• Adaptive traits in Picea abies (Norway spruce) progenies are influenced by the maternal temperatures during seed production. Here, we have extended these studies by testing the effects of maternal photoperiod and temperature on phenology and frost hardiness on progenies.
• Using eight phytotron rooms, seeds from three unrelated crosses were made in an environmental 2 × 2 factorial combination of long and short days and high and low temperatures. The progenies were then forced to cease growth rapidly at the end of the first growing season.
• An interactive memory effect was expressed the second growth season. Progenies from high temperature and short days, and from low temperatures and long days, started growth later in spring, ceased shoot growth later in summer, grew taller and were less frost hardy in the autumn than their full siblings from low temperatures and short days, and from high temperatures and long days.
• Norway spruce has developed a memory mechanism, regulating adaptive plasticity by photoperiod and temperature, which could counteract harmful effects of a rapidly changing climate.
In general, interactive effects between temperature and light on growth rhythm, hardiness and dormancy are complex. The climate will probably change rapidly in the future and knowledge about such interactions will be of prime importance for predicting the abilities of trees to adapt to changing climatic conditions. Extreme temperature is a selecting agent causing population differentiation along latitudinal and altitudinal gradients for adaptive traits such as bud set and cold acclimatization in the autumn and (Bigras & Colombo, 2001; Saxe et al., 2001). We observe, accordingly, that the Norwegian provenances of Norway spruce (Picea abies L. Karst.) form distinct and strong clinal variation patterns in both bud set (Kohmann, 1996) and frost hardiness in the autumn (Dæhlen et al., 1995).
Temperature during maternal reproduction affects adaptive traits in progenies of Norway spruce. Seed production in a cold environment advances bud set and cold acclimatization in the autumn and dehardening and flushing in spring, whereas a warm reproductive environment delays timing of these progeny traits (see reviews by Skrøppa & Johnsen, 2000; Johnsen & Skrøppa, 2001). This maternal environmental effect (memory) lasts many years in the filial generation. We have found that progenies are affected by the temperature during zygotic embryogenesis and seed maturation (Johnsen et al., 2005). Progeny performance is strongly associated with the sum of heat experienced by the maternal parent from pro-embryo to mature seeds, and we have hypothesized that temperature in this period regulates an ‘epigenetic memory’ involving differential expression of genes with putative functions in bud phenology, cold acclimatization and embryogenesis in Norway spruce. This phenomenon may improve the evolutionary sustainability under global warming conditions at northern latitudes.
Norway spruce trees growing in extreme northern autochthonic stands at the northernmost range of the species seldom produce viable seeds. Two consecutive years with warm and sunny weather are needed, in the first year floral buds are induced, and the second year sexual reproduction takes place (Owens & Blake, 1985). Plant producers in Norway still depend on a very good seed year in 1970 for their productions of planting stock used for reforestation purposes in our northernmost areas (66–67° N). These seeds give rise to seedlings with a very early growth cessation and fast development of frost hardiness in the autumn (Sandvik, 1980; Dæhlen et al., 1995; Kohmann, 1996), despite the fact that temperature during embryogenesis and seed maturation was higher than normal in 1970. This has led us to hypothesize that long days in northern areas counteract the effect of high temperature during embryogenesis and seed maturation, and we have preliminary data that progenies from seeds produced under long days are hardier than their full sibs from seeds produced under short days (Ø. Johnsen et al., unpublished).
Maternal photoperiod affects seed germination percentages and germination speed in Arabidopsis thaliana (Munir et al., 2001), but we are not aware of any published experimental data with conifers where photoperiod treatments given during maternal sexual reproduction have been shown to affect adaptive performance of progenies in subsequent years. We know that the paternal photoperiod during microsporogenesis and release of pollen does not affect progeny performance in Norway spruce (Johnsen et al., 1996). In the present paper we report results from a study where we tested the memory effect of photoperiod and temperature during sexual reproduction on 2-yr progeny performance in climate chamber, glasshouse and outdoors under nursery conditions. We made seeds from three genetically independent crosses in four contrasting environments; a 2 × 2 factorial combination of short and long days at high and low temperature in eight phytotron rooms. Seeds from these crosses gave rise to progenies, which were made phenotypically uniform by imposing a strong inductive treatment during growth cessation and bud set at the end of the first growing season. This was done to test if it was possible to erase or reduce the expression of memory effects in the progenies the second growing season compared with what we normally observe in Norway spruce (Johnsen, 1989a,b; Skrøppa, 1994; Edvardsen et al., 1996).
