How do soil resources affect herbivory in tropical plants along environmental gradients? A test using contrasting congeneric species

Plants adapted to different habitats exhibit differences in functional traits and these characteristics are influenced by soil properties. We tested the hypothesis that soil resource availability influences the functional traits of plants, affecting therefore herbivory levels. We examined three Byrsonima plant species with different life forms that occurred across a distinct edaphic habitat along the Doce River Basin, South-eastern Brazil. We characterized habitats according to soil nutrient concentration and measured functional characteristics of crown architecture, leaf nutrients, sclerophylly, leaf area and leaf density. In addition, we evaluated how these variables influenced herbivory levels of congeneric plants. Our data show that species along a gradient of soil nutrients have functional characteristics influenced by habitat, which in turn affect herbivory levels. By comparing congeners with different life forms found along a stress-gradient of continuous habitats, we describe a corresponding gradient of plant functional traits and tissue consumption by herbivorous insects.


Introduction
Plants adapt to habitat characteristics by morphological and physiological adjustments to specific abiotic conditions (Harper 1977;Mark et al. 2001;Mello et al. 2020). Soil nutrients are amongst the most influential drivers of species distribution (Prentice et al. 1992;Rodrigues et al. 2018), morphological trait selection (Cunningham et al. 1999;Ordoñez et al. 2009) and belowground interaction food webs (Laliberté et al. 2017). Edaphic factors can influence plant growth (Dighton and Krumins 2014), morphology (Bona et al. 2020) and secondary chemistry (Fine et al. 2004(Fine et al. , 2006. These morphological, physiological and/ or phenological characteristics that affect growth, survival and ultimately fitness are considered functional traits (Violle et al. 2007). Plant survival and fitness are also influenced by insect herbivory, which can be an important selective pressure (Coley and Barone 1996;Allan and Crawley 2011;Jogesh et al. 2016). Leaf damage caused by insects can impact net primary production , decrease pollinator attraction (Lehtilä and Strauss 1997;Moreira et al. 2019) and reduce reproductive capacity (Strauss et al. 1996;Schiestl et al. 2014;Kozlov and Zvereva 2017). In turn, the evolutionary investment on growth or defences is a main trade-off in life history traits related to plant adaptation to cope with herbivory (Herms and Mattson 1992;Poorter et al. 2010).
Plant chemical and physical defences are amongst the most studied factors influencing herbivory levels (reviewed by Carmona et al. 2011;, whereas other traits such as plant size, appearance and architecture have received considerably less attention in the literature (but see Castells et al. 2017;Martini et al. 2021). Variation in plant nutrient quality has been also suggested as an important cause of variation in herbivory levels (Mattson 1980;Johnson 2008). Usually, nutritious plants are those with high concentrations of macronutrients, such as nitrogen-which is essential for insect survival and reproduction (Raubenheimer and Simpson 1997;Roeder and Behmer 2014)-and are likely to occur more frequently on fertile soils (Wright et al. 2001).
Host plant structural complexity (Spawton and Wetzel 2015) and leaf nutritional quality (Maldonado-López et al., 2014) affect the quantity and quality of available food resources that in turn might directly affect herbivory levels (Campos et al. 2006;Schlinkert et al. 2015). According to the plant architecture hypothesis (Lawton 1983), plants with larger life forms (such as trees) have a more complex architecture due to the higher number of branches and leaves. Hence, they may harbour greater richness and abundance of insects with more sites for feeding and oviposition. Previous studies have reported positive effects of plant size on herbivore species richness and abundance (Ribeiro et al. 2005;Campos et al. 2006;Neves et al. 2013;, or in the distribution of herbivory within plants (Ribeiro and Basset 2007;Pereira et al. 2016;Boaventura et al. 2018). In fact, global herbivory patterns have shown higher levels of insect damage on trees compared to shrubs (Kozlov et al. 2015).
Under the resource availability hypothesis-RAH (Coley et al. 1985), resource allocation for plant defence is driven by the combination of plant growth rates and habitat quality (Coley et al. 1985;Gianoli and Salgado-Luarte 2017). Plant life forms occurring in fertile soils are capable to cope with some level of herbivory, without the associated negative impacts in fitness or growth (Coley et al. 1985;Lau et al. 2008;Lynn and Fridley 2019), mainly due to low energy cost for tissue reposition, short leaf lifespan and high return rates (Zhang et al. 2016). Life forms that are adapted to poor soils, with low resource availability-like shrubs-invest in effective defences against herbivory, as each leaf produced implies in a high-energy cost (Coley et al. 1985;Santiago and Wright 2007). Plants adapted to unfertile habitats, on the other hand, may exhibit the greatest investment in chemical defences compared to those from nutrient-rich soils, due to slower growth and the strategy of avoiding damage to valuable tissues (Herms and Mattson 1992;Fine et al. 2006;Ribeiro and Brown 2006;Poorter et al. 2010). Additionally, these plants exhibit a high C/N ratio and, thus, might have more sclerophyllous leaves than related species found on richer soils (Harbone 1980;Bryant et al. 1983, Ordoñez et al. 2009). In fact, there is a strong relationship between soil nutrient availability, plant life forms and growth strategies (Grime 1977;Gong and Gao 2019). All these factors combined affect herbivory levels experienced by plants (Olff and Ritchie 1998) and can be investigated on congeneric species that occupy contrasting habitats.
From a tropical lowland forest towards a high altitudinal grassland, there is a continuous, but remarkable change in soil conditions (Clark et al. 1999;Fine and Kembel 2011). Tropical lowlands are warmer and have high availability of light and water, accelerating decomposition and making nutrients more readily available, providing greater heterogeneity compared to any adjacent montane ecosystems from a similar geological background. There are several examples of congeneric species that occupy these contrasting habitats, providing excellent systems to study the relationships between life forms and plant characteristics whilst controlling for phylogeny (Silvertown and Dodd 1996;Sultan 2000).
The present study investigated the effects of soil nutritional quality on functional characteristics and on herbivory levels of congeneric plants of different life forms across a gradient of edaphic conditions. These are closely related species (all belonging to the genus Byrsonima sp. Rich. ex Kunth-Malpighiaceae) occurring along a gradient of decreasing soil resource availability, from a semi-deciduous tropical lowland forest towards two montane physiognomies of highlands grasslands (campo rupestre sensu Silveira et al. 2016). We tested the hypothesis that soil nutrient availability affects the functional characteristics and herbivory levels of the different plant life forms. We predict that tree species that occupy forest habitats in soils with higher nutrient availability would exhibit complex architecture, higher nutritional quality, lower sclerophylly and higher levels of herbivory than structurally simpler shrub and sub-shrub species occurring in poorer habitats.

