Short-rotation plantation forestry (SRF) is a novel and promising land-use system for the intensive production of energy and pulp wood in northern Europe (Weih 2004; Tullus et al. 2013). This type of forestry has great potential for mitigating climate change by replacing fossil fuels and sequestrating atmospheric CO2 (Paquette and Messier 2010; Haus et al. 2014; Lutter et al. 2016). Hybrid aspen (Populus tremula L. × P. tremuloides Michx.) is considered to be one of the most promising tree species for SRF and intensive pulp and biomass production in northern Europe (Tullus et al. 2012) owing to a faster growth rate compared with local European aspen (Yu et al. 2001a), the existence of long-term breeding programmes (Yu et al. 2001a; b; Stener and Karlsson 2004; Rytter and Stener 2003) and good cold resistance (Tullus et al. 2012). So far, hybrid aspen plantations in hemiboreal Estonia have been established with clonal plants that have been transferred southward, i.e. hybrids with one or both parents of northern origin (Tullus et al. 2007).

In the higher latitudes of the northern hemisphere, it is predicted that, besides the positive impact of increasing CO2 and temperature on the productivity of forests (Cole et al. 2010; Lindner et al. 2010), climate change will also extend the growth period for plants (Keenan et al. 2014), primarily through the earlier onset of springs (Jaagus 2006; Menzel et al. 2006; Schwartz et al. 2006), i.e. the date when daily mean air temperature rises above +5 °C (Jaagus and Ahas 2000; Jaagus 2006). Moreover, several recent studies also show clear trends of growth season extension in autumn, resulting in delayed leaf senescence (Garonna et al. 2014; Gill et al. 2015). Length of growing season is strongly controlled by climatic conditions (Cleland et al. 2007; Rohde et al. 2011), and tree populations display latitudinal adaptation (Pellis et al. 2004; Elferjani et al. 2016) as demonstrated by differences in phenological processes like growth initiation in spring (bud-burst) and growth cessation in autumn (bud-set) (Rohde et al. 2011). For example, northern populations are adapted to temperature-controlled delayed bud-burst in spring (Luquez et al. 2008) and photoperiod-sensitive bud-set in autumn (Way and Montgomery 2015) to avoid cold damage (Howe et al. 2003). Therefore, the warming climate and longer growing periods in autumn in northern Europe might advantage southern-origin genotypes that have adapted to a longer growing season in their local environment (Pellis et al. 2004; Way and Montgomery 2015). Northward transfer and selection of southern genotypes could improve future forest plantation biomass production and mitigate climate change more efficiently (Marris 2009; Way and Montgomery 2015).

In northern Europe, hybrid aspen progeny trials have so far mainly focused on tree growth potential (e.g. Yu et al. 2001a; Yu and Pulkkinen 2003; Stener and Karlsson 2004) and wood properties (Yu et al. 2001b; Rytter and Stener 2003). The few studies conducted on the phenological traits of hybrid aspen clones have concluded that genotypes selected in local environmental conditions with earlier bud-burst are more productive (Yu et al. 2001a; b; Jansons et al. 2014). However, given the great variability of phenological traits of Populus spp. and of climate change, hybrid aspen phenology is still understudied in progeny trials, bearing in mind that phenological traits like bud-burst, duration of leafy period, bud-set, etc. are important factors determining tree adaptability to climatic conditions, with consequent effects on productivity and survival (Cooke et al. 2012; Elferjani et al. 2016).

In order to test the suitability of hybrid aspen genotypes of different geographic origin for Estonian conditions, a progeny trial with completely randomised design was established, whereby phenological observations in spring and autumn were carried out during the third and sixth growing season since planting. The hypotheses were as follows: (1) northern transfer of hybrid aspen genotypes is superior to southern transfer because of better adaptability with an extended growing season—a consequence of climate change in northern regions; (2) genotypes with a longer leafy period also have a greater annual height increment.


Study area

A genetic trial testing a number of clones and full-sib families of hybrid aspen was established in south-eastern Estonia in the village of Agali (58° 17′ 10″ N, 27° 17′ 18″ E). Estonia is situated in the hemiboreal vegetation zone of the northern hemisphere within a transition zone of maritime to continental climate (Ahti et al. 1968). During the study period (2009–2014), the mean annual temperature was 6.1 °C and the mean annual precipitation was 676 mm according to data from the nearest weather station to the trial (Estonian Environment Agency). The mean annual temperature of the 2 years when the phenological survey was carried out was similar (6.7 °C in 2011 and 6.6 °C in 2014). The 1966–2010 long-term average temperature was 5.1 °C in this region (Tarand et al. 2013).

The trial was established in spring 2009 on flat agricultural land in the territory of the Järvselja Training and Experimental Forest Centre. The soil type of the study site according to IUSS Working Group WRB classification (2014) was fertile Eutric Glossic Retisol with a sandy loam texture in the uppermost 0–30-cm soil layer. The progeny trial was fenced to prevent animal damage. Mechanical weed control was carried out annually and neither fertiliser nor irrigation was applied after establishment.

