Autumn Frost Hardiness in Six Tree Species Subjected to Different Winter Storage Methods and Planting Dates in Iceland

Abstract

Winter storage of seedlings in freezers reduces the amount of heat sum available for growth in the following growing season compared to seedlings stored outdoors. To test the effects of a reduced growing period on the autumn frost hardiness of the six species most used in Icelandic afforestation, seedlings were stored outdoors or in a freezer during winter. In spring, the seedlings were planted on 24 May, 7 June, 21 June, and 5 July, and the frost hardiness of all treatments was tested on 12 and 26 September. In general, the probability of freezing damage increased with a later planting date, with outdoor-stored seedlings having the lowest probability of damage. The timing of frost events was of great importance; the later the freezing date, the less damage was observed. Growth cessation occurred at different times for each species, and they responded differently to the reduced heat sum. Lodgepole pine and birch accumulated the most frost hardiness in September. Sitka spruce had less autumn frost hardiness than Lutz spruce. Hybrid larch accumulated less frost hardiness than Russian larch and was most sensitive to the reduced heat sum. The results can be used to determine which species should be prioritised in frozen storage with regard to Iceland‘s short growing season.

1. Introduction

Perennial woody species native to the temperate zone undergo an annual growth cycle, including an accumulation of cold hardiness in autumn [1,2,3]. This annual cycle is determined genetically but controlled by environmental cues, in which annual temperature and photoperiod are the main factors controlling the complex physiological process of cold hardiness [2,4,5,6]. In this context, the length of the growing season is vital. It is usually defined in terms of temperature and can be quantified as duration or as a combination of temperature and time, i.e., heat sum. A relationship between heat sum and night length determines growth cessation [7], which is a prerequisite for the accumulation of cold hardiness [8,9,10,11]. Therefore, late growth cessation renders the plants more vulnerable to autumn frost damage [12,13]. These factors need to be considered when deciding on winter storage of forest seedlings [14,15].

For decades, nurseries in northern Europe and North America have used freezers to store forest seedlings during the winter [16,17,18,19]. The storage method protects seedling quality, as it reduces the effects of adverse weather conditions compared to potentially damaging outdoor temperatures [20,21]. When seedlings are stored in the dark in freezers (at −3 to −5 °C), the environmental cues (especially increasing heat) for the seedlings to decrease their dormancy in spring and begin growth are not present, and they are consequently still dormant at the time of planting [22,23]. Therefore, freezer-stored seedlings have a late onset of dehardening compared to seedlings stored outdoors during winter and the beginning of the growing season. Their hardening, in the autumn after planting, will also begin later than for seedlings stored outside, making them more susceptible to early autumn frost during their first year in the field [10,24]. Late hardening resulted in lower survival of freezer-stored Norway spruce (Picea abies Karst.) seedlings planted after mid-June in Finland [13] and had a negative impact on cold hardiness in Scots pine (Pinus sylvestris L.) and Norway spruce seedlings planted from mid-June to the end of July in Sweden [25].

The species and seed sources used in this study are commonly used in Icelandic afforestation: Downy birch (Betula pubescens Ehrh.), Russian larch (Larix sukaczewii Dyl.), hybrid larch (Larix decidua × L. sukaczewii), Sitka spruce (Picea sitchensins (Bong.) Carr.), Lutz spruce (Picea × lutzii Littl.), and lodgepole pine (Pinus contorta Dougl. Ex. Loud.) of two seed sources, Närlinge and Skagway. Since 2018, the planting of forest seedlings in Iceland has doubled [26,27], which requires greater efficiency in work processes during the nursery phase. The increased number of seedlings being produced means that it is physically difficult to grade, process, and ship stock in a short time during the spring. Traditionally, the majority of forest seedlings have been stored outdoors during the winter [28]. The use of freezers in Icelandic nurseries would be beneficial as it helps to spread out the workload since grading and packing can be completed during the autumn and early winter [29]. However, it must be kept in mind that the farther north you plant seedlings, the shorter the growing season [7]. This is also the case for Iceland, as the accumulated heat sum is higher for the western and southern parts of the country compared to the northern part [30]. Since the length of storage during winter can reduce the heat sum available for seedlings in the first growing season and thereby affect their ability to undergo the necessary physiological processes to accumulate cold hardiness, care should be taken when deciding which species will be stored in freezers during the winter and where freezer-stored seedlings should be planted to avoid frost damage during autumn and the first winter in the field.

To maximise outplanting success, forest seedling production and field establishment should be considered as a single holistic process. This also applies in context to the methods used in nurseries to store seedlings during winter [31]. Therefore, the aim of this study was (1) to detect the species/seed sources most sensitive to freezer storage during winter compared to outdoor storage in a nursery; and (2) to investigate how a delayed planting date reduces the accumulated heat sum of the growing season and how this influences autumn frost hardiness. The cumulative heat sum and the likelihood of damaging autumn frosts were also observed by examining long-term weather data from various locations in Iceland.

2. Materials and Methods

2.1. Plant Material

Seedlings for the experiment were grown in the Sólskógar nursery (65°66′ N, 18°11′ W) in Akureyri in northern Iceland (

Table 1. Origin of plant materials, sowing date, growing conditions, and information about short-day treatment in the nursery for all species and seed sources.

