Sexually dimorphic eye size in dragonfishes, a response to a bioluminescent signalling gap

Abstract

Deep-sea fishes must overcome extremely large nearest-neighbour distances and darkness to find mates. Sexual dimorphism in the size of luminescent structures in many deep-sea taxa, including dragonfishes (family Stomiidae), indicates reproductive behaviours may be mediated by visual signalling. This presents a paradox: if male photophores are larger, females may find males at shorter distances than males find females. Solutions to this gap may include females closing this gap or by males gathering more photons with a larger eye. We examine the eye size of two species of dragonfishes (Malacosteus niger and Phostomias guernei) for sexual dimorphism and employ a model of detection distance to evaluate the potential for such dimorphism to bridge the detection gap. This model incorporates the flux of sexually dimorphic postorbital photophores and eye lens size to predict detection distances. In both species, we found a significant visual detection gap in which females find males before males find females and that male lens size is larger, marking the second known case of size dimorphism in the actinopterygian visual system. Our results indicate the larger eye affords males a significant improvement in detection distance. We conclude that this dimorphic phenotype may have evolved to close the detection gap.

1. Introduction

Deep-sea fishes face low population densities and near total darkness, conditions that make locating mates extremely difficult [1]. Although it is well known that olfactory communication is important in mediating reproductive opportunities [2], the existence of sexually dimorphic luminescent structures in many of these taxa underscores the potential significance of visual signalling in mating behaviours. This is particularly evident in the dragonfishes (family Stomiidae), a meso- and bathypelagic group in which males exhibit larger luminescent organs (figure 1). The efficacy of luminescent signalling for sexual communication hinges on the strength of the signal (i.e. photon flux) being adequate to reach a conspecific as well as on the receiver’s capacity to detect this signal [4]. If size, visual sensitivity and spectral characteristics of the photophores are similar between the sexes, dragonfish females may receive a stronger male luminescent signal at a distance beyond which they can reciprocate, resulting in a signalling gap (figure 1d). Given the scarcity of reproductive opportunities owing to low population densities, we hypothesize that dragonfishes may have developed compensatory adaptations, either behavioural or physiological, to mitigate this signalling discrepancy. These mechanisms may include, for example, a female search behaviour in which she identifies a male signal and closes the signalling gap. Alternatively, or in conjunction, males might have evolved sexually dimorphic, larger eyes to enhance the detection of weaker female signals, thereby closing this gap to some extent.

Figure 1.

One of the study species, Malacosteus niger. (a) Lateral view of the head in a freshly captured specimen. The blue-emitting postorbital photophore (PO) and red-emitting anterior orbital photophore (AO) are indicated by arrows. The AO emission is rapidly absorbed by seawater [3] and, thus, it was not included in our study. (b) Schematic drawing depicting sexual dimorphism in PO size. (c) Head in $\mu$-CT reconstruction. (d) Schematic drawing of sexual dimorphism in PO size and the detection gap that results. A male (left) produces a brighter luminescent signal than a female (right) and, therefore, a female is able to detect a male at a longer distance (${\displaystyle r_{\mathrm{f}\mathrm{e}\mathrm{m}\mathrm{a}\mathrm{l}\mathrm{e}}}$) than she can be detected by the male (${\displaystyle r_{\mathrm{m}\mathrm{a}\mathrm{l}\mathrm{e}}}$).

In this study, we investigate the extent of any detection gap and the evolution of visual dimorphism in males of two dragonfish species, Malacosteus niger and Phostomias guernei, which possess larger postorbital photophores (figure 1b) [5,6]. We assessed (i) the difference in detection distance ($r$) between males and females of these two species; (ii) whether lens diameter is sexually dimorphic; and (iii) what role body size plays in determining the magnitude of the difference in detection distance. To evaluate these aspects of visual detection, we measured the eye and photophore sizes of preserved museum specimens. Using these data, we calculated $r$ across a spectrum of body sizes by employing a previously established model for detection–distance estimation.

