Effect of novel competitors
The addition of novel competitors did not significantly affect the abundance of the alpine species, and none of the four trait axes of the full PCA correlated with species’ demographic response to the treatment (Figures S16-S17, Table S5).
Discussion
As climate warming continues (IPCC 2021), predicting which species will “win” or “lose” is crucial for community- and ecosystem-level predictions. Here, we measured the abundance change of 16 alpine plant species as a response to climate warming in a manipulative field experiment, and related these responses to species traits. We found that species’ demographic responses to the manipulation could be predicted by their size and water use strategy: small species with a more prolific water use suffered most from the transplantation to warmer and drier climates. Climate change is likely to influence water availability in the Alps by increasing the frequency and severity of heatwaves (Beniston 2004). Consequently, water limitation is likely to become an important driver of community change in alpine grasslands.
The advantage of traits related to conservative water use in a warming climate has been shown before: morphological traits describing conservative water use were related to increased abundance after warming in mountain plant communities (Soudzilovskaia et al. 2013), and drier conditions are likely to promote resource-conservative species in the Arctic (Bjorkman et al. 2018) and in the Mediterranean (Pérez-Ramos et al. 2012). Soil drying can be an important driver of plant community dynamics in other alpine and Arctic systems (Harte & Shaw 1995; Chapin III et al. 1996). In our experiment, the largest change in species abundance occurred after a drought in 2018 that was particularly pronounced at the warmest site (Figure S18). Increased frequency of these types of extreme events could be critical for future plant community structure in the Alps (Jentsch & Beierkuhnlein 2008; Liu et al. 2015).
Consistent with our assertion that climate change poses more of a water than a temperature challenge in our study system, we did not observe a shift towards species with higher thermal optima (i.e. higher Topt and lower Q10). Such a shift would be expected if changes in community composition were driven by physiological adaptations to temperature per se . Previous studies have recorded “thermophilization” of alpine and Arctic plant communities due to climate warming, i.e. a shift towards “warm-adapted” species, as inferred from their distribution (Gottfriedet al. 2012; Govaert et al. 2021; Lynn et al.2021). Nevertheless, the distributional limit of a species might not reflect adaptation to temperature as such, but rather to correlated factors, such as soil moisture or competition. For example, competitive ability is likely to trade-off with cold tolerance (Koehler et al. 2012; Pellissier et al. 2018), allowing species without specific cold-adaptations to persist with higher levels of competition. In general, higher temperatures are rarely detrimental for alpine or arctic plants (Chapin 1983; Gauslaa 1984) and optimal temperatures for plant gas exchange change only a little across temperature gradients (Körner & Diemer 1987).
Predicting the winners and losers in our experiment was possible by measuring either eco-physiological (long-term water use efficiency and stomatal conductance) or size-related morphological traits, and could have been possible with relatively few easily-measured traits (e.g. width and leaf area). Even though the individual morphological traits did not correlate with physiological processes, the whole morphological trait space correlated with the physiological trait space. Consequently, using size-related morphological traits resulted in a similar estimate of winning and losing species than using the full trait data set or eco-physiological traits only (Figure 4). Nevertheless, the interpretation of the mechanism behind species either “winning” or “losing” would be different depending on which traits are used for the predictions: a correlation with size-related traits would not reveal the role of water use strategy and decreasing soil moisture as potential drivers of community change.
Traits explained 10% of the variation in cover and 12% of the variation in frequency, less than for example 40% in Wright et al. (2010) (but more than e.g., 3% in Paine et al. 2015). Nevertheless, we think the variation explained by traits should be compared to that explained by species identity (25% and 12%, respectively). In our experiment, plant cover and frequency were likely changing due to several factors which did not relate to differences between species, and thus could not be captured by trait measurements (e.g., chance events, microclimate). Our trait measurements captured almost half (40%, Table S3) of the species-specific variation in cover change and all the variation in species-specific frequency change. Thus, in our study, traits were able to effectively capture the species-level differences in demographic rates, although better for one metric (frequency) than for the other (cover).
We observed within-species changes in most of the traits measured at multiple elevations, with the exact responses differing between species and traits. Contrary to previous studies (Pérez-Ramos et al.2019; Zhang et al. 2020) and to our expectations, trait plasticity was not important in explaining demographic change (but see Block et al. 2020). A potential explanation is that the magnitude of plastic responses to (even severe) climate change is small compared to trait difference between species (Aerts et al. 2007; Soudzilovskaia et al. 2013, Table S8). Our results highlight that, despite the potential importance of plasticity, species-level trait values can be useful for demographic predictions (Albert et al. 2011; Kichenin et al. 2013).
Although communities shifted towards species with more conservative water use under warming, the plastic response showed the opposite pattern: at warmer, lower -elevation sites, individual species showed lower δ18O (higher stomatal conductance) and lower δ13C (lower water use efficiency). These trends likely follow from the increasing atmospheric pressure and temperature with decreasing altitude (Körner et al. 1991). Higher CO2 partial pressure resulting in more constrained CO2 diffusion through stomata (Smith & Donahue 1991; Terashima et al. 1995; Wang et al. 2017), higher oxygen partial pressure resulting in lower carboxylation efficiency of Rubisco (Farquhar & Wong 1984), and higher temperature resulting in decreased viscosity of water and faster transport from roots to leaves (Roderick & Berry 2001), should all result in lower δ18O and δ13C values within species at lower elevations (Körneret al. 1988, 1991). This elevation-related shift to more inefficient water use might have resulted in a larger decrease in abundance in our experiment than would have taken place due to warming alone. Nevertheless, we think the mechanism identified (decreased water availability due to warming and the importance of water use traits) holds, even if the demographic effects of the water limitation might have been overestimated.
Contrary to our predictions, traits did not explain demographic changes caused by novel competitors. This is likely because the treatment had a relatively small influence on the resident community (Figures S16-S17). The treatment was set up to mimic the initial establishment of low-elevation species into alpine communities, and the four years of the experiment were likely not enough for them to significantly affect the alpine species.