Phenotypic plasticity and insect herbivory of trees in contrasting light environments in a Mexican rainforest

Abstract

Contrasting light environments in rainforests generates changes in the characteristics of the leaves and in the herbivore community. In the present study, we carried out a reciprocal transplant experiment under natural conditions to determine the plasticity of leaf characteristics of plant species that grow in contrasting light environments in a Neotropical forest. We further explored the relationship between these traits and insect herbivory. We found that six woody species differ markedly in the phenotypic plasticity of leaf features. The specific leaf area, chlorophyll content, carbon content, nitrogen content, and leaf thickness of the most light-demanding species were highest in gaps, but their carbon/nitrogen ratios were higher under closed canopies. The herbivores were more abundant in gaps (5.9%–14.8%) than under closed canopy habitats (3.4%–6.1%) and seemingly associated to the plasticity of the leaf traits. We observed 47% more herbivores in gaps than under closed canopies. Our results suggest that the phenotypic plasticity of leaf traits depends on the identity of the plant species and its wood density, while herbivory seems to be affected by plant defence, low nutritional quality, or herbivore tolerance.

Introduction

In tropical ecosystems, light reaching the understory ranges between 1% and 3% of the total photon flux density available in the upper canopy layer (Dupuy and Chazdon 2008). Photosynthetically active radiation varies by three orders of magnitude from closed canopy understory to gaps (Stiegel et al. 2017). Therefore, light is considered a limiting resource that influences understory plant growth, reproduction, and survival (Baraza et al. 2010). Gaps produced by the fall of large branches and tall trees are an important component of forest regeneration dynamics, and it is estimated that the average hectare of tropical forest has up to eight gaps smaller than 100 m² (Poorter et al. 2023). One of the most notable changes in gaps compared to the closed forest canopy is greater light availability, so understory plants and those thriving in gaps must continually adjust to a changing light environment by modifying the expression of physiological, anatomical, and morphological traits (Lundgren and Des Marais 2020; Laughlin 2023).

Changes in leaf traits allow plants to adapt to light heterogeneity efficiently (Valladares et al. 2000). For example, light-demanding species allocate resources to rapid growth, allowing them to escape the shade of the understory by growing taller to reach areas where light is more abundant. They are characterised by thin leaves with high nitrogen content and more chloroplasts, which allow higher photosynthetic yield, keeping their leaves cool despite being exposed to intense solar radiation (Poorter and Bongers 2006; Tripathi et al. 2020). In contrast, shade-tolerant species generally have thick, nitrogen-deficient but carbon-rich leaves and a low photosynthetic rate despite their high levels of chlorophyll (Falster et al. 2018). In turn, these characteristics determine herbivory levels; for example, light-demanding leaves are consumed by insects more frequently than shade-tolerant leaves (Gianoli et al. 2023).

Phenotypic plasticity is the ability of an organism to express different phenotypes in response to its biotic or abiotic environment (DeWitt and Scheiner 2004; Adler et al. 2014; Adams et al. 2023). Many plants are phenotypically plastic and can respond to environmental cues in a variety of ways, including changes in their biochemical, physiological, and morphological traits (Agrawal 2001). Also, phenotypic plasticity can differ depending on the species and their leaf traits. For example, light-demanding species are characterised by having greater plasticity in physiological traits than in morphological traits (Valladares 2003). Phenotypic plasticity is thought to be greater in light-demanding species because they grow in a more variable environment with abundant resources (Scheepens et al. 2018). For example, Lortie and Aarssen (1996) mention that specialisation in favourable environments increases plasticity, whereas specialisation in less desirable environments decreases plasticity. This evidence suggests that light-demanding species should exhibit greater phenotypic plasticity than shade-tolerant species. This hypothesis is supported by Valladares et al. (2000), but Rozendaal et al. (2006) proposed that key functional attributes of shade-tolerant species should be more phenotypically plastic since they grow in highly varied light environments. Thus, no consensus exists as higher, similar, and lower plasticity has been found in light-demanding species compared to shade-tolerant species (Valladares and Niinemets 2008).

