental design of the common garden experiments. Elevations of origin are on the left; elevations of experiment are on the right (each having its own elevational axis). Dark gray represents bulbil-producing P. alpina plants, light gray represents seed-producing ones. Elevations are given in meters a.s.l.
Seeds and bulbils were then used to obtain at least nine offspring of every sampled “mother”-plant (genotype) in a greenhouse in Basel (276 m a.s.l.). F1-generation plants were grown from 5 September 2009 onward. On 31 October 2009, the resulting 540 young plantlets were brought to the common gardens on Mt. Calanda.
On each elevation of experiment (600, 1235, and 1850 m a.s.l.), the young P. alpina plants were placed on the ground in sowing pans to overwinter, protected from freezing by feathery wood chips and the site fenced to protect against herbivores. In the following year, on 15 June 2010, the plants were planted into the local soil in a stratified random order, with each of three replicates (i.e., from the same clone or seed family) in one of three 1 × 2 m plant beds. In total there were 540 plants, 180 in each common garden, 60 in each plant bed. The beds were weeded monthly from June to September in 2010 and from May to August in 2011. Because the plants at the 600 m elevational site acquired a rust fungal infection, they were treated three times in summer 2010 and once in July 2011 with 10 g of the fungicide Dithane NeoTec (Dow Agro Science, Indianapolis, Indiana, USA), diluted in 10 L of water. To limit plant size variability during the second year of the experiment, we removed all biomass of each plant down to 3 cm above soil level after the first growing season, on 2 October 2010.
Measurements
The reproductive biomass of each plant was harvested during the year 2011. It was harvested at monthly intervals from May to August. At each harvest time, fully grown reproductive shoots were cut off above the uppermost leaf, were collected in paper bags, and the number of shoots was counted. For each plant, one paper bag was used to collect all harvested reproductive biomass. The reproductive mode was noted as bulbil-producing or seed-producing at the beginning of the harvest season when the plant first flowered, and the reproductive mode was checked afterward every time a new shoot was collected. During the experiment, only one plant changed its reproductive mode: a seed-producing plant at the 1200 m elevational site of experiment that originated from 1630 m a.s.l. changed to the bulbil-producing reproductive mode. After each harvest, the reproductive biomass was brought to the laboratory within 24 h and was dried at 60°C for at least 48 h. The total reproductive biomass was again dried at 60°C for 48 h after the end of the growing season and was then weighed to a precision of 0.1 mg.
The vegetative biomass of each plant was harvested at the end of the season (6 August 2011). The aboveground biomass was clipped 3 cm above ground after the reproductive shoots were harvested for the last time. The vegetative biomass was then dried at 60°C for at least 48 h and weighed to a precision of 0.01 g.
The sum of the reproductive and the vegetative biomass was calculated to get the total aboveground biomass that the plants produced in 2011. The reproductive allocation (percentage reproductive biomass) was then calculated as [(reproductive biomass / total aboveground biomass) × 100].
Data analysis
Due to mortality throughout the experiment, 81 plants had to be excluded from the analysis. Another seven plants had to be excluded because they could not clearly be assigned to their identity after trampling of animals. Finally, 452 plants were included in the analysis of the total aboveground biomass. For the analysis of the reproductive allocation, only 421 plants were included because 31 of the 452 surviving plants did not flower.
Mixed-effect models were used to test whether the measured traits of P. alpina were influenced by the reproductive mode, the elevation of origin of the plants, and the elevation of experiment (i.e., the common garden). Seed-producing and bulbil-producing plants were analyzed separately after testing whether growth and reproduction of the two reproductive modes differed.
If the model residuals were not normally distributed, normality was achieved with a box-cox transformation using the R function boxcox of the package car (21). The total aboveground biomass was transformed with an exponent of 0.303, the vegetative biomass with an exponent of 0.404, and both were then analyzed with linear mixed-effect models. For the analysis of the reproductive allocation, generalized mixed-effect models with a binomial error distribution were used. For the models, the lmer function from the R package lme4 (7) was used. To test for significance of model factors, we used comparisons between models with and without factors using χ2 tests, since the lmer function does not provide p values (15).