Study sites
The study was carried out in sites located in the upper and mid Doce River Basin, in Minas Gerais State, Brazil (Fig. 1). This is an 83,400-km 2 basin, the third largest in the State. We sampled in three State Parks: Rio Doce (PERD), Itacolomi (PEIT) and Serra de Ouro Branco (PESOB). The PERD (19°45 0 S 42°38 0 W) represents the largest remnant of semideciduous Atlantic Forest of Minas Gerais with an area of approximately 36,000 ha and altitude ranging between 200-500 m. The soils in the region are mainly Ferralsols according to world reference base for soil resources (IUSS Working Group WRB). According to Köppen, the climate is Aw, which alternates between a rainy and a dry season (Alvares et al. 2014). The average annual temperature is 23°C, yearly relative air humidity is 75%, with annual average precipitation of 1,500 mm.
The PEIT (20°22 0 30 00 S and 43°22 0 30 00 W) and PESOB (20°31 0 S, 43°41 0 W) are located in the southern portion of the Espinhaço Range in the state of Minas Gerais. The two parks have an area of approximately 7,000 ha, with an altitude varying between 800 to 1772 m a.s.l. Soils are classified as Leptsols according to World Reference Base for Soil Resources (IUSS Working Group WRB 2015). According to Köppen, the climate is Cwb, which alternates between a rainy and a dry season (Alvares et al. 2014). The average annual temperature is 18°C, yearly relative air humidity is 79%, with annual average precipitation of 1800 mm. The PEIT predominant vegetation communities are distributed in quartzitic and ferruginous rocky fields, formations of campo rupestre, montane forest and natural monodominant stands of Eremanthus erythropappus (Asteraceae), a pioneer tree (Fujaco et al. 2010). The PESOB vegetation is characterized by campo rupestre in highlands and, in the lower portions of the altitudinal gradient by riparian forests and forest patches embedded inside a grassland ecosystem (Instituto Estadual de Florestas, 2015).
Considering the three areas sampled, this study was conducted along a gradient of resource availability, from a habitat of poor white sand soils (PESOB), shallow rocky outcrops (PEIT) towards deep and nutrient-richer soils (PERD).

