Skip to main content
NCBI home page
Scientific Reports logoLink to Scientific Reports
. 2025 Mar 28;15:10742. doi: 10.1038/s41598-025-94523-y

How microclimate influences the spring phenological responses to decreased snow cover in four tree species seedlings in a boreal forest

Yihan Cai 1,✉, Kiyoto Sawada 2, Kobayashi Makoto 3
PMCID: PMC11953451  PMID: 40155408

Abstract

Heat accumulation and spring freeze, both strongly influenced by snow cover, are key factors regulating the onset of spring phenology. In forest ecosystems, decreased snow cover due to climate change may differently impact heat accumulation and the occurrence of spring freezes between canopy gap and under the tree canopy, leading to varied phenological responses. In this study, we examined how spring phenology of tree seedling responds to decreased snow cover across microsites and explored whether these responses are species-specific. We conducted a manipulation experiment in a planted forest in northern Hokkaido, Japan, establishing five canopy gap and five under canopy areas, each with control and snow removal plot. Four dominant tree species were planted in each plot. Snow removal significantly advanced budburst and leaf-out in both microsites, with a more pronounced effect observed in the canopy gap. Moreover, snow removal significantly advanced the budburst and leaf-out of all four species in the canopy gap, whereas only Abies sachalinensis showed significantly earlier budburst and leaf-out in the under canopy. Overall, our study demonstrated that projected winter warming led to a greater advancement of spring phenology in tree seedlings in canopy gap compared to under the canopy, with species-specific responses.

Keywords: Winter climate change, Budburst, Leaf-out, Species-specific, Canopy gap

Subject terms: Plant ecology, Forest ecology

Introduction

Plant spring phenology is triggered by three major environmental cues, namely, heat accumulation (forcing), overwinter cold temperature accumulation (chilling), and photoperiod1, which is highly sensitive to global warming2,3. The general trend indicates that spring phenology typically advances under current rapid warming4,5, potentially leading to increased growth by an earlier onset and extending the length of the growing season6,7. However, the advancement of spring phenology does not necessarily occur in regions with decreased snow cover due to climate warming8,]9. Earlier snowmelt can advance the timing of leaf-out in tree seedlings10 because late-melting snowpacks, which keep near-surface temperatures lower than surrounding air temperatures, limit heat accumulation11,12. Conversely, early snowmelt can also delay leaf-out in tree seedlings13 due to increased spring freeze events, which occur when plants are in a dehardening phase and vulnerable to freezes14,15. Freeze damage can cancel the effect of an earlier onset of heat accumulation, diminishing or even reversing the advancement of spring phenology. Such varied spring phenological responses to decreased snow cover can be partly explained by the pattern of spring freeze occurrences under climate change when considered on a global scale16,17. Likewise, finer-scale microclimates, such as those influenced by forest structure that can outweigh the effects of macro-scale climate change on plant performance18, also potentially resulting in greater variation in the spring phenological response of tree seedlings.

Forests are known to support diverse microclimates in the understory due to the structural complexity of the tree canopy19. Given the significant differences in understory abiotic environment, such as solar radiation, air and soil temperature, and soil moisture20,21, forest microsites can be broadly categorized into under canopy and canopy gap (open areas). Snow cover typically appears later and melts earlier under the tree canopy than in canopy gaps due to the interception of snowfall by the canopy22, Roth and Nolin 2019). While the tree canopy acts as a buffer can moderate temperature fluctuations below the tree canopy to some extent23,24, the shorter snow cover duration and thinner snowpack would cause more spring freeze events below the tree canopy (Fig. S1). Whereas, decreased snow cover can also lead to earlier start of heat accumulation11. Because of the greater amount of direct solar radiation20,21, the heat accumulation would be faster in the canopy gap even though the snow cover disappeared later(Fig. 1) 25. In addition, decreased snow cover due to climate change would further increase the occurrence of spring freeze and accelerate the heat accumulation in both the under canopy and canopy gap, although the extent of these effects may vary between these microsites. As a result, decreased snow cover would lead to varied phenological responses across microsites, however, empirical research on how microclimates modulate spring phenological responses to decreased snow remains limited.

Fig. 1.

Fig. 1

Open in a new tab

Study area location and experimental design. Location of the study site (a). Position of the study areas based on an aerial photo of a 3-hectare area taken using a drone at a height of 150 m, with yellow and red rectangles representing the under canopy and canopy gap areas, respectively (b). Conditions of CC (under canopy control plot), CS (under canopy snow removal plot), GC (canopy gap control plot), and GS (canopy gap snow removal plot) based on photos taken in early spring. The map (a) was generated using QGIS (Version 3.1.0, QGIS Development Team, https://qgis.org/26) based on data from the National Land Numerical Information database (Ministry of Land, Infrastructure, Transport and Tourism, Japan, https://nlftp.mlit.go.jp/ksj-e/index.html).

Furthermore, species-specific phenological timing27 is additional factor to consider. The timing of spring phonology can vary significantly among species28 due to the trade-off between the risk of damage from spring freezes events and resource accessibility and growing season length27,29. For example, deciduous species generally exhibit earlier budburst and leaf-out compared to evergreen species30,31, which may be because evergreen species with overwintering leaves can conduct photosynthesis in early spring while deciduous species without overwintering leaves must leaf-out as early as possible to start photosynthesis31,32. Under changing climate conditions, deciduous species that initiate spring phenology earlier may be more susceptible to increased spring freeze events due to decreased snow cover. However, to the best of our knowledge, comprehensive studies have not been performed on the response of tree seedling spring phenology to decreased snow cover that consider both microclimate and species-specific flushing timings.

