From lab to ocean: bridging swimming energetics and wild movements to understand red drum (Sciaenops ocellatus) behavior in a tidal estuary
Brendan J. GibbsiD, James A. StrotheriD, C. R. Morgan, Daniele PintoniD, Alberto CanestrelliiD, James C. LiaoiD
Abstract
Understanding how fish navigate complex natural environments requires bridging fine-scale biomechanics with ecological behavior. We investigated the volitional movement and energetics of wild red drum (Sciaenops ocellatus) across laboratory, mesocosm, and field settings. Using flow-respirometry, we quantified metabolic costs and swimming kinematics under ecologically relevant flow conditions shaped by bluff bodies mimicking mangrove roots and oyster mounds. Fish swimming in turbulent wakes exhibited reduced oxygen consumption and altered tailbeat dynamics, especially at high flow speeds. In a large outdoor mesocosm, dual accelerometers revealed a rich behavioral repertoire, including maneuvering and rest, which is not easily observable in confined lab settings. Spectral analysis and clustering identified eight distinct locomotory states, highlighting the limitations of summed acceleration metrics. Field telemetry tracked wild red drum across a 54 km estuarine corridor for a three-year period through an array of 36 acoustic receivers, revealing movement patterns shaped by tidal flow and physical habitats. Hydrodynamic modeling revealed that while laboratory trials demonstrated substantial energetic savings at high flows (approaching 100 cm/s), wild fish were detected predominantly in low-velocity microhabitats (<30 cm/s) near structurally complex features. This mismatch suggests that habitat selection is an adaptive strategy driven by ecological factors such as foraging opportunities, predation refuge, and site fidelity, rather than hydrodynamic efficiency alone. Our multi-scalar approach demonstrates that while flow-structure interactions can reduce locomotor costs for fish, habitat use in the wild reflects broader ecological constraints, offering a framework for integrating biomechanics, physiology, and ecology in conservation-relevant contexts.
02 /Specification & Classification
Click any tag to browse matching papers- Citations
- 2
- FWCI
- 7.86
- Field percentile
- top 3%
- Organism
- Assay
- Research Focus
- Approach
In-depth summary
Claude Opus 5.5 · xhigh effort · 8 Nov 2026 · from the full text of version 1 (7 Apr 2026) · CC0 1.0
Red drum lowered oxygen consumption behind mangrove- and oyster-mimicking bluff bodies at flows to 100 cm/s, yet telemetered fish were detected mostly in low-velocity nearshore habitat.
Flume studies measure locomotor energetics in confined, simplified flows, but it is unclear whether flow refugia created by estuarine structures shape how wild fish use tidal habitats. This work links respirometry, mesocosm accelerometry, acoustic telemetry and a Delft3D-FLOW estuary model to test whether lab-derived flow preferences predict where fish occur in the field.
Key results
ShownHolding station in the wake of a mangrove-like (2D) or oyster-like (3D) bluff body lowered oxygen consumption compared with free-stream swimming. Savings were largest at high flow, and the 2D wake was the cheapest.
Two-way ANOVA on MO₂ (n = 5 fish; 22, 61, 100 cm/s): treatment F= 64.42, speed F= 31.36, interaction F= 18.88, all p < 0.0001. Tukey: MO₂ rose sharply with speed in free stream, changed little behind the 2D body and rose intermediately behind the 3D body; MO₂ was lower behind bluff bodies at all speeds. · Respirometry; Swimming kinematics and oxygen consumption in…; Energetic and Behavioral Strategies Across…; Figure legends
SuggestedPairing telemetry detections with Delft3D-FLOW velocities showed that wild red drum were detected disproportionately during low-to-moderate flows at nearshore receivers, even though the flume showed the largest energetic savings at high speeds.
