Introduction

Forest restoration is intended to restore tree cover in ecosystems degraded by human activities in order to recuperate, at least partially, the environmental goods and services that they originally provided (Mansourian 2005). Although forest restoration prioritizes the use of native plants that originally inhabited the target areas (Hall et al. 2011; Kettle 2012), the success of restoration practices depends on whether the early life cycle stages of species can overcome the environmental barriers that degraded habitats impose to their establishment (Rey-Benayas 2005). This has led to propose that, before planting trees, experimental studies are required to determine what local species might recruit in degraded areas (Cabin 2007; Howe and Martínez-Garza 2014).

Moderate light environments and adequate water availability promote the germination and establishment of tree species in forest understories (Rousset and Lepart 2000; Li and Ma 2003; Corrià-Ainslie et al. 2015), while excessive solar radiation and water shortages commonly prevent these processes on degraded sites (Ren et al. 2008; González-Salvatierra et al. 2013). Tree recruitment on disturbed areas can also be constrained by interspecific competition if r-strategist plants (species with widespread dispersal, fast establishment and growth, and early reproduction) that colonize these habitats outcompete tree seedlings (Davis et al. 1999; Holl et al. 2000; February et al. 2013). Thus, successfully establishing native species on degraded areas poses a number of challenges to forest restoration practitioners.

When abiotic factors prevent tree recruitment in open habitats, pioneer shrubs that ameliorate extreme environmental conditions can be used as nurse plants to facilitate their germination and establishment (Gómez-Aparicio et al. 2004; Padilla and Pugnaire 2006; Ren et al. 2008; Badano et al. 2009; Gómez-Ruiz et al. 2013). If nurse plants are not available, artificial shade structures can be used to emulate their positive effects on tree seedlings (Badano et al. 2011; Padilla et al. 2011). However, because r-strategist plants also benefit from shading, this might confer them competitive advantages on tree seedlings (Rey-Benayas et al. 2005).

As recruitment of tree species may differ in response to these environmental factors, determining which of them must be used to restore forests is critical for the success of these actions (Guariguata 2005; Clewell and Aronson 2013). In this context, the usefulness of a given species is determined by the tolerance of its early life cycle stages to physical stressors and/or to interspecific competition (Holl et al. 2000; Gómez-Aparicio 2009). However, as this information is not usually available (Löf et al. 2019), experiments evaluating how interacting abiotic and biotic factors affect seedling processes can contribute to select those species better adapted to the environmental conditions that predominate in degraded areas.

In this study, common-garden experiments were conducted to assess the effects of interacting biotic and abiotic factors on tree recruitment. This can contribute to make decisions on which species can be used for forest restoration and how environmental factors might be manipulated to increase their recruitment. For this, seedling emergence and survival of pioneer species native to seasonally dry forests of northwest Argentina were assessed on varying levels of sunlight, rainfall and interspecific competition. It was hypothesized that those species with higher tolerance to elevated levels of sunlight, water shortages and interspecific competition would be good candidates for forest restoration. Alternatively, if some species display low seedling emergence and survival under any of these conditions or their combinations, the experiments would indicate how they might be manipulated to mitigate their effects and increase the success of forest restoration.

Materials and methods

Target species

In northwest Argentina, three pioneer tree species colonize forest gaps that result from natural disturbances, including Anadenanthera colubrina (Vell.) Brenan, Ceiba chodatii (Hassl.) Ravena and Jacaranda mimosifolia (D. Don) (Grau et al. 1997; Varela and Albornoz 2013; Oyarzabal et al. 2018). Nevertheless, native forests of this region have been intensely fragmented by human caused fires over the last 50 years (Manrique et al. 2018), and the recruitment of these species in burnt areas is uncommon (Tálamo et al. 2016). Instead, vegetation on burnt sites is dominated by Wissadula tucumanensis R. F. Fr., a native r-strategist shrub that quickly colonizes disturbed habitats (Krapovickas and Tolaba 2012). According to our field measurements, which included ten transects at three burnt sites, the average cover of W. tucumanensis is 80% (± 18% SE).

