Some male cypridinid ostracods (ca. 2 mm) utilize luminescent secretions in the most unique and complex light displays yet described in marine environments. While swimming rapidly males secrete pulses of luminescence to attract females, but females do not reciprocate with light production. Females have been hypothesized to be attracted to and exhibit choice between these signals but this remained untested. Their small size and nighttime mating activity necessitated laboratory experiments that used light mimics of the male’s patterns in order to observe individual female behavior. In this study, we asked 2 simple questions: are females attracted to the light patterns that mimic conspecific male display patterns, and, if so, what do females actually do on detecting and responding to this signal pattern? Using a blue light-emitting diode array, we mimicked the male display of Photeros annecohenae in the lab and, in conjunction with infrared lighting and infrared cameras, we recorded individual behaviors of virgin females. Here, we provide direct evidence that females respond to the pulsed light mimic and approach the intermittent signal in such a way that they would cross closely above the path of an upwardly signaling male. In order to maximize her proximity to each pulse in the display, a responding female swims at a compensatory angle above the display to adjust her vector to intercept the male above each preceding pulse. These data support the hypothesis that luminescence is the initial signal used by males to attract receptive females in courting ostracods.
In many animal systems, individuals visually detect a moving target such as a conspecific or prey item, make a decision to pursue it, and subsequently approach the target in a manner so that it will ultimately result in interception. The method by which organisms identify, approach, and intercept moving visual targets has long been of interest and has been studied in organisms ranging from hoverflies (Collett and Land 1978) to humans (McBeath et al. 1995; Port et al. 1997; Lenoir et al. 1999; Fajen and Warren 2004). The majority of mammalian studies have focused on how individuals intercept moving inanimate targets such as Frisbees (dogs; Shaffer et al. 2004) and balls (humans; McBeath et al. 1995; Lenoir et al. 1999; Chardenon et al. 2002, 2005; Chohan et al. 2008), whereas invertebrate studies have primarily focused on insects, especially with regard to mate acquisition (Collett and Land 1978; Wehrhan et al. 1982; Land 1993; Boeddeker and Egelhaaf 2003, 2005; Niven 2006) and prey or host capture (Wagner 1986; Baird and May 1997; Gilbert 1997; Olberg et al. 2000, 2005, 2007; Combes et al. 2012; see Olberg 2012).
Before any interception behavior is initiated, the target must first be detected and recognized as being of interest. What aspect of the target signal prompts pursuit? Various aspects of the target could be the target’s size, shape, color, visual pattern, and speed of its trajectory through the environment. In nearly all interception studies, the target signal has been continuous (e.g., Collett and Land 1978; McBeath et al. 1995; Lenoir et al. 1999; Olberg et al. 2000; Shaffer et al. 2004). Although the signal may be observed intermittently by the tracking individual either by extraneous circumstances (i.e., by a loss of contrast between the target signal and background) or by design (i.e., tiger beetles stopping their movement to reassess their moving prey target against a static background; Gilbert 1997), the actual signal itself does not disappear. An example describing truly intermittent visual signals utilized for tracking and interception is female Photinus fireflies using luminescence for aerial guidance to intercept and consume extraspecific firefly males (Lloyd and Wing 1983). However, this firefly study was simply a report that the phenomenon occurs and did not discuss the method by which interception transpires.
After recognizing that a particular target is worth pursuing, the pursuer must move in a manner that it eventually intercepts the target. Interception can result from either plotting an interception course or copying the exact motion path of the target. Dogs and humans plot and intercept flying targets by maintaining a constant visual angle with their target (McBeath et al. 1995; Lenoir et al. 1999; Shaffer et al. 2004). Male hoverflies chasing females (Collett and Land 1978) and dragonflies chasing prey (Olberg et al. 2000; Olberg 2012) also use shortcuts by plotting interception courses to their targets. A requirement for these successful interceptions is the ability for the individual to predict, with some degree of accuracy, the future location of its target. In contrast, in order to overtake the female, males chasing females in many fly species attempt to copy the exact motion path of the target (Land and Collett 1974; Wehrhan et al. 1982; Land 1993), but they must be moving at a higher rate of speed than their target.
