Increasing antennal sensitivity for enhanced stimulus detection

Given the importance of antennae and their sensilla for insects to sense their environment, strong natural or sexual selection for the improved detection of relevant stimuli is expected [27]. One way to improve stimulus detection is by using more sensors. This could be achieved by increasing the surface area of the antennae [28,29], thus increasing the number of sensilla and their associated sensory neurons on the antennal surface [30], while at the same time increasing the air space that the antennae can sample. The surface area of the antennae can be increased in one of three ways: (1) Increased number of antennal segments, (2) increased length of individual segments, or (3) the addition of side-branches to segments [5,7]. In addition, stimulus detection could be further improved by (4) increasing the density of the sensilla on a given antennal surface area [28,31] and/or by (5) increasing the size of the individual sensilla to maximize interaction with environmental stimuli and thus their sensitivity at threshold levels [11].

Focus on Fireflies

We focus here on the antennal sensilla diversity of adult Lampyridae. Firefly species occur worldwide (except Antarctica) and exhibit a striking range of antennal diversity [4,32]. This diversity includes both antennal shape and antennomere number. Within Coleoptera, the typical number of antennomeres is eleven [33], and indeed most beetle families have this fixed number across all species [3]. Among fireflies the number ranges from 7–62 antennomeres, with eleven being the most common [4]. Within-species variation is known to occur in several firefly genera, including Alecton, Microphotus, Pleotomus, and genera of Amydetinae [4]. To date, antennal sensilla diversity in fireflies has only been studied in the bioluminescent males of one firefly species, Lu. cruciata [26] from Asia. Iwasaki et al. [26] described seven sensilla morphotypes, including four mechanosensilla, two chemosensilla and one hygrosensillum. It is unknown whether the morphotypes of Lu. cruciata are representative of other firefly species and whether there are differences in sensilla types, numbers, or densities between males and females and/or between bioluminescent (nocturnal) and non-bioluminescent (diurnal) firefly species. Filling this gap is important, because the absence or presence of adult bioluminescence has implications for how firefly antennae are used.

Nocturnal firefly species are active at dusk or at night and are bioluminescent. They use prolonged glows [34] or species-specific flash patterns [35–37] as visual signals to attract and recognize conspecific mates. In contrast, most diurnal firefly species have non-bioluminescent adults that rely exclusively on pheromones to identify and locate a conspecific mate. In both signaling systems males actively look for females during mate search, while females are sedentary, and correlated with these behaviors, males have significantly larger eyes than their conspecific females [38]. Furthermore, bioluminescent males that navigate through vegetation at low light levels to locate their bioluminescent females, have significantly larger eyes than the males of diurnal species [38]. In contrast, diurnal males have significantly longer antennae than their conspecific females and usually also significantly longer antennae than the males of bioluminescent species [38], reflecting the importance of male antennae for the detection of pheromones in diurnal species.

It is currently unknown whether this antennal size dimorphism is also reflected in antennal sensilla diversity, and/or in the number and density of different sensilla types, including olfactory sensilla that are important for pheromone detection. We also presently do not know to what extent bioluminescent species retain the use of long-range pheromones used during mate search, (e.g., to lead males close enough to females, so both can detect and respond to the light signals of their potential mates). During firefly evolutionary history, there were several independent reversals from nocturnal activity with use of light signals to diurnal activity with pheromones as the main mating signal [39–41], and such reversals also took place several times within the genus Photinus [37]. This suggests that bioluminescent species likely retain the ability to use pheromones, at least to some degree, facilitating the reversal to exclusive pheromone use. Combinations of pheromone and bioluminescent signaling has been indicated at least once in the following genera; Cyphonocerus, Pyrocoelia, Erthrolychnia, Phaenolis, Phausis, Pleotomus, and Lamprohiza [36,39–43]. Field observations suggest that the nocturnal Blue Ghost firefly, Phausis reticulata, may use both light signals and pheromones during mate search [34], and the use of both pheromones and bioluminescence has been recently confirmed with field experiments for Lamprohiza splendidula (Linnaeus, 1767) [43]. In addition, the unusual diurnal, but bioluminescent, firefly species Phosphaenus hemipterus (Geoffroy, 1762) uses pheromones as the primary sexual signal and its faint bioluminescent glow as an aposematic defense signal, rather than for mating [44,45].

While the bioluminescent mating signals of nocturnal species have been studied extensively, so far little is known about the pheromone signaling system in fireflies. The specific chemicals produced by female fireflies have only been identified for a hand full of species; Pyrocoelia oshimana Nakane, 1985 [46], two unidentified Diaphanes species, Pyrocoelia praetexta E. Olivier, 1911, and Lamprigera tenebrosa [Walker, 1858] [47]. However, functional female sex pheromones have only been identified for one species, the diurnal winter firefly P. corruscus [24]. The specific antennal sensilla that respond to sex pheromones in P. corruscus, have recently been identified as sensilla basiconica [24]. Once in close physical contact, both bioluminescent and non-bioluminescent fireflies will “antennate” each other before mating, suggesting the sampling of contact chemicals, possibly cuticular hydrocarbons (CHCs), with gustatory sensilla for a final verification of a conspecific mate. South et al. [48] found CHCs on the pronotum and elytra of the diurnal species P. corruscus and Lucidota atra (G. Olivier, 1790), but found only low or undetectable levels of CHCs in the nocturnal species Photinus greeni Lloyd, 1969, P. ignitus Lloyd, 1969, and P. obscurellus LeConte, 1851. P. corruscus males were able to distinguish between the CHCs of conspecific and heterospecific (L. atra) females, suggesting that CHCs are possibly used for reproductive isolation in this species [48].

In this study we document the morphological sensilla diversity of both males and females of seven species of fireflies, including three diurnal and four nocturnal species. Given the sensilla diversity across Coleoptera (in morphology and naming), and to make the findings for our study species more directly comparable to related taxa, we use the naming scheme for basiconica sensilla developed by Faucheux et al. [17] who synonymized sensilla nomenclature across Elateridae, a beetle family closely related to Lampyridae. In the present study, sensilla that share the same morphology with morphotypes in Faucheux et al. [17] were given the same morphotype number as those in the Faucheux et al. [17] study. Other chaetica, campaniform, coeloconica, and trichoid sensilla morphotypes were named/numbered in the order they were observed in this study.

An interesting challenge was the identification of pheromone sensilla candidates in the absence of functional testing with electrophysiology. The first pheromone sensilla for fireflies were just recently identified in P. corruscus [24] and we used this opportunity to test our set of predictions for identifying potential sex pheromone sensilla in fireflies based on external sensilla morphology, sensilla counts and sexual dimorphism in firefly behavior.

We hypothesized that the differences in mating signals between diurnal and nocturnal fireflies will be reflected in their antennal sensilla counts and possibly in their sensilla densities. Given the importance of pheromones for diurnal firefly species, we predicted to find (1) more chemosensilla in diurnal species. If nocturnal species have completely lost the ability to use pheromones during mate search, we would expect to find (a) at least one sensilla type among the chemosensilla that is only present in diurnal fireflies; if nocturnal species retain the ability to use pheromones we would expect (b) at least one sensilla type among the chemosensilla that is present in greater numbers and possibly greater density on the antennae of diurnal species, compared to nocturnal species. We further predict that (2) males have more chemosensilla than females, and especially the diurnal males, that rely on tracking a pheromone plume to conspecific females, will have greater numbers, and possibly a greater density, of pheromone chemosensilla than their females. Finally, (3) pores in the sensilla surface (terminal tip pores or wall pores) are necessary to allow volatile chemicals to enter the internal olfactory sensilla space and bind to receptor proteins on sensory neurons [49], therefore any candidate for a pheromone sensilla should have pores. In the case of gustatory chemosensilla that pick-up contact chemicals during antennation, we would not predict a difference between diurnal and nocturnal fireflies, since both diurnal and nocturnal species engage in antennation behavior before mating. We also would not predict a sex difference in gustatory chemosensilla since both sexes likely use CHCs for mate recognition; this is suggested by the lack of sex differences in diurnal firefly CHC profiles [46]. In order to detect these contact chemicals, gustatory sensilla likely have a single terminal pore and may or may not have pores on the sensillum wall [50], and based on the antennation behavior, possibly be concentrated on the ventral surface of the more distal antennomeres. Our proposed pheromone and gustatory chemosensilla candidates will facilitate future functional testing of different sensilla morphotypes, an important next step towards understanding how fireflies use their antennae to perceive their world.

