Thistledown velvet ants in the Desert Mimicry Ring and the evolution of white coloration: Müllerian mimicry, camouflage and thermal ecology

Abstract

Adaptive coloration among animals is one of the most recognizable outcomes of natural selection. Here, we investigate evolutionary drivers of white coloration in velvet ants (Hymenoptera: Mutillidae), which has previously been considered camouflage with the fruit of creosote bush (Larrea tridentata). Our analyses indicate instead that velvet ants evolved white coloration millions of years before creosote bush was widespread in North America's hot deserts. Furthermore, velvet ants and the creosote fruit exhibit different spectral reflectance patterns, which appear distinct to potential insectivorous predators. While the white coloration in velvet ants likely did not evolve as camouflage, we find that white-coloured species remain cooler than their red/orange relatives, and therefore we infer the white coloration likely evolved in response to Neogene desertification. This study shows the importance of cross-disciplinary investigation and of testing multiple hypotheses when investigating evolutionary drivers of adaptive coloration.

1. Introduction

Some of the most striking and recognizable outcomes of natural selection include the great diversity of animal coloration. While adaptive colour patterns such as aposematism and camouflage are often linked to predator/prey interactions [1], colour patterns also evolve as a means of intraspecific communication [2] and thermoregulation [3], and these causes are not mutually exclusive. For example, the aposematic coloration of Heliconius butterflies is also used in mate recognition [4,5]. Furthermore, because predators often have different colour vision from humans, assumptions about specific coloration patterns can be incorrect. Crab spiders (Thomisus), for instance, are often thought to be camouflaged with the flowers on which they hunt [6], but recent studies suggest that they reflect more UV light than the flowers, providing a contrast that actually attracts their prey [7,8]. To better understand the evolution of unique coloration patterns in animals, it is necessary to investigate multiple possible evolutionary factors.

Among patterns of animal coloration, aposematism has always drawn attention from scientists and students of biology because of the bright, contrasting colours and the association with both toxicity and mimicry [1]. Common aposematic signals include contrasting patterns of black and yellow, orange, or red [9], and some of the most well-known examples include toxic butterflies [10–12], poison dart frogs [13], venomous snakes [14] and velvet ants [15,16]. While aposematism deters predation through warning coloration [1], camouflage is another common adaptation that is effectively used by both predators and prey to avoid detection [17]. Examples of camouflage used by predators include many members of the family Viperidae [18] and big cats in the family Felidae [19,20]. Camouflaged prey species include many species of caterpillars [21] and various katydid species [22]. Other taxa, like chameleons, use their coloration for camouflage as well as intraspecific communication [2].

Despite the range of colour patterns in the animal kingdom, there are relatively few examples of white being an adaptive colour outside of arctic environments [23], with the exception of pale coloration that is thought to have evolved as camouflage in some animals that live on light sandy surfaces [24]. White coloration, particularly when paired with black, can also be aposematic, such is the case with most species of skunk (Mephitidae) [25], some species of spiny plants (Asteraceae) [26] and some wasps like the bald-faced hornet (Dolichovespula maculata) [27]. White or pale coloration can also play a role in thermoregulation in arid adapted organisms [28]. Several North American velvet ants in the genus Dasymutilla exhibit white or pale coloration, while many other species in the same genus are aposematic, yet the evolutionary and ecological drivers of light colour patterns in velvet ants are not understood.

Velvet ants (Hymenoptera: Mutillidae) are parasitic wasps with wingless females that occupy many biomes all over the globe and are known to participate in large and diverse Müllerian mimicry complexes [15,16,29]. In North America, members of the Desert Mimicry Ring are characterized by having long pale or white hairs covering most of their bodies, often with black legs or peripheral markings [15,16]. One of the most recognizable species, which has been called the ‘thistledown velvet ant,’ is Dasymutilla gloriosa (Saussure). This species and the other two thistledown species, D. pseudopappus (Cockerell) and D. thetis (Blake), are unique within the velvet ant Desert Mimicry Ring, in being completely white, with no black markings on their bodies (figure 1a).

