Introduction
Tropical dry forests (TDF) are lowland ecosystems characterised by a marked rain seasonality and alternating rainy and drought periods of 4–6 months (< 100 mm mo−1), with an annual precipitation of 250–2000 mm (Dirzo et al. Reference Dirzo, Young, Mooney and Ceballos2011, Sánchez et al. Reference Sánchez, Quesada, Rodriguez, Nassar, Stoner, Castillo, Garvin, Zent, Calvo, Kalacska, Fajardo, Gamon and Cuevas2005). Given their climatic and soil quality characteristics, TDF have been the centre of human settlement across the Americas (Calvo et al. Reference Calvo, Sánchez, Duran and Espírito-Santo2016, Sánchez et al. Reference Sánchez, Quesada, Rodriguez, Nassar, Stoner, Castillo, Garvin, Zent, Calvo, Kalacska, Fajardo, Gamon and Cuevas2005). As a consequence, the current extent of TDF in main Latin American and Caribbean countries is less than 10% of their original distribution (Calvo et al. Reference Calvo, Sánchez, Duran and Espírito-Santo2016, DRYFLOR 2016), and most of their remaining area is severely fragmented and embedded in transformed lands, mostly by agriculture and cattle ranching (Portillo & Sánchez Reference Portillo and Sánchez2010). Although dry forests are among the most threatened tropical ecosystems in the world (Dirzo et al. Reference Dirzo, Young, Mooney and Ceballos2011), only 4% of their current extent is protected, therefore being hugely under-represented in conservation areas (Calvo et al. Reference Calvo, Sánchez, Duran and Espírito-Santo2016, Pennington et al. Reference Pennington, Lehmann and Rowland2018). Thus, many authors agree that in addition to preserving the remaining TDF, there is an urgent need of restoring dry forest across borders (DRYFLOR 2016, González-M et al. 2018, Khurana & Singh Reference Khurana and Singh2001, Portillo-Quintero et al. Reference Portillo-Quintero, Sánchez-Azofeifa, Calvo-Alvarado, Quesada and Espirito Santo2015, Quesada et al. Reference Quesada, Sánchez-Azofeifa, Alvarez-Añorve, Stoner, Alvila-Cabadilla, Calvo-Alvarado, Castillo, Espirito Santo, Fagundes, Fernandes, Gaom, Lopezaraiza-Mikel, Lawrence, Cerdeira-Morellato, Powers, de S Neves, Rosas-Guerrero, Sayago and Sanchez-Montoya2009).
Examples of studies reporting on neotropical TDF restoration have highlighted the lack of information about these ecosystems (Brooks & Jordan Reference Brooks and Jordan2014, Burney & Burney Reference Burney and Burney2016, Fajardo et al. Reference Fajardo, Rodríguez, González and Briceño-Linares2013, Fundación Natura Colombia 2014, González-Tokman et al. Reference González-Tokman, Barradas, Boege, Domínguez, Del-Val, Saucedo and Martínez-Garza2018, Lanuza et al. Reference Lanuza, Espelta, Peñuelas and Peguero2020, Werden et al. Reference Werden, Alvarado, Zarges, Calderón, Schilling, Gutiérrez and Powers2018). Therefore, scientific efforts to understand ecological processes in TDF are needed to provide a baseline that improves management and restoration plans for these endangered ecosystems (Stoner & Sánchez Reference Stoner and Sánchez2009). This baseline should focus on the current abiotic and biotic pressures that TDF restoration programmes are facing. For example, the response of plants to an increase in extreme drought events and the effects of herbivory/defoliation by feral or domestic ungulates and insects, which in active restoration TDF programs, are the second cause of seed and seedling mortality after desiccation (Dimson & Gillespie Reference Dimson and Gillespie2020, Quisehuatl-Medina et al. Reference Quisehuatl-Medina, Averett, Endress and Lopez2020). This shows the need to understand plant ecological strategies for overcoming these factors in order to know which species should be prioritised for use in restoration plans.