Taxon sampling

We selected firefly species available in the Stanger-Hall Lab Lampyridae collection (no collection permits were required for the original collection of the North American species) and in the Bybee Lab (Luciolinae sp., collection with permission from Cameroon’s Ministry of Forests and Wildlife). The seven firefly species in this study were selected to represent systematic diversity (i.e., six genera in four subfamilies: Luciolinae, Lamprohizinae, Photurinae, and Lampyrinae), sexual signal diversity (i.e., bioluminescent: glows or flashes and non-bioluminescent), and differences in dial activity (i.e., nocturnal and diurnal). The seven species in this study include four bioluminescent species (in four genera): three flashing species: Photinus pyralis (Linnaeus, 1767), Photuris lucicrescens Barber, 1951, Luciolinae sp. (an unidentified species from Africa in the Luciolinae subfamily), and one glowing Phausis species, discovered as a new species by Sarah Lower [51] (voucher specimens KSH#8663 and KSH#8667 are stored in the Stanger-Hall lab at UGA), and recently named Phausis christineae Hodson, 2024 [52], as well as three non-bioluminescent species (in three genera): Lucidota punctata [LeConte, 1852], Pyropyga nigricans [Say, 1823], and Photinus corruscus (Linnaeus, 1767) (formerly Ellychnia corrusca [53]). Species were determined to be bioluminescent or non-bioluminescent as adults by the presence or absence of a light organ on abdominal ventrites 5/6 or 6/7; flashing and glowing bioluminescent species were differentiated based on field observations and records in the literature (Table 1). The antennae of three males and three females of each species were examined. Vouchers are kept in the Stanger-Hall lab at the University of Georgia (Table S1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Taxon sampling. The antennae of 3 males and 3 females of each species were examined, BL = bioluminescent, LO = light organ.

https://doi.org/10.1371/journal.pone.0323722.t001

Scanning electron microscopy (SEM)

The 42 specimens (3 males and 3 females of 7 species) were prepared by removing both the left and right antennae from the head of each specimen. Antennae of the first several specimens were washed using soapy water and an ultrasonicator to remove debris. This method was effective for several specimens, but damaged others. Damaged specimens were not further examined and replaced with a new specimen, which was washed with a different method (see Table 1; SEM was method): antennae were placed in a 0.01% solution of KOH at 50^o C for 2 hours. There was no difference in the structure of the antennal cuticle observed between the two cleaning methods. The antennae were then rinsed in distilled water and allowed to air dry for ~10 minutes (both methods), before mounting them on SEM stubs, with one antenna facing dorsal side up and the other facing ventral side up, providing both a dorsal and a ventral view of the antennae for each individual firefly. Antennae were further air dried on their stubs for a minimum of 72 hours to ensure even sputter-coating. The antennae were sputter-coated with 30 nm of gold using the Lecia ACE600. Sputter-coated antennae were then imaged using the FE-SEM Thermo Fisher Teneo with an EDT and T1 (2kV-5kV) detectors at the Georgia Electron Microscopy laboratory. For each individual antenna, SEM images were obtained for each individual antennomere. Each image contained a scale bar (generated by the Teneo software) for subsequent antennal area measurements.

Antennal area

All taxa sampled in this study have filiform or moderately serrate (L. punctata, Py. nigricans) antennae with 11 antennomeres (Fig 1), except Pha. christineae females which are paedomorphic (lacking both elytra and metathoracic wings but with paired pretarsal claws and a pair of stemmata [54] and have only three antennomeres. The area of each antennomere was measured (in mm²) using the polygon selection tool in ImageJ v1.52 [55] and the scale bars produced during imaging. The area of all (11 or 3) antennomeres was then summed for each specimen to determine the total antennal area for each side of the antenna. The area measurements for the dorsal and ventral sides of each individual antenna were then added to determine the total antennal area (per antenna) of each specimen. Since larger fireflies tend to have larger antennae, we examined the influence of body size on antenna size between our study taxa by measuring pronotum length as a proxy for body size [38] and used scaled antennal area (antenna area/pronotum length) to compare antennal areas between fireflies (note: due to the lack of pronotal expansions in Luciolinae, pronotum length will underestimate body size for Luciolinae compared to other fireflies). Pronotum length was measured from the anterior to the posterior edge along the midline of the pronotum. To visualize the relationship between pronotum length (body size) and total antenna area across the 4 nocturnal and 3 diurnal species in our study, we plotted antennal area against pronotum length for all 42 individuals in our analysis and labeled our samples as diurnal/nocturnal and male/female.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Diagram of adult firefly antennae with 11 antennomeres.

Scape and pedicel represent the first two antennomeres, antennomeres III – XI are flagellomeres. (A) filimorm antennae; (B) serrate antennae. 

https://doi.org/10.1371/journal.pone.0323722.g001

Sensilla morphotypes, diversity, and evolution

To describe sensilla morphotypes within and across our focal firefly species, we used key morphological characteristics as defined by Schneider [1] for different insect sensilla morphotypes. These characteristics included: shape of sensilla base (raised or not raised), shape of stalk (equal width throughout length or unequal width), length of stalk relative to base (stalk length equal to base or stalk length > 1x base length) and the presence or absence of grooves and/or pores. These morphological characteristics were used to separate sensilla morphotypes into the 9 major groups defined by Schneider [1]. In addition, we consulted Faucheux et al. [17] who examined the sensilla morphology of Agriotes (Elateridae) and synonymized sensilla nomenclature across Elateridae, a beetle family closely related to Lampyridae [56,57]. We used the Elateridae naming scheme to classify each of the basiconica sensilla morphotypes for Lampyridae. Sensilla that share the same morphology with morphotypes in Faucheux et al. [17] were given the same morphotype number as those in the Faucheux et al. [17], basiconica sensilla morphotypes that were not previously described or named, were named using the next consecutive number(s) of the Faucheux naming scheme [1]. All other sensilla morphotypes (chaetica, campaniform, trichoid), and other unclassified morphotypes were named based on the order they were observed in this study. We report all sensilla morphotypes found in males and females of each species.

To identify a potential phylogenetic influence on the distribution of sensilla morphotypes in our seven study species and Lu. cruciata [26], we plotted all sensilla morphotypes on a cladogram of all 8 species (pruning all other species from the phylogeny of Martin et al. [56]).

Sensilla diversity measures

Sensilla richness, or the number of sensilla types present, was counted for each specimen. To compare sensilla diversity across specimens and across firefly species we used the proportions of each sensilla type present (p_i) on the antenna of a given specimen to calculate two different diversity indices that provide different measures of diversity. The general Simpson dominance index (D) is calculated as D = ∑(p_i)² and favors dominant sensilla types over rarer sensilla types. In contrast, the Shannon Index (H) is calculated as H = ∑[-log(p_i)] (p_i) and favors rare sensilla types, allowing for a more even contribution of each sensilla type to the diversity index [58]. Since the Shannon and Simpson indices for different species cannot be directly compared or combined for analysis, we transformed all index values into their effective numbers (E) using E_D (Shannon)=10^H and E_H (Simpson)=1/D [58]. These effective numbers linearize values calculated from non-linear indices and represent how much more or less diverse in sensilla types one antenna is compared to another. We then used parametric t-Tests for E_D (normally distributed) and non-parametric Wilcoxon rank-sum tests for E_H (not normally distributed) to compare the average effective numbers (E_D) and E_H) of each species to determine if there was a difference in sensilla diversity between diurnal and nocturnal species.

