Discussion

While many species ranges are on average shifting in the direction predicted by SDMs, there is large variation among species that is difficult to explain. Microhabitat variability is increasingly recognized as influencing population dynamics across species ranges (Oldfather & Ackerly, 2019) and may be a key factor in understanding and predicting range shifts (Lembrechts, Nijs, et al., 2019; Maclean & Early, 2023; Stickley & Fraterrigo, 2023). However, how microhabitat suitability, and variability therein, is distributed along species ranges is vastly understudied, impeding an understanding of how microhabitat can facilitate or constrain range shifts. We found that microhabitat is highly variable across species ranges and is largely decoupled from the macro-scale at which SDMs are typically constructed. We further found variable, albeit weak, effects of microhabitat for different species and life stages, suggesting that the drivers of establishment are complex and difficult to detect. Our microhabitat suitability predictions show microhabitat will either constrain range expansions, or have no effect, for most species, with the range shifts of just one species (L. latifolius ) likely facilitated by microhabitat.

Complex microhabitat patterns across elevation gradients

Surprisingly, most of our microhabitat variables, even ones related to climate (summer soil moisture, summer soil temperature, summer plant height temperature) did not follow an elevational pattern representative of the macroclimate (Fig. S8) commonly used in SDMs. This is consistent with many studies finding that microhabitat is often decoupled from elevation and regional climate (Ford et al., 2013; Lembrechts, Lenoir, et al., 2019). We further found that none of the soil composition parameters we measured (fungus:bacteria, carbon:nitrogen, water holding capacity) followed an elevational pattern, even though other alpine studies have found soil microbial community (Hiiesalu et al., 2023) and soil carbon:nitrogen (Weintraub et al., 2016) to covary with elevation.
We further found that only canopy openness, winter minimum soil temperature, and spring snow days covary with elevation. Despite higher elevation areas characterized by more open forest canopies, the high variability at each site (Fig. 2) is in line with the large body of work showing that forest canopies can provide climate refugia (De Frenne et al., 2019; Haesen et al., 2021). As species light requirements can be variable, light availability and canopy cover can shape species ranges (Muñoz Mazon et al., 2023) and therefore can be an important component of range shifts (Tourville et al., 2022).

Low and variable effects of microhabitat on establishment

Our results agree with other studies, which find overall low recruitment in seed addition sites (Clark et al., 2007) and decreasing fitness of transplants beyond the range (Lee-Yaw et al., 2016; Stanton-Geddes et al., 2012). Plant establishment beyond current range edges is necessary for plant species to shift their ranges upward, however this is complicated by plants usually being recruitment limited. This means that they are limited by some aspect of their environment (e.g., microhabitat unsuitability, seed predation) and not by dispersal (Clark et al., 2007). While we did not control for seed predation, herbivory, or fungal infections, we captured mortality with our methods. Even though seed predation rates decline in higher elevation areas (Hargreaves et al., 2019), we still removed fleshy fruits around seeds to deter predation. We also did not collect data to test for effects of community composition on establishment, but with low recruitment rates, plants in our system would still be considered recruitment limited.
We did not find that establishment responses to any particular microhabitat were restricted to certain growth forms, with species-specific relationships to microhabitat. This is not surprising, as species have unique environmental requirements (Table S1) and species characteristics even impact predictive power in SDMs (Guisan et al., 2007). We also did not find differences in effects of the microhabitat aspects that are expected to be directly versus indirectly affected by climate change. Interestingly, we found that some parameters, such as winter soil temperature, increased likelihood of recruitment but these effects switched to decreasing likelihood of seedling survival. One exception was spring days with snow, which generally had positive effects, and this is similar to findings by Davis & Gedalof (2018), who found positive effects of winter snow cover on recruitment. The importance of microbial community in plant dynamics is also increasingly acknowledged (Castro et al., 2022), although soil fungus:bacteria ratio was not selected more than other variables in our system. Microbial communities could even be an important factor in determining species range shifts, such as favorable soil microbial composition mediating climate tolerance to promote tree seedling survival (Allsup et al., 2023).

