Seeds of broadleaf species that were collected from open-pollinated trees at the Petawawa Research Forest (46°00N, 77°42′W) (white birch and trembling aspen) and at Kemptville (45°02′N, 75°39W) (balsam poplar), ON, Canada, were provided by the National Tree Seed Centre at Fredericton, NB, Canada. Seeds were sown into 3.8 × 21 cm SC-10 Super Cell tubes filled with 2:1 peat moss/vermiculite (v/v) mixture in late June 2010 and grown in a greenhouse at the Ontario Forest Research Institute in Sault Ste. Marie, ON, Canada (46° 30′N, 84° 20′W). The greenhouse was programmed to provide 26°C (day)/18°C (night) temperatures and a 16-h photoperiod. Seedlings were watered as required and fertilized weekly with 20-8-20 (N-P-K) (Plant Products Co., Ltd, Brampton, ON, Canada) at 100 ppm N. Beginning early September 2010, seedlings were moved outdoors and fertilization was adjusted to 20-20-20 at 50 ppm N. Starting in mid-October, fertilization was discontinued but seedlings were watered as needed. By the end of November, leaves had abscised from all seedlings. In mid-December, seedlings were sealed in plastic bags, boxed, and stored in a freezer at -3°C. In mid-March, after 3 months in frozen storage, seedlings were moved to refrigerated storage at 2°C. In early May, seedlings were removed from the boxes and transplanted into 15-cm diameter pots. The transplanted seedlings were grown outside under natural environmental conditions for 1 year, and watered and fertilized as needed during the growing season.

Lodgepole pine seedlings were initially grown in containers in a greenhouse at Tree Time Services Inc./Coast to Coast Reforestation in Smoky Lake facility, Smoky Lake, AB, Canada, from open-pollinated seeds collected from the area southwest of Whitecourt, AB, Canada (54° 04′N, 116° 41′W). Black spruce, white spruce, and jack pine seedlings were initially grown in containers at the Millson Forestry Service Inc. at Timmins, ON, Canada, with seeds obtained from tree improvement seed orchards established for the nearby Martel Forest (47°50′–48°28′N, and 82°15′–83°25′W). Following overwinter storage, seedlings were shipped to Sault Ste Marie, ON, Canada. Upon arrival in early June 2013, the 1-year-old container seedlings were transplanted into 4″ square (10 cm side × 15 cm deep) pots filled with 2:1 peat moss/vermiculite (v/v) mixture and grown in a greenhouse at the Ontario Forest Research Institute. The greenhouse received natural photoperiods with temperatures 2 to 5°C above ambient conditions. Seedlings were watered as required and fertilized weekly with 20-8-20 (N-P-K) (Plant Products Co., Ltd, Brampton, ON, Canada) at 100 ppm N for a month before being moved outdoors in early July. Watering and fertilization continued as required until early September when fertilization was adjusted to 20-20-20 at 50 ppm N and then discontinued in mid-October.

Both broadleaf and conifer seedlings remained outdoors to undergo natural hardening and dehardening under ambient photoperiod and temperatures. Average seedling height at the outset of budburst experiments in the fall of 2014 was 121 cm for trembling aspen, 77 cm for balsam poplar, 142 cm for white birch, 54 cm for black spruce, 43 cm for white spruce, 44 cm for jack pine, and 36 cm for lodgepole pine.

A series of budburst experiments were carried out to produce different levels of chilling fulfillment and quantify the corresponding heat requirement for budburst and the subsequent quality of budburst and shoot growth. The experiments were conducted in two computer-controlled walk-in climate chambers (i.e., two replications) set at 15°C (day)/5°C (night) and a 14-h photoperiod at 350 μmol m^-2 s^-1 photosynthetic photon flux density (PPFD). This temperature and photoperiod regime is typical of late April and early May conditions in northern Ontario, which is when leaf out generally occurs in the study area (Figure 1). Batches of seedlings were sequentially moved into the climate chambers from outdoors at 10-day intervals between October 1st and January 1st and 20-day intervals thereafter to mid-April, which resulted in a time series of 16 budburst experiments for each species (15 in the climate chambers and 1 outdoors). Seedlings going into these 16 budburst experiments received different levels of chilling as they were continuously exposed to outdoor chilling temperatures before being forced to burst buds in the climate chambers. When outdoor temperatures were below -5°C, to minimize temperature shock seedlings were covered in large plastic bags and kept at 0°C overnight before their transfer to climate chambers. In each of the budburst experiments, eight seedlings from each conifer species and four seedlings from each broadleaf species were placed in each of the two climate chambers (replicates). The total number of seedlings used was 256 for each conifer species and 128 for each broadleaf species, except for balsam poplar which had 110 seedlings.

Budburst was assessed daily during the experiments. A bud was considered to have flushed when its scales were broken and new green foliage was clearly visible. Budburst was considered abnormal if it started first in lower branches, stems, and roots in broadleaves and if bud and needle expansion were constrained in conifers. The time and heat accumulations required for budburst were determined for individual seedlings. For seedlings that did not flush throughout the experiments, a maximum heat accumulation was calculated from the start of the experiment to May 31 when trees growing outdoors had begun flushing. Shoot growth was assessed on individual seedlings by measuring the length of the longest shoot 30 days after budburst.

