Herbs are different: clonal and bud bank traits can matter more than leaf–height–seed traits

Current understanding of functional differences among plant species is based on several key axes of specialization in response to environmental gradients, namely of productivity and disturbance. These axes include leaf economy spectrum, plant size and dispersal ability (Westoby, 1998; Westoby et al., 2002; Wright et al., 2004). These axes are typically proxied by easily measurable traits, namely specific leaf area, height at maturity and seed mass (leaf–height–seed (LHS) traits of Westoby, 1998) which capture well variation in a number of correlated plant characteristics (Westoby & Wright, 2006; Laughlin et al., 2010). While the first comprehensive trait-based analysis was dealing primarily with herbs (Grime, 1977), LHS-based analyses turned out to be particularly powerful to describe ecological differentiation of woody plants. In spite of the fact that woody species constitute less than a half of existing plant species (FitzJohn et al., 2014), much current understanding of plant functional tradeoffs is based on species sets that contain primarily woody species (Verdu, 2002; Kerkhoff et al., 2014; Lamanna et al., 2014). However, differentiation of herbaceous plants is likely to be shaped by factors different from woody species. They do not possess permanent aboveground structures which permits entirely different response to disturbance (Aarssen et al., 2006; Aarssen, 2008; Zanne et al., 2014), but also to other factors, such as drought (Bennett et al., 2015). While strong and infrequent disturbance kills whole plants and generates a long gradient of time-since-disturbance which favours woody species (Meiners et al., 2015), weaker and regular disturbance events leave belowground regenerative organs intact, favouring plants which can resprout from them (Bellingham & Sparrow, 2000; Mackey & Currie, 2001; Buoro & Carlson, 2014). Such disturbances preclude plants from developing long-lived structures aboveground found in trees and favour resprouting from permanent belowground structures (Bellingham & Sparrow, 2000; Vesk et al., 2004; Dietze & Clark, 2008) as found in shrubs (Bond & Midgley, 2001; Vesk et al., 2004; Dietze & Clark, 2008; Clarke et al., 2013) and herbs (Klimešová & Klimeš, 2007; Meiners et al., 2015). Further, short disturbance intervals (such as winter frost in temperate climates, regular drought or grazing) require rapid resprouting and therefore cheap aboveground tissues that are non-woody (Vesk, 2006; Zanne et al., 2014) and thus largely eliminate selective advantage provided by vertical growth and favour expansion in a horizontal dimension by clonal spread. Selective forces operating on resprouting plants are thus fairly different from those operating on trees. As a result, tradeoffs and correlations of their traits should be different from those known from woody species (Meiners et al., 2015). Still we do not know to what extent the key role of the LHS-differentiation applies to herbs as it does to woody species, and trees in particular. We also do not know whether and how traits that determine ability to resprout and spread horizontally in space (namely traits of belowground bud banks and traits of clonal growth) fit into this differentiation or whether they constitute an axis of specialization independent of the LHS traits. We examined relationships between LHS traits and traits of bud bank (size and depth) and of clonal growth (extension of lateral spread, multiplication rate and persistence of clonal connections; for an illustration see Fig. 1) of Central European species (c. 2000 species). Analysis of this data set shows that almost all relationships between LHS traits, bud bank and clonal traits are very weak (Table 1; for methodological details see Supporting Information Methods S1 and Tables S1 and S2). With the exception of bud bank depth and connection persistence, LHS traits explain < 5% of the variation in clonal and bud bank traits (and often much less). This means that the variation on traits that determine resprouting and clonal spread do not fit well into the syndromes of the LHS traits. The few significant cases of correlation of clonal and bud bank traits with the LHS traits (see Table S3; Fig. S1) are likely to be due to two different reasons. First, bud bank depth, and to a much lesser extent bud bank size and lateral spread, are positively correlated with plant height at maturity. These relationships are likely to be due to allometric effects of plant size (Enquist & Niklas, 2001; Schenk & Jackson, 2002). Tall plants are likely to have more extensive root and rhizome systems, bear more buds than short roots and rhizomes (VanderWeide & Hartnett, 2015) and are more likely to spread farther away from the mother plants. Second, negative correlation of specific leaf area with connection persistence implies that clonal plants are also differentiated in terms of leaf economics, and faster growing plants of more productive conditions tend to have short-lived connections between ramets and longer rhizome increments relative to plants that grow slower (see also Jónsdóttir & Watson, 1997; Craine et al., 2001). An independent axis of plant specialization is not characterized only by a trait which is not correlated with LHS traits, but, at the same time, it must improve substantially the predictive power of LHS traits for plant distribution in the field. Existing data show that traits of clonal growth and belowground bud bank (namely bud bank size and connection persistence) predict species distribution along ecological gradients in Central European vegetation at least as well as the LHS traits. Their net predictive power is similar to that of the LHS traits combined (Fig. 2; Table 2) and is not due to the (small but existing) covariation of these traits with the LHS traits. This leads