2. Material and methods

(a). Specimen data

Postorbital photophore area, lens size and standard length (SL) data were taken from 42 specimens of M. niger (7.6−16.4 cm SL; 25 male, 16 female; electronic supplementary material, table S1) and 63 specimens of P. guernei (4.4−14.0 cm SL; 30 male, 32 female) obtained from Harvard University’s Museum of Comparative Zoology. We took images of the head of each specimen with a Nikon D6 digital camera and measured the area of the postorbital photophore in cm${}^2$ using ImageJ [7]. Both of our study species possess an anterior orbital photophore positioned along the anteroventral margin of the eye [5,6,8]. In P. guernei, this photophore is minute in males, covered by skin in females and produces blue-green luminescence [6]. In M. niger, this photophore is similarly large in both sexes (figure 1a) and produces a far-red wavelength that may be used to detect prey [9,10]. We measured the area of the anterior photophore in specimens of P. guernei and added this value to the area of the postorbital photophore. Because the far-red emission of the anterior orbital photophore of M. niger is rapidly absorbed by seawater [3], its area was not measured.

Previous modelling studies have used pupil diameter in calculating detection distance [11]. We found that pupils of our specimens were often damaged. The lens, a very hard tissue, was never damaged and is generally the same diameter as pupil in fishes [12]. Thus, we used lens diameter as a proxy for pupil diameter. We calculated lens diameter based on $\mu$-CT reconstructions from a SkyScan1173 high-energy spiral-scan unit (figure 1c). We removed the lens from the right eye of each specimen and embedded it in closed-cell insulating foam for scanning. Parameter values for amperage, voltage, exposure time and image rotation varied between 45−56 $\mu$A, 51−73 kV, 337−730 ms and 0.06−0.07${}^{°}$, respectively. This produced voxel sizes of 12.5−18.1 $\mu$m. We performed slice reconstruction in NRecon (Micro Photonics), and segmentation and volume rendering in Mimics 15.0 (Materialise). We calculated lens diameter from these rendered volumes.

(b). Sexual dimorphism and detection distance

To evaluate sexual dimorphism in postorbital-photophore flux and lens diameter, we constructed simple linear and allometric regression models with standard length as the covariate and sex as an interacting predictor variable. Statistical significances were assessed with ANOVA.

To calculate detection distance, we implemented Warrant’s [11] model:

$$N=\frac{E{A}^2}{16r^2}{e}^{-\alpha r},$$ (2.1)

where $E$ is the flux produced by the emitting photophore, $A$ the lens diameter, $r$ the detection distance and $\alpha$ the total attenuation coefficient [13]. Solving for $r$ numerically, this reduces to:

$$r=W\left(\alpha \frac{\sqrt{\frac{E{A}^2}{16N}}}{2}\right)\frac{2}{\alpha },$$ (2.2)

where $W$ is the product log function. We used a value of 0.05 m${}^{-1}$ for $\alpha$ following Denton [14], which corresponds to the total attenuation of blue light in clear water. The number of photons needed by a fish ($N$) to detect a blue flash was set to 5 [14].

The flux ($E$) of the postorbital photophore of both species is unknown, but we presume it varies with photophore size as in other deep-sea teleosts [15]. Therefore, we predicted the flux of each individual based on data for the postorbital photophore of Aristostomias scintillans, a close relative of both species [16], from Mensinger & Case [15]. We produced a best-fitting allometric model of flux output based on photophore area of A. scintallans, through which we passed our specimens’ postorbital photophore area.

We note that our implementation of Warrant’s [11] model does not account for light in the background and that the point-source luminescent flash of a photophore would lose its contrast and eventually be overcome by this [4]. In addition, this model assumes that the point-source image falls entirely within the receptive field of an eye’s foveal ganglion cell, which is unlikely [4,11]. Therefore, calculations of $r$ must be considered near the theoretical maximum of detection distance.