Plant species
The Byrsonima is one of the most important genera of the Malpighiaceae due to its large number of species, with 135 described species restricted to tropical and subtropical regions (Anderson 1977;Davis and Anderson 2010). The genus occurs throughout the neotropical regions and in Brazil in the phytogeographical domains of the Amazon, Cerrado, Caatinga, Atlantic Forest and Pantanal (Mamede and Francener 2015). Byrsonima sericea DC. is a tree species with wide distribution (Mamede and Francener 2015). In our study system, we sampled this species in the forest habitat of the PERD, specifically in ecotonal habitats between forests and natural lakes, where its crowns grow branched towards the lakes in search for light. Byrsonima variabilis A. Juss. is a shrub species, endemic to the Brazilian mountains highlands (Mamede and Francener 2015). This species population in the PEIT was sampled in habitats of rocky outcrop in campo rupestre. Their crowns have cylindrical shapes with branches not far apart and small internodes. Byrsonima subterranea Brade & Markgr. is a sub-shrub species that occurs in campo rupestre Cerrado (latu sensu) and Amazonian Savanna (Mamede and Francener 2015). The individuals of this species in PESOB were studied in campo rupestre, in sandy quartzite soil, called white sand soil. They grow with subterranean stems and only the leaves and reproductive parts are aerial, hence not forming a typical crown.

Sampling design
Each species of Byrsonima was considered here as a different life form. We sampled three patches of B. sericea (tree) in PERD, three patches of B. variabilis (shrub) in PEIT and three patches of B. subterranea (sub-shrub) in PSOB. In each patch, we sampled 15 individuals, totalling nine plant patches and 135 individuals sampled throughout the three sites. All data were collected between January and March 2016, at the end of the rainy season, enabling the evaluation of the accumulated herbivory during the last rainy season.

Soil nutrients
For soil characterization, one sample was collected at every three individual plants, totalling five samples per population. We chose to collect the soil at 10 cm depth due to its relationship to vegetation characteristics, as suggested by Ruggiero et al. (2002). The soil parameters analysed were pH, organic matter (OM) and total concentrations of nitrogen (N), phosphorus (P), potassium (K), calcium (Ca), magnesium (Mg) and aluminium (Al). All nutrient analyses were performed according to Defelipo andRibeiro (1997), Embrapa (1999), and Raij et al. (2001), at the Laboratory of Plant Analysis of the Department of Soils of the Agricultural Sciences Centre of the Federal University of Viçosa, Minas Gerais.

Plant architecture
Due to structural differences amongst species, we adjusted the architectural complexity measurements for the three species, allowing for direct comparisons between their different life forms. The following parameters were measured: total height (cm), trunk diameter below the first branch (cm), number of growth units and mean leaf number (see Pérez-Harguindeguy et al. 2013). Growth units were defined as the set of terminal branches of each ramification from which the leaves originated (see Bell et al. 1999).

Leaf nutrients and sclerophylly
To quantify leaf nutrient content, a set of 20 mature, fully expanded leaves of each individual were collected and oven dried to quantify total N, P, K, Ca and Mg content. Nitrogen was quantified through sulphuric digestion and the quantification of K, Ca and Mg was performed through nitro-perchloric digestion (Meneghetti 2018). Leaf sclerophylly was measured as leaf thickness (Choong et al. 1992), using a Digimess digital micrometer with an accuracy of 0.001 mm. This method is correlated with other methods of measuring sclerophylly, such as dry weight, fibre content and protein content (Choong et al. 1992;Pérez-Harguindeguy et al. 2013).

Leaf area, density and herbivory
Leaf samples were taken using a random delimited volume within each crown. For such, we used three hollow cubes, thus combining the canopy cylinder method (Ribeiro and Basset 2007) with the subplot frame model (Shaw et al. 2006) to sample leaves and quantify herbivory levels. This method was selected to avoid biased sampling, commonly documented in herbivory studies (Kozlov et al. 2015). The cube volume was defined to control differences in the size of each species and to fit the approximately area occupied by the foliage of a growth unit. In B. variabilis, three cubes of 20 cm of length, width and height were used, with a total volume of 0.06 m 3 . In B. sericea three cubes of 30 cm of length, width and height, with a total volume of 0.09 m 3 were used for sampling. Each cube was positioned in three different branches (chosen at a distance of 5 m from the plant to avoid bias) of the crown of each individual and all leaves inside the cube were collected (Fig. 2). Due to its small size, the cube was not used in B. subterranea and all leaves of each plant were collected.
To quantify herbivory, all mature and fully expanded leaves per individual were counted and digitized. Leaf abundance per plant was registered as the total number of leaves sampled using the cube method, and average leaf area and the percentage of leaf damage by chewing insects (in cm 2 ) were calculated on digitized leaf images using Image JÓ 1.6.0 software. Herbivory was determined in percentage as H = (leaf area lost/total leaf area) *100.