For 5 tagged fish over the three-year simulation, nearly 90% of detections fell below 0.50 m/s and ~75% between 0.15 and 0.40 m/s, versus approximately 80% and 45% of all simulated time near receivers. Velocities were extracted within 50 m of each of 21 receivers at 30-minute intervals. · Hydrodynamic modeling; Movement and flow patterns in the wild; Energetic and Behavioral Strategies Across…
SuggestedFlume-calibrated dual accelerometers showed that mesocosm fish mostly swam slowly and expressed maneuvering and resting states that are hard to elicit in a confined flume. Spectral features separated behaviors that summed acceleration lumps together.
Tail and opercular acceleration and frequency rose with flume speed. t-SNE plus k-medoids on per-axis spectra gave 8 clusters (Swimming 1-4, Sedentary, Breathing, Slow and Fast Maneuvering) within a continuum. Swimming and Fast Maneuvering had similar summed accelerations but distinct spectra; Slow Maneuvering resembled flume 0 BL/s, Fast Maneuvering 0.94 BL/s. · Volitional behaviors in a mesocosm; Behavior in the mesocosm; Energetic and Behavioral Strategies Across…; Figure legends
ShownWake treatments changed swimming kinematics in a speed-dependent way. Tailbeat frequency fell behind obstacles at the highest speed, tailbeat amplitude was smallest behind the 2D body, and head angle differed by treatment consistently across speeds.
Treatment × speed interaction: tailbeat amplitude F = 25.25, tailbeat frequency F = 39.67 (both p < 0.001); amplitude behind the 3D body did not differ from free stream at 61 cm/s (p = 0.95). Head angle: treatment F = 32.62, speed F = 207.23 (p < 0.001), interaction F = 0.73, p = 0.58; head angle was greatest behind the 3D body. · Kinematic analysis; Swimming kinematics and oxygen consumption in…; Figure legends
SuggestedTagged fish used inshore and inlet habitats and swam both with and against tidal currents, and one fish made an offshore excursion. Most stayed near particular receivers within the zone influenced by St. Augustine Inlet, while one fish ranged more widely through the estuary.
Fish 23332 moved ~14 km around the St. Augustine inlet, including a ~9 km offshore excursion for six days in January 2020. Fish 23332 and 59225 concentrated at receivers 5 and 6, fish 23326 and 59223 at receivers 8 and 9. Fish 53427 used receivers 1 to 4 and 8-10 and moved with and against currents at simulated 0.1–0.9 m/s. · Movement and flow patterns in the wild
Methods, data and reproducibility
Intermittent-flow respirometry with DeepLabCut kinematics compared free-stream swimming against station holding behind 2D and 3D bluff bodies at three speeds. Flume-calibrated opercular and tail accelerometers recorded mesocosm behavior, which was classified by spectral t-SNE and k-medoids clustering. Acoustic detections were paired with Delft3D-FLOW velocities.
- Wild red drum line-caught in the Matanzas River for flow-tank and mesocosm experiments (40.1 ± 4.2 cm) · Sciaenops ocellatus (red drum) · n = 18
- Respirometry and kinematics subset swum at 22, 61 and 100 cm/s under free-stream, 2D and 3D bluff-body treatments (37.7 ± 1.3 cm) · Sciaenops ocellatus (red drum) · n = 5
- Dual-accelerometer recordings over 5 days in a 6.5 m diameter outdoor mesocosm with habitat structures and native prey (41.3 ± 3.1 cm) · Sciaenops ocellatus (red drum) · n = 13 (speed histograms shown for 5)
- Acoustically tagged over-slot fish (75.3 ± 8.5 cm) from the St. Augustine inlet, detected by 21 GTMNERR receivers plus partner networks, 2019 to 2021 · Sciaenops ocellatus (red drum) · n = 5
- Delft3D-FLOW simulation of the Guana Tolomato Matanzas estuary, 1 January 2019 to 31 December 2021, with grid down to 15 m × 20 m in the estuary
- Model inputs: NOAA bathymetry, USGS/USACE/NOAA LiDAR, Florida Natural Areas Inventory vegetation map, NOAA tide stations, USGS Pellicer Creek discharge and weather
- Methods & tools
- Intermittent-flow respirometry (Autoresp v.1.6, Loligo Systems), DeepLabCut markerless pose estimation, NumPy and Matplotlib (Python), Axy5 S tri-axial accelerometer loggers (Technosmart), Phantom Miro 340 high-speed camera, t-SNE dimensionality reduction, k-medoids clustering, PCA and Fourier spectral analysis in custom MATLAB scripts, V9 acoustic transmitters (Vemco) and VR2W receivers (Innovasea), VUE telemetry software and R scripts, Delft3D-FLOW hydrodynamic model, Two-way ANOVA with Tukey post-hoc tests
- Reagents & environment
- Autoresp v.1.6 (Loligo Systems); custom MATLAB scripts (Mathworks R2023a); Python with DeepLabCut, NumPy and Matplotlib; VUE software; R scripts; Delft3D-FLOW.