To evaluate differences in detection distance between sexes and the effect of lens sexual dimorphism on detection distance, we took three approaches. First, we implemented the model described above to predict detection distance for each individual based on lens diameter and the predicted flux of a size-matched individual of the opposite sex. Because our study specimens varied in size and photophore size increases with growth [5,6], we constructed species-specific allometric models of detection distance versus standard length with sex as a factor and then assessed statistical significance of sex-specific detection distance with ANOVA.

Second, to assess the effect of male lens sexual dimorphism on detection distance, we repeated the allometric modelling of detection distance for males only. In this case, we predicted male lens size based on the female regression models described above. Detection distance was then calculated based on non-sexually dimorphic lens diameter for each male specimen of both species. We then calculated the differences between detection distance with and without a sexually dimorphic lens ($\Delta r_{male}$). We subjected $\Delta r_{male}$ values for each species to a one-sided t‐test under the alternative hypothesis that values were greater than zero.

Lastly, males of our study species mature at smaller sizes than females [5,6,17] and, given their low population densities, we presume that individuals of both sexes would reproduce with counterparts of any mature size. To address the interplay between sexual size dimorphism and detection distance between males and females of different sizes, we modelled detection distance based on linear-model predictions of lens diameter and allometric models of photophore flux in the opposite sex over a range of sizes for each species (6−15 cm SL) that include the smallest sexually mature and large adults of both species. For this, we began by predicting lens diameter and postorbital flux for males and females based on sex-specific allometric regression models over our size range in increments of 0.1 cm standard length. For each sex of both species, we then modelled detection distance over this size spectrum based on predicted lens diameter and the range of predicted flux values for the opposite sex. We then calculated the detection log ratio as the difference in log-transformed detection values ($log(r_{female})-log(r_{male})$). Negative log ratios therefore indicate $r_{male}>r_{female}$, positive log ratios indicate $r_{female}>r_{male}$, while 0 indicates $r_{male}=r_{female}$.

All modelling and statistical analyses were undertaken in R [18]. Each regression model constructed in our analysis represents the best-fitting model according to corrected AIC weights (AICw) calculated for simple linear and log–log allometric regressions. In all cases, statistical significance of the best-fitting model was determined at the level of p $<$ 0.05. Morphological data and the script containing these analyses are available in the electronic supplementary material.

3. Results

For both species, our analysis revealed that a log–log allometric model fit the relationship between postorbital photophore flux and standard length best (both AICw = 0.99; electronic supplementary material, table S2). In the best-fitting models for both species, photophore flux was consistently lower in females than in males (figure 2a). Furthermore, we found a significant interaction between sex and size (p = 0.0014 for M. niger and p $<$ 0.001 for P. guernei; figure 2a), suggesting that photophore flux increases with size at a notably faster rate in males compared with females.

Figure 2.

(a) Log–log regression plots of predicted postorbital photophore flux versus standard length for males and females of M. niger and P. guernei. (b) Linear regression plot of lens diameter versus standard length. (c) Log–log regression plots of detection distance versus standard length. Detection distance was calculated based on the flux of a size-matched individual of the opposite sex.

For lens diameter, an isometric model provided the best fit for our data (AICw = 0.94 for M. niger and AICw = 0.59 for P. guernei; electronic supplementary material, table S3), with male lens diameters surpassing those of females. Specifically, for M. niger, a significant interaction was noted between sex and size (p $<$ 0.00), indicating a comparatively slower isometric growth rate in males than females, as illustrated in figure 2b. For P. guernei, although no significant sex–size interaction was found (p = 0.089), the data revealed a significantly larger average lens diameter in males (p $<<$ 0.001).

Female detection distance of a size-matched male is extraordinarily greater than the reverse. This pattern is driven by the exponentially greater fluxed produced by males (figure 2a) which, in turn, results in a much brighter signal for females. For M. niger, female detection distance ranged from 24.9 to 169.5 m and male detection distance ranged from 26.1 to 66.3 m (figure 2c). For P. guernei, detection distance ranged from 5.4 to 33.5 m in females and 2.4 to 14.5 m in males. A log–log allometric model for detection distance fit best for both species (AICw = 1; electronic supplementary material, table S4). In both species, male detection distance was predicted by a significant interaction between size and sex (both p $<$ 0.0001, figure 2c), indicating that this negative allometric relationship proceeds at a different pace between males and females.

