The studied LP stands were located in the hemiboreal forest zone, where coniferous and broadleaved trees are mixed at both the stand and landscape levels (Ahti et al. 1968; Rendenieks et al. 2022). The study area represents lowland conditions (< 312 m a.s.l.) with generally flat topography; the soils are deep glacial and post-glacial, mostly mineral (sandy and silty; Pilvere et al. 2018). According to the national forest inventory, forests cover 54% of the country, with forested areas evenly distributed across the country, and among them, 26% are dominated by Scots pine (Pinus sylvestris, SP). The total area of LP stands in Latvia is ~40 ha (State Forest Service 2025). The LPs are mostly planted next to SP or in a mixture with it.

The climatic conditions of the study area can be described as moist continental, yet the continentality increases eastwards with growing distance from the Baltic Sea (Kottek et al. 2006). During the period 1996–2025, the mean annual temperature was +6.6 °C, with February being the coldest (a mean monthly temperature of –3.2 °C) and July the warmest month (17.7 °C). The mean annual precipitation was 668 mm, and the highest monthly precipitation fell during the summer months (May–September; ca. 74 mm/month) (Harris et al. 2020). In the Baltic Sea region, due to global warming, the vegetation period (mean diurnal temperature > 5 °C) has extended by ca. 6–14 days during the last century (Meier et al. 2022). However, late spring frosts have become more frequent and intense in recent years (Meier et al. 2022). If these frosts occur immediately after germination, they can severely threaten the survival of LP seedlings (Langdon et al. 2010).

To study the self-regeneration of LP, all plantations that could produce seed, i.e., older than 20 years (Langdon et al. 2010), were selected from the National Forest Service database and the registry of introduced species hosted by LSFRI Silava. In total, 19 stands, aged 28 to 47 years, and one stand aged 87 years were selected for surveys (Fig. 1). The area of the plantations ranged from 0.2 to 1.0 ha (mean 0.58 ha). Mostly, the origin of LP was known. Most of the plantations consisted of the variety Pinus contorta var. latifolia, likely representing two provenances. Still, three plantations contained 27 provinces from different geographical areas of Canada and the USA. In most cases, LP was planted in a mixture with the same age SP (randomized blocks). The proportion of LP canopy trees in the observed plantations ranged from 25 to 93% (mean 65%). In six plantations, up to 30% of the canopy LP were dead due to fungal infections (Zaļuma et al. 2019). The vast majority of the LP plantations were also bordering with native SP or Norway Spruce (NS, Picea abies (L.) H. Karst.) stands of different ages. In general, the observed LP plantations were growing on freely draining, oligotrophic mineral soils with site types corresponding to Oxalidosa (50%), Myrtillosa (25%), and Vacciniosa (25%) (national classification according to Bušs 1997). Three plantations bordered a local wetland (swamp), located in a topographic depression.

At each plantation, four 100 m long transects were established in four perpendicular directions from a common origin, which was located close to the geometric centre of the stand, forming a cross (Fig. 1). The transects were offset by 3 m from the common origin. The directions and origin of the transects were chosen so that approximately half of each transect lay within the plantation (LP planting) and the other half was outside of it. If the transect crossed an infrastructure object, it was extended accordingly. Hence, some of the transects were interrupted. Each of the transects was split into 10 continuous plots (3 × 10 m; Fig. 1). A category, “inside the parent stand” (located within the boundaries of the LP plantation, and/or LP was present in the canopy) or “outside” was noted for each plot. In total, 714 plots were assessed: 352 inside and 362 outside.

Within each plot, seedlings and saplings of all regenerating trees that would potentially reach canopy (advance growth) and shrubs, which are not potentially reaching canopy (undergrowth), according to species, were accounted, and their height was measured with an accuracy of 5 cm. For all species, except LP, advance growth and undergrowth up to 1 m in height were recorded; for LP, self-regeneration of any height was noted. The presence of browsing damage (binary) was noted for each LP seedling (regenerating LP). Within each plot, the number of canopy (upper and lower layer/storey) trees according to species was recorded (including snags), of which the number of LP trees with cones was additionally noted. In each plot, the relative cover of undergrowth, vascular plants, and bryophyte layers, as well as bare soil, was recorded. Other site characteristics potentially affecting self-regeneration (e.g., location in canopy opening or forest road, presence of tree-throw pit or large amount of lying deadwood, locally changed moisture conditions; Engelmark et al. 2001; Ledgard 2001; Jacobson and Hannerz 2020) were noted. In cases where LP seedlings were detected in the last plots of a transect (outside the parent stand), the search for LP self-regeneration was continued, extending the transect until no LP seedlings were found. In the extended part of the transect, only the most distant LP seedling (GPS position) was noted, aiming to estimate the dispersal distance of LP seedlings (Langdon et al. 2010). The stands were surveyed at the end of the summer of 2025.