Statistical analyses
We tested whether each site differed for soil nutrients using an analyses of variance (ANOVA) for each element, considering each environment as a level of a simple fixed factor. The functional traits were explored to collinearity, and in cases when variables were positively correlated (Pearson r [ 0.70), only the variable with the greatest biological meaning for the hypotheses tested was used. Normality was tested for the models constructed based on error distribution. Non-significant explanatory variables were eliminated from analyses to obtain an adequate minimum model (Crawley 2013).
To test the hypothesis that plant functional traits and herbivory vary along the resource availability gradient, a multiple regression analysis was performed. The functional traits and herbivory were used as response variables and soil nutrients as explanatory variables. To test whether functional traits affect herbivory levels in Byrsonima sp., an analysis of covariance (ANCOVA) was performed with quasibinomial error distribution. Species and environment were considered as fixed factors, plant patches as nested randomized blocks and architectural parameters, leaf nutrients and sclerophylly as covariates. The average leaf area lost by each individual plant (herbivory levels) was the response variable. Statistical analyses were performed in R Environment v. 3.2.0 (R Development Core Team 2015) and ANOVA, multiple regression analysis and ANCOVA were conducted using generalized linear models (GLM).

Functional traits
The three species differed markedly in functional traits (Tab. 1). B. sericea is a mid-size tree, 5 m tall, its growth units were well-spaced, long and exhibited a high density of leaves (482 ± 218.5 leaves). B. variabilis is a 1.5-m high shrub with a round canopy, growth units short and close to each other and high leaf density (300 ± 134.9 leaves). B. subterranea is a low shrub (0.28 cm), with the leaves near the ground and very individualized and short growth units. In this species, the entire stem remains below ground and only a small part of the terminal branch-from which few leaves emerge (20 ± 15.9 leaves)-is exposed aboveground.
A total of 36,050 leaves distributed amongst the three Byrsonima species were sampled. B. sericea was the species with the greatest architectural complexity, Fig. 2 Schematic drawing of the cube methodology for herbivory sampling. Each cube was positioned in random parts of the crown to avoid collection bias. Art: Glória R. Soares  Fig. 3 Concentration of soil chemical elements along the environmental gradient. A Nitrogen, B phosphorus, C magnesium, D organic matter and E aluminium. The shades of grey of the boxplots represent differences between habitats being the taller (F 2 , 132 = 512.28, P \ 0.001), with larger trunk diameter (F 2 , 132 = 244.47, P \ 0.001), greater number of growth units (F 2 , 132 = 69.34, P \ 0.001) and leaf density (F 2 , 132 = 110.19, P \ 0.001), followed by B. variabilis and B. subterranea (Table 1). Plant height, trunk diameter and growth units were highly correlated (r Pearson [ 0.70) and we used only leaf density and number of growth units as explanatory variables influencing herbivory levels on the plants.

