Applied performance ecology: testing strategies of talent identification in sports using ecological systems - PMC Skip to main content An official website of the United States government Here's how you know Here's how you know Official websites use .gov A .gov website belongs to an official government organization in the United States. Secure .gov websites use HTTPS A lock ( Lock Locked padlock icon ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites. Search Log in Dashboard Publications Account settings Log out Search… Search NCBI Primary site navigation Search Logged in as: Dashboard Publications Account settings Log in Search PMC Full-Text Archive Search in PMC Journal List User Guide PERMALINK Copy As a library, NLM provides access to scientific literature. Inclusion in an NLM database does not imply endorsement of, or agreement with, the contents by NLM or the National Institutes of Health. Learn more: PMC Disclaimer | PMC Copyright Notice J Exp Biol . 2026 Apr 7;229(7):jeb251395. doi: 10.1242/jeb.251395 Search in PMC Search in PubMed View in NLM Catalog Add to search Applied performance ecology: testing strategies of talent identification in sports using ecological systems Lana A Waller Lana A Waller 1 School of the Environment, The University of Queensland, St Lucia, QLD 4072, Australia Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Validation, Visualization, Writing – original draft, Writing – review & editing Find articles by Lana A Waller 1 , Mathew S Crowther Mathew S Crowther 2 School of Life and Environmental Sciences, The University of Sydney, Sydney, NSW 2006, Australia Conceptualization, Formal analysis, Investigation, Methodology, Validation, Visualization, Writing – original draft, Writing – review & editing Find articles by Mathew S Crowther 2 , Simon P Lailvaux Simon P Lailvaux 3 Department of Biological Sciences, University of New Orleans, 2000 Lakeshore Drive, New Orleans, LA 70148, USA Conceptualization, Methodology, Validation, Writing – original draft, Writing – review & editing Find articles by Simon P Lailvaux 3 , Frank Seebacher Frank Seebacher 2 School of Life and Environmental Sciences, The University of Sydney, Sydney, NSW 2006, Australia Conceptualization, Methodology, Validation, Writing – original draft, Writing – review & editing Find articles by Frank Seebacher 2 , Robbie S Wilson Robbie S Wilson 1 School of the Environment, The University of Queensland, St Lucia, QLD 4072, Australia Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Software, Supervision, Validation, Writing – original draft, Writing – review & editing Find articles by Robbie S Wilson 1, ✉ Author information Article notes Copyright and License information 1 School of the Environment, The University of Queensland, St Lucia, QLD 4072, Australia 2 School of Life and Environmental Sciences, The University of Sydney, Sydney, NSW 2006, Australia 3 Department of Biological Sciences, University of New Orleans, 2000 Lakeshore Drive, New Orleans, LA 70148, USA ✉ Author for correspondence ( [email protected] ) Competing interests The authors declare no competing or financial interests. Roles Lana A Waller : Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Validation, Visualization, Writing – original draft, Writing – review & editing Mathew S Crowther : Conceptualization, Formal analysis, Investigation, Methodology, Validation, Visualization, Writing – original draft, Writing – review & editing Simon P Lailvaux : Conceptualization, Methodology, Validation, Writing – original draft, Writing – review & editing Frank Seebacher : Conceptualization, Methodology, Validation, Writing – original draft, Writing – review & editing Robbie S Wilson : Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Software, Supervision, Validation, Writing – original draft, Writing – review & editing Received 2025 Aug 7; Accepted 2026 Jan 13; Collection date 2026 Apr 1. © 2026. Published by The Company of Biologists This is an Open Access article distributed under the terms of the Creative Commons Attribution License ( https://creativecommons.org/licenses/by/4.0 ), which permits unrestricted use, distribution and reproduction in any medium provided that the original work is properly attributed. PMC Copyright notice PMCID: PMC13091497 PMID: 41550026 See the referenced article with doi: 10.1242/jeb.252237 . ABSTRACT Predicting success is a common goal for ecologists and sports scientists, yet these disciplines rarely interact. Sports scientists often use tests of closed-skill or game