Background of the studied hybrid aspen genotypes

The geographic origin of the studied hybrid aspen genotypes was determined by the latitudinal origin of their native European aspen (P. tremula) parent (Table 1). The exact origin of the P. tremuloides parent of these genotypes was often less precisely known but represented various states and regions along the border of the USA with Canada.

Table 1 The geographic origin of the parent trees of the studied hybrid aspen genotypes

The tested hybrid aspen progenies representing northward transfer were as follows: (1) seven clones with a P. tremula parent from Latvia (LV); (2) five clones with a P. tremula parent from Sweden (SE); (3) six different full-sib families with a P. tremula parent from Germany (DE). The trial includes also four hybrid aspen clones that represent southward transfer with a P. tremula parent from Finland (FI).

The comparison trial was established using 1-year-old container plants with 1 × 2 m spacing, planted in a completely randomised design. Each of the 22 treatments (clones or full-sib families) was represented by three replicate plots (16 trees per plot). After the fifth growing season, half of the trees (every second row) were systematically harvested to reduce within-stand competition, and the remaining trees were left growing with 2 × 2 m spacing.

Phenological observations

Phenological observations in the trial were carried out during the third (2011) and the sixth (2014) growing season by adapting the methodology described by Yu et al. (2001a) (Table 2). During both surveys, the onset (day-of-year) of spring phenological stages were recorded using a 3-day monitoring interval and that of autumn stages with a 4-day interval. Phenology was described for each replication separately and based on visual tree crown observations. For statistical analysis, the arithmetical mean day-of-year of onset of each stage was calculated for each treatment.

Table 2 Description of the stages of spring (S) and autumn (A) phenology that were recorded in the hybrid aspen progeny trial during the third and sixth growing seasons

Tree growth measurement

The height (m) of all trees growing in the trial was measured before growth initiation and after growth cessation in both study years, using an extendable measuring rod (at age 3) and a Vertex IV (Haglöf Inc.) height metre (at age 6). Current annual increment of height (CAIH, m) was calculated as the difference between initial and final height in the given year.

Statistical analysis

The effect of geographic origin (based on the origin of P. tremula parents) on phenology for each study year was tested separately with one-way analysis of variance (ANOVA). The effect of geographic origin on CAIH in each study year was tested separately with a mixed model (Eq. 1), where replication and individual treatment (clone or family) within origin were treated as random factors. The Tukey LSD test was applied to distinguish groups within each study year if a significant origin effect was detected.

$$ {y}_{ijk}=\mu +{\alpha}_i+{\beta}_j+{\gamma}_{k(j)}+{\varepsilon}_{ijk} $$

where μ is overall mean, α i is the fixed effect of geographic origin, β j is the random effect of treatment within origin and γ k(j) is random effect of replication within treatment and ε ijk is error term.

The relationships between time intervals of different phenological stages and CAIH in each study year were studied with Pearson’s linear correlation analysis.

The normality of the variables was tested with the Shapiro-Wilk test. Q-Q plots and residual distribution were applied to assess the normality of model residuals. A significance level of α = 0.05 was used to reject the null hypothesis after statistical tests. All statistical analyses were carried out using R Statistics software (R Core Team 2015).

Results and discussion

Clonal (or family-wise) mean bud-burst occurred between day 117 and day 131 in year 3 (Fig. 1) but no significant differences between geographic origins were found in the mean bud-burst date (Table 3). Also, geographic origin had no significant impact on the mean onset of other tree phenological stages or on intervals between spring and autumn phenological stages during the third growing season (Table 3).

Fig. 1
figure 1

The linear relationship between the start of the bud-burst and the number of days until leaves attained full size (p < 0.01**, p < 0.001***) in year 3 and in year 6

Table 3 Mean onset of phenological stages and the interval between spring and autumn phenological stages in third and sixth growing season of hybrid aspen with P. tremula parent from Germany (DE), Finland (FI), Latvia (LV) and Sweden (SE)

Significant impact of geographic origin on the mean onset of autumn phenological stages was revealed during the sixth growing season (Table 3). In year 6, southward transferred FI genotypes (60° 22′ N) started leaf defoliation processes and bud-set significantly earlier than northward-transferred genotypes (51° 16′ to 57° 31′ N). This is in accordance with the study by Pellis et al. (2004) reporting that poplar genotypes originating from lower latitudes shed their foliage later than genotypes from higher latitudes. Earlier leaf-fall in the genotypes of more northerly origin can be explained by their adaptation to longer photoperiod (Luquez et al. 2008; Way and Montgomery 2015). Shorter autumn days in the case of southern translocation cause them to start preparation for the dormancy season (Luquez et al. 2008; Saikkonen et al. 2012). At the same time, delayed abscission of northward-transferred genotypes could be caused by their adaptation to shorter photoperiod. This means that critical day-length triggering of autumnal senescence processes arrives later for them under northern conditions with longer photoperiod (Way and Montgomery 2015). Moreover, climate-change-induced temperature rise enables them to grow longer in the autumn than northern-origin genotypes without major risk of frost damage in autumn compared with northern-origin genotypes (Garonna et al. 2014; Gill et al. 2015).

