Water content , fibres , and herbivory in leaves of two distinct and adjacent tree communities of the Brazilian Atlantic Forest

(Water content, fibres, and herbivory in leaves of two distinct and adjacent tree communities of the Brazilian Atlantic Forest). The aims of the present study were to verify whether plants from Mussununga differ from those of Mata Alta in relation to herbivory, water content, and fibres, and to test whether plants with lower leaf water content and higher fibres are less attacked by chewing insects. The results indicated that leaf herbivory percentages and fibres were similar between the two communities, although individuals from Mata Alta showed higher leaf water content than individuals from Mussununga. Thus, the herbivory patterns found for these tree communities do not seem to not be influenced by water content and fibres. Other factors (e.g., chemical defenses and/or herbivores) may be acting as main drivers.


Introduction
Herbivory exerts an important selective pressure on plants (Coley et al. 1985) and can positively affect leaf growth and leaf production (Nascimento & Hay 1993, Nascimento & Hay 1994).However, high herbivory rates generally have negative effects on the growth, reproduction, and competitive ability of plants, consequently affecting plant community composition (Crawley 1983, Coley & Barone 1996, Marquis 2004).
Herbivory pressure allows for the evolution of different defense mechanisms (Marquis 1984, Coley et al. 1985), including physical/mechanical defenses, such as thorns (Cooper & Ginnett 1998) and nonglandular trichomes (Baur et al. 1991, Rodrigues et al. 2012, Fürstenberg-Hägg et al. 2013), and chemical defenses, such as the production of a series of toxins and defense proteins that target the physiological processes of insects, in an attempt to repel or intoxicate herbivores (Coley & Barone 1996, Mello & Silva-Filho 2002, Gullan & Cranston 2007, Trigo et al. 2012).In addition to defensive mechanisms, the nutritional quality of leaf tissues also affects the feeding of herbivorous insects (Coley 1987, Marquis 2012, Cárdenas et al. 2014), while low water and nitrogen content in leaves may lead to a decrease in the preference and performance of chewing insects (Basset 1991, Loaiza et al. 2011, Marquis 2012).
In resource-limited environments, slow-growing species invest more resources in producing immobile defenses, such as lignin or total phenols, displaying lower herbivory rates (Coley 1983, Coley et al. 1985, Endara & Coley 2011).Tree species adapted to these environments exhibit lower capacity to absorb nutrients and to perform photosynthesis (Chapin 1980).Thus, for these plants, it is probably more costly to replace lost parts than to defend them (Coley et al. 1985, Fine et al. 2004, 2006, Agrawal 2006, Fine & Mesones, 2011).Although, many studies dealing with herbivory in the Atlantic rainforest are available, most are related to the community level or seed/fruit predation (Souza et al. 2013, Rossetti et al. 2014;Galetti et al. 2015, Corrêa et al. 2016).Leaf herbivory studies considering functional characteristics of Atlantic forest trees in contrasting chemical and physical soils have not yet been performed.Therefore, the seasonally dry forest in the Northern area of the state of Espírito Santo, Brazil, can be considered an ideal forest formation to assess the relationship between herbivory, nutritional aspects and chemical defenses, because it displays two distinct and peculiar phytophysiognomies under the same climatic condition: the Mussununga, with sandy soils and lower diversity, and the Mata Alta, with clayey soils and higher floristic diversity (Simonelli et al. 2008).Mata Alta soils and humus are more nutrient rich (with a lower C/N ratio) than Mussununga soils (Garay et al. 2003(Garay et al. , 2016)).Myrtaceae is recognised as one of the most important families in both types of forests, both in terms of the number of species and individuals (Jesus & Rolim 2005, Simonelli et al. 2008, Giaretta et al. 2016).Furthermore, certain species of this family produce secondary compounds, such as terpenoids and phenols (Cooper 2001, Chaieb et al. 2007, Keszei et al. 2010) and display physical defenses, like leaf hardness, which provides them not only with chemical resistance, but also mechanical resistance to leaf-chewing insects (Sanson et al. 2001).In this context, we expected to find that Mussununga vegetation plants, with lower water content and higher fibre content, should therefore be less attacked by herbivorous insects compared to Mata Alta plants, which have higher water content and lower fibres.Only plants of the Myrtaceae family were investigated, due to their representativeness and the abundance of individuals in the studied areas.The use of plants from the same family also allows for control regarding phylogenetical aspects.