Species	Site of Seed Source	Latitude of Seed Source	Origin	Latitude of Origin	Sowing Date	Moved Outdoors	Short-Day Treatment (SDT)
Downy birch	Bolholt	1	Bæjarstaður	63°93′ N *	9 Jun	5 Jul	Without SDT
Lodgepole p.	Närlinge	60°03′ N 2	Canada	60°44′–63°40′ N 2	9 Apr	15 Jun	Without SDT
Lodgepole p.	Iceland	1	Skagway	59°27′ N *	9 Apr	15 Jun	Without SDT
Russian larch	Lassinmaa	62°04′ N 3	Raivola	60°14′ N 3	21 Jun	Early Nov	16–19 Aug
Hybrid larch	Hrymur	65°43′ N 4	4	4	21 Jun	Early Nov	16–19 Aug
Sitka spruce	Stálpastaðir	64°31′ N *	Cordova	60°32′ N 5	25 Mar	Early Nov	2–12 Aug
Lutz spruce	Haukadalur	64°19′ N *	Seward	60°06′ N 6	25 Mar	Early Nov	2–12 Aug

¹ Collected in various places in Iceland. ² Swedish seed orchard [32]. ³ Finnish seed orchard [33]. ⁴ Icelandic indoor seed orchard. Mother trees were grafts from selected European larch and Russian larch trees from various locations in Iceland ⁵ [34], ⁶ [35]. * GPS coordinates for these locations are approximations from Google Maps.

).

All species were sown in spring 2021 in plastic conical multipots (BCC, HIKO-93, Gislaved, Sweden). Each pot had a volume of 93 cm³, with 526 cells per m² (40 cavities per tray). The growing substrate used was medium–fine grade sphagnum peat, fertilised (N-P-K in the proportions of 16-9-20) and limed (1.5 kg/m³) (Kekkilä FPM 420 F6, Kekkilä Oy, Vantaa, Finland). The seeds were germinated and grown in a plastic greenhouse during the first part of the summer and moved outdoors. Fertilisation began 4 weeks after germination using electrical conductivity (EC) at a rate of 0.8–1.0 m S/cm throughout the summer. A mineral nutrient solution (Kekkila superex, NPK 19-4-20, Kekkilä Oy, Vantaa, Finland) was dissolved in the irrigation water. Fertilisation ended in late September. To induce bud induction and cold hardiness, opaque shade cloth was used to shorten the daylength (SDT) in the nursery for 16 h per day [14,20].

Half of the seedlings were packed in cardboard boxes and stored frozen (at −2.5 to −3 °C) on 24–26 January 2022. Spruce species and both seed sources of lodgepole pine were sprayed with fungicide (Rovral, Aquaflow. BASF AGRO B.V. Arnhem NL, Switzerland, active ingredient: Iprodione) before packing to prevent mould (Botrytis spp.) outbreaks during storage [36]. The other half were stored outdoors during the winter. Before planting, freezer-stored seedlings were moved to a cooler and thawed for 3 days at 10 °C in the dark in open boxes. After thawing, seedlings were stored outside in shade before planting.

The initial height of seedlings was measured before planting for 60 freezer-stored and 60 outdoor-stored seedlings per species at the first planting date. During summer 2022, while outdoor-stored seedlings were kept in the nursery until planting, the increment in annual growth was measured on 13 June, 5 July, and 19 July. These measurements were performed on 40 randomly selected seedlings per species each time. Seedling mortality and damage were counted in the field trial on 25 August.

2.2. Field Planting and Experimental Design

In the summer of 2022, the experimental seedlings were planted in a field trial on different dates so that they would receive the natural temperature sums and day lengths, as in the case of traditional planting. The field trial was located 6 km south of Akureyri (65°61′ N, 18°08′ W) on a slope facing east. The vegetation of the site was characterized by grass species (Poa sp.) and the soil was considered fertile. All vegetation was removed from the ground with an excavator, leaving bare soil for planting. Seedlings were planted in rows with 20 cm intervals and 20 cm between rows. Outdoor-stored seedlings were only planted on the first occasion (24 May) to serve as a control in the coming freezing tests.

The experiment was set up in four blocks that were ordered horizontally to the slope to reduce the effects of variation in topography and soil moisture within each block. The experimental plots were set up using a split-plot design. Each block was divided into four main plots, each representing one of the planting dates: 24 May, 7 June, 21 June, and 5 July. The main plot representing planting on 24 May was randomly divided into 14 species/seed sources plots (split plot) from both outdoor storage and freezer storage, creating treatments OS and F1 (

Table 2. Treatment names combining winter storage method and planting date, number of seedlings planted for each treatment on each date, and number of seedlings in freezing tests on 12 and 26 September for all species/seed sources. This is a total of seven, since lodgepole pine has two seed sources.

Treat-ments	Winter Storage Method	Planting Date	Number of Species and Seed Sources	Number of Seedlings in Each Species Plot	Blocks	Seedlings Planted	Usable Seedlings on 12 September Freezing Test	Usable Seedling on 26 September Freezing Test
OS	Outdoor	24 May	7	60	4	1680	748	798
F1	Frozen	24 May	7	60	4	1680	688	741
F2	Frozen	7 June	7	60	4	1680	688	747
F3	Frozen	21 June	7	60	4	1680	680	742
F4	Frozen	5 July	7	60	4	1680	695	766
					Total	8400	3499	3794

). The main plots representing 7 June, 21 June, and 5 July were divided into 7 plots each (split plot), as only freezer-stored seedlings were planted on those dates (treatments F2, F3, and F4). Thus, 60 seedlings of each species/seed source and treatment were planted in each main plot at all planting dates for a total of 8400 seedlings. Two additional field trials with similar experimental setups were established in North and South Iceland in 2022 to provide results for the growth and survival of experimental seedlings after three growing seasons.

In late May 2022, night frosts occurred in the nursery, damaging some of the terminal buds of seedlings in outdoor storage. This became visible in the spruce species later on. Although the aim was to plant only healthy seedlings, damaged buds were difficult to recognise, and some damaged seedlings were planted. Additionally, during the summer, some of the spruce seedlings’ new growth began to wither a few weeks after planting, finally resulting in seedling mortality. The survival of the spruce species was therefore lower than that of the others and resulted in fewer frost hardiness measurements for spruce species. The cause of this mortality is unknown. Grey mould (Botrytis spp.) was detected in the Skagway seed source of lodgepole pine stored in a freezer and there was some seedling mortality during summer. Only undamaged seedlings were used in the freezing tests, and the number of measurements therefore differed between species.