In mating systems where males utilize species-specific luminescent courtship displays to attract females, describing female response behavior has been restricted to only a few firefly and syllid polychaete worm species (Lewis and Cratsley 2008; Deheyn and Latz 2009 and references therein). Although luminescence is more prevalent in marine than terrestrial environments (Hastings and Morin 1991), behavioral descriptions of marine bioluminescence in courtship are rare. Here, we present not only the response but also describe the interception behavior of individual females to a distinctly intermittent visual signal, which is, to our knowledge, the first study to do so in detail. Found only in the Caribbean, males of a clade of tiny (<2 mm) myodocopid ostracods produce a series of external light pulses in spectacular, species-specific courtship displays at the end of dusk (Morin 1986; Cohen and Morin 1990; Morin and Cohen 1991, 2010). Although the male displays of many new species have been described, virtually nothing was known about how, or even if, females responded to them. Below we describe how female Photeros annecohenae respond to and intercept complex, intermittent, and moving male displays.
Photeros annecohenae is found in large numbers in shallow (1–10 m depth) seagrass beds off Southwater Caye, Belize (Torres and Morin 2007; Rivers and Morin 2008). There is clear sexual dimorphism, with females being much larger than males: males are 1.62 ± 0.05mm (±SD) in length, whereas females are 1.99 ± 0.05 mm (±SD; Gerrish and Morin 2008) (Figure 1). Adults have been kept in the lab up to 6.5 months and the time from brooded embryo to adulthood is about 3 months, which yields a total life span of about 9.5 months (Gerrish and Morin 2008). Although all life stages are able to luminesce for antipredation purposes, only males do so for courtship with the most complex displays described to date in marine systems (Morin 1986; Cohen and Morin 1990; Morin and Cohen 1991; Herring 2000; Rivers and Morin 2008, 2009, 2012).
The well-defined light organ consists of 4 types of long exocrine cells that terminate on 3 rows of nozzles and a pair of lateral tusks on the upper lip (Huvard 1993). Luciferin-containing cells lie above the medial row of nozzles, luciferase-containing cells terminate at the 2 lateral rows of nozzles and also the tusks, mucous cells terminate in the tusks, and the fourth, least common, cell type of unknown function terminates at the lateral rows. Huvard (1993) has proposed that differential contraction of specific muscles around and within the light organ rapidly extrudes luciferin and luciferase together within a rind of mucus, resulting in a discrete bolus of luminescence.
The lateral compound eyes are found on relatively short, movable stalks at the anterior-dorsal part of the body beneath the transparent carapace (Figure 1). Each eye has only 16 ommatidia with a total mean eye diameter of 0.28 mm in males and 0.22 mm in females (see Cohen and Morin 2010). Each ommatidium is approximately 25 µm in diameter (Huvard 1990).
There are a number of aspects of this system that make it tractable for analyzing mating behavior: 1) distinct sexual dimorphism; 2) females brood and produce crawl-away juveniles (there is no planktonic larval stage; Torres and Morin 2007); 3) thousands of individuals can be collected in a single night in baited traps; and 4) males exhibit apparently normal luminescent courtship behavior in aquarium tanks in the lab (Rivers and Morin 2008, 2009). It should be noted that we are able to differentiate sexes at the 5th instar (Adult-1 or last juvenile) stage, which allows us to isolate females and rear them as virgins for our experiments. Until recently, their small size and nocturnal activity have been prohibitive for answering more detailed questions about female behavior. Although infrared (IR) light attenuates rapidly in seawater, in laboratory tanks it is sufficiently reflected off the carapace of ostracods so that we can observe and record individual behavior, while at the same time not interfering with normal courtship activity.
We have previously described the courtship display and male mating behaviors (Rivers and Morin 2008, 2009). Commencing about 45 min after sunset if no moon is present in the sky and for the succeeding hour, male P. annecohenae exhibit vertical displays (up to 60cm above the top of seagrass) that consist of up to 19 bright blue pulses of light secreted from nozzles in the males’ upper lip. Each pulse is left behind the rapidly swimming, spiraling male as a discrete packet of light in the water column. Each display train has 2 phases. The first, or “stationary,” phase, which we hypothesize to be an identification and alerting phase, consists of 3–4 short (260–430ms), bright pulses with some interpulse interval variation (0.8–1.3 s), but with little upward movement (mean −0.31 cm, most <±2 cm) at or just above the top of the sea grass. The second, or “helical,” phase, which we hypothesize to be a predictable targeting phase, immediately follows as the male swims upward in a tight helix and secretes as many as 16 shorter (210ms), dimmer pulses with more consistent intensity and interpulse distances of about 4cm and interpulse intervals of about 750ms (Rivers and Morin 2008). Because the pulses of light themselves are stationary in the water column and no 2 pulses are visible at the same time, each pulse, by itself, gives no information on the directionality and speed of a displaying male. Taken together, however, the pulses yield information on the distinct, intermittent trajectory.