Plant communities in human-modified TDF are found in a mosaic of small forest patches surrounded by grasslands for cattle ranching or crops (Trejo & Dirzo Reference Trejo and Dirzo2000). Regeneration in this kind of landscapes has serious limitations. First, abiotic factors like drought may limit plant growth, leaf size, stem extension and root proliferation, since water-limited soils limit leaf water potential, which ultimately reduces survival rate (Farooq et al. Reference Farooq, Wahid, Kobayashi, Fujita and Basra2009, Poorter & Markesteijn Reference Poorter and Markesteijn2008). Additionally, climate change predictions of more extreme droughts and shifts in precipitation regimes (Allen et al. Reference Allen, Dupuy, Gei, Hulshof, Medvigy, Pizano, Salgado-Negret, Smith, Trierweiler, Van Bloem, Waring, Xu and Powers2017) make it difficult to predict how species will cope with climatic changing conditions. Second, biotic factors like herbivory affect plant survival and fitness, reducing plant growth in the long term (Marquis Reference Marquis1984) and limiting regeneration processes (Quisehuatl 2020). Furthermore, studies that assess which species are more suitable for restoration purposes (i.e. species than can establish and succeed under the limiting conditions) are critical to improve TDF restoration efforts.
Ecological restoration is often promoted through the acceleration of ecological succession by planting seedlings of desired key species (Dobson et al. Reference Dobson, Bradshaw and Baker1997). However, selecting key species for restoring TDF depends on information on their biology and ecology (Khurana & Singh Reference Khurana and Singh2001). Specifically, it is important to assess how plant functional traits can be useful to cope with TDF abiotic and biotic challenging factors at early successional stages, therefore selecting species that are more suitable for restoration plans. For instance, pioneer species are fast growing and light demanding, germinating in disturbed areas such as gaps (Alvarez & Martinez Reference Alvarez and Martinez1990, Huante & Rincón Reference Huante and Rincón1998). Given their physiological resistance to disturbance, these species have higher drought tolerance and may be good candidates for restoration plans in TDF (Vargas & Ramírez Reference Vargas, Ramírez, Pizano and García2014). However, little is known about TDF pioneer species performance under harsh abiotic and biotic conditions, such as water limitation and herbivory/predation, which are common in such ecosystems. Evaluating how these stress sources affect plant performance at early stages is key to assess which focal species would perform better for restoration purposes (Khurana & Singh Reference Khurana and Singh2001, Veenendaal et al. Reference Veenendaal, Swaine, Agyeman, Blay, Abebrese and Mullins1995).
In this study, we assessed performance (survival and growth), biomass allocation and water distribution patterns of four TDF pioneer species (Guazuma ulmifolia, Samanea saman, Pseudosamanea guachapele and Senna spectabilis) in response to abiotic (water limitation) and biotic (herbivory) stress factors commonly faced by plants in TDF restoration projects. Specifically, we addressed the following questions: (1) how does seedling performance (survival and growth) respond to water availability and herbivory? (2) how do different plant species allocate their resources (biomass and water) in response to these abiotic and biotic stress factors?
Methods
Study site
This study was conducted at Universidad Icesi in Cali, Colombia, South America, located in the Cauca river valley (3°20’29” N 76°31’38” W). The study site has TDF climatic characteristics with a mean annual temperature of 24°C and annual precipitation of 1352 mm (2000–2012) (Departamento Administrativo de Gestion del Medio Ambiente 2012). In this region, precipitation is concentrated in two rainy seasons (March–June and October–November) with a mean precipitation of ∼140–180 mm per month and two dry seasons (December–February and July–September) with a mean precipitation of ∼40–110 mm per month (Instituto de Hidrología, Meteorología y Estudios Ambientales 2014).
Seeds collection and germination
Seeds from four key species for TDF restoration (Guazuma ulmifolia Lam., and three species of legumes: Pseudosamanea guachepele (Kunth) Harms, Samanea saman (Jacq.) Merr. and Senna spectabilis (DC.) H.S. Irwin & Barneby) were collected at different dry forests of Colombia in April 2016 (Table 1). Seeds from P. guachepele, S. saman and G. ulmifolia were collected from at least three different adult trees per species in a dry forest in Armero-Guayabal, Tolima (5°04’06”N, 74°53’56”W), while seeds from S. spectabilis were collected in a dry forest from Valle del Cauca (3°21’09”N 76°31’44”W). Although originally, the idea was to include a larger number of species in the experiment, this was not possible due to historically low seed production in TDF of these two regions (probably to ENSO2015, González-M et al. Reference González, Posada, Carmona, Garzón, Salinas, Idárraga, Pizano, Avella, López, Norden, Nieto, Medina, Rodríguez, Franke-Ante, Torres, Jurado, Cuadros, Castaño, García and Salgado-Negret2020). The included species have the ability to stablish efficiently, are important in both the early and late stages of ecological succession and are widely used as nurse species, in addition to being successful in previous restoration projects (Torres-Rodríguez et al. 2019).