All species described above bloom in autumn/winter and their seeds are fully developed by late spring. For the experiments, in spring 2017, ripe fruits were collected from 30 trees of each species within local forest remnants. As all species produce large seeds (> 0.8 cm length), fruits pericarps and mesocarps were removed and seeds were visually inspected to discard those infected by insect larvae or fungi. This resulted in more than 2000 potentially viable seeds of each species. Ripe fruits of W. tucumanensis were also collected from 100 individuals in burnt areas. As this species produces capsules with 30–50 small rounded seeds (~ 1 mm diameter), seeds were released and pooled. After that, a batch of 200 seeds was weighed and the mean biomass per seed was used to estimate the total number of collected seeds. Mean seed biomass was 4.2 mg and, as we collected 150 g of seeds, about 30 thousand W. tucumanensis seeds were available for the experiments. Seeds of this species have no dormancy and thus they were preserved in paper bags at room temperature (20 °C) until their use in the experiments.

Experimental design

Common-garden experiments were conducted at the tree nursery of the National University of Salta, Argentina (24°43′42" S, 65°24′21" W, 1,200 m a.s.l.). This is an open site without vegetation with the same climate as most burnt areas of the region. Temperatures average 21 °C in summer and 11 °C in winter, while annual precipitation is between 600 and 900 mm (Bianchi and Cravero 2010). Rainfall during the period in which the experiments were conducted (January-November 2018) was 893 mm, which indicates that it was an above average year for rain. Wooden boxes (50 cm long × 30 cm wide × 30 cm deep) were filled with 10-cm surface soil from burnt sites as this is the substrate in which seeds germinate. Soil was collected in spring 2017 and sieved (2.5 mm mesh) to remove small rocks and other seeds that could potentially interfere with the experiments. The boxes were internally lined with plastic mesh to prevent the loss of soil due to drainage (Fig. 1). A total of 120 boxes were prepared in November 2017, coinciding with beginning of the rainy season, and distributed in three rectangular arrangements of 8 × 5 units spaced 2 m apart. These arrangements were randomly distributed among the three species (A. colubrina, C. chodatii, and J. mimosifolia) to conduct a common-garden experiment with each of them.

Fig. 1
figure 1

Experimental unit treated with black Raschel mesh (80% shade) and U-shaped channels of polycarbonate (30% rainfall reduction). The other combinations of light and rainfall levels resulted from excluding one of these structures or both of them

Full size image

Three factors in the experiments were evaluated, including light, rainfall and interspecific competition with two levels each. The two light levels were set out exposing the experimental boxes to full sunlight and to 80% shade to mimic light conditions of burnt areas and forest understories, respectively (hereafter, full sunlight = FL and shade = SH). Shading was randomly applied on half of the boxes by covering them with a single 80%-density layer of black Raschel mesh supported on wooden poles and lines of galvanized wire (Fig. 1). The two rainfall levels were then established by exposing the boxes to full rainfall and 70% rainfall to contrast the effects of rainy and dry years, respectively (hereafter, full rainfall = FR and reduced rainfall = RR). To reduce rainfall, the half of the boxes with and without shade were randomly selected and two U-shaped channels of transparent polycarbonate (1.5 mm thick, 10 cm wide) were fixed above them, spaced 10 cm apart with a decline of 10º between sides (Fig. 1). Finally, the two levels of interspecific competition implied including or excluding W. tucumanensis from the boxes (hereafter, without W. tucumanensis = NC and with W. tucumanensis = WC). For this, the half of the boxes with/without shade and with/without rainfall reduction were randomly selected and 1.9 g of W. tucumanensis seeds (~ 440–450 seeds) were scattered on the substrate 12 November, 2017. Seedlings of this species emerged after the first rainfall (22 November) and were allowed to grow under the light and rainfall conditions of boxes until 10 January 2018 (50 days). By this date, W. tucumanensis seedlings were 20–30 cm tall and part of them were removed to fit their cover to 80% in the experimental boxes. Therefore, the experiments consisted of eight treatments with five replicates each: full sunlight-full rainfall-no competition ( FS-FR-NC), full sunlight-full rainfall-with competition (FS-FR-WC), full sunlight-reduced rainfall-no competition ( FS-RR-NC), full sunlight-reduced rainfall-with competition (FS-RR-WC), shade-full rainfall-no competition (hereafter, SH-FR-NC), shade-full rainfall-with competition (SH-FR-WC), shade-reduced rainfall-no competition (hereafter, SH-RR-NC), and shade-reduced rainfall-with competition (SH-RR-WC).