- Reproducibility gaps
- The text has no code or data availability statement. Custom MATLAB and R scripts, accelerometry records, telemetry detections and Delft3D-FLOW outputs are not linked to any repository.
Limitations (6)
- Receivers detect fish within roughly 300 m, so a detection shows presence in a wide area, not the fish's position across the channel flow profile. Fish may use low-velocity refugia, so the flow they experienced may differ from velocities extracted at receivers. · stated by the authors
- Receivers were often fixed to nearshore docks next to oyster flats and mangroves, so detections sample shoreline sites. The authors suggest testing sensitivity to receiver placement and detection radius and adding finer-scale positioning to resolve paths. · stated by the authors
- Respirometry speeds (up to 100 cm/s) were higher than the flows in which wild fish were usually detected. The authors propose future trials at lower flows typical of the field. · stated by the authors
- Accelerometers lack the millisecond and millimetre resolution of laboratory biomechanics. Opercular signals can rise with stress as well as exertion, and the mesocosm cannot fully reproduce natural complexity. · stated by the authors
- Samples are small (5 fish in respirometry, 5 tagged in the field). Respirometry fish (37.7 ± 1.3 cm) were also much smaller than the tagged over-slot fish (75.3 ± 8.5 cm), which limits how far lab energetics transfer to wild adults. · noted in this summary
- The text is inconsistent: the abstract cites 36 receivers but the methods list 21, and detections are summarized as <30 cm/s although results place ~75% between 0.15 and 0.40 m/s. No formal statistical test compares detections with available flows. · noted in this summary
Who should read:Read this if you study fish locomotor energetics, biologging or estuarine movement ecology and want an example linking flume respirometry and calibrated dual accelerometry to telemetry and hydrodynamic models.
Reproducibility ledger
Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.
Sign in to file a reproducibility report.
Related papers
- Oxygen consumption of drift-feeding rainbow trout: the energetic tradeoff between locomotion and feeding in flow
J. L. Johansen, O. Akanyeti, J. C. Liao·bioRxiv·Dec 26, 2019
- Collective movement of schooling fish reduces locomotor cost in turbulence
- Behavior of downstream swimming brown trout in accelerating and high flow velocity - Movement matters
Falko Wagner, Ianina Kopecki, … Stefan Hoerner·PLoS ONEPeer-reviewed·Sep 30, 2026
- In-line swimming dynamics revealed by fish interacting with a robotic mechanism
R. Thandiackal, G. Lauder·bioRxiv·Jul 14, 2022
- The Effects of Flow Speeds on Smooth Pursuit Tracking and Active Sensing Movements of Weakly Electric Fish
E. Y. AYDIN, B. Unlutabak, I. Uyanik·bioRxiv·Nov 2, 2023
- Fluctuating hydrodynamic forces increase fish locomotor variability in turbulence
Michael A. Calicchia, Divya Ramesh, … Rui Ni·bioRxiv·Oct 19, 2026