2.3. Assessment of Autumn Frost Hardiness

The annual growth of seedlings was used to test the level of frost hardiness on two occasions in September. From each split plot, half of the seedlings were randomly harvested on 12 September, and the other half on 26 September, then randomly assigned a freezing treatment (4, −8, −12, and −16 °C) on each date. The stem of the seedlings was cut 2–3 cm under the start of the annual growth to ensure that all of the annual growth was included in the freezing test material. Since it was difficult to determine where annual growth began in downy birch, the uppermost 8 cm was used. The shoots were packed in plastic bags, sprayed with water to prevent drying during the freezing process, and stored in an insulated cooler. The shoots were then stored at 4 °C until freezing procedures began later on the same day. Freezing was conducted in four programmable freezers. One was used to store the control seedlings at 4 °C during the freezing process. The freezing process began at 4 °C, and the temperature was slowly lowered towards the target temperatures by 2 °C per hour. When the target temperatures were reached, the temperature was held for 2 h before thawing at the same rate. After freezing and thawing, shoots were placed in trays filled with wet peat and stored under stable growing conditions at 20 °C with constant light. To maintain a high air humidity, a plastic cloth covered the shoots, and nozzles sprayed water over the seedlings at 3 h intervals. The shoots were sprayed with fungicide (Rovral, Aquaflow) to prevent an outbreak of grey mould.

The extent of injury to each shoot was assessed after 20 days. The total length of annual growth was measured. Then, each cutting was dissected, and the length of the damaged tissue was measured. The shoot was scored as damaged if the cambium and phloem were brown and alive if they were green with no visible loss of tissue integrity. Damage was expressed as relative browning and calculated as the ratio of the total length of brown tissue to the total length of annual growth. Damage to needles on lodgepole pine and spruce species was estimated by measuring the length of discoloured and necrotic needles. To calculate the damage, the ratio of the total length of annual growth was used. This method is described thoroughly by Burr et al. [37].

2.4. Weather Data

The air temperature was registered in the field every 15 min at heights of 15 cm and 2 m using dataloggers (Dataloggers TMS, Tomst^®, Prague, Czech Republic). Heat sum calculations on long term weather data were performed according to the method introduced in Langvall and Ottosson Lofvenius [38]. The threshold value was set to +5 °C, indicating that only that part of the daily average temperature that exceeded +5 °C was added to the sum, while days with an average temperature below +5 °C did not add to the sum. The daily average air temperature at a height of 2 m was used to calculate the cumulative heat sum. Calculating heat sums at various locations in Iceland started on 1 January every year (

Table 3. Mean temperature sum (min/max) and probability of the date of first autumn frost equal to or more than −5 °C from various locations in Iceland. Observations were made on standard meteorological screens at a height of two meters during the years 1991–2022. Data from the Icelandic Meteorological Office.

			Temperature Sum 1			Probability of First Autumn Frost ≥ −5 °C
Place	Coordinates	Altitude (m asl)	Min	Mean	Max	Average Date, 50% Probability	Earliest Date, 5% Probability
Þingvellir	64°16′ N, 21°05′ E	110	587.8	735.2	931.7	3 October	8 September
Akureyri	65°41′ N, 18°06′ E	31	571.1	771.6	955.4	22 October	29 September
Reykjavík	64°12′ N, 21°90′ E	52	553.7	800.8	1044.5	14 November	15 October
Hallormsst.	65°05′ N, 14°44′ E	60	554.7	737.5	898.1	20 October	2 October
Mývatn	65°37′ N, 16°58′ E	282	382.6	559.8	758.2	10 October	10 September
Höfn	64°16′ N, 15°12′ E	5	580.8	728.3	886.3	18 November	3 October
Stafholtst.	64°38′ N, 21°35′ E	14	524	699.1	1003.2	6 October	8 September
Hella	63°49′ N, 20°21′ E	20	627	788	949	16 October	13 September

¹ Temperature sum of the thermal growing season.

).

When the probability of the first damaging frost events was calculated, the threshold for damaging first frosts was set at –5 °C. A confidence level of 95% was calculated for the date of the first frost event to show variation in the probability of the first frost event equal to or more than −5 °C. Since data for temperature sums were collected at a height of 2 m, it was possible that the probability of the first frost was underestimated. The frost was harder and occurred earlier at the height of the planted seedlings. Nevertheless, the information provided an indication of variability between locations. In addition, Table 3 shows that it is realistic to test frost resistance in the middle and second half of September, when the first frosts can often be expected. Weather data in

Table 4. Mean monthly temperatures and total precipitation in summer 2022 in Akureyri and the proportion of average (1991–2020) precipitation [39].

Month	Mean Temperature (°C)	Total Precipitation (mm)	Proportion of Average Precipitation (%)
May	6.3	50.6	200
June	9.7	19.9	96
July	11.2	47.5	140
August	10	28.6	69
September	8.7	13.8	26

is obtained from the nearest weather station.

2.5. Statistical Analysis

2.5.1. Probability of Damage

Freezing damage to the cambium resulted in an inflated zero and one distribution. Therefore, for analysis purposes, proportional damage was transformed into a binomial variable. When the damage was more than 20% of the annual growth, it was regarded as severely damaged and graded as 1 in the data. Otherwise, the damage was graded as 0. To determine where there were effects of species, winter storage treatment combined with planting date and freezing temperature used in freezing tests, the method of generalised linear mixed models (GLMM) was applied assuming a binomial distribution [40]. It is recommended that GLMM be kept simple [41]. To simplify the model, only two-way interactions were included. To reduce the number of parameters estimated, temperatures in freezing tests at −12 and −16 °C were combined into one category. The full model was as follows:

$$log\left(\frac{p_{ijk}}{1-p_{ijk}}\right)={\alpha }_i+{\beta }_j+{\gamma }_k+{\left(\alpha \beta \right)}_{ij}+{\left(\alpha \gamma \right)}_{ik}+{\left(\beta \gamma \right)}_{jk}+{\zeta }_b+m_{lb}+s_{ijb}$$

(1)

where p_ijk was the probability of damage (>20% damage) for species i receiving winter storage treatment j and freezing test temperature k. The fixed effects were as follows: α is the effect of species, β is the effect of winter storage treatment, and γ is the effect of frost temperature. The model also includes a two-way interaction between these main terms, and ζ is the effect of block. The m_lb represents main plot l in block b, and s_ijb represents the species plot for species i in winter storage treatment j in block b. These terms were included as random effects.

