a The microflume (220 × 100 × 40 mm) with a removeable working section (30 × 30 × 10 mm) was 3D printed from translucent resin and placed on top of an infrared (850 nm) LED array. b Schematic of experimental set-up. The IR light passed through the flume and the overhead camera recorded rheotaxis trials at either 200 or 60 fps onto SD cards. The timing and duration for the camera and flume pump onset and offset of was controlled by an Arduino and pump voltage (i.e., water flow velocity = 9.74 mm s⁻¹) was controlled by a rheostat. Each trial was monitored via the live camera feed displayed on the PC and all videos were copied in duplicate onto a 12TB RAID array. Dots on fish larva indicate seven body positions tracked by DeepLabCut.

a–f Representative confocal max intensity projection images of the: a–c) mid posterior lateral line (MidLL) fourth neuromast (L4); and d–f second anterior supraorbital (SO2) neuromast from the fish cohorts used for behavior experiments. Hair cells were labeled with an antibody against Otoferlin (HCS1; green a–f, gray a′–f′). Afferent neurons were labeled with an antibody against Calbindin (magenta), and cell nuclei were labeled with DAPI (blue). g Quantification of the grand mean (±SEM) number of hair cells per neuromast in intact (CTL), CuSO₄- and neomycin-treated fish. Each dot represents the mean number of hair cells from the MidLL (L3, L4, and L5) or SO (left and right) neuromasts from an individual fish. Data were collected from fish used in three experimental behavior trials; n = 4–6 fish per condition per trial. Significance values: **<0.01, ***<0.001, ****<0.0001.

a Under no flow (t = 0–10 s), groups of lateral line intact (control, n = 248) and lesioned fish (CuSO4, n = 204; neomycin, n = 222; 18 experimental sessions) have a random distribution of individual mean body angles. b, c Under flow, all groups show a statistically significant orientation to 0° ± 45°, but the distributions of the individual mean angles within the groups differ between the initial (b; t = 10–20 s) and final (c; t = 20–30 s) stimulus bins. Each dot outside the circles represents the mean body angle of an individual fish for the 10 s duration of the no flow (a) and flow stimulus conditions (b, c). The grand mean vector for each group is represented by a summary vector with an angle, theta, and a mean resultant length, rho, where the length of the vector represents the distribution of individual angles around the mean angle of the group. The length of the vector ranges from zero for uniform distributions, to one for distributions perfectly aligned with the mean angle. Consequently, the angular variance (1- rho) is inversely related to vector length. Distributions with the same lowercase letter indicate groups that do not differ statistically.

Red diamonds in each plot indicate the mean ± SE values. a Lateral line intact fish (gray = control, n = 248 fish) have a longer mean duration of rheotaxis events during flow stimulus than lesioned fish (blue = CuSO₄, n = 204 fish; green = neomycin, n = 222 fish; 18 experimental sessions). b The mean number of 0° orientation and rheotaxis events was greatest for neomycin fish and the least for CuSO₄ fish under no flow and flow conditions, respectively. c Under no flow, neomycin fish traveled a greater total distance than control and CuSO₄ fish; but under flow, neomycin and CuSO₄ fish traveled a greater total distance than control fish. d Compared to control and neomycin fish, CuSO₄ fish had the longest mean latency to the onset of the first rheotaxis event after flow stimulus presentation. Lines indicate statistical comparisons between control and treatment groups (see Supplementary Tables 4–7). The effects of treatment are indicated by long color-coded bars with branches, whereas interactions are indicated with short bars. Significance values: *=0.05, **=0.01.

a, b For all possible mean body angles (0°- 360°): a all treatment groups of fish show similar density, or total spatial use, of the arena under no flow conditions; b however, the controls (control, n = 248 fish) used the front of the arena and the lateral-line lesioned fish (CuSO4, n = 204 fish; neomycin, n = 222 fish; 18 experimental sessions) used the back of the arena under flow. c, d Filtering the data for mean body angles required for rheotaxis (0° ± 45°): c all groups clustered along the left and right sides of the arena under no flow; d but under flow, the controls used the front-left and the lesioned use the back-right portions of the arena. Dotted lines and labels in capital letters indicate spatial regions of interest (ROI). For clarity, statistical comparisons among treatment and ROIs were not included in the figure; however, there were significant fixed effects for CuSO4 and neomycin treatments, the back ROI, and interactions between treatment and ROI (Supplementary Table 8).