Sensilla counts and sensilla density

For each specimen (three females and three males for seven species = 42 specimens) we counted the total number of sensilla for each sensilla morphotype on each antennomere (both dorsal and ventral side). To calculate the total sensilla number for each morphotype on one antenna we added the counts across the 11 antennomeres (three for Phausis females) of each individual specimen. From these data we generated sensilla counts for two main functional categories (all mechanosensilla and all chemosensilla), as well as the total sensilla count per antenna.

Since the functional verification of sensilla with electrophysiology was beyond the scope of this study (due to extensive costs and required technology), there is a possibility that thermo- and/or and hygrosensilla were included in our chemoreceptor category. However, this possibility made our hypothesis testing more conservative (see below). To determine how closely sensilla were packed on the antennal surface, we calculated sensilla density by dividing the total sensilla counts (per morphotype and category) by the total antennal area of the respective specimen. For summary statistics we used JMP Pro v17.2 [59] to calculate the mean and standard deviation (x̄ ± stdev) of sensilla numbers and sensilla densities for each sex and each species (all sensilla, all mechanosensilla, all chemosensilla, and all individual sensilla morphotypes). To examine how fireflies may optimize the sensitivity of their antennae for different stimuli, we plotted how total sensilla numbers and sensilla densities change on firefly antennae as antenna size increases.

Sensilla distribution

To assess whether different parts of the antennae could be used for different functions (e.g., mechanoreception, chemoreception), we determined how the different sensilla morphotypes were distributed across the length of the antenna of each individual specimen. To visualize this distribution of the sensilla in each functional group and each morphotype, we plotted the mean number of sensilla per antennomere for each sex and each species across the 11 antennomeres.

Statistical analyses

For the statical analysis of sensilla differences between diurnal and nocturnal firefly species, and to test our predictions, we grouped the sensilla morphotypes of our 7 species into mechanosensilla (3 types) and potential chemosensilla (9 types) based on external morphology. Other sensilla types (e.g., X1), which are possible thermo- and/or hygrosensilla [15] were not included in this analysis. Since tissue staining, and especially functional testing were beyond the scope of this study, we used external sensilla morphology and putative functions from the literature to group our sensilla morphotypes into these two broad categories. We expected chemosensilla to have at least one (terminal) pore [50], however, because the gold staining in our SEM preparation obstructed small pores in at least one sensilla type (B1, see results: ‘Chemosensilla’), we included some sensilla morphotypes as chemosensilla, even if no pores were visible. By including these in the chemosensilla category we are making our hypothesis testing more conservative (i.e., less likely to detect significant differences). A possible mix of sensilla functions within our chemosensilla category can be elucidated in future functional studies and used to put our findings to the test.

We used a two-way mixed model ANOVA in JMP Pro v17.2 to test for differences in antenna size, sensilla counts, and sensilla densities between diurnal and nocturnal fireflies and between males and females. Due to our small sample sizes within firefly species (3 males and 3 females each) we used “species” as a random effect (covariate) for all our analyses, with activity and sex as fixed effects. Since males and females of diurnal and nocturnal taxa were predicted to differ for some sensilla types, we also tested for an interaction between sex and activity. For each mixed model analysis, we report the amount of variation explained by the species covariate. All fixed effects tests were based on 1 parameter (Nparm = 1) and 1 degree of freedom for the numerator (DFNum), therefore we only report the degrees of freedom for the denominator (DFDen = error term), the F-ratio (F) for testing that the effect is zero, and the p-value (Prob>F) for the fixed effects (and their interaction) for each analysis. When the interaction term was significant, we tested effect slices (activity by sex, sex by activity) with Student’s t tests to test our predicted differences in chemosensilla (but not mechanosensilla) between diurnal and nocturnal males, and between diurnal males and diurnal females. Since p-values represent a continuum, we report p-values with p < 0.05 as statistically significant, and p-values close to 0.05 as marginally significant.

Since larger fireflies tend to have larger antennae [38], we used a separate mixed model analysis to examine the distribution of body sizes (pronotum length) across our firefly sample and report the percent variation explained by species (random effect). To specifically examine the influence of body size on antenna size across our specimens, we added pronotum length as an additional (continuous) fixed effect to a mixed model analysis with activity and sex (and sex*activity) as fixed effects and species as a random effect.

Antenna size

The antennal areas across our seven species of fireflies varied greatly (Table 2). The smallest antennal area (x̄ ± stdev) was measured in Pha. christineae females (three antennomeres: x̄_Female = 0.02mm² ± 0.001) and males (11 antennomeres: x̄_Male = 0.2mm² ± 0.04). The females of nocturnal Ph. lucicrescens (x̄_Female = 2.6mm²± 0.6) and P. pyralis (x̄_Female = 2.5mm²± 0.2) had the largest measured antennal areas (Table 2), however, when body size was taken into account (scaled antenna size = antennal area/pronotum length), the males of diurnal L. punctata and nocturnal P. pyralis had the largest (scaled) antennal areas (Table 3).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Mean sensilla counts and antennal areas of male and female fireflies.

https://doi.org/10.1371/journal.pone.0323722.t002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Mean sensilla densities and scaled antennal areas of male and female fireflies.

https://doi.org/10.1371/journal.pone.0323722.t003

Body size and antennal area

Our species sample varied greatly in body size (pronotum length; Table 2). In our mixed model analysis, species (covariate) accounted for 95.06% of the variation in body size. Body sizes in our species sample did not differ significantly by activity or sex (or sex*activity). Overall, the antennal areas of the 42 firefly specimens in our study were positively correlated (R = 0.467; Spearman’s rho = 0.668, p < 0.0001) with body size (Fig 2A), which means larger fireflies tend to have larger antennae.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Firefly antennal area.

(A) Relationship between body size (pronotum length) and antennal area: antennal area is positively correlated with body size (R = 0.467; Spearman’s rho = 0.668, p < 0.0001); (B) Comparison of antennal area by activity time: there is no difference in antennal size between diurnal and nocturnal species (x̄_Diurnal = 1.50 ± 0.6 mm², x̄_Nocturnal = 1.28 ± 1.07 mm², DFDen = 5.0, F = 0.077, p = 0.7943); (C) Comparison of antennal area by sex: males have significantly larger antennae than females (x̄_Male = 1.49 ± 0.9 mm², x̄_Female = 1.26 ± 0.9 mm²; DFDen = 33.4, F = 11.58, p = 0.0017).

https://doi.org/10.1371/journal.pone.0323722.g002

In our mixed model analysis of antenna size, species accounted for 77.34% of the variation in antennal area. Body size (pronotum length) had a significant effect on antennal area (DFDen = 17.0, F = 16.65, p = 0.0008). The antennal areas of the diurnal and nocturnal (x̄_Diurnal = 1.50 ± 0.6 mm², x̄_Nocturnal = 1.28 ± 1.07 mm²) fireflies in our study did not differ significantly from each other (mixed model, DFDen = 5.0, F = 0.077, p = 0.7943, Fig 2B), but males had significantly larger antennal areas than females (x̄_Male = 1.49 ± 0.9 mm², x̄_Female = 1.26 ± 0.9 mm²; DFDen = 33.4, F = 11.58, p = 0.0017, Fig 2C). The interaction term (activity*sex) was also significant (DFDen = 32.2, F = 5.39, p = 0.0267) with significantly larger antennal areas in diurnal males compared to diurnal females (x̄_Male = 1.812 ± 0.66 mm², x̄_Female = 1.2 ± 0.38 mm²; Student’s t ratio = -3.86, p = 0.0005, Fig S1).