On the other hand, the composition and abundance of insect communities are affected by abiotic conditions that generate tree gaps (gaps vs. closed canopy), causing a filter that excludes many species (Richards and Windsor 2007). Previous studies on the effects of light environments on insect herbivores have reported increased abundance and richness in full sun environments, due to warmer temperatures of the gaps that can increase the metabolism and activity of insects (Łukowski et al. 2015). However, some studies document increased abundance and feeding in closed canopy environments as well (Karolewski et al. 2013). Therefore, the specific way in which insects respond to changes in solar radiation on their host remains open to debate.

We conducted a reciprocal transplant experiment under natural conditions to assess the leaf attributes plasticity of plants with different degrees of shade tolerance (three light-demanding and three shade-tolerant plant species) and their relationships with insect herbivores. We hypothesise that shade-tolerant plant species should have greater phenotypic plasticity than light-demanding species as a result of changing light availability (Rozendaal et al. 2006). Plasticity of structural and physiological traits was evaluated and subsequently related to insect herbivory. In addition, insects were collected from the experimental sites to learn about these communities at each site. The study questions were: (1) Which foliar traits show greater phenotypic plasticity? (2) Is herbivore damage greater in light- or shade-tolerant species? (3) Is phenotypic plasticity in foliar traits associated with the amount of leaf tissue consumed? (4) Is the abundance of insects on plant species different in gaps in closed canopy sites?

Materials and methods

Study area

The study was conducted at the Universidad Autónoma Nacional de México (UNAM) in Estación de Biología Tropical Los Tuxtlas (18°30′ N, 95°10′ W) in the northern part of the Los Tuxtlas region of Mexico. This area covers an altitudinal range of 150–700 m where the predominant vegetation type is tropical rainforest. Mean annual temperature and precipitation are 27°C and 4,700 mm, respectively (Cornejo-Tenorio et al. 2019). Lateritic soils including luvisol and acrisol predominate in up lands, while deep, fertile soils with high organic matter such as vertisol, feozem, and litosol predominate in the coastal plain (Cornejo-Tenorio et al. 2019). The dynamics of gaps created by falling branches and trees are fundamental for forest regeneration in the hot and humid tropics (Matsuo et al. 2021).

Study sites

We identified five gaps in the study site, ranging between 40 m² and 180 m² in area and separated by at least 50 m (Figure S1). In each gap, we selected a closed canopy area located 20–25 m from the gap edge. We monitored air temperature, relative humidity (10:00 am–12:00 pm), and canopy cover every two months for one year in all 10 experimental sites (five gaps and five closed canopy sites). We used a Kestrel micro-meteorological station (Model 5500, México) to measure air temperature and relative humidity and a hemispherical densitometer to assess canopy cover (Forestry Supply, Model A, USA). All measurements were done in the centre of the gaps where the experimental plants were located to the extent possible. On average, canopy coverage under the closed canopy was 75.9% ± 10.1% and in gaps was 29.9% ± 12%. The relative humidity (95.4 ± 3.9 and 94.4 ± 4.6 for closed canopy and gap sites, respectively) and air temperature (24.9 ± 2.94 and 24.7 ± 2.86 for closed canopy and gap sites, respectively) were similar in the two environments (Lu et al. 2023).

