SEARCH

SEARCH BY CITATION

Keywords:

  • bud phenology;
  • chestnut;
  • growth;
  • QTL analysis;
  • water use efficiency

ABSTRACT

  1. Top of page
  2. ABSTRACT
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION AND PERSPECTIVES
  8. ACKNOWLEDGEMENTS
  9. REFERENCES

A QTL analysis for three different adaptive traits was performed in an F1 progeny of Castanea sativa Mill. The female and male parents originated from two Turkish chestnut populations adapted to a drought and humid environment, respectively. QTLs for bud flush, growth and carbon isotope discrimination were detected over a 3-year period. Bud set was also recorded in the last year of measurement. Thirty-five individual QTLs were detected for phenology, 28 for growth and 17 for carbon isotope discrimination, most of them explaining a low to moderate proportion of the total phenotypic variance. QTLs were distributed throughout the whole genome. Temporally stable QTLs were identified for all the traits analysed, with phenology showing the higher proportion of stable QTLs. Interesting phenotypic correlations and co-localizations among QTLs for different adaptive traits were observed, allowing the formulation of an hypothesis about the genetic adaptation of the female parent to drought.


INTRODUCTION

  1. Top of page
  2. ABSTRACT
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION AND PERSPECTIVES
  8. ACKNOWLEDGEMENTS
  9. REFERENCES

Evidence of a rapid and global climate change has forced the scientific community to investigate its potential impact on animals and plants. Meta-analyses carried out on a large number of studies have recently shown that wild species are already reacting to global warming (Parmesan & Yohe 2003; Root et al. 2003). Under this scenario, the study of adaptive variation in natural populations has been recently emphasized. Forest trees, being long-living organisms with a sessile habit and generally characterized by a high phenotypic plasticity, offer a unique opportunity to study the impacts of climate changes on wild and undomesticated species and their adaptation to changing environments (Barber, Juday & Finney 2000). The tendency to longer growing seasons and the recent climate warming are expected to extensively influence the timing of bud flush and set, flowering, leaf fall and the growth rate of forest trees. Studying the genetic architecture of adaptive traits in forest trees might therefore provide important clues concerning the evolution of such traits and make it possible to predict the potential of adaptation of these species. Several studies on forest trees have used the QTL analysis approach to dissect the Mendelian factors controlling adaptive traits (reviewed by Sewell & Neale 2000; van Buijtenen 2001). One of the most relevant issues is that few loci of relatively large effect seem to control such complex traits. Despite the bias on the QTL effect estimates due to the small population sizes often analysed (Beavis 1995), a recent theoretical work on the evolution of adaptive traits showed that mutations of large effects are fixed first (Orr 1998). Hurme et al. (2000) applied the QTL approach to identify major loci influencing climatic adaptation in a natural forest of Scots pine. They concluded that during the adaptation process, natural selection fixed at least some alleles of rather large effect.

Bud phenology, growth and carbon isotope discrimination (delta, Δ, which provides an indirect measure of plant water use efficiency) are adaptive traits that show great phenotypic variation in natural populations of forest trees (Zhang & Marshall 1995; Tognetti, Michelozzi & Giovannelli 1997; Lauteri et al. 1999; Hurme et al. 2000; Jermstad et al. 2001). Initiation and cessation of the growing seasons, defined through bud flush and bud set timing, have profound implications for adaptation of perennial plants to the cold winter temperatures. Early flushing genotypes might be susceptible to spring frost damage. Likewise, bud set timing is related to the autumn cold acclimation (Howe et al. 2000). Growth traits, such as annual height and diameter increments, are important components of plant vigour and biomass production, and they are profoundly influenced by abiotic and biotic stress occurrence during the growing season. In addition, they are relevant characteristics from the economic point of view, often evaluated in breeding programmes (Bradshaw & Stettler 1995). Carbon isotope discrimination (Δ) is a parameter related to the isotopic fractionation of carbon stable isotopes during the photosynthetic process (for review see Farquhar, Ehleringer & Hubick 1989; Brugnoli & Farquhar 2000). Plant material is always enriched in 13C with respect to the isotopic composition (δ13C) of atmospheric CO2. This is particularly evident in C3 plants where the fractionation effect mostly occurs during CO2 diffusion from outside the leaf to the carboxylation sites into the chloroplasts, and during the carboxylation by ribulose 1,5-bisphosphate (RuBP) carboxylase. Due to its relationships with the diffusional path of photosynthetic gas-exchange (for both CO2 and water vapour in reverse directions) and with the photosynthetic substrate demand (CO2 fixation by RuBP carboxylation activity), Δ has been theoretically predicted and empirically demonstrated to be inversely related to plant water-use efficiency (WUE, roughly the ratio of carbon gain to water losses; for deeper insights see Farquhar et al. 1989; Brugnoli & Farquhar 2000). Despite the complexity of this trait, significant heritabilities and low genotype–environment interactions have been found for Δ in crop species (Hall et al. 1994) encouraging the use of this parameter for breeding purposes.

In this study we used the progeny of two individuals originating from two natural populations of Castanea sativa Mill. adapted to contrasting environments (Lauteri et al. 1997) to study the genetic architecture of bud phenology, growth and carbon isotope discrimination in chestnut. European chestnut has one of its major centres of origin in the Turkish region, which represented a refuge area during the Würm glaciation (Huntley & Birks 1983). A high genetic differentiation was shown between western and eastern natural chestnut populations in Turkey (Villani et al. 1991). This genetic differentiation was partially explained by some climatic factors (Pigliucci, Villani & Benedettelli 1990). Indeed, the two Turkish regions show striking differences in macroclimatic parameters, particularly rainfall, one of the most important limiting factors for chestnut expansion. Progenies from western and eastern populations were further compared in a common environment, close to the western ecotype natural environment, in the centre of Italy by Lauteri and coworkers (Lauteri et al. 1997, 1999). The western ecotype, adapted to a Mediterranean drought region, showed higher Δ (lower water-use efficiency), enhanced stomatal conductance, higher photosynthetic capacity, lower above-ground juvenile growth, more vigorous mature growth and earlier bud flush in comparison with the eastern ecotype, originating from a humid and cool region. Given the characteristics of these two populations, they were considered a suitable material to study the adaptive potential of this species in relation to climate change. A controlled cross between two individuals belonging to the two ecotypes was previously used to construct a genetic linkage map as a preliminary step for QTL analyses (Casasoli et al. 2001). The objectives of this work were to: (1) study the genetic architecture of three types of adaptive traits in chestnut: bud phenology, growth and water use efficiency; (2) verify the genetic basis of the differential adaptation of the two chestnut populations by the identification of segregating QTLs for adaptive traits in the progeny of the two parental individuals coming from contrasting environments.

MATERIALS AND METHODS

  1. Top of page
  2. ABSTRACT
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION AND PERSPECTIVES
  8. ACKNOWLEDGEMENTS
  9. REFERENCES

Plant material

The full-sib family used for QTL analysis has been previously described in Casasoli et al. (2001). The controlled cross was made in 1998 and seeds were germinated in a greenhouse (Istituto di Biologia Agroambientale e Forestale, CNR, Porano, central Italy; latitude, 42°43′, 500 m elevation). The 186 seedlings were grown in pots in the nursery of the Institute during 1999, 2000 and 2001, under natural conditions with the exception of watering from June to September In January 2002 the seedlings were transplanted in the experimental field of the Institute using a completely randomized design with a between-tree spacing of 7 m × 7 m. During the 4 years 34 plants died reducing the size of the F1 family to 152 individuals.