Sensilla morphology

We identified a total of 14 different sensilla morphotypes across the seven firefly species. Two of these 14 morphotypes were first identified in Lu. cruciata [26]. One morphotype was extremely rare and was not counted, two other morphotypes were found in only one sex, or in only a few specimens within a species. Eleven morphotypes, if present at all, were present in both sexes (Table 4). Based on their morphology [1,15], we identified 3 mechanosensilla and 8 chemosensilla morphotypes among the 11 most common sensilla morphotypes in our 7 species.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Distribution of sensilla morphotypes across species.

https://doi.org/10.1371/journal.pone.0323722.t004

Mechanosensilla.

Sensilla chaetica, also known as the bristle or spine sensilla [1], are slender sensilla arising directly from the antennal cuticle, with no modified ring around the base. These sensilla vary in length, but all gradually taper towards the distal end. The walls of these sensilla can be smooth or bear grooves that run parallel to the length of the stalk. We found two sensilla chaetica variants in fireflies (Fig 3). (1) Sensilla chaetica type 1 (C1, Fig 3A): C1 are long bristle sensilla with grooves running parallel lengthwise along the stalk. These are the longest sensilla found on firefly antennae and were located on all 11 antennomeres. They were distributed evenly across the surface of each individual antennomere. C1 sensilla were found in all seven species examined. (2) Sensilla chaetica type 2 (C2, Fig 3B): C2 sensilla have a short smooth stalk that comes to a distinct point. These sensilla were found in all seven species examined.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Mechanosensilla morphotypes.

(A) mechanosensilla chaetica type 1 (C1) on the antenna of a Py. nigricans female (also pictured are cuticular pores (CP) and chemosensilla basiconica type B3); (B) mechanosensilla chaetica type 2 (C2) of a Py. nigricans male; (C) mechanosensilla campaniform sensilla (SC) of a Luciolinae sp. female (also pictured are sensilla basiconica type B13).

https://doi.org/10.1371/journal.pone.0323722.g003

Sensilla campaniform. We identified one type of sensilla campaniform (SC; Fig 3C). These sensilla consist of a ring slightly raised above the antennal surface, with a raised disc, equal in height, inside the center of the ring. SC sensilla were found in low numbers relative to the other mechanosensilla (Table 4). They were found on the antennae of all species except Pha. christineae; they were absent in male L. punctata and female Ph. lucicrescens (Table 4).

Chemosensilla.

Sensilla basiconica, also known as peg or cone sensilla, are characterized by their raised base with a peg or cone arising from that base. The length and width of the peg varies between morphotypes. Peg and cone walls may or may not have pores and may have grooves that run parallel along the length of the peg [1]. Basiconica sensilla were found in all seven firefly species and we identified six different variants (types): (1) Sensilla basiconica type 1 (B1: Fig 4A) are peg sensilla with a well-defined dome shaped base and a short peg with a single pore at the distal end (Fig 5A). The peg length is ~ 3–4 times the height of the dome base, the width of the peg is uniform throughout the length and the distal end of the peg is rounded. We found a single pore at the apex of the B1 sensilla peg, and no pores along the length of the B1 peg, however B1 sensilla identified by Lower et al. [24] possessed pores along the sides of the peg. B1 sensilla were found in the diurnal (non-bioluminescent) species P. corruscus, Py. nigricans, and the nocturnal (bioluminescent) species P. pyralis. (2) Sensilla basiconica type 2 (B2: Fig 4A) are peg sensilla with a well-defined dome base and a moderate length peg (~4–5 times the height of the dome base). The peg is widest at the base and gradually tapers in width coming to a blunt point at the distal end, no pores were visible on the peg (Fig S2A). B2 sensilla were found in the diurnal species P. corruscus, and the nocturnal species Ph. lucicrescens and Luciolinae sp. (3) Sensilla basiconica type 3 (B3: Figs 3A, 4B) are peg sensilla with a modified dome base, with a flattened top, and a peg (~5 times the height of the base). The width of the peg is equal throughout its length and bears pores in high density along the distal 3⁄4 (Fig 5B). The peg comes to a blunt point at its distal end. B3 sensilla were found in the diurnal species P. corruscus, Py. nigricans, and the nocturnal species P. pyralis. (4) Sensilla basiconica type 7 (B7: Fig 4C, H) are peg sensilla with a dome base and peg (~2–3 times the height of the base). The peg is equal in width for the first 2/3 of its length and then tapers to a distinct point. The last 1/3 of the peg has parallel grooves that run along the length of the peg and meet at the distal point. There were no visible pores on the peg of B7 sensilla. B7 sensilla were found in all studied species except Pha. christineae. (5) Sensilla basiconica type 10 (B10: Fig 4D, E) are cone sensilla with a raised collared base and a broad cone. The length of the cone is 2 times the height of the collared base, pores are present on the cone (Fig 5C). The cone is equal in width to the collared base for the first ½ of the length and then tapers gradually to a point. B10 sensilla were only found in the diurnal species L. punctata. (6) Sensilla basiconica type 11 (B11: Fig 4E) are modified cone sensilla with a collared base and a rounded cone. The length of the cone is ~ 1.5–2 times the length of the base. The proximal end of the cone is equal in width to the base, the sides then gradually diverge, increasing the width of the cone, with a rounded end. The distal end of the cone is broadly rounded to a shallow dome shape, the cone lacks visible pores (Fig S2B). B11 sensilla were only found in the diurnal species L. punctata. (7) Sensilla basiconica type 12 (B12: Fig 4F, G) have a membranous base with a short peg that gradually comes to a point at the distal end which bears a single pore. The length of the peg is about equal to the width at the base; these sensilla were only found in the males of Luciolinae sp. (8) Sensilla basiconica type 13 (B13: Figs 3C and 4H) closely resemble B3 sensilla, but lack pores (Fig S2C); these sensilla were only found in Luciolinae sp.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Chemosensilla.

(A) sensilla basiconica types 1 and 2 (B1 and B2) on the antenna of a P. corruscus female; (B) sensilla basiconica type 3 (B3) of a P. corruscus male; (C) sensilla basiconica type 7 (B7) of a P. corruscus male; (D) sensilla basiconica type 10 (B10) of a L. punctata female; (E) sensilla basiconica types 10 and 11 (B10 and B11) of a L. punctata male; (F) sensilla basiconica type 12 (B12) and cuticular pores (CP) on a Luciolinae sp. female; (G) sensilla basiconica type 12 (B12) with an apical pore (arrow) of a Luciolinae sp. female; (H) basiconica sensilla types 7 and 13 (B7, B13) and cuticular pores (CP) of a Luciolinae sp. female; (I) sensilla trichodea (T1) of a Pha. christineae male.

https://doi.org/10.1371/journal.pone.0323722.g004

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Sensilla pores and unique sensilla.

(A) B1 sensilla with a single pore (arrow) and the sensilla peg slightly compressed into the dome base, on the antenna of a P. corruscus female; (B) multipored (arrows) peg of B3 sensilla of a P. corruscus male; (C) multipored (arrows) cone of B10 sensilla of a L. punctata male; (D) sensilla coeloconica (X1) of a P. corruscus female; (E) cone sensilla, surrounded by (broken) B3 sensilla on the antenna of a P. corruscus female.

https://doi.org/10.1371/journal.pone.0323722.g005

Sensilla trichodea (T1: Fig 4I) are simple hair like sensilla without a modified base, directly arising from the cuticle surface. We identified only one type of sensilla trichodea (T1) in fireflies and exclusively in the nocturnal species Pha. christineae. These sensilla are short, with the width of the stalk at the base ~8 times the width of the stalk at the distal end. The width of the sensilla gradually decreases before coming to a blunt point. In our SEM images they lacked visible pores (Fig S2D), but since pheromone-based male search behavior has been documented for Pha. reticulata [34], and these sensilla are the only potential chemosensilla for Pha. christineae, we included T1 sensilla as a potential chemosensilla for statistical analysis and hypothesis testing.