First, we tested whether seed-producing and bulbil-producing plants from the same elevation of origin differed in growth and reproduction. In this first model, only plants from the 2100 m site of origin were used, because only at this elevation, plants with both reproductive modes were sufficiently present and sampled. The following factors were tested: elevation of experiment (fixed factor), reproductive mode (fixed), and genotype nested in reproductive mode (random). Next, the data were split according to the plants’ reproductive modes and tested separately for seed-producing and bulbil-producing plants with a second model. In this second model, the following factors were tested: elevation of experiment (fixed), elevation of origin (fixed), genotype nested in origin (random), elevation of experiment × elevation of origin interaction (fixed), and elevation of experiment × genotype nested in origin interaction (random).
Variance components were calculated with a similar model but treating all factors as random factors. Thereby, we estimated the percentage of environmental and genetic effects that influenced the measured traits. Elevation of origin and genotype nested in origin contributed to the genetic effects, whereas elevation of experiment, elevation of experiment × elevation of origin interaction and elevation of experiment × genotype nested in origin interaction contributed to the environmental ones.
RESULTS
Reproductive mode in Poa alpina along an elevational gradient
We observed a relatively sharp shift from seminiferous to pseudoviviparous reproduction with increasing elevation, which was confirmed by the logistic regression model (z = – 2.396, df = 37, P = 0.017). The sharp shift occurred at ca. 2200 m a.s.l. on the Furka in the western Swiss Alps (Fig. 2A) and at ca. 2350 m a.s.l. on Mt. Muchetta in the eastern Swiss Alps (Fig. 2B).
Percentage of seed-producing and bulbil-producing plants depending on elevation (m a.s.l.) on (A) the Furka, and (B) Mt. Muchetta. Black portions of bars represent proportion of seed-producing, gray portions of bars the proportion of bulbil-producing Poa alpina plants.
Genetic diversity of bulbil- and seed-producing Poa alpina
Among the 60 plants that were analyzed, we detected 76 bands among the five microsatellite loci, between 6 and 24 per locus. In total, we detected 41 multilocus microsatellite genotypes. There was no significant difference in the number of multilocus-microsatellite genotypes per population, but in contrast to all the other populations, the bulbil-producing plants from the 2100 m elevation had a particularly low number of genotypes (3 compared to a mean of 6.8 genotypes per population).
The genetic diversity, measured as the number of allelic bands per individual plant, was higher for seed-producing plants (20.2 ± 0.5) than for bulbil-producing plants (17.5 ± 0.4; t = 4.11, df = 54.34, P < 0.001). The genetic diversity among the populations from different elevations was also significantly different (F5, 54 = 6.67, P < 0.001; data not shown), but these differences could not be attributed to the elevational gradient.
The factors tested in the AMOVA were significant. In seed-producing plants, 83% of the variation in microsatellite bands resided within populations and only 17% among populations, while in bulbil-producing plants, 61% of the variation in microsatellite bands resided within populations and 39% among populations.
Sexual offspring in seed-producing apomictic Poa alpina
Of the 72 offspring from eight seed-producing plants (nine offspring per mother plant), we found six that differed genetically from their mother (8.3%). Five of these six had one or two additional bands compared to the mother genotype in at least two loci and must therefore have been produced via sexual reproduction by means of outcrossing (6.9%, Fig. 3). One offspring had a missing band in two loci, and we cannot be sure whether it has been produced by outcrossing or selfing, but in any case the offspring has originated from meiosis. Four of eight of the analyzed mothers (50%) were able to perform viable meiosis.
Spreadex gel showing one mother and nine of her offspring genotyped at a polyploid microsatellite locus (GA1C3). The arrows point to two new alleles in offspring 5, which indicate sexual reproduction by outcrossing.
