Developmental plasticity in multimodal signals: light environment produces novel signalling phenotypes in a butterfly

Abstract

Developmental plasticity can alter the expression of sexual signals in novel environments and is therefore thought to play an important role in promoting divergence. Sexual signals, however, are often multimodal and mate choice multivariate. Hence, to understand how developmental plasticity can facilitate divergence, we must assess plasticity across signal components and its cumulative impact on signalling. Here, we examine how developmental plasticity influences different components of cabbage white butterfly Pieris rapae multimodal signals, its effects on their signalling phenotypes and its implications for divergence. To do this, we reared P. rapae caterpillars under two different light environments (low-light and high-light) to simulate conditions experienced by P. rapae colonizing a novel light habitat. We then examined plasticity in both visual (wing coloration) and olfactory (pheromone abundance) components of male sexual signals. We found light environments influenced expression of both visual and olfactory components and resulted in a trade-off between signal modalities. The ‘low-light’ phenotype had duller wing colours but higher abundance of the pheromone, indole, whereas the ‘high-light’ phenotype had comparatively brighter wings but lower abundance of indole. These results show that by simultaneously altering expression of different signal components, developmental plasticity can produce multiple signalling phenotypes, which may catalyse divergence.

1. Introduction

Developmental plasticity, the ability of a genotype to alter phenotypic development depending on the environment, is common across organisms and traits [1]. Such plasticity is driven by changes in developmental pathways in response to variation in environmental conditions [1] and can result in rapid generation of novel phenotypes. Consequently, developmental plasticity can play an important role in promoting divergence [2]. One relatively underexplored pathway by which developmental plasticity can catalyse divergence is through its effects on sexual signals [3–5].

Sexual signals play an important role in mate choice and mating outcomes. Consequently, changes in signalling traits can lead to assortative mating and reproductive isolation, which can facilitate divergence [4,5]. Although several studies have demonstrated plasticity in sexual signals (e.g. changes bird and frog calls [6] and changes in tactile signals in spiders [7]), the majority of these examples highlight context-dependent behavioural plasticity and not developmental plasticity. Such context-dependent behavioural plasticity can allow organisms to cope with environmental fluctuations due to a high degree of lability. However, such high lability can also be a strong buffer against divergence (see ‘behavioural inertia’) [8]. Unlike context-dependent behavioural plasticity, the effects of developmental plasticity are relatively more fixed once development is complete and individuals attain maturity [9]. This longer-term trait expression can provide greater opportunity for selection to act on traits and for divergence to occur [1].

We know that sexual signals are sensitive to developmental conditions, for instance, changes in cuticular hydrocarbons (CHCs) in crickets due to diet and temperature change [10]. However, most studies of developmental plasticity in signals focus on a single component of the sexual signal (but see [11]). Sexual signals, however, are often multimodal because they consist of multiple components across different sensory modalities such as vision, acoustic and olfaction [12]. Mate choice for such signals is also generally multivariate and does not always depend on a single component [12]. Mate choice depends instead on the effective integration of multiple components, that is the ‘signalling phenotype’. For example, in crickets, mate choice depends both on CHCs and acoustic calls [13]. Hence, to better understand the role developmental plasticity can play in promoting divergence, it is essential to examine plasticity across different signal components to assess cumulative impacts on the signalling phenotype.

Cabbage white butterflies, Pieris rapae, are an excellent model system to study links between developmental plasticity, multimodal sexual signals and divergence. Male P. rapae use multimodal signals that consist of visual and olfactory components. These components play different roles during two different stages of mate choice. In the first stage, males perform aerial courtship displays around females to advertise dorsal wing colours. Females assess wing coloration before landing and allowing males to approach [14]. Typically, males with brighter and more saturated colours are more successful at eliciting landing responses [14,15]. Females then assess a male's sex pheromones to decide whether or not to mate [16]. Although several pheromones have been identified in P. rapae, ferrulactone and indole play particularly vital roles in mate choice. Of these, ferrulactone influences mate acceptance behaviour; females reject males with low ferrulactone [17]. Indole on the other hand, is an anti-aphrodisiac pheromone, which is transferred from males to females during mating. Indole makes females appear unattractive and reduces chances of remating [18].