To describe the canopy, the density of trees by species was calculated for both the LP plantation (inside) and the surrounding area (outside). For the description of self-regeneration, the composition (proportion of total individuals), mean density, and height of each regenerating advance growth and undergrowth species were calculated for each site (plantation and surroundings). A linear mixed-effects model was used to assess the effects of tree stand and site properties on the density of LP seedlings. To account for dependencies in the data arising from the sampling design, site (plantation), and transect were used as nested random effects. A total of 28 variables most commonly found to significantly affect the LP density were tested as fixed effects. The variables were: stand age, site type, upper canopy density, lower canopy density, LP density in the canopy (upper and lower), LP seed tree density, SP density in the canopy, NS density in the canopy, silver birch (Betula pendula Roth.) density in the canopy, common aspen (Populus tremula L.) density in the canopy, pedunculate oak (Quercus robur L.) density in the canopy, Norway maple (Acer platanoides L.) density in the canopy, density of dead standing canopy trees, density of dead standing LP, density of advance growth, density of birch in advance growth, density of NS in advance growth, density of SP in advance growth, density of aspen in advance growth, density of undergrowth, density of the total understory, relative cover of bryophyte layer, relative cover of vascular plant layer, relative cover of undergrowth, relative cover of bare soil, growth condition characteristic, distance from the edge of the LP parent stand, location of plot inside or outside the parent stand. An arbitrary bidirectional selection principle was implemented when refining the set of the best-performing fixed effects. Akaike Information Criterion, as well as model performance metrics, were used during the selection. The significance of the fixed effects was evaluated by the type II Wald’s χ² test. The levels of significant categorical predictors were compared using Tukey’s post hoc test. The differences in density and height across understory species (self-regeneration) were evaluated using a linear mixed-effects model, accounting for site (plantation) and transect as nested random effects. The same model was used to determine the differences between seedling density and height ‘inside’ and ‘outside’ the parent stand. The models’ compliance with statistical assumptions was evaluated using diagnostic plots. Collinearity of the predictors was evaluated using the Variance Inflation Factor; none of the predictors showed values exceeding 5.0, indicating an absence of multicollinearity. In total, 714 plots were used in the analysis. The data analysis was conducted at the α = 0.05 significance level in R v. 4.5.3 (R Core Team 2026) using the package ‘lme4’ (Bates et al. 2015) and ‘lmerTest’ (Kuznetsova et al. 2017).

The canopy was generally dense within the studied plantations, and the surrounding areas covered by the transects. The mean density of the canopy trees was 768 ± 50 trees ha^-1, and there was also a second canopy layer with the density of 141 ± 50 trees ha^-1. As the LP plantations were studied, the upper canopies were composed of LP (mean ± SE 279 ± 44 trees ha^-1), SP (301 ± 39 trees ha^-1), NS (100 ± 34 trees ha^-1), and birch (75 ± 24 trees ha^-1). The lower canopy was formed by NS (67 ± 32 trees ha^-1) and birch (39 ± 14 trees ha^-1). In general, canopy LP trees without cones were observed only in one of 20 plantations. On average, 79% of canopy LP trees had cones (167 ± 23 trees ha^-1). Within 8 plantations, dead standing LP canopy trees were counted, where they formed 18% (9–35%) of all LP canopy trees (17 ± 6 trees ha^-1).

