Molecular Selection Tools in Adaptive Phenology of Populus trichocarpa Breeds for the Nordic-Baltic Region

Abstract

Fast-growing poplars have the potential to improve the biomass supply required for the transition to bio-based economies in the Nordic-Baltic region. As early successional trees, poplars are efficient biomass producers in relatively short rotations, when high-yielding, climate-adapted clones are available for commercial deployment. In Sweden, poplar breeding focused on adapting Populus trichocarpa to the Swedish climate by crossing parents from distant populations along latitudinal and maritime-continental clines on the Pacific coast of North America. Clonal trials with progeny from these crosses were established in the Nordic-Baltic region. Elite individuals in terms of stemwood production were used to identify candidate genes for adaptation to local photoperiod and climate in the region. The next breeding cycle utilized the elite individuals in the clonal trials to generate a training population. Genomic selection of the progeny in the training population will facilitate early selection of poplar clones for commercial deployment in the Nordic-Baltic region and reduce the time required for successive plant breeding cycles.

1. Introduction

When breeding poplars for northern latitudes, significant attention is given to the adaptation of their phenological traits to regional climates [1, 2, 3, 4]. The most important phenological events that distinguish the period of dormancy and active growth are bud flushing, growth cessation, and bud set [5]. The timing of these events is crucial for an individual genotype to adjust its growth efficiently to the temperature and photoperiod in a given region [6]. Thus, different populations within the species’ geographic range are set under intense selection pressure to adapt to latitudinal and altitudinal gradients of daylight and temperature. Individuals with phenologies that do not match their environments are at a higher risk of being frost-injured, especially during the transition periods from autumn to winter and winter to spring. However, prolonged growth can also lead to a competitive advantage by allowing for greater utilization of the growing season [7, 8]. As a result, the phenology of each population is shaped by its adaptation to the local and regional environments but still contains significant genetic variation for phenological traits [9].

Adaptation of a species to new geographic regions relies on the variation within and between populations. Populus trichocarpa (Torr & A. Gray ex Hook) offers ample opportunities for adaptation to the Nordic-Baltic region due to its wide ecological range and geographic distribution [10]. While phenological traits in natural populations adjust to climate extremes, the artificial transition is typically directed northward to extend the active growth period and increase biomass production [11]. In species like Norway spruce, population-level variation mitigates the risks of this transition. In operative poplar cultivation, however, individual clones lack the buffer contained in the variation within a deployed population, which increases risks for financial losses.

The Nordic-Baltic forest industry and energy sector have used native aspen (P. tremula L.) for decades, and it is reasonable to assume that poplars can also be utilized in current and innovative manufacturing processes. Cultivated poplars are a marginal resource in the region, and widespread cultivation would require an extensive breeding and testing program. However, molecular selection tools offer a faster and more cost-efficient way to select clones, which is particularly useful for poplars, where clone selection and testing costs are disproportionally high, considering their marginal role as a raw-material asset in the region.

For over 30 years, the Swedish University of Agricultural Sciences has been running a climate adaptation program for P. trichocarpa. Since 2007, various stakeholders have also been included in research and testing activities. The most promising clones have been planted in pilot plantations throughout Sweden, Lithuania, Latvia, and Estonia, and several clones are expected to be registered with the Swedish Board of Forestry by 2024.

2. Potential of hybrid poplars in the Nordic-Baltic wood market

Forests offer multiple ecosystem services and are a crucial resource in the transition to biobased economies. This is especially true for the Nordic-Baltic region, which has abundant forest resources and well-established wood industries. Consequently, the region is a major hub for education, research and development, testing, and production of innovative wood-based products, including textiles, biofuels, biochemicals, and engineered wood products for the construction sector [12]. However, the increasing demand for wood as a raw material is likely to be in conflict with the environmental role of forest ecosystems as carbon sinks and guardians of biodiversity [13]. Additionally, in a warming climate, forests may be more susceptible to severe droughts, fires, storms, diseases, and insect outbreaks, which could further reduce the availability of woody biomass [14].

Expanding the area of short-rotation poplar plantations is a viable option for increasing the stocks of industrial wood. Poplar plantations are usually designed for high biomass production enabling relatively frequent financial returns for the growers. This type of plantations increase the value of low-quality and marginal arable land, making them attractive investment targets for capital investors. Many EU countries have reported their anticipation of heightened private investments in fast-growing tree species [15].

