Are ants attracted to herbivorized leaves of Caryocar brasiliense Camb. (Caryocaraceae)?

Abstract Different profiles of secondary compounds are released by plants after herbivore attack. Many of these compounds are used by predators and parasitoids to locate herbivores that are damaging leaves. Such an induced indirect defense was tested with the Caryocar brasiliense-ant system in the dry season, when C. brasiliense has old leaves, and in the rainy season, when C. brasiliense has new leaves. A total of 20 plants were analyzed per season. Two opposite leaves of the same branch were selected for each plant. Approximately 40% of the area of one leaf was removed (treatment leaf) while the other leaf remained intact (control). The number of ants that visited each leaf was counted simultaneously for a period of 15 minutes and the mean difference in ant number was tested by paired t-test. The mean number of ants differed significantly between treatment and control only in the rainy season (t = 3.004, df = 19, p = 0.007). This finding suggests the presence of induced defense in this system only when the leaves are young. The study supports the Optimal Defense Theory since young leaves of C. brasiliense with artificial damage attracted significantly more ants than leaves without damage and represents the first evidence of an induced defense mechanism in the C. brasiliense-ant system.


Introduction
Plant defense strategies against herbivores can be categorized as either direct or indirect. Direct defenses negatively affect herbivore metabolism via toxic compounds and digestibility reducers (Speight et al. 1999, Marquis 2012 whereas indirect defenses occur when the plant attracts other insects that can reduce herbivore damage (Bixenmann et al. 2013). In the latter strategy, the plant provides chemical cues or rewards (food and/or shelter) to attract predators and parasitoids of the herbivores (Boege & Marquis 2005).
Leaves naturally release volatile organic compounds (VOCs) during their development. However, when damage occurs, usually due to herbivores, some plants quantitatively and/or qualitatively change their VOC profile (Paré & Tumlinson 1999, Marquis 2012. Such new VOC profiles may inhibit oviposition or reduce herbivore performance (induced direct defense), or act as a cue for predators and parasitoids to locate herbivores that are causing damage (induced indirect defense) (Turlings et al. 1995, Paré & Tumlinson 1999, Carroll et al. 2006, Heil 2014. Kost & Heil (2006) demonstrated that VOCs released by damage plants could be recognized by undamaged neighboring co-specific plants, acting as a trigger for increased secretion of extrafloral nectar.
Caryocar brasiliense Camb. (Caryocaraceae), known locally as "pequizeiro", is a typical tree of the Brazilian Cerrado. It is a semideciduous plant that loses some of its leaves during the dry season (May to July) (Araujo 1995), and grows young leaves during the beginning of the rainy season (September to March) (Vilela et al. 2008). Flowering starts soon after the expansion of young leaves, with fruiting occurring from October to February (Vilela et al. 2008).
Although several studies have verified the importance of ants as a constitutive defense in C. brasiliense (e.g., Oliveira 1997, Oliveira & Freitas 2004, the existence of an induced indirect defense in this system has not been tested. This study aimed to determine if this type of defense occurs in the C. brasiliense-ant system. Specifically, the following questions were addressed: (1) do C. brasiliense leaves with artificial damage attract more ants than leaves without damage? (2) Are young leaves more attractive to ants than mature leaves?

Material and Methods
Experiments were conducted in Emas National Park (ENP) located in the state of Goiás, Brazil (17°49'-18°28'S; 52°39'-53°10'W) during the dry (July 2012) and rainy (September 2012) seasons with 20 Caryocar brasiliense plants per season. The plants had mature leaves and considerable herbivory during the dry season (Figure 1a), while EFNs were active and young leaves did not have signs of herbivory during the rainy season (Figure 1b).
A total of 40 C. brasiliense trees of similar height, architecture and herbivory level were selected for study. One branch with fully expanded leaves and minimal herbivore damage was selected in each plant, on which two opposite leaves were identified. Approximately 40% of the leaf area of one leaf was removed with a scissors (treatment leaf, with artificial herbivory), while the other leaf remained intact (control leaf, without artificial herbivory). Both leaves were shaken prior to the start of the experiment to assure no ants were present.
The number of ants visiting paired leaves was counted simultaneously for period of 15 minutes. Ants that were observed visiting the experiment leaves were manually collected with a forceps, preserved in 70% alcohol, and identified to morphospecies. Difference in mean number of ants was tested by paired t-test, considering plants as sample units.