Understory composition within the studied stands—both inside and outside LP plantations—was remarkably rich for oligotrophic sites. A total of 11 advance (tree) and 20 undergrowth (shrub) species were listed (Table 1), with a mean density of 2381 ± 115 and 1178 ± 73 individuals ha^-1, respectively. The proportion of advance and undergrowth species were 70% vs. 30%, respectively. Species dominating in the canopy, birch and NS, also formed the densest understory (Table 1). In advance growth, the proportions of birch was 30%, but for NS, it was 28%. Oak formed 12%, but aspen and maple 11% of the advance growth, indicating explicit admixture. The total density of advance growth, as well as for most species individually, in the ‘inside’ and ‘outside’ plots was similar (Fig. 2). The most common undergrowth species were willows (Salix spp.), rowan (Sorbus aucuparia L.), alder buckthorn (Frangula alnus Mill.), and dwarf serviceberry (Amelanchier spicata (Lam.) K. Koch), which formed 8, 26, 27, and 31% of stems, respectively.

Although SP was common in the canopy of the surroundings of the plantations, the total share of SP seedlings was low (5%, i.e., 103 ± 32 seedlings ha^-1). The overall mean density of LP seedlings was higher, reaching 222 ± 74 seedlings ha^-1, indicating more sufficient regeneration. Seedlings of LP were found in 9 out of 20 plantations (occurrence 45%), and the proportion of LP seedlings (of advance growth) in these plantations was 19% (1218 ± 62 seedlings ha^-1). Eight plantations with LP self-regeneration were located in the central part of Latvia, highlighting some regional characteristics (Fig. 1). The mean height of the LP seedlings (0.4 m) was similar to that of the other advance growth species (~0.5–0.6 m, Table 1). Self-regenerated LP over 1 m high (maximum height 2.6 m) were found in 5 stands, in three of which cones on the regenerated LP were observed.

The mean density of LP seedlings in plots inside plantations was 240 ± 78 seedlings ha^-1, but outside, 234 ± 103 seedlings ha^-1 (p = 0.84, Fig. 2). However, the density of LP seedlings outside the plantations showed higher variation among the stands, indicating local specifics in spread. The location of seedlings along the transect showed that seedlings had spread at least 70–80 m outside of the LP stand (the furthest established plot). When assessing seedlings outside the transect (in the direction of the transect), a few scattered LP seedlings were found in only 3 sites, and the furthest LP seedling was recorded 89 m from the edge of a plantation.

The density of LP seedlings was significantly affected by the tested variables, and the fixed effects included in the refined statistical model were able to explain 47% of the variance in the data (R² = 0.47). The effect of the plantation was generally low, as the random variance comprised up to 10% (ICC = 0.10), indicating moderate local effects (Table 2). The refined set of the fixed effects consisted of relative cover of bryophyte and bare soil layers, mean density of advance growth, and specifically density of birch and Norway spruce in advance growth (p < 0.001; Table 2). Thus, the set of significant site properties indicated effects of competitiveness, likely via light availability, moisture regime, and growing space. Although the estimated LP seedlings density in sites with locally increased moisture (Fig. 3) was generally low, LP seedlings were significantly more abundant in sites where increased humidity and light were combined, particularly in moist canopy openings (estimated density 1214 ± 340 seedlings ha^-1), indicating the primary limiting effect of light. The highest estimated density of LP seedlings was observed on forest roadsides (1938 ± 340 seedlings ha^-1), as well as in sites with open conditions: in moist canopy openings, on forest paths (616 ± 325 seedlings ha^-1), and canopy openings (580 ± 114 seedlings ha^-1) (Fig. 3). Seedlings of LP were absent or poorly self-regenerated in sites with dense herb layer (Fig. 3), in turn, sites with bare soil and sites with high bryophyte cover had a positive effect on the abundance (p < 0.001, Table 2). The density of advance growth had a significant effect on LP seedlings density (p < 0.001, Table 2), with the density of spruce and birch seedlings having a negative impact (p < 0.001, Table 2). Negative effect was also observed for browsing, which was particularly for the tallest seedlings, as 24% of the LP seedlings over 30 cm in height had browsing damage (not shown).

The high proportion of LP canopy trees with cones (79%) and the observed density of seed trees within the plantations (166 ± 23 trees with cones ha^-1) indicated that seed appeared sufficient to ensure abundant self-regeneration (Langdon et al. 2010; Jägbrant 2014; Novotný et al. 2018). However, the average density (222 ± 74 seedlings ha^-1) of LP seedlings was low compared to regions where LP show invasive properties (e.g., Iceland, Chile Patagonia, as well as South America), where the densities exceeded 7000 seedlings ha^-1 (Peña et al. 2008; Langdon et al. 2010; Eggertsson et al. 2022). The estimated regeneration density was slightly higher than in Sweden (overall mean of 6.8 seedlings ha^-1, and 221.0 seedlings ha^-1 in plots with LP seedlings present; Jacobson and Hannerz 2020), where pre-emptive concerns about invasiveness have been raised (Despain 2001). It must be admitted that in some plantations, self-regeneration was higher, reaching 1807 seedlings ha^-1, thus indicating some potential preconditions to accelerating the spread of LP.