Populus species worldwide cover an estimated 60 million hectares, with a significant portion located in Canada and the Russian Federation in the form of natural aspen forests [15]. Planted poplars account for 8.3 million hectares, grown primarily for the production of industrial roundwood (5 million ha), environmental protection (1.9 million ha), and fuelwood (0.8 million ha). The main products derived from these plantations include veneer, plywood, pulpwood, wood panels, furniture, and matches. In Europe, poplar plantations occupy approximately 0.8 million ha, with Italy harvesting 50% of its annual hardwood biomass from poplar plantations.

In the Nordic-Baltic region, the current area of poplar plantations is approximately 5000 ha. Hybrid aspen is planted on an additional 12,000 ha. Consequently, Populus plantations are a minor source of raw material. Native aspen (Populus tremula L.), on the other hand, represents a valuable wood resource for pulping, match production, furniture, particle boards, packaging, biofuels, and many other wood-based products. Estonian Cell, for example, has a modern facility built in 2006 for processing Baltic aspen into bleached chemi-termo-mechanical pulp. They produce 170, 000 tons of pulp annually, requiring 450,000 m³ of aspen pulpwood [16]. This corresponds to a harvest of about 1000 ha and a total area of 25,000 ha of hybrid aspen plantations grown in a 25-year rotation. The Swedish Södra Cell pulp factory also uses aspen in a mix with birch and beech to produce dissolving pulp for textiles [17]. Pulpwood from Swedish poplar plantations is also used in Södra Cells’s raw-material mix.

Among the Baltic countries, Lithuania has fewer forest resources than Estonia and Latvia but more set aside arable land suitable for conversion to poplar plantations. Lithuania and Latvia are now major producers of wood-based panels in the region and supply global companies such as IKEA (Figure 1). Estonia and Latvia also produce large quantities of pellets, mainly for export to Denmark, Great Britain, and the USA (Figure 2) [19]. Poplar wood is ideal for making products like oriented strand boards (OSB) and medium-density fiberboards (MDF). These products represent a significant market opportunity for poplar wood in the region. While poplars are not likely to be used for veneer in the Baltic countries, they can be utilized for inner layers in plywood, while birch veneer is used for the surface layers.

Figure 1.

Particle board production in the Nordic-Baltic countries since 2001. Nordic countries—Denmark, Finland, and Sweden. Baltic countries—Estonia, Latvia, and Lithuania [18].

Figure 2.

Production of wood pellets in six Nordic-Baltic countries for the period 2012–2021 [18].

In the Nordic-Baltic countries, poplar plantations were initially established as a part of land conversion programs and for bioenergy purposes. Sweden has seen an increase in the area of hybrid poplar and hybrid aspen plantations in the past few decades, reaching approximately 4000 ha [20]. At the same time, the area of willow coppice has decreased from 14,000 ha in 2001 to just over 7000 h in 2016 [21]. One of the main reasons for the decrease in willow coppice is the significant increase in cereal prices since 2007. Similarly, the decline in Italian poplar cultivation has also been related to the profitability of alternative land use [22]. The potential of poplars as a raw-material resource in the region depends directly on the expansion of plantations on arable or other non-forested land. Fertile agricultural land is more likely to be utilized for cultivation of food and feed crops, while arable land and grasslands of poor quality, or fields that are difficult to access with agricultural machinery can be available for poplar plantations. However, grasslands can be a valuable form of land use, particularly on organic soils, providing high biodiversity and significant carbon storage potential [23]. Poplar plantations can also expand by replacing forest tree species planted on arable land during the early conversions in the mid-twentieth century. This replacement is more of a resilience issue in a changing climate rather than an opportunity to increase biomass production significantly. Despite these limitations, it has been suggested that between 1.8 and 4.6 million ha of land in the Nordic-Baltic region could be available for conversion to poplar plantations [24, 25].

3. Biology and ecology of Populus

The Populus genus comprises poplars and aspens, which can be found in various climatic conditions in the northern hemisphere, ranging from the sub-tropical regions to the Arctic [10]. Typically dioecious, wind-pollinated trees, poplars start to flower when 6–7 years old, producing large amounts of tiny, wind-spread cottony seeds. Poplars and aspens are pioneer species invading the fresh mineral deposits in riparian environments (poplars including white poplars from section Populus, see below) or on fire-disturbed forest sites (aspens). They also reproduce clonally through detached branches, stump shoots, and root suckers, which allows them to persist on already occupied sites. Eventually, secondary successional species will take over the sites pioneered by poplars and aspens. The exception is frequently disturbed riparian sites, where poplars and willows may regenerate cyclically. On the other hand, aspens may form pure or mixed forest stands but usually grow in small groves at forest edges.

