The clonal Geum reptans (Rosaceae) is a perennial outcrossing rosette plant, and occurs preferentially on glacier forelands and moraines on siliceous rocks between 1950 and 3800 m a.s.l. (Hegi 1995). The plant is one of the first pioneers on virgin soils, and persists during succession. In the competition experiment, early successional habitats refer to recently (several years to a few decades) deglaciated areas; late successional habitats have been free from ice for at least 100 years.
Geum reptans reproduces clonally by forming new rosettes (ramets) with adventitious roots at the end of above-ground stolons, and sexually by producing flowering stems with a single terminal flower head. Reproductive meristems are located in the axils of leaves and are preformed in the season prior to emerging. At the end of summer, stolons, daughter rosettes that do not establish, and leaves of adult plants die back.
In the competition experiment, a total of 192 plants from four early and four late successional populations were grown either with or without P. alpina: 12 individuals per population and treatment (Table 1). Plants grew in two air-conditioned greenhouse compartments with an ecologically relevant air-temperature regime of 10 °C at night and 20 °C during daytime. In the temperature experiment, a total of 208 plants from five low- and four high-altitude populations were exposed to two temperature treatments. From each population, 12 individuals grew at cold temperature (7·5 °C at night and 17·5 °C during daytime) and 12 individuals at warm temperature (12 °C at night and 22 °C during daytime) in two air-conditioned greenhouse compartments per treatment (in one population only eight individuals per treatment were available; Table 1). The temperature difference of 4·5 K equals a difference in altitude of 750 m under the estimate of a lapse rate of 0·60 K per 100 m increase in altitude (Körner 1999). Plants within greenhouse compartments were randomized every second month. Within each compartment, a temperature sensor measured air temperature continuously and regulated the connected air conditioner (Airwell Type R-407C, ACE Klimatechnik GmbH, Frankfurt, Germany) to cool the compartment to the required temperature. Temperature increased during daytime due to incident solar radiation. All plants were kept inside the greenhouse from April to October, but overwintered outside in the garden of the Institute of Botany, University of Basel, Basel (270 m a.s.l.), Switzerland, where plants experienced frost.
Table 1. Location, population abbreviation, elevation, habitat type, and number of individuals per population and treatment of 17 populations of Geum reptans in the Swiss Alps used in two greenhouse experiments with a competition treatment and a temperature treatment, respectively
|Location||Population||Longitude (m)/ latitude (m)†||Elevation (m a.s.l.)||Habitat type‡||Number of plants in treatment:|
|Competition experiment|| || || || ||Without competition||With competition|
|Scaletta early, GR||SCE||791600/175430||2500||Early ss||12||11|
|Muttgletscher, VS||MUT||674500/156600||2520||Early ss||10|| 8|
|Vadret da Radönt, GR||RAD||792585/178485||2640||Early ss||12||12|
|Vadret da Grialetsch, GR||GRI||793500/175250||2660||Early ss||12||12|
|Scaletta late, GR||SCL||791750/175500||2330||Late ss||10||10|
|Val da Cambrena, GR||CAM||797100/142300||2340||Late ss||12||12|
|Flüelapass, GR||FLU||791700/180300||2420||Late ss||12||12|
|Blauberg, UR||BLA||675030/157920||2580||Late ss|| 9|| 9|
|Temperature experiment|| || || || ||Cold||Warm|
|Val Roseg, GR||ROS||786125/142900||2120||Low|| 6|| 5|
|Val Fex, GR||FEX||781325/137730||2140||Low||13||14|
|Flüela Schwarzhorn, GR||SWH||791400/178750||2900||High||12||12|
|Piz Languard, GR||LAN||793075/151450||3080||High||10||12|
Plant material was collected in early autumn 2000 and 2001 for the competition and temperature experiment, respectively. We collected c. 30 daughter rosettes with their stolons, each from a different mother plant from each population. The randomly chosen mother plants were located along a transect, and were at least 4 m distant from each other, except in the populations at Eggishorn and Flüela Schwarzhorn where plants were aggregated along crevices. Because there were few reproducing adults in the populations at Val Roseg, Muttbach and Flüela Schwarzhorn, more than one clonal offspring from three, four and seven mother plants, respectively, were collected and allocated equally to treatments. Plants collected were kept in moist bags in the dark at 4 °C until planting. By the end of October, 24 equally sized plants per population were planted individually into single rectangular pots of 9 × 9 cm and 10 cm in height. Pots were filled with 2 cm pumice-gravel as drainage, followed by a 1 : 1 mixture of sand and pot-soil. Plants were randomly allocated to the two treatments in each experiment. After 3 weeks, in half the pots of the competition experiment four randomly chosen bulbils of P. alpina were planted as competitors in each pot, one at each edge of the pot. Poa alpina bulbils were collected at four sites at different altitudes and kept in a moist bag in the dark at 4 °C until planting. After the first winter, the number of P. alpina individuals per pot was reduced to one, and every second month individuals of P. alpina were reduced to two tillers. In each experiment plants within populations were randomly and equally partitioned to greenhouse compartments. Statistically, greenhouse compartments were treated as blocks. In the temperature experiment blocks and temperature treatments fall together.
