Effect of seedling size on post-planting growth and survival of five Mexican Pinus species and their hybrids

Introduction

Forest ecosystems and their structural characteristics are affected by anthropogenic disturbance and extreme climate variations (Bellard et al., 2012; Franklin et al., 2016; Danneyrolles et al., 2019). Most natural forests have undergone fragmentation, leading to a reduction in the amount and quality of the environmental services they supply (Mitchell et al., 2015). In Mexico, a total area of 4.2 million hectares of forest land, representing 6.2% of the total forest cover, was deforested in the period 2001–2021 (CONAFOR, 2022). In response, both federal and state governments have taken actions to reverse the reduction in forest cover, through reforestation programs (CONAFOR, 2015).

The genus Pinus is the most widely used in Mexico to restore forest areas for ecological, economic and social reasons. Several species of this genus are valuable for timber, fuel and other resources (Rzedowski, 2006; Jin et al., 2021). The species Pinus arizonica Engelm., P. durangensis Ehren., P. engelmannii Carr., P. leiophylla Schl. & Cham., and P. teocote Schiede ex Schltdl. are abundant, economically important and widely distributed throughout northern Mexico (Hernández-Velasco et al., 2021). Furthermore, P. arizonica, P. engelmannii, and P. leiophylla also occur naturally in the southwest of the USA (Haire et al., 2022). Hybrids have been reported in these tree species (Hernández-Velasco et al., 2021). However, differences in growth and survival between these hybrids and their pure species parents were hardly checked in field trails (Sánchez-Hernández et al., 2022).

Seedling survival rate in reforestation programs in Mexico is extremely low. Mexal, Cuevas-Rangel & Landis (2008) reported that survival rates of five conifer species in Central Mexico in 1995 ranged from 15% to 86% after 2 years. Seedling quality was the cause of mortality in 0–10% of cases, depending on the species. However, data from across Mexico, compiled by Prieto-Ruíz et al. (2016) for the period 2006–2014, revealed an average mortality rate of 57% in the first year of reforestation (43% in the state of Durango). Up to 9.7% of the mortality was attributed to poor quality of the seedlings. Barrera-Ramírez, López-Aguillón & Muñoz-Flores (2018) reported that the percentage of conifer mortality at national level fluctuated between 40% and 70% between 2007 and 2017, in only the first year after plantation establishment.

It is commonly accepted that success on seedling growth and survival in the plantation field greatly depends on the quality of the seedlings (especially in nursery seedlings) (Tinus, 1974). Grossnickle & MacDonald (2018) reviewed at least 171 studies on the effects of seedlings traits on plant growth (root or shoot), observing that the use of high quality seedlings can increase the chances of successful plant establishment and growth.

Larger seedlings generally have an advantage in terms of initial growth and survival due to their larger root systems and more developed tissues, which initially can access more nutrients and water (Dierauf & Garner, 1996; Garau et al., 2009). Moreover, the robustness of the seedlings seems to affect the plant performance in the field (Li et al., 2011).

However, there is no experimental evidence showing that the seedling dimensions of the economically important tree species Pinus arizonica, P. durangensis, P. engelmannii, P. leiophylla, P. teocote and their hybrids have effect on their survival and growth in reforestations and plantations in Mexico. In order to contribute to this knowledge, in the present research we hypothesized that the growth and the survival rate in the field of these five species, are also likely to be enhanced by using nursery seedlings with larger dimensions, as has been reported for other species (Dey & Parker, 1997; Prieto-Ruíz et al., 2007; Tsakaldimi, Ganatsas & Jacobs, 2013). Therefore, the aim of this work was to evaluate the influence that the initial morphological parameters (root collar diameter and height) of nursery seedlings (planted at the age of 15 months) of these five pine species and their hybrids, had on their growth and survival at 44 months after planting.

Materials and Methods

Results

Across all species and their hybrids, there was a correlation between seedling root collar diameter before planting and 44 months after planting (RCD vs. RCD₃) (r_s = 0.29, CI_95% = [0.23–0.35], p < 0.00001) as well as between height of the seedlings before planting and after 44 months of planting (H vs. H₃) (r_s = 0.40, CI_95% = [0.33–0.44], p < 0.00001). However, there was only a weak, although significant, correlation between H and SR (r_s = 0.10, CI_95% [0.06–0.15], p = 0.000013) and also between the robustness index (HRCD) and SR (r_s = 0.06, CI_95% [0.02–0.11], p = 0.007).

When the five species and their hybrids were grouped together, the RCD before planting in the two field trials was often significantly and positively correlated with the RCD₃ of the seedlings, but only slightly correlated with the H and SR 44 months after planting. By contrast, in the group of seedlings derived from pure P. teocote (PT-P), there was a significant negative correlation between RCD and SR (r_s = −0.20, p < 0.05). The H of the seedlings of the five pine species and their hybrids before planting was sometimes significantly and positively correlated with H₃ and SR of the plants, in both field trials after 44 months. In some of the studied pine species and their hybrids, 44 months after planting, HRCD was significantly and negatively correlated with RCD₃ and significant positively correlated with SR. Nevertheless, there was a high degree of variation in the magnitude of the correlation between the variables (Fig. 1, Table S3).