Sensilla diversity across species

We documented between three and eight sensilla morphotypes within an individual firefly species. The distribution of individual morphotypes varied greatly between species. Overall, three different mechanosensilla types (sensilla chaetica C1, C2 and sensilla campaniform SC), nine different chemosensilla types (sensilla basiconica types B1, B2, B3, B7, B10, B11, B12, B13, and sensilla trichodea T1), as well as one potential temperature/humidity sensilla type (sensilla coeloconica X1), were identified across the seven firefly species in this study. Two mechanosensilla types (sensilla chaetica types C1 and C2) were the only sensilla morphotypes found in all seven firefly species (Table 4). Among the chemosensilla the B7 sensilla type was shared by most species (except Phausis) and found in both males and females (Tables 4, S3). Other chemosensilla were shared by three firefly species (B1, B3), by two species (B2) or they were unique for a single species (B10, B11: L. punctata; B12, B13: Luciolinae sp.; T1: Pha. christineae). A potential thermo/hygro sensilla type (X1) with almost identical structure to the temperature sensitive sensilla coeloconica in katydids [15] was shared by three firefly species (Table 4). There was no single chemosensilla morphotype found exclusively in diurnal species, or shared by all diurnal and nocturnal species, and the different chemosensilla types varied greatly between species (Table 4). Our comparisons of sensilla diversity based on effective numbers for Shannon and Simpson diversity indices showed no significant difference in antennal sensilla diversity between diurnal and nocturnal fireflies, no matter whether rare (Shannon’s index effective numbers, t-Test; x̄_Diurnal = 2.37 ± 0.53, x̄_Nocturnal = 2.37 ± 0.86, df = 13, p = 0.99) or common (Simpson’s index effective numbers, Wilcoxon rank-sum, x̄_Diurnal= 1.62 ± 0.41, x̄_Nocturnal = 1.95 ± 0.47, df = 13, p = 0.061) sensilla types were favored.

Sex differences in sensilla types

Almost all sensilla morphotypes were found in both males and females of the respective species, except for SC and B12 sensilla. SC sensilla were found in both sexes for the two Photinus species, Py. nigricans, and Luciolinae sp., but they were only found in L. punctata females and Ph. lucicrescens males. B12 sensilla were exclusively found in Luciolinae sp. males (Table 4).

Using the phylogeny of Martin et al. [56] to extract the evolutionary relationships of the eight firefly species in six genera studied to date, it is possible to reconstruct the ancestral pattern of sensilla.

Sensilla counts and densities

The total antennal sensilla counts (x̄ ± stdev) for a single firefly antenna ranged from 49 ± 7 total sensilla per antenna in Pha. christineae females to 6984 ± 301 sensilla in L. punctata males (Table 2). Sensilla densities in our species sample ranged from an average of 1090 ± 235 per mm² in Ph. lucicrescens females to an average of 5955 ± 803 per mm² in Pha. christineae males (Table 3). Total sensilla counts were positively correlated with antennal area across the 42 antennae of the seven firefly species (N = 42, R² = 0.389, Spearman’s rho = 0.6677, p < 0.0001, Fig 7A), which means larger antennae had more sensilla. In contrast, sensilla densities were negatively correlated with antennal area (R² = 0.542, Spearman’s rho = -0.7094, df = 1, p < 0.0001, Fig 7B), which means that larger antennae had fewer sensilla per area than smaller antennae.

In our mixed model analysis, species (random effect covariate) accounted for 55.2% of the variation in total sensilla counts across firefly antennae. Diurnal fireflies had significantly more antennal sensilla than nocturnal fireflies (x̄_Diurnal = 4768 ± 1495, x̄_Nocturnal = 2349 ± 1242; DFDen = 5, F = 8.33, p = 0.0343; Fig 7C) and males had significantly more sensilla than females (x̄_Male = 3776 ± 1919, x̄_Female = 2995 ± 1638; DFDen = 38, F = 8.7, p = 0.0054; Fig 7D). The interaction (activity*sex) was not significant. Similarly, species accounted for 69.5% of the variation in sensilla densities. Diurnal and nocturnal fireflies did not differ significantly (x̄_Diurnal ==3392 ± 907 per mm², x̄_Nocturnal = 2949 ± 1766 per mm²; DFDen = 5, F = 0.187, p = 0.683 Fig S3A) in their sensilla densities, nor did males and females (x̄_Male = 3414 ± 1720 per mm², x̄_Female = 2864 ± 1128 per mm²; DFDen = 38, F = 2.92, p = 0.0956, Fig S3B). However, the interaction (activity*sex) was significant (DFDen = 38.0, F = 5.93, p = 0.0197) with significantly higher sensilla densities on the antennae of nocturnal males compared to nocturnal females (x̄_Male = 3503 ± 2145 per mm², x̄_Female = 2395 ± 1118 per mm²; Student’s t ratio = -3.17, p = 0.003, Fig S3C).

The total sensilla counts included counts of three types of mechanosensilla, nine types of chemosensilla and one possible thermo and/or hygro sensilla (Table 2). Overall, there were many more mechanosensilla (in 1000s) than chemosensilla (in 100s) on individual firefly antennae. Phausis had lower numbers (in 10s) for both sensilla types, but still ~4 times more mechanosensilla than chemosensilla (Table 2). Potential thermo/hygrosensilla (X1) were identified in only a few individuals and if present, in low numbers (N = 1–22; Table S6).

Mechanosensilla densities.

Species accounted for 78.71% of the variation in mechanosensilla densities. There was no significant difference between diurnal and nocturnal fireflies (x̄_Diurnal = 2576/mm² ± 827, x̄_Nocturnal= 2068/mm² ± 1079, DFDen = 5.0, F = 0.477, p = 0.52, Fig 9A) or between males and females in mechanosensilla densities (x̄_Male = 2265 ± 1102 per mm², x̄_Female = 2306 ± 915 per mm²; DFDen = 33.0, F = 0.558, p = 0.460; Fig 9B). However, the interaction term (activity*sex) was significant (DFDen = 33.0, F = 11.085, p = 0.0021). This was due to significantly higher mechanosensilla densities in diurnal females compared to diurnal males (x̄_DFemale = 2887 ± 571 per mm², x̄_DMale = 2265 ± 954 per mm²; Student’s t-ratio = 2.7, p = 0.0109, Fig S6A) after species differences were taken into account as a random effect. In contrast, nocturnal females tended to have lower mechanosensilla densities than nocturnal males (x̄NFemale = 1870 ± 896 per mm², x̄_NMale = 2265 ± 1243 per mm²; this trend was marginally significant: Student’s t-ratio = -1.97, p = 0.057; Fig S6B). In comparison, diurnal and nocturnal females (x̄_DFemale = 2887 ± 571per mm², x̄_NFemale = 1871 ± 896per mm²; Student’s t-ratio = 1.35, p = 0.0.23; Fig S6C) and diurnal and nocturnal males (x̄_DMale = 2265 ± 954 per mm²_, x̄_NMale = 2265 ± 1243 per mm²; Student’s t-ratio = -0.00, p = 0.999; Fig S6D) did not differ from each other. Individual mechanosensilla morphotype densities are listed in Table S4.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. Evolution of sensilla morphotypes.

Sensilla presence (filled square) and absence (empty square) for each chemosensilla type across study species are marked in the data matrix. The most parsimonious gains and losses for each morphotype are marked with a hatchmark on the corresponding branches of the species cladogram.

https://doi.org/10.1371/journal.pone.0323722.g006

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. Antennal area and sensilla abundance by sex and activity.