Phenotypic measurements

Measurements of bud flush timing, annual increments of height and diameter, and carbon isotope discrimination were carried out over a 3 year period 2000, 2001 and 2002 corresponding to the growing seasons 2, 3 and 4 since seed germination. Bud set timing was scored only in 2002. Bud flush timing was recorded in the spring. Taking into account that: (1) chestnut seedlings often showed several basal shoots; and (2) striking differences were observed among individuals for time elapsing between the day the first bud opened and the flushing time of most of the buds, two different observations were made: the date of the first observed unfolded leaf (bud) and the date when 70% of the tree buds showed an unfolded leaf (bud70). Julian days were used for data analysis. Observations of bud set were done from August to October 2002, starting from when about 30% of the plants initiated terminal bud formation. Bud setting was recorded during four observations. With the exception of the first time, observations were recorded according to a five-point scale (modified from Fernández López, personal communication): 1 = plant in active growing; 2 = terminal bud is developing, but most of the lateral shoots are in active growing; 3 = terminal bud is developing, and the plant shows more lateral shoots developing buds; 4 = terminal bud defined but the shoot is not completely developed; 5 = terminal bud completely developed. During the first bud set observation, the five-point scale was not sufficient to describe the phenotypic variability observed in the progeny. An 11-point scale was then used: 0 = any bud was setting, 1 = 0–10% of developing buds, 2 = 10–20% of developing buds and so on until 11 = 90–100% of developing buds. The sum of the scores was used for data analysis. QTL analysis performed on data from separate observations was not significantly different from analysis carried out using the sum of the scores (data not shown). It was therefore decided to present only the results obtained from this analysis.

Growth was estimated by measuring the annual height and basal stem diameter increment. Basal stem diameter (diameter) was measured on the dominant shoot at the soil level over the 3 years. In 2001 and 2002 the diameter at 10 cm above the soil level was also measured (diamb). Height was measured on the dominant shoot from ground level to the terminal bud tip (height). Total growth (heighttot and diametertot) and annual increments were used for data analysis.

Carbon isotope discrimination (Δ) was analysed on leaf dry matter harvested at the end of the growing season according to Lauteri et al. (1997). Three subsamples of each individual were separately analysed and the mean Δ-value used for QTL analysis. Isotope ratios were measured using a continuous flow isotope ratio mass spectrometer [(CF-IRMS) ISOCHROM II; VG-ISOTECH, Middlewich, Cheshire, UK], carbon isotope composition (δ13C) was referred to the PBD international standard and Δ was calculated according to the convention in Farquhar et al. (1989), assuming −8‰ as the value of δ13C of atmospheric CO2.

Trait description, number of individuals measured, means, standard deviations, range of variation, and coefficients of variation are reported in Table 1. STatistica software [STatistica for Windows (1998); Statsoft Italia S.r.l., Padova, Italy] was used to calculate phenotypic correlations (Table 2) and to test all traits for normality (Kolmogorov–Smirnov). Traits showing departure from normality (Table 1) were not normalized and raw data were always used for QTL analysis.

Table 1.  Description of the traits measured during the three years and basic statistical descriptors
Trait name (units)Trait definitionnMean (SD)MinMaxCV (%)NormalityaBlock effectaYear effecta
  1. N, number of individuals measured; SD, standard deviation; CV,coefficient of variation. aThe Kolgomorov-Smirnov test was used to test departure from normality. ns, not significant; *significant at 5% level; **at 1% level; ***at 0,1% level. The year effect was tested only for traits measured during the three growing seasons, anyway, bit was not tested for total measurements of growth parameters due to their strong autocorrelation.

Δ2000 (‰)carbon isotope discrimination 2000155 20.2 (0.96)17.1 22.8 4.75nsns*** (Δ)
Δ2001carbon isotope discrimination 2001152 20.1 (1.23)15.8 22.4 6.12nsns 
Δ2002carbon isotope discrimination 2002152 18.8 (0.94)15.3 20.8 5.00nsns 
bud2000 (Julian days)bud flush, first bud 2000174 91 (4.94)66104 5.43*ns*** (bud)
bud2001bud flush, first bud 2001153 82 (5.02)62100 6.12*ns 
bud2002bud flush, first bud 2002153 91 (5.74)78105 6.31nsns 
bud702000bud flush, 70% of unfolded leaves 2000172102 (4.52)89113 4.43*ns*** (bud70)
bud702001bud flush, 70% of unfolded leaves 2001150 90 (5.78)82115 6.42*ns 
bud702002bud flush, 70% of unfolded leaves 2002153103 (5.66)86119 5.50nsns 
budset2002 (score)bud set 2002151 19 (4.99) 6 2626.26*** 
diameter2000 (mm)basal stem diameter increment at the soil level 2000137  4.2 (2.00) 0.6 12.747.62nsns*** (diameter)
diameter2001basal stem diameter increment at the soil level 2001153  2.9 (1.47) 0.0  7.750.69nsns 
diameter2002basal stem diameter increment at the soil level 2002150  6.4 (3.35) 0.9 17.452.34ns*** 
diamb2001basal stem diameter increment at 10 cm above the soil 2001153  2.4 (1.02) 0.3  5.542.50nsns 
diamb2002basal stem diameter increment at 10 cm above the soil 2002151  4.4 (2.26) 0.8 13.451.36ns*** 
height2000 (cm)height increment 2000135 16.6 (7.80) 3.8 53.146.99nsns*** (height)
height2001height increment 2001152 11.4 (7.04) 0.8 44.261.75*ns 
height2002height increment 2002150 12.0 (10.67) 0.0 51.888.92** 
diametertot2000b (mm)basal stem diameter measured at the soil level 2000136 10.7 (2.12) 6.2 21.319.81nsns 
diametertot2001basal stem diameter measured at the soil level 2001153 13.4 (2.24) 7.4 21.316.72nsns 
diametertot2002basal stem diameter measured at the soil level 2002151 19.7 (4.21)10.3 31.621.37ns* 
heighttot2000b (cm)height 2000133 36.2 (8.36)17.9 72.123.09nsns 
heighttot2001height 2001153 49.7 (11.50)26.2103.723.14nsns 
heighttot2002height 2002152 62.6 (15.09)34.0125.824.11nsns 
diambtot2001b (mm)basal stem diameter measured at 10 cm above the soil 2001153  9.5 (1.81) 5.1 16.919.05nsns 
diambtot2002basal stem diameter measured at 10 cm above the soil 2002152 13.9 (3.24) 6.5 24.523.31ns* 

Genotyping and linkage map construction

A previously constructed genetic linkage map (Casasoli et al. 2001) based on 96 full-sibs was used to select a subset of evenly spaced molecular markers to genotype the 186 F1 individuals. Thirty-nine microsatellites markers were added to the map (Barreneche et al. 2004). The framework parental maps for QTL analysis were constructed using MapMaker V2.0 (Lander et al. 1987) using a LOD threshold of 8.0 and a maximum recombination fraction of 0.28 (corresponding to a distance of 31.64 cM, Kosambi) as grouping criteria. The marker order was assessed by the ‘ripple’ command of MapMaker (LOD > 3.0). A few distorted markers were integrated to fill some gaps when they did not disturb the optimized marker order. Genome length was estimated as described in Casasoli et al. (2001).