Material and methods
The study was performed in two areas with distinct physiognomies located within the Vale Natural Reserve (Reserva Natural Vale -RNV) in Linhares, state of Espírito Santo, Brazil, between the parallels 19º05' and 19º18'S and meridians 39º45' and 40º19'W.A 1-ha permanent plot was applied in two areas: the Mata Alta (19°09'14.2"Sand 040°02'0.004"W)and Mussununga forests (19°09'078"S and 040°02'197"W).These areas were selected considering their different soil characteristics (clay soil at Mata Alta and sandy soil at Mussununga) and physiognomies (closed canopy at Mata Alta; open canopy at Mussununga).Mata Alta is more representative of the lowland forest and corresponds to approximately 70% of the total RNV area, of which only 8% is occupied by secondary forest.It is characterised by closed canopy trees, reaching up to 40 m in height, and a poorly developed understory.Its soils are classified as clayey or clayeysandy (Peixoto et al. 2008, Rolim et al. 2016b), being dystrophic yellow podzol or yellow argisol (ultisol), and are characterised by differences in grain size with depth and low fertility (Garay et al. 2003).
The Mussununga forest is surrounded by the Mata Alta forest, forming an enclave within the latter (Peixoto et al. 2008, Rolim et al. 2016b), and occupies 7.9% of the RNV area (Peixoto et al. 2008).It is composed of trees with less dense canopies, allowing higher light penetration to the soils.Soils are classified as hydromorphic podzol with a sandy texture, with a shallow water table that rises to the surface during certain times of the year (Garay et al. 2003, Simonelli et al. 2008, Saporetti-Junior et al. 2012, Rolim et al. 2016b).The trees reach approximately 10 m in height, although some emergent species can grow up to 20 m (Simonelli et al. 2008, Rolim et al. 2016b).This forest displays low species richness in comparison to Mata Alta and has a lower abundance of xerophile elements, lianas and bromeliads, with characteristics similar to those of the restinga forest (coastal sandy-soil forests) (Garay et al. 2003, Simonelli et al. 2008).
Mata Alta shows higher species diversity (H' = 5.04 nats/sp; DBH ≥ 10 cm) when compared to Mussununga (H' = 3.36 nats/sp; DBH ≥ 5cm).The Shannon diversity index for Mata Alta is one of the highest ever recorded in the Atlantic Forest (Jesus & Rolim 2005).A similarity analysis using the Jaccard index based on species lists published by Peixoto et al. (2008), Simonelli et al. (2008), and Jesus & Rolim (2005) indicates that the similarity between the tree species of Mata Alta and Mussununga is low (SJ = 0.067).However, at the family level, Myrtaceae shows high representativeness in both physiognomies, being the family with the highest number of species in both Mata Alta (14%) (Jesus & Rolim, 2005) and Mussununga (24%) (Simonelli et al. 2008).
The soils that dominate the Mata Alta and Mussununga regions are, in general, low fertility homogeneous soils, with a sandy surface horizon unfavourable for nutrient retention.Therefore, organic matter originating from the vegetation may be essential for the maintenance of soil structure and fertility (Garay et al. 2003).Total carbon in the hemiorganic A12 horizon of the soil in the RNV is 0.72% at Mata Alta and 1.16% at Mussununga, whereas N concentrations are 0.08% at Mata Alta and 0.07% at Mussununga, with a C/N ratio of 8.9 for Mata Alta and 16.9 for Mussununga (Garay et al. 2003).
