Academia.eduAcademia.edu

Harvesting-induced evolution of collective behavior in a fish

Profile image of Pascal KlamserPascal Klamser

2019

Abstract

Many fisheries around the globe preferentially capture large individuals with implications for the evolution of exploited populations. Fisheries-induced evolution may alter collective behavioral phenotypes through individual-level adaptations that affect boldness, swimming speed and tendency to follow social vs. environmental cues. Studying the behavioural mechanisms that give rise to possible changes in shoaling and other collective outputs is challenging in the wild, but first insights into whether intensive and size-selective harvesting could alter collective phenotypes and shoaling can be gathered through experiment of size-selective harvesting conducted in the laboratory. We present a multi-generation harvest selection experiment with zebrafish (Danio rerio) as a model species and demonstrate that large size-selective harvesting typical of global fisheries decreases risk-taking behavior of individuals, and surprisingly also decreases shoal cohesion. This counter-intuitive effec...

Figures (18)
Fig. 1. Group level results for time spent feeding at the surface of the water during a diving test (A), mean group nearest-neighbor distance (expressed in body length; B) and inter-individual distance (expressed in body length; inset B). Individual level results for burst rate (C), speed (expressed in body length per second; inset C) and body size (D). The selection lines are indicated by different color (red=LH: large- harvested; black=RH: random-harvested; blue=SH: small-harvested). Bold colors (C , D) refers to the entire data set (N = 48), while shadowed colors refers to the subsample of size-matched individuals (N = LH: 24; RH: 36; SH: 40). Points rep- resent the median and vertical lines represent the first (25th percentile) and third (75th percentile) quartiles.
Fig. 1. Group level results for time spent feeding at the surface of the water during a diving test (A), mean group nearest-neighbor distance (expressed in body length; B) and inter-individual distance (expressed in body length; inset B). Individual level results for burst rate (C), speed (expressed in body length per second; inset C) and body size (D). The selection lines are indicated by different color (red=LH: large- harvested; black=RH: random-harvested; blue=SH: small-harvested). Bold colors (C , D) refers to the entire data set (N = 48), while shadowed colors refers to the subsample of size-matched individuals (N = LH: 24; RH: 36; SH: 40). Points rep- resent the median and vertical lines represent the first (25th percentile) and third (75th percentile) quartiles.
Fig. 2. (A) Schematic model description showing that at each burst an individual can follow with probability P;.- social cues (i.e. behaves according to a three-zone model: r=repulsion-, o=orientation-, a=attraction-zone) or with probability 1 — Psoc environmental cues. (B) Optimization results for fitting the burst-coast model to ex- perimental data related to the probability to follow social cues P;,-. The selection lines differed, next to P;.< in their specific burst rate -y that was estimated from ex- perimental data. The selection lines are indicated by different color (red=LH: large- harvested; black=RH: random-harvested; blue=SH: small-harvested). No error bars exist for the optimization. See methods for the robustness of the optimization re- sults.
Fig. 2. (A) Schematic model description showing that at each burst an individual can follow with probability P;.- social cues (i.e. behaves according to a three-zone model: r=repulsion-, o=orientation-, a=attraction-zone) or with probability 1 — Psoc environmental cues. (B) Optimization results for fitting the burst-coast model to ex- perimental data related to the probability to follow social cues P;,-. The selection lines differed, next to P;.< in their specific burst rate -y that was estimated from ex- perimental data. The selection lines are indicated by different color (red=LH: large- harvested; black=RH: random-harvested; blue=SH: small-harvested). No error bars exist for the optimization. See methods for the robustness of the optimization re- sults.
Fig. 3. Results of the natural predator scenario (A, C), the single fishing agent (A, D) and the multiple fishing agent scenario (B, E). N = 30 shoaling fish were simulated in a box of size L = 100 cm with periodic boundary conditions and swimming with a speed of (v) + 15 cm/s. The predator moved with une = [15, 22.5, ,30, 37.5] em/s while the fishing agents with unc: = [0, 7.5, 15, 22.5] cm/s indicated by solid, dashed, dash-dotted and dotted line, respectively. Exploitation rates were reduced by the mean and divided by the standard deviation of the random-harvested line to standardize their values. Snapshots of a typical simulation are shown for the predator- and single-fishing-agent scenario (A) and for the multiple-fishing-agents scenario (B); the shoaling fish are represented in black and the predator and fishing agents in red. The relative mean exploitation rate (~Yexp) of the three selection lines (red=LH: large-harvested; black=RH: random-harvested; blue=SH: small-harvested) was shown for each o' the simulated scenarios: predator (C), single fishing agent (D), and multiple fishing agents (E). The error bars indicated the standard deviation relative to the control line (RH) using a sample of 40 simulations.
Fig. 3. Results of the natural predator scenario (A, C), the single fishing agent (A, D) and the multiple fishing agent scenario (B, E). N = 30 shoaling fish were simulated in a box of size L = 100 cm with periodic boundary conditions and swimming with a speed of (v) + 15 cm/s. The predator moved with une = [15, 22.5, ,30, 37.5] em/s while the fishing agents with unc: = [0, 7.5, 15, 22.5] cm/s indicated by solid, dashed, dash-dotted and dotted line, respectively. Exploitation rates were reduced by the mean and divided by the standard deviation of the random-harvested line to standardize their values. Snapshots of a typical simulation are shown for the predator- and single-fishing-agent scenario (A) and for the multiple-fishing-agents scenario (B); the shoaling fish are represented in black and the predator and fishing agents in red. The relative mean exploitation rate (~Yexp) of the three selection lines (red=LH: large-harvested; black=RH: random-harvested; blue=SH: small-harvested) was shown for each o' the simulated scenarios: predator (C), single fishing agent (D), and multiple fishing agents (E). The error bars indicated the standard deviation relative to the control line (RH) using a sample of 40 simulations.