QTL analysis

Phenotypic data were tested for the block effect by analysis of variance (SPlus software; Math Soft, Inc. Seattle, WA, USA). A significant block effect was found in 2002 for bud set, height, diameter, diamb, diametertot, and diambtot (Table 1). These data were adjusted to the block effect prior to QTL detection. The year effect was also tested for the traits measured at different ages (bud, bud70, growth increments and Δ). All traits showed a significant year effect (Table 1). A two-way analysis of variance was performed for these traits considering all loci and years in order to examine the marker, year, and marker–year interaction effects. The following model was applied: Yijk =µ + Yeari + Mj + Yeari × Mj + ɛijk, where Yijk is the phenotypic value of the individual k measured in the year i and characterized by the marker j, µ is the overall mean, Yeari is the effect of year i, Mj is the effect of marker j, Yeari × Mj is the interaction effect, and ijk the residual effect. QTL analysis was performed both using raw data for the 3 years (trait name is followed by the year, see Table 1) and adjusted data to the year effect. In this case the name of the trait is not followed by the year. Analysis carried out on adjusted data makes it possible to take into account the environmental variability registered during the three years allowing to confirm in some cases temporal stability of some QTLs.

QTL analysis was performed under the two-way pseudo-test cross strategy (Grattapaglia, Bertolucci & Sederoff 1995) using the MultiQTL software (Britvin et al. 2001; http://esti.haifa.ac.il/(poptheor). Multiple Interval Mapping (MIM, Kao, Zeng & Teasdale 1999) was used to detect QTLs with the ‘fitting cofactors’ option of the software. According to the principles of MIM, a QTL with the highest effect is first found and it is taken as a cofactor to control the genetic background, while another QTL is searched in a different position. This procedure, iteratively repeated until no further QTL is found, takes into account effects due to the QTLs present in other chromosomes. Therefore, the precision and the power to detect QTLs are increased. Significance thresholds were computed by permutation (1000 permutations, Churchill & Doerge 1994). A genome-wise type I error of 5% was chosen to declare the presence of ‘significant’ QTLs, and 20% for ‘suggestive’ QTLs. Confidence intervals for QTL position were estimated by bootstrap (1000 random samples, Visscher, Thompson & Haley 1996). The proportion of the phenotypic variation explained by the QTL (PEV) was calculated as: PEV = 1/4(d 2/σph2), where d is the estimated substitution effect of the QTL and σph the phenotypic variance of the trait. When a pseudo-testcross strategy is used, the allele substitution effect for each QTL is calculated taking into account the single-dose markers present in one parent, linked to the QTL, and absent in the other. This estimate is averaged over the two alleles inherited from the other parent. It is important to note that due to the lack of knowledge of the phase between linkage groups (coupling or repulsion) and the marker re-coding procedure used during map construction (Nelson, Nance & Doudrick 1993), signs of the allelic substitution effect can be compared only for QTLs present on the same linkage group. A joint analysis of the trait scores across years was also performed using the MultiE option available in the MultiQTL software (data not shown). No important improvements of QTL detection results were observed in comparison with the joint analysis carried out on adjusted data to the year effect.

RESULTS

  1. Top of page
  2. ABSTRACT
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION AND PERSPECTIVES
  8. ACKNOWLEDGEMENTS
  9. REFERENCES

Linkage maps for QTL analysis

The female and male framework maps were constructed using a total number of 109 (66 RAPDs, 19 ISSRs, two isozymes and 22 SSRs) and 108 (76 RAPDs, 11 ISSRs, one isozyme and 20 SSRs) markers, respectively. The female map was 848.6 cM (Kosambi) in length, covering 84% of the estimated genome length. The male map spans 832.9 cM, which corresponds to a genome coverage of 75%. The average marker spacing was of 8.7 cM in the female and male framework maps. Genome length estimates and average marker spacing were similar to the first version of the C. sativa genetic linkage map (Casasoli et al. 2001). The new microsatellite markers integrated into the map increased the genome coverage. Furthermore, they provided fully informative markers to complete the alignment of the 12 chestnut linkage groups (data not shown). A few differences were observed in marker order (linkage group, LG, two female, F, and male, M, LG11F and LG12F), mainly occurring among tightly linked markers. Due to the larger number of genotyped individuals and the more stringent conditions used during the marker ordering step, the statistical support of the present maps was greatly improved.

Phenotypic data

For all the traits measured over the 3 year period, a large phenotypic variation within the F1 progeny was observed, with minimum and maximum values often ranging more than two phenotypic standard deviations from the mean (Table 1). A significant year effect was observed for these traits. This is not surprising since seedlings were grown in a nursery during the first 2 years of measurement (2000–01) and then transferred to the field for the third growing season (2002). In addition, plants were grown open air and faced very different climatic conditions during the 3 years; for example, year 2000 was characterized by an unusually high annual mean temperature, whereas 2002 was characterized by particularly abundant summer rainfalls. Therefore, both environmental conditions and ontogenesis contributed to the year effect.

A cluster analysis was performed in order to point out clusters of phenotypic correlations between traits. Traits were presented in the order obtained after the clustering analysis (Table 2). Overall, growth, Δ and bud set were more highly correlated with each other than with bud flush measurements. Bud flush measurements were significantly correlated in 2001 and 2002, but not in 2000 and 2001. Bud and bud70 were always significantly correlated in the same year. Bud set was significantly and negatively correlated with height and diameter increments in 2002 (an early bud set corresponded to a lower growth rate). Significant phenotypic correlations were observed for Δ measurements over the 3 year period. Growth parameters were always negatively correlated with Δ, in several cases at a significant level. Very weak or non-significant phenotypic correlations occurred among annual growth increments. Nevertheless, diameter and height increments were always significantly correlated in the same year. Phenotypic correlations between bud flush and growth and between phenological traits and Δ showed a more complex pattern. They varied with no discernible trend across years suggesting influences due to the variations of environmental conditions and ontogenetic effects.

QTL analysis

Overview of the detected QTLs
Total number of QTLs

A total of 80 QTLs were detected (Fig. 1), 62 resulting from the analysis on a yearly basis and 18 from data adjusted to the year effect. Fifty QTLs were detected at the significant level (5% at the genome level), whereas 30 were classified as suggestive QTLs (20% at the genome level). Twenty-five suggestive QTLs co-localized with significant QTLs for the same trait, or several suggestive QTLs for the same trait were identified in the same chromosomal region. Only in five cases were suggestive QTLs localized in a chromosomal region where no other QTLs for the same trait were detected.

image

Figure 1. QTL positions for traits analysed in 2000, 2001 and 2002. See Table 1 for description of trait names. Linkage groups (LG, F: female, M: male) and molecular markers are named as in Casasoli et al. (2001). Kosambi units are used for genetic distances; scale is reported near the LG1F. Only linkage groups containing QTLs are reported. Solid bars denote the most probable position of the QTL obtained from the highest LOD score position given by the CIM and the bootstrap. Confidence intervals, represented as lines, were estimated on 1000 bootstrap samples. The figure was drawn using MapChart software (Voorrips 2002).

Download figure to PowerPoint

Unbalanced distribution of QTLs between both parents

A striking difference in the number of QTLs detected was observed for the two parental maps. Sixty QTLs (75%) were identified in the female linkage groups, whereas only 20 (25%) concerned the male. This difference was observed for all traits studied. Although genome coverage reached 84 and 75% in the female and male maps, respectively, this difference could not be explained by this factor alone.

Redundancy of the detected QTLs

Overall in the two parental maps, a total of 35 QTLs were detected for phenology, 28 for growth and 17 for Δ. Some QTLs were detected in both the yearly based and adjusted data. Such QTLs sharing the same map position (on the same linkage group in one parent or on homologous groups between parents) and having the same direction of substitution effect (when localized in the same parental linkage group) were considered as unique QTLs. Following this criterion 14 unique QTLs were detected for phenology, 12 for growth and eight for Δ. However, the large confidence intervals, often larger than 40 cM, did not permit the precise localization of most of the QTLs, hampering a clear definition of unique QTLs. It is possible that two linked QTLs were also present on the same linkage group. We tested this possibility using the option present in the MultiQTL software to detect two linked QTLs (data not shown). No significant test was obtained, suggesting that the resolution power of our experiment was quite low to detect two linked QTLs.