Across all species and their hybrids under study, there were correlations between the seedling root collar diameter before planting and 44 months after planting (RCD vs. RCD₃) (r_s = 0.29, CI_95% [0.23–0.35], p < 0.00001) as well as between height of the seedlings before planting and 44 months after planting (H vs. H₃) (r_s = 0.40, CI_95% [0.33–0.44], p < 0.00001). But there was only a weak, although significant, association between H and SR (r_s = 0.10, CI_95% [0.06–0.15], p = 0.000013) and also between the robustness index (HRCD) and SR (r_s = 0.06, CI_95% [0.02–0.11], p = 0.007).

Analysis of the classification and regression models using Random Forests resulted in mean significant, unbiased conditional R²c values of RCD, H and HRCD, with RCD₃, H₃ and SR between 0.10–0.18 (Table 2) and 0.14–0.16 (Table 3), 0.10–0.15 (Table 4), respectively. Furthermore, the predictor variable “seed provenance” (23 provenances) also significantly explained the variation in SR, RCD₃ and H₃ (unbiased conditional R²c = 0.06–0.17). By contrast, the variables “tree species and hybrids” (10 classes) and “field trial” (two classes) were the predictor variables that explained the least amount of variation in these three response variables. Besides, the predictor variable “seedling of pure or hybrid parents” (two classes) was unable to explain the variation of SR, RCD₃ and H_3. The biased marginal values R²m often differed greatly from R²c (Tables 2–4).

The GLM resulted in mean significant, although very low unbiased conditional R²c values of RCD, H and HRCD, with RCD₃ and H₃ between 0–0.03, respectively. Moreover, the predictor “seed provenance” significantly contributed to explain the variation in RCD₃ and H₃ (R²c = 0.03 and 0.05; (CI_95% [0.01–0.05] and CI_95% [0.03–0.07])). Finally, the GLM predictor variable “field trial” significantly contributed to explain the variation in SR with R²c equal 0.09 (CI_95% [0.06–0.12]).

When grouped in classes of 2 mm, 2 cm and 0.3 cm/mm² width, the values of RCD, H and HRCD of the 15-month seedlings (before planting) affected RCD₃, H₃ and SR 44 months after planting in both field trials. On average, we observed positive relationships between RCD before planting and 44 months later, between H before planting and 44 months after planting, and between both H and HRCD before planting and SR 44 months after planting, but not between RCD and SR (Figs. 2–3).

The low width of the CI_95% of the study results, especially in correlations, shows that the trial fields had environmental conditions quite homogeneous (Table S3; Tables 2–4).

Discussion

In accordance with our hypothesis, the results indicate that the seedling root collar diameter (RCD) and height (H) of 15-month-old seedlings of the pure species Pinus arizonica, P. durangensis, P. engelmannii, P. leiophylla and P. teocote, as well as their hybrids, in several cases had a significant and positive (Fig. 1, Table S3), although weak effect on the same respective dimensions and on the survival rate (SR) 44 months after planting (explained between 8% and 18% of the variance in the plants’ corresponding dimensions and explained between 12% and 15% of the survival rate based on Random Forests models; Tables 2–4, Figs. 2–3). In both field trials, consequently, the advantages of large over small seedlings were observed, similar to Garau et al. (2009), Landis & Dumroese (2010) and Grossnickle & MacDonald (2018).

Two important factors to explain this low association between the pre-planting pine seedling dimensions and the corresponding dimensions and survival rates after 44 months in the trial fields could be the strong damage observed caused by pocket gopher, Thomomys spp. and the typically low winter-spring precipitation in both field trials, which was particularly low in 2019 and 2021 (<100 mm in each year) (CONAGUA, 2024). In practice, larger pine seedlings may be more susceptible to damage from the burrowing activities of Thomomys spp., as their root systems could be disrupted by tunnels (Knight, 2000), and they could also be more attractive to gophers due to their size and higher nutrient content (Hunt, 1992). Besides, larger seedlings have a larger transpiration demand, which may have a negative impact on their growth and survival during periods of drought or in particularly dry locations. Therefore, larger seedlings may suffer more from water stress compared to smaller seedlings (Oliet et al., 2019).