This study aimed to enhance our understanding of how microclimates within the forest ecosystem influence plant phenological responses to decreased snow cover and whether these responses are species-specific. We hypothesized that 1) decreased snow cover would advance spring phenology in canopy gap but result in no change or even a delay under the canopy and 2) late-flushing evergreen species would show clearer advancements in spring phenology in response to decreased snow cover compared to early-flushing deciduous species. To accomplish the goal, we conducted a manipulation experiment in a boreal forest in northern Hokkaido, Japan. We first established experimental sites in both the canopy gap, which were created by clear-cutting trees, and under the tree canopy. Since global warming has caused snow cover to appear later and disappear earlier, resulting in a shorter snow cover duration over the past few decades in this region33, we performed a snow removal experiment to delay snow cover appearance and advance snowmelt in both microsites. To examine how the flushing time of tree seedlings influences spring phenological responses to decreased snow, we selected two deciduous and two evergreen species that are dominant in this region and planted them in each manipulation plot. Since spring phenology can affect the growth and survival of tree seedlings by determining the start of the growing season and the likelihood of freeze damage34, understanding how the spring phenology of tree seedlings with different flushing timings in both canopy gap and under canopy areas can provide crucial insights into future tree regeneration.

Results

Snow cover under varying manipulation

The appearance of snow cover was delayed by approximately 10 days later in first experimental year and 14 days later in the second experimental year in the snow removal plot compared to the control plots in both the under canopy and canopy gap. Snow depth was significantly thicker in the canopy gap compared to the under canopy (P < 0.001) and thinker in the control plots compared to the snow removal plots (P < 0.05) in both first and second experimental years. Snow cover disappeared significantly earlier under the canopy than in the canopy gap (P < 0.05), and it also disappeared significantly earlier in the snow removal plots compared to the control plots (P < 0.001). Snowmelt occurred 12.6 ± 2.4 days earlier in 2023 and 9.6 ± 1.7 days earlier in 2024 in the snow removal plots compared to the control plots in the under canopy (Fig. 2). Similarly, snowmelt occurred 11.4 ± 3.8 days earlier in 2023 and 9.4 ± 1.9 days earlier in 2024 in the snow removal plots compared to the control plots in the canopy gap (Fig. 2).

Fig. 2.

Fig. 2

Open in a new tab

Snowmelt date (day of year) in in under canopy control plot (CC), under canopy snow removal plot (CS), canopy gap control plot (GC), and canopy gap snow removal plot (GS) in 2023 (a) and 2024 (b). The median (horizontal line), first and third quartiles (box edges), and the maximum and minimum vales were showed in the figure. Different letters indicate statistically significant differences among the manipulation plots.

The number of soft spring freeze occurred greater in snow removal plots compare to control plots, however, the significancy was only detected in the under canopy (P < 0.05; Fig. 3a), but not in the canopy gap (P = 0.67; Fig. 3a). The number of hard spring freeze also occurred greater in snow removal plots compared to control plots, however no statistical significance was detected (P > 0.05; Fig. 3b). Heat accumulation started earlier in the snow removal plots compared to the control plots in both the under canopy and the canopy gap (Fig. 4). And snow removal had the significant positive effect on heat accumulation (10-days summed) in both the under canopy and canopy gap, leading to greater heat accumulation in snow removal plots compared to the control plots since the late April (e.g. DOY 111: F1,17 = 11.94, P < 0.01; Fig. 4). Although heat accumulation (10-days summed) began earlier under canopy than in the canopy gap due to the earlier snowmelt, the heat accumulation in the canopy gap significantly surpassed that under canopy once the snow cover disappeared from the canopy gap (e.g. DOY 141: F1,18 = 13.428, P < 0.01; Fig. 4). Chilling accumulated significantly greater below tree canopy compared to canopy gap (F1,283 = 7.57, P < 0.01; Fig. S5).

Fig. 3.

Fig. 3

Open in a new tab

The number of soft (a) and hard (b) spring freezes in under canopy control plot (CC), under canopy reduced snow plot (GS), canopy gap control plot (GC), and canopy gap reduced snow plot (GS) in 2023 (a) and 2024 (b). The median (horizontal line), first and third quartiles (box edges), and the maximum and minimum vales were showed in the figure. Different letters indicate statistically significant differences among the manipulation plots.

Fig. 4.

Fig. 4

Open in a new tab

10-days summed heat accumulation (growing degree hours) in under canopy control plot (CC), under canopy snow removal plot (CS), canopy gap control plot (GC), and canopy gap snow removal plot (GS). The small figure is an enlargement of the day of year (DOY) period from 91 to 131.

Differences in spring phenology response to reduced snow between the canopy gap and under canopy areas

The absence of tree canopy had significant negative effect on budburst (F1,1381 = 34.98, P < 0.001) and leaf-out (F1,1381 = 43.79, P < 0.001), leading to the earlier budburst and leaf-out in the canopy gap compared to the under canopy (Fig. 5, Table 1). The snow removal also had significant negative effect on budburst (F1,1380 = 106.98, P < 0.001) and leaf-out (F1,1379 = 97.77, P < 0.001), leading to the earlier budburst and leaf-out in the snow removal plots compared to the control plots (Fig. 5, Table 1). In addition, significant interactions between the absence of tree canopy and snow removal were detected for the budburst (F1,1380 = 4.35, P < 0.05) and leaf-out (F1,1380 = 8.03, P < 0.01), indicating that snow removal advanced budburst and leaf-out more in canopy gap compared to the under canopy (Table 1, Fig. 5).

Fig. 5.