Phenology

Thirty-five individual QTLs were identified for phenology distributed over 11 linkage groups (Table 3). The phenotypic variance explained by each QTL ranged from 4.2 to 17.1%. In most of the cases QTLs for bud and bud70 shared the same map position; nevertheless some differences were observed. For instance, two distinct QTLs were identified in LG3F for bud in 2000 and bud70 in 2002. The different map position and the opposite direction of the substitution effect suggest that these correspond to two different QTLs affecting bud flush. QTLs were defined as being temporally stable if detected in at least two different years and/or confirmed by the analysis on adjusted data. Following this criterion, stable QTLs were detected on LG1F, LG4F, LG6M, LG9F, and LG12F. Interestingly, among the four QTLs detected for bud set, three co-localized with QTLs for bud flush.

Table 3.  Bud phenology
MapTraitLGaMap position (cM) bLODP-value (significance %)cddPEV (%)e
  • a

    LG: linkage group.

  • b

    b Map position: distance in cM (Kosambi) from the top of the LG.

  • c

    P-value at the nominal level (genome-wise significance obtained from 1000 permutations).

  • d

    d, substitution effect.

  • e

    e PEV: phenotypic variance explained by the QTL.

femalebud2000 33.32.820.0009 (5%)2.837 8.1
1238.43.050.0000 (5%)2.785 7.9
bud2001 128.33.660.0000 (5%)2.844 8.5
 743.71.930.0196 (20%)−2.074 4.5
bud2001 923.92.370.0047 (10%)2.286 5.6
bud2002 122.74.610.0000 (5%)4.09512.2
 923.93.050.0009 (5%)2.995 6.8
1219.93.000.0009 (5%)2.903 6.3
bud702001 422.13.480.0009 (5%)3.72710.3
 612.82.920.0000 (5%)3.435 8.8
bud702002 127.13.480.0000 (5%)2.998 6.7
 289.33.620.0000 (5%)3.436 8.5
 348.23.910.0000 (5%)3.039 7.1
 525.53.410.0000 (5%)2.988 6.7
 923.96.320.0000 (5%)3.88911.6
1219.91.990.0121 (20%)2.046 3.2
budset2002 264.73.120.0019 (5%)2.822 8.9
 833.22.180.0131 (15%)2.335 6.1
bud 120.65.420.0000 (5%)2.69911.8
 433.52.500.0084 (10%)−1.573 4.2
 923.94.500.0000 (5%)2.184 8.2
1240.55.300.0000 (5%)2.51010.7
bud70 419.23.310.0009 (5%)2.002 7.1
 923.92.350.0084 (20%)1.736 5.3
1224.13.610.0000 (5%)2.119 8.0
malebud2002 645.52.360.0084 (10%)2.849 6.3
11 9.62.620.0056 (15%)3.214 7.8
bud702001 621.62.230.0121 (15%)3.057 6.9
 928.62.490.0028 (5%)3.228 8.6
bud70200211 8.02.410.0047 (10%)3.235 8.0
budset2002 650.83.440.0009 (5%)2.916 9.6
 921.76.640.0000 (5%)3.87017.1
bud 342.12.180.0140 (20%)1.848 5.8
 641.03.390.0000 (5%)2.165 8.1
bud70 245.72.080.0178 (15%)1.896 6.3
Growth

Sixteen traits were used to identify QTLs for growth. A relatively low number of QTLs (in most of the cases either one or two QTLs) was identified for each trait (Table 4). Twenty-eight individual QTLs, explaining from 5.5 to 17.0% of the phenotypic variance, were detected. Ten linkage groups contained QTLs for growth traits with clusters on LG1F for height and on LG10F, in which seven QTLs detected during the 3 years both for height and diameter were mapped. With the exception of the LG10F and LG11F, QTLs for diameter and height did not co-localize. Although data adjusted to the year effect proved the temporal stability of some QTLs (i.e. LG1F, LG10F), several QTLs seem to be year specific. QTLs for annual increments and total measurements of growth parameters co-localized in LG1F and LG10F.

Table 4.  Growth
MapTraitLGaMap position (cM) bLODP-value (significance %)cddPEV (%)e
  • a

    LG: linkage group.

  • b

    b Map position: distance in cM (Kosambi) from the top of the LG.

  • c

    P-value at the nominal level (genome-wise significance obtained from 1000 permutations).

  • d

    d, substitution effect.

  • e

    e PEV: phenotypic variance explained by the QTL.

femaleheight2000 293.12.870.0037 (5%)4.867 9.6
 7 0.03.820.0000 (5%)5.29411.8
1010.72.720.0000 (5%)4.419 8.0
1115.42.200.0065 (15%)3.919 6.2
height2001 1 0.02.740.0056 (5%)3.729 7.2
height2002 878.43.330.0019 (5%)8.35617.0
heighttot20001017.51.940.0093 (20%)−4.344 7.0
heighttot2001 1 3.42.830.0037 (5%)6.667 8.3
heighttot2002 1 6.14.930.0000 (5%)10.29011.5
1023.43.180.0000 (5%)7.947 7.1
diameter200111 6.52.010.0056 (15%)−0.784 7.1
diamb200111 4.52.210.0056 (15%)−0.540 7.2
diameter2002 748.92.480.0047 (5%)1.684 7.7
diamb2002 5 0.02.750.0047 (10%)1.171 7.8
diametertot20011023.42.350.0028 (5%)1.090 5.9
diambtot20011023.42.050.0093 (20%)−0.881 5.9
diambtot2002 623.52.20.0121 (15%)−1.589 6.4
1023.41.950.0110 (20%)−1.458 5.5
height 133.83.090.0009 (5%)2.876 7.3
 874.32.560.0065 (10%)−3.49010.6
1023.42.490.0009 (5%)2.506 5.5
maleheight20021117.83.460.0000 (5%)7.02011.5
diameter2001 740.22.780.0019 (5%)0.829 8.1
diamb2001 739.32.330.0028 (5%)0.503 6.3
diameter2002 615.62.080.0121 (15%)−1.879 9.8
 932.62.140.0140 (20%)1.468 5.9
diametertot2001 333.72.430.0047 (10%)−1.201 7.1
diambtot2001 333.73.570.0000 (5%)1.16610.3
Carbon isotope discrimination

Seventeen individual QTLs were detected for carbon isotope discrimination on eight different linkage groups explaining from 4.1 to 13.2% of the phenotypic variance (Table 5). Linkage groups 3F and 7F contain temporally stable QTLs whereas QTLs on LG1F, LG3M, LG4F, LG4M, LG6F, LG8F, LG10M and LG12F were year specific. Most of the QTLs were detected in 2000 and 2001, mainly located in the female linkage groups; only one QTL was detected in 2002.

Table 5.  Carbon isotope discrimination (Δ)
MapTraitLGaMap position (cM) bLODP-value (significance %)cddPEV (%)e
  • a

    LG: linkage group.

  • b

    b Map position: distance in cM (Kosambi) from the top of the LG.

  • c

    P-value at the nominal level (genome-wise significance obtained from 1000 permutations).

  • d

    d, substitution effect.

