Flower color preferences of insects and livestock: effects on Gentiana lutea reproductive success

Introduction

Despite pollinators being considered the principal driver of floral diversification and speciation (Bradshaw & Schemske, 2003; Herrera, Castellanos & Medrano, 2006; Whittall & Hodges, 2007; Van der Niet, Peakall & Johnson, 2014), non-pollinator agents also interact with plants and promote floral evolution (Strauss & Whittall, 2006; Andersson, 2008). This is the case with antagonist animals, such as herbivores, which damage plants as a result of resources extraction (Strauss & Whittall, 2006; Whittall & Carlson, 2009), affecting fitness and, therefore, potentially causing natural selection on plant traits (Wise & Hébert, 2010; Agrawal et al., 2012; De Jager & Ellis, 2014).

Plant reproductive success can be shaped by the balance between mutualistic and antagonistic interactions which may maintain floral trait variation (Herrera et al., 2002; Lavergne, Debussche & Thompson, 2005; McCall & Irwin, 2006; Bartkowska & Johnston, 2012). Herbivory can have a negative effect on plant fitness in a greater or lesser extent depending on pollinator presence or absence (Herrera, 2000; Herrera et al., 2002), and the positive effect exerted by pollinators may depend on herbivore presence (Gómez, 2003; Gómez, 2005a). Thus, phenotypic variation of a floral trait, such as flower color, can result from the balance between the positive selection pressures exerted by pollinators and seed dispersers, and the negative selection pressures exerted by herbivores or seed predators (Herrera, 2000; Herrera et al., 2002; Irwin et al., 2003; Asikainen & Mutikainen, 2005; Frey, 2004; Strauss, Irwin & Lambrix, 2004; Andersson, 2008; Bartkowska & Johnston, 2012). Selective pressures exerted by abiotic factors (Galen, 2000; Warren & Mackenzie, 2001; Streisfeld & Kohn, 2007) and historical processes or genetic drift (Mitchell-Olds, Willis & Goldstein, 2007) may also influence phenotypic variation.

Invertebrate herbivory (mainly insects and gastropods) negatively affects plant fitness causing natural selection on plant traits, including floral traits, such as flower color (Fineblum & Rausher, 1997; Frey, 2004; Whittall & Carlson, 2009; Bartkowska & Johnston, 2012). Additionally, vertebrate herbivory (typically represented by large mammal herbivores) affect plant fitness and community composition (Herrera, 1984; Agustine & McNaughton, 1998; Knight, 2003; Gómez, 2005b; Allsup, 2014). Interaction with mammals can also have a positive effect for the plants. Reproductive success of many flowering plant species relies on browsing ungulates activity, since they play an essential role if ungulates act as seed dispersal agents within plant communities (for example, Herrera, 1984; Herrera, 2000). Thus, the effect of mammal herbivory can be either positive, if they act as seed dispersal agents, or negative if they act as herbivores. Many plant communities interact not only with wild fauna but also with domesticated mammals. Domesticated ungulates such as cattle, horses and sheep often feed on wild plant populations (for example during transhumance practices) potentially exerting natural selection on them.

The type and strength of selection exerted on plant attributes depend on herbivore preferences during foraging (Agustine & McNaughton, 1998; Asikainen & Mutikainen, 2005; Agrawal, Lau & Hambäck, 2006). Livestock could discriminate between flower color morphs (Phillips & Lomas, 2001), and thus might be able to show color preferences. But, even if livestock could not detect flower color differences, it still may differentially feed on a particular color morph if it shows preferences for any trait correlated with flower color (see Lande & Arnold, 1983). As it could be the case for some herbivory defenses which are known to be correlated with floral pigments (Simms & Bucher, 1996).

The effect of livestock (negative or positive) is expected to be stronger than the exerted by insect herbivores or pollinators due to the amount of damage or seed dispersal that large vertebrates are able to exert. Thus, it is necessary to take their effect into account in order to have a holistic view of the biotic forces exerting selective pressures on traits such as floral color (Johnson, Campbell & Barrett, 2015).

