Introduction

Plants generally incur potential costs in the production, storage, and metabolism of defense traits used to provide resistance to herbivores (Gershenzon 1994; Heil et al. 1997; Karban and Baldwin 1997; Strauss et al. 2002; Yamawo et al. 2015), and these defensive costs may divert limited resources that could be allocated to plant growth and reproduction (Heil and Baldwin 2002; Strauss et al. 2002). Accordingly, many plants have evolved “induced defenses” against herbivory to reduce the costs of expressing constitutive defenses by reducing the window of time during which these defense traits are active (Agrawal et al. 1999; Herms and Mattson 1992; Karban 2010; Karban and Baldwin 1997). Against the prominent merits of an induced defense versus constitutive defense strategy, disadvantages in the former may still occur. The most likely disadvantage is a time lag until affected plants fully activate their defenses. Until this activation occurs, plants may find themselves exposed and vulnerable to herbivory (Karban 2010; Karban and Baldwin 1997; Metlen et al. 2009). Another crucial matter is the timing to stop or “turn-off” the induced defense (i.e., relaxation timing) (De Witt et al. 1998; Karban 2010), because maintaining an induced defense may be wasteful after the attacking herbivores have been repelled. Therefore, to achieve the most effective induced defense to be most effective, plants fine-tune the defense activation as well as the decay time for the expression of their defense traits (Karban 2010). Despite its obvious importance in determining the costs and benefits of induced defenses, our knowledge of how the induction of defense traits are fine-tuned by plants in response to herbivory remains quite limited.

Plants have evolved numerous morphological and chemical defense traits (Walters 2011). Such traits can directly defend plants from herbivores (e.g., spines, trichomes, or alkaloids). Several studies have reported on systems for which the complete time course of induction, and particularly the relaxation, of direct defense traits in plants, such as needs and chemical compounds (Gomez et al. 2010; Young et al. 2003). In addition, some plants have evolved indirect defense traits those typically takes a biotic form; for example, plants bearing extrafloral nectaries (EFNs), food bodies (i.e., lipid-rich particles), and/or domatia to attract natural enemies of herbivores (Koptur 1989). Several studies have reported that the extrafloral nectar secretions and formation of EFNs may be induced by leaf damage, with the former already increasing within 24 h (Heil et al. 2000, 2001; Koptur 1989; Lach et al. 2009; Radhika et al. 2008, 2010) and completing within 6 days after herbivore damage (Ness 2003; Wäckers et al. 2001). Although the induction of EFN formation has been investigated in some plant species (Mondor and Addicott 2003; Pulice and Packer 2008; Wooley et al. 2007), the relaxation of induction of EFN formation has not yet been reported. To better understand how plants induce an indirect defense through ants, we investigated both the induction and relaxation of extrafloral nectar secretions and EFN formation.

Here, we focused on the Mallotus japonicus (L.) Muell. Arg. (Euphorbiaceae). This plant species utilizes EFNs as an indirect defense trait (Yamawo et al. 2012c). To determine how the secretion of extrafloral nectar is induced and relaxed, we grew M. japonicus seedlings and investigated the temporal variation in the extrafloral nectar volume and the number of ant workers on experimental plants whose leaves were artificially damaged. We discuss the characteristics of induced defense of indirect defense by attracting ants.

Materials and methods

Study species

We used seedlings of Mallotus japonicus (L.) Muell. Arg. (Euphorbiaceae), a pioneer plant growing in gaps and disturbed areas in temperate and subtropical regions of eastern Asia. Mallotus japonicus shows indirect defense traits against herbivores (Yamawo et al. 2012c). The plant bears EFNs on its leaf edges and food bodies (pearl bodies) on its leaf and stem surfaces as indirect defense traits (Yamawo et al. 2012c). EFNs contain primary sugars, and food bodies are lipid-rich particles (O’Dowd 1982). These indirect defense traits attract ants, which remove herbivores from the plant (Yamawo et al. 2012c, 2014).

Cultivation of young M. japonicus plants

Approximately two hundred seeds of M. japonicus were collected from 10 trees growing in Okayama Prefecture, western Japan (33°41′N, 133°55′E), in September and October 2008. Two plastic containers (45 cm wide, 35 cm long, 5 cm deep) were filled with moist, soil to 4 cm in depth, and into each container 100 seeds were sown at a depth of 1 cm. Containers were placed under a 12-h light/35 °C and 12-h dark/25 °C photoperiod for 30 days, because M. japonicus seeds germinate in response to high temperature (Washitani and Takenaka 1987). The containers and seedlings that emerged were watered every other day.

One hundred plants that had reached approximately 4 cm in height were selected on 1 April 2009. Each plant was transplanted into a plastic pot (20 cm × 20 cm × 25 cm) containing 70% tuff loam and 30% humus. These pots were placed in a greenhouse at Saga University (33°24′N, 130°29′E), where the temperature, light regime, and humidity were not regulated. Each plant had approximately 10 leaves.