  • e

    e PEV: phenotypic variance explained by the QTL.

femaleΔ2000 318.63.300.0019 (5%)0.531 7.3
 4 0.04.620.0000 (5%)0.61510.1
 727.62.160.0093 (10%)0.415 4.6
1216.52.100.0131 (20%)0.425 5.8
Δ2001 353.65.400.0000 (5%)0.89413.2
 623.53.370.0009 (5%)0.667 7.3
 712.72.410.0075 (10%)0.560 5.1
 854.73.390.0019 (5%)0.739 8.8
Δ2002 1 5.14.510.0000 (5%)0.66112.2
Δ 123.53.250.0009 (5%)0.497 7.9
 357.73.800.0000 (5%)0.498 8.1
 4 0.02.080.0140 (15%)−0.359 4.1
 655.42.920.0037 (5%)0.430 6.0
 716.32.180.0140 (15%)0.366 4.5
maleΔ2000 339.72.400.0019 (5%)0.463 5.7
 425.32.110.0093 (10%)−0.432 4.9
1011.64.190.0000 (5%)0.63110.4

DISCUSSION

  1. Top of page
  2. ABSTRACT
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION AND PERSPECTIVES
  8. ACKNOWLEDGEMENTS
  9. REFERENCES

QTL analysis

Distribution of the QTLs over the genome

QTLs were detected on the 12 linkage groups. Eleven, 10 and eight linkage groups contained at least one QTL for phenology, growth and Δ, respectively. This result agrees with the complexity of these adaptive traits showing that most of the linkage groups contained genes explaining a portion of the phenotypic variance. However, considering the groups with the higher number of individual QTLs for each trait, linkage groups 1, 6, 9 and 12 contain the higher number of QTLs for phenology, 3, 4 and 7 for Δ and 1, 7, 10 and 11 for growth, respectively. Despite several co-localizations, the most stable QTLs for the three traits were located on different linkage groups.

Unbalanced distribution of QTLs highlights a differential genetic adaptation of the two parents

The most striking result of the QTL analysis was the presence of most QTLs detected (75%) on the female parental map. Even considering the difference in map saturation between the two parental maps (84 and 75% for the female and the male map, respectively), this result remains to be explained. The female and male parents come from two distinct and contrasting regions from the climatic point of view. The female parent belongs to a Mediterranean drought-adapted ecotype originating from western Turkey, whereas the male parent comes from eastern Turkey from a chestnut population adapted to a humid and cool region (Pigliucci et al. 1990; Villani et al. 1991). The two populations showed a very high genetic differentiation but a similar genetic variability was observed at the intrapopulation level (Villani et al. 1991). Thus, a comparable heterozygosity level is expected for the two parental individuals, even if heterozygosity at the quantitative trait loci could be not directly linked to our observations from isozyme analysis. However, the two populations were shown to be adapted to contrasting environments by Lauteri et al. (1997, 1999). This behaviour could explain our result. The progeny was grown in a Mediterranean environment with macro-climatic conditions (amount of annual precipitation and mean temperature) very similar to those registered in the origin region of the female parent (Lauteri et al. 1997). Having the parent adapted to a Mediterranean climate, the segregating alleles in the female individual might mainly contribute to the F1 full-sib phenotypic variability in this particular environment. To definitely demonstrate this hypothesis, QTLs for the same full-sib family may have to be detected in a humid environment. In this case, supposing that the male parent is heterozygous at the quantitative trait loci, most of the QTLs should be detected in the male parental map, this individual being adapted to a wet environment.

Stability of QTLs over the years

Detecting stable QTLs across several years, environments and genetic backgrounds, is a good way to verify results of a QTL experiment. In our experiment adaptive traits were measured over a period of 3 years to verify the temporal stability of the QTLs detected. It is well accepted that traits such as growth can be considered as developmental traits for which a temporal shift of the underlying genes is expected (Kremer 1992). Thus, both stable as well as year-specific QTLs are expected. The large confidence intervals observed for most of the QTLs hamper an accurate location of QTLs along the chromosome, influencing conclusions about co-localization of QTLs. However, if we consider that a QTL is temporally stable in our experiment when it is detected at least in two different years and/or confirmed by the analysis on adjusted data, then we found stable QTLs on LG3F and LG7F for Δ, LG1F, LG4F, LG6M, LG9F and LG12F for phenology, and LG1F and LG10F for growth. In LG9 a QTL for bud phenology was identified several times in both parental maps affecting both bud flush and bud set and the data adjusted to the year effect permitted a more precise location. Such stable QTLs could represent alleles that played a major role in the adaptability of this species. Complexity of Δ and juvenile growth and low heritabilities can explain the lower number of stable QTLs detected for these traits in comparison to phenology. Previous work on QTL detection for tree growth showed that QTLs varied from one developmental stage to another (Plomion, Durel & O’Malley 1996; Emebiri et al. 1998; Tsarouhas, Gullberg & Lagercrantz 2002). Therefore, growth has to be considered a trait determined by epigenetic QTLs; that is loci expressed in a specific stage of the developmental process (Wu et al. 2002). Thus the higher proportion of year-specific QTLs for growth, compared to phenology, is not surprising. Conner, Brown & Weeden (1998) proposed that QTLs for growth detected on mature trees could be useful to identify genomic regions controlling tree vigour, whereas a QTL analysis performed in single growing seasons may detect QTLs implied in the regulation of the tree growth. Comparing the location of QTLs for annual increments and total measurements of growth parameters shows that in most of the cases the two types of QTLs do not coincide. They are co-localized only on LG1F and LG10F, where stable QTLs that affect LG10F in both height and diameter, were mapped, suggesting the presence of major genes controlling juvenile tree growth.

Phenotypic variance explained by the detected QTLs

The ranges of heritability values on forest trees reported in the literature for the three adaptive traits analysed vary from moderate to high values for phenology (0.5–0.8, Bradshaw & Stettler 1995; Jermstad et al. 2001), from low to high for growth (0.20–0.80, Bradshaw & Stettler 1995; Byrne et al. 1997) and from low to moderate for Δ (0.20–0.50, Brendel et al. 2002; 0.31 in chestnut, Lauteri et al. personal communication). In chestnut, Pliura & Eriksson (2002) reported a rather high heritability for juvenile height growth (0.48–0.89). If we consider the cases where the highest total phenotypic variance was explained by the QTLs (bud702002 female 43.8%, height2000 female 35.6% and Δ2001 female 34.4%, respectively; these values, being simply the sum of PEV values for each single QTL, are probably slightly overestimated), then the QTLs detected would explain from 55 to 88% of the genetic variance for phenology, from 45 to 100% for growth and from 69 to 100% for Δ. Estimations of phenotypic explained variances (PEV) are biased by the small population size, and as reported in Zeng, Kao & Basten (1999) the higher the value of heritability, the lower the bias on estimation of the proportion of the variance explained by the QTL. Thus, estimates for phenology would be quite reliable, whereas an overestimation of PEV values was likely for growth and Δ QTLs. However, QTLs for adaptive traits explaining a large amount of the PEV have been identified for all traits analysed, supporting the hypothesis that natural populations adapted to contrasting climatic conditions constitute a suitable material to dissect the genetic architecture of these traits, and that some QTLs of large effect are always present.