(A) Positive correlation between antennal area and total sensilla number (N = 42, R² = 0.389, Spearman’s rho = 0.6677, p < 0.0001): larger antennae tend to have more sensilla; (B) Negative correlation between antennal area and total sensilla density (N-42, R² = 0.542, Spearman’s rho = -0.7094, df = 1, p < 0.0001): larger antennae tend to have a lower sensilla density; (C) Comparison of mean sensilla number between diurnal and nocturnal species: diurnal species have significantly more sensilla than nocturnal species (x̄_Diurnal = 4768 ± 1495, x̄_Nocturnal = 2349 ± 1242; DFDen = 5, F = 8.33, p = 0.0343); (D) Comparison of mean sensilla number between females and males: males have significantly more sensilla than females (x̄_Male = 3776 ± 1919, x̄_Female = 2995 ± 1638; DFDen = 38, F = 8.7, p = 0.0054).

https://doi.org/10.1371/journal.pone.0323722.g007

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 8. Mechanosensilla and Chemosensilla counts by activity and sex.

(A) Mechanosensilla counts (both sexes) by activity time: there is no significant difference between diurnal and nocturnal species (x̄_Diurnal = 3503 ± 993, x̄_Nocturnal = 1753 ± 1009; DFDen = 5.0, F = 4.88, p = 0.078); (B) Mechanosensilla counts (both activity times) by sex: there is no significant difference in mechanosensilla number between males and females (x̄_Male = 2586 ± 1248, x̄_Female = 2420 ± 1418; DFDen = 33.0, F = 0.21, p = 0.279); (C). Chemosensilla counts (both sexes) by activity time: there is no significant difference between diurnal and nocturnal species (x̄_Diurnal = 1261 ± 1114, x̄_Nocturnal = 595 ± 282; DFDen = 5.0, F = 3.81, p = 0.108); (D) Chemosensilla counts (both activity times) by sex: males have significantly more chemosensilla than females (x̄_Male = 1188 ± 1041, x̄_Female = 573 ± 307; DFDen = 33.0, F = 15.66, p = 0.0004).

https://doi.org/10.1371/journal.pone.0323722.g008

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 9. Mechanosensilla and Chemosensilla densities.

(A) Mechanosensilla densities (both sexes) by activity time: there is no significant difference between diurnal and nocturnal species (x̄_Diurnal = 2576/mm² ± 827, x̄_Nocturnal = 2068/mm² ± 1079, DFDen = 5.0, F = 0.477, p = 0.52); (B) Mechanosensilla densities (both activity times) by sex: males have a significantly higher density of mechanosensilla than females (x̄_Male = 2265 ± 1102 per mm², x̄_Female = 2306 ± 915 per mm²; DFDen = 33.0, F = 0.558, p = 0.460); (C) Chemosensilla densities (both sexes) by activity time: there is no significant difference between diurnal and nocturnal species (x̄_Diurnal = 813 ± 486 per mm², x̄_Nocturnal = 881 ± 914 per mm²; DFDen = 5.0, F = 0.025, p = 0.880); (D) Chemosensilla densities (both activity times) by sex: males have a significantly higher density of chemosensilla on their antennae than females (x̄_Male = 1147 ± 952 per mm², x̄_Female = 556 ± 273 per mm²; DFDen = 33.0, F = 11.05, p = 0.0022).

https://doi.org/10.1371/journal.pone.0323722.g009

Sensilla Distribution on Firefly Antennae

To determine which portions of firefly antennae may be used for mechanoreception and/or chemoreception, we examined the distribution of sensilla numbers (mechanosensilla, chemosensilla, and each sensilla morphotype) across antennomeres. Mechanosensilla were most abundant at the base of the antenna. C1 mechanosensilla occurred in high numbers on the scape, dropping to the lowest numbers on the pedicel (the second and smallest antennomere) in all species, except Pha. christineae (Fig S7G). C1 sensilla sharply increased in number from the pedicel to the third antennomere and occurred at similar numbers between antennomeres 4 and 10 for all species. Numbers further increased in P. corruscus, Py. nigricans, L. punctata, and P. pyralis on antennomere 11 (Fig S7A-D). C2 sensilla were exclusively located on the scape and pedicel in all species, with about twice as many C2 sensilla found on the scape (Fig S7H-N). SC sensilla were absent in Phausis and in some sexes of other species (Fig S7O-T). SC sensilla were found on all 11 antennomeres but occurred in low numbers and were not evenly distributed. The antennomeres with the highest SC numbers differed between individuals of the same species and between species.

In contrast, chemosensilla were entirely absent from the scape and pedicel. They tended to be most abundant in the middle of the firefly antennae. B1 sensilla increased in number between antennomeres 3–6 and then decreased from antennomere 6–10 in P. corruscus and P. pyralis. In contrast, the distribution of B1 sensilla in Py. nigricans were overall consistent across antennomeres 3–8 (Fig S8A-C). The number of B2 sensilla also increased between antennomeres 3 and 6 and then decreased from antennomere 6–11 (Fig S8D-E). B3 sensilla numbers varied between antennomeres 3–8 and gradually decreased in number in each subsequent antennomere (Fig S8F-H). B7 sensilla were found on antennomere 3 in all species except P. pyralis. B7 numbers varied between antennomeres 3–10 with a sharp increase on antennomere 11 in all species, except P. pyralis (B7 decreased on antennomere 11; Fig S8I-N). B10 sensilla (L. punctata) were most abundant between antennomeres 4 and 6 (Fig S9A), slowly decreasing in number towards the distal antennomere; males have about twice as many B10 on each antennomere (peak average of 99 sensilla on antennomere 4) compared to their females (peak average of 56 on antennomere 6). B11 sensilla (L. punctata) were rare on female antennae with most (average 9 sensilla) located on the last antennomere; in contrast males consistently averaged between 300 and 350 B11 sensilla on antennomeres 3–11 (Fig S9B). B12 sensilla (Luciolinae sp.) were present in low numbers on antennomeres 3–10 of male antennae, with the highest number (average 2.7 sensilla) on antennomeres 3 and 4 and the lowest number (average 0.33 sensilla) on antennomere 5 (Fig S9C). B12 sensilla were absent in females. B13 sensilla (Luciolinae sp.) were present on antennomeres 3–11 on male antennae, with the highest number on antennomere 5 (average 81 sensilla), then slowly declining in numbers to antennomere 11 (average 17 sensilla; Fig S9D). In parallel to males, B13 sensilla of females were present on antennomeres 3–11, with the highest number on antennomere 5 (average 71 sensilla), slowly declining to antennomere 11 (average 14 sensilla; Fig S9D). T1 sensilla (Pha. christineae) were present on the 2nd antennomere (of 3) on female antennae (3–10 sensilla each; Fig S9E). Phausis males had T1 sensilla on antennomeres 3–11, and they were more or less evenly distributed (ranging on average 55–70 sensilla on each antennomere) between antennomeres 3–11 (Fig S9E). X1 sensilla were only found in three species: P. corruscus, Py. nigricans, and Ph. lucicrescens, and only on individual fireflies. In Py. nigricans and P. corruscus, X1 sensilla were present exclusively on the ventral side of the antenna, and on both dorsal and ventral sides in a single female of Ph. lucicrescens.