The age of the studied stands did not affect LP self-regeneration (Table 2), and even though the surveyed stands were at an age (28–87 years) when large seed yields are typical (Rejmanek and Richardson 1996; Langdon et al. 2010; Jägbrant 2014; Novotný et al. 2018), and the subsequent rapid growth of new plants (Grotkopp et al. 2002) is expected, relatively low occurrence (45% of stands) of LP seedlings was found. Similarly, in Sweden, 53% of plantations over 25 years contained LP seedlings (Jacobson and Hannerz 2020).

The observed distribution of LP contradicted distance dependant dispersion of seedlings (Langdon et al. 2010; Sjödin 2012; Jacobson and Hannerz 2020), as the distance from the edge of the lodgepole pine parent stand did not affect LP seedling density (p = 0.95, Table 2), nor LP seedling density inside and outside the plantations was similar (Fig. 2). Alternatively, there might be an alternative spread mechanism depending on landscape heterogeneity (Milani et al. 2020). Considering that the seeds are airborne, it must be admitted that the observed dispersal distance was low (Ledgard 2001). In Sweden, seedlings have been similarly found mostly 0–15 m from the planting edge (Sjödin 2012; Jacobson and Hannerz 2020), but in Iceland, where the spread has been faster, 0–40 m (Eggertsson et al. 2022). In contrast to several studies where the dispersion coincided with the predominant winds (Engelmark et al. 2001; Peña et al. 2008; Langdon et al. 2010; Eggertsson et al. 2022), the dispersal direction in Latvia was random. The highest spread of 70–80 m outside the plantations, however, corresponded to the most common seed fall distance 50–100 m (Engelmark et al. 2001; Langdon et al. 2010; Eggertsson et al. 2022). Seedling occurrence according to the extended transects confirmed that seedlings spread over long distances only in rare cases (three of 20 plantations) (Langdon et al. 2010; Eggertsson et al. 2022). The most distant seedling was found 89 m from the edge of the LP plantation, which can be considered a relatively short distance, since, for example, in Iceland, the farthest seedlings were counted 760 m from the seed source during 10-year period (Eggertsson et al. 2022) and the mean annual spread rate of LP in the last decade was 61.6 ± 40.2 m/year (Wasowicz et al. 2025). According to the criteria for invasiveness proposed by Richardson et al. (2000) (i.e., dispersal distances > 100 m over < 50 years), the LP population in Latvia can soon be considered invasive; however, repeated studies should be conducted to confirm this.

Although LP is considered one of the most invasive conifer species (Despain 2001; Richardson and Rejmanek 2004), the studies on its self-regeneration and factors affecting it in Europe, except Sweden and Iceland, have been rare. LP tolerates very low temperatures (Langdon et al. 2010) and is already broadly common in planted stands in Sweden’s harsher climate (Elfving et al. 2001; Engelmark et al. 2001; Sjödin 2012; Jacobson and Hannerz 2020), the northern timberline in Finland (Ruotsalainen 2006), and subarctic Iceland (Eggertsson et al. 2022; Wasowicz et al. 2025). Therefore, it was assumed that Latvia’s climate would not strictly limit the regeneration of LP. Considering the long self-regeneration opportunities (age 28 to 47 and 87 years), and the fact that low temperature does not seem to be a significant factor in seedling mortality (Despain 2001), site properties were limiting (Jacobson and Hannerz 2020).

The presence and effect of serotine variety when cones open after being subject to intense heat, generally after a forest fire (Langdon et al. 2010; Novotný et al. 2018), has not been assessed, maintaining uncertainties regarding the invasive potential of the studied LP. The high representation of serotinous cones is characteristic of variety latifolia (Klinka 2002; Langdon et al. 2010), which was mostly present in the studied plantations. The genetics of the regenerating LP might also have had the pace of the spread (Carlin et al. 2025), though such an investigation was outside the scope of the current study. Still, it can be speculated that provenance has low effects on dispersion of LP as plantations contained a variable set of provenances, yet did not differ explicitly by seeding densities.