Study species

Fabaceae and Moraceae (Fabidae) and Rubiaceae (Lamiidae) are the three most prominent plant families in Los Tuxtlas (Cornejo-Tenorio et al. 2019). These three families include species with light-demanding and shade-tolerant species. We selected two species from each of these three families, pairing a light-demanding and shade-tolerant species to account for phylogenetic relatedness, at least in part, and considering that reciprocal wood density (WD) correlates positively with growth rate and regeneration mode of woody species in tropical rainforest (King et al. 2006; Chave et al. 2009). We used data from the global WD database (Zanne et al. 2009) to select the six species. From Rubiaceae, we selected Palicourea faxlucens (Lorence and Dwyer) Lorence (WD = 0.36) and Faramea occidentalis (L.) A. Rich. (WD = 0.55), which show a 52% difference in WD. From Fabaceae, we paired Pterocarpus rohrii Vahl (WD = 0.32) and Cynometra retusa Britton and Rose (WD = 0.85) with a 165% difference in WD. From Moraceae, we chose Poulsenia armata (Miq.) Standl. (WD = 0.4) and Pseudolmedia glabrata (Liebm.) C.C. Berg (WD = 0.61) with a 53% difference in WD.

Experimental design

In March 2017, we selected 40 saplings (20 cm ≤ 35 cm in height) of each focal species within a 10 m radius of 10 parental trees. The selected parental trees were at least 30 m away from other conspecifics. Each of the 240 (40 individuals × six species) saplings was transplanted into 5 L pots and kept under the parental tree canopy for 101 days to acclimate (Delph 2017). By June 2017, 152 out of the 240 saplings were alive, and we distributed them in the experimental sites (gaps and closed canopy sites) balancing as much as possible the number of saplings of each species in each site. The actual number of saplings at each site ranged between 15 and 17, and there were no fewer than two individuals of each species in each site (with the exception of one with a single sapling of Pa. faxlucens); the maximum number of saplings of a single species at any site was four. The total number of saplings of each species ranged between 23 and 28.

Leaf functional traits

After 276 days in the experimental sites (March 2017–March 2018), we recorded eight leaf traits from all surviving saplings. Functional traits were measured in the middle region of each sapling in the experiment. The thickness of the leaf blade was taken with a digital calliper (MDC-1 MX, Chicago, USA). We used a portable chlorophyll metre to estimate chlorophyll content (atLEAF, Chicago, USA), and SPAD units were transformed into concentration units (mg/cm²) following the linear model proposed by Zhu et al. (2012). Leaf toughness, measured as (g/cm²) required to break the leaf blade, was estimated with a penetrometer (KSE-2006072, Mexico). We kept the leaf blade fully extended between two pieces (7 cm × 7 cm) of plexiglass (3 mm thick) with a central hole (1 cm diameter), and the rod of the penetrometer held vertically, avoiding the first- and second-order veins, and then pressed firmly to perforate the blade. We used a rod with a diameter of 3 mm for all species except Pt. rohrii, for which we used a rod with a diameter of 5 mm because of its softer leaves. The measurements were later transformed to compare with those of the other five species. We estimated the fresh leaf blade area, fresh leaf mass, and dry mass with a portable flatbed scanner (CI-202 Bio-Science, Boston, USA). In addition, we dried the leaves in the oven at 60°C for 72 h to measure dry mass. To estimate the specific leaf area (SLA, cm²/g), we divided the fresh leaf area by the dry mass.

Water content of leaves and soil nutrients

To estimate the water content of the leaves (g ^g−1), we calculated the difference between fresh and dry masses and then divided by the dry mass and multiplied by one hundred. For each species at each site, we collected 10 intact leaves from each marked plant. The nitrogen and carbon contents (% of dry mass) were measured with an automated combustion analyser (TruSpec CN, Leco Corporation 2002, Michigan, USA) in the soil, water, and plant chemical analysis laboratory at Instituto de Ecología (INECOL).

