In all studied sites, lying deadwood supported the occurrence of some vascular plants species and its characteristics influenced their species composition. While typical understory species tended to colonize deadwood at advanced decomposition stages, the initial stages of deadwood decomposition were usually more suited to species of open habitats or invasive species. A significant proportion of vascular plant species observed (on average 40%) showed substrate preference for deadwood or for soil, often across multiple sites.
No vascular plant species exclusively colonised deadwood as a substrate. This has similarly been demonstrated in other studies from temperate (Chećko et al. 2015; Chmura et al. 2016) as well as from boreal (Kumar et al. 2017) forests. However, species composition on deadwood usually differs from that on mineral soil substrate (Kennedy and Quinn 2001; Six and Halpern 2008; McDonald 2013; Kumar et al. 2017). The specificity of deadwood substrate thus increases the microsite diversity of forest ecosystems because each decaying log is unique for its species identity, decomposition stage (and the combination of both), dimensions, location, mortality type and the time when it initially fell down (Crites and Dale 1998; Pyle and Brown 1999, 2002; Skubała and Duras 2008; Bunnell and Houde 2010). This complex interplay of factors then has a strong effect on the species composition of plant colonizers and partly also on the density of their stems. As deadwood decomposition proceeds, the species composition of vascular plants tends to gradually converge with that on mineral soil (Kumar et al. 2017). Among vascular plants, substrate preference for deadwood is most often reported for the germinants of Picea abies and also some other tree species (Chećko et al. 2015, Orman and Szewczyk 2015, Orman et al. 2016). Preferences of herbs for deadwood substrate have only rarely been mentioned in other studies and mostly in connection with ferns (McGee 2001; Unar et al. 2017). Nevertheless, it seems that preferences of vascular plants for a specific substrate are much more common than previously thought. We identified species that exhibited a preference for deadwood substrate and this preference was usually consistent between sites. For instance Dryopteris carthusiana agg. showed preference for deadwood in all forest types except the mountain spruce forest. Conversely, soil substrate was unambiguously preferred e.g. by Athyrium filix-femina. Variation in species composition between deadwood and soil substrates can thus be explained by whether and how much vascular plants tend to colonize deadwood (Dittrich et al. 2014; Chećko et al. 2015; Unar et al. 2017). Our findings suggest that deadwood is often colonized by early successional species. For instance, Impatiens noli-tangere, I. parviflora, Epilobium angustifolium, Rubus idaeus, R. fruticosus agg. and Urtica dioica are mentioned in relation to Sambuco-Salicion capreae, an alliance comprising mesophilic shrub communities of forest clearings, canopy gaps and disturbed sites (Chytrý ed. 2013). The highest proportion of species preferring deadwood being observed in beech-spruce-fir forest (Žofín) may thus be attributed to recent severe disturbances at this site. This was also reflected by the fact that deadwood was often colonized by shade-intolerant species (Epilobium angustifolium, Rubus idaeus, Rubus fruticosus agg. Taraxacum sect. Ruderalia). Species preferring soil as a substrate (e.g. Athyrium filix-femina, Carex remota, Galeobdolon montanum, Galium odoratum, Gymnocarpium dryopteris and Veronica montana) were generally more shade-tolerant. The optimum of preferences of Epilobium angustifolium and Taraxacum sect. Ruderalia for deadwood substrate in the first decay stage likely results from their natural tendency to occur in recently disturbed areas and thus fresh deadwood may be of secondary importance for their occurrence.
The highest proportion of species preferring soil was observed in the mountain spruce forest. This forest type is, however, very specific in many aspects, compared to the other study sites. The absence of accumulated broadleaved litter with its suppressing effects on herbaceous species (especially in beech dominated forests) is very important for the dynamics of the herb layer (Wulf and Naaf 2009). Deadwood reaching above the compact layer of undecomposed beech litter thus locally becomes the only substrate which can be colonized by vascular plants (Orman et al. 2016). In dispersed mountain spruce forests, the herb layer is not so limited by litter (Semenyuk et al. 2020), which may explain the highest proportion of species with their optimum on soil substrate in this forest type.