Competing males exhibit 2 alternative behaviors with respect to a displaying male: they may entrain by producing their own competing, similar luminescent pattern nearby and in loose synchrony with the initiator, or they may sneak by swimming above and close to the displaying male but without luminescing (Morin 1986; Rivers and Morin 2009). Therefore, a single male is capable of initiating, entraining, and sneaking, and they switch rapidly between tactics during the hour-long display period (Rivers and Morin 2009).
Unlike fireflies where there is often a luminescent call-and-response “dialogue” between males and females (Lloyd 1966), in P. annecohenae, females do not engage in luminescent duetting (Morin 1986; Morin and Cohen 1991). This lack of luminescence by females is more akin to many auditory mating systems, such as crickets and frogs (Greenfield 1994; Wagner and Reiser 2000), than to other systems utilizing light as a signal. We hypothesize that by remaining dark, females can exhibit more choice in the courtship arena (water column) where the operational sex ratio is highly skewed toward males (>175:1; Rivers and Morin 2008) so that they do not attract overwhelming male attention. The lack of dialogue suggests that females, of necessity, have a highly developed and species-specific swimming behavior in order to track, intercept, and copulate with her target male, without having to rely on any initial behavioral change in a displaying male.
In our initial attempts to observe female behavior, we found that when we combined males and females in a tank, the confined spaces and absence of refuges for females allowed males to find females and attempt to force copulations prior to any female choice or tracking behavior. Because no broods resulted from these attempts, we concluded that they were unsuccessful. Placing males and females in separate tanks was not ideal for 2 reasons. First, the rate and number of displays were somewhat unpredictable, with sometimes up to 15–30 min before the first displays commenced, and no guarantee of large numbers of displays. Second, the IR light levels necessary for tracking females throughout a large (70 cm [height] × 60 cm [width] × 15 cm [depth]) laboratory tank often overpowered display luminescence detected by our equipment, so that only the first few pulses of a male display were detectible on video; thus, we were unable to determine whether or not a female was reacting to luminescence. Therefore, in order to allow unhindered observation of female behavior to light signals that we are able to detect with our cameras, we designed a blue light-emitting diode (LED) array that mimicked male courtship luminescence patterns in space, time, and pulse characteristics. The mean average pulse durations, interpulse intervals, and interpulse distances of 10 natural male displays in the lab were used to construct the LED array “code.” Females responded to these artificial signals by a change in swimming behavior.
Females used for the LED trials were collected (along with juveniles and males) off the south shore of Southwater Caye, Belize (16°48.74′N, 88°4.98′W) in small 4-cm-diameter × 8-cm-long PVC pipe traps with 500-µm mesh funnels at each end and using fish muscle as bait. Virgin subadult females (5th of a total of 6 instars), termed A(adult)-1 juveniles, were isolated and reared in natural seawater without males in 15-cm (ca. 0.5L) Tupperware containers (and fed Tetramin tropical fish flakes every 2 days) until each molted to adulthood (6th instar). Trials using these virgin females were carried out from June to July 2005 and January to March 2006.
LED display mimic
The LED display that was used to mimic the helical phase of male ostracod displays consisted of a wand of 12 blue (460nm) LEDs (E934MBD, eLED, Walnut, CA) arranged vertically 3cm apart to mimic the spatial patterns of the display trains of males collected in the field (see Rivers and Morin 2008). The wand was located immediately outside and behind the aquaria and approximately 20cm (one-third of the way) from the left side (Figure 2). By using resistors and 3 sheets of 66% reduction neutral density filters, we closely matched the light output of each LED to the intensity of male displays. Pulse durations and interpulse intervals (the time between the start of one pulse to the start of the next) were controlled by an Atmel ASTK500-ND AVR pulse microcontroller starter kit (Atmel Corporation), using an AT90S8515 8-PC microcontroller (Atmel Corporation). The microcontroller was programmed with the CodeVision AVR C Compiler program (HP InfoTech) to mimic the temporal patterns of the display trains of males collected in the field (Rivers and Morin 2008).