The taxonomy of Populus is rather complex, the complexity being a result of low barriers to gene flow between species and even between sections of the genus leading to the emergence of intermediate forms (hybrid swarms) among sympatric poplar species [26]. Out of the six sections in the Populus genus, only three—Aigeiros, Tacamahaca, and Populus (previously known as Leuce)—contain commercially valuable species. Interspecific hybridization within sections is common but is rare or absent between sections, except for Aigeiros (black poplars) and Tacamahaca (balsam poplars), which can easily be crossed, producing valuable hybrids.

Black cottonwood (Populus trichocarpa) is a commercially essential species in the Tacamahaca section. It is naturally found in the western regions of North America, from California in the south to Alaska in the north (Figure 3). Black cottonwood typically grows in moist environments on moraines, alluvial soils, and riparian habitats. It can also thrive in interior valleys and canyons up to 2000 m in altitude. Strait stems, narrow crowns, and impressive dimensions, reaching heights over 40 m and more than 1 m in diameter [28], characterize the species. The species is easy to propagate with dormant stem cuttings. Harvested trees can resprout from stumps or even produce root suckers.

Figure 3.

Natural distribution of Populus trichocarpa (Source: [27]).

4. Nordic-Baltic poplar cultivation practices

Operative poplar cultivation in the Nordic-Baltic region was first introduced during the early 1990s. Typically, plantations were established at a 3 × 3 m spacing (≈ 1000 trees ha⁻¹), aiming at a rotation of 20 years (Figure 4). In forest-dominated areas of central Sweden, planting densities ranging from 1100 to 1700 trees per hectare are also used, with higher densities serving as a buffer against losses due to the browsing pressure. The great majority of Swedish poplar plantations were established with clone ‘OP42’ (P. maximowiczii × trichocarpa) bred in the USA at the beginning of the 1930s. In Lithuania, approximately 1700 ha of poplar plantations have been planted since 2014, mostly using a 3 × 2 m spacing (1600 trees per hectare), with the expected rotation period of no longer than 10 years (Figure 5).

Figure 4.

A 19-year-old trial with climate-adapted black cottonwood (P. trichocarpa) grown in Central Sweden at 59°N. The trial contains more than 100 clones, with the three best clones reaching a mean diameter of 400–450 mm at breast height. The chapter’s main author for scale beside a tree of clone 722.16. The trial was established with 3.5 × 3.5 m spacing and thinned systematically after 9 years. Several candidate clones for commercial deployment are currently planted in pilot plantations and operational plantations on agricultural soils in Sweden, Latvia, Estonia, Lithuania, and Romania. The best clones will be registered at the Swedish Forest Agency in 2024. Photo by Almir Karačić.

Figure 5.

Seven-year-old poplar plantation in Lithuania established at 3 × 2 m spacing. The clone used was the Italian AF7, which suffered from repeated frost injuries on this site. This stand was established using 1.5 m long poles and harvested a couple of months after the photo was taken. Lithuanian poplar plantations are established without the application of herbicides and have been certified with FSC standards since August 2023 [29]. Photo by A. Karačić.

The intensity of work to prepare land for planting varies depending on its previous use but typically involves using herbicides and cultivating the soil. Various types of planting materials are used, ranging from rooted plants (either bare root or containerized) to cuttings of varying lengths, including unrooted 2-year-old poles. In some cases, cuttings are planted through degradable plastic mulch that can be replaced by biodegradable mulch paper. However, most Nordic and Baltic entrepreneurs rely on using robust planting material and mechanical weed control after planting.

Poplar growers in the Nordic-Baltic region face specific challenges due to the relatively short growing season and cold winters. Poplar clones with better frost hardiness are required compared to the material available from central/southern European breeding programs [3]. Moreover, the market for poplar wood is limited, which makes it difficult to justify cost-intensive tending. To overcome these obstacles, poplar growers need to optimize management by combining denser initial spacing, more robust plant material, and less frequent mechanical weeding. Additional challenges are present due to high browsing and rodent populations, particularly in small, isolated poplar plantations in forest-dominated landscapes.

