experimental setup for testing behavioral responses to olfactory stimulation in individual, 6 weeks old, Astyanax mexicanus larvae.

A, Each fish is placed individually in 150ml water in a rectangular testbox placed on an infrared light emitting table. Control or odorant solution are delivered at the two extremities of the box, inside tubes covered with a net. Medium to high throughput behavioral testing is achieved by parallel recording of 32 testboxes placed under 8 infrared recording cameras (4 testboxes per camera).

B, After one hour or 24 hours habituation, the test consists of one hour IR video recording. The first 30 minutes provide the control/baseline behavior of individual fish. Odorant stimulation is given at 30 minutes, and the behavioral responses are recorded for 30 more minutes.

basal swimming behaviors of 6 week-old CF, SF and F2.

A, Examples of typical and distinctive basal swimming patterns exhibited by Astyanax mexicanus morphotypes. The behavioral patterns (Random, Wall Following, Circles, Thigmotactism) and the individual fish shown are indicated. Each pattern is best described by the combination of 3 types of representations. (1) the top left track shows the displacement of the fish along the x-axis of the box, representing back-and-forth swims along its long dimension. (2) the bottom left track is the top view, 2D representation of the trajectory. (3) the right track is the 2D plus time (color-coded) representation and helps, for example, to discern between wall following and thigmotactism.

B, Distribution of basal swimming pattern displayed by SF, F2, and CF after 1h or after 24h of habituation. Numbers of fish tested are indicated. The elementary color code is indicated (R in blue, WF in red, C in green, T in yellow). Exact swimming patterns and combinations are given on the colored plots (Suppl. Fig7 for the full color code). Fisher’s exact tests for statistical comparisons between groups. Of note, in this and subsequent figures, the swimming pattern color code does not relate whatsoever with the time color code used in the 2D plus time representation of swimming tracks such as in panel A.

C, Boxplots showing the swimming speed in SF (blue), F2 (green) and CF (red), after 1h or 24h of habituation. Values are mean speed calculated over a period of 15 minutes. Numbers of fish tested are indicated. Mann-Whitney tests p-values are shown.

D, Examples of the stability of the basal swimming pattern over 4 experimental days in 3 individuals, with 24h of habituation time. One SF displaying Random pattern, one CF displaying Wall Following and one F2 displaying Thigmotactism + Random swim are shown.

Controls on the olfactory setup and solution delivery method.

A, Lack of behavioral response in a representative CF individual after injection of water on one side of the testbox.

B, Lack of change in qualitative swimming pattern displayed by SF, CF and F2, before and after water injection on one side. Fisher’s exact tests.

C, Lack of change in quantitative swimming kinetics in SF (blue), CF (red) and F2 (green), before (pale colors) and after (dark colors) water injection on one side. In this and the following graphs, boxplots show the swimming speed, the fish position along the X-axis of the testbox, and the numbers of back-and-forth swims along the X (length) and Y (width) axes of the box. All values are averaged over a 15 minutes period, either before or after the injection. A thin line links the value « before » and the value « after », for each individual. Numbers of fish tested are indicated on the “speed” boxplots. Numbers next to dots indicate the identity of outlier individuals.

Behavioral responses to alanine of 6 week-old CF, SF and F2.

A,B,C: Representative individual responses of CF, SF and F2 after injection of alanine at the indicated concentrations. In the two left graphs of each panel, the blue color indicates the water/control injection side and the yellow color indicates the alanine injection side.

D,E,F,G: Box plots showing swimming speed (D), mean position along the X-axis (E) and the number of back and forth trips in X and Y (F,G) in SF (blue), CF (red) and F2 (green), before (lighter color) or after (darker color) the injection of alanine at the indicated concentration. Values are calculated over a 15 minutes period. Black lines link the « before odor » and « after odor » value for each individual. Numbers close to black dots indicate the identity of outlier individuals. P-values from paired Mann-Whitney tests are shown. The number of fish tested is indicated.

H, Change in swimming patterns exhibited by SF and CF after injection of alanine 10-4M. Fisher’s exact tests.