On days when there were no experimental trials, females were maintained in their containers under natural light conditions (approximately 12–12 light–dark cycle). During experimental nights, females were illuminated with a 15-W fluorescent light used to simulate daylight until they were selected for recording. For each trial, 8 virgin females were allowed to swim freely together in a 70 cm (height) × 60 cm (width) × 15 cm (depth) clear acrylic tank filled with clean seawater (70 L) collected on the lagoon-side of the island off the dock and within about 100–200 m of the display grounds. Females were left in the tank with an overhead incandescent light on for 20 min, after which the light was extinguished. We then waited for an additional 20 min before recording. IR illumination for filming was supplied from above (1 cm above the waterline) by 3 rheostat-controlled 15-W red frosted incandescent bulbs further restricted by an IR barrier filter. A high-sensitivity (0.00015 lux) low-light 1.25-cm CCD camera (Watec LCL-902K) with a 12-mm aspherical low-light TV lens (Computar HG1208FCS-HSP) was connected to a Sony DCR VX-2000 miniDV camcorder, which was used in VCR mode. After 5 min of recording without displays, the LED display was turned on for 15 min. The 5 min of darkness and 15 min of displays were repeated twice more for a total trial time of 1 h. At least ten 1-h trials were performed during each field season, for a minimum of 20 trials (totaling 20 h).
If females are responding to and approaching the light signals, there are only 4 variables that must be determined: the female’s swimming speed, her interception angle (A), her position (F) at the start of an approach, and the male apparent swimming speed. The male apparent swimming speed was determined by Rivers and Morin (2008) and provided the parameters for the LED display model. We analyzed 4 main components of the female swimming pattern that would indicate a response to the courtship display mimic: 1) height of interception point (IP in Figure 2) above the LED to show trajectory; 2) the linearity of swimming (line b in Figure 2) between pulses to show consistency of the trajectory; 3) the angle difference (A in Figure 2) from a straight-line trajectory to the LED; and 4) the overall patterns and slope of swimming to show, if linear and targeted, a focused directionality. We applied these a priori stringent requirements to over 20 h of display and nondisplay treatments. We selected and analyzed 25 separate swimming patterns: 11 from virgin females that indicated positive behavioral responses to the LED display trains by showing a change in swimming direction, angle, or speed (e.g., see Figure 3a), and, as a control, we analyzed 14 comparably positioned randomly chosen (but initially oriented in the same direction as responding females) swimming patterns of females when no LED displays were being emitted. We examined the female response to each of the 12 pulses in a display for every train we analyzed, but only while she was still approaching the display. Once she passed the LED array, we no longer analyzed the patterns; thus, some patterns had fewer than the 12 pulses possible. Because females were almost always swimming in the upper half of the tank, and also to maximize observable behaviors free of sidewall interference prior to a female intercepting the LED display, we restricted our analyses to only those females (both experimental and control) that were initially in the upper right-hand side of the tank furthest from the wand mimic.
We analyzed responding female swimming patterns by marking the female’s position every 2 frames (66.7ms), as well as the location and timing of the individual LED pulses. For analysis, pulse height above the bottom was used instead of pulse number to control for the slight variation of pulse distances during a display. Positions and distances were measured using Image-J 1.32J image-analysis software (NIH). Interception points were calculated by plotting the vector of the female 166.7ms (5 frames) after each pulse (to allow time for swimming correction) and extrapolating a straight line (line b in Figure 2) to the interception point (IP in Figure 2) of the female (F in Figure 2) above the vertical display on the y axis. Using the projected interception trajectory, we also calculated the angle difference (A in Figure 2) of the female swimming trajectory from a straight-line approach to the previously illuminated LED. Similarly, angles and distances were also calculated by using Image-J. The swimming pattern was digitized, and best-fit lines were calculated to determine the overall swimming slope and linearity of each female trajectory (r2). The linearity of swimming between pulses was calculated from the r2 of best-fit line calculated 166.7ms after each pulse.
Because our computer-controlled displays were constant in the experimental trials, we analyzed the responses of the control females using the equivalent timing of the display to mark where the female would be if, in fact, a display mimic were occurring. Thus, the protocols were identical except for the actual emission of the displays from the LED array.
All statistical analyses were performed in SAS (9.2), where we confirmed that all assumptions of the models were met.
When exposed to an LED display that closely mimics the helical phase of a P. annecohenae male luminescent display, conspecific females will orient to and then track the display train so as to intercept the path just above the display (Figure 3). When we examined the display versus dark (control) periods, we saw an overall difference in swimming behavior. For example, in 1 typical 15-min display period, we identified 27 potential behavioral responses to luminescence by females throughout the tank. Of these, we selected only the ones in the upper right-hand portion of the tank for analysis as described above. In contrast to display periods, in a 15-min dark, nondisplay control period, we only detected 3 swimming patterns that even superficially resembled responses to luminescence. When an LED display commenced, responding females that were initially swimming away (e.g., Figure 3b; Supplementary Video) turned to track the light signal after the second or third pulse; a single pulse of light was not sufficient to trigger a behavioral response. Subsequent pulses produced further changes in direction as females oriented toward the display. We analyzed the approach angles, interception trajectories, and swimming linearity of females only after they proceeded to swim toward the display; swimming trajectories prior to orienting to the displays were not analyzed.