Fig. 5

Open in a new tab

Budburst and leaf-out date in under canopy control plot (CC), under canopy reduced snow plot (GS), canopy gap control plot (GC), and canopy gap reduced snow plot (GS) when considering all four species together in 2023 (a) and 2024 (b). The median (horizontal line), first and third quartiles (box edges), and the maximum and minimum vales were showed in the figure. Different letters indicate statistically significant differences among the manipulation plots.

Table 1.

Results of mixed-model ANOVA on timing of budburst and leaf-out including all species.

Tree canopy Snow removal Tree canopy Inline graphic snow removal
df F P df F P df F P
Budburst 1,1381 34.98  < 0.001 1,1380 106.98  < 0.001 1,1380 4.35  < 0.05
Leaf-out 1,1381 43.79  < 0.001 1,1379 97.77  < 0.001 1,1380 8.03  < 0.01
Open in a new tab

Bold value indicates the significance of the effects.

Species-specific responses

Significant differences in budburst and leaf-out dates were observed among the species (F1,1380 = 732.3, P < 0,001; Fig. 6). Specifically, A. pictum exhibited the earliest budburst and leaf-out, followed by Q. crispula, A. sachalinensis, and P. glehnii. The absence of tree canopy had significant negative effect on budburst and leaf-out of Q. crispula (leaf-out: F1,250 = 6.72, P < 0.05), A. sachalinensis (budburst: F1,397 = 50.59, P < 0.001; leaf-out: F1,396 = 44.01, P < 0.001), and P. glehnii (budburst: F1,350 = 120.50, P < 0.001; leaf-out: F1,351 = 82.27, P < 0.001), leading to the earlier budburst and leaf-out in canopy gap compared to under canopy, except for budburst of Q. crispula (F1,251 = 3.52, P = 0.062; Table 2). On the other hand, the absence of tree canopy had the significant positive effect on budburst of A. pictum (F1,366 = 2.03, P < 0.01; Table 2), leading to the later budburst in canopy gap compared to under canopy. Snow removal had a significant negative effect on the budburst (A. pictum: F1,367 = 11.56, P < 0.001; Q. crispula: F1,252 = 15.38, P < 0.01; A. sachalinensis: F1,397 = 92.11, P < 0.001; P. glehnii: F1,350 = 31.93, P < 0.001) and leaf-out (A. pictum: F1,367 = 10.45, P < 0.01; Q. crispula: F1,251 = 12.61, P < 0.01; A. sachalinensis: F1,397 = 69.02, P < 0.001; P. glehnii: F1,351 = 35.42, P < 0.001) of all four species (Table 2, Fig. 6). Additionally, significant interactions between the absence of tree canopy and snow removal were detected for the budburst and leaf-out of A. sachalinensis (budburst: F1,391 = 6.39, P < 0.05; leaf-out: F1,350 = 8.44, P < 0.01) and P. glehnii (budburst: F1,397 = 6.04 P < 0.01; leaf-out: F1,351 = 7.67, P < 0.01), indicating that snow removal advanced budburst and leaf-out more pronouncedly in canopy gap compared to the under canopy (Table 2). By conducting the multi-comparison, we found significant earlier budburst in the snow removal compared to the control plot for Q. crispula (P < 0.01), A. sachalinensis (P < 0.001), and P. glehnii (P < 0.001) in canopy gap, however only A. sachalinensis (P < 0.001) showed significant earlier budburst in snow removal plot also in the under canopy (Fig. 6a). Likewise, we found significant earlier leaf-out in the snow removal compared to the control plot for all four species (P < 0.05), however only A. sachalinensis (P < 0.001) showed significant earlier leaf-out in snow removal plot also in the under canopy (Fig. 6b).

Fig. 6.

Fig. 6

Open in a new tab

Budburst (a) and leaf-out (b) of A. pictum, Q. crispula, A. sahalinensis, and P. glehnii in under canopy control plot (CC), under canopy reduced snow plot (GS), canopy gap control plot (GC), and canopy gap reduced snow plot (GS) in 2023 (a) and 2024 (b). The median (horizontal line), first and third quartiles (box edges), and the maximum and minimum vales were showed in the figure. Different letters indicate statistically significant differences among the manipulation plots.

Table 2.

Results of mixed-model ANOVA on timing of budburst and leaf-out for each species.

Tree canopy Snow removal Tree canopy Inline graphic snow removal
df F P df F P df F P
Budburst A. pictum 1,366 4.62  < 0.01 1,367 11.56  < 0.001 1,364 0.0080 0.93
Q. crispula 1,251 3.52 0.062 1,252 15.38  < 0.01 1,249 0.56 0.46
A. sachalinensis 1.397 50.59  < 0.001 1,397 92.11  < 0.001 1,397 6.39  < 0.05
P. glehnii 1.350 120.50  < 0.001 1,350 31.93  < 0.001 1,350 8.44  < 0.01
Leaf-out A. pictum 1,366 2.03 0.15 1,367 10.45  < 0.01 1,364 0.34 0.56
Q. crispula 1,250 6.72  < 0.05 1,251 12.61  < 0.01 1,253 2.11 0.15
A. sachalinensis 1,396 44.01  < 0.001 1,397 69.02  < 0.001 1, 397 6.04  < 0.05
P. glehnii 1,351 82.27  < 0.001 1,351 35.42  < 0.001 1,351 7.67  < 0.01
Open in a new tab

Bold value indicates the significance of the effects.