performances, but these are critiqued for their inherent uncertainties in predicting success. In contrast, ecologists embrace variance, measuring traits under controlled conditions to make probabilistic predictions of success. Integrating ecological perspectives could enhance team selection efficiency in youth sports. Here, we demonstrate this concept using territorial contests in crayfish. As in sports, individual traits in crayfish can be measured rapidly but do not perfectly predict contest outcome. First, we simulated populations of 100 male and 100 female crayfish that competed in 20 rounds of contests and estimated how many individuals must be selected to ensure the top 10% of performers are included. Selections were based on individual traits (body length, claw size and strength) and/or contest outcomes. When few contests had occurred, the top 10% of individuals were most efficiently selected on individual traits but increasingly more on contests as rounds progressed. Empirical data supported these theoretical simulations. We staged 10 rounds of contests among 27 male and 32 female Cherax destructor . After two rounds, ∼21 individuals were needed to capture the top 3; by round 10, ∼5 were required. Taken together, our study provides an initial but compelling demonstration of how ecological models can help improve talent identification strategies in sport. Such an adaptive selection framework efficiently narrows down selection of high-performing individuals under uncertainty and has the potential to be applied to reintroduction and translocation strategies in conservation. Keywords: Ecological models, Individual success, Contests, Crayfish, Traits Summary: Use of a framework informed by ecological models demonstrates how combining trait-based assessments with contest outcomes can improve talent identification in sport by increasing the likelihood of selecting top-performing players, even under uncertainty. INTRODUCTION ‘ The important thing in life is not the triumph, but the fight; the essential thing is not to have won, but to have fought well. ’ Pierre de Coubertin – Founder of the Modern Olympics Ecologists have long studied which traits lead to individual success in nature and how these traits evolve across species and time ( Arnold, 1983 ; Green et al., 2022 ; Lailvaux and Irschick, 2006 ; Violle et al., 2007 ). Just as athletes compete under the rules of their sport, species in nature engage in their own evolutionary contests, shaped by behaviour, environment and ecological constraints ( Briffa and Sneddon, 2007 ; Hardy and Briffa, 2013 ). In some, success depends on speed; in others, stealth or cooperation. Yet, the goal is the same: to survive and reproduce ( Dawkins and Krebs, 1979 ; Stearns, 1998 ). In this way, nature resembles a vast arena of overlapping competitive events, each with distinct challenges, strategies and champions. Evolution ‘rewards’ those best suited to their specific game – just as sporting victories go to those most capable under competitive pressure. While nature and sport share these structural parallels, the disciplines that study them – ecology and sports science – have remained surprisingly separate. Both fields grapple with understanding how excellence emerges, is measured and can be predicted, yet there has been limited crosstalk between them. This paper aims to bridge that gap. We argue that conceptual frameworks from ecology – particularly those used to study adaptation and performance – can provide valuable tools for sports science. Specifically, we suggest that sport can be viewed as a form of applied performance ecology, where success is determined by how well individuals meet the demands of complex, competitive environments. To illustrate this perspective, we offer a proof of concept. Drawing from ecological models of territorial competition, we tested an approach for talent identification in competitive team sports using data from animal contests. This dual application demonstrates how ecological thinking can test strategies for decision making in sport. Talent identification in youth sports aims to recognise individuals most likely to succeed at the professional level. In soccer, expert scouts – relying on match and training observations – remain the dominant evaluators of youth potential ( Christensen, 2009 ; Den Hartigh et al., 2018 ). However, these assessments are often based on limited observations and are prone to bias and inconsistency ( Meylan et al., 2010 ; Williams and Reilly, 2000 ). To address this, sports