Leaf herbivory

We marked all intact leaves of each plantlet with metal rings in June 2017 and every two months thereafter from June 2017 to March 2018. We also recorded cumulative evidence of herbivory following the method by Dirzo and Domínguez (1995) for non-destructive quantification. We accounted for damage left by chewing herbivores (incomplete margins, perforations, and missing areas inside the leaf blade) and recorded areas scraped off the surface of the leaf (e.g. by miners or suckers). The following damage categories were used: zero (intact leaves), one (leaves with barely perceptible damage and up to 6% of the leaf area), two (more than 6% and up to 12% in damage), three (12%–25%), four (25%–50%) and five (more than 50% in damage). The score for each leaf was used to calculate an index of damage (ID) as

$$ID = \mathop \sum \limits_{i = 0}^5 \left( {{x_i}{N_i}} \right)N$$

where X _i is the i scores of damage (0 to 5), n _i is the frequency of leaves with score X _i , and N is the total number of scored leaves. The herbivory index was transformed into a percentage according to Ruiz-Guerra et al. (2017). Specifically, this method is based on the percentage of average expected herbivory (EH) for each category of damage:

$$EH{\rm{\;}}\left( {\rm{\% }} \right) = {H_{i{\rm{\;}}max}} {{{{H_{i{\rm{\;}}max}} - {H_{i{\rm{\;}}min}}}} \over {2}}$$

where H _{i max} and H _{i min} correspond to the upper and lower asymptotic limit of herbivory for each category of damage, respectively. In this way, we obtained the expected average herbivory percentages of 0%, 3%, 9%, 18.5%, 37.5%, and 75% for damage categories 0–5, respectively. Finally, through a cubic linear regression with an intercept adjusted to zero, the herbivory percentage was obtained for any value of the index of damage (ID):

$$Herbivory = {\rm{\;}}\left( {\rm{\% }} \right) = 5.6131{\rm{\;}}ID - 2.4505{\rm{\;}}I{D^2} + 0.8691{\rm{\;}}I{D^3}$$

Insect community

Insect sampling at all sites was carried out in March, June, and August 2018 during dry and rainy seasons in the morning (09:00–12:00 h) and evening (21:00–24:00 h). During each sampling event, five randomly selected trees of the study species (2–8 m height) were firmly hit five times (without fatally damaging the branches) using a stick, and all falling arthropods were collected in a conical 1 m² beating sheet emptying into a 50 ml container with 10 ml of ethanol. All taxa were identified to the order level and classified according to their feeding guilds by taxonomist María Luisa Castillo (MLC) from INECOL (Cibrián et al. 1995; Buck et al. 2009).

Phenotypic plasticity index

To estimate the phenotypic plasticity of the species found in gaps versus closed canopy habitats, we modified the procedure by Valladares et al. (2000). The plasticity index was calculated as the difference between the trait value in the closed canopy habitat and its value in the gaps, divided by the largest of the two values. We calculated all possible combinations between shade and light environments for each of the six species. We kept track of the identity of the sites where the plant belonged in each calculation to use it as a random factor, together with the specific identity of the plants. Calculated in this way, positive values of the plasticity index will indicate that the trait is more variable in the shaded environment, while negative values indicate greater plasticity in gaps.

Statistical analysis

To evaluate whether the values of the phenotypic plasticity index were statistically different from zero, we used a mixed effects model in which tree species (six levels) was the fixed factor, and the identity of the pair of sites for plasticity estimates was defined as a random factor to control for the variance for each species and avoid pseudo-replication. In all cases, the variance was modelled within the fixed factor with a constant variance equal to one, while we used the raw estimated variances for the random factor. To explore the overall pattern of phenotypic plasticity, we used non-metric multidimensional scaling between species and traits. We also used cluster analysis with the Ward clustering method and Euclidian distances.

To evaluate the effect of the light environment and the specific identity of the plants on cumulative herbivory, we used a mixed effects model with repeated measurements. The fixed factors were habitat (two levels) and species (six levels) with time as a covariate. The identity of the gap and closed canopy sites was used as a random variable analysed within each time (repeated measures model). A logarithmic transformation of the percentage of herbivory was used.

To evaluate the correlation between the phenotypic plasticity of the plant traits and cumulative herbivory in the two light environments, we used a co-inertia analysis based on principal components. The co-inertia analysis, like other multivariate techniques including canonical correlation and redundancy analysis, seeks to correlate two data matrixes that share the identity of the observation units; however, the co-inertia analysis is more robust when dealing with collinearity among variables within each of the matrixes (Dray et al. 2003). We used a Monte Carlo test to establish the significance of the observed correlation between the two tables (Heo and Gabriel 1997).