When a female approached close to the display mimic, she continued swimming diagonally through the vertical axis of the display trajectory rather than altering course to swim upward along the display path presumably because the LED mimic was outside the tank and no additional sensory cues other than light were available to a female.
Overall slopes and linearity
Females responding to LEDs (e.g., Figure 3a,b) swam at a downward trajectory from horizontal, toward the LEDs situated below them (mean slope = −0.48 ± 0.26 [SD]), whereas control females (e.g., Figure 3c) swam with a nearly flat trajectory and stayed at a constant depth in the tank (mean slope = −0.03 ± 0.13 [SD]; general linear model [GLM]: F23 = 15.57, P = 0.0006). Responding females also swam significantly more linearly during this time, with a mean r2 value of 0.96 versus 0.48 in control females (GLM: F23 = 18.28, P = 0.0004). Horizontal distance from the position of the LED had no significant effect on the overall slope of either responding (GLM: F10 = 1.31, P = 0.2853) or control females (GLM: F13 = 0.24, P = 0.6407). However, although vertical distance above the display was not significant in control females (F10 = 5.02, P = 0.0555), it was significant in responding females (F13 = 6.32, P = 0.0361). The higher the females were in the tank when displays started, the steeper their overall downward swimming trajectory.
Interception point distances
Between pulses, females responding to a display always swam so that they shortened the distance toward the display, whereas control females swam variably, often increasing the distance from the display (Figure 3). Factors that might influence the female intercept height above the LED were 1) height of a pulse above the bottom (which corresponds to pulse number), 2) the horizontal distance of the female from the LED, 3) the female’s vertical distance either above or below the LED, 4) whether the pattern was from the control or response treatment, and 5) various interactions of the above factors. Because we had a maximum of 12 LED pulses/display per individual (some individuals showed fewer because they passed the LED array prior to the 12th pulse), we used a random coefficient mixed model to account for individual variance of 234 total observations. We used the Kenward–Roger degree of freedom approximation to compute the denominator degrees of freedom.
We found significant differences in interception patterns for responding compared with control females (Figure 4). Note that in order to provide the most conservative indication of behavioral changes, we chose to analyze only those (both control and responding) females whose orientations and swimming directions were approaching the display. Because displays are produced vertically upward, both experimental and control females showed a decrease in the interception point height above the previous LED (Figure 3), but their patterns differed. We found, unsurprisingly, that there was a significant effect of the height of the pulse above the bottom (which corresponds to the pulse number) and interception point distance above the previous pulse in both responding and control females: as pulse height increased, interception point distances decreased (F1,17.7 = 62.04, P = <0.0001). However, although interception point distances decreased with pulse number in both responding and control females, the interception point distances were significantly smaller in responding compared with the control females (F1,23.9 = 3.95, P = 0.0006). In addition, there was a significant interaction difference between the interception point height and treatment (F1,18.1 = 7.74, P = 0.0123). Not only are the interception trajectories closer to the pulse in responding females but also the rate at which females reduce their interception point distance was also significantly lower than in control females. This difference indicates that responding females were correcting their swimming vectors (Figure 4).
The actual position of the female in the tank did not significantly affect the interception height above the LED in either the horizontal (F1,97.1 = 1.37, P = 0.2453) or vertical direction (F1,83.3 = 0.17, P = 0.6787). However, the interaction between pulse height off the bottom (number) and horizontal distance was significant (F1,37.7 = 4.49, P = 0.0408); as horizontal distance and pulse height (number) increased, there was a decrease in interception point distance. The interaction between pulse height and vertical distance was also significant (F107 = 50.35, P = <0.0001); as vertical distance and pulse height increased, there was an increase in interception point distance.
Linearity between pulses
For the same reasons outlined above, we used a random coefficient mixed model to account for individual variance of 234 observations of linearity (r2) between pulses of the 11 responding and 14 control females. Responding females were swimming significantly more linearly (r2 = 0.89 ± 0.19 [SD], N = 95) between pulses than control females (r2 = 0.70 ± 0.32 [SD], N = 145; F1,25 = 11.28, P = 0.0025).