Fig. S.1. Simulations with varying social update probability (P;o-) for the different harvesting lines, defined by their specific burst rate +. Solid circles represent the Psoc for each selection line used to compute the smallest error Eq. (S.12). The triangles represent the best fit to experimental data for (A) the nearest neighbor distance and (B) the speed v. The resolution for Psoc was 0.025. The burst rates + for the different selection lines were yz, 47 = 3.45, ypy = 3.51 and yyy = 2.91. Each average value was computed from 40 samples. The selection lines are indicated by the following abbreviations (LH: large-harvested; RH: random-harvested; SH: small-harvested).
Fig. S.1. Simulations with varying social update probability (P;o-) for the different harvesting lines, defined by their specific burst rate +. Solid circles represent the Psoc for each selection line used to compute the smallest error Eq. (S.12). The triangles represent the best fit to experimental data for (A) the nearest neighbor distance and (B) the speed v. The resolution for Psoc was 0.025. The burst rates + for the different selection lines were yz, 47 = 3.45, ypy = 3.51 and yyy = 2.91. Each average value was computed from 40 samples. The selection lines are indicated by the following abbreviations (LH: large-harvested; RH: random-harvested; SH: small-harvested).
Fig. S.2. Angular change dy/dt depending on the velocity v. The blue dots (A, B) represent the mean values while orange dots (B) show the maximum observed value for the binned data. The blue line was f(v) = 1/v and the orange line was f(v) = 5/v and therefore the data agreed well with Eq. (S.2c).
Fig. S.2. Angular change dy/dt depending on the velocity v. The blue dots (A, B) represent the mean values while orange dots (B) show the maximum observed value for the binned data. The blue line was f(v) = 1/v and the orange line was f(v) = 5/v and therefore the data agreed well with Eq. (S.2c).
Table S.2. Upper parts shows fixed parameters estimated in section (F). The lower part shows optimized parameters by the covariance matrix adaptation evolution strategy (CMA-ES) minimizing Eq. (S.12). The selection lines are indicated by the following abbreviations (LH: large-harvested; RH: random-harvested; SH: small-harvested). Table S.1. Upper part are parameters of the burst-coast model and their estimates from the experimental data. The lower part are non model parameters which are used to fit the model. The selection lines are indicated by the following abbreviations (LH: large-harvested RH: random-harvested; SH: small-harvested).
Table S.2. Upper parts shows fixed parameters estimated in section (F). The lower part shows optimized parameters by the covariance matrix adaptation evolution strategy (CMA-ES) minimizing Eq. (S.12). The selection lines are indicated by the following abbreviations (LH: large-harvested; RH: random-harvested; SH: small-harvested). Table S.1. Upper part are parameters of the burst-coast model and their estimates from the experimental data. The lower part are non model parameters which are used to fit the model. The selection lines are indicated by the following abbreviations (LH: large-harvested RH: random-harvested; SH: small-harvested).
Fig. S.3. A: Burst-duration (t,) and B: burst-acceleration (F’) estimated from experimental data of the different selection lines that are indicated by the following abbreviations (LH: large-harvested; RH: random-harvested; SH: small-harvested).
Fig. S.3. A: Burst-duration (t,) and B: burst-acceleration (F’) estimated from experimental data of the different selection lines that are indicated by the following abbreviations (LH: large-harvested; RH: random-harvested; SH: small-harvested).
Fig. S.4. Estimation of friction coefficient from experimental data. The red lines represent linear fits of the mean and minimum acceler- ation for a specific velocity which results in a slope of —beta = 0.082 for the mean and —beta = 0.88 for the minimum acceleration.
Fig. S.4. Estimation of friction coefficient from experimental data. The red lines represent linear fits of the mean and minimum acceler- ation for a specific velocity which results in a slope of —beta = 0.082 for the mean and —beta = 0.88 for the minimum acceleration.
Fig. S.5. Non-standardized results of the natural-predator-like scenario (A, C), the angling-like scenario (A, D) and the trawling-like scenario (B, E) are shown. Here N = 30 fish swim in a box of size L = 100cm with periodic boundary conditions. Compared to the shoaling fish, moving with a speed of (v) ~ 15cm/s, the predator moves with unet = [15, 22.5, ,30, 37.5]cm/s while the fishing agents with une = [0, 7.5, 15, 22.5]em/s indicated by solid, dashed, dash-dotted and dotted line respectively. A, B show snapshots of a typical simulation; C, D, E the mean exploitation rate of the predator and fishing agents (exp) and F, G the mean distance to the closest fishing agent (dif). In contrast to the figure in the main-text no relative measures are shown. The selection lines are indicated by different color (red=LH: large-harvested; black=RH: random-harvested; blue=SH: small-harvested).
Fig. S.5. Non-standardized results of the natural-predator-like scenario (A, C), the angling-like scenario (A, D) and the trawling-like scenario (B, E) are shown. Here N = 30 fish swim in a box of size L = 100cm with periodic boundary conditions. Compared to the shoaling fish, moving with a speed of (v) ~ 15cm/s, the predator moves with unet = [15, 22.5, ,30, 37.5]cm/s while the fishing agents with une = [0, 7.5, 15, 22.5]em/s indicated by solid, dashed, dash-dotted and dotted line respectively. A, B show snapshots of a typical simulation; C, D, E the mean exploitation rate of the predator and fishing agents (exp) and F, G the mean distance to the closest fishing agent (dif). In contrast to the figure in the main-text no relative measures are shown. The selection lines are indicated by different color (red=LH: large-harvested; black=RH: random-harvested; blue=SH: small-harvested).