Time series of the proportion of individual fish within each group that performed rheotaxis during flow presentation. A greater proportion of lateral line intact fish (gray = control, n = 248 fish) performed rheotaxis than lesioned fish (blue = CuSO₄, n = 204 fish; green = neomycin, n = 222 fish; 18 experimental sessions). The data for all treatments converges after 17 s of flow presentation.

(Note: the relative velocity and acceleration periodicity data mimicked the patterns observed in relative movement; see Supplementary Fig. 4). Gray = control (n = 248 fish), blue = CuSO₄ (n = 204 fish), green = neomycin (n = 222 fish). Spectral decomposition of the observed data (a, d, g) removed the noise (Supplementary Fig. 3g, j, k) to reveal the overall underlying trends (b, e, h) and the periodicity, or recurring fluctuations (c, f, i) that occurred during any given 1 s of the experiment. The periodicity waveform peaks (c, f, i) indicate the average amount (amplitude), number, direction (positive = increasing; negative = decreasing), and order of occurrence for these cyclic fluctuations as a function of unit time (1 s). The overall trends were that CuSO₄ treated fish had the least relative movement (b), while the control fish more rapidly oriented to 0° (e) and swam with more angular variance (h; 1 – mean length of the resultant vector) compared to lesioned fish. The periodic fluctuation in relative distance moved (c) was greatest in CuSO₄ treated fish compared to control or neomycin treated fish. However, the fluctuation in mean body angle (f) was greatest in neomycin treated fish compared to control and CuSO₄ fish, while the fluctuation in mean length of the resultant vector (i.e the angular variance; i) was greatest in control fish compared to lesioned fish.

Because frequency and period are inversely related, the low frequency peaks to left of the periodograms indicate cycles with longer periods, and vice versa The amplitude of the peaks indicates the spectral density, or the number of movement events at a given frequency that occurred during the experiment. The peaks with the greatest amplitude indicate the fundamental or dominant frequencies of fluctuation in the periodicity data. The frequency and amplitude of three most dominant peaks were summed to calculate the net shifts in frequency and power. Relative to controls (gray, n = 248 fish), lesioned fish (blue = CuSO₄, n = 204 fish; green = neomycin, n = 222 fish) had a net downshift in the three dominant frequencies of a relative movement, b velocity, and c acceleration and a net upshift in d mean body angle of larval zebrafish during rheotaxis. For the dominant frequencies of e mean resultant length, there was a net downshift and upshift for CuSO₄- and neomycin-treated fish, respectively. Furthermore, relative to lateral line intact fish, the peaks of lesioned fish are clustered into fewer peaks of greater amplitude over a relatively narrow range, which indicates that lateral line ablation increased the number yet reduced the temporal variation of changes in movement.

The correlograms depict how an above average increase in relative movement (a, b), relative velocity (c, d), or relative acceleration (e, f) were significantly cross correlated with above average increases or decreases in the mean body angle (a, c, e) or mean resultant length (b, d, f). In the figures for mean body angle (a, c, e), the positive and negative peaks indicate that fish were oriented to the right or left of the oncoming flow vector, respectively. In the figures for mean resultant length (b, d, f), the positive and negative peaks indicate fish that had a lesser or greater variance of the mean body angle, respectively. The X-axis indicates the relative timing, or lag, of the cross correlation between the angular parameter with respect to an above average increase in the linear parameter (zero = simultaneous occurrence; negative = angular change occurs before linear change; and positive = angular change occurs after linear change). For example, interpret panel 9a as: In control (gray, n = 248) fish, above average increases in relative movement are most significantly correlated with changes in mean body angle to the left of the flow vector that previously occurred. In CuSO₄-treated (blue, n = 204) fish, above average increases in relative movement are most significantly correlated with above average changes in mean body angle that subsequently occurred. In neomycin-treated (green, n = 222) fish, above average increases in relative movement are most significantly correlated with below average changes in mean body angle that subsequently occurred.