The function glmer in package lme4 [42] in statistical software R version 4.2.2. [43] was used to estimate the parameters in the models. Backward selection was used to select the final model, and based on a likelihood ratio test, non-significant main terms (α = 0.05) were removed from the full model. Random and block effects were not removed from the model. Asymptotic confidence intervals were estimated with the emmeans package [44]. This was performed separately for the two freezing times (12 and 26 September). At the time of the earlier freezing test, the Närlinge seed source of lodgepole pine hardly suffered any damage, and mostly had measurements graded 0. This resulted in numerical estimation problems [40], resulting in no parameter estimation. Therefore, the seed source was excluded from the analysis of the earlier freezing test. This was also the case for both seed sources of lodgepole pine at the later freezing test date, and they were then excluded from further analysis.

2.5.2. Annual Growth in Field Trials

To test whether there were effects of species and winter storage treatment combined with planting date on annual growth, a mixed model method was applied. The full model was as follows:

$$y_{ijb}={\alpha }_i+{\beta }_j+{\left(\alpha \beta \right)}_{ij}+{\zeta }_b+m_{lb}+s_{ijb}+{ϵ}_{ijb}$$

(2)

where y_ijb is the annual growth of species i receiving planting treatment j in block b. The terms in the model were the same as in Equation (1). Prior to the analysis of annual growth, normality was checked by inspecting the histogram of residuals, and no violation of normality was observed. Backward selection based on the likelihood ratio test was used to select the final model. Asymptotic confidence intervals were estimated with the emmeans package [44]. Downy birch was not included in the analyses, as it was not possible to detect where annual growth started on the stem of the seedlings.

2.5.3. Mortality and Initial Seedling Height

The average proportional mortality of species within each treatment (OS, F1, F2, F3, and F4) and block was calculated and then analysed by the non-parametric Kruskall–Wallis test.

Anova was used to test if there was a difference in the average initial height of seedlings of different species. The initial height was transformed with a logarithmic transformation to fulfil the assumption of normality.

3. Results

3.1. Initial Height and Annual Increment Measured in the Nursery

The measurements of initial height were found to be the same for seedlings stored in freezers during winter and seedlings stored outdoors. Therefore, no distinction was made between the initial height of seedlings with different winter storage methods, as shown in Figure 1. There was a significant difference in the average initial height of lodgepole pine seed sources. Skagway was about 7 cm, and Närlinge was 11.8 cm. Russian larch was about 1 cm higher than hybrid larch, on average, and there was no difference in initial height between the spruce species. As early as 13 June, lodgepole pine from seed source Närlinge had grown 83% of the total annual increment, and Skagway had grown 84% of the total annual increment. In later measurements, the increment was much lower, indicating reduced growth. The growth rhythm was different in spruce species, as their growth was more evenly distributed over the period until 19 July. Then, measurements showed that the rate of growth decreased. In contrast, measurements for larch species and downy birch on 19 July did not indicate slower growth.

3.2. Seedling Mortality

All dead seedlings were counted on 25 August in the field trial. Survival was good in birch, the larch species, and the Närlinge seed source. Mortality in the spruce species and the Skagway seed source was noticeable (

Table 5. The average proportion of dead seedlings in Sitka spruce, Lutz spruce, lodgepole pine seed sources Skagway and Närlinge, hybrid larch, Russian larch, and downy birch with different planting dates and winter storage methods. Asterisks after treatment names indicate significant differences between species planted at different dates and stored over the wintertime outdoors or in a freezer. OS = outdoor-stored seedlings planted on 24 May, F1 = freezer-stored seedlings planted 24 May, F2 = freezer-stored seedlings planted 7 June, F3 = freezer-stored seedlings planted 21 June, and F4 = freezer-stored seedlings planted 5 July.

	Treatments
Species	Mortality (%)
	OS **	F1 **	F2 ***	F3 ***	F4 **
Sitka spruce	10.8	37.5	43.3	44.6	39.2
Lutz spruce	12.9	31.0	28.8	25.0	23.3
Lodgepole pine, Skagway	5.0	13.7	8.3	19.2	10.8
Lodgepole pine, Närlinge	3.9	1.3	0.4	2.5	2.1
Hybrid larch	0.8	6.3	0.0	0.4	1.3
Russian larch	0.4	0.8	1.3	0.0	0.4
Downy birch	0.0	1.3	0.0	0.0	0.0

Note: ** p < 0.01, *** p < 0.001.

). Sitka spruce had 41.2% mortality, on average, in freezer-stored treatments. The OS treatment had 10.8% mortality. In Lutz spruce, the average mortality in freezer-stored treatments was 27%, while the OS treatment had 12.9% mortality. Skagway had 13% mortality, on average, in F1–F4 treatments, and the OS treatment had 5% mortality. There was a significant difference in mortality between species in all treatments (p < 0.01).