Predicting microhabitat suitability

We found that microhabitat suitability beyond the leading range edge was species- and life stage-specific, with most species showing either decreased suitability or no pattern with elevation. We posit that the role that microhabitat may play in facilitating or constraining range shifts for any given species is closely tied to how microhabitat itself varies with elevation. For example, if a favorable microhabitat parameter for establishment decreases with elevation, establishment will overall not be favored and a range shift could be constrained. We found this in our system, with increasing likelihood of L. latifoliusrecruit counts and seedling survival with warmer winter minimum soil temperatures, but soil temperature decreases with elevation. If a favorable microhabitat parameter increases with elevation, however, this could lead to a facilitated range shift. This is the case for canopy openness, which increases with elevation and also increases likelihood of L. latifolius establishment. Since microhabitat suitability increases beyond the range for this species, these cases highlight the complex ways in which microhabitat parameters may interact to mediate range shifts. The alternative scenarios, where an unfavorable microhabitat parameter either declines with elevation to inversely favor establishment or increases with elevation to inhibit establishment, can also facilitate or inhibit range shifts, respectively.
Macroclimatic parameters vary predictably with elevation (Fig. S8) in our study system, however only a third of our microhabitat variables show an elevational pattern. Since species distributions can be driven by microhabitat, understanding how microhabitat parameters vary across species’ ranges can give insights into how microhabitat may facilitate or inhibit range expansion (Lembrechts et al., 2017; Tourville et al., 2022; Kemppinen et al., 2023). Incorporating microhabitat suitability into ecological studies is not new, including quantifying habitat suitability in the field (Jabis & Ayers, 2014), with remote sensing (Falco et al., 2019), and at the leading range edge (Mamet & Kershaw, 2013). Transplant studies often find decreasing macroclimate suitability beyond the range (Lee-Yaw et al., 2016), but to our knowledge no studies exist that assess microhabitat suitability beyond species range edges. Despite increasing work showing the necessity of including microhabitats in SDMs (Lembrechts, Nijs, et al., 2019), the vast majority of SDMs still focus on macroclimatic variables and this may lead to the mismatches between predicted and observed range shifts, or lack thereof.
In our system, we found that suitable microhabitat is unchanged or reduced beyond the leading edge for many species, yet we also find that macroclimatic conditions enable an upwards shift in species’ recruitment optima at the community level (unpublished data ). However, this pattern is only evident at the community level, with high species-level variability and no effect of canopy cover. This matches the large variability, and low explanatory power, seen in our microhabitat suitability predictions, and suggests that species can find pockets of suitable microhabitat to recruit beyond their macroclimatic cold range edge. Microhabitat suitability may be just high enough to allow for successful establishment, but is overall very low and the complicated ways in which microhabitat acts to affect species establishment makes it difficult to detect patterns. We also found continued species recruitment at the warm edge of the range (unpublished data ) and this buffering of contractions at the warm edge could be due to microhabitat refugia (De Lombaerde et al., 2022) created by the range of microhabitat found at the lower elevation sites (Fig. 2)

Limitations

Our experimental design is not without limitations. Seed provenance might cause different recruitment responses across microhabitat gradients, however variation in germination rates was not biologically meaningful (Fig. S1). Since we sowed all seeds in the same year, we also cannot test if we sowed in favorable versus unfavorable years and aimed to capture a large germination window by recording recruitment over three years. Finally, we collected some of our data on species responses and microhabitat measurements asynchronously and therefore emphasize that only differences between sites, not absolute values, should be interpreted for their effects. Finally, we strongly urge future studies to incorporate increased replication at each microhabitat value for higher statistical power in explaining results.

Conclusion

To our knowledge, our work is one of the first studies to measure key demographic life stages of a large group of species along a microhabitat gradient both within and beyond their current range limit. As such, this work yields a more comprehensive understanding of the mechanisms that set species range limits and indicates ways in which species shift their distributions with climate change. Our results suggest that complex ways in which microhabitat parameters influence early life stages are complex and difficult to detect, which complicates range shift predictions. A greater understanding of the role of macrohabitat in shaping species’ distribution limits will ultimately improve predictions of how species distributions will shift with climate change. We emphasize that predictions accounting for complex microhabitat drivers are necessary to create tailored conservation and management decisions in order to mitigate ongoing biodiversity loss.