Fig. S.6. Results from five-dimensional parameter-scan used to validate the results from the Covariance Matrix Adaptation Evolutionary Strategy (CMA-ES) of the main-text. Error according to Eq. (S.12) for the one percent of the tested parameter configurations with the smallest errors. The result of the CMA-ES was that the probability to do a social burst Psoc differed in between the selection lines as Psoc(LH) < Psoc( RH) < Psoc(SH). A parameter configuration was consistent if these inequalities hold, partly consistent if one of the inequalities was an equality, unclear if both inequalities were equalities and the configuration was inconsistent if one or both inequalities were opposite. The consistent, partly consistent, unclear and inconsistent cases are color coded as green, blue, black and red respectively.
Fig. S.6. Results from five-dimensional parameter-scan used to validate the results from the Covariance Matrix Adaptation Evolutionary Strategy (CMA-ES) of the main-text. Error according to Eq. (S.12) for the one percent of the tested parameter configurations with the smallest errors. The result of the CMA-ES was that the probability to do a social burst Psoc differed in between the selection lines as Psoc(LH) < Psoc( RH) < Psoc(SH). A parameter configuration was consistent if these inequalities hold, partly consistent if one of the inequalities was an equality, unclear if both inequalities were equalities and the configuration was inconsistent if one or both inequalities were opposite. The consistent, partly consistent, unclear and inconsistent cases are color coded as green, blue, black and red respectively.
Fig. S.7. A: Part of the trajectory of a swimming zebrafish generated from raw-data of the ID-Tracker. The black dots indicate the burst events computed from the smoothed trajectory. B: Acceleration F’ along heading direction é;. The vertical black lines indicate burst events which are computed by the first positive acceleration of the smoothed data. C: Acceleration F’ perpendicular to heading direction Ey, i.e. direction changes. The burst events were mostly accompanied by changes in the direction. bioRxiv preprint doi: https://doi.org/10.1101/809442. this version posted October 17, 2019. The copyright holder for this preprint (which was nc certified by peer review) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
Fig. S.7. A: Part of the trajectory of a swimming zebrafish generated from raw-data of the ID-Tracker. The black dots indicate the burst events computed from the smoothed trajectory. B: Acceleration F’ along heading direction é;. The vertical black lines indicate burst events which are computed by the first positive acceleration of the smoothed data. C: Acceleration F’ perpendicular to heading direction Ey, i.e. direction changes. The burst events were mostly accompanied by changes in the direction. bioRxiv preprint doi: https://doi.org/10.1101/809442. this version posted October 17, 2019. The copyright holder for this preprint (which was nc certified by peer review) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
Fig. S.8. Number of burst events of a randomly selected trajectory spanning 2000 frames. The horizontal black line marks the number of manually counted burst events. Blue dots mark automatic counts of burst events on a smoothed trajectory using a moving window with a Gaussian kernel of varying standard deviation osmoo. The vertical black line indicates the standard deviation osmoo = 1.13fms where the automatic burst detection equals the manually counted one.
Fig. S.8. Number of burst events of a randomly selected trajectory spanning 2000 frames. The horizontal black line marks the number of manually counted burst events. Blue dots mark automatic counts of burst events on a smoothed trajectory using a moving window with a Gaussian kernel of varying standard deviation osmoo. The vertical black line indicates the standard deviation osmoo = 1.13fms where the automatic burst detection equals the manually counted one.
Fig. S.9. Results of the alternative predator scenario described in section 2 without confusion effect. The other parameters were identical to the predator scenario of the main text. Here N = 30 fish swim in a box of size L = 100cm with periodic boundary conditions Compared to the shoaling fish, moving with a speed of (v) = 15cm/s, the predator moves with une = [15, 22.5, ,30, 37.5]cm/s indicated by solid, dashed, dash-dotted and dotted line respectively. (A) Relative mean exploitation rate of the predator with respec to the control line (random-harvested line). The observed values are made relative to the control line by subtracting their mean anc dividing by their standard deviation. The error-bars indicate the standard deviation relative to the control line. (B) the mean exploitatior rate of the predator (exp). The selection lines are indicated by different color (red=LH: large-harvested; black=RH: random-harvested blue=SH: small-harvested).
Fig. S.9. Results of the alternative predator scenario described in section 2 without confusion effect. The other parameters were identical to the predator scenario of the main text. Here N = 30 fish swim in a box of size L = 100cm with periodic boundary conditions Compared to the shoaling fish, moving with a speed of (v) = 15cm/s, the predator moves with une = [15, 22.5, ,30, 37.5]cm/s indicated by solid, dashed, dash-dotted and dotted line respectively. (A) Relative mean exploitation rate of the predator with respec to the control line (random-harvested line). The observed values are made relative to the control line by subtracting their mean anc dividing by their standard deviation. The error-bars indicate the standard deviation relative to the control line. (B) the mean exploitatior rate of the predator (exp). The selection lines are indicated by different color (red=LH: large-harvested; black=RH: random-harvested blue=SH: small-harvested).