Evolution of sensilla types

Given our taxon sampling of eight taxa in six genera and their phylogenetic relationships [56], it appears that mechanosensilla types are relatively conserved and shared by all (C1) or almost all (C2, SC) firefly species studied to date. In contrast, their chemosensilla types are much more diverse and only a single sensilla type (B7) is present in all species and thus seems to be inherited from the common ancestor of our study species and subsequently lost in the Phausis lineage (Fig 6). B1 and B3 sensilla most likely evolved in the last common ancestor of Pyropyga and Photinus fireflies. Other sensilla types evolved independently in individual lineages, e.g., B2 evolved independently in Ph. lucicrescens and P. corruscus, B10 and B11 evolved in the Lucidota lineage, B12 and B13 evolved in the Luciolinae sp. lineage, and T1 evolved in the Phausis lineage. X1 evolved independently in the Photuris, Pyropyga and P. corruscus lineages; alternatively, it could have evolved in Photuris and the common ancestor of Pyropyga and Photinus and was subsequently lost in P. pyralis. For Lu. cruciata, unique pored sensilla chaetica, capitular sensilla, and gemmiform sensilla are reported in the literature [26]. The spotty distribution of the other chemosensilla types across firefly genera suggests a high evolutionary lability of chemosensilla types among the six firefly genera in our study. However, these genera are relatively distantly related and belong to three different lampyrid subfamilies. On the other hand, the two Photinus species share many of the same sensilla types, despite their different (diurnal/nocturnal) activity times. To further investigate this seeming evolutionary lability of chemosensilla types across genera and the conservation of sensilla types within a genus, more species in each genus need to be studied, as well as additional genera that are closely related to our study taxa [56].

Optimizing sensitivity

Antennal sensitivity to relevant environmental stimuli can be increased through additional sensilla, thus increasing the probability of capturing physical or chemical stimuli. This can be achieved by increasing antenna size (surface area) or by packing sensilla more tightly (density). In the present study diurnal fireflies had significantly more sensilla on their antennae than nocturnal fireflies and thus a higher sensitivity for environmental stimuli. Similarly, males had significantly more sensilla on their antennae than females. Neither diurnal and nocturnal fireflies, nor males and females differed in overall sensilla densities. Thus, in fireflies overall antennal sensitivity (sensilla numbers) seems to be mainly increased by increasing the antennal surface area. Larger antennae also have the advantage of sampling a larger airspace, which increases the probability of detecting the pheromones of a conspecific female [27]. Not surprisingly, firefly males - as the actively searching sex - have relatively larger antennae (relative to their body size) and significantly higher total sensilla counts compared to their females. The evolution of larger firefly antennae as an improved sampling area seems to be supported by the negative correlation between antennal area and sensilla density: as firefly antennae increase in size, sensilla numbers do not increase in a 1:1 relationship with antennal area, resulting in a lower density. However, there may be other limiting factors for sensilla density on firefly antennae, such as limited space in the antennomeres that supports a limited supply system for the receptor neurons in each sensillum [1].

Larger body size in insects supports larger antennae [82,83], including in fireflies [36]. Not surprisingly, across our seven species, body size (pronotum length) had a significant effect on antennal area, with species accounting for almost all (95.06%) of the variation in body size. The body size of fireflies in our species sample did not differ significantly by sex or activity, thus body size did not influence our statistical analysis of sex and activity effects on sensilla counts or densities. This means that the significantly larger antennal areas of firefly males compared to females cannot be explained by a difference in body size. Instead, the evidence points towards direct selection on male antenna size, most likely sexual selection resulting in increased antennal sensitivity of male antenna to chemical stimuli. This is supported by significantly larger antennal areas of diurnal males compared to diurnal females.

In contrast, diurnal and nocturnal fireflies in our study overlapped greatly in their antennal areas, and did not differ significantly from each other. We suspect that this is likely due to our relatively small species sample. A previous large-scale analysis of eye and antenna sizes in 101 firefly taxa showed that the males of the 26 diurnal firefly taxa tended to have significantly longer antennae (and smaller eyes) than the males of the 75 nocturnal taxa [38] with activity time (and type of mating signals) accounting for 13% of the observed variation in male antenna size, while phylogenetic relatedness (genus) accounted for 63% [38]. Similarly, in a phylogeny-based analysis of 43 North American firefly species, antenna length (phylogenetic signal lambda = 0.859) was significantly correlated with body size (pronotum length, p < 0.0001) and activity time (mating signal: p = 0.037). Combined with the results from the present study, it appears that the evolution of a larger antenna size is an important factor in increasing sensilla counts, and thus antennal sensitivity, in diurnal fireflies.

Interestingly, when fireflies change their activity time from nocturnal to diurnal over evolutionary time (and switch their mating signals from bioluminescent signals to the exclusive use of pheromones), the selection on antenna size seems to be relatively weak, resulting in a relatively slow increase in antenna size compared to the reduction in eye size [38]. For this reason, Stanger-Hall et al. [38] suggested that the first response to selection for improved pheromone detection may occur through sensilla numbers instead, resulting in an increased sensilla density on relatively short antennae. An analysis of our data for P. pyralis (N = 6) and P. corruscus (N = 6) shows that this is indeed the case. The diurnal firefly P. corruscus, which belongs to a clade of fireflies (formerly Ellychnia [51]) that split from their common ancestor with nocturnal Photinus fireflies between 6 and 25 mya [84,85] has significantly shorter antennae (scaled with body size: Wilcoxon rank-sum test: S=56, Z = 2.64, p = 0.008) than the nocturnal P. pyralis, but it has significantly more sensilla (Wilcoxon rank-sum test: S=21, Z = -2.80, p = 0.005) and a significantly higher sensilla density (Wilcoxon rank-sum: S=21, Z = -2.8, p = 0.005). This is due to significantly higher mechanosensilla numbers (x̄ = 4506 ± 456 versus x̄ = 3015 ± 222.5, Wilcoxon rank sum N = 6,6, S=21, Z = -2.8, P = 0.005) and densities (x̄ = 2851 ± 352 per mm² versus x̄ = 1381 ± 263 per mm²; (Wilcoxon rank sum N = 6,6, S=21, Z = -2.8, P = 0.005). Interestingly, the total chemosensilla counts of P. corruscus and P. pyralis did not differ (N = 6,6, x̄ = 794 ± 154 versus x̄ = 806 ± 155; Wilcoxon rank sum N = 6,6, S=41, Z = 0.24, P = 0.81), but P. corruscus (with its shorter antennae) has a significantly higher chemosensilla density (x̄ = 504 ± 109 per mm² versus x̄ = 365 ± 65 per mm²; Wilcoxon rank sum N = 6,6, S=25, Z = -2.16, P = 0.031) as predicted by Stanger-Hall et al. [38]. Given the similar chemosensilla numbers for P. corruscus and P. pyralis, this raises the question whether (or to what degree) P. pyralis may still utilize pheromones during mate search.

Testing predictions

We hypothesized that the differences in mating signals between diurnal and nocturnal fireflies will be reflected in their antennal sensilla counts and possibly in their sensilla densities. Given the importance of pheromones for the mate search of diurnal firefly species, we predicted (1) more chemosensilla (including pheromone sensilla) in diurnal species compared to the bioluminescent nocturnal species, and (2) more chemosensilla in males compared to females, specifically that diurnal males, which are tracking pheromone plumes to females, would have more chemosensilla than their females. In comparison, we predicted no differences in mechanosensilla counts and densities between diurnal and nocturnal fireflies and/or between males and females.

Species accounted for most (81.75%) of the variation in mechanosensilla counts across firefly antennae, reflecting large species differences in sensitivity to mechanical stimuli, possibly due to different microhabitats. Diurnal and nocturnal fireflies did not differ in their mechanosensilla counts, nor did males and females. There was also no significant interaction between sex and activity in mechanosensilla counts. This finding was not unexpected, since both diurnal and nocturnal fireflies (and both sexes) need mechanosensilla to navigate through their physical environment during mate search, to deposit fertilized eggs, and to and from their daily resting places. Similarly, species accounted for most (78.71%) of the variation in mechanosensilla densities. There was no significant difference between diurnal and nocturnal fireflies or between males and females, however the interaction (sex*activity) was significant with significantly higher mechanosensilla densities in diurnal females compared to diurnal males. This is likely due to the significantly smaller antennal areas of diurnal females compared to diurnal males, resulting in a denser packing of similar mechanosensilla numbers on a smaller area.