Site and clone selection are the two most important decisions in poplar silviculture. Poplars perform best on deep, moist, well-drained, and light-textured soils. Dry sandy soils and waterlogged soils are unsuitable as they can result in poor growth and establishment. However, the productivity of dry soils can be improved by fertilization and irrigation with wastewater, while waterlogged soils can be ditched. Wet sites are typically located in lowland terrain on organic soils with abundant ground vegetation. There is a relative abundance of these sites in the Nordic-Baltic region, so farmers prioritize these areas when converting land into tree plantations. However, there is a high likelihood of frequent frost, high browsing pressure, vole populations, and serious competition from herbaceous vegetation occurring on these sites. Combined, these factors represent a severe challenge to poplar growers who must employ simultaneous measures to be successful, including heavy soil preparation (mounding, for example), herbicide treatments, frequent mechanical weeding, and sometimes plant protection.

Deploying highly productive clones is the most efficient measure to increase the yields of poplar plantations. However, even for a small set of productive clones, there can be substantial variations in clonal performance [8, 30]. Some clones, such as ‘OP42’, are generalists and can grow well on many different soils. ‘OP42’ is outperformed by better-adapted clones on some sites or towards the edges of its deployment range. Planting clone mixtures is sometimes recommended to resolve the uncertainties related to clonal performance in different environments. This approach has both advantages and disadvantages, and the optimization of mixtures regarding the number and growth pattern of the clones in a mixture is necessary [31].

The Nordic-Baltic poplar plantations established with 1000 trees ha⁻¹ yield total biomass at an average of 10 tons ha⁻¹ year⁻¹ dry weight over a 20-year rotation. This corresponds to 8 tons ha⁻¹ year⁻¹ of stem biomass or 20–25 m³ ha⁻¹ year⁻¹ [25]. During the second half of a rotation, individual trees face intense competition, which can impact their vitality, especially during periods of summer drought [32]. Therefore, thinning is commonly utilized to promote diameter growth and vitality in the remaining trees. The recommended final stocking of Nordic-Baltic poplar stands is 650–800 trees per hectare. The timing of thinning will depend on the initial density and site productivity, usually occurring between ages 8–14 when the mean height is 15–18 m. However, thinning poplar stands pose a risk as heavy machines can damage their shallow root system. The final harvest should be performed between December and April to ensure successful resprouting from stumps and roots. Harvesting during May–September may result in the failure of the coppice regeneration [33].

5. Current genetic resources and past breeding activities

Populus breeding and selection began in Sweden during the 1930s [34], with a focus on P. tremula and its hybrids with P. tremuloides. Initial poplar trials also included P. ×euramericana hybrids to demonstrate the general production potential of poplars in southern Sweden [35]. These activities were eventually abandoned in the 1960s. Still, a portion of the tested material was transferred into several clone archives and later used in the breeding of P. trichocarpa at the Swedish University of Agricultural Sciences.

In the 1970s, interest in poplars and other fast-growing species increased due to energy crises. Several trials were established in southern Sweden, mainly with material imported via Germany and Holland. Karlsson et al. [36] evaluated some of these trials, suggesting that hybrids with species from the Tacamahaca section were best suited for southern Swedish conditions. The Swedish Forestry Research Institute (SkogForsk) continued the work with imported poplar material, expanding testing to include agricultural fields and forest soils between 56°N and 65°N [30, 37, 38]. So far, SkogForsk has registered five poplar clones with the Swedish Forest Agency, including ‘OP42’.

In the late 1980s, Ilstedt [3] conducted a test on a collection of Belgian P. ×generosa and P. trichocarpa in central Sweden. The material showed good growth potential but was sensitive to early autumn frost due to prolonged growth, leading to disruption of bud formation and winter hardening. The same material was tested in the milder climate of southernmost Sweden (55°40’N), where P. ×generosa clones failed due to stem canker [39]. Some Belgian P. trichocarpa clones have shown satisfactory growth compared to ‘OP42’ in a trial established by the Swedish west coast [8]. P. trichocarpa has been used in poplar breeding programs since the 1920s, producing commercially valuable hybrids in crossings with P. maximowiczii and/or P. deltoides. The hybrids with P. deltoides (P. ×generosa) have been widely employed in resource-intensive poplar cultivation in the Pacific Northwest [40]. In Europe, black cottonwood was also utilized in the Belgian poplar breeding program, producing high-yielding, leaf rust-resistant P. ×generosa hybrids. However, towards the end of the 1980s, the leaf rust resistance eventually collapsed, making this material unsuitable for large-scale operations. As a result, the breeding strategy was adjusted to entail tolerance rather than resistance to Melampsora leaf rust [41].