Effect of leaf damage on the production of extrafloral nectar

On July 18, 20 plants were placed in a greenhouse free of any nectar-collecting insects at Saga University and randomly assigned to either the control (n = 10) or leaf damage treatment (n = 10; see damage details below). We estimated the production of extrafloral nectar from the nectar volume secreted by two EFNs on the third leaf from the apex of the plants because nectar secretion was active on this leaf (Yamawo et al. 2012c). The volume of nectar from each EFN was measured once every 24 h for 11 days, beginning on July 20. All the leaves were washed with distilled water to remove any accumulated nectar at 09:00 on July 19, and the nectar that had been secreted for 24 h was collected in 0.5-µl micro-capillary tubes (Drummond Scientific, Broomall, PA, USA). The volume was estimated by multiplying the proportion of the total tube length filled with extrafloral nectar by 0.5 µl.

Studies of leaf damage have imposed a range of damage levels, from as low as 1–2% (Heil et al. 2001) to as high as 75% (Wooley et al. 2007). Previous field studies showed that 10–80% of the leaf surface area of young M. japonicus plants is damaged by the inchworm Ascotis selenaria Butler (Yamawo et al. 2012c, 2014). Hence, we selected a damage level of 50% for our experiment. On July 21, after nectar removal, we used scissors to cut away 50% of each leaf on the 10 seedlings in the leaf damage treatment because the induction of EFN formation and extrafloral nectar secretion does not depend on specific types of herbivory (Heil et al. 2000; Pulice and Packer 2008). Seedlings in the control group were left undamaged.

Effect of leaf damage on ant attendance in the field

On July 18, we placed another set of 20 potted M. japonicus plants in the grassland site at Saga University and randomly assigned them as control and damage treatments (n = 10 each). On June 19 (after counting the ants), we cut away 50% of each leaf on the seedlings in the damage treatment with scissors as previously described; seedlings in the control group were not cut. Between June18 and June 25, we counted the number of ants visible on each plant once a day between 11:00 and 12:00 [a commonly used approach to quantify the temporal variation in ant abundance on plants (e.g., Yamawo et al. 2017)]. In some obligate ant–plant mutualism systems, volatile emissions from damaged leaves or stems may attract the ants to damaged tissue areas (e.g., Brouat et al. 2000; Mayer et al. 2008). However, our experimental system is a facultative ant–plant mutualism, so any volatile emissions from the damaged leaves did not influence ant visitation in frequency or number (Yamawo 2009). Therefore, we considered any temporal change in ant abundance on the plants as dependent on the changes in nectar volumes.

Effect of leaf damage on the formation of EFNs

On July 18, we placed 60 potted seedlings in an experimental field at Saga University and randomly assigned them to control and damage treatments (n = 30 each). At the start of the experiment on July 21, we used scissors to remove 50% of each leaf on the seedlings in damage treatment. Although following this leaf damage treatment, the plants produced approximately seven new leaves, their leaf production activity was reduced in autumn. Therefore, we consecutively numbered these seven leaves (post-leaf damage treatment) and counted the leaf margin EFNs on every leaf.

Statistical analysis

A repeated-measures ANOVA was used to compare extrafloral nectar secretion volume and number of ant workers per plant. Although this data sets follow a normal distributions, the data of each day does not follow it. Thus, the extrafloral nectar secretion volume and number of ant workers were compared between treatments on each experimental day by Mann–Whitney U test. Multiple comparisons were performed using the false discovery rate (Benjamini and Yosef 1995). The number of EFNs per leaf was analyzed using two-way ANOVA. The extrafloral nectar secretion volume, number of ant workers and number of EFNs, were compared between treatments by t test. A P value < 0.05 was considered to indicate a significant difference.

Results

Extrafloral nectar secretions

The time-sequential pattern of nectar secretions per EFN differed significantly between the control and damage treatments (treatment × day, F = 14.981, P < 0.0001; Fig. 1). The volume was significantly greater in damaged plants than in undamaged plants at 1 to 7 days after damage treatment, but not at 0, 8, or 9 days (Fig. 1a).

Fig. 1
figure 1

Temporal changes in a volume of extrafloral nectar secretion and b number of ant workers on young Mallotus japonicus plants (mean ± SD). *Significant difference between control and damaged plants (Mann–Whitney U test, P < 0.05)

Full size image

Ant attendance in the field

The workers of two ant species, Pheidole noda Motschoulsky and Nylanderia flavipes Smith, visited the plants during the experiment. The frequency of visits by P. noda workers was high, but that by N. flavipes workers was conspicuously low. Time-sequential pattern of total number of ant workers on the plants differed significantly between the control and damage treatments (treatment × day, F = 13.84, P < 0.0001; Fig. 1b). The number of ant workers was significantly greater on damaged plants than on undamaged plants at 1 to 7 days after damage treatment, but not at 0, 8, or 9 days.