Although it has been long recognized that both pollinators and herbivores play an important role on plant evolutionhttp://brahmaputra:8080/tf/texfolio/APP/connector/0/791/source/peerj1685-main14580233070791.png (Ehrlich & Raven, 1964), their effects are usually studied independently and under different scopes (Johnson, Campbell & Barrett, 2015). Moreover, the effects of grazing mammals on flower traits have been disregarded (but, see Juenger & Bergelson, 1997; Gómez, 2003; Gómez et al., 2009; Ågren et al., 2013), and the potential effects of domesticated mammal herbivores on floral traits are unknown. Here we explore for the first time the simultaneous effect of wild and domesticated animals on plant traits, particularly regarding their potential effect on flower color variation.

Gentiana lutea flower color varies continuously from orange to yellow within and among populations in NW Spain (Sobral et al., 2015)—where livestock interacts with plant communities (see Blanco-Fontao, Quevedo & Obeso, 2011) and commonly feed on G. lutea. This corolla color variation has a genetic basis (Zhu et al., 2002; Zhu et al., 2003) and is not related to abiotic factors such as radiation, altitude, temperature or rainfall (Veiga et al., 2015a). Two flower color varieties are described (Laínz, 1982; Renobales, 2003) and a partial hybridization barrier exists between the yellow G. lutea var. aurantiaca and the orange G. lutea var. lutea (Losada et al., 2015). G. lutea is strongly dependent on pollinators which, together with seed predators, show flower color preferences causing selection on flower color (Losada et al., 2015; Veiga et al., 2015b).

Here we test if insect herbivores—mainly adults belonging to Orthopthera and Coleoptera orders, and larvae from different insect groups—and large mammal domesticated herbivores (local extensive livestock, mainly cows and horses) affect flower color variation in Gentiana lutea—while having into account the effect of pollinators and seed predators. For this purpose, the following questions were formulated: (i) Do insect herbivores and livestock show flower color preferences when feeding on G. lutea? (ii) Do mutualists (pollinators) and antagonists (seed predators, insect and livestock) jointly affect G. lutea reproductive success?

Materials & Methods

Study area

Our study area covered the distribution of Gentiana lutea in the Cantabrian Mountains, NW Spain (see Fig. 1). In 2010, we visited 8 populations (Cebreiro, Ancares, Leitariegos, Torrestío, Ventana, San Isidro, Señales and San Glorio) and 12 in 2011 (Queixa, San Mamede, Loureses and Pontón were new). All studied populations were haphazardly selected along a 230 km longitudinal gradient from the San Mamede population (42°12′N, 7°30′W; at the western limit) to the San Glorio population (43°04′N, 4°45′W; at the eastern limit), and localized at high altitudes from 1,100 m to 1,700 m.a.s.l., on grassy pastures and hillsides used extensively by local livestock. For this research, we received a field permit from the Environmental Territorial Service of León, Territorial Delegation of Government of Spain, Regional Government of Castilla and León (ID:12_LE_325_RNA_PuebladeLillo_INV; Reference: 06.01.013.016/ROT/abp; File number: AEN/LE/103/12).

Results

Insect and livestock preferences for flower color

Insect herbivores showed simultaneous preferences for flower color and stalk height when feeding on Gentiana lutea (Table 1). Overall, these insects preferred to feed upon yellow-flowering individuals (Fig. 2) and this flower color preference depended on stalk height, as the significant interaction between these two plant traits suggests. Insects prefer yellow-flowering individuals and, among these, shorter individuals were more herbivorized than longer ones (Table 1 and Fig. 3). The relationship between the intensity of herbivory and the interaction between flower color and other plant traits also happens within some of the populations (see Cebreiro, Torrestío and Ventana populations, Appendix S1).

Livestock did not show preferences for flower color within each studied population; although, the probability of livestock herbivory marginally depended on the interaction between flower color and leave length in a population (Cebreiro population; see Appendix S2). Note that significant effects were also not found if the analyses were split by year.