Fig. S.10. Results of the alternative predator scenario described in Sect.(2) with confusion effect. The other parameters are identical to the predator-like scenario of the main text. Here NV = 30 fish swim in a box of size L = 100cm with periodic boundary conditions. Compared to the shaoling fish, moving with a speed of (v) + 15cm/s, the predator moves with unet = [15, 22.5, ,30, 37.5]em/s indicated by solid, dashed, dash-dotted and dotted line respectively. (A) Relative mean exploitation rate of the predator with respect to the control line (random-harvested line). The observed values are made relative to the control line by subtracting their mean and dividing by their standard deviation. The error-bars indicate the standard deviation relative to the control line. (B) the mean exploitation rate of the predator (exp). The selection lines are indicated by different color (red=LH: large-harvested; black=RH: random-harvested; blue=SH: small-harvested). bioRxiv preprint doi: https://doi.org/10.1101/809442. this version posted October 17, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
Fig. S.10. Results of the alternative predator scenario described in Sect.(2) with confusion effect. The other parameters are identical to the predator-like scenario of the main text. Here NV = 30 fish swim in a box of size L = 100cm with periodic boundary conditions. Compared to the shaoling fish, moving with a speed of (v) + 15cm/s, the predator moves with unet = [15, 22.5, ,30, 37.5]em/s indicated by solid, dashed, dash-dotted and dotted line respectively. (A) Relative mean exploitation rate of the predator with respect to the control line (random-harvested line). The observed values are made relative to the control line by subtracting their mean and dividing by their standard deviation. The error-bars indicate the standard deviation relative to the control line. (B) the mean exploitation rate of the predator (exp). The selection lines are indicated by different color (red=LH: large-harvested; black=RH: random-harvested; blue=SH: small-harvested). bioRxiv preprint doi: https://doi.org/10.1101/809442. this version posted October 17, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
Fig. S.11. Results of the natural predator (A), the angling (B, D, F) and the trawling scenarios (C, E, G). Here N = 30 fish swim ina box of size L = 100cm with periodic boundary conditions. Compared to the shoaling fish, moving with a speed of (v) = 15cm/s, the predator moves with une = [15, 22.5, ,30, 37.5]em/s while the fishing agents with une: = [0, 7.5, 15, 22.5]em/s indicated by solid, dashed, dash-dotted and dotted line respectively. B, C show snapshots of a typical simulation; A, D, E the relative mean exploitation rate of the predator and fishing agents (exp) and F, G the relative mean distance to the closest fishing agent (dif). The observed values are made relative to the control line by subtracting their mean and dividing by their standard deviation. The error-bars indicate the standard deviation relative to the control (random-harvested line) using a sample of 40 simulations. The selection lines are indicated by different color (red=LH: large-harvested; black=RH: random-harvested; blue=SH: small-harvested). bioRxiv preprint doi: https://doi.org/10.1101/809442. this version posted October 17, 2019. The copyright holder for this preprint (which was no certified by peer review) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
Fig. S.11. Results of the natural predator (A), the angling (B, D, F) and the trawling scenarios (C, E, G). Here N = 30 fish swim ina box of size L = 100cm with periodic boundary conditions. Compared to the shoaling fish, moving with a speed of (v) = 15cm/s, the predator moves with une = [15, 22.5, ,30, 37.5]em/s while the fishing agents with une: = [0, 7.5, 15, 22.5]em/s indicated by solid, dashed, dash-dotted and dotted line respectively. B, C show snapshots of a typical simulation; A, D, E the relative mean exploitation rate of the predator and fishing agents (exp) and F, G the relative mean distance to the closest fishing agent (dif). The observed values are made relative to the control line by subtracting their mean and dividing by their standard deviation. The error-bars indicate the standard deviation relative to the control (random-harvested line) using a sample of 40 simulations. The selection lines are indicated by different color (red=LH: large-harvested; black=RH: random-harvested; blue=SH: small-harvested). bioRxiv preprint doi: https://doi.org/10.1101/809442. this version posted October 17, 2019. The copyright holder for this preprint (which was no certified by peer review) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
Fig. S.12. Results of the predator scenario (A), the angling scenario (B, D, F) and the trawling scenario (C, E, G) are shown. Here N = 30 fish swim in a box of size L = 100cm with periodic boundary conditions. Compared to the shoaling fish, moving with a speed of (v) & 15cm/s, the predator moves with unet = [15, 22.5, ,30, 37.5]cem/s while the fishing agents with unez = [0, 7.5, 15, 22.5)cm/s indicated by solid, dashed, dash-dotted and dotted line respectively. B, C show snapshots of a typical simulation; A, D, E the mean exploitation rate of the predator and fishing agents (exp) and F, G the mean distance to the closest fishing agent (dif). In contrast to the figure in the main-text no relative values are shown. The selection lines are indicated by different color (red=LH: large-harvested; black=RH: random-harvested; blue=SH: small-harvested). bioRxiv preprint doi: https://doi.org/10.1101/809442. this version posted October 17, 2019. The copyright holder for this preprint (which was nc certified by peer review) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.
Fig. S.12. Results of the predator scenario (A), the angling scenario (B, D, F) and the trawling scenario (C, E, G) are shown. Here N = 30 fish swim in a box of size L = 100cm with periodic boundary conditions. Compared to the shoaling fish, moving with a speed of (v) & 15cm/s, the predator moves with unet = [15, 22.5, ,30, 37.5]cem/s while the fishing agents with unez = [0, 7.5, 15, 22.5)cm/s indicated by solid, dashed, dash-dotted and dotted line respectively. B, C show snapshots of a typical simulation; A, D, E the mean exploitation rate of the predator and fishing agents (exp) and F, G the mean distance to the closest fishing agent (dif). In contrast to the figure in the main-text no relative values are shown. The selection lines are indicated by different color (red=LH: large-harvested; black=RH: random-harvested; blue=SH: small-harvested). bioRxiv preprint doi: https://doi.org/10.1101/809442. this version posted October 17, 2019. The copyright holder for this preprint (which was nc certified by peer review) is the author/funder. It is made available under a CC-BY-NC-ND 4.0 International license.

