DISCUSSION

This is the first study that tested how above- and belowground functional traits of grassland species relate to their success across multiple spatial scales. Among the 242 study species, low root tissue density was identified as important for all success metrics. However, the effects of the other traits were frequently dependent on the success metric considered. For example, while seed weight was negatively related to the occurrence frequency in Germany, it was positively related to the abundance in German grasslands. This illustrates that some traits may have opposite effects on different dimensions of species rarity and commonness. Moreover, we found that many of the effects had significant non-linear effects, in most cases with the highest success-metric value at intermediate trait values. While most previous studies on relationships between functional traits and species distributions focussed on easy-to-measure aboveground traits , our study shows that belowground traits can also explain a significant amount of variation in species success.
The spatial scale of success ranged from local abundance in 16 m² grassland plots in Germany (GPs and EPs) to the global occurrence outside the native range (number of GloNAF regions). While the local scale abundance data are restricted to a single habitat type, the occurrence at larger spatial scales also covers other habitat types (e.g. 86% of the area in Germany is not used as grassland; BMEL 2017). Each habitat type might select for different values of a trait , resulting in no clear relationship between the trait and the success metric overall. This context specificity could explain why the models of the success metrics at larger spatial scales had overall fewer traits with significant effects than the models on abundance in the grassland plots.
Accordingly, we found varying degrees of consistency in the trait values of successful species across spatial scales. Root-tissue density was the only trait with a consistent effect on all success metrics. Probably, a low root-tissue density, which is indicative of a high resource-acquisition-rate strategy , is beneficial in nutrient-rich habitats, which are locally and globally widespread as a consequences of agriculture. On the other hand, the effect of specific leaf area, an aboveground trait associated with the resource-acquisition strategy (Onoda & Wright, 2018), depended on the spatial scale of the success metric. Occurrence frequencies at all spatial scales tended to asymptotically increase with specific leaf area, which is in line with the frequent observation that high specific leaf area promotes invasion success . However, abundance in the grassland plots was negatively related to specific leaf area, possibly reflecting that persistence under highly competitive pressures in dense grasslands could require a more conservative growth strategy.
Bud-bank size had significant effects on all success metrics, but the direction and shape of the relationship varied a lot. Species with a large bud-bank had the highest abundance and occurrence frequency in the grassland plots. A large bud-bank is essential for regrowth of long-lived perennials after e.g. grazing or mowing . Species with small bud-banks, on the other hand, had the largest naturalized ranges. This dual effect of bud-bank size on success was also evident from the fact that the highest occurrence frequencies in the native range and in Germany were found for both species with high and low bud-bank sizes. Although buds themselves are not very costly , they require bud-bearing organs and nutrient reserves, which may trade-off with seed production . Thus, species with smaller bud-banks may invest more in seed production, resulting in a higher dispersal ability and larger native and naturalized distributions.
The effect of seed weight on species success metrics ranged from positive at the plot scale to negative or absent at larger spatial scales. The finding that species with heavy seeds tended to be more frequent and abundant in the grassland plots, most likely reflects that large amounts of resources stored in seeds increase seedling survival under the strong competition in grasslands . Species with light seeds, on the other hand, might have a higher reproductive output , could potentially disperse over longer distances , but see Thomson et al. , 2011), and could persist longer in the seed bank . At the larger spatial scales, this benefit of small seeds could have compensated or overcompensated the reduced seedling survival chances.
Maximum rooting depth was related to success metrics at the largest spatial scales, where it had positive effects. Deep roots allow a plant to take up water with nutrients from deeper soil layers, increasing survival and growth, particularly when the surface soil regularly dries out . As most of the agriculturally used grasslands in Germany are mesic , this could explain why rooting depth was not significantly associated with occurrence frequency and abundance in the grassland plots. At the larger spatial scales, which also cover other habitat types, species with deep roots might be more persistent. For naturalization success, however, there was a significant non-linear effect of rooting depth as both deep-rooting and superficially rooting species were most successful. This could indicate that the alternative strategy of lateral spread to acquire resources and avoid competition with deeper rooting species might also be beneficial at the global scale.
Height of the plants was not related to success of the species at larger spatial scales. This is surprising, particularly for naturalization success, as numerous studies on naturalization and invasion success found that tall species were more successful . Most of those studies, however, considered woody species that are overall much taller than the grassland herbs in our study. Interestingly, while plant height increased abundance in the German grassland plots (at least in the GPs), it decreased the occurrence frequency in those grasslands. On the one hand, tall plants, when they occur somewhere, might be competitively superior and become dominant, whereas, on the other hand, small plants might be less at risk of losing reproductive organs due to mowing or grazing.
We found that species that are abundant in grasslands are typically characterized by thin low-density roots, which promote the uptake of belowground resources. The patterns for success metrics at larger spatial scales, at least those in the native ranges, were less clear. Indeed, plant traits explained large proportions of the variation in local abundance and occurrence frequency in the grasslands (>25%; Table 1), whereas the proportion of variation in occurrence frequency in the native range and in Germany was very low (2% and 11%, respectively; Table 1). This suggests that plant traits could be good predictors of species success if one considers a single habitat type, but that this is less the case for success metrics at large spatial scales that are not habitat specific. However, a notable exception is the global naturalization success of the species, as 41% of the variation in occurrence frequency outside the native range was explained by the plant functional traits. Possibly, this reflects that most naturalizations happen in anthropogenic environments , and thus largely in a single habitat type.
The plant economics spectrum postulates that the high specific leaf area typical for “acquisitive” plants should be mirrored belowground by a high specific root length, low root-tissue density and a low root diameter . Indeed, specific leaf area was negatively correlated with root-tissue density, but it was not significantly correlated with specific root length and diameter of the first-order roots (Appendix S3). This decoupling from the plant economic spectrum has previously been found for grassland plants as well as tree seedlings . Seed weight and plant height, the other aboveground traits frequently used in studies on functional ecology of plants, were also not very strongly correlated with the belowground traits in our study. These belowground traits explained a considerable proportion of variation in the success metrics, in addition to the variation explained by the three aboveground traits (Table 1). Indeed, for all success metrics, except occurrence frequency in the EPs, the belowground traits explained at least as much of the variation in success metrics as the three aboveground traits did. Therefore, our results show that aboveground traits cannot substitute for belowground traits in studies on plant functional ecology.