scientists have introduced some structured, more data-informed methods. These often rely on closed-skill tests – controlled assessments of technical abilities isolated from match environments ( Ali, 2011 ; McCalman et al., 2022 ; Williams and Reilly, 2000 ) – as well as physical attributes (e.g. body size) and other non-technical performance tests (e.g. sprinting or jumping) ( Ali, 2011 ; Camata et al., 2025 ; Reilly et al., 2000 ; Wilson et al., 2025b ). When such tests correlate with in-game performance, they can offer valuable predictive information ( Phillips et al., 2010 ; Wilson et al., 2025a ). However, critics argue that these tests oversimplify the dynamic nature of match play and typically explain only a modest proportion of variance in complex, game-realistic performances ( Bergkamp et al., 2019 ; Travassos et al., 2013 ). Thus, players who excel in drills do not always translate their abilities effectively in real matches. Although performance evaluations from game-based assessments may offer richer insights into an individual's ability in matches, these data introduce their own challenges ( Heilmann et al., 2022 ; McCalman et al., 2022 ). Within games, performance is affected by teammates, opponents and chance ( Reilly et al., 2000 ). Furthermore, it is logistically difficult to run enough games to reliably evaluate all players in large youth cohorts and provide the required statistical confidence on an individual's ability ( Wilson et al., 2021 ). As a result, few professional clubs rely on quantitative approaches alone. Yet, from an ecological perspective, discarding imperfect predictors simply because they do not offer certainty is a missed opportunity. In nature, ecologists routinely measure animal traits under controlled conditions, despite their artificial scenario, because these traits can still explain a proportion of variance of performance in complex environments ( Husak et al., 2006 ; Lailvaux and Husak, 2014 , 2017 ). Predictive accuracy in ecology often depends on recognising uncertainty and integrating multiple sources of information ( Hendry, 2023 ; Milner-Gulland and Shea, 2017 ; Mouquet et al., 2015 ). Performance traits never provide perfect predictions, but they can still improve decision making when used appropriately. For ecologists, success lies in balancing what can be measured with what can be reliably inferred, managing risk by minimising false positives while still capturing signals of promise. Recent work by Wilson et al. (2025b) applied this approach to sport. They developed a data-driven model to identify top performers in youth soccer based on outcomes from 1v1 contests – a simplified, yet dynamic version of real matches. By combining a closed-skill measure (dribbling speed) with win/loss data from over 1300 1v1 games, they showed that initial predictions could be made from closed-skill tests and refined over time with contest outcomes. This adaptive selection model efficiently reduced the pool of players for further scouting while managing uncertainty throughout the process. Their approach mirrored ecological principles: use imperfect traits to guide early decisions and improve predictions through ongoing competitive evaluation. Beyond offering conceptual tools, ecology can also provide real-world systems to test and refine talent identification strategies. Many animals engage in contests over resources such as mates or territories, often using signals and displays to predict performance ( Abalos et al., 2024 ; Hardy and Briffa, 2013 ; Maynard Smith and Harper, 2003 ; Walter et al., 2011 ). For example, male red deer roar to intimidate rivals ( Clutton-Brock and Albon, 1979 ), while paper wasps use black facial patterns as signals of status to assess rivals before combat ( Tibbetts and Dale, 2004 ; Tibbetts and Lindsay, 2008 ). In these systems, outcomes of the actual contest are clear (win or loss), and traits associated with contest outcomes, such as size or strength, can be measured rapidly. Crayfish are especially well suited to this kind of analysis. These freshwater crustaceans engage in ritualised contests using their claws, which serve both as signals of strength and as weapons for combat ( Bywater et al., 2008 ; Graham and Angilletta, 2020 ; Walter et al., 2011 ; Wilson et al., 2007 ). When claw size differs, contests are often resolved through visual display alone; when opponents appear evenly matched, strength usually determines the winner ( Wilson et al., 2007 ). Crucially, claw size does not always predict