Behavioral responses to serine and cysteine 10-2M.

A,B and H,I, Representative individual responses of CF and SF after injection of serine (A,B) or cysteine (H,I) at high concentration (10-2M). In the two left graphs, the blue color indicates the water/control injection side and the green (serine) or red (cysteine) color indicates the injection side.

C-F and J-M, Box plots showing swimming speed (C,J), mean position along the X-axis (D,K) and the number of back and forth trips in X and Y (EL and FM) in SF (blue), CF (red) and F2 (green), before (lighter color) or after (darker color) injection of serine 10-2M (C-F) or cysteine 10-2M (J-M). Values are calculated over a 15 minutes period. Black lines link the « before odor » and « after odor » value of each individual fish. Numbers close to black dots indicate the identity of outlier individuals. P-values from paired Mann-Whitney tests are shown. The number of fish tested is indicated.

G and N, Change in swimming patterns elicited after injection of 10-2M serine (G) or cysteine (N) in SF and CF (F2 not shown). Fisher’s exact tests.

Behavioral responses to lysine, histine and leucine 10-2M.

A-D, Box plots showing swimming speed (A), mean position along the X-axis (B) and the number of back and forth trips in X and Y (CD) in SF (blue), CF (red) and F2 (green), before (lighter color) or after (darker color) the injection of the indicated amino acid. Values are calculated over a 15 minutes period. Black lines link the « before odor » and « after odor » value of each individual fish. Numbers close to black dots indicate the identity of outlier individuals. P-values from paired Mann-Whitney tests are shown. The number of fish tested is indicated. See Supplemental Figure 4 for examples of representative individual responses.

E-G, Change in swimming patterns elicited after injection of 10-2M lysine (E) or histidine (F) or leucine (G) in SF, CF and F2. Fisher’s exact tests.

Summary of the results.

A, Population level summary. For each amino acid and each concentration tested, arrows indicate whether the considered parameter has changed (increased or decreased). ns indicates no significant change, - indicates the condition was not tested in the fish type (SF in blue, CF in red, F2 in green).

B, Individual level summary. The percentage of fish displaying an individual olfactory score superior to 1.5 is indicated.

Individual olfactory scores of SF, CF and F2 for different odors.

A, Graphs representing index values (i.e., the variation between the ‘before’ and the ‘after’ odor condition; the value 0 corresponds to no change) for the 4 response parameters (speed, thigmotactism in X and Y, position and swim pattern) used to calculate the total individual olfactory score for each fish (last column of points on each graph). Each fish is depicted by a colored line linking its 4 different indexes and its final individual olfactory score. The dotted line at value 1.5 indicates the score threshold above which a fish is considered as a responder. The percentages in black and grey indicate the proportion of responders/non-responders, respectively. The colored lines of responders are bright, those of non-responders are pale. Amino acids and concentrations are indicated. Top row: CF; bottom row: SF.

B, Distributions of individual olfactory scores of SF (blue), CF (red) and F2 (green) for different odors. The threshold score (1.5) is indicated by a dotted line.

The individual olfactory score is related to individual fish swimming personality.

A, Box plots showing individual olfactory scores in SF (blue), F2 (green) and CF (red) as a function of their individual baseline swimming pattern. As stated in Methods, only amino acid conditions for which more than 40% of fish were responders (score >1.5) are pooled and plotted on this graph. Mann-Whitney two-tailed with Bonferroni post hoc tests were performed on each morphotype.

B, Regressions to explore the correlation between baseline swimming speed displayed by individual fish before odor injection and their individual olfactory score. Fish and amino acid type and concentrations are indicated. Linear correlations were calculated with Spearman’s rank correlation test followed by Student test for the p values. Conditions for which significant correlation was found are labelled with an asterisk.

C, The PVE (Percentage of Variation Explained) on the individual score is plotted as a function of individual baseline behavior, to probe the predictability of the behavioral output as a function of “swimming pattern personality”.

D, The PVE (Percentage of Variation Explained) on the individual score is plotted as a function of individual swimming speed, to probe the predictability of the behavioral output as a function of “locomotor activity personality”. Linear correlations were calculated with Spearman’s rank correlation test followed by Student test for the p values. Conditions for which significant correlation was found are labelled with an asterisk. SF (blue), F2 (green) and CF (red).