To analyse possible differences in the abundance of invertebrate trophic guilds found in the two environments, we used a contingency table with six columns (trophic guilds) and four rows (environments in daytime and night-time). For the purpose of this analysis, we summed all samples for daytime and night in each environment. All analyses were done in R version 4.3.2 (R Core Team 2023).

Results

Phenotypic plasticity

Non-metric multidimensional scaling found that SLA and leaf water content were the most variable traits among species (Figure 1). The intraspecific variation between environments, although limited, suggested a concordant increase in nitrogen and chlorophyll in gaps.

Figure 1. Functional traits of six woody, tropical rainforest tree species in Mexico. The non-metric multidimensional scaling plot in the central panel highlights clusters of each species: Cynometra retusa = Cy. retusa (green) and Pterocarpus rohrii = Pt. rohrii (red) in Fabaceae, Pseudolmedia glabrata = Ps. glabrata (blue) and Poulsenia armata = Po. armata (orange) in Moraceae, and Faramea occidentalis = Fa. occidentalis (purple) and Palicourea faxlucens = Pa. faxlucens (magenta) in Rubiaceae. Solid lines encircle points representing saplings grown under a closed canopy, and dotted lines surround points from saplings in gaps. Peripheral panels represent the phenotypic plasticity index for each foliar trait (SLA = specific leaf area). The data are presented as (values in closed canopy) – (values in gaps). Means, whiskers, and error bars (standard error) in the grey section represent traits with higher average values under a closed canopy. Means, whiskers, and error bars in the white section represent traits with higher average values in gaps. *P = 0.01, **P = 0.001, ***P = 0.0001 significance level.

Leaves of shade-tolerant species were on average 11% tougher when grown under the canopy of the forest compared to when they were grown in gaps (t = 4, df = 26, P < 0.001). Nonetheless, when separated by species, the leaf toughness of Cy. retusa and Fa. occidentalis increased by 13.6% (t = 4.2, df = 34, P < 0.001), while there was no significant difference for Ps. glabrata (t = 0.3, df = 68, P = 0.73). In contrast, the leaf toughness of the light-demanding species did not change between light environments (t < 0.1, df = 23, P > 0.05).

Similarly, two out of the three shade-tolerant species (Cy. retusa and Fa. occidentalis) developed thicker leaves (11% and 7%), respectively (Table S1), under closed canopy, compared to leaf thickness in gaps (t = 3.7, df = 23, P = 0.001). The leaf thickness of Ps. glabrata did not differ significantly between environments (t = 0.5, df = 68, P = 0.59). There were also mixed responses for the light-demanding species. Poulsenia armata leaf thickness did not differ between environments (t = 1.6, df = 68, P = 0.10) but increased in Pt. rohrii and Pa. faxlucens (t > 2.8, df = 68, P < 0.005). Leaves of Pt. rohrii were 18% thicker under closed canopy than in gaps, meanwhile Pa. falcons had 33% thicker leaves in the gaps than closed canopy forest.

The leaf water content of Ps. glabrata (t = 0.4, df = 68, P = 0.71) did not change between environments, while for the two other shade-tolerant species, it changed (t = 10.2, df = 34, P < 0.001). The leaf water content of Cy. retusa was 6% higher in the gaps than under closed canopy (Table S1). In contrast, leaf water content was 28% higher in Fa. occidentalis under closed canopies than in gap environments. Leaf water content increased under forest canopies for the light-demanding species Poulsenia armata (22%) and Pa. faxlucens (20%) compared to gaps (t = 15.2, df = 34, P < 0.001). There was no change between environments in Pt. rohrii (t = 0.5, df = 68, P = 0.60).