The chilling–heat relationship was determined by species with hourly temperature data from a HOBO weather station (three sensors set at 0.80 m height) installed beside the outdoor seedlings and records from inside climate chambers. The chilling accumulation was calculated using the Sarvas Chilling Rate Model (Sarvas, 1974) which was reformulated by Hänninen (1990) and Kramer (1994),

where CH is weighted chilling hours, T is the actual temperature imposed on seedlings, and -3.4, 3.5, and 10.4°C are three threshold values for lower, optimum, and upper limits of chilling rate. Because of the possible negative effect of high temperatures on the effectiveness of chilling (Young, 1992; Arora et al., 2003), chilling accumulation was calculated from late September to the time of transfer to the climate chambers. Similarly, possible chilling in the climate chambers under cooler night temperature (5°C for 8 h) was not considered. For the budburst experiment with outdoor seedlings, chilling accumulation continued until late April when outdoor daily temperature was generally below 15°C. As jack pine, lodgepole pine, and balsam poplar started budburst before late April, their chilling accumulation to the time of budburst was calculated for individual trees.

The accumulation of heat (expressed as cumulative growing degree hours above 0°C; see Man and Lu, 2010) included high temperatures in the climate chambers and under outdoor conditions between March 1 and the time seedlings were brought into the climate chambers (Figure 1).

The relationship between chilling accumulation and heat requirement for budburst was investigated by fitting a 3-parameter exponential decay curve following Cannell and Smith (1983) and Hannerz et al. (2003) in the form:

where y is the heat requirement for budburst, x is the chilling accumulation (i.e., cumulative weighted chilling hours), e is the base of natural logarithm, and a, b, and c are the model parameters. The parameter c reflects the rate of chilling completion, i.e., a greater value indicates earlier and faster completion of chilling (Figure 2). We used the method documented by Hannerz et al. (2003) to determine chilling requirements for budburst, which was set as 1.05^∗a, a point slightly before the lowest value of the fitted curves (Figure 2). This method did not work well for trembling aspen and balsam poplar as they had negative estimates for parameter a. For these species, a was estimated as the lowest value from the observed heat requirement for budburst.

Heat requirement for budburst and shoot growth of individual seedlings were subjected to a 2-way analysis of variance based on the linear model:

where y_ijkl is the observation of the l^th seedling of the j^th species tested in the k^th replication of the i^th experiment; μ is the overall mean; E_i is the i^th experiment; S_j is the j^th species; ES_ij is the interaction between i^th experiment and j^th species; and 𝜀_ijkl is the residual. One-way ANOVA was performed to test differences among tree species in the rate of chilling completion (parameter c of the exponential curve) and chilling requirement determined from chilling–heat relationships for tree species and replications (individual climate chambers). Multiple contrasts were conducted to examine differences in heat requirement for budburst and shoot growth among experiments within a species or among species within an experiment, and in rate of chilling completion and chilling requirement among species. The P-values were corrected using Tukey’s method in SAS 9.3 (SAS Institute Inc., 2011).

Analysis of variance indicated that both budburst experiments (i.e., timing) and tree species significantly affected heat requirement for budburst (p < 0.001 for species, budburst experiments, or species by experiment interaction). The amount of heat required for budburst decreased progressively with experiment start time in all species, and was greater in broadleaf than conifer species in experiments conducted before December 10, and greater in spruce than pine across almost all budburst experiments (see Supplementary Table S1). Among species, the change in heat requirement became insignificant over time, occurring first for black spruce and white spruce (by October 31) and last for trembling aspen (by December 20) (Figure 2 and Supplementary Table S1).

The rate of chilling completion (i.e., estimate of parameter c in the fitted exponential curves) was highest for black spruce and white spruce and lowest for trembling aspen and balsam poplar, with intermediate values for white birch, jack pine, and lodgepole pine (Figure 2 and Supplementary Table S2). The chilling requirement (amount of chilling accumulation when chilling need was fully met) was less than 600 chilling hours (prior to January 1) for black spruce and white spruce and over 1100 chilling hours for trembling aspen and lodgepole pine. Balsam poplar, white birch, and jack pine were intermediate, completing their chilling by late March (about 900 chilling hours).

The heat requirement for budburst after the completion of chilling, determined from the fitted exponential curves and expressed as cumulative growing degree hours was, in descending order, 9204 for black spruce, 6831 for white spruce, 6109 for white birch, 4816 for trembling aspen, 3062 for balsam poplar, 2691 for lodgepole pine, and 2114 for jack pine (Figure 2 and Supplementary Table S2).

Mean heat requirements and coefficient of variations (CV) were substantially higher before chilling requirements were fulfilled, especially in broadleaves (Figures 2, 3 and Supplementary Tables S1, S2). Budburst experiments started before the end of November, when chilling accumulation was < 500 chilling hours, produced abnormal budburst in broadleaves, which typically included early and erratic burst of buds in lower branches, as well as the burst of basal buds at lower stems and on roots (aspen only); some aspen and balsam poplar had not burst buds after 243 days in the climate chambers (Figure 4). Among the three broadleaf species, trembling aspen had significantly longer new shoots (from lower stem sprouting and suckering) in the fall budburst experiments (<500 chilling hours) than the winter and spring budburst experiments (>500 chilling hours) compared to balsam poplar and white birch (p < 0.01 for species, budburst experiments, or species by experiment interaction, see Figure 5).

Insufficient chilling also delayed the time of budburst, increased variations among individual trees, and caused abnormal budburst in conifers (Figures 2–4 and Supplementary Table S1). The typical patterns of abnormal budburst in conifers included constrained bud expansion, shortened new needle length, and shorter terminal than lateral shoots. The effect was strongest in lodgepole pine and weakest in jack pine, with the spruces intermediate. Insufficient chilling did not, however, stop conifers from bursting bud, but when chilling accumulation was less than 300 chilling hours significantly shortened the length of new shoots (Figure 5).