Folivory has long‐term effects on sexual but not on asexual reproduction in woodland strawberry

Abstract Plant fitness is often a result of both sexual and asexual reproductive success and, in perennial plants, over several years. Folivory can affect both modes of reproduction. However, little is known about the effects of folivory on resource allocation to the two modes of reproduction simultaneously and across years. In a 2‐year common garden experiment, we examined the effects of different levels of folivory by the strawberry leaf beetle, Galerucella tenella, on current growth, as well as current and future sexual and asexual reproduction (runners) of perennial woodland strawberry, Fragaria vesca. In addition, we measured the chlorophyll content in leaves in the year of experimental damage to determine whether there was increased photosynthetic activity, and, thus, a compensatory response to herbivory. Finally, we tested whether the previous year's folivory, as a result of its effect on plant fitness, affected the level of natural herbivory the plant experienced during the subsequent year. In the year of experimental damage, plants that were exposed to moderate and high levels of folivory (25% and 50% leaf area consumed, respectively) increased their photosynthetic activity compared to control plants. However, only plants exposed to high folivory exhibited negative effects, with a lower probability of flowering compared to control plants, indicating that plants exposed to low or moderate folivory were able to compensate for the damage. Negative effects of folivory were carried over to the subsequent year. Plants that were exposed to moderate folivory (25% leaf area consumed) during first year produced fewer flowers and fruits in the subsequent year. Runner production was consistently unaffected by folivory. The effects of experimental folivory on the level of natural herbivory were mediated via its effects on plant fitness. Our results show that the negative effects of folivory only influence sexual reproduction in woodland strawberry. Furthermore, even though woodland strawberry can tolerate moderate amounts of folivory in the short term, the negative effects on fitness appear later; this highlights the importance of studying the effects of herbivory over consecutive years in perennial plants.

Many plant species reproduce both sexually and asexually, and both forms of reproduction contribute to plant fitness. Asexual fitness is most commonly a result of vegetative reproduction (i.e., clonal growth or vegetative propagation). However, most studies on tolerance have been conducted by concentrating on traits related to sexual fitness, such as number of flowers, fruits, or seeds (e.g., Carmona & Fornoni, 2013;Muola, Mutikainen, Laukkanen, Lilley, & Leimu, 2010;Puentes & Ågren, 2012); thus, less information is available pertaining to long-lived plants employing both sexual and asexual reproduction. Because sexual and asexual reproduction often occur simultaneously, an allocation trade-off may exist between the two modes of reproduction (Abrahamson, 1980). Given that herbivore consumption can directly damage plant meristems and alter the availability of resources, it could also affect the balance and the potential allocation trade-offs between sexual and asexual reproduction. Compared to asexual reproduction, sex increases and maintains genetic variation through recombination, and, thus, is likely to promote adaptations; such traits may include those that function as defence against herbivores (Hamilton, Axelrod, & Tanese, 1990;Johnson, Smith, & Rausher, 2009). However, sexual reproduction is, in general, considered to be more costly than asexual reproduction.
In contrast, asexual reproduction has several ecological advantages and most of them can be seen as beneficial when the plant faces damage by herbivores. These advantages include persistence in habitats unfavorable for sexual reproduction, the ability of clones to forage for resources in heterogeneous environments, and opportunities to spread the risk of death among ramets (Caraco & Kelly, 1991;Stuefer, DeKroon, & During, 1996).
Metabolic compensation for herbivore damage may force plants to increase their use of stored resources or reduce their ability to store resources for future use, which may, in turn, defer the negative effects on fitness until subsequent growing seasons . Furthermore, in perennial plants, the negative effects of herbivore damage on current and/or future reproduction may be confounded by the costs of reproduction (Venecz & Aarssen, 1998). For instance, if herbivore damage inhibits current reproduction, resources that would have been invested in reproduction may instead be stored and invested in reproduction in the following year. Thus, damaged plants could, using stored resources, achieve higher fitness in the subsequent year compared to undamaged plants. Alternatively, if damaged plants immediately regrow and reproduce to mitigate the costs of herbivore damage on current reproduction, this investment may reduce future fecundity as a consequence of the costs of compensatory reproduction. In addition to more direct effects on plant performance and fitness, herbivory during 1 year might at least partially determine the probability or amount of herbivory during subsequent years. For instance, many herbivores prefer larger or more vigorous plants (Price, 1991;Schlinkert et al., 2015). Plants that are negatively affected by herbivory are likely to be less vigorous compared to undamaged plants. Even though the amount of herbivory that plants experience is affected by many other factors than plant vigor, herbivores' preference to feed on vigorous plants is likely to add to the long-term effects of herbivory and potentially also to the plants' ability to compensate for the herbivore damage. Given these potential long-term impacts, it is important to study the effects of herbivore damage over several years in perennial plants.
In this study, we measured the ability of woodland strawberry (Fragaria vesca) to tolerate leaf damage by the strawberry leaf beetle Galerucella tenella (Coleoptera: Chrysomelidae), hereafter strawberry leaf beetle (SLB). This was done by measuring the effects of different levels of leaf damage on survival, growth, and chlorophyll content, as well as sexual and asexual reproduction of woodland strawberry across two years. Nothing is yet known about the effects of leaf feeding although florivory by SLB has previously been shown to have negative effects on the sexual fitness of woodland strawberry (Muola et al., 2017).
Woodland strawberry is a perennial herb that reproduces both sexually and asexually through the production of runners (hereafter clonal reproduction; Angevine, 1983). Flower initiation of woodland strawberry occurs in the autumn (Heide & Sønsteby, 2007); therefore, resources lost to herbivores may affect the plant's ability to initialize flowers and, thus, flowering in the subsequent growing season. Our main question was as follows: How much damage is woodland strawberry able to tolerate without negative effects on current and future growth, survival, chlorophyll content, and reproduction? We expected low to moderate amounts of damage to be readily tolerated, while high amounts of damage are likely to have negative effects over subsequent years. In addition, we were able to investigate whether damage in 1 year affects the level of natural herbivory that the plants experience the following year.