Figure 1.

(a) Dasymutilla gloriosa (the thistledown velvet ant) next to a fruit of creosote bush. (b) External temperatures (with s.e.) for a velvet ant in the Western Mimicry Ring (D. scitula), a member of the Desert Mimicry Ring (D. pseudopappus), and background sand. (c) Internal temperatures (with s.e.) for a velvet ant in the Western Mimicry Ring (D. vestita) and a member of the Desert Mimicry Ring (D. gloriosa). (d) Time-calibrated phylogenetic tree from Williams [30] with ancestral state reconstructions showing probable ancestral coloration. Nodes with a Bayesian posterior probability (bpp) above 50% are shown as solid coloured circles, indicating the ancestor at this node was likely to exhibit that colour pattern (nodes with bpp lower than 50% are show as a pie chart displaying uncertainty of coloration).

It has been widely assumed that thistledown velvet ants mimic fruits of the creosote bush (Larrea tridentata [Sesse and Moc. ex DC] Coville) as a form of camouflage [31–33] (figure 1a). Creosote bush, however, evolved in South America and only migrated to North America relatively recently, likely in the mid-Pleistocene (reviewed by [34]). Therefore, it is unclear if thistledown velvet ants evolved their white colour as a creosote camouflage or if different selective pressures are responsible. The fact that these velvet ants participate in a large mimicry complex raises the possibility that their white coloration could be related to a unique version of warning coloration. Most white velvet ant species, however, are closely related to aposematically coloured yellow or orange species [15], suggesting that factors other than aposematism may have driven the evolution of white coloration. An alternative explanation for the evolution of white coloration is related to thermal ecology. Because these white species are found in some of the warmest and most arid parts of North America, the white coloration could be an adaptation associated with thermoregulation.

Here we use phylogenetic hypotheses with ancestral state reconstructions, along with body temperature and reflectance data (processed through three different visual systems) to ask the following questions: (i) did the white coloration of velvet ants in the Desert Mimicry Ring evolve to mimic the fruits of creosote bush? (ii) does the white colour pattern have thermoecological benefits compared to related red/orange species in the Western Mimicry Ring?

2. Material and methods

(a). Ancestral state reconstruction

To explore the phylogenetic history of velvet ant coloration, we trimmed the phylogeny estimated by Williams [30] to focus on the genus Dasymutilla. Each terminal taxon on the phylogeny was assigned a colour representing the associated Mimicry Ring [15,16], with polymorphism assigned multiple states. Ancestral trait reconstruction was performed using RASP v. 2.0b [35] with the Bayesian Binary MCMC analysis (BBM). The F81 model of evolution was implemented and different rates of change among ancestral states were allowed to reduce constraints, and default values were selected for all other parameters.

(b). Reflectance spectrometry

Spectral reflectance data of white velvet ants in the Desert Mimicry Ring were obtained using an Ocean Optics SD2000 spectrometer with a WS-1 diffuse white reflectance standard and a PS-2 xenon light source. Because live specimens are hard to consistently collect in large numbers, pinned preserved specimens of D. gloriosa (n = 5), D. nocturna (n = 5), D. pseudopappus (n = 5) and D. sackenii (n = 5) were analysed by focusing the coaxial probe at a distance of 2 mm above the abdomen on the surface of the second tergite (the largest abdominal segment) of each specimen. Reflectance data of fresh creosote fruits (n = 9) were similarly obtained. The mean (±s.e.) peak reflectance (λ_max) of each species and an average reflectance spectrum were calculated.