Materials and Methods
Origin of parents, floral induction and crosses performed
The clonal parents, propagated as grafts, originated from the south-eastern part of Norway (61° N, 100–300 m above sea level). The potted grafts were grown in Biri Nursery and Seed Improvement Centre, Biri, Norway, as described earlier (Johnsen et al., 1994; Owens et al., 2001). Ramets of each clone were 5–7 yr from grafting, 2.5 m tall at time of flower induction treatments, and were grown in 50-l pots. Male and female cone buds were induced on grafts of separate clones the year before seeds were produced (2000), as described in Johnsen et al. (1994) and Owens et al. (2001). Male cones were forced in the glasshouse at Biri nursery during April 2001, as described by Owens et al. (2001). Pollen was extracted, cleaned and dried before approx. 7–8 ml samples were filled into 15 ml glass vials and sealed. Thus, fresh and viable pollen for the crosses were ready-made before female cones were receptive. Female grafts were transferred to the Phytotron at the University of Oslo, Norway, well before female cones were open. They were then moved to eight phytotron rooms on May 7, 2001, using two ramets per clone per room. Female cones were isolated with paper bags, and pollen was sprayed into the bag using a syringe tightly connected to the sealed vial containing the pollen. Pollination was performed once a day in the period from 18 to 24 May in the high temperature treatment, and from 25 May to 2 June in the low temperature treatment.
Maternal treatment during sexual reproduction
Three genetically unrelated crosses were made in each of the eight phytotron rooms. Four rooms were programmed with northern (66° N) daylengths (long days; LD) and four different rooms with southern (52° N) daylengths (short days; SD; see Fig. 1). The SD light intensity was 300 µmol m−2 s−1, but in LD a difference was programmed between high and low light intensity (300 and 100 µmol m−2 s−1). Low light intensity was given during mornings and evenings (dusk and dawn effects), varying from 1.5 to 4 h before and after the high light intensity periods; the longer the total light period the longer the low light intensity periods. High- and low-temperature (HT and LT) regimes were programmed to rooms with long days and short days in a factorial way. The combinations HT-SD, HT-LD, LT-SD and LT-LD were distributed to two rooms each. Day and night temperatures were programmed to coincide with periods of light and darkness, but in a way that avoided confounding temperature and daylength. Night temperatures were programmed to be 2–6°C lower than day temperatures at LT and 3–6°C lower at HT, depending on the programmed season. The heat sum accumulation (using 5°C as a threshold; Sarvas, 1968) is shown in Fig. 1, for both high- and low-temperature treatments. Relative humidity was 60%, and grafts were fertilized and watered as described in Owens et al. (2001). Cones were collected on September 20 in the high-temperature regimes, and October 9 for the low-temperature treatments.
The first year, seeds were sown in multipot containers in a mixture of peat and Perlite 75% : 25% (volume-based proportions). Growing conditions in the Phytotron at the University of Oslo were as described by Johnsen et al. (1996). Plants were grown in continuous light until they were 15–20 cm tall, then potted in 2-l pots using the same root medium and given a 16-h light/8-h dark cycle immediately after repotting. This treatment combination resulted in a rapid development of terminal buds; the first pioneers had terminal buds 11 d after repotting and 100% of the plants had produced terminal buds within 19 d. Plants were then given a 12-h light/12-h dark cycle for additional 6 wk before being cold stored at 2°C for 8 wk.
After cold storage, plant material was divided in three groups to be tested in three environments; in a phytotron, a glasshouse and an outdoor nursery. Constant vs variable temperatures as well as artificial vs natural light may affect physiological processes differently (Bamberg et al., 1967), and we wanted to test if was possible to extrapolate test results from growth rooms to glasshouse and field conditions. The full set of progenies from all combinations of maternal treatment and families were placed in the phytotron and glasshouse experiment, but only two families were fully represented in the outdoor nursery experiment. The phytotron experiment started in the middle of December 2002 and the plants were given a 15 h photoperiod (200 µmol m−2 s−1) at 10°C/5°C (light/dark) for 3 wk. During the following three weeks plants were given an 18-h photoperiod (250 µmol m−2 s−1) at 15°C/10°C (light/dark). The plants were then given 22°C and continuous light (300 µmol m−2 s−1) for 3 wk followed by a growth cessation and cold acclimatization treatment. Cessation of growth was induced by increasing the dark period by 1 h per week, and cold acclimatization started when the dark period was 8 h with the temperature set to 10°C. When the dark period was 12 h, temperature was set to 5°C. Light period temperature was 22°C throughout this cold acclimatization period.