On 23 January 2018, 25 seeds per box were sown in the experimental arrangement assigned to each tree species. Seeds were sown at 5-cm depth and positions were marked with small aluminum stakes to facilitate monitoring emergence and survival of seedlings. These variables were assessed every seven days until the beginning of the next rainfall season (11 November 2019). Emergence of hypocotyls above the substrate was used as indicator of seedling emergence, while survival was assessed monitoring alive individuals across time. In this later case, it was assumed that fully withered seedlings that did not resprout on further monitoring dates were dead.

Statistical analyses

For each tree species, failure-time-analyses were used to compare seedling emergence and survival rates between levels of each experimental factor. For this, a zero-value was assigned to all seeds at the beginning of the experiments, which were turned into one-values on the monitoring date in which the hypocotyl emerged (failure = hypocotyl emergence). Seeds that did not show the hypocotyl retained the zero-value until the end of the experiment. These binary data were used to compute standardized seedling emergence rates with the Kaplan–Meier’s method and were compared between levels of each factor with Gehan-Wilcoxon tests for two samples (Kleinbaum and Klein 2005). The same statistical procedures were used to compare survival rates between levels of each experimental factor but, in this case, one-values were assigned to emerged seedlings and turned into zero-values on their death dates (failure = seedling death). Seedings that survived until the end of the experiments retained the one-values until the last monitoring date.

After analyzing the effects of the main experimental factors, similar analyses were used to compare emergence and survival rates among simple combinations of factor levels (i.e., double factor interactions) and among the treatments that resulted from combining levels of the three factors (i.e., triple factor interactions). These comparisons were conducted with Gehan-Wilcoxon tests for multiple samples and, when differences were found, Gehan-Wilcoxon tests for two samples were used to perform pairwise comparisons (Kleinbaum and Klein 2005). All statistical analyses were run in R 4.0.2 (R Development Core Team 2020) and the datasets supporting them are available in the Zenodo repository (https://doi.org/10.5281/zenodo.4621291).

Results

Seedling emergence rates

Seedling emergence in all tree species increased during the first 50 days (Fig. 2). Comparisons between levels of the main experimental factors indicated that emergence rates of C. chodatii were higher at shade (SH) than at full sunlight (FS) (c2 = 4.512, df = 1, p < 0.001), while no differences between light levels were found for A. colubrina (c2 = 1.253, df = 1, p = 0.210) and J. mimosifolia (c2 = 1.135, df = 1, p = 0.256). These analyses also indicated that emergence rates of A. colubrina were higher with reduced rainfall (RR) than with full rainfall (FR) (c2 = 3.964, df = 1, p < 0.001), but the opposite occurred when this variable was compared between rainfall levels for C. chodatii (c2 = 2.311, df = 1, p = 0.021). No differences between rainfall levels were found on emergence rates of J. mimosifolia (c2 = 0.537, df = 1, p = 0.591). There were also no effects of the presence/absence of the competitor on the emergence rates of A. colubrina (c2 = 1.122, df = 1, p = 0.262) and C. chodatii (c2 = 1.601, df = 1, p = 0.109), but J. mimosifolia seedlings emerged faster and in higher numbers in absence of the competitor (NC) than in its presence (WC) (c2 = 2.980, df = 1, p = 0.003).