Table 1. Characteristics of the four pioneer plant species used in the experiment, including their seed mass after being dried for 3 days at 60 °C (mean ± sd; n for G. ulmifolia = 24; S. spectabilis = 29; P. guachapele = 69; and S. saman = 62) and resprout ability.
Two weeks after collection, seeds were sterilised using 0.6% sodium hypochlorite (NaClO) for 15 min, for reducing seed pathogens that may negatively impact seed germination. Once sterilised, seeds were washed with distilled water and seeded to germinate in a 2:3 mixture of sand and soil steamed sterilised at the International Centre for Tropical Agriculture (CIAT), and then re-sterilised at Universidad Icesi with a 1-hour autoclave cycle.
Experimental design
After 3 months of growth (July 2016), 32 seedlings of each species were transplanted into individual 1 L pots filled with a mixture of soil and rice husk (60% soil 40% sand) in an open-top greenhouse protected from direct sunlight by a mesh cover (35% sunlight); pots were randomised every other week to control for differences in microclimatic conditions. Eight individual plants from each species were randomly assigned to each of the following four treatments: control (C), irrigation (Ir), herbivory (H) and control × herbivory (C × H), for a total of 128 plants. All seedlings were watered two times a week for 1 month before treatments started.
Irrigation
Starting in August 2016, plants from Ir and H treatment were watered two times a week with 250 mL of water (the equivalent to weekly precipitation in a dry forest with 1600 mm annually, estimated based on the pot volume). In contrast, plants in the Control (C) treatment and C × H were not watered and received only rain (natural conditions). During the eight months of the experiment, precipitation was approximately ∼966 mm (Instituto de Hidrología, Meteorología y Estudios Ambientales 2017), slightly under the historical precipitation of 1109 mm (Figure 1) (Departamento Administrativo de Gestión del Medio Ambiente 2012). However, precipitation during the experiment was half the historic record in the driest months of the year (July, August, January and February; Figure 1).
Figure 1. Monthly precipitation at Universidad Icesi. Gray bars represent mean historical precipitation (2000–2012) (Departamento Administrativo de Gestiín del Medio Ambiente 2012), while black bars represent monthly precipitation during the experiment (July 2016–April 2017). Values below the red dashed line represent months of drought (< 100 mm).
Herbivory
To simulate the effect of extreme herbivory, all seedling leaves were manually removed in the H and C × H treatments every 2 months. Leaves were removed three times during the experiment (September 2016, December 2016 and February 2017).
Control × herbivory
Plants under this treatment received only rainwater and were manually defoliated every 2 months.
Data collection
Plant survival was measured as the number of seedlings that survived until the end of the experiment divided by the number of seedlings that were initially planted. Eight months after the treatments started (April 2017), all plants were harvested. Fresh weight (FW) and dry weight (DW) of leaves, stems, roots and root nodules (for P. guachapele and S. saman) were measured. FW was measured the same day the experiment was harvested, before plants were dried in the oven at 60°C for 3 days. Once dried, DW was measured in the lab.
Survival, final DW, root-shoot ratio and water allocation to root vs. shoot were estimated with the following equations:
$$Survival\; = {{\# \;of\;surviving\;seedlings} \over {\# \;of\;planted\;seedlings}} \times 100$$
$$Final\;dry\;weight\; = D{W_{Stem}} + D{W_{Leaf}} + D{W_{Root}}$$
$$Root - shoot\;ratio = {{D{W_{Root}}} \over {D{W_{Leaf}} + D{W_{Stem}}}}$$
$${Water\;allocation\;to\;root\;vs\;shoot\; = {{\left( {F{W_{Root}} - D{W_{Root}}} \right)} \over {\left( {F{W_{Stem}} - D{W_{Stem}}} \right) + \left( {F{W_{Leaf}} - D{W_{Leaf}}} \right)}}}$$
Statistical analyses
Survival
A two-way ANOVA was performed using arcsine transformation values to test for differences in survival across species and treatments (survival ∼ species + treatments).
Final DW
We conducted a three-way ANOVA with square root transformed values of final seedling DW to compare seedling biomass across species and treatments (irrigation and herbivory) (final DW ∼ species × irrigation × herbivory). We then performed a priori contrasts to compare species DW with and without irrigation in the two-way interaction of species and irrigation. These contrasts evaluated the difference in DW between the restricted water availability (C and C + H) treatments and the non-restricted water availability treatments (Ir and H); however, S. saman biomass under control treatment was not included in the analyses given that only one individual survived. To compare species growth across the three experimental factors, we ran a Tukey’s test in the three-way interaction of species, irrigation and herbivory.