Results
Nine morphospecies of five subfamilies of ants occurred on the leaves of C. brasiliense during our study. Three of the morphospecies occurred only in the dry season, two only in the rainy season and four in both seasons. Different morphospecies of ants patrolled the experimental plants, but in most cases only one morphospecies was found on a plant. Camponotus was the most diverse genus with three morphospecies (Table 1).
The Optimal Defense Theory (ODT) predicts that resource allocation for defense should be higher in the most valuable plant parts (measured in terms of reduction in fitness by its removal) and parts more likely to be damaged by herbivores (Mckey 1974, Zangerl & Bazzaz 1992. Many studies have found young leaves to have higher concentrations and a greater diversity of secondary compounds than mature leaves (e.g., van Dam et al. 1995, Read et al. 2003, which supports ODT. Radhika et al. (2008), for example, applied jasmonic acid to castor bean leaves and found that the release of volatiles and extrafloral nectary secretion was higher in young than in old leaves. The present study also supports ODT, since only the young leaves of C. brasiliense with artificial herbivory attracted significantly more ants than leaves without artificial herbivory. Likewise, Pacheco & Del-Claro (2018) also detected a higher number of Pseudomyrmex concolor ants on young damaged leaves of Tachigali myrmecophila in the Amazon.
Although the present study did not analyze the composition and concentration of leaf compounds, the response of ants (higher recruitment in damaged young leaves) indicates chemical signaling by the plant after damage, suggesting that induced defense is only evident when leaves are young. The present results may be an underestimation because of the type of artificial herbivory employed (damage by cutting with scissors), because some studies have shown that natural herbivory elicits greater defense induction than mechanical damage (Hartley & Lawton 1987, Massey, Ennos & Hartley 2007, Quigley & Anderson 2014. Oral secretions provide herbivore-specific cues for defense induction by many insects (Alborn et al. 1997, Tian et al. 2012. Components of insect saliva, plant cell wall fragments and other cues create a signal cascade that triggers a defense response that increases concentrations of secondary metabolites (Stam et al. 2014). While counting ants in the present study, many were observed touching the region where the leaves had been cut with their antennae and/or mandibles, suggesting that liquid substances, in addition to volatile compounds, may also be acting as signals for the ants.
Plants of C. brasiliense are constantly visited by ants that feed on extrafloral secretions of flower buds and young leaves (Oliveira & Brandão 1991). Several studies have shown that ants reduce herbivore damage to C. brasiliense (e.g., Oliveira & Freitas 2004), thereby increasing plant fitness (Oliveira 1997). The observations of the present study suggest that there is induced defense in young leaves of C. brasiliense and represent the first evidence of an induced defense mechanism in the C. brasiliense-ant system. Additional studies, including quantitative and qualitative chemical analyses, are recommended to further investigate the role of this type of defense in C. brasiliense. Additional research on VOCs in other non-myrmecophytic systems are needed. Such studies could lead to greater insight into whether particular chemical traits are associated with the evolution of ant-plant relationships. There was no significant difference in the mean number of ants between treatment and control leaves in the dry season (t = 1.011, df = 19, p = 0.325; mean = 5.75 and 4.20, respectively) ( Figure 2a), but there was significantly more ants on treatment (mean = 4.15) than control (mean = 2.10) leaves in the rainy season (t = 3.004, df = 19, p = 0.007) (Figure 2b).

Discussion
Although the mean number of ants on treatment leaves was greater than that of control leaves in both seasons, the difference was only significant in the rainy season, suggesting induced defense in this system only when leaves are young.
Herbivory of young leaves tends to have a greater effect on plant fitness than does herbivory on mature leaves (Jurik & Chabot 1986). In addition, young leaves have not yet contributed enough in terms of photosynthetic production to offset the high costs of their construction (Radhika et al. 2008). Because young leaves are softer and more nutritious than mature leaves they tend to be more attractive to herbivores (Kursar & Coley 2003). For example, Coley & Barone (1996) found that 70% of leaf herbivory of rainforest plants occurs during leaf expansion.