Mean bud-burst started 6–10 days later in the northern-origin genotypes than in southern ones in year 6 but the differences were not significant (Table 3). This result is contrary to what has been observed in hybrid aspen genotypes grown in more northern regions (Yu et al. 2001a). The earlier onset of spring because of climate warming (Jaagus 2006; Menzel et al. 2006; Schwartz et al. 2006) suggests that northward-transferred trees adapt to initiate a growth period earlier than northern populations (Pellis et al. 2004; Luquez et al. 2008) as spring bud-burst is more controlled by temperature than photoperiod (Laube et al. 2014). The timing of bud-burst is considered to be an important characteristic that determines the growth rate of different hybrid aspen genotypes (Yu et al. 2001a; Jansons et al. 2014) and other fast-growing Salicaceae family species (Lennartsson and Ögren 2004; Weih 2009; Müller et al. 2013). Genotypes with early bud-burst are able to develop a higher number of leaves by capturing priority resources (Bergh et al. 2003; Müller et al. 2013) and therefore grow much faster than late flushing genotypes (Yu et al. 2001a; Bergh et al. 2003; Müller et al. 2013; Jansons et al. 2014).

Also, the mean day-of-year when trees obtained full-sized leaves at year 6 varied only by 4 days among all the studied genotypes, and this was also not significant (Table 3). However, genotypes with later bud-burst developed full-sized leaves much faster compared with early bud-burst genotypes (Fig. 1). Our results for hybrid aspen are in correspondence with the study by Yu et al. (2001a) about hybrid aspen in Finland and other studies about willow (Weih 2009) and silver birch (Possen et al. 2014) in northern European regions. Leaf unfolding and leaf development are controlled by temperature (Davi et al. 2011; Laube et al. 2014), and later-flushing genotypes are able to develop leaves faster than early flushing ones because they are adapted to avoid late-frost damage in northern growing conditions (Howe et al. 2003).

Earlier abscission of FI genotypes shortened their leafy period (S1–A3) compared with southern genotypes (DE, SE, LV) at year 6. The period from bud-burst to defoliation ranged from 177 to 188 days for southern-origin genotypes and was 27 days shorter for northern FI genotypes compared with the longest growing SE genotypes (Table 3). The intervals between several spring (growth initiation) and autumn (growth cessation) phenological processes were significantly and positively correlated with CAIH in both study years: in year 3, the strongest relation was observed with the interval between S3 to A1 (Fig. 2a); in year 6, the strongest relation was observed with the interval between S2 to A1 (Fig. 2b). That means that faster height growth of northward-transferred genotypes (SE and LV but not DE; Fig. 3) at both 3 and 6 years is explained, at least in part, by their longer leafy period compared with southward transferred FI genotypes (Fig. 2). It has already been found that the fast growth of hybrid aspen compared with the European aspen is partly explained by a longer growth period (from bud-burst to half of the leaves yellowed) (Yu et al. 2001a). However, correlation between the length of leafy period and height growth among different hybrid aspen genotypes after latitudinal transfer has not been reported previously.

Fig. 2
figure 2

Linear relationship between the ‘interval between spring and autumn phenological processes’ and current annual height increment (CAIH), where a the relationship with the interval that showed the strongest correlation during year 3, i.e. S3–A1, and b the relationship with the interval that showed the strongest correlation during year 6, i.e. S2–A1, are presented (p < 0.01**, p < 0.001***)

Fig. 3
figure 3

Comparison of mean current annual height increment (CAIH) and standard errors between the third and sixth growing seasons (p < 0.001***; ns not significant) for different P. tremula parent origins. Letters denote significant differences in the given year according to the Tukey LSD test

At the same time, wide latitudinal transfer should be approached cautiously (Marris 2009). For example, CAIH of DE genotypes representing the southern-most origin in our trial (51° 16′ to 52° 16′ N), did not differ significantly from the northern-origin FI genotypes (60° 22′ N) during both phenological surveys (Fig. 3) based on the Tukey LSD test. However, it must be remembered that DE genotypes represent full-sib families, meaning that they are less intensively selected than the clones from the other countries. Given the results after 6 years, northward-transferred SE (55° 53′ to 57° 31′ N) and LV (56° 06′ to 57°10′ N) genotypes can be regarded as most suitable for Estonian conditions as they showed significantly higher growth potential than southward transferred FI genotypes (Fig. 3).


A warming climate has already caused extension of the growing season in the hemiboreal region (a trend predicted to continue in future). Consequently, hybrid aspen genotypes with a southern origin responded well to a northward transfer, whereby they had a longer leafy period and greater annual height growth than southward transferred genotypes. The main differences among genotypes of different geographic origin were revealed in autumn phenology, when northern genotypes started earlier defoliation and bud-set. The effects of geographic transfer on performance of hybrid aspen genotypes should be studied in various environmental conditions for a longer time to draw final conclusions in terms of selecting the most suitable genotypes for commercial propagation and large-scale plantation establishment.