Distribution of phenotypic variance explained by the QTLs

The foundation of adaptive traits evolution is still largely unknown. QTL analysis was expected to answer one of the questions about the genetic basis of adaptation: are alleles at a large number of loci of small effect fixed by natural selection (infinitesimal model, Fisher 1930), or rather, do few loci of relatively large effect play a major role in adaptation (oligogenic model, Orr 1998)? Orr's findings show that during adaptation directional selection fixes new mutations of large effect first and the distribution of factors fixed towards the optimum assumes an exponential form. As already shown by Tanksley (1993) and Kearsey & Farquhar (1998), the distribution of individual QTL effects showed an L shape (Fig. 2). Two main conclusions can be drawn from such a result: first, limits of the QTL analysis do not allow the detection of very small effect QTLs; second, all adaptive traits analysed show a clear L-distribution of QTL effects, with most of the QTLs explaining from low to moderate percentages of the phenotypic variance and few QTLs of large effect. Despite the bias due to the impossibility to identify in one experiment all loci involved in a complex trait, the genetic architecture of adaptive traits seems to be characterized by many more factors of small and moderate than large effect. Overall, this result agrees with previous findings. Several studies have dealt with the identification of QTLs influencing bud phenology (Bradshaw & Stettler 1995; Frewen et al. 2000; Jermstad et al. 2001; Saintagne et al. in preparation), growth (Bradshaw & Stettler 1995; Plomion et al. 1996; Conner et al. 1998; Emebiri et al. 1998; Wu 1998; Kaya, Sewell & Neale 1999; Lerceteau, Plomion & Andersson 2000; Tsarouhas et al. 2002; Weng et al. 2002) and Δ (Brendel et al. 2002) in forest trees. Generally, few QTLs of large effect were detected. However, due, mainly, to the difference of the experimental design, few QTLs of large effect were identified in poplar for bud phenology (Frewen et al. 2000), whereas 33 different QTLs, each being heavily significant and showing a relatively small effect, were identified in Douglas-fir (Jermstad et al. 2001), some of which were shown to be temporally stable.

image

Figure 2. Percentage of total phenotypic variance explained by the QTLs. All the adaptive traits analysed showed much more QTLs of small and moderate than large effect. Due the small population size QTLs accounting for a very small proportion of phenotypic variance could not be detected.

Download figure to PowerPoint

Clustering and co-localizations of QTLs

Although the large confidence intervals of QTLs can influence conclusions about co-localizations, interesting observations can be drawn from co-localizations of QTLs for different traits.

Co-localizations between different traits measuring the same adaptive trait
Phenology

Co-localizations among QTLs for different characters measuring the same adaptive trait were identified. For instance, bud flush was assessed by two distinct observations: bud and bud70. Of course, co-localizations between these QTLs are expected, being two different observations of the same trait and significantly correlated (discussed later). Thus, QTLs for bud and bud70 in LG1F, LG4F, LG6M, LG9F, LG11M, and LG12F, sharing the same map position and showing the same direction of the substitution effect, might correspond to the same QTL. However, on LG3F the two QTLs identified in 2000 and 2002 for bud and bud70, respectively, show a different map location and opposite sign of the substitution effect. In addition, separate QTLs for the two observations were identified on LG2F, LG5F (bud70), and LG7F (bud). These findings suggest that observations recorded at different stages of the bud flushing can allow the identification of QTLs affecting different developmental stages of leaf formation. Mapping candidate genes for bud flushing will be an interesting perspective to shed light on this process.

In 2002, bud set was also recorded. Bud set and bud flush are two phenological traits representing two different stages of bud activity. Although many factors contribute to both processes, it is known that timing of bud flush is primarily controlled by temperature, whereas short days seem to mainly influence bud set (Howe et al. 2000). Frewen et al. (2000) found that three QTLs for bud set were localized in linkage groups also containing QTLs for bud flush, but on a different chromosomal interval. In our case, despite the very weak phenotypic correlation between bud flush and bud set, among the four QTLs identified for bud set, three co-localized with QTLs for bud flush on LG2F, LG6M and LG9M, whereas one did not (LG8F), suggesting both common and specific genes. Following Lin, Shertz & Paterson (1995), the probability that these three matches would occur by chance is P = 0.0063, supporting the hypothesis of single QTLs affecting both traits. Both pleiotropy and tightly linked genes can explain these QTL co-localizations. In most cases fine mapping studies are needed to distinguish between the two possibilities. For the three co-localizations a similar trend in the substitution effects was observed (later bud flushing corresponding to an earlier bud set or vice-versa), indicating putatively underlying genes that inhibit or enhance the bud activity.

Growth

A similar situation was observed for growth that was assessed by measuring annual increments of height and diameter. Due to the complexity of the juvenile tree growth, the distinct physiological processes involved in height and diameter growth, and the fact that a shift of genes involved was suggested for tree growth over different years by Kremer (1992), the complex pattern of the QTLs detected was expected. However, we observed at one occasion a clustering of QTLs on the same chromosome (LG10F) with parallel substitution effect both for height increments and total diameter and height measurements over the 3 years, suggesting the presence of a unique and major QTL showing a pleiotropic effect on vigour. Moreover, on LG7F and LG11F, the detected QTLs for height and diameter had an opposite sign in the substitution effect and a quite clear different position, suggesting that two different loci could account for these QTLs. Genes which are involved in a general regulation of the growth process or structural genes implied on basal biochemical pathways (for instance on LG10F) as well as stage and tissue specific genes, such as primary versus secondary meristem-specific genes (for instance on LG7F and LG11F) are expected to be responsible for the QTLs identified. Furthermore, in this case, mapping candidate genes could help to draw some hypotheses about the molecular pathways involved.

Clustering of QTLs for different adaptive traits

A clustering among QTLs for different adaptive traits was also observed. Several linkage groups contained QTLs for the three adaptive traits analysed. This is not surprising, given the often significant phenotypic correlations among them and the wide distribution of QTLs over the 12 linkage groups that can lead to some co-localizations just by chance.

Growth and Δ: a possible adaptive compromise?

Co-localizations among QTLs may be expected when significant genetic correlations are observed among quantitative traits. In such a case, identification of co-localizing QTLs may be explained by a QTL having a pleiotropic effect on different traits. Furthermore, the sign of the substitution effect of co-localizing QTLs should be in this case in agreement with that of the correlation coefficient. In our experiment growth parameters were generally and negatively correlated with D. In five cases QTLs for D and growth-related traits coincided on the same linkage group. Opposite signs of the substitution effect were observed on the LG1F, LG6F, LG7F (the three QTLs for D and height2000) and LG8F, whereas concordant signs were observed on LG3M and LG7F (considering the QTL for diameter2002). Overall, this result agrees with the negative and weak phenotypic correlation between D and growth suggesting a genetic correlation between the two traits. Recently, Lauteri et al. (personal communication) found negative and strong genetic correlations between D and growth traits in chestnut seedlings, highlighting the interesting adaptive value of this kind of relation. However, because of the complex determinism of D, the sign of the relationship between plant growth and D itself is likely to be dependent on genotype–environment interactions and difficult to predict (Farquhar et al. 1989). For instance, Brendel et al. (2002) found a positive phenotypic correlation between d13C and mean ring width in Pinus pinaster (which corresponds to a negative correlation with D). A negative correlation between D and tree height was also found in Picea mariana on a dry site, but the relation was not confirmed in a site characterized by higher water availability (Flanagan & Johnsen 1995). Brendel et al. (2002) did not obtain significant genetic correlation between d13C and mean ring width. No co-localization of QTLs for the two traits was found in their case. In their study on P. pinaster a rather moderate genetic control was reported for both ring growth and d13C of cellulose. Furthermore these authors hypothesize a dependence of d13C, but not of growth, from photosynthetic assimilation, supporting the lack of co-localization of QTLs for these traits in P. pinaster. In the present study, the chestnut F1 progeny was obtained from a female parent originating from a drought-adapted population, which is expected to present genetic adaptation to Mediterranean-type. Lauteri et al. (1997) showed that this drought-adapted ecotype was well adapted to the Italian site where the full-sib family was grown, showing higher photosynthetic capacity, and higher stomata and mesophyll conductance when compared with the wet-adapted population which the male parent came from. In such a situation we can suppose that allele combinations for or of adaptation to drought conditions were detected by our QTL analysis in the female map. The negative correlation between D and growth and the opposite sign of four co-localized QTLs found in the female map are in agreement with a whole-plant structure adaptation to drought conditions proposed by Lauteri et al. (1997). Following this hypothesis, the relatively higher photosynthetic performances of the drought-adapted ecotype may be linked to an increased carbon allocation to roots, beneficial to take up deep and reliable soil water sources, and to a lower above-ground juvenile growth. Indeed, the opposite trends of D and growth, taking into account the four co-localized QTLs in the female parent, are consistent with this adaptive outline, confirming a whole-plant evolutionary strategy in this chestnut population.