Discussion

Our study clarified that the absence of tree canopy can advance budburst and leaf-out (Table 1, Fig. 2). Snow cover disappeared later, and the onset of heat accumulation occurred later in the canopy gap compared to the under canopy (Fig. 2, 3). However, once the snow cover had completely disappeared, heat accumulation in the canopy gap exceeded that in the under-canopy (Fig. 4), driven by higher air temperatures resulting from direct solar radiation (Fig. S421). This explained the earlier budburst and leaf-out in the canopy gap (Fig. 5) 25. As the tree canopy absorbs red light, seedlings growing beneath the canopy experience a lower red/far-red ratio compared to those in the canopy gap35. The reduction in red light may weaken the perception of day length36, potentially leading to a delayed onset of spring phenology below tree canopy (Fig. 5). Additionally, since insufficient chilling accumulation delays the onset of spring phenology by increasing the amount of forcing temperature required for leaf-out27,37, it can be expected that budburst and leaf-out would have occurred even earlier in the canopy gap if chilling accumulation had been greater (Fig. S5). Alternatively, it could simply be that, although chilling accumulation was lower in the canopy gap, it was still sufficient to meet the requirement for seedlings to become increasingly sensitive to warmth. Furthermore, we found that the decreased snow cover can advance budburst and leaf-out in both the under canopy and canopy gap, with significantly greater advancement in the canopy gap (Table 1, Fig. 5), consistent with our first hypothesis. Snow removal would advance the onset of heat accumulation and in both the canopy gap and the under canopy compared to the control. As more direct solar radiation reaches the canopy gap, while the tree canopy intercepts solar radiation19, we expected that snow removal would accelerate heat accumulation, with a more pronounced effect in the canopy gap. However, our results did not support this expectation, as the effect of snow removal on the heat accumulation was similar in two microsites (Tree canopy × Snow removal: P > 0.05; Fig. 4). Unlike heat accumulation, we found that the significantly greater number of spring freezes in the snow removal plots compared to the control plots in the under canopy, but not in the canopy gap (Fig. 3a). The increased spring freeze events elevate the risk of freeze damage to plants in the dehardening phase, when they are particularly vulnerable to freezing, and partially diminish the advancement of spring phenology15,27. This likely explains the more pronounced advancement of budburst and leaf-out in the snow removal plots in the canopy gap compared to the under canopy (Table 1). However, it should be noted that the contribution of both spring freezes and heat accumulation due to decreased snow cover can vary depending on the interannual variations in spring temperature dynamics and regional climate differences. Spring temperature dynamics vary from year to year. If the spring temperatures are extremely low, decreased snow cover would lead to more spring freezes. In this case, the effect of increased spring freezes may outweigh the effects of faster heat accumulation, resulting in delayed spring phenology. Furthermore, the risk of spring freezes varies globally and is expected to increase in Asia and Europe under climate change16,17. Therefore, in these regions, the effects of spring freezes may outweigh those of heat accumulation, leading to delayed spring phenology under decreased snow cover. Additionally, it is important to investigate how altered spring phenology may influence the performance of tree seedlings in the subsequent growing season. Advanced budburst and leaf-out can enhance plant survival and growth by increasing carbon assimilation in early spring7,38. However, increased spring freeze events may have carry-over negative effects on the survival and growth of tree seedlings27. As the next generation of the forest ecosystem, it is crucial to examine how altered spring phenology, resulting from decreased snow cover, may influence the recruitment of tree seedlings across different microsites.

In addition, we also observed species-specific spring phenological responses to decreased snow cover between the canopy gap and the under canopy (Table 2, Fig. 6). Contrary to our second hypothesis, early-flushing species A. pictum and Q. crispula also showed the significant advancement of leaf-out in response to decreased snow cover in the canopy gap, similar to late-flushing species A. sachalinensis and P. glehnii (Fig. 6b). However, A. sachalinensis was the only species that showed a significant advancement of budburst and leaf-out in the snow removal plot compared to the control plot in the under canopy (Fig. 6). The lack of significant responses in A. pictum and Q. crispula to decreased snow cover under the canopy is likely explained by the timing of spring freezes, which coincide with their dehardening and flushing phases, making their winter buds and/or leaves vulnerable to damage14, thus negating the effect of faster heat accumulation. On the other hand, A. sachalinensis, with its relatively late leaf-flushing timing, can avoid freeze damage to swollen winter buds, which likely explains its significant advancement of spring phenology in response to decreased snow cover, even under the canopy. Overall, our study demonstrated that spring phenological responses to decreased snow cover in two microsites can be species-specific, which partially explained by the species-specific phenological timings. However, it is important to note that snow removal did not have a significant effect on the leaf-out of A. pictum and Q. crispula based on the results of the Bayesian generalized linear model. This suggests that the two-year field experiment may have been insufficient to fully capture the patterns in phenological responses to decreased snow cover across microsites with varying abiotic conditions (e.g., thermal and light dynamics influencing phenological responses25,39 or to disentangle the underlying mechanisms driving these responses. Therefore, long-term observations that account for interannual variability in climatic conditions are needed to develop a more comprehensive understanding of how microclimates influence the phenological responses of understory vegetation to decreased snow cover, as well as the mechanisms underlying these responses.

Lastly, although we did not consider the effects of browsing damage on spring phenology, it may be a significant negative factor that delays the timing of spring phenology40. Reduced snow can make it easier for herbivores, such as deer and rodent, to locate tree seedlings41, increasing the risk of browsing on winter bud, which can delay budburst42,43. Given the range expansion and increased population of deer, along with their associated browsing damage in Japan—particularly in the Hokkaido region where we conducted our research44—a comprehensive study that incorporates the effects of herbivory is needed for more accurate predictions of seedling spring phenology in response to decreased snow.