Growth of Poa alpina in the common garden
Mortality in the common garden was 15% in total (81 of 540 plants died during the experiment). Mortality was highest at 600 m (10.4%), lowest at 1235 m (0.2%), and intermediate at 1850 m (4.4%). During the experiment, 93.1% of the surviving 452 plants were reproducing.
Seed-producing plants originating from 2100 m a.s.l. were more vigorous across the three common gardens than bulbil-producing plants from the same elevation, as the seed-producing plants produced more total aboveground biomass than bulbil-producing plants (P = 0.04; Table 1; Fig. 4A).
| Total biomass | Vegetative biomass | Reproductive allocation | |||||||
|---|---|---|---|---|---|---|---|---|---|
| Traits | df | χ2 | P | df | χ2 | P | df | χ2 | P |
| Elevation of experiment | 2 | 77.44 | <0.001 | 2 | 79.96 | <0.001 | 2 | 1.51 | 0.469 |
| Reproductive mode | 1 | 4.13 | 0.042 | 1 | 8.68 | 0.003 | 1 | 1.39 | 0.238 |
| Genotype nested in reproductive mode | 1 | 1.03 | 0.311 | 1 | 1.17 | 0.280 | 1 | 0.00 | 1.000 |
| Elevation of experiment × reproductive mode | 2 | 1.78 | 0.410 | 2 | 3.29 | 0.193 | 2 | 0.16 | 0.925 |
Growth and reproduction of Poa alpina in the common garden. (A) Total aboveground biomass of individuals (g). Dark gray bars represent vegetative biomass; light gray bars represent reproductive biomass. (B) Reproductive allocation (%). L, M, and H stand for lowest (600 m a.s.l.), medium (1235 m a.s.l.) and highest (1850 m a.s.l.) elevation of experiment, respectively. Error bars show SE. Means of seed families and bulbil families, respectively, were calculated and subsequently used to compute mean and SE of each population (elevation of origin) for the three elevations of the experiment separately.
Seed- and bulbil-producing plants did not differ in their investment in reproductive biomass, suggesting that both have the same reproductive allocation (P = 0.469; Table 1; Fig. 4A).
The percentage of reproducing plants was similar in seed- and bulbil-producing plants. A total of 92.4% of the seed-producing and 94.0% of the bulbil-producing plants formed reproductive structures.
The total and vegetative biomass of seed- as well as bulbil-producing plants was strongly influenced by the elevation of experiment, indicating strong plasticity (P < 0.001 for seed-producing and bulbil-producing plants for total and vegetative biomass; Table 2A, 2B). Biomass was much larger at the highest elevation of experiment (1800 m) compared to the other elevations of experiment for plants of all elevations of origin, and this was the case for plants with either type of reproduction (Fig. 4A).
| Total biomass | Vegetative biomass | Reproductive allocation | |||||||
|---|---|---|---|---|---|---|---|---|---|
| Traits | df | χ2 | P | df | χ2 | P | df | χ2 | P |
| A) Seed-producing plants | |||||||||
| Elevation of experiment | 2 | 135.53 | <0.001 | 2 | 144.56 | <0.001 | 2 | 3.41 | 0.182 |
| Elevation of origin | 2 | 0.46 | 0.796 | 2 | 1.76 | 0.415 | 2 | 0.92 | 0.632 |
| Genotype nested in origin | 1 | 3.49 | 0.062 | 1 | 4.46 | 0.035 | 1 | 0.00 | 1.000 |
| Elevation of experiment × elevation of origin | 4 | 12.26 | 0.016 | 4 | 13.01 | 0.011 | 4 | 0.15 | 0.997 |
| Elevation of experiment × genotype nested in origin | 1 | 2.93 | 0.087 | 1 | 5.99 | 0.014 | 1 | 0.00 | 1.000 |
| B) Bulbil-producing plants | |||||||||
| Elevation of experiment | 2 | 124.20 | <0.001 | 2 | 136.11 | <0.001 | 2 | 1.58 | 0.454 |
| Elevation of origin | 2 | 0.90 | 0.637 | 2 | 0.47 | 0.792 | 2 | 2.10 | 0.350 |
| Genotype nested in origin | 1 | 0.00 | 0.946 | 1 | 2.05 | 0.152 | 1 | 0.00 | 1.000 |
| Elevation of experiment × elevation of origin | 4 | 18.24 | 0.001 | 4 | 18.63 | <0.001 | 4 | 0.12 | 0.998 |
The reproductive allocation was not influenced by the elevation of experiment (P = 0.18 for seed-producing plants, Table 2A; P = 0.45 for bulbil-producing plants, Table 2B; Fig. 4B).