3.3. Weather Data

Frost events occurred during the growing season. The first one took place on 28 May, with a minimum temperature of −0.9 °C measured at a 15 cm height (Figure 2). The next event, −0.7 °C, occurred on 27 June. Five frost events occurred in August. On 5 August, the minimum temperature was −0.9 °C. On 16 and 17 August, the minimum temperatures were −1.9 and −1.6 °C, respectively. On 22 August, the minimum temperature was −1.3 °C, followed by −2.1 °C on 27 August. The next frost events occurred on 14 and 15 September, when minimum temperatures were −2.1 and −2.6 °C, respectively. On 18 September, the minimum temperature at a 15 cm height was −1.5 °C, and the night before the samples were harvested for the later freezing test, the minimum temperature was −3.8 °C. At the same time, the minimum temperature measured at a 2 m height was 0.21 °C. Measurements from a 2 m height showed warmer temperatures in most cases than measurements from a 15 cm height.

3.4. Cumulative Heat Sum

The cumulative heat sum for the seedlings stored outdoors during winter 2022 was 680.4 day degrees (d.d) on 12 September and 730.7 d.d on 26 September (

Table 6. Cumulative heat sum according to planting dates for treatments subjected to different winter storage methods and planted at different times during the summer of 2022, and the proportion of cumulative heat sum to total heat sum for each treatment. Weather data were collected from the nearest weather station in Akureyri at a 2 m height (65°41′ N, 18°06′ W).

Time Periods in 2022	Explanation	Heat Sum	Treatments and Planting Dates
			OS	F1	F2	F3	F4
			24 May	24 May	7 June	21 June	5 July
1 January–24 May	C.h.s until 24/5	106	106	0	0	0	0
25 May–7 June	C.h.s between planting dates 1 and 2	62.5	168.5	62.5	0	0	0
8 June–21 June	C.h.s between planting dates 2 and 3	81.9	250.4	144.4	81.9	0	0
22 June–5 July	C.h.s between planting dates 3 and 4	43.7	294.1	188.1	125.6	43.7	0
6 July–12 September	C.h.s between planting date 4 to 12 September	386.3	680.4	574.4	511.9	430	386.3
13 September–26 September	C.h.s between freezing tests	50.3	730.7	624.7	562.2	480.3	436.6
27 September–31 December	C.h.s throughout the year after 26 September	38.1	768.8	662.8	600.3	518.4	474.7
	Total cumulative heat sum for 2022	768.8		662.8	600.3	518.4	474.7
	Proportion of total heat sum for 2022		100%	86%	78%	67%	62%
	Proportion of total heat sum 12 September		100%	84%	75%	63%	57%
	Proportion of total heat sum 26 September		100%	86%	77%	66%	60%

C.h.s = Cumulative heat sum. OS = outdoor-stored seedlings planted on 24 May, F1 = freezer-stored seedlings planted 24 May, F2 = freezer-stored seedlings planted 7 June, F3 = freezer-stored seedlings planted 21 June, and F4 = freezer-stored seedlings planted 5 July.

). The decrease in the heat sum was clear as the planting date was delayed. Thus, the F1 treatment received 84% of the total cumulative heat sum on 12 September, and F2, F3, and F4 received 75, 63, and 57%, respectively. The numbers for the freezer-stored treatments on 26 September were 86, 77, 66, and 60%, respectively. The heat sum accumulated 12 September for seedlings that were stored in the freezer and planted on 24 May (F1) was reduced by 16% compared to seedlings that were stored outdoors and planted at the same time (OS).

3.5. Probability of Frost Damage

The freezing tests were conducted at the time of the year when frost hardiness is known to be increasing in young seedlings [22]. It was, therefore, expected that there would be a difference between the outcome of the two freezing occasions, and so the data from the two freezing occasions were analysed separately. However, the statistical analysis applied to the data for the freezing test on 12 September did not result in the same final model as the freezing test on 26 September. The main factor, frost exposure, was significant in the model (0.0013) but only the interaction between treatments and species was significant in the model for the earlier freezing test (0.0005). In the later test, there was a significant effect of the interaction between species and the temperatures (0.0050) used in the freezing test and the interaction between treatments and temperatures (<0.0001) used in the freezing test. In the model for the earlier freezing test, Skagway was included, but in the later test, both seed sources of lodgepole pine were excluded. This may have caused the results from the earlier test to be less conclusive than those for the later one.

3.5.1. Probability of Frost Damage in Downy Birch

The results from 12 September for downy birch (Figure 3) showed a probability of damage in control seedlings (4 °C), indicating mild natural freezing occurrence before the freezing test. This was consistent with weather data showing frost events prior to 12 September (Figure 2). The date of planting had a significant effect on the probability of damage. For downy birch seedlings planted on 24 May, OS and F1 had a 2 and 5% probability of damage, but seedlings planted on 5 July (F4) had a 17% probability of damage. This was also the case when birch seedlings were frozen to −8 °C. Seedlings planted at the earliest dates had a lower probability of damage than seedlings planted later. The OS treatment had a 6% probability of damage, while the seedlings from the later planting dates F3 and F4 had a significantly higher probability of damage (27 and 36%, respectively). All seedlings frozen to −12 and −16 °C had severe damage ranging from 69 to 95% probability of damage.

In general, the probability of damage for downy birch in the later test was less than in the earlier test. In control seedlings and in seedlings frozen to −8 °C, the probability for the five treatments ranged from 0 to 8%. When seedlings were frozen to −12 and −16 °C in the later freezing test, the outdoor-stored seedlings had the significantly lowest probability of damage (6.8%). Probabilities of damage in other treatments were 14, 24, 45, and 69%, respectively.

3.5.2. Probability of Frost Damage in Larch Species

The natural frost that occurred before 12 September increased the probability of damage in hybrid larch more than in Russian larch, and the probability in both species increased as the planting date was delayed (Figure 4). This was observed in the control in the former freezing test for hybrid larch, where the OS treatment had a 15% probability of damage, while the other treatments had 20, 28, 39, and 68% probabilities of damage. The same numbers for Russian larch ranged from 4% for the OS treatment to 10, 15, 11, and 56%, respectively, for treatments F1–F4. A similar trend was observed when seedlings were frozen at −8 °C, showing that hybrid larch had a higher probability of damage than Russian larch. Severe damage was observed in both larch species when frozen to −12 and −16 °C on 12 September. In the second freezing test, the probability of damage in both species was much lower, indicating increased frost hardiness between the tests.