Along with other studies (Zielonka 2006, Dittrich et al. 2014; Chećko et al. 2015) we show that the process of deadwood colonization by vascular plants already begins in the first stages of its decomposition. Christy and Mack (1984) suggested that the decomposition stage of deadwood plays a more important role in how long seedlings survive than in their initial ability to colonize the substrate. Our results that the species and decay class of deadwood may shape species composition and density of colonizers have also been demonstrated in other studies (Staniaszek-Kik et al. 2014; Chećko et al. 2015; Kumar et al. 2017). In this respect, we show that more acidic substrate of Abies alba and Picea abies (Asplund et al. 2015; Táborská et al. 2015; Kaufmann et al. 2021) exhibited a different species composition of colonizers and lower numbers of their stems in comparison to the broadleaved Fagus sylvatica. This disproportion was also previously reported in other colonizer groups, such as lichens, mosses and liverworts (McAlister 1997; Rambo 2001; Táborská et al. 2015). On the other hand Quercus robur in alluvial forest shows higher bark acidity compared to other broadleaves (Larsen et al. 2007; Jüriado et al. 2009), but we found the highest stem densities of colonizers on decaying logs of this species.. Possibly large log dimensions, long residual time (Vrška et al. 2015) and rough, uneven bark in which propagules easily get stuck (Chećko et al. 2015) greatly facilitate its colonisation.
Symbiotic interactions of vascular plants with other organisms likely play an intriguing role in the process of deadwood colonization. Associations with mycorrhizal fungi have been described in many plant species preferring a deadwood substrate, e.g. Dryopteris carthusiana (Rünk et al. 2012), Dryopteris filix-mas (Cooper 1977), Oxalis acetosella (Koorem et al. 2012) or Vaccinium myrtillus and V. vitis-idaea (Bonfante-Fasolo et al. 1981). Deadwood provides colonizers with higher light availability and favourable temperature regimes, thereby supporting the infection of plant roots by mycorrhizal fungi. Hayman (1974) described this phenomenon in Oxalis acetosella, Circaea alpina and other plant species. Unrug and Turnau (1999) pointed to greater mycorrhizal colonization of Arum-type for individuals on deadwood than for individuals on soil. The characteristics of substrate may thus determine when mycorrhizal interactions develop as well as the set of species involved in these interactions (Lehnert and Kessler 2016). Whether the species composition of symbionts or mycorrhizal type of plants growing on soil and deadwood substrate differ needs further investigation in future research.
Forests in Central Europe, especially those at productive sites, have long been affected by intense anthropic pressure. After traditional forest management practices are abandoned deadwood accumulation may take dozens of years to reach the levels commonly found in the existing old-growth forests (Vandekerkhove et al. 2009). This is why our study does not cover the lowland Quercus-Carpinus forests, although the areal extent of these communities is not negligible. Even though many forest reserves can be classified as this forest type, most of them were designated for the active protection of biodiversity and not for the protection of the natural dynamics of forest communities. These forests were mostly under strong anthropic pressure in the past and have only a short history of protection, resulting in a minimal amount of deadwood left for complete decomposition (Kapusta et al. 2020).
Although we attempted to bridge the differences in sampling procedures on mineral soil and decaying logs in our analyses, these differences could still lead to some biases. The sampling plots on mineral soil and decaying logs mainly differed in size and geometry. The discrepancies in sampling areas (25 m2 for mineral soil vs. an average of 6.8 m2 for decaying logs) would mainly lead to a higher probability of species occurrences in larger plots (i.e. mineral soil), but this was mitigated by our modelling approach. Another potential bias could be related to higher sampling omission rate in larger plots. However, this issue should not be serious at our relatively species poor study sites. In terms of plot geometry, mineral soil sampling plots were squares, but decaying logs had an elongated shape. Elongated plots, compared to compact ones, can be expected to encompass more heterogeneous environmental conditions, which may promote higher alpha diversity (Dengler 2008; Bacaro et al. 2015). This pattern is more important on larger landscape scales than on local scales (Kunin 1997), such as the scales of decaying logs. Still, this could potentially bias our results, thereby overestimating species occurrence probability on decaying logs. Nevertheless, we did not only compare species occurrence between mineral soil and decaying logs but rather investigated more nuanced patterns along the gradient of deadwood decomposition, where the effect of such bias should be relatively low.