The climate in the region is classified as type Aw according to the Köppen climate classification (Peixoto et al. 2008), i.e. hot and humid, marked by seasonality, with a wet season in the summer (October to March) and a dry season in the winter (April to September).The mean annual rainfall is of 1.227 mm year -1 , with mean temperature ranging from 20 °C to 26 °C (Rolim et al. 2016a).The native vegetation in the state of Espírito Santo is generally classified as being a lowland dense rainforest (RADAMBRASIL 1987).However, some studies have analysed the vegetation based on functional traits related to seasonal aspects and have categorized it as being a lowland seasonal semideciduous forest or semideciduous rainforest (Peixoto & Gentry 1990) or lowland seasonal evergreen forest (Rolim et al. 2016a).
Tagged Myrtaceae trees of the understory between 8 and 10 m in height were randomly selected for leaf collection in the 1-ha plots of each forest type, regardless of species.Six to 10 mature leaves from each selected tree in both areas were collected in August 2010 and February 2011 for subsequent herbivory and functional traits analyses.Mature leaves were considered as being the leaf below the fourth leaf counting from the apex to the base of the branch, being dark-green in coloring (Nascimento & Proctor 2001, Nascimento at al. 2011).The total leaf area consumed by herbivores (leaf-chewing insects) was visually estimated using a system of seven classes of percentile damage (1 = absence of damage by herbivory; 2 = ]0,5]; 3 = ]5,10]; 4 = ]10,25]; 5 = ]25,50]; 6 = ]50,75]; and 7 = ]75,100]).The mean herbivory rate for each tree was calculated by multiplying the number of damaged leaves within each herbivory class by the mean interval of classes into which the leaves were classified, and subsequently dividing the result by the total leaf samples (Nascimento & Proctor 2001, Nascimento at al. 2011).
Percentage of herbivory = Σ(ni x i)/N Where, Ni: number of mature leaves within the herbivory class; I: mean interval of the herbivory class; N: total number of sampled mature leaves.
The collected leaves (6 to 10) were separated from the branch, placed in plastic bags and Styrofoam boxes and taken to the laboratory, where they were weighed using a semi-analytical scale.The leaves were subsequently left in a forced-air oven at 105 °C until reaching constant weight.Water content was calculated by finding the difference between the dry weight and the initial weight and multiplying the result by 100, to obtain percentage values (Allen 1989).
The chemical analyses for the determinations of the neutral detergent fibre (NDF), acid detergent fibre (ADF), cellulose (CEL) and lignin (LIG) were performed by sequential analysis at the Laboratory of Animal Science and Animal Nutrition (Laboratório de Zootecnia e Nutrição Animal -LZNA) of the Agricultural Science and Technology Centre/State University of the North of Rio de Janeiro (Centro de Ciências e Tecnologias Agropecuárias/Universidade Estadual do Norte Fluminense Darcy Ribeiro, CCTA/ UENF), following the method described by Van Soest et al. (1991), using 0.5 g of the dry and pulverised plant material.Lignin content was determined by the 72% sulphuric acid method, followed by incineration of the obtained residue at 550 °C for 3 h.Cellulose content was determined by calculating the difference between ADF and lignin.The NDF includes cellulose, hemicellulose, and lignin as its main components, in addition to heat-damaged proteins, cell wall proteins, and minerals (ashes).In contrast, the ADF consists primarily of cellulose and lignin (lignocellulose) and heat-damaged proteins, with small amounts of cell wall proteins and insoluble minerals (ashes).
Description of the statistical analyses -The Mann-Whitney U-test was used to compare herbivory rates and water and fibre content median values between the investigated areas.The analyses were not performed to the species level because of the low sampling of species common to both areas.
A linear regression analysis was performed to assess the relationship between herbivory, fi bre concentrations and water content in the sampled from Myrtaceae species from each forest type.All analyses were performed using the R software package, version 3.2.4revised ( 2016).