Recognizing the potential of P. trichocarpa, a project was initiated to adapt this species to the climate of central Sweden [42]. A series of more than 100 crossings involved 20 female and 30 male parent trees originating from the Pacific coastal areas of North America and more continental locations between 44°N and 60°N. Approximately 900 selected clones underwent testing for several years before planting 100 individuals in the long-term clone trial near Uppsala at 59°N. Since 2007, selected parts of this material have been tested in multiple locations in Sweden, Estonia, Latvia, Lithuania, and Poland [6, 7, 43, 44]. Since 2013, 10 candidate clones for commercial deployment have been tested in pilot plantations.

6. Adaptive phenology is one of the breeding goals

Flowering and bud flushing of poplar trees co-occur early in spring and are regulated mainly by temperature (Figure 6) [45]. In contrast, the timing of growth cessation is primarily governed by changes in day length and is known to have a significant genetic component (Figure 7) [46]. Bud formation is initiated after growth cessation and is also affected by the temperature [5]. Due to a rapid temperature cline along increasing latitudes in the Nordic-Baltic region, one of the primary goals of poplar climate adaptation is to identify genotypes with optimal timing of active growth initiation and termination.

Figure 6.

Stages of bud burst (1—buds swelled but no green leaves visible; 6—leaves fully flushed with an initial shoot increment >10 mm) for four clones grown in a common garden in Uppsala, central Sweden (59°N). Clone ‘OP42’ (P. maximowiczii × trichocarpa) initiates growth earlier than the other three clones. Clone ‘722.16’ is adapted to growth in central Sweden (see Figure 4). The Italian clone ‘AF8’ is not adapted to the Nordic climate and requires a higher temperature threshold to initiate bud burst. The Icelandic clone ‘Idunn’ is adapted to maritime conditions and will also have somewhat delayed bud flushing. The trial included 40 clones planted in 10 single-tree plots (blocks), and the number of observations for the four presented clones was ‘OP42’ = 4, ‘722.16’ = 6, ‘AF8’ = 7, and ‘Idunn’ = 7. Error bars are standard deviations.

Figure 7.

Stages of bud set (3—fully growing shoots; 1.5—initiation of bud formation; 0.5—bud formed but not matured; 0—matured bud) for four clones grown in a common garden in Uppsala, Central Sweden (59°N). The growth cessation and bud set were inventoried in the year of planting. The first-year growth can be prolonged compared to the already established plants. Data from the second growing season (2018) was not useful due to an extremely dry summer causing atypic bud set in many clones, and the inventory was not possible in 2019 due to height of trees (up to 8 m). For example, the Icelandic clone ‘Idunn’ would normally initiate a bud set already in July. Clones ‘OP42’ and ‘AF8’ are late to set bud and will be at risk of frost damage at this latitude. Early frosts can induce frost damage to the top shoot. Cold temperatures also inhibit bud formation leading to incomplete dormancy and the risk of frost damage during the winter. The trial included 40 clones planted in 10 single-tree plots (blocks), and the number of observations for the four presented clones was ‘OP42’ = 4, ‘722.16’ = 6, ‘AF8’ = 7, and ‘Idunn’ = 7. Error bars are standard deviations.

These traits are controlled by many genes with minor effects, e.g., they are quantitatively inherited. New progeny genotypes obtained by cross-pollination of individuals from distant heterotic groups will display a spectrum of phenological characteristics. Consequently, breeding for climate adaptation involves such crosses to achieve hybrid vigor, which is a common strategy in European poplar breeding programs [47, 48]. In the Nordic-Baltic region, an example of a specific breeding goal for autumn phenology is to select clones that exhibit active growth throughout August but also have a rapid bud formation and maturation in September. This allows the new poplar clones to take advantage of favorable growing conditions with optimum temperature and rainfall in August while ensuring frost hardiness. A good benchmark for the timing of bud set in poplar clones can be developed from the Swedish horticultural growing zones (Figure 8), where bud formation should be accomplished approximately 1 week earlier within each successive growing zone. However, it is important to keep in mind that the timing of frost occurrence within each growing zone can vary significantly, so local conditions and premises for clone deployment should be considered. Nonetheless, having a reasonable estimate of bud set for a particular clone and zone is essential, as the timing for a clone is likely to differ across different growing zones.

Figure 8.

The latest date for forming a stable bud is autumn within the six growing zones in Sweden. The horticultural growing zones (left) are used as guidance for clone deployment based on the timing of bud set (map from the Swedish Garden Association, with permission). The range of average dates for frost occurrence (right) is an essential benchmark for evaluating clone hardiness (map from the SMHI [49]).