3.5.3. Probability of Damage in Lodgepole Pine Seed Sources

Natural frost had little effect on the seedlings of the lodgepole pine from Skagway in the earlier freezing test (Figure 5A). In control seedlings, treatments planted on 24 May had a significantly lower probability of damage than those planted on 7 June and later with 6, 7, and 7% probability of damage, respectively. Freezing to −8 °C produced similar results, except that the probability of damage in F3, F4, and F5 had then increased to 16, 18, and 17%, respectively. When freezing to −12 and −16 °C, the probability of damage increased, but the injury was least in treatments planted on 24 May.

Due to the low probability of damage in the later freezing test, it was not possible to perform a significance test with the same model on the data for Skagway; therefore, only averages are shown for those measurements. Natural frost events did not have any effect on the probability of damage in control seedlings, and it was only in seedlings planted on 5 July that the average rose to 4% probability of damage when freezing to −8 °C. When freezing to −12 and −16 °C, the probability of damage was progressive from 2% for the OS treatment, increasing with later planting dates to 5, 11, 27, and 52%, respectively.

The probability of damage to the lodgepole pine from Närlinge (Figure 5B) was only noticeable in the earlier freezing test. However, the small amount of damage (too many zeros) resulted in no convergence of the models and that seed source was therefore removed from the statistical analyses. It was only when freezing to −12 and −16 °C that any damage occurred. The OS treatment did not have any probability of damage, but the probability of damage in the other treatments increased as the planting date was delayed. In the later freezing tests, the probability of damage only occurred in F4 when freezing to −12 and −16 °C. When comparing the two seed sources of lodgepole pine, Närlinge had the most frost hardiness in both freezing tests.

3.5.4. Probability of Damage to Spruce Species

In Sitka spruce, natural frost events increased the probability of damage in the earlier freezing test. In the control seedlings, the OS treatment had 2% damage. The other treatments had a significantly higher probability of damage (14, 25, 14, and 23%) (Figure 6A). When freezing to −8 °C, the results were similar, except that the probability of damage in OS was 6%, while the other treatments had 32, 47, 30, and 45% probability of damage. Again, it mattered if the seedlings had been stored in a freezer or outdoors. When seedlings were frozen to −12 and −16 °C, OS also had high damage (68% probability), while the other four treatments had an average probability of damage of 95.5%. In the later freezing test, 26 September, all treatments for all temperature exposures had less probability of damage, indicating more frost hardiness.

The results from the earlier freezing test for Lutz spruce did not show as clear results as those seen in Sitka spruce (Figure 6B). The F3 treatment had a lower probability of damage than treatments planted earlier, except for the OS treatment. When comparing Sitka spruce to Lutz spruce in the later freezing test, the probability of damage for Sitka spruce was 3–34% for the five treatments in the control seedlings but only 1–20% for Lutz spruce. The same numbers for seedlings frozen to −8 °C were 0.5–50% probability of damage in Sitka spruce and 0.5–10% for Lutz spruce, indicating more frost hardiness in Lutz spruce than Sitka spruce.

3.5.5. Annual Growth in Field Trial

The species responded differently in annual growth regarding winter storage and planting dates (Figure 7). The interaction between treatments and species was significant (<0.0001). This indicates that treatments affected the species in different ways. Therefore, each species and seed source were analysed separately. In general, the lodgepole pine seed sources had similar annual growth for all treatments. The Närlinge seed source grew 14.3–16.3 cm, on average, with the OS treatment growing significantly less than the freezer-stored treatments. The Skagway seed source grew 12.9–14.1 cm in height, on average. Larch species showed large differences in annual growth between treatments. Hybrid larch from the OS treatment grew 50% more than the F4 treatment. This difference was even larger in Russian larch, where the OS treatment grew 16.6 cm on average, but the F4 treatment grew only 6.6 cm. Growth was also reduced in spruce species with later planting dates.

4. Discussion

The general pattern of freezing damage in all species was of increased damage with later planting dates. These results are in accordance with results from Luoranen et al. for Norway spruce [45] and Scots pine [13,25]. The main reason for these results is the reduction in the length of the growing season. The accumulation of heat sum was greatly reduced with later planting, resulting in only 60% of the total available heat sum for the F4 treatment, which had the highest probability of damage.

Night length is a dominant factor in inducing growth cessation [7], which is a prerequisite for the accumulation of cold hardiness [8,9,10]. Frost hardiness increased between the two freezing occasions in September, when the night length is increasing rapidly. This was observed in the lower probability of damage in seedlings tested in the later freezing test. This is in accordance with the review of Bigras et al. [22], who described frost hardiness acclimation as a process that is observed concomitantly with the cessation of growth in late summer and fall, induced by decreasing day length and mild freezing occurrence causing more rapid hardening. After 12 September, several mild frosts are likely to have induced freezing tolerance in this study.

Since night length alone cannot induce growth cessation before species-specific heat sum is accumulated [6,10,24], and since growth cessation determines the onset of hardening, it is worthwhile to investigate the growth rhythm of the species included in this project. To facilitate discussion of the results, each genus is discussed in separate sections.

4.1. Lodgepole Pine

Both seed sources of lodgepole pine showed the overall lowest probability of damage and had, therefore, developed the greatest frost hardiness of the tested species in September. Närlinge hardened earlier than Skagway. Adaptation connected to the origin of species and provenances affects how quickly their frost hardiness is developed [46,47], and this was evident for the lodgepole pine in this project, as Skagway is originally from 59°27′ N but the origin of Närlinge is even more northerly at 60°44′ N–63°40′ N.