strength, as some individuals have large but weak claws ( Walter et al., 2011 ; Wilson et al., 2007 ). This mismatch between signal size and true ability mirrors a key challenge in sport: easily measurable traits are not always reliable indicators of competitive success and there are multiple factors that ultimately dictate success ( Bergkamp et al., 2019 ; Vaeyens et al., 2007 ; Williams and Reilly, 2000 ; Wilson et al., 2017 ). Thus, crayfish offer a natural proof of concept for testing talent identification models using ecological systems. Their contests are observable, outcomes are binary and traits can be measured quickly and non-invasively. By integrating trait data with contest outcomes, researchers can build and test predictive models just as one would in sport, while accounting for uncertainty and variance. Crayfish therefore provide an ecologically valid, tractable system for advancing strategies for predicting ecological success and which in this case is analogous to talent identification in sport. Building on this logic, we used crayfish as a proof of concept to test the adaptive selection model for talent identification proposed by Wilson et al. (2025b) . Our goal was to assess how efficiently different selection strategies could identify the top 10% of individuals within a population, based on either traits or contest outcomes, or a combination of the two. We first conducted simulations to estimate the minimum number of individuals that would need to be selected to ensure with 95% confidence that the true top 10% were always selected. Using populations of 100 individuals, we simulated 20 rounds of contests for both males and females, generating a new population for each of 20 replicate simulations. Relationships between body length and claw size and strength were simulated (with variance) using unpublished data (L..A.W. and R.S.W.) drawn from male and female Cherax destructor . Each individual was assigned a true dominance value that determined their actual rank in the performance hierarchy, with contest outcomes reflecting these underlying abilities. Contest ability was modelled using a Bradley–Terry model ( Bradley and Terry, 1952 ), and we systematically varied the relative weighting of trait and contest information using an alpha (α) parameter to assess how optimal selection strategies shift as contest data accumulate. We hypothesised that in early rounds, when contest data are sparse, trait-based selection (body length, claw size or claw strength) would be more efficient, requiring fewer individuals to reliably include the top performers. As the contest data set increased, however, we expected contest outcomes to gain increasing predictive power, improving selection precision. To mimic more realistic constraints common in youth sports, we also simulated a second scenario where only the top 40 individuals by trait ranking were allowed to compete in contests. We predicted that this pre-filtering step would further improve efficiency by focusing resources on likely high performers. Finally, we tested these hypotheses using empirical data from staged contests in 27 male and 32 female C. destructor . We measured morphological traits (body length, claw size), claw strength and recorded outcomes across 10 rounds of 1v1 contests. Individual dominance scores were estimated using the Bradley–Terry model, and we tracked how the number of individuals that needed to be selected to include the top performers (i.e. the top three by dominance score) changed as contest data accumulated. Consistent with our simulations, we expected the number of required individuals to decline as contest data accumulated, demonstrating the practical value of an adaptive, ecologically grounded selection strategy. By applying this framework to both simulated and empirical crayfish contests, we extend the model of Wilson et al. (2025b) to an ecological system and demonstrate how adaptive selection strategies that integrate multiple information sources while accounting for variance can efficiently identify top performers. MATERIALS AND METHODS Simulating traits and true dominance We simulated traits that predicted dominance for 100 male and 100 female crayfish, Cherax destructor (Clark 1936), and we repeated these simulations 20 times per sex. Trait equations were based on empirical data describing morphology and claw strength in C. destructor ( n =71 males, n =74 females) (see Supplementary Materials and Methods, Eqns S1–S8 ; L.A.W. and R.S.W., unpublished data). For both sexes, body length was drawn from a normal distribution (mean=0, s.d.