Specific leaf area in light-demanding species was consistently higher in gaps than in closed canopy environments (t = 22.3, df = 23, P < 0.001). The specific leaf area of Pt. rohrii increased by 33% in gaps and increased by 19% and 17% in Po. armata and Pa. faxlucens, respectively, in gaps (Table S1). In contrast, the shade-tolerant species Cy. retusa (19%) and Fa. occidentalis (4%) had higher specific leaf area under closed canopies compared to gaps (t = 11.4, df = 34, P < 0.001), while Ps. glabrata (17%) showed the opposite response (t =12, df = 68, P < 0.001).

Chlorophyll content showed the same pattern as leaf water content in shade-tolerant species. The chlorophyll content did not change in Ps. glabrata between environments (t = 1.6, df = 68, P = 0.08), while Cy. retusa and Fa. occidentalis had opposite responses. The chlorophyll content of Cy. retusa increased by 8% when grown in the gaps compared with closed canopies (t = 2.3, df = 68, P = 0.012), while the chlorophyll content was 14% higher in Fa. occidentalis under closed canopies compared to gaps (t = 4.3, df = 68, P < 0.001) (Table S1). Similarly, results were mixed for light-demanding species. The chlorophyll content of Pa. faxlucens did not change between environments (t = 1.7, df = 68, P < 0.09), while Pt. rohrii and Po. armata had 49% and 13% more chlorophyll in gaps and closed canopy forest (t > 5.5, df = 34, P < 0.001) (Table S2).

The three light-demanding species had consistently higher leaf carbon content (4%–10%) when grown in gaps than when grown under forest canopies (t > 9.4, df = 34, P < 0.001) (Table S1). Similarly, the nitrogen contents of Pt. rohrii (27%) and Pa. faxlucens (32%) were higher in gaps than under closed canopy (t > 10.3, df = 34, P < 0.001). In contrast, the nitrogen content of Po. armata did not change between environments (t = 0.4, df = 68, P = 0.65). Because of that, the C:N ratio in Po. armata also did not change between gaps and closed forest canopy (t = 0.7, df = 68, P = 0.46), while the C:N ratio was higher (20%–31%) in the other two light-demanding species under closed forest canopy compared to gaps (t > 6.3, df = 68, P < 0.001). In shade-tolerant species, the carbon content changed significantly between environments only in Ps. glabrata (2%, t = 2.1, df = 68, P = 0.048), while the nitrogen content changed significantly in all three species, albeit with mixed responses. In Cy. retusa (7%) and Ps. glabrata (9%), the nitrogen content was higher when grown in gaps than under closed canopy. In contrast, Fa. occidentalis (7%) had higher nitrogen content under closed canopy compared to gaps (t = 2.4, df = 68, P = 0.017). Consequently, the C:N ratios of shade-tolerant species were variable. The C:N ratio was higher in Cy. retusa (7%) and Ps. glabrata (9%) in closed canopies than in gap environments (t = 3.5, df = 34, P = 0.001), while the C:N ratio was higher (7%) in Fa. occidentalis when grown in the gaps compared to the closed forest canopies (t = 2.6, df = 68, P = 0.01) (Table S1).

A cluster analysis of plant trait plasticity found that Fa. occidentalis and Po. armata were similarly variable (dissimilarity less than 30%) (Figure 2). Also, with a dissimilarity of 40%, Cy. retusa and Ps. glabrata formed a second pair of species with similar responses and an average dissimilarity of ca. 60% in relation to the cluster with Fa. occidentalis and Po. armata. The main difference between these two pairs was in the phenotypic plasticity of specific leaf area. Finally, Pt. rohrii and Pa. faxlucens were 90% dissimilar to the other four species. Coincidentally, the groups corresponded to high, mid, and low WD.