Total biomass of both seed- and bulbil-producing plants was influenced more by environmental effects (variance components for elevation of common gardens, elevation of common gardens × elevation of origin interaction, and elevation of experiment × genotype nested in origin interaction) than by genetic effects (variance components for elevation of origin and genotype nested in origin). Environmental effects accounted for 94.6% (bulbil-producing) and 96.6% (seed-producing) of the measured variance components. Genetic effects on the other hand accounted only for 3.4% (seed-producing) and 5.4% (bulbil-producing) of the measured variance components.
Aside from the biomass difference between seed- and bulbil-producing Poa alpina, plants from a particular elevation of origin were not generally more vigorous than others, since there was no significant main effect of elevation of origin on total biomass (Fig. 4A, P = 0.80 for seed-producing plants, Table 2A; P = 0.64 for bulbil-producing plants, Table 2B) nor on vegetative biomass (P = 0.42 for seed-producing plants, Table 2A; P = 0.79 for bulbil-producing plants, Table 2B) or on reproductive allocation (P = 0.63 for seed-producing plants, Table 2A; P = 0.35 for bulbil-producing plants, Table 2B).
Nevertheless, the elevation of origin influenced total and vegetative growth, but not the reproductive allocation, of both seed- and bulbil-producing plants in the three common gardens (interaction of elevation of experiment × elevation of origin; for total biomass P = 0.001 for bulbil-producing plants, Table 2B; P = 0.02 for seed-producing plants, Table 2A; for vegetative biomass P < 0.001 for bulbil-producing plants, Table 2B; P = 0.01 for seed-producing plants, Table 2A). This interaction was probably due to the seed-producing plants from the lowest elevation of origin growing best at the lowest elevation of experiment and least at the highest elevation of experiment. In bulbil-producing plants, plants from the intermediate elevation of origin grew similarly well at the highest and lowest elevation of experiment, whereas bulbil-producing plants from other origins revealed strong differences among the three elevations of experiment (Fig. 4A).
For seed-producing plants, the genotypes (P = 0.035, Table 2A) as well as their interaction with elevation of experiment (P = 0.014, Table 2A) had a significant influence on the vegetative biomass.
DISCUSSION
We found that elevation strongly influenced the plants’ distribution and their growth and reproduction. In the field, bulbil-producing plants occurred at higher elevations than seed-producing plants. Seed-producing plants had a higher genetic diversity and were molecularly less differentiated than bulbil-producing plants. In the common gardens, seed-producing plants grew better than bulbil-producing plants, but the reproductive allocation was similar for both types. The elevation of origin did not influence the biomass of seed- or bulbil-producing plants, but we found a strong phenotypic adjustment due to environmental conditions at the different elevations of experiment indicating that plants of Poa alpina are highly plastic and not well adapted to elevations below 1850 m a.s.l.
Occurrence of seed- and bulbil-producing Poa alpina
At both locations, bulbil-producing plants were more common than seed-producing plants at higher elevations. This result is in line with the hypothesis of an adaptive advantage of vegetative reproduction of Geum reptans in the harsher conditions at higher elevations (47). The shorter growing period and the harsher conditions at higher elevations can endanger seed production and establishment in seed-producing plants (9; 10). Offspring of bulbil-producing plants can grow into mature plants faster because of their greater size and nutrient capital (26) and may therefore be more likely to survive at higher elevations with short growing seasons. In a common garden experiment, 20 also found a higher proportion of bulbil-producing plants among samples from higher elevations.