Fig. 2
figure 2

Seedling emergence rates (± 95% confidence intervals) of Anadenanthera colubrina, Ceiba chodatii and Jacaranda mimosifolia estimated for the levels of each experimental factor. The left column shows the effects of light intensity (FS = full sunlight, SH = 80% shade), the center column the effects of rainfall (FR = full rainfall, RR = 30% reduced rainfall) and the right column the effects of interspecific competition (NC = no competition, WC = with competition)

Full size image

Emergence rates differed among all pairwise combinations of the experimental factors for the three species (p < 0.05 in all Gehan-Wilcoxon tests for multiple samples). The analyses of the combined effects of light and rainfall indicated that emergence rates of C. chodatii were higher in SH-FR than in other combinations of these two factors, while the opposite occurred with A. colubrina (Fig. 3). No differences in emergence rates of J. mimosifolia were found between FS-FR and SH-RR, which had higher emergence rates than other combinations of light and rainfall (Fig. 3). The combined effects of light and interspecific competition indicated that emergence rates of A. colubrina were higher in FS-NC than in other between-level combinations of these two factors, while emergence rates of C. chodatii and J. mimosifolia were higher in SH-NC (Fig. 3). The combined effects of rainfall and interspecific competition indicated that emergence rates of A. colubrina were higher in boxes with reduced rainfall (RR-NC and RR-WC) than in boxes exposed to full rainfall (FR-NC and FR-WC), regardless of the competition level (Fig. 3). No differences in emergence rates were found between FR-WC and FR-NC for C. chodatii, but these values were higher than in RR-WC (Fig. 3). Emergence rates of J. mimosifolia were higher in boxes where the competitor was absent (FR-NC and RR-NC) than in boxes with the competitor (FR-WC and RR-WC), regardless of the rainfall level (Fig. 3).

Fig. 3
figure 3

Seedling emergence rates (± 95% confidence intervals) of Anadenanthera colubrina, Ceiba chodatii and Jacaranda mimosifolia estimated for the between-level combinations of experimental factors. The left column shows the combined effects of light intensity and rainfall, the center column the combined effects of light intensity and interspecific competition, and the right column the combined effects of rainfall and interspecific competition. Codes in panels are FS = full sunlight, SH = 80% shade, FR = full rainfall, RR = 30% reduced rainfall, NC = no competition and WC = with competition. Different letters on the side of codes indicate significant differences in seedling emergence rates (Gehan-Wilcoxon critical α for pairwise comparisons = 0.05)

Full size image

Emergence rates of the tree species differed among treatments that resulted from combining levels of the three experimental variables (p < 0.01 in all Gehan-Wilcoxon tests for multiple samples). Pairwise contrasts of treatments indicated that emergence rates of A. colubrina were higher in full sunlight in absence of the competitor, whatever the rainfall level (FS-FR-NC and FS-RR-NC), and in shade with reduced rainfall, regardless of competition (SH-RR-NC and SH-RR-WC) (Fig. 4). Seedlings of C. chodatii emerged faster and in higher numbers in shaded boxes exposed to full rainfall, irrespective of the competition level (SH-FR-NC and SH-FR-WC), and in shaded boxes with reduced rainfall and without the competitor (SH-RR-NC) (Fig. 4). Emergence rates of J. mimosifolia were higher in shade with reduced rainfall without competition (SH-RR-NC) (Fig. 4).

Fig. 4
figure 4

Seedling emergence rates (± 95% confidence intervals) of Anadenanthera colubrina, Ceiba chodatii and Jacaranda mimosifolia estimated for the triple-level combination of experimental factors. Codes in panels are FS = full sunlight, SH = 80% shade, FR = full rainfall, RR = 30% reduced rainfall, NC = no competition and WC = with competition. Different letters on the side of code combinations indicate significant differences in seedling emergence rates (Gehan-Wilcoxon critical α for pairwise comparisons = 0.05)

Full size image

Seedling survival rates

Seedling survival decreased with time in the three target species (Fig. 5). Comparisons between the main experimental factors indicate that survival rates of A. colubrina were higher in full sunlight (FS) than in shade (SH) (c2 = 4.802, df = 1, p < 0.001), while the opposite was found for C. chodatii (c2 = 5.910, df = 1, p < 0.001) and J. mimosifolia (c2 = 6.914, df = 1, p < 0.001). No effects of rainfall were found on survival rates of A. colubrina (c2 = 0.407, df = 1, p = 0.684) and C. chodatii (c2 = 0.550, df = 1, p = 0.582), but J. mimosifolia had higher survival rates with full rainfall (FR) than with reduced rainfall (RR) (c2 = 11.915, df = 1, p < 0.001). Survival rates of the three species were lower with competition (WC) than without competition (NC) (A. colubrina: c2 = 12.514, df = 1, p < 0.001; C. chodatii: c2 = 9.813, df = 1, p < 0.001; J. mimosifolia c2 = 11.915, df = 1, p < 0.001).