Formation of EFN

Before treatment, plants did not differ in the number of EFNs per leaf (t test, P = 0.2441). The number of EFNs on the first leaf of plants in the leaf damage treatment was larger than that in control plants (Fig. 2). In contrast, the number of EFNs on the sixth leaf of plants in the damage treatment was smaller than that in control plants (Fig. 2).

Fig. 2
figure 2

Number of extrafloral nectaries (EFNs) produced by young Mallotus japonicus plants (mean ± SD) in leaves produced sequentially following mechanical leaf damage. *Significant difference between control and damaged plants (t test, P < 0.05)

Full size image

Discussion

Our results indicated a pattern of induction followed by relaxation of the extrafloral nectar secretions in M. japonicas plants. Nectar secretion was induced and attracted many ant workers within 1 day following the leaf damage treatment, with the secreted volume peaking at day 3 (Fig. 1a). This increased volume might increase the efficiency of indirect defense through ants, which strongly depends on the actual ant abundance on the plant (Ness 2003; Yamawo et al. 2014). Most studies of several plant species have also reported that extrafloral nectar secretion is induced within 24 h (Heil et al. 2000, 2001; Jones and Koptur 2015; Koptur 1989; Lach et al. 2009; Radhika et al. 2008, 2010) and completed within 6 days (Ness 2003; Wäckers et al. 2001). Our results are consistent with those of these previous studies.

Following the leaf damage treatment, the induction of EFN formation was the greatest on the first new leaf produced (after approximately 1 week; Fig. 2). A similar result was reported for Vicia faba (Mondor and Addicott 2003). In other plant species, the trichomes, a morphological defense trait, were induced over a 5–10-week period (review in Schaller 2008). Within approximately 2 months, Acacia drepanolobium browsed by large mammals produced larger spines than did the unbrowsed trees (Young and Okello 1998). EFN formation is perhaps induced more rapidly in response to herbivory than other morphological defenses, since the latter depends on the production of new tissues, and many plant species with EFNs are pioneers that grow rapidly (Bentley 1976, 1977; Heil and McKey 2003).

Plants can respond to herbivory by inducing extrafloral nectar secretion, but why would they increase their number of EFNs that require more time for induction than does nectar secretion? This probably happens because more EFNs enhance the effectiveness of the constitutive aspect of this indirect defense by attracting a greater number of ants onto the plants, as previously reported for M. japonicus (Yamawo and Hada 2010; Yamawo et al. 2012a, b). Another distinct possibility is that increasing EFNs serves to prime the plant for induced defenses against further bouts of herbivory. Plants that have already been damaged by herbivores may be primed to respond rather than to activate a fully induced response, which may be too costly for them (Conrath et al. 2006; Frost et al. 2008). Plants that are primed by previous cues typically do not express measurable changes in their phenotypes unless they are subsequently attacked by herbivores. When attacked, however, they can respond more rapidly and more strongly than those plants that have not been primed. Priming is assumed to accrue some small yet incremental costs overall; otherwise, all plants would constantly remain in the primed state (Karban 2010).

Mallotus japonicus plants responded to the leaf damage treatment by reducing their EFN formation on the sixth leaf produced. This result suggests that the induction of EFN formation on newly produced leaves may be costly. The plants may reduce this cost by decreasing EFN formation on leaves later produced. In fact, because the total number of EFNs during the experiment did not differ among treatments, this suggests that plants, whether damaged or not, made similar overall investments in EFNs during the experiment.

Several studies reported that induced nectar secretion was negatively correlated with constitutive secretion, thus suggesting a potential trade-off in these two herbivore defense traits (Heil et al. 2004; Holland et al. 2009). In our study, the induction of EFN formation was relaxed in those leaves later produced, following the imposed artificial leaf damage (Fig. 2). The number of EFNs on second leaves did not differ between damaged and undamaged plants. The fact that the number of EFNs on the control plants increased with their leaf number indicates that the level of constitutive investment in EFNs also increases with plant growth. Hence, these results suggest that some EFNs were induced at an earlier stage than the normal constitutive investment in EFNs stimulated by leaf damage.

In conclusion, both the EFN formation and extrafloral nectar secretion were induced and relaxed in young M. japonicus plants. Nevertheless, additional studies must be conducted to reveal the rate of induction and defense types involved. Generally, induced responses to herbivores are complex processes that depend on many interacting factors, including the characteristics of the attacking herbivore (Agrawal 2000; Mattiacci et al. 2001; Zong and Wang 2007), aspects of ant attendance (Yamawo et al. 2015), as well as local abiotic factors, such as nitrogen and water availability to plants (Mondor and Addicott 2003; Olson et al. 2009). Further research is clearly required to robustly test the induction of EFN formation in plants through long-term field experiments that include natural herbivores and ant communities.