Bud burst (flushing) was recorded twice a week according to Krutzsch (1973) until all buds burst, and shoot extension was measured weekly until elongation ceased and terminal buds were formed. Plants from the phytotron were freeze-tested on May 20 of 2003, when darkness was 13 h and night temperature 5°C. This was done by collecting twigs from each plant, both leader shoot and lateral shoots, and inserting detached twigs in humid peat in multipot container for freeze testing. Trays with twigs were frozen at −11°C, −12°C and −13°C in three programmable freezing chambers. The twigs were subjected to 5°C for 1 h, and then cooled at 2°C h−1 until test temperature was reached. They were held at the test temperatures for 4 h, thawed at 2°C h−1 to 5°C, and then held at 5°C for at least 3 h. After freezing the twigs developed freezing injury symptoms (browning) for 3 wk at continuous light and 20–24°C in a misting chamber (> 95% relative humidity was kept).
Needle injury was classified visually on individual twigs according to the following classes: 0, no visible injury, all needles green; 1–10, 10% classes of brown or discoloured needles (10 = from 90 to 99%); 11, all needles completely brown. To score injury to cambium and buds, a scalpel was used to slice the twig longitudinally, revealing the cambium and both terminal and lateral buds.
Cambial injury was scored according to the following classes: 0, no visible injury; 1–3, 33% classes of brown tissue; 4, complete injury along the entire stem.
Buds were classified as either dead or alive (dried brown or green succulent). The total number of buds was counted on each twig and the proportion of dead buds per twig was calculated. Each twig was then categorized according to the following classes: 0, all buds alive; 1–4, 25% interval classes of dead buds.
The plants to be grown in the glasshouse and the outdoor experiment were moved from the cold store to the glasshouse and outdoor nursery on 14 May, 2003. Plants were grown as described elsewhere (Johnsen et al., 2005). Bud burst was assessed twice a week until all plants had started to elongate their shoots, and leader shoot extension was measured weekly thereafter.
Experimental design and statistics
All progeny tests were designed as complete blocks, replicated two to six times depending on experiment. Family was used as a contiguous main plot, and the maternal treatments within family plots were fully randomized, placing five plants from each maternal treatment noncontiguously randomized. Development of terminal buds the first growing season was studied in two phytotron rooms (replicates) with 7 × 3 family main plots in each room. The phytotron test the second season contained six replicates, the glasshouse trial had three and the outdoor experiments had four replicates.
When twigs were placed in multipot trays for freezing, we decided to use the original experimental layout from the phytotron room in the multipot system. One twig per plant was placed in the same randomized position as the donor plant. Thus, at each of the three test temperature (−11°C, −12°C, and −13°C) we sorted out twigs which represented the full experiment in the phytotron room.
Seedlings were examined until a terminal bud appeared the first growing season. The number of days from start of short day treatment until this happened was analysed statistically by the following linear model, based on individual seedlings:
Yikjl = µ + Ti + Dj + Tdij + Fk + Rl + eijkl
(Yikjl is the numbers of days from start of short-day treatment in a plant from maternal temperature i, daylength j, family k and replicate l; µ is the total mean; Ti is the fixed effect of maternal temperature i (i = HT, LT); Dj is the fixed effect of maternal day length j (j = lD, SD); Tdij is the fixed interaction between maternal temperature and daylength; Fk, is the fixed effect of family k (k = 1, 2, 3); Rl, is the random effect of room l (l = 1, 2); eijkl is the residual error).
The residual error was used as the denominator when testing the significance of fixed effects and LSD-test was used to test differences between maternal treatments.
A similar model was used to analyse the data from the second growth season in the phytotron room, glasshouse and the outdoor nursery trials. We substituted the random effect of room with replicate. The variables analysed by this model were shoot elongation (mm) during early (4–9 cm) stage, and the final extension (mm) from the median stage of leader elongation (17–18 cm) to growth had terminated.
