Skip to main content

Quantitative genetic architecture of adaptive phenology traits in the deciduous tree, Populus trichocarpa (Torr. & Gray)

Cite this dataset

Richards, Thomas (2020). Quantitative genetic architecture of adaptive phenology traits in the deciduous tree, Populus trichocarpa (Torr. & Gray) [Dataset]. Dryad.


In a warming climate, the ability to accurately predict and track shifting environmental conditions will be fundamental for plant survival. Environmental cues define the transitions between growth and dormancy as plants synchronise development with favourable environmental conditions, however these cues are predicted to change under future climate projections which may have profound impacts on tree survival and growth. Here, we use a quantitative genetic approach to estimate the genetic basis of spring and autumn phenology in Populus trichocarpa to determine this species capacity for climate adaptation. We measured bud burst, leaf coloration, and leaf senescence traits across two years (2017- 2018) and combine these observations with measures of lifetime growth to determine how genetic correlations between phenology and growth may facilitate or constrain adaptation. Timing of transitions differed between years, although we found strong cross year genetic correlations in all traits, suggesting that genotypes respond in consistent ways to seasonal cues. Spring and autumn phenology were correlated with lifetime growth, where genotypes that burst leaves early and shed them late had the highest lifetime growth. We also identified substantial heritable variation in the timing of all phenological transitions (h2 = 0.5-0.8) and in lifetime growth (h2 = 0.8). The combination of abundant additive variation and favourable genetic correlations in phenology traits suggests that cultivated varieties of P. Trichocarpa have the capability to create populations which may adapt their phenology to climatic changes without negative impacts on growth.


Population characteristics and phenotypic measurements

Here we present a quantitative genetics analysis of phenology measurements taken from 564 mature P. Trichocarpa trees in a plantation at Krusenberg (59°44'44.2“N 17°40'31.5“E) in central Sweden. Material in this plantation was originally generated from 9 female and 10 male trees collected over a latitudinal range from 44 to 60o in North America (Supplementary Table 1), which were randomly crossed to produce 34 families. From these families, a total of 120 half sib, full sib, and unrelated trees (onward referred to as genotypes) were clonally replicated with between 1 and 20 (median = 6) individual trees per genotype represented in the study population. The plantation was established in 2003 on a flat, homogeneous area of agricultural field ~275 m × 40 m. Individuals were planted at 3.5 m quadratic spacing in a randomised design. The experiment was systematically thinned in March 2013 leaving 564 trees in an ~3.5 × 7 m diamond spacing.

To assess variation in phenology we included the traits that best define the major milestones of phenology during the annual growth cycle. For spring development, bud burst (bb) was scored on a scale from 1 to 5, with stage bb2 representing initial shoot emergence, bb3 leaf primordia exposed, bb4 leaves half shed with bud scales dropped and bb5 leaves completely shed. Autumn phenology was scored on a scale of 1–8 based on a continuous range of crown colouring from col1 (100% green) to col8 (100% yellow). Leaf senescence (ls) was measured on a 3-point scale where ls1 = full foliage, ls2 = half leaves remaining and stage ls3 full defoliated. We estimated growing season length by the duration of photosynthetically active leaf canopy. Canopy duration (CD) was defined as the period between the beginning of budburst (bb2) and beginning of leaf yellowing (co3). Phenology measurements were taken every 2–5 days during the spring and autumn seasons in 2017 and 2018. Lifetime growth was determined by measuring diameter at breast height (DBH) in 2017

While late season growth cessation is best described by bud set, this trait is difficult to accurately measure in mature trees. Due to the difficulty of determining the exact transitions between beginning of growth, cessation of growth and bud development in grown trees, we split these seasonal transitions into 5 biologically relevant proxy stages which describe the start and end of seasonal transition; (bb2) the first bud emergence, (bb5) full leaf emergence, (col2) first stage of yellowing, (col5) complete yellowing, (ls3) day of full leaf shed. These phenology traits are combined with growth of the tree at 14 years of age as measured by cross calliper measurement (DBH) in 2017.

Data imputation

Screening was conducted at intervals of 2–5 days meaning that individual trees occasionally passed through developmental stages between screenings. To account for these missing estimates of the day of transition we estimated the transition days for each developmental stage using local regression models (LOESS) fit to each individual tree. Models were fit through the data point of the first day an individual tree was observed at a stage transition and day estimates were calculated for any stage transitions for which there was no direct observation (Fig. S2). This method estimates a non-linear developmental curve which is not constrained to fit any a-priori mathematic distribution for each individual tree and allows estimation of the day in which trees passed each developmental stage and inclusion of individuals which were not observed at important developmental transitions in the analysis. As extrapolation beyond the range of the observed data is unreliable using this method, we only retained estimates that were bounded by observations on either side. We validated this method by comparing estimated values with direct observations to ensure that estimated transitions accurately represented the observed data (Pearsons R = 0.98-1).

Usage notes

There are two data sets associated with this manuscript. The Full.csv, which contains all phenotypic measures and design of the field experiment, and newped.csv, which describes the pedigree structure of the individuals in the field experiment. Variables are described below, but I encourage cross referencing with the description in the manuscript:

Full.csv - data on phenology transitions in poplar 


"row"     - planted row in plantation

"key"     - combination row- place in row in plantation (spatial description) 

"clone"   - clone name

"yr"      - year of measurment

"bb2"     - day initial stage of bud burst (see methods/ paper for further description of all phenotypic measurements)

"lo4"     -  day stage 4 of bud burst

"lo5"     - day final stage 5 of bud burst - leafs full open

"estbb2"  - Imputed/ estimated day for stage 2 bud burst

"estbb4"  - Imputed/ estimated day for stage 4 bud burst

"estbb5"  - Imputed/ estimated day for stage 5 bud burst

"yel3"    - day stage 3 of leaf yellowing 

"yel8"   - sday tage 8 of leaf yellowing (final stage tree completely yellow)

"estco3"  - imputed/estimated day stage 3 of leaf yellowing 

"estco8"  - imputed/estimated day stage 8 of leaf yellowing 

"ls2"     - day leaf shed stage 2

"ls3"     - day leaf shed stage 3

"estls2"  -imputed/estimated day leaf shed stage 2

"estls3"  -imputed/estimated day leaf shed stage 3

"len"     - length of growing season 

"animal"  - indentifier to link individuals to pedigree file (ID)

"dbh16"   - diameter breast height (DBH) 2016

"dbh17" - diameter breast height (DBH) 2016


newped.csv - pedigree file for trees in dataset

"id"     -- indentifier to link individuals to pedigree file (ANIMAL)

"mother" - dam ID

"father" - sire ID