The growth of lodgepole pine has been classified as predetermined growth. Shoot elongation is completed by mid-summer, and terminal bud formation begins then [48,49]. Measurements of annual growth in the nursery showed that both lodgepole pine seed sources had finished the majority of their shoot elongation on 13 June. This was not the case for the other species. Additionally, terminal buds were easily detected when measurements were taken in September, and their shoot elongation was similar despite being planted at different times, indicating that growth cessation had been reached during the growing season. The accumulation of heat sum was enough for growth cessation and, thereby, early onset of frost hardiness accumulation. Only in the lowest freezing temperatures (−12 and −16 °C) in the earlier freezing test was considerable damage evident in Närlinge. Since such hard frosts are very unlikely in Iceland in September, the probability of damage of this kind is negligible for this seed source.

4.2. Spruce

The high mortality rate of spruce seedlings remains unexplained. A possible reason could be fungal infection. The spruce was cultivated in a different place in the nursery than the other species and might have been more exposed to contamination by a fungal disease of some kind. Most diseases known to cause problems in the production of forest seedlings are fungal and gain access to the host plant during periods of physiological stress or through localized sites of damaged or diseased tissue [36]. Another possible explanation of the mortality is physiological stress from storage conditions. Seedlings in frozen storage are still physiologically active despite the low temperatures. This is reflected in decreased root growth potential, delays in days to bud break, and reduced freezing tolerance and carbohydrates as the storage lengthens, which can negatively affect planting success [14,15]

Annual growth was reduced in the spruce species with later planting. This resulted in later growth cessation and consequently a higher probability of damage in freezing tests. Although there was no significant difference in the annual growth among the treatments planted on 24 May in Sitka spruce, there was a difference in the frost hardiness of these treatments in the earlier freezing test. Sitka spruce seedlings stored in a freezer had a significantly higher probability of damage in freezing tests than those stored outdoors but planted at the same time. By storing seedlings in a freezer, the accumulation of the thermal growing season was reduced by 16% of the total heat sum if planted on 24 May. Sitka spruce is the only species in this study to show such sensitivity to a reduced heat sum. This was not the case for Lutz spruce. It accumulated more frost hardiness than Sitka spruce, as there were no significant differences in the probabilities of frost damage between treatments planted on 24 May. These results are in accordance with Skulason et al. [50]. They studied frost hardiness in three-year-old Sitka spruce and Lutz spruce from various provenances in Iceland. Sitka spruce gained frost hardiness later in autumn than the Lutz spruce, and a large variation in frost hardiness was observed in Lutz spruce between provenances and between individuals within provenances. A large variance in the frost hardiness of Lutz spruce provenances and within provenances was also found by Dietrichson [51] and Junttila and Skaret [52]. This is also evident in this study, as the probability of damage in Lutz spruce was not linear in the results from 12 September. When Sitka spruce and Lutz spruce were compared in the freezing test on 26 September, Sitka spruce had not developed as much frost hardiness as Lutz spruce. Therefore, more emphasis should be placed on Lutz spruce in Icelandic afforestation to ensure less damage is caused by autumn frosts, especially if the spruce is to be stored frozen during the winter. Duong [34] recommended a Lutz spruce provenance from Iniskin Bay, Alaska. In that study, results were obtained from a spruce provenance study that started in 1995 and 1996 at six locations in Iceland. The Sitka spruce provenance ‘Stálpastaðir’ (also used in this study) had a 65% survival rate, on average, and a mean height of 381 cm, while ‘Iniskin Bay’ had a 94% survival rate, on average, and a mean height of 392 cm. The difference between these species is not great, although Sitka spruce is generally considered a faster-growing species than Lutz spruce. The Lutz spruce provenance ‘Iniskin Bay’ performed better in frost-prone (flat landscape) and non-frost-prone (sloped landscape) sites than Sitka spruce. This underlines the importance of planning afforestation from the start by using knowledge of how to read the landscape to select a non-frost-prone site for spruce species in such a way that there is as small a risk of frost damage as possible or that damage can be avoided by assigning suitably hardy genotypes to frost prone sites.

As the minimum heat sum for the species used in this study is not available and the variation in heat sum from year to year is considerable, it is difficult to determine when freezer-stored spruce should be planted during the spring. The total accumulation of heat sum was 768.8 d.d during the summer of 2022 in Akureyri, while the average heat sum for Akureyri, based on 30 years of weather data, was 771 d.d. The minimum heat sum in the 30-year period was 571.1 d.d, and the maximum heat sum was 955.4. There are also differences in heat sum between sites, and with altitude above sea level. The results of this study, which was carried out in a summer with an average heat sum, showed that Sitka spruce is particularly sensitive to a reduced heat sum. In an average year, autumn frost damage can be expected in freezer-stored Sitka spruce planted after 24 May. However, the extent of damage is influenced by the time of planting and when the first significant frost occurs. Planting as early as possible in spring is strongly recommended for spruce.

Tree species require different amounts of accumulated heat sum for breaking buds in spring, depending on the adaptation of the species to its long-term historical origins [53,54]. As freezer-stored seedlings are dormant at planting, the species’ inherent heat sum requirement to break buds can influence the growth rhythm during the growing season, including growth cessation and thereby the beginning of hardening. The timing of budbreak for spruce species in the spring is primarily controlled by the accumulated time of exposure to warm air temperatures [55]. According to Skulason et al. [50], Lutz spruce is more sensitive to spring frost than Sitka spruce. This indicates a higher heat sum requirement for Sitka spruce than for Lutz spruce to start growing. This is also a factor affecting frost hardiness during autumn, as Sitka spruce requires a higher heat sum accumulation to reach growth cessation.