Related papers

Evolutionary implications of size-selective mortality on the ontogenetic development of shoal cohesion: a neurochemical approach using a zebrafish, Danio rerio, harvest selection experiment
Svante Winberg

Behavioral Ecology and Sociobiology, 2022

Size-selective mortality may evolutionarily alter life-history as well as individual behavioral and physiological traits. Moreover, size-selective mortality can affect group behavioral traits, such as shoaling and collective properties (e.g., shoal cohesion), which are relevant for finding food and reducing risk of predation. Here, we present experimental evidence using selection lines of zebrafish (Danio rerio) that were exposed to positive (large-harvested), negative (small-harvested), and random (control) size-selective mortality for five generations, followed by eight generations during which harvesting was halted to remove maternal effects and to study evolutionarily fixed outcomes. We investigated changes in shoal cohesion and turnover in monoamines in zebrafish through ontogeny. To that end, we repeatedly measured inter-individual distance in groups of eight fish and the turnovers of dopamine and serotonin in brains of fish from juvenile to the adult stage at 40-day intervals. We, firstly, found that shoal cohesion was overall consistent through ontogeny at group levels suggesting the presence of collective personality. Secondly, we found a decrease in shoal cohesion through ontogeny in the small-harvested and control lines, while the large-harvested line did not show any ontogenetic change. Thirdly, the selection lines did not differ among each other in shoal cohesion at any ontogenetic stage. Fourthly, dopamine turnover increased through ontogeny in a similar way for all lines while the serotonin turnover decreased in the large-harvested and control lines, but not in the small-harvested line. The large-harvested line also had higher serotonin turnover than controls at specific time periods. In conclusion, intensive size-selective mortality left an evolutionary legacy of asymmetric selection responses in the ontogeny of shoal cohesion and the underlying physiological mechanisms in experimentally harvested zebrafish in the laboratory. Significant statement The evolution of animal behavior can be affected by human activities both at behavioral and physiological levels, but causal evidence is scarce and mostly focusing on single life-stages. We studied whether and to what extent size-selective harvesting, a common selection pattern in fisheries, can be an evolutionary driver of the development of shoal cohesion during ontogeny. We used a multi-generation experiment with zebrafish to study cause-and-effects of opposing size-selection patterns. We quantified shoal cohesion, and serotonin and dopamine turnover in the brain. We found that shoal cohesion emerged as a collective personality trait and that behavioral and physiological responses were asymmetrical with respect to the opposing selection patterns.