Discussion

Insect herbivores preferred to feed on yellow and short stalk-flowering individuals but livestock did not show flower color preferences along the Gentiana lutea range studied (Table 1 and Fig. 2; Appendix S2). In some G. lutea populations pollinators and seed predators also prefer the yellow morphs (see Veiga et al., 2015b) but the color preferences vary between populations (see Sobral et al., 2015). Pollinators might visit the more herbivorized individuals in some locations whereas, on other locations, pollinator and herbivores might prefer different corolla colors. This is likely to depend on the pollinator and herbivore community composition as well as on the range of color variation in particular populations.

Floral color in G. lutea varies among individuals depending on carotenoids concentration (Zhu et al., 2002). These pigments, which regulate color expression from orange to yellow, are involved in the synthesis of volatile compounds considered to be defenses against insect herbivory (Lakshminarayan, 2013). Alternatively, anthocyanines (which participate in red color expression) may play a dual role, both attracting pollinators and alerting herbivores of a high content of chemical secondary compounds that confer toxicity or, at least make plant tissues difficult to metabolize (Lev-Yadun & Gould, 2009). The fact that insect herbivores preferred to feed upon yellow-flowering individuals might suggest that yellowish corolla color pigments are related to lower amounts of chemical deterrents than orange pigmentation.

Livestock herbivory negatively affected the reproductive success of G. lutea, whereas pollination had a positive effect on G. lutea’s seed output. We found the effect of livestock herbivory on seed production to be stronger than the effect of pollinators (see β values, Table 2). The few studies dealing with the joint effect of herbivores and pollinators on floral traits show that selection by herbivores is often (in 70% of the cases) as strong (or stronger) than selection exerted by pollinators on flowers’ characteristics—herbivores potentially affect floral traits as much as pollinators do (see Johnson, Campbell & Barrett, 2015). Flower color variation in other species is maintained by the balance of selective pressures exerted by mutualisms and antagonisms; for example in Raphanus sativus (McCall et al., 2013), Iris lutescens (Wang et al., 2013), Ursinia calenduliflora (De Jager & Ellis, 2014) and Geranium thunbergii (Tsuchimatsu, Yoshitake & Ito, 2014). But, we cannot confirm that either insects or livestock play a role on G.lutea’s flower color variation. This is because we found effects of livestock on seed production but not color preferences, and insect herbivores showed color preferences but did not affect seed production. However, insect herbivory also depended on correlations among vegetative traits (such as stalk height or leaf length) and flower color (Table 1; Appendix S1). Thus, if leaf length or stalk height were related to seed production, insects could still be playing an indirect role on the maintenance of flower color variation through pleiotropic effects among correlated plant traits (Fineblum & Rausher, 1997; Herrera et al., 2002; Strauss, Irwin & Lambrix, 2004; Narbona et al., 2014; Johnson, Campbell & Barrett, 2015).

Livestock and insect herbivores do not play a role shaping flower color variation in G. lutea (unless through indirect selection). But, pollinators show color preferences and affect G. lutea’s fitness. Thus, flower color variation in G. lutea might be originated, or at least reinforced, by the selective pressures exerted by pollinators. Flower color variation in polymorphic species may originate from selection by animals which could favor isolation between different color morphs and cause sympatric diversification. However, other reasons could additionally explain flower color variation; for example, the geographic isolation produced by the Quaternary climatic changes has been identified as the main cause of divergence in several mountain plant species (e.g., Martín-Bravo et al., 2010; Alarcón et al., 2012; Blanco-Pastor & Vargas, 2013; Fernández-Mazuecos et al., 2013).

Domesticated animals feeding on G. lutea do not show preferences for flower color. Despite it, we argue for the importance of considering the effect of domesticated animals on plant conservation and evolution. Livestock shape plant communities and ecosystems through their interaction with particular species (for example López-Sánchez et al., 2014; López-Sánchez et al., 2015). Human activities are known to have many effects on biodiversity (for example, Dirzo et al., 2014). We might have overlooked a potential indirect effect that human could play on diversity conservation and evolution—through the interactions of domesticated mammals with plant communities. Considering livestock effects on plant communities could better our nature understanding and management.

Supplemental Information