Fig. 5
figure 5

Seedling survival rates (± 95% confidence intervals) of Anadenanthera colubrina, Ceiba chodatii and Jacaranda mimosifolia estimated for each experimental factor. The left column shows the effects of light intensity (FS = full sunlight, SH = 80% shade), the center column the effects of rainfall (FR = full rainfall, RR = 30% reduced rainfall) and the right column the effects of interspecific competition (NC = no competition, WC = with competition)

Full size image

All between-level combinations of experimental factors had significant effects on survival rates of tree species (p < 0.05 in Gehan-Wilcoxon tests for multiple samples). Pairwise comparisons indicated that survival rates of A. colubrina were higher in full sunlight (FS-FR and FS-RR) than in shade (SH-FR and SH-RR) regardless of rainfall levels, while C. chodatii and J. mimosifolia had the opposite response (Fig. 6). Between-level combinations of light and interspecific competition indicated that survival rates of A. colubrina were higher in FS-NC than in other combinations of these experimental factors, while survival rates of C. chodatii and J. mimosifolia were higher in SH-NC (Fig. 6). The combined effects of rainfall and interspecific competition indicated that survival rates of all tree species were higher in absence of the competitor regardless of the level of rainfall (Fig. 6).

Fig. 6
figure 6

Seedling survival rates (± 95% confidence intervals) of Anadenanthera colubrina, Ceiba chodatii and Jacaranda mimosifolia estimated for the between-level combinations of experimental factors. The left column shows the combined effects of light intensity and rainfall, the center column the combined effects of light intensity and interspecific competition, and the right column the combined effects of rainfall and interspecific competition. Codes in panels are FS = full sunlight, SH = 80% shade, FR = full rainfall, RR = 30% reduced rainfall, NC = no competition and WC = with competition. Different letters on the side of codes indicate significant differences in seedling survival rates (Gehan-Wilcoxon critical α for pairwise comparisons = 0.05)

Full size image

Seedling survival of the target species differed among treatments as a result of combining levels of the experimental factors (p < 0.01 in all Gehan-Wilcoxon tests for multiple samples). For A. colubrina, survival rates were higher in full sunlight without the competition, regardless of level of rainfall (FS-FR-NC and FS-RR-NC) (Fig. 7). Survival rates J. mimosifolia were higher in shaded boxes exposed to full rainfall without competition (SH-FR-NC) (Fig. 7), and survival rates of C. chodatii were higher in shaded boxes without competition regardless the rainfall level (SH-FR-NC and SH-RR-NC) (Fig. 7).

Fig. 7
figure 7

Seedling survival rates (± 95% confidence intervals) of Anadenanthera colubrina, Ceiba chodatii and Jacaranda mimosifolia estimated for the triple-level combination of experimental factors. Codes in panels are FS = full sunlight, SH = 80% shade, FR = full rainfall, RR = 30% reduced rainfall, NC = no competition and WC = with competition. Different letters on the side of codes indicate significant differences in seedling survival rates (Gehan-Wilcoxon critical α for pairwise comparisons = 0.05)

Full size image

Discussion

This study focused on analyzing how the recruitment of different tree species may be influenced by abiotic and biotic factors for the restoration of disturbed sites. It is recognized that these are ex situ experiments because they were performed at the University campus instead of field sites, and this might make the results not as scientifically rigorous as desired. Nevertheless, this lowers experimentation costs because it reduces workforce hours, and the economic resources required for establishing and monitoring the experiments, as compared with the costs of performing field experiments. Further, it is important to note that our results are applicable to direct seeding restoration practices, which also constitute a less expensive technique than using seedling transplants (Florentine and Westbrooke 2004; Ceccon et al. 2016). Therefore, as forest restoration programs are often constrained by budgets, the methodology used in this study could increase cost–benefit ratios.