The reason for the sharp shift from seed- to bulbil-producing plants with elevation remains elusive. Possible reasons could be sharp boundaries in abiotic conditions, human interference or positive feedback in biological processes within a plant community (67). 66 found a shift of reproduction from bulbil- to seed-producing mode with increasing vegetation density in P. alpina along a primary succession gradient. Further studies, which correlate environmental and land-use factors with the occurrence of the two reproductive modes along the elevational gradient, could shed more light on the exact causes.
Genetic diversity of seed- and bulbil-producing Poa alpina
Seed-producing plants had a higher genetic diversity than bulbil-producing plants did. Vegetative reproduction often leads to a loss of genetic diversity through extinction of genotypes over time (56). Therefore it was expected that bulbil-producing plants, which reproduce strictly vegetatively, had a lower genetic diversity than seed-producing plants. Even though seed-producing plants turned out to reproduce predominantly through apomixis, bulbil-producing plants remained more genetically uniform. The observed rate of sexually produced offspring in seed-producing plants seems high enough to conserve the higher level of genetic diversity in these plants. The effect of facultative apomixis on plant species’ genetic diversity was explored in a model by 39, who found that the genetic diversity of facultative apomicts that are capable of some sexual reproduction was found to be comparable to their sexual counterparts. In general, a mixture of vegetative with sexual reproduction is sufficient to maintain genetic variability. In Titanotrichum oldhamii, which produces bulbils and flowers on the same plant, 65 also found a genetic diversity comparable to that of outcrossing plants. In Polygonum viviparum, 16 found that bulbil-producing plants that occasionally produced seeds sexually had a higher genetic diversity than plants that reproduced strictly asexually.
Additionally, we found in our analysis that bulbil-producing populations had a greater variation in microsatellite bands among populations than seed-producing populations did and were thus molecularly more differentiated, which suggests that gene flow among bulbil-producing populations is very low, as can be expected for populations with strictly asexually reproducing plants (25).
Apomixis in seed-producing Poa alpina
The molecular analysis of seed-producing P. alpina mothers and their offspring revealed that the majority of the analyzed plants produced their seeds apomictically, but some were partly outcrossing, as assumed by 35, 36) and 55. Four of eight analyzed mothers were able to perform meiosis, suggesting that the plants were facultative apomicts. In various taxa, populations of facultative apomicts have been found to consist of multiple genetically distinct clones (3), e.g., in Poa pratensis (32), Hypericum perforatum (6), Townsendia hookeri (62) and Ranunculus kuepferi (14; see also 50). Facultative apomixis could be an advantage because genotypes that are well adapted to current conditions could be exploited, while a certain level of genetic variation is still maintained for future adaptation (6).
Growth of seed- and bulbil-producing Poa alpina in the common garden
In all three common gardens, seed-producing plants were more vigorous than bulbil-producing plants. Considering that our common gardens were located at lower elevations than the natural elevational range of bulbil-producing P. alpina, this result was not surprising even though we initially expected to find a higher fitness of bulbil-producing plants compared to seed-producing plants in the highest common garden. Seed-producing plants seemed to have a generally higher fitness than bulbil-producing plants at the elevations of our common gardens. Due to the lack of a common garden at elevations that correlate more with the natural elevational range of bulbil-producing plants, we were unable to detect whether the opposite is true at particularly high elevations where bulbil-producing plants are predominant. But according to our observations in the field, where bulbil-producing plants were much more common at higher elevations, we hypothesize that at such sites bulbil-producing plants perform better than seed-producing plants, probably mainly as a consequence of advantages due to their reproductive mode. At higher elevations, offspring of bulbil-producing plants probably can establish easier and more quickly than offspring of seed-producing plants because grass plantlets generally have a greater size and nutrient capital than seeds (26; 59). The advantage of bulbil-producing plants at higher elevations is accompanied by a cost of fitness at lower elevations. Because of its two reproductive modes, which seem to have contrasting advantages at different elevations, P. alpina seems to be able to occupy a broad ecological niche across varying elevations.