We performed two models for the two nitrogen-fixing legumes species that form nodules (P. guachapele and S. saman); in the first model, nodule DW was used as a covariate in a GLM (final DW ∼ species × irrigation × herbivory + nodule DW), and in the second model, nodule DW was used as a response variable in an univariate linear model with C, Ir, H and C + H in a single factor (nodule DW ∼ treatments) to perform a multiple comparison Tukey’s test. Finally, we ran a correlation test and a linear regression using nodule DW as a predictor and seedling final DW as response variable for these two species (nodule DW ∼ final DW).
Root-shoot ratio
Seedling biomass allocation was compared across species and treatments using Johnson’s transformed (Jones Reference Jones2014) root-shoot values in a three-way ANOVA (R-S ratio ∼ species × irrigation × herbivory) followed by Tukey’s tests. Additionally, we performed a priori contrasts to analyse the two-way interactions testing for the effects of irrigation on the biomass allocation patterns of the species; however, S. saman biomass under C treatment was not included given that only one individual survived.
Water allocation to root vs. shoot
Values of water allocation to root vs. shoot were normalized using a Johnson’s tranformation and compared between treatments with a three-way ANOVA (water aloc. to root vs. shoot ∼ species × irrigation × herbivory) and Tukey’s tests.
All statistical analyses were performed within the R environment for statistical computing using the packages dplyr for data manipulation; gridExtra, ggplot and ggthemes to plots creation; and Johnson to conduct the Johnson’s transformations (R Core Team 2018) and JMP, ver. 15 for ANOVAs, a priori contrasts, regression analysis and Tukey’s tests.
Results
Performance
Survival
After 8 months, 104 seedlings out of the 128 planted survived. Most species had a survival greater than 60%, except S. spectabilis in the control treatment with 50% survival, and S. saman in the C treatment, for which only one individual survived (12.5%) (Figure 2). Additionally, the survival rate was impacted by experimental treatments (F3,9 = 5.29, P = 0.022) – specially by lack of irrigation (control treatment) – but did not vary across species (F3,9 = 1.19, P = 0.37; Fig. S1).
Figure 2. a) Seedling final dry biomass (and seedling survival percentage in the upper part of each box), b) root-shoot ratio and c) water allocation to root vs. shoot of four plant species of tropical dry forest (G. ulmifolia, S. spectabilis, P. guachapele and S. saman) as a response to four experimental treatments (C: control; Ir: irrigation; H: herbivory; C + H: control × herbivory). Values higher than one indicate investment in roots, while those below one suggests investment in shoot (panels b and c). Results of S. saman under the ND treatment are not shown because only one individual survived. Different letters over boxes indicate statistical differences across treatments. Letters in b) and c) were not depicted because three-way interactions were not statistically significant for root:shoot ratio (Table 2) and did not have enough degrees of freedom to be run for water allocation to root vs. shoot.
Final DW
Final DW varied among species, but most importantly, across Ir and H treatments (Table 2). In general, species had higher biomass when irrigated, and herbivory affected more plant biomass than the lack of a constant supply of water (F3,98 = 21.42, P < 0.001), but all species performed worst with the combination of no irrigation, and herbivory (Figure S2). Growth of G. ulmifolia and S. spectabilis was similar across treatments, but significantly decreased in the C × H treatment (F1,26 = 9.22, P < 0.001; F1,18: 8.41, P = 0.001, respectively; Figure 2), while P. guachapele showed a significantly lower biomass with both stress factors compared to control (F3,16 = 11.45, P < 0.001) and S. saman under H and C × H (F2,20 = 13.82, P < 0.001) treatments (Figure 2a). In contrast, the effects of drought varied with respect to other treatments, as, while the DW of S. spectabilis was not affected by this factor alone (F1,85 = 0.45, P = 0.50), that of G. ulmifolia and P. guachapele was clearly reduced (F1,85 = 14.85, P < 0.001; F1,85 = 27.80, P < 0.001, respectively). Similarly, herbivory also reduced seedling biomass across all species (G. ulmifolia: F1,85 = 27.45, P < 0.0010; S. spectabilis: F1,85 = 11.84, P < 0.0010; P. guachapele: F1,85 = 23.49, P < 0.0010 and S. saman: F1,85 = 40.66, P < 0.0010). Additionally, despite lack of irrigation (C) and herbivory having different effects on seedling biomass across species, the combination of these two factors significantly reduced growth of all species (Figure 2a).