View PDFchevron_right
Unexpected harvesting-induced evolution of collective behavior in a fish
Pascal Klamser

edoc Publication server (Humboldt University of Berlin), 2020

First, I would like to thank Pawel Romanczuk, who created an unique working environment, whose advice has always helped me and who has supported me and inspired me in every possible way. Also for bringing the scientific world to us so vividly through a variety of conferences and collaborations. I owe many unique encounters, both human and scientific, to the CoCCoN network he initiated. Especially in these times of isolation it is a pleasure to think back. I would like to thank the people who shaped my daily work at the lab. It was a great pleasure for me to work with Winnie Poel, explore the research field of collective behavior and pursued the path to dissertation together. I owe Yinong Zhao many enthusiastic conversations about related scientific fields. Haider Klenz has humorously lit up the office and was a highly valued sitting neighbor. I was fortunate to explore the role of a supervisor with the talented interns Conor McBride, Alexandra (Olenka) Jain and Noam Miller. I would like to thank

View PDFchevron_right
Size-selective mortality may trigger maladaptive changes of shoaling behaviour in a fish
Pascal Klamser

bioRxiv, 2020

40 cally capture fisheries, profoundly alter the natural mortality-41 based fitness landscape by increasing adult mortality to levels 42 hardly ever experienced in the evolutionary history of most 43 fish populations (3). Size-selective mortality imposed by 44 fishing can foster adaptive responses in terms of life-history 45 as well as physiological and behavioral traits (4-7). Fishing 46 can also affect shoaling tendency and collective behavior of 47 fish (8, 9), with potential consequences for population dy-48 namics and food-web functioning (10). However, the mech-49 anisms through which size-selective harvesting affects shoal-50 ing behavior are largely unexplored. 51 Shoaling behavior has strong adaptive value in many fish 52 species, reducing predation risk, increasing foraging effi-53 ciency, and reducing energetic costs (11-13). For example, 54 a classic expectation is that fish increase shoal cohesion in 55 the presence of predation risk because more cohesive, denser, 56 and coordinated groups can more effectively confound preda-57 tors and consequently reduce their predation efficiency (14). 58 Shoaling behavior is complex and can be simultaneously in-59 fluenced by several variables including predation risk (15), 60 food availability (16), and light levels (17). In particular, the 61 trade-off between vigilance (i.e. the frequency of environ-62 mental scans) and time spent feeding is a crucial aspect in 63 determining shoaling behavior in fishes (15, 18). However, 64 because coordinated movement demands attention to social 65 cues, there is also a trade-off between shoaling and vigilance. 66 For example, in three-spined sticklebacks, Gasterosteus ac-67 uleatus, more aligned individuals respond slower to an ex-68 ternal cue (19), and in herring, Clupea harengus, solitary 69 individuals respond faster to an external cue than shoaling 70 individuals (20). Therefore, an increase of vigilance could 71 weaken social coordination and hence shoal cohesion. Vig-72 ilance can also be affected by group size (21-23) and so-73 cial behavior (e.g. territorial aggression; 24). Shoaling be-74

View PDFchevron_right
Quantification of shoaling behaviour in zebrafish (Danio rerio)
Robert Gerlai

Behavioural Brain Research, 2007

Zebrafish has been a favourite of developmental biologists and numerous genetic tools have been developed for this species. In recent years, zebrafish has become an increasingly popular subject of neuroscientists and behavioural scientists. One of the typical characteristics of zebrafish is shoaling, individuals forming a tight group in which fish swim together. The biological mechanisms of social behaviours are complex and not well understood in vertebrates, and zebrafish, due to its highly social nature and the genetic tools developed for it, may represent an excellent animal model with which these mechanisms may be studied. Improvement of behavioural quantification methods would facilitate research in this area. We describe a custom software application that allows the precise quantification of several parameters of group cohesion in zebrafish. We also present three experimental examples to illuminate the use of our methodology, and show how group cohesion changes in response to manipulations of the environment.

View PDFchevron_right
Inter and intra-population variation in shoaling and boldness in the zebrafish (Danio rerio)
Dominic Wright

Journal of Fish Biology, 2003

Population differences in anti-predator behaviour have been demonstrated in several species, although less is known about the genetic basis of these traits. To determine the extent of genetic differences in boldness (defined as exploration of a novel object) and shoaling within and between zebrafish (Danio rerio) populations, and to examine the genetic basis of shoaling behaviour in general, we carried out a study that involved laboratoryraised fish derived from four wild-caught populations. Controlling for differences in rearing environment, significant inter-population differences were found in boldness but not shoaling. A larger shoaling experiment was also performed using one of the populations as the basis of a North Carolina type II breeding design (174 fish in total) to estimate heritability of shoaling tendency. A narrow-sense heritability estimate of 0.40 was obtained, with no apparent dominance effects.