Conclusion

Our study showed that in a boreal forest ecosystem, decreased snow cover can advance the budburst of tree seedlings in both under canopy and canopy gap with more pronounced advancement in the canopy gap. The varied influence of decreased snow cover on the occurrence of freeze events in spring between canopy gap and under canopy may explain the difference of spring phenological responses to decreased snow between these microsites. Moreover, while the decreased snow cover caused significant advancement on budburst and leaf-out in the canopy gap for all four species, only A. sachalinensis showed the significant earlier budburst and leaf-out in the snow removal plot compared to the control plot also in the under canopy. Overall, our study demonstrated the varying responses of spring phenology to decreased snow cover across microsites and species. However, further research is needed to confirm the trends observed in this study and to examine how shifts in spring phenology due to decreased snow affect tree seedling performance in the subsequent season.

Materials and methods

Study site and experimental design

We conducted a field experiment in a Picea glehnii plantation forest where the canopy consisted solely of P. glehnii (40 years old, mean diameter at breast height of 17.8 cm) in the Teshio Experimental Forest of Hokkaido University (45o03’N, 142o06’E, 66 m above sea level, Fig. 1a) in the northern part of Hokkaido, Japan. The study area situates is located within an ecotone between boreal and cool-temperate forest with drastic vegetation transitions45,46. The mean monthly temperature ranges from − 5.4 to 18.1 °C (January and July, respectively), and the annual precipitation is approximately 1000 mm47. The soil at this site was classified as Dystric Cambisol (Food and Agriculture Organization Soil Taxonomy).

In July 2022, canopy gap sites were artificially established by clearing canopy trees from five 20 × 10 m areas within the forest (Fig. 1b). These dimensions represent ordinary gaps in forest ecosystems48,49. Additionally, five under canopy areas (20 × 10 m) were established 10 m away from their corresponding canopy gap sites. At each site, two 5 × 5 m plots were established as control and reduced snow plots at 5 m intervals, which resulted in four distinct plots: under canopy control plot (CC, n = 5), under canopy reduced snow plot (CS, n = 5), canopy gap control plot (GC, n = 5), and canopy gap reduced snow plot (GS, n = 5) (Fig. 1c). In each plot, 20 tree seedlings from each of the four representative boreal mixed forest species (Acer pictum Thunb. Subsp. Mono (Maxim.), Quercus crispula var. crisp, Abies sachalinensis, and Picea glehnii) were planted in October 2022 (1,600 tree seedlings). A. pictum and Q. crispula were brought from OSAKA NURSERY, and A. sachalinensis and P. glehnii were brought from SASAKI SANGYO Co., Ltd., both located in the Hokkaido region. Seedlings with heights ranging from approximately 20 to 40 cm were selected for plantation in the field to prevent ontogenetic differences in phenological responses50.

Snow removal process

The analysis of average monthly (November) air temperature (Horokanai, Japan Meteorological Agency, https://www.data.jma.go.jp/stats/etrn/index.php) and snow cover appearance data (long-term observations of snowpack and snowfall depth in the Uryu Experimental Forest (UREF), northern Japan; https://db.cger.nies.go.jp/JaLTER/metacat/metacat/JaLTER-Hokkaido-Kita.2.5/jalter) from 1977 to 2012 in the northern part of Hokkaido demonstrated that a 1 °C rise in air temperature can delay snow cover forming for approximately 4.9 d (Fig. S2a). Additionally, an analysis of the average spring (March–May) air temperature data and date of snowmelt from 1996 to 2014 in the northern part of Hokkaido demonstrated that a 1 °C rise in air temperature may advance snowmelt for approximately 6.7 d (Fig. S2b). In this study, we conducted snow removal based on the Representative Concentration Pathways (RCP) 6.0 scenario, which projects a global mean surface temperature increase of 2.2 °C (with a likely range of 1.4 to 3.1 °C) between 2081 and 2100 (IPCC 2021). Accordingly, we aimed to delay snow cover formation by approximately 11 days (ranging from 7 to 14 days) and advance snowmelt by approximately 15 days (ranging from 9 to 21 days). During the onset of winter, snow was removed immediately after the appearance of snow cover to delay snow cover forming. We repeated the snow removal manipulation to ensure the 11 days (ranging from 7 to 14 days) delay in snow cover forming compared with that of the control plot. Consequently, we conducted snow removal twice in December 2022 and 2023 (Fig. S3). To prevent snow removal process-induced damage to the tree seedlings, a 5 cm layer of snow cover was left undisturbed. In March 2023 and 2024, we removed snow to half the depth of the control plot to advance snowmelt by approximately 15 days (ranging from 9 to 21 days) compared to that of the control plots. All snow-removal activities were manually performed using shovels.

Snow depth was measured weekly during the snow removal manipulation in December and after the snow removal manipulation in March. At other times, snow depth was measured monthly. To obtain precise data on snowmelt, TLC200 time-lapse cameras (Brinno, Taiwan) were used to record the condition of each plot daily until the snow cover completely disappeared.

Phenology observation

Following the Biologische Bundesanstalt, Bundessortenamt und CHemische Industrie (BBCH) scale51, we recorded budburst (BBCH stage 07, which is defined as the onset of sprouting or bud breaking) and leaf-out (BBCH stage 11, which is defined as first leaves unfolded) twice or thrice a week until the tree seedlings reached leaf-out in 2023 and 2024. We randomly selected 5 to 7 seedlings of each species from each manipulation plot at five sites to observe their phenology in 2023 (560 seedlings in the first stage). In 2024, we targeted all surviving seedlings for phenological observations (totally 1383 seedlings in the first stage). Tree seedlings that were browsed by mammals or had died before leaf-out were excluded. . Consequently, a total of 460 and 957 seedlings were included in the subsequent analyses for 2023 and 2024, respectively.

