Model species
Lysmata shrimp are protandric simultaneous hermaphrodites. An individual matures as a functional male and later on, with increasing body size, gains the female function to become a fully functional simultaneous hermaphrodite [18, 21]. L. amboinensis De Man (Caridea: Hippolytidae) was classified as 'pairs' species by Bauer [18] and as 'tropical-pairs' species by Baeza [35] due to its social monogamy. It is small in size (approx. 6 cm total length) [21] and occurs in the tropical waters of the Indo-Pacific or the Red Sea. It is often referred to as cleaner shrimp due to the cleaning behaviour it performs on fish clients [17, 22].
As in all caridean species, individuals mating as females undergo a moult prior to mating and spawning [36]. Only for a few hours after moulting, hermaphrodites are receptive and can receive sperm from a male phase or another hermaphroditic phase partner [37]. Thus, the sex roles are defined by the moulting cycle of each individual. The mean intermoult interval reported from Fiedler [21] ranged from 20.8 to 18.3 days, but was much shorter in our experiment (approx. 15 days in pairs). Fertilization is external [36]. The male role can be attained anytime, even while incubating eggs. A mating event is generally non-reciprocal [18, 38, 39]. Although it is mechanically possible self-fertilization does not occur [21, 39]. There is no long-term sperm storage [14, 38, 39].
Maintenance of shrimp
L. amboinensis was obtained from De Jong Marinelife B.V. (Lingewaal, Netherlands) and imported from the Philippines. Shrimp were allowed to acclimatize to laboratory conditions for 31 days. We did not have precise information about the age of each shrimp, but individuals should not have differed in their behaviour as we assured by the presence of eggs that the hermaphroditic phase was accomplished in all individuals. The single treatment was a control to assign the frequency of moulting in the absence of a mating partner. Subsequently, shrimp were separated into treatment groups in a size-controlled randomized manner. Carapace length (CL) was used to assort candidates of approximately the same body size into the same tank to minimize size effects. The different body size classes were spread evenly among treatment groups. The remaining size differences between the largest and the smallest individuals within one tank were (mean ± SE) 0.038 mm ± 0.010 mm in pairs, 0.148 mm ± 0.065 mm in triplets and 0.211 mm ± 0.047 mm in quartets.
Individuals were allocated to one of four treatments: single shrimp were kept in 15 l (25*25*25 cm), pairs in 30 l (35*35*25 cm), triplets in 45 l (42*42*25 cm) and quartets in 60 l (49*49*25 cm) aquaria, with the same water level (height of tanks = 25 cm), resulting in an equal surface area per individual. All tanks were connected to a single water circulation system (1440 l). The ground of each aquarium was covered with coral sand. Shrimp were maintained at 24 to 25 °C water temperature, 34 to 35 ‰ salinity (Tropic Marin® Sea Salt) and a 12 h light:12 h dark cycle. Partial water change (200 l) in the circulatory system and removal of debris were carried out weekly.
Each treatment consisted of 10 replicates (pairs: N = 20 shrimp, triplets: N = 30 shrimp, quartets: N = 40 shrimp) except for the single treatment, which consisted of 6 aquaria (N = 6 shrimp). Aquaria were spatially randomized across a three-level rack. Opaque PVC plates between tanks prevented visual contact. To minimize visual stress within tanks they were compartmentalized into zones with incomplete opaque dividers. The number of zones was equal to the number of shrimp in the tank. Individuals were allowed to move freely throughout the aquarium. Every zone contained one plastic perch with a uniform overhang. Personal observations of L. amboinensis have shown that individuals prefer to sit on some structures (rocks, corals or artificial perches) rather than simply on the ground. Offering perches increased the likelihood that shrimp show their natural behaviour. Shrimp were fed daily (Tropical® Shrimp Sticks), resulting in ad libitum food condition.
Measurements
The size of all shrimp was measured before the experiment started. We used carapace length (CL), defined as the distance from the posterior-most margin of the eye orbit to the mid-dorsal posterior margin of the carapace [40], measured to the nearest of 0.01 mm. We used digital photographs (Olympus CAMEDIA C-8080), which were taken in a standardized way and measured CL using the image analysis software ImageJ 1.39e.
Shrimp were marked individually within each tank by clipping short parts of the distinct antennae to record moulting events. Antennae regenerated after every moult and were re-clipped in the same way. We could not detect signs of stress or harm caused by this treatment. Clipping allows indentifying exuviae from moulted individuals, which is not true for coloured elastomer tags, which are injected into the abdominal musculature [39]. Each individual was checked daily, noting each moult the morning after. Exuviae were removed from the aquaria. The date and shrimp ID were noted in case of mortality.
Video observations of newly assembled shrimp quartets were carried out after the experiment in order to reveal the reason for mortality in groups. We used quartet tanks and groups of 4 individuals accordingly (N = 6 replicates). Recordings were done at night using infrared illumination (two infrared spotlights: LED - Infrared Illumination, Model: IR880/12), five cameras (35x Zoom Day&Night IR Cut Auto Iris Vari Focal Camera, Model: CCD1000H35/3.6-126) and a digital video recorder (8 Channel Triplex Digital Recorder, Model: DVR008TPX/SA800NC). Observation and data collection took place for 84 days starting September 24, 2007.
Statistical analyses
To test whether mortality is dependent on the number of conspecifics in the tank, i.e. whether mortality of the first individual occurred sooner in groups of four compared to three individuals, we compared the lifespan of the first individual that died in triplet and quartet tanks using a survival analysis. We did not compare survival between all individuals and all treatment groups, as the data points within one tank were not independent of each other.
Although shrimp within the same tank were chosen to be of approximately same body size, the remaining small size differences may explain mortality. Alternatively, moulting sequence might have determined mortality within each tank. To evaluate mortality in groups, the relationships between body size, moulting sequence and mortality were quantified for triplets and quartets by computing partial Kendall's τ correlation coefficients between lifespan, size rank and moulting sequence within tanks. We calculated the coefficients within tanks first, to avoid pseudoreplication, as the individuals in one tank were not independent of each other. The largest individual in each tank obtained the highest rank regarding size (= 4). The individual that moulted first in each tank obtained the lowest rank regarding moulting sequence (= 1). We used a Kruskal-Wallis-Test to see, if the correlation coefficients (Kendall's τ) of triplets and quartets were significantly different from each other. At the same time we tested if Kendall's τ was significantly different from zero, i.e. we compared three groups in the Kruskal-Wallis-Test: triplets, quartets and zero. If the overall differences were statistically significant, a post-hoc analysis was performed using the Steel-Dwass test for multiple comparisons. We applied non-parametric tests since the residuals of the correlation coefficients were not normally distributed.
To test whether group size had an effect on moulting, we compared the daily number of moults per survivor and tank for triplets and quartets when they were in groups (more than two individuals) or in pairs (only the two survivors) after mortality of conspecifics within the tank. We used a multivariate analysis of variance (MANOVA) for repeated measurements with the daily number of moults per survivor and tank of groups and of the resulting pairs as dependent variables, the group size as between subjects factor and time as within subjects factor including the interaction between treatment and time. Subsequently, we used a Welch ANOVA to compare the daily number of moults per survivor and tank among all treatments after the group treatments had changed to pairs to test whether the remaining pairs originating from triplets and quartets showed a difference in the number of moults compared to singles and pair treatments where no mortality had occurred. All p-values stated are two-tailed. All statistical analyses were carried out using JMP® Version 9.0.2 © 2010 SAS Institute, Inc.