=1). Claw size was generated as a linear function of body length with added Gaussian noise (males s.d.=0.26, females s.d.=0.22). Claw strength was derived from claw size with additional noise (males s.d.=0.54, females s.d.=0.69). We defined a trait-based predictor of contest success for each individual using the combination of traits that best explained outcome in empirical staged contests, as identified by a Bradley–Terry model (L.A.W. and R.S.W., unpublished data). The Bradley–Terry model estimates the probability of an individual winning a contest based on their traits and/or contest outcomes ( Bradley and Terry, 1952 ). We used this model in preliminary analyses to determine which of body length, claw size and claw strength – traits previously shown to influence contest outcome in crayfish ( Bywater et al., 2008 ; Graham et al., 2020 ; Walter et al., 2011 ; Wilson et al., 2007 ) – were most predictive of success in empirical staged contests among C. destructor crayfish (as detailed below). For males, the predictor was calculated as an equally weighted sum of claw size and claw strength. For females, the predictor included body length and claw strength. Individuals were then ranked within each sex and replicate according to their trait score (see Table 1 for definitions). Table 1. Definitions of rank types used in simulated and empirical analyses Rank type Simulation definition Empirical definition Trait rank Ranking based on sum of simulated traits (e.g. body length, claw size and/or claw strength), each with equal weighting Ranking based on sum of measured traits (e.g. body length, claw size and/or claw strength), each with equal weighting Contest rank Ranking derived from simulated contest outcomes using Bradley–Terry model Ranking derived from staged contest outcomes using Bradley–Terry model True rank Ranking based on simulated true dominance value, calculated by adding noise to trait predictor Final ranking based on estimated dominance scores from Bradley–Terry model after all contest rounds Open in a new tab An individual's true dominance was simulated by adding noise (s.d.=0.7) to the trait-based predictor ( y = x +error) to reflect variation in competitive ability. Individuals were then ranked within each replicate according to their simulated dominance values to generate their true rank ( Table 1 ). Simulating contests and dominance ranks We simulated 20 rounds of contests per replicate. In each round, 50 pairs of individuals were formed randomly without replacement. Contest outcomes were resolved deterministically such that the individual with the higher true dominance value always won. We recorded the winner and loser of each contest. To estimate dominance abilities from contest outcomes, we fitted a Bradle–Terry model every two rounds (i.e. rounds 2, 4…20), where each model used cumulative contest outcomes up to and including the respective round. Estimated dominance scores for each individual were extracted from the Bradley–Terry models at each respective round and individuals were ranked accordingly to generate contest-based ranks ( Table 1 ). Evaluating selection strategies using trait, contest and hybrid ranks We aimed to quantify how effectively different selection strategies could identify the top 10% of individuals in a simulated population. Each individual was assigned three ranks: a true rank based on their simulated true dominance value, representing their underlying ability; a trait rank based only on measurable traits, calculated as the sum of claw size and claw strength for males or body length and claw strength for females; and a contest rank, reflecting performance in staged contests which was estimated using Bradley–Terry models fitted to cumulative contest outcomes every two rounds. To compare selection strategies, we calculated how many of the top-ranked individuals needed to be selected to ensure that all of the true top 10% (i.e. top 10 based on true dominance) were included. These strategies included selection based on trait ranks only, contest performance only, and a hybrid score combining the two: (1) where α (alpha) controls the weighting of trait versus contest information, ranging from 0 (contest only) to 1 (trait only), in increments of 0.1. We calculated the hybrid score under two scenarios: (i) contests involving all 100 individuals, and (ii) contests conducted after a preliminary cut, where the bottom 60 individuals based on trait rank were removed, such that contests only