To examine the effects of developmental plasticity on P. rapae signals, we reared caterpillars of wild caught females under two different light environments ‘high-light’ and ‘low-light’. Light conditions in the low-light environment were similar to conditions in closed-canopy shady habitats such as woodlands and forests, whereas the light conditions in the high-light environment were comparable to an open habitat such as grassland and agricultural fields (see electronic supplementary material). We then examined plasticity in wing coloration and abundances of two sex pheromones, indole and ferrulactone, and its combined effects on male P. rapae signals. Here, we chose to focus on plasticity in response to light environments because butterfly development is sensitive to light conditions and changes in light are known to influence expression of several traits in butterflies [19]. Additionally, P. rapae, a non-native species in North America, generally prefer open environments, but experience variation in light conditions as they colonize novel habitats such as closed-canopy forests and woodlands [20], for instance to make use of invasive garlic mustard as a host [21,22].

2. Material and methods

(a) . Egg collection and caterpillar rearing

Gravid P. rapae females (N = 23) were captured between June and August 2020 from agricultural fields around University of Minnesota, Saint Paul campus. Eggs from these females were collected and allowed to hatch. Second instar caterpillars were then transferred to artificial diets in cups. Caterpillars reared in the low-light environment were transferred to black cups with transparent lids, whereas caterpillars reared under high-light were transferred to fully transparent cups. Assignment of caterpillars to black or transparent cups was done at random. All cups were maintained in a climate chamber at about 24°C, 55% humidity and 14 h photoperiod. Caterpillars in the low-light environment experienced approximately 62% reduction in total light intensity compared to high-light environment (see electronic supplementary material, figure S1). For every male butterfly, we measured development time (number of days between transferring caterpillars to the artificial diet until eclosion) and length of upper wings from standardized photos using Fiji software. We also captured wild males from the same area where gravid females were captured. Wing lengths, coloration and pheromone abundances of wild males were also measured and used as a reference to gauge the direction and magnitude of plasticity in signals in response to light environments. Additional details of female maintenance, caterpillar rearing and light environments are in the electronic supplementary material.

(b) . Colour measurements and visual modelling

We used a JAZ-A2474 spectrophotometer with PX lamp, Ocean Optics, to measure the dorsal wing coloration of males. To examine whether differences in wing coloration were perceivable to female P. rapae, recorded spectra were used to calculate chromatic (hue and saturation differences between wings and the background) and achromatic (luminance difference between wings and the background) contrasts of wing coloration using the photoreceptor sensitivities of female P. rapae [15]. Details of spectrophotometry and visual modelling are provided in the electronic supplementary material.

(c) . Pheromone extraction and quantification

Freshly eclosed males were held in insect cages for 5 days after which they were sacrificed by holding them at −80 °C for 5 min in fresh glassine envelopes. The abdomen and one set of wings were separated and transferred to 4 ml opaque brown borosilicate vials containing 500 µl of dichloromethane (DCH, CH₂Cl_2.) to extract pheromones. Samples were held at room temperature for 30 min, shaken and then stored at −80 °C until further analyses. GC-MS was performed on an Agilent chromatograph model with a DB-5MS UI capillary column coupled to an Agilent 5973 quadrupole mass spectrometer. One millimolar of 10 µl caryophyllene (Sigma Aldrich) was added to each sample and used as internal standard. Chromatograms were analysed in ChemStation software (Agilent Technologies) and observed peaks were identified by comparing their mass spectra with those of known compounds found in P. rapae (spectral comparison provided in the electronic supplementary material) [17,23]. Peak areas including that of the internal standard (caryophyllene) were obtained using automatic peak integration function in ChemStation. Abundance of internal standard was used to normalize abundance of indole and ferrulactone to facilitate cross-sample comparisons. Normalized abundances were also standardized by wing lengths to control for body size effects. Additional details are provided in the electronic supplementary material.

(d) . Statistical analyses

To examine the effect of light environments (low-light versus high-light) on development time, we used generalized linear models with Poisson distributions. For wing size, chromatic and achromatic contrasts, indole, ferrulactone and total sex pheromone abundance (sum of indole and ferrulactone abundances), differences were examined using linear models. Assumptions of data normality and homoscedasticity were examined visually using ‘sm.density’ function and using qq-plots. Pairwise differences were examined using the ‘emmeans’ function. All statistical analyses were performed using R statistical software (version 3.6.2).

