Sampling data
The study spanned five years, from 2011 to 2015, and was conducted in an experimental garden near the town of Bechyně, Czech Republic (coordinates 49°18′36″N, 14°30′14″E). The mean annual temperature during this period was 8.8 °C. Monthly temperature variations ranged from a minimum of -0.1 °C in January to a maximum of 19 °C in July. Cumulative annual precipitation was variable, reaching a minimum of 431 mm in 2015 and a maximum of 818 mm in 2013 (Douda et al. 2018).
We established a mesocosm experiment designed to simulate changes in the hydrological regime and interspecific interactions within a wetland ecosystem. Wetlands are characterised by cyclical flooding and periods of drought, creating highly productive environments typically dominated by one or a few species. Such ecosystems often have dense herbaceous canopies, indicating strong asymmetric competition for light (Keddy 2010). In our study, we established plant communities consisting of four herbaceous species that commonly coexist in central European wetland forests (Douda et al. 2012). These communities were grown in 48 large (90 L) plastic containers (360 mm high, 660 mm diameter; IKO90 CONTAINER).
Each container always contained three subordinate species:Calamagrostis canescens (Weber) Roth, Carex elongata L. and Deschampsia cespitosa (L.) P. B., together with a dominant species, Carex elata All. Dominant species averaged at least 2.6 times higher biomass than subordinate species (Doudová & Douda 2020). All four species are characterised as wind-pollinated, long-lived, clonal perennials that produce multiple ramets. Specifically, C. canescens produces guerrilla genets with rhizomes connecting widely spaced ramets, whereas the other three species produce phalanx genets with densely clustered ramets (Douda et al. 2018). All plant species begin sexual reproduction in the second year. This is when the varying number of ramets within each genet begin to produce inflorescences and achenes. As is common in graminoids, ramets that produce inflorescences invest more in stem height but less in total leaf biomass (Reekie & Bazzaz 1987).
We started the experiment by sowing stratified seeds in trays in March 2011. Then, in April 2011, uniformly sized seedlings were transplanted into experimental containers (see Douda et al., 2018, for details on sampling and stratification of seeds). Seedlings were systematically placed at precisely defined positions in a grid, with an approximate spacing of 14 cm. If any plant died within the following 4 weeks, replacements were planted. Replacements were of the same species and similar size. The placement of the species within the pots was randomised. Each pot contained a total of 20 plants, distributed as five individuals for each of the four species. The pots were filled with a mixture of commercial soil and sand in a 2:1 ratio (v:v; pH = 5.5). The pots were seasonally fertilised with 10.6 g of CERERIT fertiliser (8-24-11, N-P-K + micronutrients; AGRO CS).
We implemented a factorial design with three hydrological regimes: (a) a well-watered control, (b) an interannual drought, and (c) a permanent drought, coupled with a dominant species biomass removal treatment (i.e., with (A) and without (B) the dominant species). Treatments were replicated in eight fully randomised blocks, each containing the six (3 × 2 factorial) treatments (a, b, c × A, B). Prior to planting, pots were modified to establish water levels based on the specific requirements of each treatment (see Douda et al. 2018 for technical details). In the well-watered control treatment, the soil remained fully saturated with water throughout the experiment, reflecting the natural conditions of wetland habitats where groundwater is at the soil surface for most of the year (Hulík & Douda 2017). In the permanent drought treatment, the water surface was maintained at 25 cm below the soil surface (since 2012), while in the interannual drought treatment, the water surface was decreased from the optimal level in some years (2011, 2013, 2015) to the permanent drought level in other years (2012, 2014) at the beginning of each growing season. The mean volumetric soil water content was 39.9% ± 1.2% (mean SE, n = 16) in the permanent water-lowering pots compared to 77.2% ± 1.8% in the high-water pots (TRIME-EZ, TRIME TDR System moisture sensors, Imko GmbH, Ettlingen, Germany).
During the 2011 season, all pots received full irrigation to facilitate community establishment prior to the initiation of the three water level treatments in 2012. To assess the effect of interspecific interactions between dominant and subordinate species on reproductive allocation, we used a biomass removal method. In June 2013, we clipped all ramets of the dominant Carex elata (24 pots) that had grown alongside the three subordinate species during the previous two years (2011–2012). This treatment was maintained by monthly re-clipping, considering the occasional re-sprouting of C. elata during the first year.
To assess species allocation to generative reproduction, we counted the number of flowering ramets (shoots) and total ramets (including flowering and non-flowering shoots) instead of aboveground biomass (whole-plant biomass could not be determined without destructive sampling) (Douda et al. 2018). Over five consecutive years, we counted the number of flowering and all ramets of each species per container annually; for flowering ramets, at the time of maximum flowering of each species, and for total ramets, at the end of each growing season in August - a period coinciding with peak plant biomass. To measure individual plant reproductive investment over years, we also counted the number of fruiting and all ramets per plant of tussock species where individual plants are easily identified (i.e., Carex elata ,C. elongata , and Deschampsia cespitosa ). We confirmed that the number of all ramets was strongly correlated with aboveground biomass (Douda et al. 2018), and the number of flowering ramets with fruit mass when we compared the number of ramets and yield in the last year of the experiment (Supporting information S1). Controlling the confounding effects of developmental rate and environment is necessary to interpret reproductive allometry in plant populations (Supporting information S2). In particular, our experimental design removed the confounding effects of individual plant age on plant development by creating an even-aged community.