Swimming angles between pulses
The angle from the LED to the interception point with the female as the focal point was significantly different between responding and control females (t-test of unequal variance, t157 = 9.13, P < 0.0001). The mean angle of control females above the LED was 54.9 ± 20.9° (SD; N = 129), whereas their swimming vector was essentially flat (0.1 ± 17.3° [SD; N = 129]). In comparison, the mean angle of responding females 166.7ms after a luminescent pulse (to allow for angular correction) was 23.9.7 ± 16.1° (SD; N = 60). Unlike control females in the dark who swam horizontally, responding females swam downward at an average of 20.8 ± 12.5° (SD; N = 60) from horizontal, with females higher up in the water column during the first few pulses having a steeper downward angle (Figure 5).
In addition, when responding females did correct their swimming angles (indicated by an inflection point along their swim path) while approaching the display, they always did so by reducing their vector (swimming slope became more horizontal; mean = 9.8 ± 20.8 [SD; N = 95]); we did not see females increasing their downward angle (Figure 6). By contrast, control females showed no consistent pattern (57 increased; 88 reduced their vector) with an overall angular correction of −0.35 ± 18.33 (SD).
Because males respond to another male’s display by increasing their swimming speed by nearly 50% (Rivers and Morin 2009), it might be expected that females would respond similarly. However, unlike males, we found no difference in the mean speeds of control (4.53cm/s) versus responding females (4.58cm/s; t20 = −0.167, P = 0.869).
Male behavioral responses to females
A different behavior was observed from males confined together with females in aquaria in the dark. With IR videography we have documented multiple (n > 20) instances of males making dramatic shifts in swimming direction or pattern in the presence of females, which were either swimming or stationary on the side of a tank. Many of these males approached and grasped the female and the pair then dropped toward the bottom of the tank, spinning rapidly with the male sometimes luminescing intermittently, but each female was always able to dislodge the male (sometimes after several minutes, but most within seconds), after which the male would usually begin to display. After each of these cases, the females were isolated, but none produced a brood, thus indicating no probable insemination and hence copulation failure (n = 20).
Female responses to luminescence
We have identified certain criteria that had to be met, in proper sequence, to demonstrate that a female is responding to the light mimic. First, the female must ultimately swim in the direction of the display. Secondly, the calculated interception points must be at or above the previous pulse in the train. Thirdly, the female must compensate her trajectory toward the display to always reduce, and not increase, the distance between the female and each successive pulse. Finally, the female responding to a display should follow a nearly linear, nonerratic trajectory (after allowing for a course correction). Deviation from any of these criteria would indicate that the female might not be responding to the display. Using these strict criteria, we were able to confirm more than 50 additional responses of females to the displays across multiple trials and reject the patterns of 3 control females that superficially resembled female responses.
Requiring both responding and control females be swimming in the direction of the display meant that interception height above the display would necessarily decrease for both control and responding females. Even so, the data show that the rate of decrease of interception point distance is significantly lower in responding females compared with control females (Figure 4), which indicates that a female responding to an actual display is continually compensating her swimming vector to intercept the train above the next pulse, thus ensuring that when she does intercept the display she will be within a 2–3 cm of the succeeding pulse. Furthermore, the trajectory of responding females is closer to the LED pulse and with less angle variation than females (Figure 5a–c).
An animal can intercept a moving target either by copying the exact motion path of the target, also called “pursuit” (Land and Collett 1974; Wehrhan et al. 1982; Land 1993), or by calculating some sort of course intercept (e.g., Collett and Land 1978; Olberg et al. 2000; Shaffer et al. 2004; Olberg 2012). In cases where the target is moving in a very predictable manner, the fastest, most direct method of interception would be to follow a linear optical trajectory (e.g., McBeath et al. 1995; Shaffer et al. 2004) or a constant bearing angle. This method has been well documented in both vertebrates and invertebrates (McBeath et al. 1995; Lenoir et al. 1999; Olberg et al. 2000; Shaffer et al. 2004; Olberg 2012). Similarly, when approaching the series of light pulses, female P. annecohenae swim in a relatively linear fashion but at a compensatory angle to the last pulse. They demonstrate course corrections that maintain a certain bearing angle with respect to each subsequent pulse.
