Variation in plant communities is likely to modulate the feeding and oviposition behavior of herbivorous insects, and plant associated microbes are largely ignored in this context. Here we take into account that insects feeding on grasses commonly encounter systemic and vertically transmitted (via seeds) fungal Epichloë endophytes, which are regarded as defensive grass mutualists. Defensive mutualism is primarily attributable to alkaloids of fungal origin. To study the effects of Epichloë on insect behavior and performance, we selected wild tall fescue (Festuca arundinacea) and red fescue (Festuca rubra) as grass-endophyte models. The plants used either harbored the systemic endophyte (E+) or were endophyte-free (E-). As a model herbivore, we selected the Coenonympha hero butterfly feeding on grasses as larvae. We examined both oviposition and feeding preferences of the herbivore as well as larval performance in relation to the presence of Epichloë endophytes in the plants. Our findings did not clearly support the female’s oviposition preference to reflect the performance of her offspring. First, the preference responses depended greatly on the grass-endophyte symbiotum. In F. arundinacea, C. hero females preferred E+ individuals in oviposition choice tests, whereas in F. rubra, the endophytes may decrease exploitation, as both C. hero adults and larvae preferred E- grasses. Second, the endophytes had no effect on larval performance. Overall, F. arundinacea was an inferior host for C. hero larvae. However, the attraction of C. hero females to E+ may not be maladaptive if these plants constitute a favorable oviposition substrate for reasons other than the plants’ nutritional quality. For example, rougher surface of E+ plant may physically facilitate the attachment of eggs, or the plants afford greater protection from natural enemies. Our results highlight the importance to consider the preference of herbivorous insects in studies involving the endophyte-symbiotic grasses as host plants.

We collected seeds from four E+ and four E- free-pollinated mother plants of red fescue (Festuca rubra) and tall fescue (Festuca arundinacea). As fescues were both wind- and cross-pollinated, we assumed that the offspring of an individual mother plant were half-siblings. Thus, this data set is based on four half-sibling groups in both E+ and E- plants of both species. The minimum of three pots (8 cm x 8 cm in size; each with more than five individuals) of each half-sibling group (50 pots in total, later referred to as “plants”) were grown in greenhouse conditions (stable ~15 °C; ambient light; potting soil: Biolan for saplings; fertilizer: 1 dl of Biopon Rose every other week according to instructions) at Ruissalo Botanical Garden for two months before they were transferred to growth chambers [16 h: ~22 °C, light; 8 h: ~16 °C, dark] at the University of Tartu, Estonia. All plants were cut to a height of 5 cm 2 weeks before the first experiments to induce tillering so as to obtain more plant material.

The adult female insects used were caught from the wild in Estonia in the beginning of June 2016. When they were not used in the experiments, the wild-captured females were kept individually in small jars in a refrigerator. Food (sugar solution) for them was always present on tissue in the jars. The larvae in the feeding experiments were first-generation, laboratory-born offspring of the wild females used in the oviposition experiments. The eggs were kept in small jars and the newly-hatched larvae were not fed before the experiments so as to avoid biasing their preference for the food given beforehand.

Oviposition-choice test

To examine the oviposition preference of the C. hero females relative to the presence of endophytes, we conducted trials that introduced the butterflies individually to terrarium boxes (35 cm x 25 cm x 20cm) that gave them the choice between E+ and E- F. arundinacea or F. rubra plants. The pots with plants were placed under the terraria, and the above-ground parts of the potted E+ and E- plants were inserted in opposite corners of the terrarium through holes made in the bottom of the box. Both plants inside the same terrarium were approximately equal in biomass and similar in appearance. We investigated all possible combinations of E+ and E- half-sibling plant groups. During the experiment, some of the plant individuals were used twice with different counterparts making all the plant combinations unique. The positions of the E- and the E+ plants were alternated in the terraria, and an adjustable table lamp was put above each to provide the females with light and warmth (about 27 °C inside the terraria). The light was on daily from approximately 8 a.m. to 8 p.m. One female at a time was kept in a terrarium with the E- and E+ plants until it had laid several (> 5) eggs on the plants. A different female was used in each trial. After the end of the experiment, we recorded the number of eggs attached to each plant. We examined the behavior of at least five females from the four insect populations on both plant species.

Oviposition-rate test

To determine the attractiveness of the E- and E+ plants as oviposition substrates, we conducted trials to measure the oviposition rate of the C. hero females exposed to a single plant, either E- or E+, for a fixed amount of time (single-substrate design). We assumed that the oviposition rate indirectly reflects the host’s quality, because the female should save her eggs if there is no preferred plant available. In contrast, the oviposition rate should be high if an ideal plant is present. The number of eggs laid in a fixed period provided an approximation of both the plant’s suitability for oviposition and the preference of the female.

The experiment was conducted in growth chambers [17 h: ~27 °C, light; 7 h: ~20 °C, dark]. The adult females were placed in transparent plastic boxes (1.5 liters) into which the shoots of  potted E+ or E- plants of F. arundinacea or F. rubra had been inserted through a hole in the bottom. Small holes in the lids let excess humidity evaporate from the box. After 21 hours, the females were removed from the experimental arena, and the eggs they had laid were counted. Each female was used in this experiment up to four times if her condition allowed it. Each was exposed to all four plant types (two species with both E+ and E- individuals), and the first and second plants were of the same species before the plant species was switched. The starting plant type was randomly chosen for each female to take into account the possible effect of previous experience on oviposition decisions.

Food-plant preference test

To examine larval food preference, we conducted trials that enabled the mobile, newly-hatched C. hero larvae to choose between E- and E+ plants. The leaves of living, potted E+ and E- plants were inserted from opposite directions between the lid of a Petri dish, and a single larva was placed in the center of the Petri dish . After 18 h, the experiment was terminated, and the leaves were examined for signs of larval consumption (missing leaf parts).

Larval-growth and survival test

To compare the quality of the E- and E+ plants for larvae, we conducted a no-choice rearing experiment that mimicked the situation in which the larvae have no possibility of switching to another plant after hatching. We measured the growth and survival of the larvae that were feeding on either the E- or E+ plants of F. rubra or F. arundinacea in a growth chamber. The leaves of a potted, living plant were placed so as to grow through the mesh tubes (12 cm x 10 cm) in which the larvae were reared. The amount of plant material enclosed in the tubes was more than sufficient to support all the larvae. These mesh tubes were closable at each end and had a wooden ring in the middle that kept the tube from collapsing. Each mesh tube contained five newly-hatched and active larvae from the same population. After two weeks, the living larvae were weighed using a precision balance, and the weight was used as an index of individual performance. Any larvae missing from the mesh tubes were counted as having died.

There were four plant types F. rubra E+, F. rubra E-, F. arundinacea E+ and F. arundinacea E-. E+ stands for endophyte-symbiotic and E- is endophyte-free.

Numbers like 2_7 or 38_10 are just plant codes. Plants with same plant code are at least half siblings.

"d" in mass means dead (not weighted).