7. Genetic architecture of photoperiodic traits

The genetic architecture of photoperiodic traits is complex, involving many genes with minor effects. The alleles either decrease or increase the additive effect of the different genes. Many loci in the poplar genome have been identified through QTL mapping that can explain variation in bud set traits [50, 51, 52, 53, 54, 55, 56, 57, 58, 59]. The identified genes regulate the active growth and dormancy cycle based on the perception of photoperiodic and dormancy signals [60]. The circadian rhythm regulates the active growth cycle, a biological process that displays an endogenous adjustable oscillation of approximately 24 h. External cues such as daylight and temperature determine the circadian rhythm in a specific environment. The interplay between various photoreceptors and the circadian clock is important for induced growth cessation and bud set in deciduous trees [61, 62]. The large number of candidate genes involved in adaptation to photoperiod and climate highlights the quantitative nature of these traits (Figure 9).

Figure 9.

Overlap between candidate genes of spring and autumn phenology in populations of Populus trichocarpa. Orange: wild accessions from Northwest coast of North America [63]. Lilac: wild accessions from western North America [64]. Green: hybrids between P. trichocarpa provenances originating from the populations along the latitudinal and longitudinal clines on the Pacific Coast of North America, bred at Swedish University of Agricultural Sciences and introduced to progeny trials in the Baltic Sea region in northern Europa [65]. Modified figure from Apuli et al. [65].

7.1 Circadian clock

Circadian clock genes are genes whose protein products generate and regulate circadian rhythms in different organisms. In poplar, several circadian clock genes have been identified: LHY1, LHY2, TOC1, PRR5, FLOWERING LOCUS T (FT), PhyA, etc. [52, 57]. The circadian clock resets at dawn and dusk. It involves the action of the red/far-red and blue light-receiving phytochromes (phy) and cryptochromes (cry), as well as blue light receptors from the ZEITLUPE (ZTL) family [66]. P. trichocarpa has one PHYA and two PHYB genes [67] which are involved in the control of growth in Populus [53, 68]. Many other genes are involved in the interplay of the circadian clock with the tree’s growth [60].

7.2 Heritability of critical daylength

Typically, a day shorter than the specific length required for growth evokes growth cessation and bud set. The critical day length is when about 50% of individuals of a particular clone are setting buds [60]. Fabbrini et al. [46] and Rohde et al. [5] have demonstrated that traits related to the timing of growth cessation in black poplars (P. nigra) had significantly higher broad sense heritability (H² = 0.64–0.84) compared to the traits associated with the duration of bud maturation (H² = 0.10–0.67). Similar results have been obtained in European aspen (P. tremula) for populations sampled across a latitudinal gradient in Sweden [69]. The broad sense heritability for two different wild populations of P. balsamifera was 0.37 and 0.55 for bud flush and 0.44 and 0.72 for bud set when grown in two common gardens in Canada [45]. Several studies have used genome-wide association studies to dissect the genetic basis of growth cessation in response to critical day length in several Populus species [63, 64, 65, 70, 71]. Many of these studies have identified an association between the timing of growth cessation and core genes from the photoperiod pathway and the circadian clock. In particular, the gene FLOWERING LOCUS T2 (FT2) has been implicated as a major effect locus in both P. trichocarpa [63] and P. tremula [70], indicating that parallel adaptation to similar environmental gradients target similar developmental pathways and sometimes even the same gene in different species.

7.3 Molecular markers for selection of operational varieties

Populus was early established as a model genus for studying the biology of woody perennials [72, 73]. This resulted in the early development of large-scale genomics resources and, ultimately, the first de novo draft genome for a forest tree [74]. The availability of large-scale genomics resources also resulted in early efforts aimed at studying genetic variation at the nucleotide level, first achieved by targeting short genomic regions [75, 76, 77, 78] and ultimately through whole-genome re-sequencing [63, 79].

Thus, abundant genomic information is available for many species in the genus Populus that can be used for developing molecular breeding tools. The most abundant genetic marker type is single nucleotide polymorphisms (SNPs), which can be detected in the millions in most Populus species, e.g., Wang et al. [79]. To date, large SNP data sets are available for most species in the genus Populus. Methods for rapid and cost-efficient genotyping have also been developed in a few other species [80, 81, 82, 83]. In addition, many genetic markers have also been associated with traits of interest through genome-wide association studies [64, 70, 84], making it possible to directly target variation underlying traits that are of interest as targets in e.g., breeding programs [65].

