Personality has been identified in a range of animal taxa during the last few decades, with important ecological and evolutionary implications. Investigating the effects of environmental factors during early life can provide important insights into the ontogeny of animal personality. We reared newborn mosquitofish, Gambusia affinis, in tanks of different structural complexities, and measured their behavioral traits (i.e., shyness, exploration, and sociability) when they reached sexual maturity. Univariate linear mixed-effects models were fitted to test the effects of environmental complexity and sex on population-average behavior, whereas multivariate models were fitted to quantify behavioral repeatability (i.e. personality) and among-individual correlations (i.e., behavioral syndromes). On average, females were shyer and more social than males, and the fish reared in complex environments were shyer, less explorative, and more social than those reared in open environments. Among-individual differences were consistently large across trials for all behaviors, indicating that personality variation was present in mosquitofish of both sexes reared in different environments. Repeatability did not differ among behaviors, and there were no differences in repeatability in any behavior between sexes or among environments. A negative correlation between shyness and exploration was found in mosquitofish from all treatments at both phenotypic and among-individual levels, with the latter indicating a strong shyness-exploration behavioral syndrome. Our study provides robust evidence that average levels of personality might vary when mosquitofish are raised in different levels of structural complexity during early life.

Study species and rearing conditions

A total of 150 gravid female mosquitofish were caught during May 2018 in artificial ponds on the campus of Anhui University (117.18°E, 31.77°N) and were housed in three 200-L holding tanks filled with oxygenated tap water at approximately 26°C. They were fed fine-grained commercial food (TIDDLER, Weifang YEE Pet Products Co., Ltd., China; 42% crude protein, 5% crude fat, 5% crude fibers, 11% ash) twice per day and exposed to a natural photoperiod (~ 14:10 L:D). We checked the housing tank frequently for newborn fry and transferred them on the day after birth to 30 net tanks (blocks: 80 × 80 × 80 cm, mesh size: 0.177 mm), which were placed in an undisturbed artificial pond (60 × 10 m) on the same campus. To avoid potential genetic effects, we randomly and evenly assigned the fry born in each day to the 30 tanks. Due to this assignment procedure, the number of mothers contributing to the fry and the sex ratio of the fry in each tank were unknown but were assumed to be approximately equal. In total, 1500 newborns were collected within one week and were reared, that is, 50 individuals in each tank with a maximum age gap of one week, until they were sexually mature. No predation on the fish occurred during the rearing period, and the mortality was assumed to not differ among the rearing tanks.

The 30 rearing tanks were equally divided into three groups with different levels of structural complexity, that is, open (A), medium (B) and complex tanks (C). Two locally common plants, Nymphaea tetragona and Alternanthera philoxeroides, were placed into the medium and complex tanks to enrich the environment, with approximately 30% of the water area covered in the medium tanks and 70% in the complex tanks. Three complexity levels were found in the ponds where the mother fish were captured. To maintain the level of complexity, plant coverage inside the tanks was checked visually each day, and the extra parts due to plant growth were manually removed. To avoid the effect of the tank location, tanks housing different groups were placed alternately (i.e., ABCABCABC). There was little plant coverage in the rearing pond with a water depth of approximately 40 cm, and a large natural population of mosquitofish lived in this pond for several years. The water temperature ranged from 20°C to 32°C and the pH ranged from 7.4 to 7.6 during the rearing period. The fish in the tanks were fed brine shrimp nauplii and zooplankton during early ontogeny and then fine-grained commercial food two weeks later. Apart from being fed additional food, the ambient environment of the reared fish was identical to that of the natural population. Such a design allows fish to experience a near-identical environment in the field, thus avoiding behavioral abnormalities that might arise in laboratory conditions (Carere et al. 2015).

Two months after the experiment started, we randomly caught four females and four males from each rearing tank, for a total of 240 individuals at an average age of 50 days, to characterize their behavioral traits in the following experiments. The sampled subjects were individually kept in labeled, black, opaque cylindrical tanks (height: 9 cm; diameter: 15 cm; hereafter, the holding container) with transparent lids filled with oxygenated tap water before and between experiments. A black, opaque, cylindrical refuge chamber (height: 5 cm; diameter: 7 cm; hereafter, temporary refuge) was placed in the center of each holding container, and fish were given 24-34 h (varying due to test orders on the first day of the experiment) to acclimate to the chamber before experiments. All fish were kept in a laboratory room with a natural day-night photoperiod, and were not fed 12 h before the experiments.

In China, mosquitofish is an invasive species that is prohibited from being released into natural water bodies; thus, ethics committee review was not required to capture mosquitofish (Ministry of Environmental Protection of the People’s Republic of China 2016, Index No.000014672/2016-01463). The behavioral experiments were approved by the Institutional Animal Care and Use Committee of Anhui University (IACUC, AHU). No harm was done to the fish during the experiments.