Growth and bud set: an example of an adaptation strategy

Another interesting correlation from the adaptive point of view was observed among growth parameters and bud set in 2002. Both height and diameter increments were negatively and significantly correlated with bud set, which means that an early bud set (higher score in our measurement) corresponds to a lower growth rate and vice-versa. Three QTLs for bud set on LG6M, LG8F and LG9M were co-localized with height and diameter increments in 2002. The co-localization was very precise on the LG9M, but only approximate on the LG6M and LG8F. In all cases opposite signs for the allelic substitution effects were observed, possibly indicating, particularly in the case of LG9M, a unique QTL with a pleiotropic effect or tightly linked QTLs. This type of correlation was already observed in poplar (reviewed by Howe et al. 2000). As already observed in poplar, its adaptive value is easily understandable, genotypes which stop growing early, will be better acclimated to winter chilling, whereas genotypes showing a longer growth season, despite the higher tree vigour, are prone to frost damage. Reaching an optimum for a given environment is an important adaptive strategy for winter survival. This of course will require some common genetic mechanisms influencing both processes. It is worth observing that the exact co-localization on LG9M concerns the male parent, namely the ecotype coming from a humid and cool region, and occurs in 2002 when the rainfall was very abundant. However, since bud set was recorded only in 2002, caution is needed in interpreting this result.

CONCLUSION AND PERSPECTIVES

  1. Top of page
  2. ABSTRACT
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION AND PERSPECTIVES
  8. ACKNOWLEDGEMENTS
  9. REFERENCES

Our findings are consistent with previous experimental and theoretical issues on adaptation (Orr 1998; Hurme et al. 2000). We showed that individuals belonging to populations adapted to contrasting environments are a good material to dissect quantitative trait loci for adaptive traits. We found that adaptive traits seem to be regulated by many genetic factors of low and moderate effect. Nevertheless, few QTLs of large effect are always detected. These results give rise to an L-shaped distribution of the QTL effects which is predicted by Orr's model. We also found that most of the linkage groups contain factors controlling adaptive traits as expected considering their complexity. Furthermore, several co-localizations among QTLs for different adaptive traits confirm that adaptation is a process in which the reaching of an optimal value for a given trait can simultaneously influence by pleiotropy the optimum for another trait. This has to be considered particularly when selection strategies for adaptive traits are envisaged.

Comparative QTL mapping is a useful strategy to validate QTLs (Sewell & Neale 2000) allowing the identification of corresponding chromosomal regions affecting the same quantitative trait in different species. Correspondence of QTLs would support the hypothesis that some genes affecting quantitative traits have been conserved over a long period of evolutionary distance. Recently, a first comparison of the chestnut linkage map with that of the phylogenetically related species Quercus robur has been carried out using microsatellite markers (Barreneche et al. 2004). QTLs for bud phenology, growth and Δ have been localized in Q. robur (Saintagne et al. in preparation; Brendel et al. in preparation). The alignment of the 12 linkage groups of the two species and the mapping of candidate genes should make it possible to identify conserved QTLs for these important adaptive traits, and possibly the major genes involved in the adaptation of these two forest tree species.

ACKNOWLEDGEMENTS

  1. Top of page
  2. ABSTRACT
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION AND PERSPECTIVES
  8. ACKNOWLEDGEMENTS
  9. REFERENCES

We thank Antoine Kremer and Paul Sisco for critical comments on this manuscript; Marcello Cherubini, Giovanni De Simoni, Margherita Casasoli, Antonia Prudenzi and Adolfo Casasoli for assistance on field work; Simone Pacetti for computer assistance. This work was supported by the EU research project ‘CASCADE’(EVK2-CT-1999–00006). The authors declare that the experiments conducted for this publication comply with the current laws of France and Italy.