Table 2. Results of the linear models for the dependence of seedling final dry weight (ANOVA), root-shoot ratio (ANOVA) and water allocation to root vs. shoot (ANOVA) (Response (Final dry weight, Root-shoot ratio, Water allocation) ∼ species × Ir × H) of four plant species (G. ulmifolia, S. spectabilis, P. guachapele and S. saman). Data of S. saman under C treatment were excluded from the analysis because only one individual survived.
Significant P-values are indicated by *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001.
For the two nitrogen-fixing legumes (P. guachapele and S. saman), nodule DW was an important factor determining total seedling DW (F1,42 = 7.10, P = 0.010), and there was a significant positive correlation between nodule DW and total plant DW (t = 6.5, DF = 48, P < 0.0001) (Figure 3a). Nodule DW was reduced by herbivory (H), water availability (C) and C × H (F1,45 = 3.61, P = 0.020, Figure 3b). Finally, much of the variation in seedling DW was explained by nodule DW in S. saman (R2 = 0.51, P = 0.001) and P. guachapele (R2 0.4, P < 0.0001).
Figure 3. a) Pearson’s correlation between nodule dry weight and seedling final dry weight for two N-fixing legume species (P. guachapele and S. saman). R2 represents the determination coefficient of the linear regression Seedling Final Dry Weight ∼ Nodule Dry Weight. Asterisks (*) indicate significant models (P < 0.05). b) Nodule dry weight of these same species as a response to four experimental treatments (C: control; Ir: irrigation; H: herbivory; C+H: control × herbivory). Different letters over the boxes indicate statistical differences between treatments.
Allocation patterns
Root-shoot ratio
Biomass allocation patterns also differed across species and treatments (Table 2). In general, experimental treatments had no effect on species allocation patterns (F3,98 = 2.38, P = 0.075) (Figure S2b). Furthermore, species showed a more or less homogeneous biomass allocation to roots and aboveground tissues, with root-shoot ratio values close to one across treatments (Figure 2). However, G. ulmifolia showed root-shoot values greater than one (Figure 2), indicating that this species stores more resources in roots than in aboveground tissues. In general, there was no effect of C treatment on root-shoot ratios (Table 2), although S. spectabilis allocated higher biomass to shoots when stressed by lack of a water supply (C) (F1,18 = 5.81, P < 0.001) (Figure 2). Interestingly, in the C+H treatment G. ulmifolia allocated 65% more biomass to roots compared to the Ir treatment (Figure 2b). In contrast, herbivory tended to increase biomass allocation to plant roots (Table 2), although having only a significant effect on S. spectabilis (F1,18 = 10.87, P = 0.004, Figure 2b). With regard to the Ir and H interaction, H was the treatment under which all species allocated more resources to roots (root-shoot > 1), followed by C+H, C and Ir (Tukey’s test: A, B, BC and C, respectively).
Water allocation to root vs. shoot
In general, experimental treatments changed the way plant species allocated water resources (F3,80 = 5.0, P = 0.003), with a higher investment in roots when irrigated, under herbivory and water limitation with herbivory. Surprisingly, under water limitation only, there was a similar water allocation to shoots and roots across species (Figure S2c). Moreover, species stored more water in their roots across all treatments, except S. spectabilis under Ir and C treatments, for which the amount of water stored in aerial tissues was greater than in the roots (Fig. 2c). Furthermore, seedlings exposed to herbivory allocated more water resources to roots than to shoots (Table 2), except for S. saman (Figure 2c).
Discussion
Assessing which plant species are the most useful for TDF restoration exercises should depend on information on how they cope with stressful abiotic and biotic conditions at early developmental stages (Khurana & Singh Reference Khurana and Singh2001, Veenendaal et al. Reference Veenendaal, Swaine, Agyeman, Blay, Abebrese and Mullins1995). This information should be focused on the current pressures that TDF active restoration programs are facing, such as high seedling mortality resulting from an increase in extreme drought events and the effects of herbivory (Dimson & Gillespie Reference Dimson and Gillespie2020). In this study, we compared the performance of four pioneer species broadly used in TDF restoration projects under two stresses: water limitation by natural drought events – (ENSO) and extreme herbivory/predation (complete defoliation). We found that G. ulmifolia performed the best and allocated the highest amount of biomass to its roots across treatments (Figure 2a). On the other hand, growth of two nitrogen-fixing legumes (P. guachapele and S. saman) was correlated with nodule DW, showing their dependence to nitrogen fixing for an optimal performance (Figure 3). Finally, all species allocated a high proportion of water to roots, suggesting that belowground water reserves may be key for their survival under abiotic and biotic stress.