| Study species
Woodland strawberry, F. vesca (Rosaceae), is a perennial plant that occurs throughout the Northern Hemisphere, growing in various habitats such as forest clearings, forest edges, roadsides, and paths (Hancock, 1999). Flower initiation is controlled by temperature and photoperiod and occurs in the autumn (Heide & Sønsteby, 2007).
In the study area, the initiated flower buds develop in the following spring, and the main flowering period lasts from late May to early July. Flowers are hermaphrodite and self-compatible to various degrees (Egan, Muola, & Stenberg, 2018;Angevine, 1983). Woodland strawberry is insect pollinated, and wild plants reproduce both sexually and clonally through formation of aboveground runners (Hancock, 1999).
The strawberry leaf beetle (SLB), G. tenella L. (Coleoptera, Chrysomelidae), is a common oligophagous herbivore that feeds on many species of Rosaceae (Olofsson & Pettersson, 1992;Stenberg & Axelsson, 2008; Figure 1). Adult SLBs overwinter in the soil and emerge in April to May. They forage on both leaves and flowers and can eat petals, seeds, and fruits. Oviposition starts in late May to early June and eggs are laid on the leaves. Eggs hatch in June to July.
Larvae feed for 2-4 weeks on aboveground tissues before pupating in the ground. After approximately eight days, the pupae hatch and the adults that emerge feed on leaves before overwintering from mid-September (Olofsson & Pettersson, 1992;Stenberg, Witzell, & Ericson, 2006). All SLB individuals used in the experiment were collected from a natural population close to SLU Ultuna campus (N59.810°, E17.667°) during late April and early May 2013.