occurred amongst crayfish with the top 40 trait ranks. For each scenario, contest round and α value, we were therefore able to compare the minimum number of individuals required to ensure the true top 10 were captured. Empirical data Study animals and experimental design We used C. destructor sourced from a commercial aquaculture facility (AustSilvers, Swan Bay, NSW, Australia). Individuals were housed at a laboratory at The University of Queensland (St Lucia, QLD, Australia) in individual plastic containers (18×17×18 cm) within larger tanks (61×30×38 cm), at a density of 3–4 crayfish per tank. Tanks contained dechlorinated water, a 2 cm base of gravel and a box filter. Crayfish were fed twice per week with commercial crayfish food pellets [Aqua One Vege Wafers, Kong's (Aust.) Pty Ltd, Ingleburn, NSW, Australia] and were maintained in the laboratory for 12 months before use in the study. We selected 32 females and 27 males that were intermoult and had intact claws for use in experiments. Morphology and claw strength were measured on the first 2 days, followed by staged 1v1 contests that were conducted over a 2 week period. Morphology We measured the body length and claw size of each crayfish using photographs taken with a digital camera (Casio EXLIM EX-100F, Casio Computer Co. Ltd, Tokyo, Japan), with a scale object included for calibration. Images were analysed using ImageJ ( Schneider et al., 2012 ). Body length was measured as the distance between the tip of the rostrum and the end of the telson. We took seven measurements to quantify claw size, which included the width of the propodus at the carpus and dactyl joints, the width of the pollex and dactylus and the length of the propodus, pollex and dactylus ( Bywater et al., 2008 ; Walter et al., 2011 ). We used a principal components analysis (PCA) to combine the seven highly correlated claw dimensions into new orthogonal variables, with separate analyses completed for the left and right claws. The first principal component (PC1) was used as a measure of claw size as all claw dimensions were loaded in the same direction, and it explained over 74% of the variation in the claw data. As both claws were approximately symmetrical, we used the mean PC1 score from the left and right claws as the measure of an individual's claw size. Claw strength We measured the claw strength of crayfish using a custom-built force transducer, as described in detail in Wilson et al. (2007) . The transducer consisted of two outer metal plates that were separated by a central metal plate, with a strain gauge attached to the outer side of one plate. The strain gauge measured the change in resistance when the metal plates bent as force was applied by the claws, which was measured via a Wheatstone bridge connected to a BridgePod amplifier (ADInstruments, Bella Vista, NSW, Australia) and recorded using PowerLab software (ADInstruments). The strain gauge was calibrated using known weights to convert the millivoltage output to newtons. We induced crayfish to grab the metal plates at a standardised location and obtained three measures per claw for each individual, with the process repeated the following day. We used the mean of the maximum force generated by the left and right claws as a measure of an individual's maximum claw strength ( Bywater et al., 2008 ; Walter et al., 2011 ; Wilson et al., 2007 ). Contests We staged contests between pairs of same-sex crayfish (male contests n =122; female contests n =162; Fig. S1 ). Contests occurred in a 70 l tank (61×30×38 cm) containing gravel and dechlorinated tap water at 25°C. In each round, two randomly selected crayfish were simultaneously placed into the arena and allowed to freely interact ( Walter et al., 2011 ; Wilson et al., 2007 ). Interactions between two crayfish either involved only signalling (use of antennae and/or tapping of claws) or escalated to fights (grappling and pushing). Contests ended when one crayfish retreated without re-engaging, and this individual was determined as the loser of each pairing. Individuals that lost a claw were excluded from subsequent contests. Ten rounds of contests were conducted, where each individual participated in 8.9±2.2 contests (mean±s.d.). An eleventh round of contests was conducted in which individuals were paired according to their preliminary ranks after round 10, estimated by a Bradley–Terry model (e.g. rank 1 versus rank 2, rank 3 versus rank 4). This additional round enabled us to improve the reliability of the contest-based rankings. We then