Discussion
We found strong evidence that the variation of functional characteristics is influenced by soil nutrient content and that, consequently, these functional characteristics affect herbivory levels experienced by plants. In the present study, the average rates of foliar consumption by insect herbivores were generally low compared to the global patterns of herbivory, which show an average of 4.5% of leaf area lost (Kozlov et al. 2015). The tree species here studied exhibited the highest level of herbivory amongst the three species occurred in a high-productivity forest habitat and with the highest nutrient content in its leaves. The shrub and sub-shrub life forms, on the other hand, occupied oligotrophic habitats and were less consumed by herbivorous insects. These results indicate different strategies of resource use and acquisition (Wright et al. 2004). Tree species have characteristics (such as many growth units and high leaf nutrient content) associated with the ability to quickly capture light and nutrients (Ordoñez et al. 2009;Poorter et al. 2010;Reich 2014).
On the other hand, the shrub and sub-shrub species exhibit a conservative strategy, as they have functional characteristics (like sclerophylly) linked to tissue protection (Díaz et al. 2004;Reich 2014). This continuum amongst acquisitive and conservative strategies can be mediated by soil nutrients (Hernández-Vargas et al. 2019). Our results indicate that soils with higher concentration of macronutrients and organic matter (i.e. more productive soils) positively affect plant height, growth units, leaf nutrient concentration and leaf density. In contrast, these richer soils lower the levels of plant sclerophylly. Soils with high concentrations of toxic elements-such aluminium-on the other hand, were associated to increased sclerophylly (Feller 1995;Brady et al. 2005;Ribeiro et al. 2016). Consequently, growth units, leaf nutrient concentration and leaf density positively affected herbivory levels, whereas sclerophylly was inversely related to the amount of tissue removed by insects. Similar results were found by Lynn and Fridley (2019) suggesting an effect of soil on functional characteristics and consequently on plant herbivory. The three studied species of Byrsonima occurred along a soil nutritional gradient, from the lowland forest-the most fertile habitat-to the campo rupestre, a comparably poorer habitat. Life form and architecture here described are probably consequences of adaptive demands linked to abiotic environmental conditions, as suggested by Grime (1977) and Crawley (1997). The number of growth units, leaf density and higher leaf nutrient concentration in tree species are typical characteristics of plants from productive environments (Wright et al. 2004). The smaller number of growth units of shrubs and sub-shrubs may be related to a trade-off in biomass investment amongst resources for growth and protection of plant tissues. Both species occurred as a shrub or sub-shrub life forms and are highly sclerophyllous, compatible with what is expected for species occurring in the campo rupestre (Ribeiro et al. 1999;Negreiros et al. 2014;Silveira et al. 2016). Therefore, the functional traits of the plants respond to the gradient of resource availability of the habitats, and consequently herbivory levels follow this gradient, supporting our hypothesis.
Our results also corroborate the Resource Availability Hypothesis (Coley et al. 1985;Lau et al. 2008) and the plant architecture hypothesis-PAH (Lawton 1983). The crown architecture (measured in our study by the number of growth units) has a strong relationship with herbivory levels Pereira et al. 2016). Some previous studies have shown, for example, that the size of the plant positively influences the availability of niche for the survival and feeding of herbivores (Hannunen 2002;Campos et al. 2006;Randlkofer et al. 2009). In addition, our results are in line with the Plant Apparency Hypothesis, which predicts that larger plants are more likely to be found by herbivorous insects (Feeny 1976;Rhoades andCates 1976, Smilanich et al. 2016). Our data corroborate other studies that found the positive effect of apparency and herbivory (Castagneyrol et al. 2013;Strauss et al. 2013;Smilanich et al. 2016). Our results, however, indicated a positive relationship between number of growth units and herbivory only for shrubs compared to sub-shrubs. This relationship between apparency and levels of herbivory can be even more important in open environments, such as the rupestrian field, where visible plants are even more easily found by herbivorous insects.
Leaf nutrient content also influenced positively herbivory levels in Byrsonima species, as previously reported (e.g. Casotti and Bradley 1991;Meloni et al. 2012;Silva et al. 2015). Soil nutrient availability and leaf nutritional content have been related to the production of simple and low-cost tissues, which facilitates insect feeding (Coley 1983;Coley et al. 1985;Price 1991), and may clearly affect plant selection by insects (Mattson 1980). We did not find a relationship between sclerophylly and herbivory levels, contrary to other studies (e.g. Pennings and Paul 1992;Ribeiro and Basset 2007;Malishev and Sanson 2015), although sclerophylly in tree species was the lowest. However, Byrsonima shrub and subshrubs exhibited high sclerophylly, as commonly expected for species that occur in the campo rupestre (Negreiros et al. 2014). We attributed this result to the fact that the campo rupestre is a geologically old biome, accumulating 65 million years of species adaptation and speciation (Silveira et al. 2016), where high sclerophylly is expected to be a default trait (Ribeiro et al. 1999). It is likely that herbivorous species have had time to evolve strategies to deal with sclerophylly in these habitats. Similar results were reported by Meloni et al. (2012), indicating that concentrations of defensive compounds in Cerrado plants did not inhibit herbivory, probably due to a very long period for adaptation. Therefore, data suggest that as herbivores have already overcome the barrier of leaf toughness, they can choose the most nutritious plants, regardless of the levels of sclerophylly. It is important to acknowledge, however, that sclerophylly in our study was measured as leaf thickness, and other leaf traits, such as leaf specific area and leaf specific mass, should also be addressed as potential drivers for the levels of herbivory in plants.
Our data have shown that species along a gradient of soil nutrients have functional characteristics influenced by habitat, ultimately affecting herbivory levels. Differences in herbivory levels amongst species of distinct life forms can be explained by functional characteristics, which are an adaptive response to habitat type. By comparing species from different life forms but within the same genus, along a stressgradient of continuous habitats, we described a corresponding gradient of plant tissue consumption by herbivorous insects. The confounding effects between habitat resource availability and plant life form were coherent with theoretical predictions. Data amongst species, however, is suggestive of evolutionary filters that may further constrain herbivores in high-altitude grasslands.