3. Results

(a) . Development time and wing lengths

We found no difference in the development time (z = −0.405, p = 0.686; figure 1a) or the wing lengths (F_2,151 = 2.67, p = 0.07; figure 1b) of wild, low-light and high-light males (table 1).

Figure 1.

(a) Development time (b) wing lengths and (c) chromatic contrast (d) achromatic contrast expressed in just noticeable differences units (JND units) of low-light (LL), high-light (HL) and wild (W) P. rapae males. Error bars are 1 s.e. Pieris rapae illustration source: Das kleine Schmetterlingsbuch: Die Tagfalter, Insel-Bücherei No. 213.

Table 1.

Estimates and sample sizes of measured traits.

trait	treatment
	low-light	high-light	wild
development time (days)	20.35 ± 0.20	20.14 ± 0.20	n.a.
	(N = 162)	(N = 161)
wing length (cm)	2.30 ± 0.01	2.25 ± 0.01	2.23 ± 0.03
	(N = 65)	(N = 61)	(N = 28)
chromatic contrast (JND units)	11.11 ± 0.35	11.89 ± 0.32	13.85 ± 0.49
	(N = 65)	(N = 61)	(N = 28)
achromatic contrast (JND units)	13.32 ± 0.30	16.12 ± 0.16	16.08 ± 0.29
	(N = 65)	(N = 61)	(N = 28)
sex pheromone abundance	0.43 ± 0.07	0.43 ± 0.03	0.55 ± 0.09
	(N = 44)	(N = 43)	(N = 26)
indole abundance	0.12 ± 0.01	0.0006 ± 0.0006	0.07 ± 0.02
	(N = 44)	(N = 43)	(N = 26)
ferrulactone abundance	0.31 ± 0.06	0.32 ± 0.03	0.48 ± 0.08
	(N = 44)	(N = 43)	(N = 26)

(b) . Wing coloration

We found that chromatic contrast of wild males was significantly higher than the chromatic contrast of low-light (t = −4.529, p < 0.001) and high-light males (t = −3.216, p = 0.004). There was, however, no difference in chromatic contrast of low-light and high-light males (t = −1.625, p = 0.23; figure 1c). This suggests that wing colours of wild males appear more conspicuous to P. rapae females than those of high- and low-light males. When we examined differences in achromatic contrast, we found that low-light males had significantly lower achromatic contrast than wild (t = −6.429, p < 0.001) and high-light males (t = −8.268, p < 0.001), but achromatic contrast of wild and high-light males was similar (t = 0.910, p = 0.995; figure 1d) indicating that wings of low-light males appear significantly duller to P. rapae females compared to wild and high-light males (table 1).

(c) . Pheromone abundances

There were no differences in the total sex pheromone abundance (sum of indole and ferrulactone abundance; F_2,110 = 2.741, p = 0.068) and ferrulactone abundance (F_2,110 = 1.813, p = 0.167) of wild, low-light and high-light males (figure 2a,b,d). However, indole abundance of high-light males was significantly lower than wild (t = −3.660, p = 0.001) and low-light males (t = 6.639, p < 0.001). There was no difference in indole abundance of wild and low-light males (t = 2.080, p = 0.0987; figure 1d; table 1).

Figure 2.

(a) Total sex pheromone abundance (b) indole abundance (c) ferrulactone abundance (d) three signalling phenotypes of P. rapae males separated across achromatic contrast and indole abundance axes. Grey squares, yellow circles and green squares are low-light (LL), high-light (HL) and wild (W) P. rapae males, respectively. Error bars are 1 s.e. Pieris rapae illustration source: Das kleine Schmetterlingsbuch: Die Tagfalter, Insel-Bücherei No. 213.