All categorical scorings (bud burst, freezing injury) were transformed to normal scores. We assigned midpoint x-values of cumulative frequency distributions of plants within each replicate (and test temperature for injury data), based the original liability classes of earliness or injury, to the individual plants (Gianola & Norton, 1981; Falconer, 1989; Ericsson, 1994). The transformed single seedling data formed approximate normal distributions with a mean = 0 and variance ranging from 0.8 to 0.95, and assigned seedling values within the range ± 2.45. Negative values represent late bud burst or hardier seedlings and positive values earlier bud burst or less hardy seedlings than average. Analyses of variance of normal scores were made by the same model as above, except for replicates, which were omitted. As an example of such analyses, the result from the anova of bud burst normal scores in the beginning of the second growth season in the glasshouse environment is shown in Table 1.
Table 1. The anova of bud burst (normal scores) of Picea abies (Norway spruce) in the beginning of the second growth season in the glasshouse test environment
Source of variation
Degrees of freedom
Maternal temperature (T)
Maternal day length (DL)
T × DL
The seedlings formed terminal buds very fast in phytotron rooms the first growth season (Fig. 2). Even though the main effects of maternal temperature and daylength as well as their interactions were significant (Table 2), the maximum difference in mean performance was less than 1 d. Despite the very small experimental error found, the family difference was not significant (Table 2). However, after dormancy release the seedlings expressed significant interactive effects of the maternal treatments the second growth season (Figs 3 and 4) and a significant difference between the three families was found (Tables 1, 2 and 3). The main effects of maternal temperature and daylength were small because of the large interactions. Progenies from HT-SD and LT-LD flushed significantly later (3–4 d) than those from LT-SD and HT-LD in all test environments (Fig. 2 upper part). Because bud burst was delayed, the elongation of leader shoots lagged behind at the beginning of elongation period in HT-SD and LT-LD (Fig. 2 middle part). The cessation of growth, expressed at final extension (mm) from median elongation to termination of growth, was also delayed in the HT-SD and LT-LD (Fig. 2, lower part), and this resulted in longer total shoot length in the second season (data not shown). Twigs from the seedlings that were growing in the phytotron room were freeze tested during cold acclimatization. Figure 4 shows that progenies from HT-SD were the least hardy ones, those from LT-LD were significantly hardier, but nevertheless significantly less hardy than their full sibs from HT-LD and LT-SD. The maximal difference between high and low temperature at short days (Fig. 4) is comparable to 2.5 latitudinal provenance difference and about 250 m elevation at 61° N in Norway (Dæhlen et al., 1995; Ø. Johnsen & T. Skrøppa, unpublished). In all tests, both first and second growth season, the two-way interactions between family and maternal treatments were nonsignificant.
Table 2. P-values from the statistical tests of maternal treatments (MT) and families (F) from first and second growth season in the tests environments (TE) phytotron (p), glasshouse (g) and outdoors (o) of Picea abies (Norway spruce)
Bud set (BS) was recorded in the first growth season; bud burst (BB), early elongation (EE) and growth cessation (GC; final extension (mm) to growth has ceased) were recorded in the second season. Maternal treatments were temperature (T), day length (DL) and the interaction (T × DL).
T × DL
Table 3. P-values from the statistical tests of freezing injury (normal scores) after tests at −11 to −13°C in the second growth season of Picea abies (Norway spruce) in the phytotron room
Day length (DL)
T × DL
The maternal photoperiod and temperature influence the progeny in an interactive way. The data explains why the northern performance is maintained, despite high temperature during seed production in the northernmost areas of Norway spruce (Figs 3 and 4). The species has developed a mechanism that relies on both temperature (Johnsen et al., 2005) and photoperiodic signals during sexual reproduction, indicating the presence of a very flexible regulation of adaptive plasticity (Donohue & Schmitt, 1998). This may also explain why we observe a rapid shift in adaptive performance from one generation to the next in progenies from continental (50–52° N) Norway spruce grown at 64° N in Norway. They perform more like Norwegian provenances than continental provenances (T. Skrøppa et al., unpublished). We think that this ability probably will make Norway spruce less vulnerable to adverse effects of climate change in the future (Hänninen et al., 2001; Saxe et al., 2001).