As in our case, Li et al. (2011) in a 3-year study and Tsakaldimi, Ganatsas & Jacobs (2013) in a 2-year study also found positive, but stronger correlations of the same variables, probably caused by their reported active intervention in the trail fields (Li et al., 2011) and by the shorter study time (Tsakaldimi, Ganatsas & Jacobs, 2013). Dey & Parker (1997) described similar results in red oak (Quercus rubra L.) seedlings underplanted in a central Ontario shelterwood. Li et al. (2011) reported similar findings for Larix olgensis seedlings. Alike to the observations of Li et al. (2011) and consistent with our assumption, we hardly find correlation between robustness index (HRCD) before planting and the plants dimensions 44 months after planting, with the only exception of a weak negative, although significant correlation with the seedlings diameter (Fig. 1, Table S3). Prieto-Ruíz et al. (2018) studied the seedlings behavior of Mexican Pinus cooperi 13 months after planting, comparing seedlings with two initial morphological conditions (RCD 3.5 mm and H 10–14 cm vs. RCD 4 mm and H 15–25 cm), and found that the seedling size did not influence survival, but higher values of RCD and H favored growth.

In reforestation practices, many traits and drivers have been identified and proposed as important factors for the near-optimal state of forest plantations (Prieto-Ruíz et al., 2016; Grossnickle & MacDonald, 2018). Nevertheless, the results regarding the assessment of the predictors evaluated in this study suggest that other variables, such as seed provenance (Tables S1–S2 and Tables 2–4), were more important than the initial seedling dimensions for the survival and plants’ size after 44 months in the field. According to our results regarding the width of CI_95%, the variation in microhabitat conditions slightly influenced the magnitude of variable importance, but not the ranking of importance (Table S3, Tables 2–4).

Given that the average seedlings per each seed provenance were grown under similar conditions of temperature and humidity, both in the greenhouse and after planting, any initial differences in seedlings size and also 44 months later in the field trials, as well as differences in survival rates can be partly attributed to genetics and maternal effects. The influence of two genetic variables (“seed provenance” and “tree species and hybrids”) was taken into account in this study, Tables S1–S2 and 2–4).

Thus, the weak correlation between nursery seedlings size before and after 44 months of planting could also be due to adaptation factors. Since, considering the climatic conditions of each of the 23 seed provenances, some seedling genotypes may have grown better under the tropical humid, nutrient-rich greenhouse conditions of the nursery (Hernández-Velasco et al., 2021) than if they had grown in the natural colder and drier environment where the respective mother tree lives. However, after the nursery stage, when the same seedlings were re-transplanted to the colder, drier and nutrient-poorer environment of the trial fields (Sánchez-Hernández et al., 2022), the mortality rate of large seedlings may have been greater than that in shorter seedlings. Large seedlings could be less genetically adapted to the planting site by epigenetic memory effects (Gömöry et al., 2017). The highly artificial nursery conditions could also have altered genotype selection (e.g., homozygote excess) in the seedlings (Gillet, Ziehe & Gregorius, 2016) which may have selectively influenced the survival of individual genotypes after planting (Campbell & Sorensen, 1984).

The comparison of the initial development of the RCD and H of our test seedlings with the current Mexican standard NMX-AA-170-SCFI-2016 (Secretaría de Economía, 2016) shows that the RCD almost always exceeded the standard minimum sizes after a period indicated for each species in nursery conditions. However, the majority of the seedlings in our study did not reach the minimum standard’s H in such period. This also means that most of these seedlings initially had a better (higher) robustness index (HRCD) than the aforementioned mentioned standard suggests.

Our results also show that the Random Forests models (RF) detected a higher unbiased conditional R²c than the generalized linear models (GLM) used. RF overfitting was unlikely, due to the very small confidence intervals found with such models (Tables 2–4). These differences may be attributed to a larger robustness, flexibility and to the ability of RF to capture complex patterns in the data (Breiman, 2001). On the other hand, the flexibility of GLM is limited because these models are usually linear with respect to the predictors, even if they can model nonlinear relationships through appropriate link functions and polynomial or interactive terms (Hardin & Hilbe, 2007).

The large difference found between the squared Spearman correlation coefficient (r_s²) of individual variables, biased marginal R²m and the unbiased conditional R²c of classification and regression models (Table S3 vs. Tables 2–4), which no longer included the influence of the other specific predictors variables (also referred to here as “confounding variables”), highlights the importance of calculating R²c in order to mitigate the effects of the studied confounding variables. However, R²c may change if more confounding variables are added to the study.

Conclusion

In this study, the effects of initial morphological attributes such as seedling diameter, height, and robustness index were significant, although in general not good predictors of growth and survival 44 months after planting those seedlings in field trials. Moreover, the genetic variables “seed provenance” and “species affiliation” were just as important as the seedling dimension for growth and survival. Consequently, the adequate selection of seed origins and tree species and their hybrids might be an important factor for the success of reforestation in the Sierra Madre Occidental.

Since the experiment was conducted using only regional provenances in just two trial fields, the representativeness of the results is limited. However, our findings offer preliminary insights that could help refine standards for plant nurseries seeking to enhance survival and growth rates in reforestations. Specifically, our results suggest that the Mexican standard NMX-AA-170-SCFI-2016 should be revised to incorporate genetic factors and consider not only seedlings dimensions but also the climatic conditions of the seed provenance and the site of final plantation. Further research is needed to address these gaps in knowledge.

Supplemental Information