Additional controls for experimental design.

AA’, Distribution of swimming patterns exhibited by SF, F2 and CF (number of fish tested is indicated) along 4 testing days, with 24 hours habituation time (A). Fisher’s exact tests. The color code for the different swimming patterns and combinations is given in the inset below. The table in A’ shows statistics summary for paired comparison of individual’s mean position, swimming speed and the total number of round trips in X plus Y over 4 testing days, using Friedman tests. The results of additional pairwise Wilcoxon comparisons between different days are as follows. All these pairwise comparisons were NS except for CF roundtrip numbers (d1/d2 p=0.04; d1/d3 p=0.0009; all other comparisons NS) and F2 position (d1/d3 p= 0.022; d3/d4 p= 0.042). These results suggest that fish display individual swimming behaviors that are stable through time and may be considered as individual swimming personality or temperament.

B, Lack of change in swimming pattern displayed by SF, CF and F2, before and after water injection on both sides. Fisher’s exact tests.

C, Global lack of change in quantitative swimming kinetics in SF (blue), F2 (green) and CF (red), before and after water injection on both side. Mean speed and mean position are shown.

D, Lack of change in quantitative swimming kinetics in SF, CF and F2 when no injection is performed at all. Mean speed and mean position are shown. The result suggests that an individual’s behavior is stable throughout the one-hour test.

E, Lack of change in swimming speed or position on the box when SF are recorded in the dark (dark blue), in the light (light blue), or when the light is switched off after 30 minutes.

Additional examples of behavioral responses to alanine

A, Examples of CF responses to alanine 10-3M (intermediate concentration). Note the reproducible switch in position toward the odorant side in the box (yellow shading) and the change in behavioral pattern. Note that CF n°8, which was tested twice with the same concentration of alanine, switched from wall following to small circles and thigmotactism along the Y border of the box in both cases. CF n°4 on the other hand swam in large circles before injection, and changed its pattern to display intense thigmotactism accompanied with a few circles.

B, Examples of SF responses to alanine 10-2M (high concentration). Upon injection, SF show changes in swimming patterns and increase their swimming speed, suggesting that the stimulus was perceived. SF n°22 swam very actively randomly. SF n°4 switched from random with small circles to circles plus thigmotactism. SF n°12 added thigmotactism to its random pattern upon stimulus. Some SF, like SF n°8, did not seem to respond, even at such high alanine concentration.

C, Examples of F2 hybrids responses to alanine 10-3M (intermediate concentration). F2 n°14 did not appear to move its position to the odorant part of the box, but a striking change in swimming pattern from intense thigmotactism to random swim was observed. F2 n°7 explored close to the odor source and decreased swimming speed. F2 n°3 and 1 switched from wall following to circles with thigmotactism associated to a change in position towards the odor side.

D, Change in swimming patterns displayed by SF, F2 and CF after injection of alanine 10-2M, 10-3M and 10-5M. Fisher’s exact tests.

Additional examples of behavioral responses to serine and cysteine

A, responses to serine 10-2M (high concentration). Note the change in behavioral pattern and position in the box observed in CF and F2 but not in SF. The two F2 shown had different baseline swimming patterns: circles in F2 n°71, thigmotactism in F2 n°79. Yet, both individuals showed a change in swim pattern and an attraction to the serine side of the box.

B, responses to cysteine 10-2M (high concentration) and 10-3M (intermediate concentration). CF n°73 gives an example of an individual that did not respond to 10-3M cysteine: no change in position, no change in pattern, no change in swim kinetics. CF n°71 and CF n°72 on the other hand did respond to cysteine 10-2M, the former by an increase in activity, the later by a transient change in position. SF n°34 is a rare example of a SF that was attracted to cysteine.

Examples of behavioral responses to lysine (ABC), histidine (DEF) and leucine (GHI) 10-2M (high concentration)

In each panel, in the two left graphs, the blue color indicates the water/control injection side The lysine injection side is shown in red, the histidine injection side is yellow, the leucine injection side is pink.