estimated each individual's contest rank using all contest outcomes up to and including this final round, which we used as the ‘true’ dominance rank to enable comparison with simulated data ( Table 1 ). As in the simulations, we calculated the minimum number of crayfish that needed to be selected to ensure inclusion of the top 10% (i.e. top 3) based on trait and contest ranks. Statistical analyses To estimate individual dominance abilities from pairwise contest outcomes, we fitted maximum-likelihood Bradley–Terry models using the BTm function from the BradleyTerry2 package in R ( Turner and Firth, 2012 ). For the simulated data, we fitted models every two rounds (rounds 2–20 in steps of 2) using cumulative contest data up to and including each round. The response variable was the win–loss outcome of each contest, with the probability of individual i beating individual j modelled as a logistic function of the difference in their abilities. We extracted individual dominance estimates from the models using the BTabilities function and converted these to rankings. Separate models were fitted for each sex and replicate. To examine the relationship between the number of rounds of contests and the number of individuals needed to be selected to include the true top 10% in the simulations, we fitted generalised additive models (GAMs) using the gam function from the mgcv package in R ( Wood, 2017 ). The GAM modelled the number of individuals needed to include the top 10% as a smooth function of the number of contest rounds, using a basis dimension of k =10 to avoid overfitting. Models were fitted using restricted maximum likelihood (REML) as it provides a more robust and stable estimation of smoothing parameters ( Wood, 2011 , 2017 ). Rounds 2–20 in steps of two were analysed to improve model fit. Model diagnostics and fit were assessed using the gam.check function. Separate models were fitted for males and females. We fitted additional GAMs to evaluate how different selection strategies (by hybrid score) influence the number of individuals required to ensure inclusion of the top 10%. A tensor product spline ( te function) was used to model a potential non-linear interaction between α (0 to 0.1 in steps of 0.1) and round (4–20 in steps of 2). Smoothing parameters were estimated using REML, with basis dimensions set to k =10 for α and k =8 for round. Model diagnostics and fit were assessed as above. Separate models were fitted for males and females, and for the two selection scenarios: (i) when all 100 individuals were included and (ii) when contests after the bottom 60 individuals were cut based on trait rank. We examined the empirical contest data to assess how well traits predicted dominance. Contest rankings were estimated using Bradley–Terry models, fitted at every second round (rounds 2–10, and 11) using pairwise contest outcomes, following the same methods used in the simulations. We first tested the association between contest rank and the composite trait score, calculated as an equally weighted sum of claw size and strength for males, and of body length and claw strength for females, reflecting the weighting used in the simulations. We fitted a linear regression model using the lm function with contest rank as the response variable and trait score as the predictor variable. To determine the actual relative contribution of each trait to dominance in the empirical data, we fitted additional multiple regression models with claw size and claw strength (males) and body length and claw strength (females) as predictors, with true rank as the response variable. All traits were standardised (mean=0, s.d.=1). All simulations and analyses were conducted in R version 4.4.3 ( https://www.r-project.org/ ). We used dplyr ( https://CRAN.R-project.org/package=dplyr ) and tidyr ( https://CRAN.R-project.org/package=tidyr ) for data manipulation and ggplot2 ( Wickham, 2016 ) for data visualisation. RESULTS Simulated selection across rounds On average, 34.3 males (95% confidence interval, CI [28, 40.6]) and 35 females (95% CI [28.7, 41.2]) were required to include the top 10% of individuals based on pre-contest trait ranks ( Fig. 1 ). Using contest data, the number of individuals required decreased non-linearly with the number of contest rounds for both males ( P <0.001; Table 2 ) and females ( P <0.001; Table 2 ). After 4 rounds, an average of 38.3 males (95% CI [34, 42.6]) and 41.7 females (95% CI [36.6, 46.8]) had to be selected to guarantee that the top 10 were included. This number