The elevation of the experiment generated a highly plastic response in total aboveground and vegetative biomass, as biomass strongly increased from the two lower to the highest elevation of experiment and plants at the highest elevation of experiment grew best. The variation in growth in the common garden was probably due to environmental factors, most likely related to the elevations of the common gardens. The highest elevation of experiment seemed to be the most suitable elevation for growth of P. alpina, and it also came closest to the distribution center of P. alpina regarding elevation. Independently from their elevation of origin, all populations rapidly adjusted similarly to the elevation of experiment, through active or passive phenotypic plasticity. According to 59, phenotypic plasticity is crucial for the survival of, and might be particularly pronounced in, alpine plants. Strong phenotypic plasticity has often been found in alpine plants, e.g., by 52 in the specific leaf area of Campanula thyrsoides or by 24 in all measured traits of Festuca eskia. In a common garden experiment located at the same study site as ours, 45 found a high phenotypic plasticity in growth and fitness of Scabiosa columbaria.
We did not find differences in growth in the common garden related to the elevation of origin, which would have suggested genetic adaptation to elevation of origin among the seed-producing populations and among the bulbil-producing populations. The only exception was that seed-producing plants from the lowest elevation showed some advantage at the lowest and some disadvantage at the highest experimental site. This effect, indicating some adaptation to elevation, was, however, weak compared to plastic adjustment of the plants. In a common garden experiment, 20 also found that the vegetative, but not the reproductive biomass of P. alpina was independent of the elevation of origin. 27 found genetic differentiation in growth and reproduction of P. alpina related to the elevational origin in a transplant experiment along an elevational gradient from 425 to 1921 m a.s.l., but the plants were not locally adapted to their growing site. It could be that we did not find stronger effects of the elevation of origin on growth of P. alpina because of the high plasticity in growth of our plants and because the plants’ ability to respond to the elevation of experiment varies strongly. Another possible reason why we did not observe patterns suggesting strong genetic adaptation to elevation could have been that the growth conditions in the common gardens were much better than in the natural habitat of the tested plants, potentially obscuring genetic differences. Under more stressful conditions, possible genetic differences might have appeared more clearly. Finally, a factor that could have influenced our result is adaptation to the local conditions among the habitats of P. alpina populations independent of elevation, which could have had a greater impact on the populations than clinal changes along the elevational gradient on the Furka, as has been found in other alpine species (59).
Reproductive allocation did not differ among plants in the common gardens. Neither genetic differentiation among elevations of origin nor plasticity in response to elevational factors could be observed in this trait. It has been shown that reproductive allocation in Poa alpina is adapted to different types of land use (20). Our plants all originated from the same land use type, which could explain why we did not find any adaptive differences. In some studies, reproductive allocation of Poa alpina decreased with increasing elevation. 20 found a lower biomass allocation to reproduction in populations from higher elevations, and 27 found a decreasing proportion of reproductive tillers with increasing elevation. However, the decrease of reproductive tillers was accompanied by a simultaneous size decrease of the plant (27), while we found an increase of growth with elevation, which might have counteracted effects of increasing elevation. 19 found that alpine plants sometimes prioritize reproductive allocation over vegetative growth.
Conclusion
We found that the reproductive mode of Poa alpina strongly influenced the growth and reproduction of plants in the common gardens, suggesting adaptation of reproductive modes to elevation. The greater occurrence of bulbil-producing plants at higher elevations was accompanied by high costs of fitness for these plants transplanted at lower elevations. Within reproductive modes, we found only a weak indication for adaptation to different elevation, but a high capacity for plastic adjustment to environmental conditions. The low rate of sexually produced offspring seemed sufficiently high to maintain a higher genetic diversity of seed-producing P. alpina plants. Thus, by means of the two different reproductive modes and the capacity to adjust plastically, P. alpina seems to be able to occupy a broad ecological niche across a large elevational range. This ability could help the important fodder grass P. alpina to largely persist despite future climate change.