The memory persists
The memory effect endures for many years in the filial generation (Johnsen, 1989a,b; Johnsen et al., 1989; Skrøppa, 1994; Edvardsen et al., 1996). It seems futile trying to erase the memory of the past time reproductive environment by using strong forcing treatment to mask the differences between progenies the first growth season. The programmed difference between maternal treatments and the inherent difference between the families reappear after dormancy release, in all three environments (Fig. 3). It was shown in an early study that the memory is maintained even after a cycle of cutting propagation (Johnsen, 1989a). By contrast, no evidence for memory of photoperiod treatment the first growing season could be seen in the same cutting material years later (Johnsen, 1989a). Nevertheless, the vernalization phenomenon in plants (e.g. in Arabidopsis; Sung & Amasino, 2004) and the after-effects of autumn temperature on timing of bud burst in spring (Heide, 1974, 2003) are examples of short-term memory effects of past temperature condition on subsequent performance in the same plants.
The interaction is not easily explained
The unexpected nonadditive effect of low temperature in long days is hard to explain, and our observations certainly call for further studies (Figs 3 and 4). Natural daylength at latitude 59–61° N gives the opposite results; low temperature during seed production always results in advance timing of bud burst, bud set and frost hardiness development (Johnsen et al., 1995, 1996, 2005; Figs 3 and 4). The only factor we can think of is that the genotypes used in our study may become maladapted to the programmed northern climate in the phytotron. They originate from 59 to 61° N and the phytotron mimics conditions at 66–67° N. We may speculate that some unnatural physiological imbalances occur under the low-temperature and long-day regime, beyond the natural ecological limits of the maternal genotypes. This hypothesis can probably be tested by comparing hormonal levels in the developing embryos under long and short days and high and low temperature, using mother trees originating from both northern and southern areas.
In an interesting study with shortleaf pine (Pinus echinata), Schmidtling & Hipkins (2004) used allozymes to tests for segregation ratio change between two contrasting reproductive environments. They documented significant differences between the two maternal environments in segregation ratios for many cross–loci combinations. However, the environmental irregularities in allozyme alleles did not relate well to differences in progeny performance induced by the crossing environments. Johnsen et al. (2005) did not find any progeny differences in Norway spruce that could be related to temperature differences during prezygotic stages and fertilization. By contrast, progeny performance was strongly associated with the temperature difference from pro-embryo to mature seeds. Selection among a maximum of four genetically different, competing and developing embryos inside a seed (Sarvas, 1968; Owens & Blake, 1985) would only create a minimum proportion selected of 0.25. This theoretical selection intensity of 1.27 (Falconer, 1989) is rather low, and possible changes due to directional selection during embryo competition cannot alone explain the large phenotypically change generally observed in Norway spruce. Selective events during reproduction associated with a warm and a cold maternal environment have also been studied in two independent crosses of Norway spruce (G. Besnard et al., unpublished). One family expressed large and the other small phenotypic differences between the crossing environments. However, strongly distorted alleles were observed depending on the crossing environment in both crosses, i.e. the altered allele frequencies could not explain the phenotypic difference well. Nevertheless, even if it is difficult to believe that postzygotic selection may be a significant explaining factor, we cannot rule out that a single genetic selection at a locus could impose large effects of other genes through epistasis, especially if distortion happens in regions where regulatory genes are positioned.
An alternative hypothesis is that a memory of photoperiod is programmed during zygotic embryogenesis. This memory is probably epigenetic, because the percentage of 5-methylcytosine is higher in DNA extracted from progenies from a warm embryogenesis than from their full sib counterparts with a cold embryogenic history (R. Baumann et al., unpublished). This is correlated with a reduced transcription level of genes that may regulate bud phenology, cold acclimatization and embryogenesis in progenies with a memory of elevated embryonic temperatures (Johnsen et al., 2005). We thus hypothesize that the maternal photoperiod and temperature interactively regulate a molecular epigenetic and long-lasting memory in the progeny, which affects climatic adaptation in Norway spruce.
This work was supported by the EU grant #QLK5 – CT-2000-00349, The Research Council of Norway Grant #1137.51.01, Grant #155041/140 and Grant #155873/720. We thank the staff of the Phytotron at the University of Oslo for taking care of grafts and progeny plants, and for their friendly service during the experimental period.