Figure 2. Grouping of six tree species of the tropical rainforest in Mexico under two light environments based on the plasticity of eight functional leaf traits. Cynometra retusa = Cy. retusa, Pterocarpus rohrii = Pt. rohrii, Pseudolmedia glabrata = Ps. glabrata, Poulsenia armata = Po. armata, Faramea occidentalis = Fa. occidentalis, and Palicourea faxlucens = Pa. faxlucens.

Herbivory levels

Leaf damage by herbivores over 259 days differed significantly among species, and there was interaction with the type of environment (Χ ² = 13.96, df = 6, P < 0.03, Figure 3). In the closed canopy habitat, average herbivory ranged from 3.4% to 6.1%, while in the gaps, it ranged between 5.9% and 14.8%. Herbivory was higher in forest canopy sites for Ps. glabrata (Moraceae), Pa. faxlucens (Rubiaceae), and both species in the Fabaceae (Cy. retusa and Pt. rohrii), compared to the herbivory observed in the closed canopy habitat (t > 3.2, P < 0.002) (Figure 3). For Po. armata and Fa. occidentalis (Moraceae and Rubiaceae, respectively), there were no significant differences in herbivory between habitat types (t < 0.4, P > 0.33).

Figure 3. Cumulative herbivory over 259 days of field exposure for six tree species of the tropical rainforest in Mexico in two light environments. Cynometra retusa = Cy. retusa, Pterocarpus rohrii = Pt. rohrii, Pseudolmedia glabrata = Ps. glabrata, Poulsenia armata = Po. armata, Faramea occidentalis = Fa. occidentalis, and Palicourea faxlucens = Pa. faxlucens.

Correlation of phenotypic plasticity and insect herbivory

Based on the co-inertia analysis, we observed an overall correlation coefficient of 64% (P = 0.021) between leaf trait plasticity and the percentage of herbivory observed in gaps and closed canopy habitats. Species with a high C:N ratio in closed canopy forests but a low C:N ratio in gaps (Pa. faxlucens, Ps. glabrata, and Pt. rohrii) were more heavily attacked by herbivores in gaps. In contrast, herbivory on Po. armata, Fa. occidentalis, and Cy. retusa did not differ between environments. They also had the least variable C:N ratios.

Insect community

Overall, we collected 1,202 insects from seven different feeding guilts (Table S3). They were similarly abundant in closed canopy sites (607) and gaps (595). Herbivores collected at night accounted for 59% and 65% of the insects in closed canopies and gaps, respectively. The abundances of trophic guilds changed significantly between habitat types (X ² = 111.2, df = 15, P < 0.001). The residual analysis demonstrated that herbivores were overrepresented in the closed canopy sites during the night census but underrepresented in the morning census (z = 2.4, P = 0.016 and z = 2.7, P = 0.007, respectively). In contrast, abundances of herbivores in gaps were no different than by chance (z < 0.1, P = 0.92). However, the overall abundances of herbivores were similar in closed canopy sites and gaps, respectively. Detritivorous taxa in closed canopy sites, unlike herbivores, were overrepresented in the daytime (z = 6.7, P < 0.001) and underrepresented at night (z = 3.2, P = 0.001); however, their abundances did not differ from random expectation in gaps (z < 1.4, P > 0.161). Omnivores in closed canopy sites were overrepresented in the night census (z = 3.3, P < 0.001) and underrepresented during daytime in gaps (z = 2.1, P = 0.035). Also, mycetophagus taxa were overrepresented in gaps during daytime (z = 2.6, P = 0.009). Overall, all trophic guilds were similarly abundant in different habitat types (z = 1.9, P > 0.057).

Discussion

We found that six Mexican tropical rainforest tree species differed markedly in the phenotypic plasticity of eight leaf traits. Two of the light-demanding species, which also had the lowest wood densities, had the most phenotypically plastic responses. Specific leaf area and carbon contents were highly variable in Pt. rohrii, and leaf thickness and nitrogen content were plastic in Pa. faxlucens. The Moraceae species Ps. glabrata and Po. armata, along with Pt. rohrii and Pa. faxlucens, had higher specific leaf area in gaps than under the closed canopy. Herbivory of Po. armata was significantly higher in gaps than under closed canopies. On the other hand, C:N

ratios do not differ between light environments, yet C:N ratios are lower in low-light environments (Luo et al. 2022). The carbon gain hypothesis posits that saplings tend to maximise net carbon gain to perform optimally in low-light conditions, for example, to alter plant morphology to maximise light capture (de Mello Prado and da Silva 2017).

