Limitations of the protocol
The focus of the protocol on mountain roads provides excellent opportunities to disentangle the effects of climate and road construction on plant species and community redistributions. However, the protocol also has three important limitations (Figure 5).
First, the protocol excludes the most pristine environments that exist far from roads and at elevations above where roads reach, so does not monitor mountain biodiversity as a whole. As such, the protocol is a complement to the GLORIA protocol, which focuses on long-term climate change-related vegetation shifts on undisturbed mountain summits (Pauli et al., 2015). Nevertheless, one regional study has shown that, at least in northern Scandinavia, the effect of roads on mountain plant diversity disappears beyond 25 m from the roadside (Lembrechts et al., 2014), suggesting that the vegetation surveyed in the MIREN survey plot furthest from the roadside (Figure 3b; 50-100 m distance) may indeed be representative of regional biodiversity. Yet, using these data beyond the 100 m reach of the sample site could bring issues for some applications, such as spatial modelling, where extrapolations for locations away from the road will suffer from increased uncertainty (Kadmon et al., 2004). Coupled to this, the restriction of the protocol to mountain roads means that, depending on the heterogeneity of the landscape, not all habitat types are necessarily covered relative to their distribution in the ecosystem. Plot locations may be biased towards valleys and less steep terrain if road construction favours such areas. Additionally, while roads represent the most prominent dispersal pathway present in mountains, they are not the only one (e.g. rivers, mountain trails, powerline cuttings, cable cars; Foxcroft et al., 2019). However, the protocol could be easily adapted for other pathways (as done for trails (Liedtke et al., 2020) and rivers (Vorstenbosch et al., 2020)), and we suggest that this would be of particular interest in regions with sparse roads and/or where most of the common non-native species are wind or water dispersed.
Second, MIREN adopts a discrete temporal and spatial sampling approach. Specifically, since the protocol focuses on community dynamics and large-scale patterns it lacks the spatio-temporal resolution to monitor individual species and populations over time. The relatively low spatial sampling intensity (i.e. few plots for each elevational belt) and sometimes large distances between elevational increments (e.g. on average c. 75 m steps across current MIREN regions) can limit understanding of local processes, while also biasing sampling against rarer plant species or habitats. Furthermore, while repeated surveys facilitate investigation of species range dynamics under global change, the complete design does not explicitly consider dispersal dynamics (e.g. through seed rain or seed bank sampling, or seed tracking), instead assessing such dynamics indirectly through repeated snapshots of plant community composition.
Third, the standard protocol emphasises simplicity to be as inclusive as possible and to keep resource use to a minimum. The approach thus focuses chiefly on plant community composition and coarse estimates of species abundance (see Supporting Information S1). Other important variables such as biomass, functional traits, community 3D-structure, species interactions and other abiotic and biotic variables thus require additional sampling effort. For the same reason, the protocol is limited to vascular plants, excluding bryophytes and other taxonomic groups of potential interest.
Finally, the assumption that elevation can serve as a proxy for climate is of particular relevance here. Testing how the elevation gradient correlates with fine-grained climatic gradients requires validation using high-resolution climate data produced either using in-situmeasurements or downscaling of climate models (Lembrechts et al., 2019). We therefore recommend participants to include at least one add-on study that deploys temperature data loggers to allow linking of vegetation patterns with microclimatic gradients (Lembrechts et al., 2019) – although this would already add cost.