View PDFchevron_right

Related topics

  • Collective Behavior
  • Biology

    • Related papers

    Collective departures and leadership in zebrafish
    Axel Séguret

    PLOS ONE, 2019

    In social animals, morphological and behavioural traits may give to some individuals a stronger influence on the collective decisions, even in groups assumed to be leaderless such as fish shoals. Here, we studied and characterized the leadership in collective movements of shoals of zebrafish Danio rerio by observing groups of 2, 3, 5, 7 and 10 zebrafish swimming in a two resting sites arena during one hour. We quantified the number of collective departures initiated by each fish and the number of attempts that they made. To do so, we developed an automated pipeline that analysed the individual trajectories generated by the tracking software. For all shoal sizes, the leadership was distributed among several individuals. However, it was equally shared among all the fish in some shoals while other groups showed a more asymmetrical sharing of the initiation of collective departures. To quantify this distribution, we computed the entropy associated with the time series of the identity of all initiators for each experiment and confirmed the presence of a continuum between a homogeneous and a heterogeneous distribution of the leadership. While some fish led more departures than others, an individual analysis showed that all fish had actually the same success rate to lead the shoal out of a resting site after an attempt. Thus, some individuals monopolized the leadership by attempting more often than others to exit a resting site. Finally, we highlight that the intra-group ranking of a fish for the initiative is correlated to its intra-group ranking for the average speed with mobile individuals more prone to lead the shoal. These results demonstrate that the collective behaviour of a shoal can be mainly driven by a subset of individuals even in the absence of higher influence of a fish on its congeneers.

    View PDFchevron_right
    Context-dependent group size choice in fish
    Daniel Hoare

    Animal Behaviour, 2004

    View PDFchevron_right
    What drives mixed-species shoaling among wild zebrafish? Role of predators, food access, abundance of conspecifics, and kin familiarity
    Anuradha Bhat

    Mixed-species groups commonly occur across a wide range of faunal communities and are known to provide several benefits to members. While zebrafish have often been observed to form mixed-species shoals with coexisting species, the factors determining their occurrence are not yet understood. Using laboratory-based experiments, we decipher the main ecological drivers of mixed-species shoaling in these tropical fish communities. Shoals comprising zebrafish (Danio rerio), flying barbs (Esomus danricus), and whitespots (Aplocheilus panchax) were collected from a stagnant canal at Haringhata (West Bengal, India). Experiments to assess foraging efficiency were conducted where single or mixed-species shoals (comprising 5 individuals) were given low or high amounts of food. Shoal choice experiments were also conducted to assess the preferences of test subjects (zebrafish individuals) for forming associations based on shoal composition and familiarity. Results from experiments on feeding effi...

    View PDFchevron_right
    Thirty-five experimental fisheries reveal the mechanisms of selection
    Anna Sturrock

    Fisheries have been described as large-scale evolutionary experiments; yet such “experiments” tend to be poorly replicated and therefore lack the predictive power essential for designing appropriate management strategies to minimize the effects of fisheries-induced selection. Large-scale removal of non-native trout from 35 montane lakes in California provided repeated experimental fisheries that allowed us to explore how environmental parameters affect the three potential contributors to overall selection: the fitness-trait correlation, trait variability, and fitness variability. Our results demonstrate that fishing rapidly altered the size structure of harvested populations, and that the magnitude of change was primarily driven by the fitness-trait correlation (net selectivity). Fishing-induced selection was repeatable overall but was also influenced by environmental (lake size and quality) and demographic (size structure) parameters. Decomposing fishing-induced selection into its ...

    View PDFchevron_right
    The influence of sex and phenotype on shoaling decisions in zebrafish
    Jennifer Snekser

    Fish typically choose shoalmates with similar phenotypic characteristics to themselves, thus creating shoals for which predators have difficulty identifying and attacking one specific individual. And while shoaling should provide similar anti-predator benefits to both males and females, the two sexes do not always make the same shoaling decisions. Here we explore the effect of phenotype on sexspecific shoaling in three varieties of zebrafish (Danio rerio) and the closely related pearl danio (Danio albolineatus). We hypothesized that males and females of each type of zebrafish (wildtype, golden mutants and leopard mutants), as well as male and female pearl danios, would choose to shoal rather than be alone and, when given a choice of shoalmates, would shoal with fish of their own phenotype rather than dissimilar fish. As expected, our results show that most fish preferred to shoal rather than be alone. However, while both sexes of wildtype zebrafish responded identically to shoaling decisions, male and female mutant zebrafish and pearl danio fish differed in their response to such choices. When given a choice of shoalmates, wildtype zebrafish of both sexes showed no discrimination between different D. rerio strains, although they did choose to shoal with wildtype conspecifics rather than pearl danios. The shoalmate preferences of the mutant zebrafish revealed that males showed no discrimination between shoals of their own variety and wildtype shoals, while mutant females preferred shoals of their own strain. Similarly, male pearl danios showed no discrimination between shoals of their own species and shoals of wildtype zebrafish, while pearl danio females preferred their own species. These results demonstrate the complex influence of sex and phenotype on shoaling behavior.