8. Timeline for the implementation of molecular markers in selection

Genomic selection (GS), initially proposed by Meuwissen et al. [85], is a method that uses genetic markers with genome-wide coverage and phenotypic information from a training population to develop a model that can be used to predict the breeding values for individuals from a breeding population with only genotype information available. GS has many advantages that increase the method’s utility in modern plant breeding. First, GS can be used to perform early selection, often even at the seedling stage, without the need for extensive field testing and screenings. This facilitates early selection, which reduces the time required for successive plant breeding cycles and thus contributes to an increased genetic gain per unit of time. Second, the estimated marker effects from GS are precise, and unbiased prior marker selection is avoided [86, 87]. GS is particularly suitable for trees due to their long generation times and traits characterized by a genetic architecture consisting of many underlying genes, each with relatively small effects. Furthermore, traits of interest for tree breeding are often expensive to phenotype or are displayed late in the life cycle, further increasing the utility of GS.

Implementing a genomic selection program for tree breeding encompasses two stages; the first stage relies on a “training population” consisting of individuals that are both phenotyped and genotyped. The combined genotype and phenotype data are used to develop a predictive model that links variation in genetic markers to variation in phenotypes of interest. Training populations are usually derived from existing progeny trials or breeding populations, where previously selected ‘elite’ parents have been crossed, and their progenies have been extensively tested in field trials. Breeding populations in forest trees usually contain 1000–2000 individuals and have effective population sizes (N_e) in the range of 30–100. The larger the training population, while keeping N_e in the appropriate range, the more precisely marker effects are estimated, ultimately resulting in a more accurate predictive model [88].

GS is then employed by genotyping a large number of individuals that form ‘selection candidates’, usually consisting of full- or half-sib families derived from individuals that are part of the training population. The genotype information is used with the prediction model to estimate genomics-based genotypic values (GEGVs) for all selection candidate individuals. Top-ranking individuals from the selection candidates, based on the GEGVs, are then selected and used to establish the next generation of the breeding program. To further enhance testing, top-ranked selection candidates can also be clonally propagated and tested in clonal trials, where elite clones are eventually selected for operational plantation, especially for tree species that rely on clonal deployment, such as Eucalyptus and Populus. A random subset of the selection candidates can also be planted in an experimental design trial and phenotyped at the target age to update the GS model and ensure that the accuracy of GS predictions remains high over successive generations.

Large training population sizes have thus far characterized studies of GS in forest trees, and large numbers of genetic markers have also typically been used, especially when compared to GS studies in crops. Recent studies in forest trees include a diverse array of species, including eucalypts [89, 90], spruces [91, 92], pines [93], and Populus [40, 94, 95]. These studies have shown that the ability to predict complex traits in forest trees is high and suitable for implementing GS as a breeding tool to increase the efficiency of tree breeding programs. Furthermore, in species that utilize clonal deployment, such as eucalypts and poplars, GS can reduce or eliminate the progeny trials and the time and costs of clonal testing trials by minimizing the number of selected genotypes propagated as clones.

GS’s utility depends on fundamental population and quantitative genetics aspects as well as more practical and logistical aspects relating to resource allocation and cost-benefit considerations. The accuracy of a genomic prediction model is the most important aspect of the success of GS, and many factors immediately impact GS model accuracy. The first important aspect is the effective size of the training population (N_e), which directly influences the extent of linkage disequilibrium (LD), which, in turn, dictates the necessary genetic marker density needed for successful model building. The following essential aspect is phenotyping accuracy in the training population, which sets an upper limit to how much variation can be explained by marker effects. Similarly, the heritability and genetic architecture of the traits of interest are essential, and traits with high heritabilities generally have higher prediction accuracies. Finally, the statistical methods used for model building and deployment are also relevant, and this is currently an area of active research. Simulation studies have partially assessed all of these factors and have provided some general guidelines for implementing GS in forest trees [86, 89].