One major difference between the P. annecohenae interception strategy and most others is the only periodic availability of the visual information. When chasing a fly ball or a Frisbee, a human or dog can visually sample the target at will (barring contrast issues with the background). However, for the intermittent pulses of the ostracod displays of P. annecohenae, the amount of visual input available for the approaching female is dictated not by the female, but by the male. Recall that, between pulses, there is a period of about 500ms where there is absolutely no visual signal. In addition, because an individual luminescent pulse from a male P. annecohenae display is stationary and extracellular, it demonstrates neither directionality nor speed of the displaying male. Therefore, either 1) the female is predisposed to aim her trajectory above a single luminescent pulse at a particular angle or 2) the female calculates her trajectory from information garnered from multiple pulses. Because females do not significantly change their orientation until after the second or third pulse in a train, multiple pulses in close proximity are at least necessary. This approach pattern lends credence to our hypothesis that the first “stationary” phase of male displays acts to attract the attention of both receptive females and competing males (Rivers and Morin 2008, 2009). However, the question remains whether, once attracted, a female integrates information from multiple pulses to calculate interception or bases her response to individual, not multiple, pulses. Because in P. annecohenae, each subsequent pulse is distinct from the previous one, no 2 pulses overlap (Rivers and Morin 2008). Thus, integration of multiple pulses would require the female to remember the location of the previous pulse in space while she is moving toward it in the water column. Although this is arguably possible, it would require more complex processing than merely utilizing a preset constant bearing angle above each pulse. As long as the female swims toward and above the last detected pulse, she will pass in close proximity and just above the signaling male. On the other hand, the display trains of many other luminescent displaying ostracod species in the Caribbean consist of multiple overlapping pulses that are visible at one time (Morin 1986; Morin and Cohen 1991, 2010), which makes the integration theory in these particular species both plausible and perhaps even probable.
The simplest hypothesis for how a female responds to a display that is consistent with our results is that after orienting toward a male’s display during the bright “alerting” phase of the display (Rivers and Morin 2008), she updates her approach trajectory with each new pulse by aiming a preprogrammed angle above each pulse without integrating multiple pulses. The constant correction to a relatively consistent angle is similar to how dragonflies adjust their interception trajectory to maintain a constant bearing angle when their target deviates from a straight-line flight path (Olberg et al. 2007; Olberg 2012).
Distance to targets can be calculated by a variety of methods; animals can use the size of the target image, the intensity of the image (when the signal is luminescent), stereopsis, motion parallax, or any combination of the above. We are currently unsure of what methods P. annecohenae utilize. The compound (or lateral) eyes of luminescent cypridinids (Figure 1), which likely aid in detecting low-level point light sources (Nilsson 1996), occur on short, movable stalks just under the transparent carapace in the dorso-anterior part of the body (Huvard 1990). P. annecohenae have only 16 ommatidia in each lateral eye, with large facet diameters (ca. 50–60 µm) (Huvard 1990; Torres and Morin 2007).
Arguably the simplest method for gauging distance is target size such as is used by hoverflies (Collett and Land 1978). However, in comparison to the insects, ostracod eyes are simpler and the small pulses of light may only subtend a single ommatidium, making distance approximation using this method only very general. An intriguing alternative to target size to gauge distance, female ostracods might use the intensity of the pulse. However, because we have found some ostracod displays over 70% brighter than others (Rivers and Morin 2008), this distance approximation would also likely only be a rough estimate.
Ostracods are likely to be able to utilize stereopsis (Morin JG, Rivers TJ, unpublished data), as found in mantids (Rossel 1983, 1986), but further study on the spatial geometry of the eyes is necessary to determine how much binocular overlap individuals have. Alternatively, they may use motion parallax, as has been found in many flying insects such as honeybees (Srinivasan et al. 1989), perched insects such as grasshoppers (Eriksson 1980; Sobel 1990), and perhaps in dragonflies (in conjunction with stereopsis; Olberg et al. 2005). Because P. annecohenae eyes are on movable stalks, not fixed in the head (Huvard 1990), they could be using either vibrations of the eye, their own movement in the water column, or a combination of both along with stereopsis to determine their distance to a luminescent pulse via motion parallax.
Finally, in the unlikely event that females are integrating multiple pulses for their approach, they may be able to gauge the distance via the angles subtended between 2 pulses. The highly predictable interpulse distances (Rivers and Morin 2008) and linear swimming patterns of females make integration feasible. However, this strategy would be much simpler in species that utilize a sit-and-wait strategy (e.g., Olberg et al. 2000). Female ostracods swim over 20 body lengths a second and, because their eyes are not stationary in space as in sit-and-wait predators, the processing required to integrate 2 pulses not concurrently visible to gauge distance would be much greater than a simple preprogrammed angular correction to each individual pulse.