Effects of water limitation and herbivory on seedling performance and allocation patterns
Water availability is the most important abiotic factor determining tree species richness, composition and distribution in TDF (Poorter & Markesteijn 2007). Thus, current and future intensification of drought events will probably have a great effect on TDF plant growth and survival (Cunze et al. Reference Cunze, Heydel and Tackenberg2013, Velazco et al. Reference Velazco, Villalobos, Galvão and De Marco Júnior2019), given the marked seasonality in rain regimens that these forests already experience. The seedling stage is generally considered to be the most important bottleneck for successful regeneration in dry areas, given the limited root system of small plants, and the fact that drought results in a reduction in leaf water potential and gas exchange, leading to a deceleration in growth and an increase in mortality (Poorter & Markesteijn 2007). Accordingly, we observed the lowest survival rate (overall average of 56%), and a decrease in growth for all species when they were exposed to water limitation (Figure 2a). This may be due to water loss through transpiration and xylem cavitation, the most important cause of plant mortality in dry ecosystems (Poorter & Markesteijn 2007). In particular, S. saman had only one surviving individual under this treatment. In contrast, we found that G. ulmifolia had the highest survival among the four species under this treatment (87.5%) (Figure 2a), and correspondingly, the highest biomass investment in roots (Figure 2b). Additionally, we found that under drought conditions, S. spectabilis sent its hydric resources to the shoot, while G. ulmifolia and P. guachapele allocated most water towards their roots, which could help these species overcome dry periods by limiting water loss and maintaining a favourable water status (Hoffmann et al. Reference Hoffmann, Orthen and Franco2004)
Similar to water limitation, herbivory may also result in reduced plant biomass and slower development (Marquis Reference Marquis1984), in addition to being one of the major causes of seedling mortality in TDF restoration projects (Dimson & Gillespie Reference Dimson and Gillespie2020, Gerhardt Reference Gerhardt1998). We found that herbivory decreased growth of all species (Figure 2a). However, surprisingly, and although we applied extreme herbivory (complete defoliation), no individual seedling died under this treatment. Thus, abiotic factors such as water stress may play a more important role in determining seedling survival in TDF regeneration (Pearson et al. Reference Pearson, Burslem, Goeriz and Dalling2003). In fact, previous studies have shown that pioneer species show a fast recovery after herbivory events due to their life-history and ecological strategies (Atkinson et al. Reference Atkinson, Burrell, Rose, Osborne and Rees2014). Correspondingly, all species used in this experiment can resprout; S. spectabilis, P. guachapele and S. saman showed a fast post-herbivory recovery, investing their resources to the shoot and growing back their leaves soon after being completely defoliated (Figure 2b). In contrast, G. ulmifolia prioritised the investment of biomass to its roots as a response to herbivory (Figure 2b), showing a preference to resistance over fast recovery.
Species’ strategies to cope with the interaction of abiotic and biotic stresses
Although the effects of water limitation and herbivory on seedling growth and survival have been previously evaluated (Atkinson et al. Reference Atkinson, Burrell, Rose, Osborne and Rees2014, Gerhardt Reference Gerhardt1998, Marquis Reference Marquis1984, Poorter & Markesteijn 2007, Slot & Poorter Reference Slot and Poorter2007), there is a current lack of information on the effects of their interaction on seedling performance, which is key to understand plant establishment dynamics (Gerhardt Reference Gerhardt1998, Quentin et al. Reference Quentin, O’Grady, Beadle, Mohammed and Pinkard2012). For this reason, it is crucial to identify what are some of the strategies plants may use to cope with these stress factors.
Plants have a range of ecological strategies to cope with environmental adverse conditions. For example, dry forest trees usually have resistance mechanisms to cope with drought such as having compound leaves, strong stomatal control and high stem dry matter content, while minimising leaf area and maximising biomass investment to roots (Parkhurst & Loucks Reference Parkhurst and Loucks1972, Poorter & Markesteijn 2007). In fact, Poorter and Markestein (2007) found that deciduousness accounts for almost 70% of the interspecific variation in seedling drought survival in a Bolivian TDF. Hence, deciduousness is an important trait that prevents water loss during dry conditions, but is more commonly found in adults than in seedlings (Poorter & Markesteijn 2007, Reich & Borchert Reference Reich and Borchert1984). Accordingly, we found the lowest survival rate for all species under the water limitation treatment (C) (56.2%), but, surprisingly, all species had higher than 80% survival rate under drought and herbivory (81.25% on average for C+H) (Figure S2b). Thus, it is possible that the induced defoliation in the herbivory treatment conferred resistance to drought due to lower photosynthesis rates, and therefore, reduced transpiration and cavitation risk, combined with higher water allocation to roots (Figure 2b) but lower growth rate (Figure 2a) (Poorter & Markesteijn 2007, Reich & Borchert Reference Reich and Borchert1984), which could be advantageous for TDF seedlings facing drought.