Results and Discussion
The percentage of leaf areas lost to herbivores did not differ between Mata Alta and Mussununga (Mann-Whitney U-test, p = 0.29), with mean values ranging from 8 to 11% (table 1).These values are close to the upper limit recorded for humid tropical forests (ca.15%), evaluated using a similar methodology (Sterck et al. 1992, Nascimento & Proctor 2001, Cárdenas et al. 2014).Cases of individuals with extreme leaf areas lost by herbivory (> 25%) were rare (less than 2% of trees) in both sampled areas (fi gure 1).Signifi cant differences were observed between Mata Alta and Mussununga Myrtaceae species relative leaf water (Mann-Whitney U-test, p = 0.03), with leaves from the former showing signifi cantly higher values compared to leaves from the latter (table 1), as expected, since Mussununga has a sandy soil (Garay et al. 2003(Garay et al. , 2016)).
The analysed fi bre concentrations (NDF, ADF, CEL and LIG) did not differ signifi cantly between the areas (Mann-Whitney U-test, p> 0.05) (table 1).Although fi bre contents in leaves may act as a barrier regarding the use of these leaves as a food source (Coley, 1983;Clissold et al. 2009, Cárdenas et al. 2014), the present study did not corroborate this statement, since simple linear regressions performed between herbivory and fi bres (ADF, NDF, CEL and LIG) indicated no signifi cant relationship between herbivory and any of the variables (fi gure 2).
Plants display a combination of defense systems that involve physical and chemical barriers and/or the recruitment of predators and parasitoids that prevent or hinder herbivores from completely exploring their host plants (Rasmann & Agrawal 2009, Marquis 2012).It is well known that leaf fi bre content may act as a barrier  regarding the use of leaves as a food source against small mammals and insects (Coley 1983, Clissold et al. 2009, Cárdenas et al. 2014).Several studies have found significant negative correlations between leaf herbivory and fibre, cellulose and lignin concentrations (Coley 1983, Lacerda et al. 1986, Coley 1987, Coley & Barone 1996).Sinclair & Hughes (2008) studied 88 Myrtaceae species and observed that the presence of mining insects was positively and significantly correlated to species presenting thinner leaves and high phenol concentrations.These authors also found a marginal significance for specific area and water content, as well as no significant correlation for leaf area, hemicellulose or tenacity.In the present study, Mata Alta trees showed higher water content than trees from Mussununga, suggesting that environmental conditions such as soil characteristics are the main driving factor, as observed by Marquis et al. (2002).
The lack of differences in herbivory percentages between these forest types and the non-significant relationships between herbivory and fibre and water content suggest that attacks by herbivorous insects can be limited by other chemical leaf aspects, such as sugars and nitrogen content, among others, as also suggested by Cornelissen & Fernandes (2001).
The hypothesis that plants with lower water and/ or higher fibre content are less attacked by chewing insects was, thus, not accepted.Although Myrtaceae trees from Mata Alta had higher water content in leaves when compared to Mussununga trees, herbivory values were similar.Thus, the herbivory pattern in these forest types seems to not to be strongly influenced by fibre concentrations and relative water content, suggesting that other factors (i.e. chemical defenses and/or herbivores) might be acting as main agents.

Figure 2 .
Figure 2. Simple linear regressions between herbivory and water content neutral detergent fi bre (NDF), acid detergent fi bre (ADF), cellulose (CEL) and lignin (LIG) in the leaves of the investigated Myrtaceae species (DBH > 10 cm) in the understory of the Mata Alta (N = 16 species, a, c, e, g and i) and Mussununga (N = 10 species, b, d, f, h and j) areas located in the RNV, Linhares, Espírito do Santo State, Brazil.

Table 1 .
Mean values of herbivory, water content (WC) and fibre components (%) in leaves of Myrtaceae species from Mata Alta and Mussununga, located in the RNV, Linhares, ES, Brazil.Fibre components: neutral detergent fibre (NDF), acid detergent fibre (ADF), cellulose (CEL) and lignin (LIG).Different letters in the columns indicate significant differences between both study areas.Mann-Whitney U-test, p < 0.05.