Personality assays

We measured shyness, exploration, and sociability for each subject in an individual arena, which was a white opaque plastic tank (37 cm long × 30 cm wide × 20 cm high; Figure 1). A camera (Sony HDR-CX510, 55× extended zoom, Sony Corporation, Tokyo, Japan) was suspended above each identical arena to record the behavior of the fish. The tanks were filled with oxygenated tap water (3-cm deep for shyness and exploration, 5-cm for sociability), and the water was exchanged between trials to exclude any effect of chemical signals released by previous individuals on subsequent animals. Behavior recordings were carried out in a laboratory with sufficient light (for clear observations), constant temperature (26°C), and no disturbances. To avoid observer effects, the experimenters were shielded from the tested subjects by a 1.5-m high opaque curtain during each trial. Behavioral trials for each subject were repeated three times every other day, with the test of shyness and exploration on alternate days to the test of sociability. To achieve a large sample size (i.e., three times for the three behaviors of 240 individuals) within one week, experiments were performed by six experimenters with each in charge of four arenas simultaneously. To avoid potential observer effects, the experimenters were trained before experiments to follow the standard procedures detailed as follows, and the fish observed by a certain experimenter were exchanged between days. The trial orders were randomized and fish were fed commercial food after each trial. At the end of all three behavioral trials, standard body length (accurate to 0.1 mm) was measured for each subject before fish were released into the artificial pond where they had been reared.

Assay 1: shyness and exploration

Shyness and exploration for each subject were tested in one continuous trial. Before the experiment, one black, opaque, cylindrical refuge chamber (the same refuge placed in the holding container, hereafter referred to as starting refuge) was fixed at the middle position of one end of each tank. There was a sliding trapdoor (3 × 3 cm) on the side of the starting refuge facing the arena. A piece of fishing line was connected to the trapdoor, and hence, experimenters behind the opaque curtain could pull the fishing line to remotely open the refuge without disturbances to allow fish to swim out of the chamber and toward the arena. The rest of the tank (approximately 27 × 30 cm) was furnished with novel objects to enrich the environment; thus, the fish could not see through the arena (Figure 1a). To avoid potential effects of habituation toward the arena, we used different novel materials, that is, gravel, plastic sheets, and fresh lotus leaves in the three repeated trials. The materials were scattered in the same configuration for different subjects.

At the beginning of each trial, a randomly selected fish was gently (in water) coaxed into the closed starting refuge, and then the camera began recording. The subject was allowed to acclimate for 5 min, after which the observer remotely opened the trapdoor of the starting refuge and left it open until the end of the trial. The time taken by the subject to emerge out of the starting refuge onto the arena (latency time) was used to measure shyness, with shyer subjects emerging later from the shelter. Whereas some researchers have defined shyness-boldness as an individual’s reaction to explicit risks (Réale et al. 2007), we followed many previous studies (Brown and Braithwaite 2004; Brown, Burgess et al. 2007; Brown, Jones et al. 2007; Toms et al. 2010; Cote et al. 2011; Huntingford et al. 2013; Polverino, Cigliano et al. 2016; Polverino, Ruberto et al. 2016) that measured shyness in small fish as an individual’s latency to emerge from a familiar refuge outside which there might be predation risk. We considered a fish to have emerged from the starting refuge when its entire body crossed the trapdoor. All tested fish swam out of the refuge within 13 min. The camera continued recording the movement of the subject for another 10 min after it left the refuge. After the exploration assay, the subject was immediately transferred back to its holding container. We extracted 600 image stacks from the 10-min movement videos (one frame per second) and used Image J (http://rsbweb.nih.gov/ij/) to mark the location of the fish head in each frame to delineate its movement pathway. Similar to Edenbrow and Croft (2013), we used the total pathway length to quantify the exploration score of the focal subject. We acknowledge that the measurement of exploration could not be unambiguously differentiated from activity because animals should be active while exploring (Carter et al. 2013). While this measurement might concurrently include activity, we referred to this test as exploration because it was performed in an unfamiliar environment in which the distance covered has been proposed to be a measure of exploration (Sih et al. 2004; Réale et al. 2007).

Assay 2: sociability

Sociability was tested on the day following the assay for shyness and exploration. The arena was not furnished with novel structures, and no starting refuge was placed in it. Similar to the method described by Sommer-Trembo et al. (2017), a perforated, transparent, plastic, cylindrical bottle (diameter: 6 cm; length: 23 cm) containing 10 non-experimental conspecific individuals (five females and five males) as a stimulus shoal was fixed in the middle of the arena. One dark line was marked on each side of the bottle to define a social zone (width: 13 cm; Figure 1b). The width of the social zone at each side of the stimulus shoal was smallest near the water surface (i.e., 3.5 cm, approximately twice the average body length of the test fish) and increased with water depth. A randomly selected subject was gently introduced into one side of the arena (exchanged between trials) and was given 5 min to habituate to the new environment. After 5 min, the camera was switched on to record the location of the subject for 10 min. We determined the sociability of each subject as the time the focal individual spent in the social zone during the 10-min period.

There are no missing values. We encourage anyone wanting to use this data to contact the authors.