The moss species studied have widespread distributions that span several continents47,48. All six species are common in the Iberian Peninsula, where they frequently coexist (e.g. refs. 52, 55). They differ in ecological requirements regarding humidity (D. scoparium is hygrophyte-mesophyte, H. cupressiforme and P. capillare are mesophyte-xerophytes and H. aureum, S. ruralis and T. squarrosa are xerophytes) and soil type (e.g. D. scoparium and P. capillare are more humicolous, and H. aureum is more arenicolous) (see ref. 47). Besides, they also differ in their bearing of archegonia and growth form; H. aureum and H. cupressiforme are pleurocarpous mosses that grow flat to the ground, whereas P. capillare, S. ruralis, T. squarrosa and D. scoparium are acrocarpous species that grow normally erect. Additionally, P. capillare can develop specialized asexual propagules in rhizoids, stems and leaves56.
The study consisted of two experiments, a pilot experiment to test different methods of artificial fragmentation and experimental conditions, and the main experiment, in which we tested the effect of propagule size on the establishment success.
In the pilot experiment we tested the six species, two different substrates (rockwool vs. felt), and four classes of fragments (fragments resulting from wet milling, and three size intervals of dry-milled fragments). We tested three replicates of all combinations (6 species x 2 substrates x 4 fragment types), thus we had a total sample size of 144. For wet milling, we first saturated the samples with water and milled them for 10 seconds in a regular electric coffee grinder with a single blade. Note that wet milling does not allow the separation of fragments of different size, and it may affect survival rates. For dry milling, we air-dried the samples and milled them in several successive steps until we obtained enough material of the different propagule size classes. To separate the fragments by size we sieved the dry-milled propagules through meshes of different size (small: Ø 0.25 mm – 0.16 mm, medium: Ø 0.45 mm – 0.25 mm, large: Ø 0.75 mm – 0.63 mm). This method separates propagules by their maximum width but note that propagules can substantially vary in length, especially in the cases of long and narrow propagules. The propagules were sown in 4 x 4 cm seedling pots and kept in a growth chamber in controlled conditions (temperature of 10 °C, photoperiod 10/12 h, and a PAR radiation light of 31 µmol m-2s-1), for 8 weeks. The results of the pilot experiment showed that there were no substantial differences between the establishment success of wet and dry fragments. Since dry milling allowed an easy separation of propagules by size, we used this milling method in the main experiment. Also, we used rockwool as a substrate because it is an inert material that does not interfere with growth and required less watering.
The main experiment consisted of a full factorial design including as factors the 6 species and three size classes of dry-milled fragments (small: Ø 0.25 mm – 0.16 mm, medium: Ø 0.45 mm – 0.25 mm, large: Ø 0.75 mm – 0.63 mm). Total sample size for the main experiment was 216 (6 species x 3 fragment sizes x 12 replicates). Each sample consisted of 0.004 g (±0.0002 g) of fragments sown in a 5 x 5 cm pot filled with rockwool. The experiment was done in the same culture conditions as the pilot experiment. The plants were allowed to grow for 2 months. At the end of the experiments, we made a qualitative assessment on the germination of the species observing whether the established shoots arose from leaf or shoot fragments.
We measured several parameters at the end of the experiment as indicators of establishment success: the number of established shoots, the percentage of surface covered by the established shoots in relation to the sample surface (colonized surface), the biomass of the established shoots (viable biomass) and non-viable biomass (non-established shoots and the rest of organic matter). Biomass was weighted after collecting and oven-drying the established shoots at 60ºC for 48 h. The biomass data (both viable and no viable biomass, which account for final biomass) was used to estimate the percentage of viable biomass respect to the cultured biomass and the Relative Growth Rate (RGR), calculated as follows:
These indicators account for local population growth as in Söderström & Herben (1997), taking into account that the shoots can be newly generated during the culture period, or be already present in the cultured propagules. This formula allows focusing only on newly generated biomass.
Propagule trait characterization
To characterize the traits of the propagules, we measured 30 propagules of each species and fragment size class (540 propagules in total) under the optical microscope. First, we annotated several morphological traits in hydrated state. Then, we calculated the proportion of shoots per every propagule size class, as it is known that shoots have greater germination potential than leaves18. Afterwards, we measured a viability index (hereafter apparent viability) according to some observable features, and assigning each propagule to values from 1 (large number of collapsed cells, apparent senescence) to 6 (few empty/collapsed cells, cytoplasm with plastids) in an ordinal scale. Additionally, we measured several quantitative size and shape traits in dry and hydrated states. For this purpose, we took photos of the same set of propagules, both dry and hydrated, and analysed the images using the software ImageJ58. We measured the area of the propagules in mm2; the maximum length in mm and the circularity, a shape descriptor that is size-independent and varies between 0 and 1, 1 being a perfect circle. Additionally, we compared the area (Equation 3) and maximum length (Equation 4) of hydrated and dry propagules:
To select the most relevant propagule traits we explored the correlations between the studied traits and the establishment success indicators using a correlogram and a network plot. The p-values of the correlations were adjusted using post hoc Holm adjustment method.
We used analysis of deviance (Type III test) to determine the effects of 1) propagule size (classes large, medium and small), 2) species cultured (D. scoparium, H. aureum, H. cupressiforme, P. capillare, S. ruralis and T.squarrosa) and 3) the interaction between these two factors on colonization. We measured establishment by using four indicators as dependent variables in the different analyses: number of established shoots, colonized surface, viable biomass and relative growth rate (RGR). We tested the deviations from normality and homoscedasticity of model residuals using Kolmogorov-Smirnov and Bartlett's tests for the main effects, and Levene's tests for the interactions. Since the tests showed large deviations from normality, we used robust models with IWLS (iteratively (re)weighted least squares) for all the indicators, and a box-cox transformation on number of established shoots, colonized surface and viable biomass, to cope with the violations of normality in these variables. Also, we used a type III analysis of variance with HC4 (heteroscedasticity-consistent) robust sandwich variance estimator to cope with heteroscedasticity for all the indicators. RGR transformation was not need, being enough with the robust model and robust sandwich estimator. Then, we performed planned contrasts for post hoc pairwise t-tests with Holm adjustment method for the p-values, to assess properly the interactions between the two factors. Finally, after adjusting the models we assessed that there was no overdispersion in the data. All statistical analyses and graphs were performed in R environment59, using the packages: car60, rstatix61, psych62, corrr63 , corrplot64, cowplot65 and MASS66.