Introduction
The diversity of color patterns found in the Heliconius butterfly radiation is a striking example of the power of natural selection to generate biodiversity. However, while the most popular theory describing the evolution of these vivid color patterns proposes a framework dissuading from wing pattern diversity, we in fact find dozens of established color patterns throughout the neotropics (Müller, 1879; Joron and Mallet 1998; Mallet and Joron 1999; Moest et al., 2020).
Franz Müller (1879), suggested in his theory that mimicking organisms which are unpalatable, venomous or toxic to predators, benefit from reduced predation by converging on common warning patterns. As these organisms become all the more similar over time, Müllerian mimicry theory predicts that the weight of predation will be optimally shared amongst the mimicking populations. Furthermore, the evolution of stark warning colorations (aposematism), increases the effectiveness of this evolutionary strategy by providing memorable patterns and colors to predators (Su et al., 2015). Examples of animals that through natural selection have trodden this evolutionary journey are familiar to many of us for their striking aspects (pitvipers, poison-dart frogs, bumblebees and wasps (Sanders et al., 2006; Symula et al., 2001; Williams, 2007; Boppré et al., 2017). The main mechanism driving this mimicry is known as positive frequency-dependent selection (pFDS), where the most common warning signal is more likely to spread through a population as it will be the most avoided by predators (Müller, 1879). In the past decades, empirical evidence has largely validated pFDS to be a principal selective force maintaining such phenotypic convergence throughout the animal kingdom (Mallet and Barton, 1989; Symula et al., 2001; Dumbacher and Fleischer, 2001; Sanders et al., 2006; Noonan and Comeault, 2009; Borer et al., 2010; Miller and Pawlik, 2013; Chouteau et al., 2016).
Heliconius butterflies are a renowned example of Müllerian mimicry. However, as first described by Henry Walter Bates (1862), the genus clearly demonstrates a diverse array of warning color patterns established throughout several mimicry rings. This presents a challenge to Müller’s theory which predicts that the selective pressures enacted by predators attacking novel color patterns should force the convergence of many warning signals into few easily recognizable color patterns. In contrast to this expectation, the co-mimics Heliconius erato andHeliconius melpomene diverged into over 25 geographic color pattern morphs. (Bates, 1862; Turner, 1975; Mallet and Gilbert 1995; Van Belleghem et al., 2020). These mimicry rings maintain homogenous local warning color patterns within their borders through localized pFDS mostly driven by a few insectivorous birds such as rufous-tailed Jacamars and tyrant-flycatchers (Benson, 1972; Chai, 1986; Langham, 2004; Mallet and Barton, 1989; Pinheiro, 2011). However, at the boundaries of these mimicry rings hybridization frequently occurs and results in narrow regions of intermediate color patterns (Mallet, 1986a; Thurman et al., 2019; Edelman et al., 2019). Such phenomena can also be observed in vertebrate Müllerian mimics such as the dendrobatid poison-dart frog radiation (Roland et al., 2017).
In contrast to the homogenous local warning color patterns, some species have evolved the ability to maintain multiple mimetic warning phenotypes in a single population, a phenomenon known as “polymorphic mimicry” (O’Donald and Pilecki, 1970). In these populations, distinct morphs are locally adapted to their environment by sharing distribution with other Müllerian co-mimics (Arias et al., 2016). The selective pressures that allow polymorphic mimicry to evolve and be maintained remains a largely unresolved question. Historically, polymorphy was considered to be a random occurrence with no obvious advantages to the organism bearing it. However, initial evidence in banded land snails (Cain and Sheppard, 1954) and later in a variety of other organisms such as spiders, guppies and wolves (Hendrickx et al., 2015; Hedrick et al., 2016; Hughes et al., 2013), has indicated that polymorphism may serve an adaptive role that can be maintained through sexual selection and possibly promote speciation (Jamie and Meier, 2020). Such a system has been described in Heliconius numata , where polymorphism is considered the result of competing selective pressures on the genomic architecture underlying the trait (Jay et al., 2021).
In this study with test sites throughout Central and South America, we set out to characterize the ecological pressures that drive polymorphism in aposematic butterflies. The Müllerian mimic Heliconius dorisis known for being polymorphic across its entire geographic distribution that spreads across most of South and Central America (Mallet, 1999; Constantino et al., 2005), with both red and blue color morphs found throughout its range. While these two morphs are ubiquitous to allH. doris populations, personal observations point out blue morphs being more abundantly found than red morphs in coastal areas of French Guiana. Additionally, red morphs show a divergence in the red rayed pattern where rays have a broader shape in Central America where red banded co-mimics are common and thinner rays in South America which perfectly match those of the thin red ray mimicry ring of the amazon basin (see Figure 1). Here, we tested if red and blue morphs of H. doris reflect predictions of a balanced polymorphism, where we expected both morphs to experience similar predation pressure wherever they are both local. We also used the regional difference in the red H. doris morphs between South and Central America to assess the ability of pFDS to drive adaptive divergence of a balanced polymorphism at varying geographic scales. Furthermore, we tested if the differences in co-mimic frequency in French Guiana from rayed phenotypes in the interior to non-rayed on the coast (Blum, 2008), can drive local differences in predation on H. dorismorphs.
Even though Müllerian Mimicry theory predicts warning signal monomorphy over time, we have found the selective pressures that allow H. doris to maintain multiple warning colors across its range. We have further observed how the same selective forces maintaining this polymorphism also act to drive divergence in warning coloration at large geographical scales.