(c). Visual modelling

Because the predators of velvet ants are not known with certainty, we took advantage of visual models in the R package pavo v. 2 [36] to examine the appearance of focal velvet ants and fruits using three colour models: dichromatic (from the dog, Canis familiaris), trichromatic (ornate dragon lizard, Ctenophorus ornatus) and tetrachromatic (generic avian). We do not assume that those models have specific biological relevance to velvet ants, rather we use them to cover a range of perception. Smoothed curves (span = 0.2) were processed using the vismodel function (with ‘canis’, ‘ctenophorus’ and ‘avg.uv’ specified, but other options left as defaults) and then the colspace function (with ‘di’, ‘tri’ and ‘tcs’ specified, leaving other defaults). Finally, projplot was used to convert the three-dimensional avian space into a two-dimensional projection.

(d). Temperature analyses

Temperature measurements were obtained from D. gloriosa and D. pseudopappus and related orange members of the Western Mimicry Ring (D. scitula and D. vestita). External temperatures were measured using a FlirOne thermal imaging camera. Preserved specimens of D. pseudopappus (n = 5) and D. scitula (n = 5) were removed from their pins and placed into a Petri dish filled with sand. The Petri dish was then placed 25 cm beneath a 75 w incandescent heat lamp (Exo Terra Heat-glo infrared spot lamp). Because incandescent bulbs do not provide even lighting at the macro scale, we marked the location with the brightest light and placed the Petri dish in the same location for each trial. External temperatures were taken every minute for 5 min.

Because external body temperatures do not necessarily reflect internal temperatures (most relevant to physiological processes), we used a K/J thermocouple (Sper Scientific 800007) to measure internal abdominal temperatures of preserved velvet ant specimens. The abdomens of preserved specimens of D. gloriosa (n = 7) and D. vestita (n = 7) were removed by separating the majority of the metasoma from the petiole (between the first and second metasomal segments). This left a fairly intact abdomen with a small opening (approx. 1 mm) at its base in which we inserted the probe. Because of the destructive nature of this procedure, we used D. gloriosa to represent the Desert Mimicry Ring and D. vestita to represent the Western Mimicry Ring as these species are well represented in our collection. Each dissected abdomen was then placed 25 cm under a 75 w incandescent heat lamp (suspended on the thermocouple probe) making sure to place the specimen under the brightest part of the lamp. Internal temperature measurements were taken every minute for 15 min.

3. Results

(a). Ancestral state reconstruction

Based on the well-supported phylogeny from Williams [30], the genus Dasymutilla originated around 21 Ma (figure 1d). Ancestral state reconstructions indicate that the early Dasymutilla ancestors had Madrean colour patterns but quickly evolved the tropical pattern. The reconstruction suggests the Desert (white) colour pattern first evolved around 6 Ma (figure 1d), likely in more than one clade, although the number of times the trait evolved is less important for the questions being asked here as compared to the timing of trait evolution relative to the arrival of creosote bush.

(b). Reflectance spectrometry and visual modelling

All of the velvet ants (D. gloriosa, D. nocturna, D. pseudopappus and D. sackenii) had similar reflectance spectra, with light broadly reflected across wavelengths from 300–700 nm (figure 2a), which includes ultraviolet light and ‘visible’ light. In other words, they are truly white but with variation among species. The creosote fruits, however, showed a different reflectance pattern, with less UV light reflected and more light at higher wavelengths (figure 2b). Similarly, when modelled under vertebrate visual models, creosote fruits were visually distinct from the velvet ants with the exception of D. sackenii, which tended farthest from white (figure 2c–e).

Figure 2.

Spectral reflectance curves with s.d. for (a) four velvet ant species and (b) creosote fruit, and the same reflectance data visualized with three visual systems: (c) trichromatic (note that the larger triangle is an expansion of the full visual space shown in the upper left smaller triangle), (d) tetrachromatic (using two-dimensional Mollweide projection of the three-dimensional space) and (e) dichromatic. In the trichromatic graph, whiter colours are towards the lower left of the larger triangle; in the tetrachromatic projection, whiter colours are at the lowest point; and on the dichromatic line whiter colours are to the left (‘s’, ‘m’ and ‘l’ in (c) and (e) refer to short, medium and long wavelengths). See main text for additional details on visual models and associated model organisms.