| Experimental design
To determine the effects of SLB damage on survival, growth, chlorophyll content, and sexual and clonal reproduction of woodland strawberry, we conducted a 2-year common garden experiment.
Experimental plants were divided into four treatment groups (described below) representing different damage levels: control, 10%, 25%, and 50% leaf area consumed by SLB. Temperatures and weather conditions in the caged area followed outdoor temperatures and conditions. Wild pollinating insects were able to enter the area, but the mesh (diameter: 20 × 20 mm) excluded birds that could consume the ripened fruits. One plant per genotype was randomly assigned to one of four damage treatments (n = 70 per treatment) representing different damage levels: control, 10%, 25%, and 50% leaf area consumed by SLB. In June 2013, each experimental plant was enclosed in a perforated (hole diameter 0.5 mm) polythene bag (Baumann Saatzuchtbedarf, Waldenburg, Germany). To generate the herbivore damage, two beetles, one female and one male, were placed inside the bag and they were allowed to feed until the required damage level was reached (maximum eight days). To score the damage level, plants were monitored daily and the damage level was estimated visually (by eye). The used herbivore density corresponds to natural beetle densities found in a previous study (Stenberg, 2014).
Due to an unfortunate human error, some of the experimental plants were sprayed with pesticides and had to be excluded from the experiment. After excluding the plants that had been exposed to pesticides, we ended up with a total of 186 plants and the following number of replicates in different treatments: control (n = 66), 10% (n = 36), 25% (n = 52), and 50% (n = 32) leaf area consumed by SLB.
To measure the effect of SLB leaf damage on growth, chlorophyll content, and sexual and clonal reproduction of woodland strawberry, we determined the number of leaves prior to the damage treatment in June 2013 and 4 weeks after the damage treatment in July 2013. We calculated the number of flowers and fruits, and recorded whether each plant was producing runners (i.e., reproducing clonally) in July and early September 2013. Given that increased photosynthetic activity following herbivore damage is one of the most widely acknowledged mechanisms of tolerance (Rosenthal & Kotanen, 1994;Strauss & Agrawal, 1999;Tiffin, 2000), we measured the foliar chlorophyll content as an estimate of compensatory response to SLB damage. We used a hand-  (Cross et al., 2001). We assessed the total amount of damage each plant had experienced, meaning that we did not quantify damage by different herbivores separately. According to the amount of damage, plants were classified into one of four classes: (a) 1% or less of leaf area damaged, (b) 5% or less of leaf area damaged, (c) 5% to 20% of leaf area damaged, and (d) 20% or more of leaf area damaged.  To investigate whether previous year's experimental herbivore damage, as a result of its effects on plant fitness, affects the level of natural herbivory a plant experiences during the following year, we tested the association between natural herbivory and the number of flowers. We calculated the correlation coefficient for number of flowers against natural damage using Spearman's partial correlation in order to remove the potential effect of block on natural herbivory. Because number of flowers was negatively affected by the experimental damage in the year before (see Results, Table 2 and In all the models, interactions with covariate and explanatory variables were tested and nonsignificant interactions were removed from the final models. Block was included to eliminate the potential effects of environmental variation due to the different positions of experimental plants first in the caged area and, later, in the common garden in Krusenberg. Block was treated as a fixed factor since we had only three blocks. Differences between control plants and plants that experienced different amounts of SLB damage were examined with pairwise a priori contrasts, specified as control versus each damage level in all the analyses where the main effect of controlled damage treatment was statistically significant. Bonferroni correction was used to correct for multiple comparisons. All analyses were conducted with SAS 9.4 (SAS Enterprise Guide 6.1/SAS 9.4 Cary, NC).  Figure 2b). In addition, plant size affected the probability of flowering (Table 1): Flowering individuals were, on average, larger (initial size 6.0 ± 0.2 leaves) than those that did not flower (initial size 5.  Table 1). Runner producing plants had on average 7.9 ± 9.5 (mean ± SD) runners.

| Year following experimental damage
Strawberry leaf beetle damage had a negative effect on flower and fruit production in the year following experimental damage (Table 2, Figure 3). Compared to control plants, plants with 25% of their leaf area consumed had significant, and plants with 50% of their leaf area consumed nearly significant, reduction in their flower production during the year following experimental damage. The reduction in flower production was 31% for plants that had 25% of their leaf area consumed (contrast: F 1, 173 = 7.65, Bonferroni corrected p = 0.0189; Figure 3a) and 30% for plants that had 50% of their leaf area consumed (contrast:  Table 2). Initial plant size affected both the number of flowers and the number of fruits but not fruit set: Plants that were larger to begin with produced, on average, more flowers and fruits even during the following growing season (data not shown; Table 2). Experimental damage did not affect runner production during the following year (control 22.3 ± 1.1 g [mean ± SE], 10% leaf area consumed 23.5 ± 1.6 g, 25% leaf area consumed 21.9 ± 1.3 g, and 50% leaf area consumed 18.5 ± 1.7 g; Table 2). We found a positive correlation between natural herbivory and the number of flowers, indicating that herbivores preferred plants with more flowers (r = 0.3585, n = 180, p<0.0001).