4. Discussion

To examine how developmental plasticity affects multimodal sexual signal development, we reared caterpillars of P. rapae under different light environments. We found that light environment had no effect on development time and only a weak effect on wing lengths. We also found that both low-light and high-light males had similar but lower chromatic contrast than wild males, indicating the possible role of other factors such as artificial diet and lab conditions in promoting plasticity in sexual signals. Interestingly, however, exposure to different light environments did result in a plastic trade-off between wing brightness and indole abundance in low-light and high-light males, generating two novel phenotypes. Together, these results show that by producing a diversity of signalling phenotypes, developmental plasticity may open avenues for divergence in new habitats [2,4,5].

Exposure to novel habitats is one avenue that can facilitate divergence via signal change (see ‘ecological speciation’ [24]). For instance, increasing evidence indicates that P. rapae are colonizing shaded forests and using garlic mustard, Alliaria petiolata, as a host plant. Alliaria petiolata grows in highly shaded habitats and P. rapae frequently lay eggs on them [21,22]. Caterpillars developing on shade-tolerant A. petiolata may experience light conditions similar to our low-light environment (photon flux = ∼390 µmol s⁻¹ m⁻²) [22]. This can lead to males developing signals similar to our low-light phenotype (dull wings and high indole). Furthermore, in such shaded habitats, females may be unable to evaluate differences in wing coloration due to ambient light conditions [25]. Such a scenario may favour phenotypes with higher indole abundances (low-light phenotypes) over high-light phenotype, as indole will reduce chances of females remating with other males. If populations continue to inhabit shaded habitats (see ‘natal habitat preference induction’ [26]), over time, divergent selection could result in reinforcement and divergence [27].

Variation in female choice is yet another avenue that can contribute to divergence. Although we currently know little about variation in P. rapae female choice, there is ample evidence of inter and intra-population variation in choice in other taxa [3,28]. Access to novel signalling phenotypes can provide females the opportunity to exercise pre-existing variation in their choice, which may favour novel phenotypes. This can result in disruptive selection, which can facilitate early stages of divergence [27].

In the colonization example above, it is important to note that divergence occurs because higher indole abundance is favoured in shaded environments. Focus on one component (e.g. wing colours) and the lack of examination of other components (e.g. pheromones) could lead to underestimation of the role developmental plasticity can play in divergence. This reaffirms the need to examine plasticity across components [29], especially since multimodality in sexual signals is a norm rather than an exception. Although plasticity in a single component can drive divergence, simultaneous plasticity in multiple components has a greater potential to generate a diversity of novel phenotypes, especially if magnitude and direction of plasticity (like the trade-off observed between colours and pheromones in this study) differs across components [2]. Such rapid generation of novel signals can present greater opportunity for divergence to occur via either variable female choice or exposure to novel environments.

In summary, we highlight the need for examining plasticity across components of sexual signals to accurately gauge how developmental effects could facilitate divergence. We also show that developmental effects can influence multiple signal components to rapidly increase the diversity of signalling phenotypes. Taken together, we show that developmental plasticity can indeed be a powerful mechanism promoting divergence via changes in sexual signals.

Ethics

Experimental work on non-native invertebrates such as cabbage white butterfly Pieris rapae does not require IACUC approval. However, utmost care was taken to ensure that all individuals were handled with care and sacrificed humanely.

Data accessibility

Data supporting the findings highlighted in this article have been uploaded as electronic supplementary material [30].

Authors' contributions

A.M.Z.: conceptualization, data curation, formal analysis, funding acquisition, investigation, methodology, project administration, resources, software, supervision, validation, visualization, writing—original draft, writing—review and editing; L.B.: data curation, investigation, methodology, project administration, writing—review and editing; J.S.: data curation, formal analysis, investigation, methodology, project administration, writing—review and editing; A.D.H.: funding acquisition, investigation, methodology, project administration, resources, software, supervision, validation, writing—review and editing; E.C.S.-R.: conceptualization, formal analysis, funding acquisition, investigation, methodology, resources, software, supervision, validation, writing—review and editing.

All authors gave final approval for publication and agreed to be held accountable for the work performed therein.

Conflict of interest declaration

We declare we have no competing interests.

Funding

This study was funded by Department of Ecology, Evolution, and Behaviour, University of Minnesota, Research Award to A.M.Z. A.D.H. and J.S. were supported by funds from the Luby Family Honeycrisp Endowed Chair for Fruit Crop Innovation.