Because our experiments were done with display mimics and not live displaying males, we do not know whether they use additional behaviors once she locates herself above a signaling male, at which point she is also often in the vicinity of several sneaker males (Rivers and Morin 2008). Thus for a female to locate and approach a displaying male and then control the interaction, she must 1) rely solely on her ability to intercept the displaying male based on her swimming trajectory; 2) change her swimming behavior on detecting a second stimulus (e.g., chemical trail or turbulence) produced by the male; and/or 3) leave her own stimulus (e.g., a chemical trail or turbulence) so that the male can approach and intercept her. Chemical cues are known to play important roles in many invertebrate mating systems, including crustaceans (Yen et al. 1998; Vickers 2000). A plausible hypothesis supported by all these data, but not yet tested, is that a female approaching a display intercepts the vertical path above the displaying male, and then she signals to the male by producing a chemical trail to attract the displaying male. The closer she is to the luminescent pulse when she intercepts the display trajectory, the more likely that only the signaler, and not sneaker males, will detect her cue. Thus, luminescence may be the initial mating signal whereby the male calls the female, and then a secondary signal from either the male, or more likely the female, may play a role in the fine-tuning of the final approach between the individuals. This sequence would represent intersensory, sequential duetting.
Expected male behavior
Observing female responses to light pulses provides insight into what luminescent behavior we should expect to see from males. Females approach the light displays without luminescing rather than engaging in a luminescent dialogue, suggesting that it would benefit a displaying male to make as many pulses as possible during each display train. In the field, however, male–male competition is extreme and competing males may be attempting to interfere with a displaying male to minimize the number of pulses per rival display (Rivers and Morin 2009). In laboratory observations, male display train length (pulses per train), but not interpulse distance, was inversely correlated with male density (Rivers and Morin 2009). Train length determines the maximum distance a female can respond to and still intercept a male. Females swim at a mean speed of about 4.5cm/s, and the longest male train duration observed was about 16 s and about 60cm in total length (Rivers and Morin 2008). Thus, the female could be up to about 3 quarters of a meter from the termination point of the train and still intercept a male exhibiting the longest display. The average display in the field, though, contains about 12 pulses and is 10 s long (Rivers and Morin 2008). In these conditions, a female would have to be almost 30cm closer to the display in order to intercept the male, a considerable decrease in effective range. If a female were approaching the male’s display, which stopped because of other male interference, the male may be able to regain the female’s attention by displaying as soon as possible in the same area (Rivers and Morin 2008). Indeed, we see individual males display multiple times in succession in the lab in the same general location within a tank (Rivers and Morin 2009).
In the field, the display arena of the males is a flat, fairly homogeneous open grassbed unbounded by any vertical sides. However, in our experiments, females were confined in tanks in the lab, which may have introduced potential artifacts such as 1) tank edge effects, which could limit or block normal female behavior to displays, 2) the clear acrylic sides of the tank, which produce reflections of the LED display and could make it difficult for females (and us) to determine whether the female response was to the primary light source or a reflection of that light, and 3) the simple fact that the LED displays, although immediately next to the tank, were still outside the water. Because we are not yet able to track the tiny (ca. 2 mm) females in the field in the dark, we do not know the initial location of receptive, responding females in natural situations during a courtship display in the field. In the lab, the females in most cases tended to swim back and forth in the upper third of the tank, which may suggest that receptive females do the same in the field.
The behavior of P. annecohenae females to luminescent courtship displays differ significantly from other known courtship systems. Both females and males are moving many times their own body length every second (>20) in 3D space. In order to intercept a male, a female swims without luminescing and must recognize and react to multiple, rapid (ca. 200 ms) signals spaced about 4cm apart and produced every 700ms. Each male pulse, by itself, does not reveal the velocity or direction of the male. Only by integrating multiple luminescent pulses or by having a preset swimming behavioral response that closely matches the swimming pattern of the displaying male will a female be able to intercept her chosen male in the dark, open sea. We suggest that there is likely a multimodal dialogue (luminescence for males, pheromones or other cues for females and perhaps males) adds an even higher level of complexity to this system.
Supplementary material can be found at http://www.beheco.oxfordjournals.org/
This research was supported by Cornell University Chapter of Sigma Xi and Andrew W. Mellon Foundation research grants, and a Mario Einaudi Center for International Studies research travel grant to T.J.R.
We thank Jennifer Hall, John MacDougal, and others at International Zoological Expeditions (IZE) on Southwater Caye, Belize, for help and accommodations during our field research. This research was completed in accordance with permits received from the Belize Department of Fisheries. We thank Tom Seeley, Paul Sherman, Steve Shuster, Cole Gilbert, and Gretchen Gerrish for research and design help and comments; Francoise Vermeylen and Myra Shulman for statistical support. This study was completed in partial fulfillment of T.J.R.’s PhD thesis to the Department of Ecology and Evolutionary Biology, Cornell University.