A, The 2 CF individuals shown decrease their swimming activity upon injection. Moreover, n°81 and 87 shift their position to the odor side of the box. A change in swimming pattern from intensive wall following to thigmotactism with circles and some random swimming is also observed.

B, The 2 SF individuals shown do not change their swimming pattern (Random) or kinetics in response to lysine.

C, F2 n°35 increased swimming activity upon lysine injection but did not show attraction to odorant side. F2 n° 37 kept its random pattern of swimming and was not attracted to the odorant side.

D, Whether the cavefish individual was initially displaying thigmotactism (CF n°8) or swimming in large circles (CF n°4), the response to histidine included decreasing the swimming activity and restricting locomotion to the odorant side of the box.

E, The two surface fish individuals shown swam slowly and randomly as baseline pattern. Both increased their activity and spent more time in the odorant side.

F, F2 hybrids n°11 and n°1 were both attracted to histidine odorant side and they changed their swimming pattern.

G, CF n°65 showed no response, neither in activity, in position nor in pattern. CF n°4 drastically changed all swimming parameters.

H, Surface fish showed little reaction to leucine. SF n°43 and n°40 persisted in random swimming patterns, without change of position or swimming kinetics.

I, F2 hybrids n°16 and n°13 had markedly different baseline patterns (large circles versus random swim, respectively). The former changed its pattern to less active and small circles, suggesting that it perceived leucine injection. The later showed no behavioral response.

Individual olfactory scores of SF, CF and F2 in control and experimental conditions.

The graphs represent index values for the 4 response parameters used to calculate the total individual olfactory score for each fish. Each fish is depicted by a colored line linking its 4 different indexes and its final individual olfactory score. Experimental conditions and fish types are indicated. The threshold value for olfactory score was set at 1.5 (dotted line), because very few individuals of the 3 fish types have scores above this value in control experiments when water is perfused in the box on both sides. The percentages in black and grey indicate the proportion of responders/non-responders, respectively. The colored lines of responders are bright, those of non-responders are pale.

Relationships between the number of round trip in X and the olfactory score

A, Regressions to explore the correlation between baseline number of round trips in X displayed by individual fish before odor injection and their individual olfactory score. Fish and amino acid type and concentrations are indicated. Linear correlations were calculated with Spearman’s rank correlation test followed by Student test for the p values. Conditions for which significant correlation was found are labelled with an asterisk.

B, The PVE (Percentage of Variation Explained) on the individual score is plotted as a function of individual swimming parameter (numbers of round trips in X), to probe the predictability of the behavioral output as a function of “locomotor activity personality”. Linear correlations were calculated with Spearman’s rank correlation test followed by Student test for the p values. Conditions for which significant correlation was found are labelled with an asterisk. SF (blue), F2 (green) and CF (red).

Complete color code for the description of baseline swimming patterns.

Correspondence Analyses (CA) for assessment of behavioral pattern change after different stimuli.

Confident ellipses for each condition in the behavioral space are shown in blue for SF, green for F2 hybrids, red for CF. The conditions are indicated at the top of the graphs.

A, Control experiments. CA shows that behavioral patterns in the 3 types of fish are clearly distinctive in the behavioral space: confidence ellipses are strictly non-overlapping. Moreover, CA confirms that for the 3 types of fish, the swimming pattern does not vary along the one hour test when no injection is made (nothing/nothing), when water is injected on one side (water/nothing), or when water is injected on both sides ((water/water). In all cases, confidences ellipses overlap in the “before” and “after” condition. These analyses fully confirm the results shown in Figures 2B and 3B with the colored bar plot representation.

B, Odor stimulation experiments. Changes in behavioral patterns is the behavioral space are illustrated for the 6 different amino acid tested. Overlap of confidence ellipses indicate a lack of change in behavioral pattern after odor stimulus, whereas separated confidence ellipses indicate a change in behavioral pattern after odor stimulus. The CA fully confirms the results shown in Table1 and Figures 4-5-6 with the colored bar plot representation.