The remaining ramets were potted the same way. During the first winter, 14 and 21 individuals died in the competition and temperature experiment, respectively. Whenever possible dead plants were replaced by individuals of the same population or, in the temperature experiment, of the same elevation. Plants that died later were not replaced. On 15 July 2002, the 161 remaining plants were brought to 2000 m a.s.l. near Davos, Graubünden, Switzerland, to compare their reproductive behaviour with the plants in the greenhouse.
In the greenhouse, plants received additional light from 1 kW lamps for 3–5 h during daytime, according to day-length. All plants were watered twice a week and received the same amount of full fertilizer (in 2002, 9 kg N ha−1; in 2003, 9 kg N ha−1 plus 23 kg N ha−1 with a higher K content, applied in equal portions during the vegetative period). Potassium promotes the growth of reproductive meristems.
Numbers of leaves per plant were recorded at the beginning of the experiment and analysed to test for equal plant size between treatments and blocks. This measure was also used as a covariate in statistical analyses.
The final harvest for both experiments was in July and August 2003, 3 and 2 years after the beginning of the competition experiment and the temperature experiment, respectively. Numbers of rosettes and of green and dead leaves produced in 2003 were counted. Above- and below-ground plant biomass was measured after drying at 80 °C for 48 h. Axils of leaves formed in the year of the final harvest were checked for reproductive meristems and, if present, they were cut and stored in 50% alcohol to determine if they were future flowers or stolons. In the greenhouse, few individuals reproduced in 2003 (8·7%), whereas 73% of the remaining plants kept at 2000 m a.s.l. reproduced, despite a similar number of preformed reproductive meristems in both groups. Therefore we judged preformed meristems as a more precise measurement of the reproductive potential of populations in the greenhouse than the actual number of flowers and stolons. In the common garden, vernalization was probably too short to initiate enough preformed meristems to grow because of the milder climate during winter compared with the conditions at 2000 m a.s.l.. Reproduction was constantly low in the earlier years and should not have influenced reproductive allocation in the year of the final harvest.
Vegetative and reproductive measurements were used to test for effects of origin (habitat type and population) and environment (treatment) by hierarchical analysis of covariance (ancovas). In the competition experiment, habitat type (early and late succession) was tested against population, treatment against the interaction of treatment and population, and population, as well as the interaction of habitat type and treatment against the remaining residual. In the temperature experiment, treatment was tested against block, habitat type (low and high altitude) against population, and population, the interaction of treatment and habitat type, as well as the interaction of treatment and population against the remaining residual. Plant measurements were ln-transformed to test for relative differences, and the proportion of stolons on all reproductive organs was arcsin-transformed. Because several individuals produced neither stolons nor flowers, the requirement for the ancova could not be fulfilled with individual data points. Instead, we used means of reproductive meristems per population within treatment and block for the ancova. Correlations were also calculated with population means with a non-parametric Spearman's rho (rs) procedure instead of parametric Pearson's product-moment (R) tests.
Phenotypic measurements of experimental plants included a genetic and a maternal environmental component. The maternal environment can influence the nutrition and growth of clonal offspring. Therefore we used the number of leaves at the beginning of the experiment as a measure of maternal effects. In both experiments, this measure did not affect the number of reproductive meristems (P > 0·1), and did not influence plant size at the final harvest in the competition experiment, but did influence size at final harvest in the temperature experiment (R = 0·22, P < 0·01 for leaves and above-ground biomass). To minimize any maternal effects we used the number of leaves at the beginning of the experiments as a covariate in all the analyses.
Plants were classified according to their reproductive meristems in individuals with stolons, with flowers, with stolons and flowers, or non-reproducing individuals. A Poisson log-linear model was used to test whether reproductive behaviour differed between habitat types or treatments.
All statistical analyses were carried out with R (Ihaka & Gentleman 1996), a shareware package similar to s-plus.