The self-regeneration and spread of LP was affected by site rather tree level characteristic (Table 2, Fig. 3) as observed in Sweden (Richardson et al. 1994), indicating dependence on the availability of seedbeds and competition, as well as symbionts (Despain 2001). The observed intensive browsing damage on the larger LP seedling highlighted the explicit limiting effects of ungulates (Zamora Nasca et al. 2018). Previous studies in Latvia have already shown that 68% of the 30-year-old LP had browsing damages, and the density of the ungulate had a significant impact on the growth and survival of pines (Arhipova et al. 2015; Done et al. 2025). In Sweden, however, 60% smaller browsing damages were registered in LP stands than in SP stands (Horstkotte et al. 2023). Therefore, further monitoring of browsing damage is needed.

Due to competition, forests with dense understory and shrublands are less favourable for LP invasion (Richardson et al. 1994). Although the density of undergrowth does not affect the density of LP, the effect of advance growth was significant (Table 2). Negative effect (Table 2) of the most common spruce and birch advance growth (Fig. 3) indicated interspecies competition, which shapes the regeneration of LP within the native range (Stuart et al. 1989). In subarctic Iceland, where expansion of LP in native birch woodlands and heathland has been observed, the ground cover vegetation is mostly absent or very sparse (Eggertsson et al. 2022; Wasowicz et al. 2025), thus favouring regeneration of LP (Bravo-Monasterio et al. 2016; Eggertsson et al. 2022) as LP establishment is obstructed by ground cover (Ledgard 2001; Peña et al. 2008). In the studied stands, however, only in the plots with high raspberry cover, the regeneration of LP was limited (Fig. 3), indicating the effects of competition (Ulanova et al. 2007).

Light is considered one of the main factors affecting the establishment and survival of LP seedlings (Lotan and Critchfield 1990; Coates 2000). Still, the total density of the first and second canopy layers (alive trees), that is, the primary light-limiting factor, was not significant, likely as the canopy density was still high (923 trees ha^-1). On the other hand, canopy dieback has been shown to have a limited effect (McAlpine et al. 2016), hence, the mortality of separate canopy trees did not affect the regeneration of LP. Possibly, the gaps formed due to separate trees or small groups were too small to provide sufficient light. The same relationship has been observed in the subalpine zone of Colorado (Veblen 1986), as substantial openings (two canopy heights) are needed for the establishment (Bernier 1987).

Higher estimated self-regeneration in open conditions (e.g., canopy openings, forest path, Fig. 3), locally disturbed areas (presence of bare soil), and sites with less competitive undergrowth species (e.g., higher cover of bryophyte rather than vascular plants or undergrowth), are fully consistent with other studies (Engelmark et al. 2001; Ledgard 2001; Jacobson and Hannerz 2020). Although fire has been mentioned as the main disturbance facilitating regeneration of LP (Elfving et al. 2001; Engelmark et al. 2001; Coates 2000), in Latvia, forest fires have become increasingly rare in recent decades (Donis et al. 2017), shifting focus to other disturbances. In Sweden, 63–96% of sites containing LP seedlings were found on disturbed grounds (wheel tracks, roadsides, sites with soil preparation) (Sjödin 2012; Jacobson and Hannerz 2020). The positive relationship between LP density and proportion of bare soil in the plot (Table 2) complied with the effects of the disturbance (Elfving et al. 2001; Engelmark et al. 2001), potentially facilitating regeneration and invasions. Though preconditions for invasions have been predicted by the high ability of LP to colonise most of the disturbed sites also beyond the seed shadow (Richardson et al. 1994; Langdon et al. 2010), which was not the case in this study.

The studied plantations corresponded to theoretical invasion sites, as LP primarily invades areas where SP and partly NS occur (Langdon et al. 2010). However, the regeneration densities (Table 1, Fig. 2) were generally too low for effective initiation of invasion (Peňa et al. 2008; Langdon et al. 2010; Eggertsson et al. 2022). Still, the competitiveness of LP is higher in areas where both pine species occur together, especially in moist and shady sites, although in such conditions, LP may be less fertile (Elfving et al. 2001; Knight et al. 2001).