Species accounted for a relatively small part (32.27%) of the variation in chemosensilla counts across firefly antennae. Overall, the diurnal fireflies did not significantly differ in their chemosensilla counts from the nocturnal fireflies in our study, but males had significantly more chemosensilla than females and the interaction term (activity*sex) was also significant. As predicted, diurnal males had significantly more chemosensilla than nocturnal males. Diurnal males also had significantly more chemosensilla than their females, who release pheromones, but have not been documented to respond to them (e.g., to pheromones of other females). The females of diurnal and nocturnal fireflies did not differ significantly in their chemosensilla counts, nor did nocturnal females and their males, who rely much less (if at all) on pheromones than the diurnal males. Species accounted for 46.3% of the variation in chemosensilla densities. Overall, diurnal fireflies did not significantly differ from nocturnal fireflies, but males had significantly higher chemosensilla densities than females. Since males also have significantly larger antennal areas than females, these results suggest that sexual selection in firefly males can act through both, a significantly larger antennal area and/or a higher density of chemosensilla to optimize the sensitivity of male antennae to chemical stimuli. For example, Phausis and Luciolinae sp. males have relatively small antennae with high densities of chemosensilla, and Lucidota males have both large antennae and a high chemosensilla density (Tables 2, 3).

Gustation during antennation behavior

The intensive antennation behavior that precedes mating in fireflies, led us to hypothesize that individuals sample chemical cues via gustatory (contact) chemosensilla to verify a conspecific mate, immediately before mating. Based on this reciprocal behavior we predicted no sex differences in sensilla that detect these cues. We presently do not know to what extent individual firefly species use contact chemicals during mating (or whether antennation represents a purely physical stimulus in preparation for mating), and which sensilla could be used as gustatory sensilla. For a gustatory (low volatile or contact chemical) sensilla types we would predict single pores and the absence of sexual dimorphism in sensilla numbers, or possibly skewed towards females, because they incur a larger cost for a mating mistake. Based on these criteria, we propose the following candidates for future functional testing as gustatory sensilla: B2 or B3 for P. corruscus, B3 for P. pyralis, B1 for Py. nigricans, B10 for L. punctata, and B13 for Luciolinae sp. If no pheromones are used by Ph. lucicrescens, B2 would test as a gustatory sensilla, with a sexual dimorphism based on the predatory behavior of females. For Pha. christineae with a single chemosensilla type, T1 sensilla would most likely function as pheromone sensilla [34].

Possible other uses of chemosensilla in fireflies include the localization of milkweed for nectaring. Fireflies from three genera (Photinus, Pyropyga, Photuris) have been observed to collect nectar and pollen from milkweed, choosing only the freshest aromatic flowers, and two additional genera (Lucidota, Pyractomena) have been seen on milkweed [93]. In addition, P. corruscus has been observed feeding on the floral nectaries of Norway maples [94]. Both males and females show this behavior, therefore no sex differences in the respective chemosensilla would be expected.

As an important next step in understanding how fireflies perceive the world through their antennae, and how this influences their behavior, all of the pheromone and gustatory sensilla candidates need to be tested in functional studies. One key question in this context is the extent to which the pheromone sensilla candidates remain functional in nocturnal firefly species. Our morphological data suggest that B1 sensilla may function as pheromone sensors in both Photinus species: P. pyralis, a nocturnal bioluminescent species, and P. corruscus, which split from nocturnal Photinus ~6–25 mya [84,85] and returned to diurnal activity with the exclusive use of pheromones for mate search [37,40]. Other repeated losses of bioluminescence and reversals to mate search exclusively with pheromones during firefly evolutionary history [39,40], suggest that pheromone sensilla may remain functional at least in some nocturnal species, especially in those clades with recent reversals to diurnal activity (e.g., Photinus), facilitating the switch from nocturnal to diurnal activity. If functional in nocturnal species, pheromones would increase mating opportunities, because they would attract and direct males towards the bioluminescent display sites of females. Once they are close enough, they can use visual cues to locate and/or identify conspecific females over shorter distances.

Pheromones are especially important in environments that are not conducive to visually finding females. For example, considering their antennal areas, the males of diurnal Lucidota and of nocturnal Phausis both have relatively high chemosensilla (but not mechanosensilla) densities on their antennae. L. punctata (5–6mm body length) and Pha. christineae, like Pha. reticulata (6–9mm) are relatively small fireflies and the males of these species search for conspecific females in the low vegetation on the forest floor, navigating around leaves to find their females [95]. The diurnal L. punctata males have relatively large antennae, combined with a relatively high chemosensilla density to locate their tiny females in shady forests. The males of nocturnal Pha. reticulata (blue ghosts) and Pha. christineae fly at night and glow while trying to locate their tiny flightless females. Phausis females emit a weak glow (visible within 10 feet, but almost invisible when moonlight is reflected from wet leaves [95]), and Phausis males seem to use female pheromones to get close to females. If the male cannot locate the female by her weak glow, males may use their own glow as “spotlight” to locate the female [34].

Why are chemosensilla so diverse across fireflies?

The individual firefly species in our study had between one to four different chemosensilla types on their antennae. The number of types seem to be correlated with diurnal or nocturnal activity, however, a puzzling insight from our study is the diversity of antennal sensilla types in fireflies, and how little overlap there is between genera. The adult fireflies in our study tend to survive for only a few weeks (except for the winter firefly P. corruscus, which may live several months [96]) for the sole purpose of mating, and diurnal and nocturnal species face very similar challenges: prevent dehydration, locate a conspecific mate, mate, and in the case of females, find a suitable egg deposition site. So why are their antennal chemosensilla so different? We identified three different pheromone sensilla candidates in our three diurnal firefly species (B1, B3, B10/B11), which raises the question of how it is possible that these different sensilla morphotypes converged on an apparently similar function? Or alternatively: why did the external morphology (accessory structure) of pheromone receptors diverge across species?

In Drosophila trichoid sensilla are required for pheromone detection, while their basiconica sensilla mostly detect food-derived odors [49]. Similarly, trichoid sensilla respond to female pheromones in click beetles (Elateridae) [65] and in Asian longhorned beetles (Cerambycidae) [97], while placoid sensilla detect pheromones in Japanese beetles (Scarabaeidae) [98]. In fireflies (Lampyridae) we identified trichoid sensilla as pheromone sensilla candidates for nocturnal Phausis, however, trichoid sensilla are absent in all other firefly species studied to date. Instead, sensilla basiconica are used for pheromone detection in the firefly P. corruscus [24]. P. corruscus females emit (1S)-exo-3-hydroxycamphor (hydroxycamphor), which in single sensillum recordings elicited a neuronal response from a pheromone-sensitive olfactory sensory neuron in a basiconica sensillum on the male antenna [24]. We identified this sensillum as sensilla basiconica type 1 (B1). A possible model for how pheromone detection can switch between major sensilla types, is the Asian longhorned beetle Anoplophora glabripennis (Cerambycidae), whose pheromone consists of two components [98]. Trichoid sensilla are the pheromone sensors that respond to both components, but basiconica sensilla respond to one of the two components (along with plant compounds that enhance male attraction). Even though there is no pheromone-specific information relayed by these basiconica sensilla, this observation suggests a possible mechanism for sensilla type switching as pheromone compounds diverge in closely related species. Similarly, in the moth Ostrinia nubilalis (Hübner, 1769; Crambidae) pheromone sensilla that respond to different pheromone compounds are split between two olfactory sensory neurons in two sensillum subtypes, which is thought to reflect an ongoing evolution of this sensillum type as two O. nubilalis strains diverge [99]. A similar mechanism could account for the diversification of B-sensilla in fireflies (Lampyridae) and the diversity of pheromone sensilla candidates across diurnal firefly species. Sensilla type switching may not be limited to fireflies, but reflect the evolutionary dynamics of pheromone signals and their sensilla (with their sensory neurons) in insects in general; evolutionary studies with a broad taxon sampling across closely related species in different genera will be required to test this in studies that combine external and internal sensilla morphology, including receptor neurons, with functional testing.