The extent of LD is perhaps the most important aspect influencing the accuracy of GS, and LD, in turn, depends on the N_e of the training population. The extent of LD directly affects the marker density needed for successful implementation of GS, and marker density has been shown to tightly scale with N_e of the training population, where larger populations need more markers. The level of LD between markers and the unknown quantitative trait loci (QTLs) controlling traits of interest can be increased by reducing N_e. Previous studies have shown that to maintain reasonable levels of LD while maintaining sufficient genetic diversity to sustain long-term genetic gains in breeding, N_e in the range of 40–100 is typically recommended. These N_es typically correspond to census sizes of 100–200 related individuals; for example, most advanced GS populations in Eucalyptus have N_es around 30–60 [86, 96]. For such small values of N_e, the accuracy of GS can be maintained with marker densities averaging 2–3 markers per centiMorgan (cM). For a genome of 1500–2000 cM, typical for many species of, e.g., Eucalyptus and Populus, ~5000 SNPs would be sufficient for most practical applications of GS. However, as N_e increases, the density of markers also must increase to maintain GS accuracy, and up to 20 markers/cM can be needed to maintain accuracy as N_e approaches and exceeds 100–200 individuals [86], requiring genotyping methods that routinely and reliably can genotype 20,000–50,000 informative markers depending on the size of the target genome. Another critical aspect of marker density worth emphasizing is that higher marker densities will facilitate the preservation of rare alleles and thereby contribute to long-term gains from selection in the breeding population [97].

The design of the training population depends on the actual breeding strategy that is adopted. Training populations are established using trees from existing progeny trials based on crosses among elite parents. The relatedness between the training population and selection of candidates is another factor that has great importance for the success of GS. Increasing the genetic relationships between the training population and selection candidates also leads to higher prediction accuracies in a manner similar to the effects of reducing N_e [98]. Similarly, the genetic architecture underlying trait variation has significant consequences for the accuracy of GS. A small number of loci controlling a large proportion of the phenotypic variation allows for more variation to be captured compared to more complex genetic architectures involving more loci, each with a smaller effect size [99]. GS accuracy drops with the increasing number of QTLs contributing to a trait, and this effect is more pronounced when marker density is low or in populations with large N_e [86]. Traits heritability only has minor impacts on GS accuracy as long as the training population size is large enough to estimate marker effects adequately.

Based on the issues outlined in the preceding paragraphs, the prospects for implementing GS in Populus are good. There are abundant genomic resources, including large SNP data sets, available in many Populus species. What perhaps is lacking are cheap and reliable methods for large-scale genotyping of SNP data sets of the size that are useful for establishing GS (5–50k SNPs). Although SNP chips have been developed for some species of Populus [80], no commercial alternatives are available akin to the EuCHIP60K that has been developed in Eucalyptus and can be used across several species in the genus [100]. However, several genotyping methods suitable for targeted genotyping have recently started to become commercially available at costs that make them feasible for commercial GS programs. In addition, many species of Populus have extensive clone collections available, and many species also have comprehensive progeny testing programs ongoing, meaning that ample material is available for use in breeding programs and in establishing training populations.

9. Conclusions

Extensive collaboration between forestry sector and academia in the Nordic-Baltic region over the past decades has resulted in readiness for successful hybrid poplar breeding programs in the region. The existing collections of Populus clones with large genetic variations in adaptive phenology originating from different geographical regions and breeding programs in the northern hemisphere have paved the ground for the establishment of training populations with effective population sizes for successful genomic selection. Deployment of genomic selection models in poplar breeding reduces the duration of breeding cycles and leads to availability of complementary woody raw material for the forestry sector in the Nordic-Baltic region. Poplars are well-suited for genomic selection due to the abundance of genomic resources available, including large panels of genetic markers. Also, target phenology traits, such as growth cessation, have generally high heritabilities suggesting that selection will be efficient. Hybrid poplars as early successional trees grown in short rotations of 10–20 years provide a significant complementary wood source for Nordic-Baltic forest industries that today rely mostly on secondary successional tree species grown in rotations of 50–100 years.

Acknowledgments

The authors will greatly acknowledge our partners from the forestry sector who have been driven in innovations in the procurement of woody raw material in the Nordic-Baltic region: Anders Ekstrand from Södra Skog AB, Lars-Georg Hedlund from Södra Latvia SIA and Mindaugas Šilininkas from Euromediena UAB. This chapter has been written within three following projects during the last decade: (1) an EU project within the EUREKA program Eurostars E! 8443—SnowTiger during 2014–2016, (2) grant number 942-2016-20001 from Swedish Research Council FORMAS “Climate-adapted poplar through more efficient breeding and better tools for matching genotype and site—developing the poplar bio-economy market in Sweden and the Baltic Region” during 2016–2021. Finally, (3) this chapter was finalized within the “NutriBiomass4LIFE” project (LIFE17/ENV/LV000310) co-financed by the Swedish Energy Agency (P-45082-1).