| D ISCUSS I ON
We found that damaged plants increased their chlorophyll content, thereby compensating for folivory. As a result, only plants with large amounts of leaf damage experienced negative effects on their sexual reproduction in the year of experimental damage. Given that most resistance traits reduce but do not completely eliminate damage, even relatively well-defended long-lived perennials often suffer severe biomass losses to herbivory (Haukioja & Koricheva, 2000;Turcotte, Turley, & Johnson, 2014). Thus, the long life span and the long exposure time to herbivores may favor the evolution of tolerance in perennial plants (Haukioja & Koricheva, 2000). Interestingly, in the year following experimental damage, negative consequences for flower and fruit production were more pronounced for plants with moderate (25%) damage than for plants with high (50%) damage. Our findings indicate that although woodland strawberry seems F I G U R E 3 The effect of Galerucella tenella (SLB) damage on (a) number of flowers and (b) number of fruits in the year following experimental damage (LSMEANS ± SE). Statistical significance between control and different damage levels was obtained using a priori contrasts and is denoted with an asterisk. *p < 0.05 able to tolerate moderate amounts of leaf damage relatively well in the short term, the negative effects of folivory appear later and are specifically focused on sexual reproduction. Given these findings, studying only the short-term effects of folivory in perennial plants may lead to misleading conclusions. Thus, the effects of herbivory are important to study over longer time periods in perennial plants.
In the year of damage, their growth, and sexual and clonal reproduction did not differ significantly from those of control plants.
However, since strawberry flowers are initiated during the previous growing season (Heide & Sønsteby, 2007), it is likely that, due to compensating for herbivore damage without compromising the current sexual reproduction, plants with only 25% damage had fewer available resources to initiate flower production and this delayed the negative fitness effects of SLB damage until the following growing season.
The observed carry-over effect of leaf damage is probably due to differences in resource reallocation patterns after different amounts of damage, and it may be further confounded by the potential costs of (sexual) reproduction (Primack, Miao, & Becker, 1994;Puentes & Ågren, 2012;Vallius & Salonen, 2000). Our findings emphasize the importance of following the effects of herbivore damage over more than one growing season in perennial plants. In addition, we found that even if leaf damage decreased the probability of flowering and the number of flowers and fruits, it did not affect the probability of fruit production or fruit set. Our results, thus, indicate that plants do not respond to damage by adjusting fruiting probability of flowering plants or fruit set. Fruit setting may contribute to the costs of sexual reproduction but, at least according to our results, at a similar level in all experimental groups.
The negative effects of herbivory are supposed to be more pronounced when plants are developing reproductive structures that are related to sexual reproduction (e.g., Tiffin, 2000).
In accordance with this, our results showed that while herbivore damage had negative effects on sexual reproduction, no effects at all were found on clonal reproduction. This might be due to the allocation trade-off between sexual and clonal reproduction.
Allocation trade-offs can be either resource-based or more direct when clonal organs replace sexual structures, and they are likely to limit the fitness gain through either reproductive mode (Ronsheim & Bever, 2000;Vallejo-Marín, Dorken, & Barrett, 2010 (Berenbaum, Zangerl, & Nitao, 1986;Mauricio, 1998;Juenger, Lennartsson, & Tuomi, 2000;Baucom & Mauricio, 2004). Sexual reproduction can, thus, be considered beneficial to plants exposed to herbivory (Hamilton et al., 1990;Johnson et al., 2009). Our results showed that although only a small proportion of the experimental plants flowered and set fruits during the first year, even some of the plant individuals that experienced the largest amount of experimental damage (50%) did so. This might indicate that even when facing severe defoliation plants maintain at least minimal levels of sexual reproduction in order to ensure the benefits of sexuality. Further studies are needed to understand the roles of sexual and asexual reproduction when plants are facing herbivory, and whether the patterns observed in this study are due to resource allocation trade-offs between the different modes of reproduction.
In the year following damage, experimental plants growing in the common garden were damaged by herbivores present at the site. The highest amount of natural leaf damage during the second year was around 20% leaf area consumed. We found a positive correlation between natural herbivory and the number of flowers, indicating that herbivores preferred more vigorously growing plants (Price, 1991). The same phenomenon has also been found in other plant species (Haysom & Coulson, 1998;Neuvonen & Niemelä, 1981;Schlinkert et al., 2015). Plants that experienced more experimental damage produced, on average, fewer flowers during the following year. If herbivores do prefer vigorous woodland strawberries, this finding could indicate that plants suffering the negative effects of herbivory during subsequent years may experience less damage, and, thus, have more resources to allocate for compensation.

| CON CLUS IONS
The results of this study show how the negative effects of folivory depend not only on the amount of damage but also on the time frame used to study the effects. This is likely to be because of differences in resource allocation after herbivore damage. This problematizes our view of tolerance-what appears to be tolerance in 1 year actually turns out to be a cost that is deferred in time. Our results, thus, underline the importance of studying the effects of herbivore damage over several consecutive years in perennial plants. In addition, our results show how negative effects of herbivory are consistently channeled to sexual reproduction, while clonal reproduction remains unaffected. Given that both modes of reproduction usually have different ecological and evolutionary consequences for different aspects of plant life, it is important to consider the role of both these modes when studying the effects of herbivory.

ACK N OWLED G M ENTS
We thank J.-E. Englund for valuable statistical discussions; C.