By comparing our models of male detection distance with those that predict distance without a dimorphic lens ($\Delta r_{male}$), we found that a larger eye increases detection distance in both species examined (figure 3a, both p $<<$ 0.001). The mean gain in detection distance in males was 1.6 m for M. niger and 1.4 m for P. guernei.

Figure 3.

(a) Difference between detection difference in males with and without a sexually dimorphic lens size ($\Delta r_{male}$) in M. niger and P. guernei. (b) Predictions of detection log ratio as it varies with size in males and females of both species. Hot colours, log ratio $>$ 0, indicate a larger male gap ($r_{male}<r_{female}$), while cool colours, log ratio $<$ 0, indicate a larger female gap ($r_{male}>r_{female}$). Note that the log ratio of 0 indicates $r_{male}=r_{female}$. Vertical and horizontal lines represent the minimum length at sexual maturity for males and females of both species according to Marks et al. [17].

In our analysis of detection distance between males and females of different sizes, we observed large differences in $r$ and log ratios that depend on a combination of size and sex in both species, as illustrated in figures 3b and 1. Notably, when females reach sexual maturity and a small male is involved, there is no detection gap in P. guernei or even a female detection gap ($r_{male}>r_{female}$) in M. niger. This pattern shifts to a large male gap ($r_{male}<r_{female}$), as indicated by a transition from low to high log-ratio values from left to right along the male size axis in figure 3b. This trend continues as females get larger; however, the gap decreases when larger males are involved as indicated by lower log-ratio values along the female size axis in figure 3b. This pattern suggests that when both mature females and males are small, the detection gap is smallest and when females are small, but males are large, the gap is largest with females detecting males at much greater distances. When both sexes are small, the photophore size and resulting flux is less dimorphic, resulting in similar detection distances (figure 2a). When females are small and males are large, the photophores size and flux are most dimorphic and this results in a large male gap ($r_{male}<r_{female}$).

4. Discussion

Our study focused on modelling visual detection distances, taking into account lens diameter and the photophore flux from size-matched individuals as well as across a range of sizes in both species. Our findings reveal several intriguing patterns. First, owing to the allometry of sexual dimorphism in photophore size and resulting flux, sexually mature females have a significant advantage in detecting luminescing males before being detected themselves (i.e. a male detection gap). Second, for both species, the lens diameter of males exceeds that of females. Combined, our findings suggest that this so-called ‘dead zone’ [1] has resulted in the evolution of relatively larger lens diameters in males of M. niger and P. guernei.

To our knowledge, this is one of only a few documented cases of eye-size dimorphism in vertebrates and perhaps only the second for ray-finned fishes. Previous observations of this phenomenon are limited to a few species of reptiles, including snakes [19,20], geckos [21] and chickens [22] and species of the scorpaenid fish genus Sebastes [23]. The implications of dimorphism in reproductive behaviours for these cases remain unexplored. Notably, within some species of the deep-sea teleost family Myctophidae, males possess a unique dimorphic ocular filter that likely enhances the visibility of luminescent signals [24]. Similar to our focus species, these species also feature sexually dimorphic photophores [2], a pattern that may represent an evolutionary parallel to our results.

We also note, however, that sexual dimorphism of the visual and luminescent systems in deep-sea fishes may have arisen for reasons unrelated to reproduction. In particular, ecological drivers that place different natural selective pressures on the two sexes can result in sexual dimorphism [25–27]. Differences in prey resource use, for instance, have been implicated in the evolution of sexually dimorphic feeding structures in several other groups of vertebrates, including squamates and freshwater fishes [28,29]. Whether any ecological factor plays a role in the evolution of sexually dimorphic luminescent and visual systems of deep-sea fishes is unknown, but this form of selection should not be overlooked.