REFERENCES

  1. Top of page
  2. ABSTRACT
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION AND PERSPECTIVES
  8. ACKNOWLEDGEMENTS
  9. REFERENCES
  • Barber V.A., Juday G.P. & Finney B.P. (2000) Reduced growth of Alaskan white spruce in the twentieth century from temperature-induced drought stress. Nature 40, 668673.
  • Barreneche T., Casasoli M., Russell K., Akkak A., Meddour H., Plomion C., Villani F. & Kremer A. (2004) Comparative mapping between Quercus and Castanea using simple-sequence repeats (SSRs). Theoretical and Applied Genetics 108, 558566.
  • Beavis W.D. (1995) The power and deceit of QTL experiments: lessons from comparative QTL studies. In Proceedings of the 49th Annual Corn and Sorghum Industry Research Conference, (ed D.B. Wilkinson), pp. 250266. ASTA, Washington DC.
  • Bradshaw H.D. & Stettler R.F. (1995) Molecular genetics of growth and development in Populus. IV. Mapping QTLs with large effects on growth, form, and phenology traits in a forest tree. Genetics 139, 963973.
  • Brendel O., Pot D., Plomion C. & Rozenberg. & Guehl J.M. (2002) Genetic parameters and QTL analysis of δ13C and ring width in maritime pine. Plant, Cell and Environment 25, 945953.
  • Britvin E., Minkov D., Glikson L., Ronin Y. & Korol A. (2001) MultiQTL, an interactive package for genetic mapping of correlated quantitative trait complexes in multiple environments, Version 2.0. (Demo). In Plant and Animal Genome Conference IX, San Diego, CA, http://www.intl-pag.org/pag/9/abstracts/c01_01.html.
  • Brugnoli E. & Farquhar G.D. (2000) Photosynthetic fractionation of carbon isotopes. In Photosynthesis: Physiology and Metabolism. Advances in Photosynthesis (eds R.C.Leegood, T.D.Sharkey & S.Von Caemmerer), Vol. 9, Chapter 17, pp. 399434. Kluwer Academic Publishers, Dordrecht, The Netherlands.
  • Van Buijtenen J.P. (2001) Genomics and quantitative genetics. Canadian Journal of Forest Research 31, 617622.
  • Byrne M., Murrell L.C., Owen J.V., Kriedemann P., Williams E.R. & Moran G.F. (1997) Identification and mode of action of quantitative trait affecting seedling height and leaf area in Eucalyptus nitens. Theoretical and Applied Genetics 94, 674681.
  • Casasoli M., Mattioni C., Cherubini M. & Villani F. (2001) A genetic linkage map of European chestnut (Castanea sativa Mill.) based on RAPD, ISSR and isozyme markers. Theoretical and Applied Genetics 102, 11901199.
  • Churchill G.A. & Doerge R.W. (1994) Empirical threshold values for quantitative trait mapping. Genetics 138, 963971.
  • Conner P.J., Brown S.K. & Weeden N.F. (1998) Molecular-marker analysis of quantitative traits for growth and development in juvenile apple trees. Theoretical and Applied Genetics 96, 10271035.
  • Emebiri L.C., Devey M.E., Matheson A.C. & Slee M.U. (1998) Age-related changes in the expression of QTLs for growth in radiata pine seedlings. Theoretical and Applied Genetics 97, 10531061.
  • Farquhar G.D., Ehleringer J.R. & Hubick K.T. (1989) Carbon isotope discrimination and photosynthesis. Annual Review of Plant Physiology and Molecular Biology 40, 503537.
  • Fisher R.A. (1930) The Genetical Theory of Natural Selection. Oxford University Press, Oxford, UK.
  • Flanagan L.B. & Johnsen K.H. (1995) Genetic variation in carbon isotope discrimination and its relationship to growth under field conditions in full-sib families of Picea mariana. Canadian Journal of Forest Research 25, 3947.
  • Frewen B.E., Chen T.H.H., Howe G.T., Davis J., Rohde A., Boerjan W. & Bradshaw H.D. (2000) Quantitative trait loci and candidate gene mapping of bud set and bud flush in Populus. Genetics 154, 837845.
  • Grattapaglia D., Bertolucci F.L. & Sederoff R.R. (1995) Genetic mapping of QTLs controlling vegetative propagation in Eucalyptus grandis and E. urophylla using a pseudo-test cross strategy and RAPD markers. Theoretical and Applied Genetics 90, 933947.
  • Hall A.E., Richards R.A., Condon A.G., Wright G.C. & Farquhar G.D. (1994) Carbon isotope discrimination and plant breeding. Plant Breeding Review 4, 81113.
  • Howe G.T., Saruul P., Davis J. & Chen T.H.H. (2000) Quantitative genetics of bud phenology, frost damage, and winter survival in an F2 family of hybrid poplars. Theoretical and Applied Genetics 101, 632642.
  • Huntley B. & Birks H.J.B. (1983) An Atlas of Past and Present Pollen Maps for Europe, 0–13000 Years Ago. Cambridge University Press, Cambridge, UK.
  • Hurme P., Sillanpää M.J., Arjas E., Repo T. & Savolainen O. (2000) Genetic basis of climatic adaptation in Scot pine by Bayesian quantitative trait locus analysis. Genetics 156, 13091322.
  • Jermstad K.D., Bassoni D.L., Jech K.S., Wheeler N.C. & Neale D.B. (2001) Mapping of quantitative trait loci controlling adaptive traits in coastal Douglas-fir. I. Timing of vegetative bud flush. Theoretical and Applied Genetics 102, 11421151.
  • Kao C.H., Zeng Z.B. & Teasdale R.D. (1999) Multiple interval mapping for quantitative trait loci. Genetics 152, 12031216.
  • Kaya Z., Sewell M.M. & Neale D.B. (1999) Identification of quantitative trait loci influencing annual height- and diameter-increment growth in loblolly pine (Pinus taeda L.). Theoretical and Applied Genetics 98, 586592.
  • Kearsey M.J. & Farquhar A.G. (1998) QTL analysis in plants: where are we now? Heredity 80, 137142.
  • Kremer A. (1992) Prediction of age-age correlations of total height based on serial correlations between height increments in maritime pine (Pinus pinaster Ait.). Theoretical and Applied Genetics 85, 152158.
  • Lander E.S., Green P., Abrahamson J., Barlow A., Daly M.J., Lincoln S.E. & Newburg L. (1987) MAPMAKER: an interactive computer package for constructing primary genetic linkage maps of experimental and natural populations. Genomics 1, 174181.
  • Lauteri M., Monteverdi M.C., Sansotta A., Cherubini M., Spaccino L. & Villani F. (1999) Adaptation to drought in European chestnut. Evidences from a hybrid zone and from controlled crosses between drought and wet adapted populations. Acta Horticolturae 494, 345353.
  • Lauteri M., Scartazza A., Guido M.C. & Brugnoli E. (1997) Genetic variation in photosynthetic capacity, carbon isotope discrimination and mesophyll conductance in provenances of Castanea sativa adapted to different environments. Functional Ecology 11, 675683.
  • Lerceteau E., Plomion C. & Andersson B. (2000) AFLP mapping and detection of quantitative trait loci (QTLs) for economically important traits in Pinus sylvestris: a preliminary study. Molecular Breeding 6, 451458.
  • Lin Y.R., Shertz K.F. & Paterson A.H. (1995) Comparative analysis of QTLs affecting plant height and maturity across the Poaceae, in reference to an interspecific sorghum population. Genetics 141, 391411.
  • Nelson C.D., Nance W.L. & Doudrick R.L. (1993) A partial genetic linkage map of slash pine (Pinus elliottii Engelm. var. elliottii) based on random amplified polymorphic DNAs. Theoretical and Applied Genetics 87, 145151.
  • Orr A.H. (1998) The population genetics of adaptation: the distribution of factors fixed during adaptive evolution. Evolution 52, 935949.
  • Parmesan C. & Yohe G. (2003) A globally coherent fingerprint of climate change impacts across natural systems. Nature 421, 3742.
  • Pigliucci M., Villani F. & Benedettelli S. (1990) Geographic and climatic factors associated with the spatial structure of gene frequencies in Castanea sativa Mill. from Turkey. Journal of Genetics 69, 141149.
  • Pliura A. & Eriksson G. (2002) Genetic variation in juvenile height and biomass of open-pollinated families of six Castanea sativa Mill. populations in a 2 × 2 factorial temperature × watering experiment. Silvae Genetica 51, 152160.
  • Plomion C., Durel C.E. & O'Malley D.M. (1996) Genetic dissection of height in maritime pine seedlings raised under accelerated growth conditions. Theoretical and Applied Genetics 93, 849858.
  • Root T.L., Price J.T., Hall K.R., Schneider S.H., Rosenzweig C. & Pounds J.A. (2003) Fingerprints of global warming on wild animals and plants. Nature 421, 5760.
  • Sewell M.M. & Neale D.B. (2000) Mapping quantitative traits in forest trees. In Molecular Biology of Woody Plants (eds S.M.Jain & S.C.Minocha), pp. 407423. Kluwer Academic Publishers, Dordrecht, The Netherlands.
  • Tanksley S.D. (1993) Mapping polygenes. Annual Review of Genetics 27, 205233.
  • Tognetti R., Michelozzi M. & Giovannelli A. (1997) Geographical variation in water relations, hydraulic architecture and terpene composition of Aleppo pine seedlings from Italian provinces. Tree Physiology 17, 241250.
  • Tsarouhas V., Gullberg U. & Lagercrantz U. (2002) An AFLP and RFLP linkage map and quantitative trait locus (QTL) analysis of growth traits in Salix. Theoretical and Applied Genetics 105, 277288.
  • Villani F., Pigliucci M., Benedettelli S. & Cherubini M. (1991) Genetic differentiation among Turkish chestnut (Castanea sativa Mill.) populations. Heredity 66, 131136.
  • Visscher P.M., Thompson R. & Haley C.S. (1996) Confidence intervals in QTL mapping by bootstrapping. Genetics 143, 10131020.
  • Voorrips R.E. (2002) MapChart: Software for the graphical presentation of linkage maps and QTLs. Journal of Heredity 93, 7778.
  • Weng C., Kubisiak T.L., Nelson C.D. & Stine M. (2002) Mapping quantitative trait loci controlling early growth in a (longleaf pine × slash pine) × slash pine BC1 family. Theoretical and Applied Genetics 104, 852852.
  • Wu R.L. (1998) Genetic mapping of QTLs affecting tree growth and architecture in Populus: implication for ideotype breeding. Theoretical and Applied Genetics 96, 447457.
  • Wu R., Ma C.X., Zhu J. & Casella G. (2002) Mapping epigenetic quantitative trait loci (QTL) altering a developmental trajectory. Genome 45, 2833.
  • Zeng Z.B., Kao C.H. & Basten C.J. (1999) Estimating the genetic architecture of quantitative traits. Genetic Research 74, 279289.
  • Zhang J. & Marshall J.D. (1995) Variation in carbon isotope discrimination and photosynthetic gas exchange among populations of Pseudotsuga menziesii and Pinus ponderosa in different environments. Functional Ecology 9, 402412.