    View PDFchevron_right
    Ecological change alters the evolutionary response to harvest in a freshwater fish
    Michael Fox

    Ecological Applications, 2018

    Harvesting can induce rapid evolution in animal populations, yet the role of ecological change in buffering or enhancing that response is poorly understood. Here, we developed an ecogenetic model to examine how ecological changes brought about by two notorious invasive species, zebra and quagga mussels, influence harvest-induced evolution and resilience in a freshwater fish. Our study focused on lake whitefish (Coregonus clupeaformis) in the Laurentian Great Lakes, where the species supports valuable commercial and subsistence fisheries, and where the invasion of dreissenid (zebra and quagga) mussels caused drastic shifts in ecosystem productivity. Using our model system, we predicted faster rates of evolution of maturation reaction norms in lake whitefish under pre-invasion ecosystem conditions when growth and recruitment of young to the population were high. Slower growth rates that occurred under post-invasion conditions delayed when fish became vulnerable to the fishery, thus decreasing selection pressure and lessening the evolutionary response to harvest. Fishing with gill nets and traps nets generally selected for early maturation at small sizes, except when fishing at low levels with small mesh gill nets under pre-invasion conditions; in this latter case, evolution of delayed maturation was predicted. Overall, the invasion of dreissenid mussels lessened the evolutionary response to harvest, while also reducing the productivity and commercial yield potential of the stock. These results demonstrate how ecological conditions shape evolutionary outcomes and how invasive species can have a direct effect on evolutionary responses to harvest and sustainability.

    View PDFchevron_right
    Shoal choice in zebrafish, Danio rerio: the influence of shoal size and activity
    Victoria Pritchard

    Animal Behaviour, 2001

    Shoaling fish are expected, in many cases, to gain fitness benefits from being in a larger shoal and previous experiments have shown that fish are indeed capable of choosing between shoals of different sizes. We investigated the influence of shoal activity on shoal size preference in the zebrafish. We gave test fish the choice between shoals of one to four stimulus fish, presented at two different water temperatures, and so differing in their activity levels. Where all stimulus fish were in water of the same temperature, test fish generally preferred the larger shoal. However, this preference could be reduced by presenting the larger shoal in colder water and so reducing its activity. We discuss these findings with reference to the factors that may influence shoal activity, the effect of temperature on shoaling behaviour and the mechanisms that may be used by fish to discriminate shoal size.

    View PDFchevron_right
    Recreational fishing selectively captures individuals with the highest fitness potential
    Steven J. Cooke, R. Arlinghaus

    Proceedings of the National Academy of Sciences, 2012

    Fisheries-induced evolution and its impact on the productivity of exploited fish stocks remains a highly contested research topic in applied fish evolution and fisheries science. Although many quantitative models assume that larger, more fecund fish are preferentially removed by fishing, there is no empirical evidence describing the relationship between vulnerability to capture and individual reproductive fitness in the wild. Using males from two lines of largemouth bass (Micropterus salmoides) selectively bred over three generations for either high (HV) or low (LV) vulnerability to angling as a model system, we show that the trait "vulnerability to angling" positively correlates with aggression, intensity of parental care, and reproductive fitness. The difference in reproductive fitness between HV and LV fish was particularly evident among larger males, which are also the preferred mating partners of females. Our study constitutes experimental evidence that recreational angling selectively captures individuals with the highest potential for reproductive fitness. Our study further suggests that selective removal of the fittest individuals likely occurs in many fisheries that target species engaged in parental care. As a result, depending on the ecological context, angling-induced selection may have negative consequences for recruitment within wild populations of largemouth bass and possibly other exploited species in which behavioral patterns that determine fitness, such as aggression or parental care, also affect their vulnerability to fishing gear.

    View PDFchevron_right
    The evolutionary legacy of size-selective harvesting extends from genes to populations
    Arne Ludwig, Giovanni Polverino

    Evolutionary applications, 2015

    Size-selective harvesting is assumed to alter life histories of exploited fish populations, thereby negatively affecting population productivity, recovery, and yield. However, demonstrating that fisheries-induced phenotypic changes in the wild are at least partly genetically determined has proved notoriously difficult. Moreover, the population-level consequences of fisheries-induced evolution are still being controversially discussed. Using an experimental approach, we found that five generations of size-selective harvesting altered the life histories and behavior, but not the metabolic rate, of wild-origin zebrafish (Danio rerio). Fish adapted to high positively size selective fishing pressure invested more in reproduction, reached a smaller adult body size, and were less explorative and bold. Phenotypic changes seemed subtle but were accompanied by genetic changes in functional loci. Thus, our results provided unambiguous evidence for rapid, harvest-induced phenotypic and evolutio...

    View PDFchevron_right
    Motion cues tune social influence in shoaling fish
    Bertrand Lemasson

    Scientific reports, 2018

    Social interactions have important consequences for individual fitness. Collective actions, however, are notoriously context-dependent and identifying how animals rapidly weigh the actions of others despite environmental uncertainty remains a fundamental challenge in biology. By exposing zebrafish (Danio rerio) to virtual fish silhouettes in a maze we isolated how the relative strength of a visual feature guides individual directional decisions and, subsequently, tunes social influence. We varied the relative speed and coherency with which a portion of silhouettes adopted a direction (leader/distractor ratio) and established that solitary zebrafish display a robust optomotor response to follow leader silhouettes that moved much faster than their distractors, regardless of stimulus coherency. Although recruitment time decreased as a power law of zebrafish group size, individual decision times retained a speed-accuracy trade-off, suggesting a benefit to smaller group sizes in collecti...

    View PDFchevron_right