Group size and aquatic vegetation modulates male preferences for female shoals in wild zebrafish, Danio rerio

Ethics statement

The study complied with the existing rules and guidelines outlined by the Committee for the Purpose of Control and Supervision of Experiments on Animals (CPCSEA), Government of India, the Institutional Animal Ethics Committee’s (IAEC) and guidelines of Indian Institute of Science Education and Research (IISER) Kolkata. All experimental protocols followed here have been approved by the Institutional Animal Ethics Committee’s (IAEC) and guidelines of Indian Institute of Science Education and Research (IISER) Kolkata, Government of India. No animals were euthanized or sacrificed during any part of the study, and behavioral observations were conducted without any chemical treatment on the individuals. At the end of the experiments, all fish were returned to stock tanks and continued to be maintained in the laboratory.

Procuring subject animals and maintenance

We used wild-caught zebrafish (from Howrah district, West Bengal, India), bought from a commercial supplier. The fish were maintained in the laboratory in mixed-sex groups of approximately 60 individuals in well-aerated holding tanks (60 × 30 × 30 cm) filled with filtered water. The lighting in the laboratory was maintained at 14 hL:10 hD to mimic the natural LD cycle in zebrafish. They were fed commercially purchased freeze-dried blood worms once a day alternating with brine shrimp Artemia. The holding tanks were provided with standard corner filters for circulation. They were maintained in the laboratory for six months before experiments were conducted to ensure they were all adults and were reproductively mature. Holding room temperature was maintained between 23 and 25 °C.

Experimental setup

The experiments were conducted in a square glass arena (83 × 83 cm), with a half-diagonal of the square from the center that approximated ten fish standard body lengths (i.e. 40 cm, assuming one body length of adult zebrafish to be about 4 cm) (Fig. 1). Each corner of the arena was provided with a square chamber (of sides 10 cm) built from transparent mesh (using synthetic fish nets) for housing the females. This design allowed for the stimuli females to be localized in the patches and not escape into the arena while simultaneously ensuring that the test males can have visuo-chemical communication with the females. The center of the arena was provided with a removable chamber (with holes) for acclimation of the males prior to the trial.

Three sets of experiments were performed to test their association preferences under (1) only varying female densities (2) increasing female and vegetation densities and (3) increasing female densities with decreasing vegetation.

Association preference experiment with varying female densities

For this experiment, each small chamber within the arena housed two (low number), four (medium number), eight (high number) or no (blank) females. These chambers represented patches of varying female numbers. The position of the female-containing chambers, as well as the composition of females within each patch, was randomized between trials. A total of 20 males were tested for their association preferences. Details on the data collected are provided in Supplementary File S1.

Association preference experiment with vegetation

For this experiment, the female-housing chambers (patches) were provided with vegetation (using artificial plants) of varying density (Fig. 2). Each subject fish was tested under two experimental settings. In E1, the number of females was proportional to the density of associated vegetation cover. We used four different densities of females, each associated with different densities of plants

1. 1.

   one female + no plants (no vegetation—N)

2. 2.

   two females + two plants (low vegetation—L)

3. 3.

   four females + three plants (moderate vegetation—M) and

4. 4.

   eight females + five plants (high vegetation—H).

For E2, we interchanged in the vegetation cover for the two and eight female patches. The patch composition in E2 set were as follows

1. 1.

   one female + no plants (no vegetation—N)

2. 2.

   two females + eight plants (high vegetation—H)

3. 3.

   four females + three plants (moderate vegetation—M) and

4. 4.

   eight females + two plants (low vegetation—L).

All test males were tested in E1 and E2 on consecutive days in no particular order. Details on the data collected are provided in Supplementary Files S2 and S3.

Experimental protocol

For the experiment involving association preferences with only varying female numbers a total of 20 males were tested, while 24 males were tested for experiments on the association preferences in varying female numbers combined with vegetation density gradients (E1 and E2 experiments). The experiments were performed two months’ apart to ensure the fish do not retain any memory from the first experiment, and thus they could be treated as two independent sets. We isolated subject males of comparable sizes and kept them in individual isolation in 500 ml jars for four days prior to experiments as that allowed us to keep track of individual fish and also stimulated mate-seeking behavior^21,22. They were fed freeze-dried blood worms every day at constantly maintained feeding times. The gravid females that were used for the experiment as stimuli for association were isolated (about 22 females) in a small holding tank (30 × 20 × 20 cm) with a feeding regimen similar to the test males. Before the start of each trial, we introduced the females into each chamber (patch) randomly (according to the experimental setup described above) and left them there for 15 min. for acclimation. A single male individual was then gently introduced into the central cylindrical chamber (with a hand-net), open at both ends (made of transparent plastic and provided with holes). After a five-minute acclimation period, the chamber was slowly removed to allow the male to swim freely in the arena and video recording was commenced. Video recordings were done using a camera (Sony DCR-PJ5, Sony DCR-SX22) placed perpendicularly above the arena. The test fish (males and females) were fed only after the end of experimental trials, on each day of experiments. At the end of the trials, the fish were returned to their holding tanks. No subject male fish were tested more than once per experimental setup and trial. The females used for the patches, were housed together (but separate from their male counterparts) in a smaller tank. Before the trials the females were picked randomly and assigned into each patch. During the experiment, the position of females being used was randomized between trials from patch to patch, to avoid the possibility of bias among the subject males for any particular females in the patches.

We recorded the behavior of each test fish for 10 min. All videos were analyzed using the software BORIS²³. A single visit to any of the patch was denoted when the male approaches within 6 cm (1.5 times their average body length) of the patch. We collected data on three parameters: total number of visits to each patch, the total amount of time spent in each patch and the mean time spent per visit within each patch. The same overall protocol was followed for all sets of experiments.

Statistical analyses

We noted the total number of visits to each patch, the total duration of time spent in each patch and mean time spent per visit per patch for the entire ten minutes duration of video recording for each test male. We calculated preference index (I) the total number of visits (I_visit) and total time spent (I_time) for each patch as proportion of the total visits made to all four patches²⁴.

• I_visit for patch A = No. of visit to patch A/(visit to patch A + visit to patch B + visit to patch C + visit to patch D).

• I_time for patch A = time spent in patch A/(time spent in patch A + time spent in patch B + time spent in patch C + time spent in patch D).

All statistical analyses were performed in R studio (version 1.1.463)²⁵. We developed generalized linear mixed models (GLMMs) using package glmmTMB (version 0.2.3)²⁶ with ‘fish’ as the random factor and ‘Patches’ as the fixed factor, with four levels representing the four choices for the test (male) fish. Preference for total number of visits (I_visit) as well as total time spent (I_time) were found to fit beta distribution with values ranging between 0 and 1. For data fitting, we added 0.0001 to every value, to remove zeroes. Relevelled models were used to compare the parameters between the four patches. Link = logit was used under beta family to construct the GLMM models.

For analyzing the data for the second and third experiments involving varying female densities along with vegetation densities (E1 and E2), we followed a similar procedure of constructing a GLMM followed by post hoc tests. GLMM models were constructed with a single independent variable, “patch”, that had four levels, designated as H (high vegetation density), M (moderate vegetation density), L (low vegetation density) and N (no vegetation).

Source: Ecology - nature.com