Phenotypic plasticity of C:N ratios was high in Cy. retusa and Ps. glabrata, and higher trait values were observed under a closed canopy. The high C:N ratio under a closed canopy translates into low nutritional values of the leaves for herbivores (Coley and Barone 1996). Because the overall abundance of herbivores between environments showed did not differ significantly, the evidence suggests that these species shifted their antiherbivore defence or could have differed in their host specialisation, size, etc. Defence syndromes are defined as the orchestrated expression of functional traits to minimise the impact of herbivores (Agrawal and Fishbein 2006). Recent studies find that defence syndromes range from tolerance to herbivory. Physical and chemical defences, along with low nutritional quality, deter herbivores (Ruiz-Guerra et al. 2021; Gianoli et al. 2023).

Overall, we observed 47% more herbivory in gaps than under closed canopy. This is consistent with the findings by Piper et al. (2018) who, in a meta-analysis, found almost 50% more herbivory in tropical forest gaps compared to closed canopy environments, a pattern also reported by De la Cruz and Dirzo (1988), Boege and Dirzo (2004), and Ruiz-Guerra et al. (2021) in a Mexican tropical rainforest.

Our results contradict the findings of Rozendaal et al. (2006), who concluded that shade-tolerant species are more phenotypically variable than light-demanding species. We observed the greatest phenotypic plasticity in two of the light-demanding species. This agrees with the hypothesis of increased phenotypic plasticity in resource-rich environments (Lortie and Aarssen 1996) and with Valladares et al. (2000) who investigated phenotypic plasticity in 16 Psychotria species on Barro Colorado Island. Nonetheless, we observed that each species had some traits that differed between environments, and the most frequent differences are seemingly related to the species’ WD. WD is arguably a surrogate for growth rate (King et al. 2006; Chave et al. 2009) and plant regeneration mode (light-demanding and shade-tolerant species). Specific leaf area and high nitrogen content were the most variable in the two species with the lowest WD, Pt. rohrii and Pa. faxlucens. The high values of these traits in gaps indicate high photosynthesis and nutrient assimilation and are consistent with the high levels of herbivory observed in these species.

In general, the carbon content of light-demanding species was highly variable between environments and was generally higher when they grew under forest canopy sites. However, leaf toughness did not change significantly between light environments. This suggests that the increase in carbon content in gaps may be in the form of non-structural carbon compounds such as starch, sucrose, and fructose (Hoch et al. 2003). In Cy. retusa and Fa. occidentalis, phenotypic plasticity was observed in leaf toughness. Leaves were tougher when grown under a forest canopy. In contrast, carbon content did not change between light environments, suggesting at least a partial substitution of some structural carbon compounds accumulated under the closed canopy of the forest for non-structural carbon compounds in gaps. One way in which this relocation of compounds may occur is by the incorporation of soluble sugars from the cell apoplast into structural defences (Zhang et al. 2021).

In conclusion, we demonstrate that phenotypic plasticity in foliar traits of woody tropical rainforest species occurred in light-demanding and shade-tolerant species, and the most frequently observed response patterns seem to be associated with the species’ WD. The trait is commonly used as a surrogate for growth rate and regeneration mode. Consistent with the resource availability hypothesis (Gianoli and Salgado-Luarte 2017), we found that herbivory was higher in gaps than in closed canopies. We hypothesise that this is a consequence of an environmentally mediated shift from herbivore tolerance in gaps to chemical defence-related herbivore resistance (or low nutritional quality) under a closed canopy.