(c). Temperature analyses

In our experiments, white members of the Desert Mimicry Ring remained cooler, both internally and externally, relative to orange members of the Western Mimicry Ring (figure 1b,c).

4. Discussion

We find that members of the velvet ant Desert Mimicry Ring did not evolve their white coloration to mimic the fruits of creosote bush. Instead, our data suggest the white coloration evolved as an adaptation to desert environments and was then potentially reinforced through Müllerian mimicry. The well-supported phylogenetic and ancestral state reconstructions show that velvet ants evolved white coloration around 6 Ma, before creosote bush arrived in North America in the mid-Pleistocene (approx. 1.5 Ma) [34]. Even intraspecific genetic analyses within D. pseudopappus and D. gloriosa indicate that these individual species were white earlier than 2 Ma [30]. While recent adaptive introgression between species has been reported from other mimetic taxa [37], this seems unlikely in the distantly related members of the Desert Mimicry Ring. Furthermore, even though creosote fruits and white velvet ants appear similar to humans (figure 1a), they have different reflection profiles, and for the most part appear distinct under modelled non-human visual systems (figure 2), some of which have broader spectral vision that can see near ultraviolet light [38]. One exception is D. sackenii, which has more of a yellow hue compared to many other white velvet ants. There is a possibility that invertebrate predators, like spiders, might have visual systems distinct from those modelled here, but this will have to await further study.

Instead of predation pressure, we suggest that white coloration is an adaptation to hot desert environments. The white coloration of the Desert Mimicry Ring evolved from a red/orange ancestral state at least once but possibly multiple times around 4–6 Ma (figure 1d). Many parts of the North American deserts were becoming warmer and dryer at that time owing to mountain building events across the west [39]. All 17 members of the Desert Mimicry Ring are endemic to North America's hot deserts (Mojave, Sonoran, Chihuahuan) and are most common in the hottest parts of these deserts [15,16]. Furthermore, most of the species in this Mimicry Ring are active during the hottest part of the year (May–September). Both internal and external temperatures of the white species remain cooler compared to orange species (figure 1b,c), which could be associated with longer foraging and a selective advantage to being white in the hot deserts. Once the light hairs were successful through thermoregulatory advantage, Müllerian mimicry could have reinforced the common visual signal and facilitated diversification in the Desert Mimicry Ring (figure 1d).

5. Conclusion

Evolutionary biologists have long recognized that we cannot assume origins of traits based on superficial similarity of extant species, and this is especially true when sensory systems of non-human animals are involved. Our study illustrates the power of combining phylogenetic inference with experimental measurements to elucidate the origin of a conspicuous phenotype in velvet ants. Rather than evolving as crypsis, it seems more likely that the white hairs of the velvet ants and the white creosote fruits have evolved as a kind of inter-kingdom convergence on an desert thermoregulatory strategy [15,16], which is all the more remarkable because velvet ants are otherwise known for their contrasting aposematic coloration. Future analyses should investigate the length of the hairs both in velvet ants and in creosote fruits to see if hair length also has a thermoecological benefit. We hope our results inspire other cross-disciplinary investigations and broaden our knowledge of velvet ants as some of the most remarkable and diverse insects.

Ethics

No requirements for ethical approval were needed for this study as it primarily involved preserved insect specimens.

Data accessibility

Data available from the Dryad Digital Repository: https://dx.doi.org/10.5061/dryad.dr7sqv9vw [40].

Authors' contributions

The experiments were designed by J.S.W., J.S.S., K.A.W. and M.L.F. They were conducted by J.S.W., J.S.S., K.A.W. and M.L.F. who were supervised by J.P.P. Specimens and materials were provided by J.S.W. and J.P.P. Analyses and figures were done by J.S.W. and M.L.F. The manuscript was written by J.S.W., J.S.S. and M.L.F., with input from K.A.W. and J.P.P. All authors gave final approval for publication and agree to be held accountable for the work performed therein.

Competing interests

We declare we have no competing interests.

Funding

This work was supported partially by funding through Utah State University.