Calculation of chilling, forcing, and spring freezes

Air temperature and light data loggers (HOBO Pendant, Onset Computer, Bourne, MA, USA) were installed in the centre of each plot and set up in October 2022 to record data every 30 min. To measure the air temperature experienced by the tree seedlings, data loggers were located 20 cm above the ground level. Chilling was recorded as days with daily mean temperatures between 0 and 5 °C from 1st November until leaf-out52. Growing degree days, which is used as a measure of heat accumulation (Reaumur, 1735), was calculated by accumulating the average daily temperature above a base temperature of 5 °C according to previous studies53. The accumulation of growing degree days starts from 1st January54 and continues until leaf-out. The data did not accumulate when the mean daily temperature was below 5 °C. We started recording the spring freeze after snow removal manipulation in March until the end of May. Following budburst, -1.7 ℃ represented the threshold of soft freeze that can damage certain forest areas28 while -2.2 ℃ represented the threshold of hard freeze that can damage large forest areas55. Following these criteria, we counted the number of soft- and hard-spring freezes that occurred in each plot.

Statistics

We performed linear mixed-effects models (LMMs) to estimate the effects of tree canopy absence, snow removal, and their interaction on snow maximum depth, snowmelt timing, heat accumulation, chilling accumulation, and the number of soft and hard spring freeze (‘lme4’ package56). We also performed LMMs to estimate the effects of tree canopy absence, snow removal, and their interaction on budburst and leaf-out timing for each species. The models included site and year as random effects. As exceptions, species, site, and year were treated as random effects when analysing budburst and leaf-out with the full dataset and day of year and site were treated as random effects when analysis chilling accumulation. All the response variables were log-transformed to ensure the normality. Multi-comparisons were performed for snow maximum depth, snowmelt timing, heat accumulation, chilling days, the number of soft and hard spring freeze, budburst, and leaf-out. We initially calculated the estimated marginal means (EMMs) for the parameters (‘emmeans’ package, Lenth 202557), then, performed the multi-comparisons using the pairs function and applied Bonferroni correction to account for the increased risk of Type I errors associated with multiple tests. We also used a Bayesian generalized linear mixed model with a Gaussian family (Bayesian Regression Models using Stan, v.2.20.458) to predict budburst and leaf-out dates for each species. The model included microsite (under canopy, canopy gap), treatment (control, snow removal), and their interaction as fixed effects, with site and year as random effects. The posterior distribution was estimated based on a non-informative prior distribution using a flat uniform distribution across the parameter values of the Bayesian model using the Markov chain Monte Carlo (MCMC) method, which was implemented in RStudio using the package rstan, version 2.3159. Non-informative priors play a crucial role in objective Bayesian analysis (Shemyakin, 2014). For each model, we ran four chains, each with 10,000 iterations, among which 2,000 were warm-up iterations. We considered the effects to be statistically significant if the lower and upper 95% confidence intervals did not cross zero. The Bayesian generalised linear mixed models were all convergent (Rhat < 1.1). All statistical analyses were conducted using R software, v.4.1.060.

Supplementary Information

Supplementary Information. (1MB, docx)

Acknowledgements

We thank Dr Kentaro Takagi and the technical staffs of Teshio Experimental Forest, Hokkaido University, for their support on this study. We also thank Haruka Kobayashi, Akira Nakaya, Hina Haratani, Ruiqi Zeng, Hiroya Kataoka, Rihito Hosoda, Zimo Su, Moe Murakami, Tomoya Kawakami, and Shun Hizume for their help for the fieldwork.

Author contributions

YC and MK formulated the idea, YC developed methodology, YC and MK conducted fieldwork, YC and KS performed statistical analyses, YC wrote the manuscript, MK provided editorial advice.

Funding

This work was supported by JST SPRING (Grant Number JPMJSP2119), a Grant-in-Aid for Transformative Research Area (A) (Grant Number 21A403), and a Grant in Aid for Basic Research (B) (Grant Number 24K01791).

Data availability

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Declarations

Competing interests

The authors declare no competing interests.

Footnotes

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-94523-y.