Another strategy of persistence for woody species in dry forests is resprouting, an important adaptation to face disturbances like biomass removal by fires or herbivores and shoot dieback due to drought (Poorter et al. Reference Poorter, Kitajima, Mercado, Chubiña, Melgar and Prins2010, Strauss & Agrawal Reference Strauss and Agrawal1999). This mechanism is correlated with resource allocation patterns and guarantees the recovery of biomass needed for plant development after damage (Poorter et al. Reference Poorter, Kitajima, Mercado, Chubiña, Melgar and Prins2010). In our study, G. ulmifolia had the highest survival rate of all species with 93.75% survival across treatments and was the only species that invested more biomass to roots than shoots overall. Moreover, G. ulmifolia allocated up to 65% of its biomass to roots when exposed to both abiotic and biotic stress (C+H) (Figure 2). Such strategy ensures a better resprout and persistence rate compared to other species, hence, better establishment, since resources allocated to roots are better protected and may be readily used at any time. In addition, a higher root biomass enhances water uptake and access to nutrients and water (Slot & Poorter Reference Slot and Poorter2007). This results in an increase in plant N concentration, and consequently, higher photosynthetic capacity per unit leaf area, and reduced water loss (Slot & Poorter Reference Slot and Poorter2007). Accordingly, G. ulmifolia, P. guachapele and S. saman tended to allocate their water resources in the roots (Figure 2c), possibly reflecting a desiccation avoidance strategy, increasing access to water and water storage and reducing water loss (Pineda-García et al. Reference Pineda-García, Paz and Meinzer2013).
In addition to water, the storage of non-structural carbohydrates (NSCs) in stems and roots is another important trait related to drought and herbivory resistance (O’Brien et al. Reference O’Brien, Leuzinger, Philipson, Tay and Hector2014, Strauss & Agrawal Reference Strauss and Agrawal1999). Although NSCs were not measured in this study, previous work has shown that these resources help seedlings survive drought and herbivory events, as their high mobility inside the plant assures root reserves that are critical for an optimal resprout (Hoffmann et al. Reference Hoffmann, Orthen and Franco2004, Khurana & Singh Reference Khurana and Singh2004, O’Brien et al. Reference O’Brien, Leuzinger, Philipson, Tay and Hector2014, Poorter & Markesteijn 2007, Strauss & Agrawal Reference Strauss and Agrawal1999, Verdaguer & Ojeda Reference Verdaguer and Ojeda2002). In fact, previous studies have shown that resprouting species show higher belowground biomass allocation and higher starch concentrations in roots than in shoots (Zeppel et al. Reference Zeppel, Harrison, Adams, Kelley, Li, Tissue, Dawson, Fensham, Medlyn, Palmer, West and McDowell2015). In addition, high biomass invested in roots enhances water and nutrient uptake (Poorter & Markesteijn 2007). Despite the fact that pioneer species do not usually have big NSCs reserves (Atkinson et al. Reference Atkinson, Burrell, Rose, Osborne and Rees2014), we found a general pattern of biomass allocation to the roots of G. ulmifolia, and P. guachapele under the Ir+H treatment, but much higher in G. ulmifolia (Figure 2), which could indicate allocation of NSC reserves, and an advantage for post-stress resprouting.
Finally, although previous studies have shown that TDF species with larger seeds have higher survival than small-seeded pioneer species under water stress (Khurana & Singh Reference Khurana and Singh2004), we did not detect differences in survival across treatments between four species with seed size varying two orders of magnitude (Table 1, Fig. S1). In fact, G. ulmifolia, the species with the smallest seed in our experiment, had the highest survival rate (≥ 87.5%) across treatments compared to other species (Figure 2a). Moreover, all species had reduced survival under water stress (C) and the combination of water stress and herbivory (Fig. 2a and S1), treatment under which S. saman, the species with the largest seed, had only one seedling surviving. Thus, seed size did not appear to impact survival in several-months old seedlings, although being important for initial seedling performance.