to a conclusion that clonal and bud bank traits, that is, traits that determine resprouting and clonal multiplication, indeed form axes of specialization of temperate herbs fairly independent of the LHS traits. However, it should be noted that the combined predictive power of both groups of traits reaches only c. 10% of the total variation in the species habitat distribution in the field. First, the distribution along ecological gradients is best predicted by bud bank size and connection persistence. In contrast to the LHS traits, which (namely leaf traits) capture the role of physiological performance and the pace of life, the bud bank size and clonality per se expresses primarily morphological/developmental constraints on the potential to resprout. Although the herbaceous habit probably did not evolve in temperate regions (Zanne et al., 2014), prevalence of resprouting and/or clonally growing herbs is the determining feature of temperate vegetation, both in (semi-)natural habitats and in communities regularly disturbed by human activities such as arable fields, mown and pastured grasslands, and ruderal places (Klimešová & Herben, 2015). The resprouting ability in herbs is likely to have evolved under a number of disturbance regimes, such as large ungulate grazing, fire, flooding and avalanches (Klimeš et al., 1997; Bellingham & Sparrow, 2000; Klimešová & Klimeš, 2007). Second, the differences due to connection persistence and correlated clonal traits are likely to reflect the gradient of stress/productivity as clonal plants with persistent connections are known to occur in stressed and unproductive conditions (Jónsdóttir & Watson, 1997; Klimešová & Herben, 2015). Slow growing clonal plants tend to have long-lived connections between ramets and shorter rhizome increments relative to plants that grow faster (see definition of different clonal growth strategies in de Kroon & Schieving, 1990; Jónsdóttir & Watson, 1997; Craine et al., 2001). This parallels development of costly, but long-lived leaf tissues in slowly growing plants in stressed conditions and cheap short-lived tissues in fast growing species under productive conditions (Reich et al., 1992; Westoby & Wright, 2006). Our analyses support the hypothesis that functional differentiation within herbaceous plants thus is based, to an important extent, on dimensions different from those that are used in large-scale studies that include primarily woody species. This is due to the two fundamentally different strategies that plants use to cope with disturbance: tolerance (resprouting) and avoidance (by dispersing to disturbed habitats by seeds, either in time or in space). The LHS system captures seed and growth-related traits that are primarily important for disturbance avoidance, whereas milder and more frequent disturbances select for the second strategy, which is associated with clonal and bud bank traits that supply reserve meristems for survival. Further, it should be noted here that while the predictive power of clonal and bud bank traits relative to that of the LHS trait is high, neither of them is a very strong predictor of species habitat distribution. This may be partly due to the restricted range of plants studied here, as traits of temperate herbs have smaller ranges relative to what is found across the plant kingdom, namely in height, but also in seed size (see e.g. Thompson & Rabinowitz, 1989; Moles et al., 2007). However, we would like to point out here the low predictive power of all commonly collected traits for species distribution along ecological gradients which is often not paid sufficient attention (but see e.g. Coyle et al., 2014; Moles et al., 2014). Finally, we would like to point out that clonal and bud bank traits should be given more prominence in trait-based analyses of species habitat distribution and coexistence. Clonality is a key potential of plant bodies (Mogie & Hutchings, 1990; Klimešová et al., 2015) with deep interspecific differences in the way in which it is expressed (Klimeš et al., 1997), but we are only beginning to appreciate its full ecological implications. Temperate herbs show rather a broad range of clonal growth types (J. Klimešová, 2015, unpublished data), but clonal growth is also found in trees and woody species, albeit in lower proportions (Aarssen, 2008). While collecting data on its traits is often more difficult than collecting data on LHS traits, it is feasible and has been done both for herbs and woody species (Bond & Midgley, 2001; Klimešová & de Bello, 2009; Clarke et al., 2013; VanderWeide & Hartnett, 2015). Comparative analyses of their role relative to other traits, and their mutual relationships across different biomes and different floras would provide better understanding about how plants respond to a number of factors, namely disturbance. Comparison of the role of clonal growth and bud banks in herbs and woody species would also shed more light on differences between these two fundamental ecological strategies in the plant kingdom. The authors thank Irena Šímová, Robin Pakeman, Benjamin Blonder and three anonymous referees for thoughtful comments on earlier versions of this paper. The research was partly supported by the Grant Agency of the Czech Republic (projects GA P505/12/1007, P505/13-17118S, Centre of Excellence 14-36079G, PLADIAS). J.K. and T.H. planned and designed the research, J.K. and O.T. assembled data, T.H. analysed data, T.H. wrote the manuscript with contributions of J.K. and O.T. Please note: Wiley Blackwell are not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing material) should be directed to the New Phytologist Central Office. Fig. S1 Pairwise plots of leaf–height–seed (LHS) traits and clonal traits. Table S1 Bud bank and clonal growth traits used in the analyses Methods S1 Data sources and data analysis. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.