Leaving aside what has led to the evolution of dimorphism in photophore size and the consequential bioluminescent signal gap, it is plausible that the dragonfish dead zone exerted selective pressure on narrowing this signalling gap. The distance that separates conspecifics in these two species is rather high compared with that of other fishes [1,30]. The nearest-neighbour distance of P. guernei is around 25 m [1,31]. M. niger is captured more rarely [31,32] and we expect the inter-individual distance to be even greater. Given that sex ratios approach 1 : 1 for both species [17], and assuming all individuals are sexually mature, the spatial range for locating the nearest potential mate exceeds at least 50 m, presumably resulting in rare reproductive encounters. Consequently, the development of larger eyes in males, which on average enhances detection distance by a few metres as demonstrated in figure 3a, may effectively reduce the dead zone and improve reproductive encounter rates.

We also reveal that, in smaller males ($<$8 cm for M. niger and $<$7.5 cm for P. guernei), the bioluminescent signalling gap closes and even inverts for M. niger (figure 3). Notably, males of both species attain sexual maturity at these smaller sizes [17]. The absence of a detection gap in small and newly mature males suggests the evolution of a signalling adaptation aimed at maximizing reproductive prospects as soon as males become capable of reproduction.

Indeed, larger males are present in populations of both species and they would likely seek to mate with females of any size (and vice versa) to maximize reproductive chances amid sparse densities. This, coupled with the prevalent luminescence detection gap, suggests that luminescent signalling is not the sole mechanism for mate detection. Complementing luminescent cues, female dragonfishes could narrow this gap by actively seeking out males. Simultaneously, both males and females might use olfactory cues to find one another. Given their small size and presumably limited swimming capabilities [33], locating a male after detecting a bioluminescent signal could waste valuable time [11]. The employment of olfactory cues, a common strategy among many fishes in mate recognition [34,35], could not only minimize the energy spent in searching, but may represent the initial cue received by either sex.

In several groups of deep-sea fishes, males possess an enlarged olfactory apparatus relative to females [33,35–37], suggesting that males may actively seek out females by following chemical trails [38]. The presence of olfactory dimorphism and whether such searching behaviour applies to dragonfishes remains unexplored. Nonetheless, it is intriguing to consider the possibility that male dragonfishes signal their location to females with a luminescent flash, and upon visual detection, females might then use olfactory cues to navigate towards the male, closing the distance before spawning.

The reason behind sexual dimorphism in photophores is unclear and we encourage any future research that assesses this phenomenon across a wider range of species as it relates to encounter rates, size dimorphism and ecological and sexual selection more generally. Should a female be capable of discerning the distance of a signal, the larger and more intense light emitted by male photophores could signal superior fitness [2], suggesting that this dimorphism could have arisen through sexual selection. However, the idea that females choose mates based adornment (i.e. photophore size), which would typically necessitate higher population densities and male competition [39], appears improbable. It seems more plausible that photophore size dimorphism has evolved as a mechanism for sex recognition [2]. In this context, the sexually dimorphic eye and the greater detection distance it affords males may have also evolved as a means to mitigate differences in the range at which sex could be determined between potential mates.

Ethics

This work did not require ethical approval from a human subject or animal welfare committee.

Data accessibility

Code and data for our analyses are available at [40] and the Dryad Digital Repository [41].

Supplementary material availabe online [42].

Declaration of AI use

We have not used AI-assisted technologies in creating this article.

Authors’ contributions

T.V.: conceptualization, data curation, formal analysis, investigation, methodology, project administration, writing—original draft, writing—review and editing; H.E.: conceptualization, formal analysis, investigation, methodology, project administration, writing— original draft, writing—review and editing; V.D.S.: conceptualization, formal analysis, investigation, methodology, project administration, supervision, visualization, writing—original draft, writing—review and editing; C.P.K.: conceptualization, data curation, formal analysis, funding acquisition, investigation, methodology, project administration, resources, supervision, validation, visualization, writing—original draft, 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 interest.

Funding

Funding for this study was provided by the Morrissey College of Arts and Sciences at Boston College and NSF grant 1103761 awarded to C.K.