References

  • 1.Körner, C. & Basler, D. Phenology under global warming. Science (80-. ). 327, 1461–1462 (2010). [DOI] [PubMed]
  • 2.Fu, Y. S. H. et al. Variation in leaf flushing date influences autumnal senescence and next year’s flushing date in two temperate tree species. Proc. Natl. Acad. Sci. U. S. A.111, 7355–7360 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Piao, S. et al. Plant phenology and global climate change: Current progresses and challenges. Glob. Chang. Biol.25, 1922–1940 (2019). [DOI] [PubMed] [Google Scholar]
  • 4.Cleland, E. E., Chuine, I., Menzel, A., Mooney, H. A. & Schwartz, M. D. Shifting plant phenology in response to global change. Trend. Ecol. Evol.22, 357–365 (2007). [DOI] [PubMed] [Google Scholar]
  • 5.Vitasse, Y. et al. Leaf phenology sensitivity to temperature in European trees: Do within-species populations exhibit similar responses?. Agric. For. Meteorol.149, 735–744 (2009). [Google Scholar]
  • 6.Augspurger, C. K. Early spring leaf out enhances growth and survival of saplings in a temperate deciduous forest. Oecologia156, 281–286 (2008). [DOI] [PubMed] [Google Scholar]
  • 7.Lee, B. R., Yancy, A. J. & Heberling, J. M. Phenological Escape and Its Importance for Understory Plant Species in Temperate Forests. Int. J. Plant Sci.185, 321–333 (2024). [Google Scholar]
  • 8.IPCC. Climate Change 2021: The Physical Science Basis. Masson-Delmotte, V. et al. (eds.) Working Group I Contribution to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change. 2391 (Cambridge University Press, 2021).
  • 9.Gottlieb, A. R. & Mankin, J. S. Evidence of human influence on Northern Hemisphere snow loss. Nature625, 293–300 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Livensperger, C. et al. Earlier snowmelt and warming lead to earlier but not necessarily more plant growth. AoB Plants8, 1–15 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Kelsey, K. C. et al. Winter snow and spring temperature have differential effects on vegetation phenology and productivity across Arctic plant communities. Glob. Chang. Biol.27, 1572–1586 (2021). [DOI] [PubMed] [Google Scholar]
  • 12.Tyystjärvi, V. et al. Variability and drivers of winter near-surface temperatures over boreal and tundra landscapes. Cryosphere18, 403–423 (2024). [Google Scholar]
  • 13.Sanders-DeMott, R., McNellis, R., Jabouri, M. & Templer, P. H. Snow depth, soil temperature and plant–herbivore interactions mediate plant response to climate change. J. Ecol.106, 1508–1519 (2018). [Google Scholar]
  • 14.Chamberlain, C. J., Cook, B. I., García de Cortázar-Atauri, I. & Wolkovich, E. M. Rethinking false spring risk. Glob Chang. Biol.25(2209), 2220 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Lenz, A., Hoch, G., Vitasse, Y. & Körner, C. European deciduous trees exhibit similar safety margins against damage by spring freeze events along elevational gradients. New Phytol.200, 1166–1175 (2013). [DOI] [PubMed] [Google Scholar]
  • 16.Liu, Q. et al. Extension of the growing season increases vegetation exposure to frost. Nat. Commun.10.1038/s41467-017-02690-y (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Zohner, C. M. et al. Late-spring frost risk between 1959 and 2017 decreased in North America but increased in Europe and Asia. Proc. Natl. Acad. Sci. U. S. A.117, 12192 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Zellweger, F. et al. Response to comment on forest microclimate dynamics drive plant responses to warming. Science370, 772–775 (2020). [DOI] [PubMed] [Google Scholar]
  • 19.De Frenne, P. et al. Forest microclimates and climate change: Importance, drivers and future research agenda. Glob. Chang. Biol.27, 2279–2297 (2021). [DOI] [PubMed] [Google Scholar]
  • 20.Malcolm, D. C., Mason, W. L. & Clarke, G. C. The transformation of conifer forests in Britain - Regeneration, gap size and silvicultural systems. For. Ecol. Manage.151, 7–23 (2001). [Google Scholar]
  • 21.Muscolo, A., Bagnato, S., Sidari, M. & Mercurio, R. A review of the roles of forest canopy gaps. J. For. Res.25, 725–736 (2014). [Google Scholar]
  • 22.Varhola, A., Coops, N. C., Weiler, M. & Moore, R. D. Forest canopy effects on snow accumulation and ablation: An integrative review of empirical results. J. Hydrol.392, 219–233 (2010). [Google Scholar]
  • 23.Chen, S. et al. Macroclimate and canopy characteristics regulate forest understory microclimatic temperature offsets across China. Glob. Ecol. Biogeogr.33, 1–10 (2024). [Google Scholar]
  • 24.De Frenne, P. et al. Global buffering of temperatures under forest canopies. Nat. Ecol. Evol.3, 744–749 (2019). [DOI] [PubMed] [Google Scholar]
  • 25.Vitasse, Y. et al. Impact of microclimatic conditions and resource availability on spring and autumn phenology of temperate tree seedlings. New Phytol.232, 537–550 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.QGIS Development Team QGIS geographic information system. Open Sour. Geospat. Found. Proj. https://qgis.org/ (2023).
  • 27.Chamberlain, C. J. & Wolkovich, E. M. Late spring freezes coupled with warming winters alter temperate tree phenology and growth. New Phytol.231, 987–995 (2021). [DOI] [PubMed] [Google Scholar]
  • 28.Augspurger, C. K. Reconstructing patterns of temperature, phenology, and frost damage over 124 years: Spring damage risk is increasing. Ecology94, 41–50 (2013). [DOI] [PubMed] [Google Scholar]
  • 29.Basler, D. & Körner, C. Photoperiod sensitivity of bud burst in 14 temperate forest tree species. Agric. For. Meteorol.165, 73–81 (2012). [Google Scholar]
  • 30.Osada, N. Relationships between the timing of budburst, plant traits, and distribution of 24 coexisting woody species in a warm-temperate forest in Japan. Am. J. Bot.104, 550–558 (2017). [DOI] [PubMed] [Google Scholar]