4.3. Larch

The larch species in this study showed the highest probability of damage caused by the freezing tests, and the hybrid larch was more sensitive than the Russian larch. According to Simak [56], autumn frost hardiness was strongly connected with the growth rhythm–photoperiodic response of the European larch, with an interaction between photoperiod and thermoperiod determining the growth rhythm. Thus, the days in northern Scandinavia were longer and the temperatures during autumn were lower than the climate of the place of origin for European larch. Therefore, the accumulation of frost hardiness during autumn was insufficient for the larch. The measurements in the nursery on 19 July revealed that larch species growth was not reduced as in spruce and pine and that growth cessation occurred later in larch species. In a provenance trial with European larch and Russian/Siberian larch, Eysteinsson and Skúlason [57] found that European larch provenances required a higher temperature sum to commence growth in spring than Russian/Siberian larch provenances. Thus, European larch has a later onset of growth during spring and continues to grow longer during autumn and is therefore more sensitive to autumn frost damage than Russian larch [58]. Our findings are in accordance with the results of Skulason and Jonsdottir [59] who studied autumn frost hardiness in 10–15-year-old trees of Russian larch ‘Lassinmaa’ and the hybrid larch ‘Hrymur’. They found that on 21 September, 7.5% of Russian larch scions frozen to −12 and −16 °C had freezing damage, while 85% of hybrid larch had damage. Further measurements on 27 September and 4 and 11 October showed more frost hardiness in Russian larch. Since hybrid larch is less resistant to autumn frosts than Russian larch, regardless of whether it is stored in the freezer or outdoors, the emphasis on its planting should be in relatively higher heat sum areas and directed to less frost-prone sites. Crookedness due to autumn frost damage in Russian larch is a known problem in Iceland [60]. To avoid freezing damage during autumn, freezer-stored Russian larch should be planted early in spring.

4.4. Downy Birch

Birch is the only native species in this study, and it is therefore well adapted to the short growing season and low summer temperatures in Iceland [61]. Nevertheless, natural frosts increased the probability of damage to birch. Since the frost damage appeared in freezer-stored seedlings planted on 21 June and 5 July, a frost event occurring after these planting dates must have caused natural damage. It is most likely that the frosts that occurred in August caused the damage since almost the same probability of damage appeared in the control temperature as when freezing to −8 °C in the later test. It was not until freezing to −12 and −16 that a significant difference in the probability of damage appeared in birch. The probability increased with later planting. The treatment that was stored outdoors during the winter had only a 5.8% probability of damage, even though the freezing temperatures were −12 and −16 °C. The measurements conducted on 19 July in the nursery showed that downy birch had not reduced growth. This did not affect frost hardiness as negatively as seen in larch species. With this, the birch showed good adaptability as it continued to grow further into the autumn, even though the photoperiod had started to shorten, and with less risk of increasing frost damage compared to larch and spruce. This indicates that downy birch is adapted to make good use of the growth period when opportunities arise [7].

4.5. Weather Data and Heat Sum Accumulation

According to Iceland’s Meteorological Agency, the summer of 2022 was an average summer in terms of temperature. However, since there can be great fluctuations in the weather between years, it is not possible to draw a conclusion about when freezer-stored seedlings should be planted. Risk analysis based on long-term climatic data gives more general information about potential frost risks for different planting dates for various sites. Based on 50-year climate data from Central Finland, Hänninen et al. [10] showed that Norway spruce seedlings planted around 10 June will be damaged by autumn frosts every tenth year, that those planted on 20 June will suffer from frost every fifth year, and that those planted on 30 June will suffer autumn frost damage as much as every second year. In their study, they used daily mean temperatures measured at a height of 2 m and the daily minimum temperature measured at a height of 5 cm from the soil surface. This is necessary to be able to predict the risk of damaging frost events with accuracy since heat measurements from 2 m height can underestimate the actual temperatures at the young seedling’s height. When the sky is clear, the rate of thermal radiation loss from the ground is 5–10 times greater than on a cloudy night, which can lead to a 4–5 °C decrease in minimum air temperature at seedling height. The temperature near the soil surface decreases rapidly and frost occurs even though the air temperature at 1.3 m is well above the freezing point [55,62]. This was visible in this study in the temperature data from the field site, in which temperature measurements at a height of 15 cm showed lower nighttime temperatures, in most cases, than measurements at a height of 2 m. For example, when the minimum temperature at a 2 m height was registered as 0.21 °C on 26 September, the minimum temperature at a 15 cm height was −3.8 °C. To predict the probability of frost as well as Hänninen et al. [10], there would have to be available data from weather stations at a height of 5 cm. If only data from a height of 2 m is used, there is a risk that the forecast will be inaccurate, i.e., damaging frosts would not be predicted until too late in the year. Additionally, to predict damage based on the accumulation of heat sum, the minimum heat sum requirement for the species used in this study must be known. Further research on this topic should focus more on that. Due to considerable variation in temperature sum from year to year (Table 3), it is difficult to state a date that should be the latest date for planting seedlings that have been stored in a freezer. However, the data and related publications show that the heat sum varies depending on height above sea level, land areas, and landscape [30], and can therefore be used to decide which area should be avoided when planting freezer-stored seedlings.

5. Conclusions

The total available heat sum for growth and maturation of freezer-stored seedlings was reduced compared to seedlings stored outdoors and the reduction increased with later planting. This caused an increased probability of autumn frost damage with a later planting date. However, the species in the project were differently sensitive to the decreasing heat sum and the study revealed they had different growth rhythms. Lodgepole pine and birch were the least sensitive, showing the lowest probabilities of frost damage in September. The lodgepole pine finished shoot elongation early, resulting in the early onset of frost hardening, and native birch is well adapted to utilise the short growing season. Spruce species had less frost hardiness in autumn than pine and birch, and Sitka spruce did not have as much frost hardiness during autumn as Lutz spruce. The larch species accumulated the least frost hardiness in September. The growth rhythm of spruce and larch affected the timing of growth cessation and, subsequently, their accumulated frost hardiness. Therefore, those species, if frozen and stored during winter, should be prioritised in planting in early spring. Lodgepole pine and birch better withstood the effect of the reduced growing season and were therefore better suited to being stored in a freezer over the winter than larch and spruce.