The response of N-fixing plants to water limitation and herbivory
Historically, seedlings of pioneer species have been widely used in restoration exercises due to their fast growth and adaptability to harsh conditions after disturbance (Florentine & Westbooke Reference Florentine and Westbrooke2004). In particular, legume pioneers have been prioritised, since most of them are also N-fixers, guaranteeing a constant flow of N without depending on the environment, and adding N-rich litter to the system (Batterman et al. Reference Batterman, Hedin, Van Breugel, Ransijn, Craven and Hall2013, Vitousek et al. Reference Vitousek, Cassman, Cleveland, Crews, Field, Grimm, Howarth, Marino, Martinelli, Rastetter and Sprent2002). This is an advantage given that N is a key nutrient for plant development (Leghari et al. 2016), and sites under restoration generally exhibit N limitation (Moreira dos Santos 2006). Thus, although we only worked with two N-fixing legume species (P. guachapele and S. saman), we expected them to show the best performance under stress, given their capacity of avoiding N deficiency (Vitousek et al. Reference Vitousek, Cassman, Cleveland, Crews, Field, Grimm, Howarth, Marino, Martinelli, Rastetter and Sprent2002). In support of this idea, we found that seedling growth of N-fixing species was strongly correlated with nodule DW (Figure 3). However, these species did not show an advantage with respect to non-N-fixing species (G. ulmifolia and S. spectabilis) under hydric (C) or mechanical (H) stress (Figure 2). In fact, nodule production decreased with stress factors of water limitation and herbivory (Figure 3); likewise, P. guachapele and S. saman were dependent on a relative environmental welfare to have a relatively good performance, indicating that nodules probably do not help these species’ seedlings in dealing with hydraulic or mechanical stress factors, but are key for assuring an optimal initial growth of these species (Figure 3). These results suggest that if the purpose of restoration is to increase N inputs into the soil, P. guachapele and S. saman should be less exposed to the adversity of the environment allowing for an effective nodulation and growth, using other species with better performance like G. ulmifolia as nurse plants. Second, there are other ecological traits (i.e., high stem and root biomass, low leaf ratio and compound leaves) that are appeared to be more important for drought and herbivory resistance of these species, at least in the seedling stage.
Additionally, it has been recently demonstrated that as N-fixing species grow in regenerating rain forests, they limit the growth of their neighbours, given that they auto-recycle and acquire most N back through leaf litter production (Taylor et al. Reference Taylor, Chazdon, Bachelot and Menge2017). Therefore, N fixation might not facilitate the establishment of other species due to the competitive behaviour of legumes. Owing to this competition behaviour, exploring how legume species used in TDF restoration projects interact with their neighbours as they grow is important to better understand competition interactions in successional TDF.
Conclusion
Seedling functional traits are key for coping with abiotic and biotic stresses in TDF, and this information is necessary to postulate species that may be ideal for restoration projects (Khurana & Singh Reference Khurana and Singh2001, Vargas & Ramírez Reference Vargas, Ramírez, Pizano and García2014). Based on our findings, we recommend the use of G. ulmifolia, since it was the species that best responded to stress, with an optimal survival rate, and biomass and water allocation to belowground structures. Furthermore, G. ulmifolia works as a nurse species (Torres et al. Reference Torres, Díaz, Villota, Gómez and Avella2019) and even as a feeding supplement for livestock (Castrejón et al. Reference Castrejón, Martínez, Corona, Cerdán and Mendoza2016, Manríquez et al. Reference Manríquez, López, Pérez, Ortega, López and Villarruel2011). In contrast, legume species used in this study seemed to have no advantage in terms of performance during seedling stage. However, nitrogen fixers (P. guachapele and S. saman) might be useful in restoration projects due to their ability to supply nitrogen to the soil, as long as they are not exposed to environmental adversity (low water availability and high herbivory rates), allowing for an optimal nodulation and guaranteeing an optimal N entry to the soil. In this study, we only used three legume species (two N-fixers and one non-fixer) and one non-legume species, which calls for further studies that include more species and explore other functional traits and additional abiotic and biotic pressures such as trampling by domestic ungulates, competition, light exposure and transplant shock. These results may significantly improve restoration practices of the highly threatened TDF.
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/S0266467422000074
Acknowledgements
We are grateful for logistical and economic support provided by Universidad Icesi. We also thank Eucaris Escobar, Manuela Chávez and Sarah González for their assistance in the greenhouse and the laboratory when processing plants from the experiment.
This study was funded by a grant from Universidad Icesi to CP.