  • 31.Panchen, Z. A. et al. Leaf out times of temperate woody plants are related to phylogeny, deciduousness, growth habit and wood anatomy. New Phytol.203, 1208–1219 (2014). [DOI] [PubMed] [Google Scholar]
  • 32.Davi, H. et al. Diversity of leaf unfolding dynamics among tree species: New insights from a study along an altitudinal gradient. Agric. For. Meteorol.151, 1504–1513 (2011). [Google Scholar]
  • 33.Makoto, K., Templer, P. H., Katayama, A., Seki, O. & Takagi, K. Early snowmelt by an extreme warming event affects understory more than overstory trees in Japanese temperate forests. Ecosphere13, 1–12 (2022).38988721 [Google Scholar]
  • 34.Forrest, J. & Miller-Rushing, A. J. Toward a synthetic understanding of the role of phenology in ecology and evolution. Philos. Trans. R. Soc. B. Biol. Sci.365(3101), 3112 (2010). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Turnbull, M. H. & Yates, D. J. Seasonal variation in the red/far-red ratio and photon flux density in an Australian sub-tropical rainforest. Agric. For. Meteorol.64, 111–127 (1993). [Google Scholar]
  • 36.Zohner, C. M. & Renner, S. S. Perception of photoperiod in individual buds of mature trees regulates leaf-out. New Phytol.208, 1023–1030 (2015). [DOI] [PubMed] [Google Scholar]
  • 37.Beil, I., Kreyling, J., Meyer, C., Lemcke, N. & Malyshev, A. V. Late to bed, late to rise—Warmer autumn temperatures delay spring phenology by delaying dormancy. Glob. Chang. Biol.27, 5806–5817 (2021). [DOI] [PubMed] [Google Scholar]
  • 38.Lee, B. R. & Ibáñez, I. Spring phenological escape is critical for the survival of temperate tree seedlings. Funct. Ecol.35, 1848–1861 (2021). [Google Scholar]
  • 38.Ward, S. E., Schulze, M. & Roy, B. A long-term perspective on microclimate and spring plant phenology in the Western Cascades. Ecosphere10.1002/ecs2.2451 (2018). [Google Scholar]
  • 40.Guiden, P. W., Connolly, B. M. & Orrock, J. L. Seedling responses to decreased snow depend on canopy composition and small-mammal herbivore presence. Ecography (Cop.)42, 780–790 (2019). [Google Scholar]
  • 40.Giroux, M. A., Dussault, C., Tremblay, J. P. & Côté, S. D. Winter severity modulates the benefits of using a habitat temporally uncoupled from browsing. Ecosphere10.1002/ecs2.1432 (2016). [Google Scholar]
  • 42.Carroll, A. L. & Quiring, D. T. Herbivory modifies conifer phenology: Induced amelioration by a specialist folivore. Oecologia136, 88–95 (2003). [DOI] [PubMed] [Google Scholar]
  • 43.Quiring, D. T. & McKinnon, M. L. Why does early-season herbivory affect subsequent budburst?. Ecology80, 1724–1735 (1999). [Google Scholar]
  • 44.Takatsuki, S. Effects of sika deer on vegetation in Japan: A review. Biol. Conserv.142, 1922–1929 (2009). [Google Scholar]
  • 45.Hiura, T., Go, S. & Iijima, H. Long-term forest dynamics in response to climate change in northern mixed forests in Japan: A 38-year individual-based approach. For. Ecol. Manage.449, 117469 (2019). [Google Scholar]
  • 46.Sato, H., Shibuya, M. & Hiura, T. Reconstructing spatiotemporal dynamics of mixed conifer and broad-leaved forests with a spatially explicit individual-based dynamic vegetation model. Ecol. Res.38, 465–478 (2023). [Google Scholar]
  • 47.Naka, M. et al. Long-term Consequences on Soil Fungal Community Structure: Monoculture Planting and Natural Regeneration. Environ. Manage.10.1007/s00267-023-01917-7 (2023). [DOI] [PubMed] [Google Scholar]
  • 48.Schliemann, S. A. & Bockheim, J. G. Methods for studying treefall gaps: A review. For. Ecol. Manage.261, 1143–1151 (2011). [Google Scholar]
  • 49.Wu, C. D. et al. Gap shape classification using landscape indices and multivariate statistics. Sci. Rep.6, 1–10 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 50.Vitasse, Y. Ontogenic changes rather than difference in temperature cause understory trees to leaf out earlier. New Phytol.198, 149–155 (2013). [DOI] [PubMed] [Google Scholar]
  • 51.Meier, U. German federal biological research centre for agriculture and forestry - growth stages of mono- and dicotyledonous plants. BBCH Monogr.115http://pub.jki.bund.de/index.php/BBCH/article/view/515 (2001).
  • 52.Peaucelle, M. et al. Spatial variance of spring phenology in temperate deciduous forests is constrained by background climatic conditions. Nat. Commun.10, 1–10 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 53.Clark, J. S., Salk, C., Melillo, J. & Mohan, J. Tree phenology responses to winter chilling, spring warming, at north and south range limits. Funct. Ecol.28, 1344–1355 (2014). [Google Scholar]
  • 54.Murray, M. B., Cannell, M. G. R. & Smith, R. I. Date of budburst of fifteen tree species in Britain following climatic warming. J. Appl. Ecol.26, 693. 10.2307/2404093 (1989). [Google Scholar]
  • 55.Schwartz, M. D., Reed, B. C. & White, M. A. Assesing satellite-derived start-of-season measures in the conterminous USA. Int. J. Climatol.22, 1793–1805 (2002). [Google Scholar]
  • 56.Bates, D., Maechler, M., Bolker, B. & Walker, S. Fitting linear mixed-effects models using lme4. J. Stat. Softw.67, 1–48 (2015). [Google Scholar]
  • 57.Lenth, R. emmeans: Estimated Marginal Means, aka Least-Squares Means. R package version 1.10.7-100003.https://rvlenth.github.io/emmeans/ (2025).
  • 58.Bürkner, P. brms: An R package for Bayesian multilevel models using Stan. J. Stat. Softw.80, 1–28 (2017). [Google Scholar]
  • 59.Stan Development Team. RStan: The R Interface to Stan. R package version 2.32.6 (2024).
  • 60.R Core Team (2021) R: A language and environment for statistical computing. Vienna, Austria.: R. Found. Stat. Comput.

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Information. (1MB, docx)

Data Availability Statement

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Articles from Scientific Reports are provided here courtesy of Nature Publishing Group

RESOURCES