Our results suggest that the early life cycle stages of the three tree species (A. colubrina, C. chodatii and J. mimosifolia) differentially respond to varying light intensity, rainfall and interspecific competition, also indicating that each species tolerates particular combinations of these factors. However, although these species are considered pioneer trees in seasonally dry forests of South America (Varela and Albornoz 2013; Oyarzabal et al. 2018), none of them tolerate elevated levels of light, water shortages and interspecific competition together. This implies that using these species for reforestation and restoration programs requires controlling the environmental factors that more critically constrain their recruitment in burnt areas.

The experiment with A. colubrina indicated that the emergence of seedlings was promoted under reduced rainfall conditions, while seedling survival was higher in full sunlight. This suggests that A. colubrina seedlings tolerate water shortage conditions under elevated solar radiation and these are desirable features of tree species for restoring degraded areas (Khurana and Singh 2001; Vallejo et al. 2012). Similar recruitment patterns were reported for this species in seasonally dry forests from central Bolivia, where seedling survival was greater in open areas with elevated sunlight and low soil moisture compared with survival rates recorded in forest understories (Fredericksen et al. 2000). Thus, our results suggest that A. colubrina can be used to restore forests in burnt areas of northwest Argentina, as most plants in extensively deforested sites are commonly exposed to reduced water availability and elevated solar radiation (Rey-Benayas et al. 2005; Badano et al. 2015; González-Salvatierra et al. 2013).

Seedling emergence and survival rates of C. chodatii suggest that this species does not tolerate elevated sunlight during recruitment. Further, because emergence rates in shaded boxes were higher under full rainfall, it appears that this species also requires elevated levels of soil moisture to germinate. In shaded boxes survival rates were similar under both rainfall levels (FR and RR), suggesting that C. chodatii seedlings can tolerate drought and shade. These are common life traits of tree species that establish in forest understories and endure in these habitats until disturbances generate canopy gaps (Givnish 1988; Valladares et al. 2016). Considering these results, it is suggested that using C. chodatii for restoring burnt areas requires sowing seeds beneath nurse plants or artificial shade structures to promote the emergence and survival of seedlings, while rainfall would not constrain these processes.

Seedling emergence rates of J. mimosifolia were similar between boxes exposed to full sunlight and shaded boxes, as well as between boxes exposed to full and those exposed to reduced rainfall. This species appears to have no exacting requirements for germination and this concurs with studies conducted in northern Argentina and Brazil reporting that J. mimosifolia seeds quickly germinate under a wide variety of environmental conditions (Colombo-Speroni and De Viana 2000; Socolowski and Takaki 2004). Nevertheless, the survival of J. mimosifolia seedlings was strongly reduced when exposed to elevated sunlight and reduced rainfall. This suggests that using J. mimosifolia to restore degraded forests would require sowing large numbers of seeds beneath nurse plants or artificial shade structures and regularly watering the seedlings. Similar results have been reported for oaks in central Mexico where their establishment in deforested areas required planting seedlings beneath nurse plants and providing a reliable water supply during the dry season (Badano et al. 2009).

Although the emergence rates of seedlings of the target species were not significantly affected by the presence of W. tucumanensis, survival rates decreased when they were grown with this r-strategist competitor. As this was found in all light and rainfall levels, it suggests that young seedlings of A. colubrina, C. chodatii and J. mimosifolia are poor interspecific competitors with W. tucumanensis in burnt areas. Similar effects have been reported in other deforested areas where herbaceous plants that early colonize open sites outcompete tree seedlings and reduce the success of forest restoration (Holl et al. 2000; Rey-Benayas et al. 2005; February et al. 2013; Weidlich et al. 2020). In this study, these results suggest that forest restoration requires controlling the population sizes of potential competitors in burnt areas, at least until tree seedlings are fully established.

Conclusions

This is not an exhaustive study of factors impeding the recruitment of tree species on disturbed areas but it shows that sunlight, water availability and interspecific competition with pioneer plants are important drivers that determine the success of forest restoration. Therefore, studies to determine which tree species can recruit under the environmental conditions that prevail on these sites becomes an important component of restoration. It is recognized that some restoration practitioners may perceive experimentation as a time-consuming process that delays the “let’s plant trees” phase, but they should analyze the feasibility of including it in their project schedules, as this can lead to more effective restoration actions.