decreased with additional rounds, reaching 20.8 males (95% CI [17.2, 24.4]) and 25.4 females (95% CI [22.6, 28.2]) after 10 rounds, and plateaued at approximately 17 individuals for both sexes after 14 rounds. Fig. 1. Open in a new tab Mean number of individuals needed to include the true top 10 in simulated contests. ‘Pre-contests’ refers to selection based on trait ranks only for (A) males and (B) females. Rankings across rounds were estimated using Bradley–Terry models with cumulative contest data. Points, error bars and shading represent means±95% confidence intervals across 20 replicate simulations, each with 100 simulated crayfish. Table 2. Summary of six generalised additive models (GAMs) predicting the number of individuals required to ensure the top 10% are included, with either round or interaction between α and round Model Smooth term e.d.f. F P Deviance explained (%) Male (Round) s(Round) 7.71 122 <0.001 84.7 Female (Round) s(Round) 7.14 115 <0.001 83.2 Male hybrid (100) te(Alpha, Round) 17.6 37.8 <0.001 32.2 Female hybrid (100) te(Alpha, Round) 20.6 37.6 <0.001 35.6 Male hybrid (40) te(Alpha, Round) 18.9 62.3 <0.001 45.2 Female hybrid (40) te(Alpha, Round) 19.8 54.1 <0.001 43.3 Open in a new tab EDF, effective degrees of freedom. Selection based on hybrid scores When contests involved all 100 individuals, the number of males required to ensure inclusion of the top 10% depended on a non-linear interaction between the weighting of trait ranks versus contest ranks (α) and the number of rounds ( P <0.001), with the model explaining 32.2% of the deviance ( Table 2 , Fig. 2A ). This interaction indicates that the α value associated with the smallest number of individuals that need to be included decreased as the number of rounds increased. For example, at round 4, the optimal α was 0.6, requiring selection of 28 males on average (95% CI [24.5, 31.5]), whereas by round 20, the optimal α decreased to 0.1, requiring only 14 males (95% CI [12.7, 15.2]). Fig. 2. Open in a new tab Influence of α (weight on trait rank) on mean number of Cherax destructor individuals required to ensure inclusion of the top 10 performers (100 individuals per group). α determines weighting between trait-based (α=1) and contest-based rankings (α=0). Results are shown for (A) males and (B) females for a subset of rounds. Points and shading represent means±s.e.m. across 20 replicates, each with 100 simulated crayfish. Females exhibited a similar pattern, where the number of individuals needed depended on a non-linear interaction between α and the number of rounds ( P <0.001; Table 2 , Fig. 2B ). The GAM explained 35.6% of the deviance ( Table 2 ). At round 4, the optimal α was 0.5, where 29.6 females needed to be selected to ensure the top 10 were included (95% CI [25.5, 33.6]). By round 20, the optimal α decreased to 0.1, where only 14.2 females, on average, needed to be selected (95% CI [12.7, 15.7]). When contests involved only the top 40 individuals based on trait ranks, the relationship between α and the number of rounds was still non-linear for males and females (both P <0.001; Table 2 ). For males, the optimal α decreased from 0.7 at round 4 to 0.2 by round 20, reducing the number of individuals needed to include the top 10 from 24 (95% CI [21.5, 26.5]) to 12.7 (95% CI [11.8, 13.5]) ( Fig. 3A ). Similarly, for females, the optimal α decreased from 0.7 to 0.1 from round 4 to round 20, with the number required declining from 24.9 (95% CI [22.5, 27.2]) to 13.8 females (95% CI [12.3, 15.3]; Fig. 3B ). Optimal α values were slightly higher when contests were restricted to the top 40 individuals compared with when all 100 individuals competed. Fig. 3. Open in a new tab Influence of α (weight on trait rank) on mean number of individuals required to ensure inclusion of the top 10 performers (top 40 trait rank). The bottom 60 individuals were excluded prior to contests based on trait rank. α determines weighting between trait-based (α=1) and contest-based rankings (α=0). Results are shown for (A) males and (B) females for a subset of rounds. Points and shading represent means±s.e.m. across 20 replicates, each with 40 simulated crayfish. Empirical trait and contest data Males with higher trait scores, reflecting larger claw sizes and greater claw strength, had better (lower score) contest rankings ( R 2 =0.35, F 1,25 =23.23, P =0.001; Fig. 4A ). Similarly, females with higher trait scores, and thus larger body lengths and greater claw strength, tended to exhibit better rankings ( R 2 =0.51, F 1,30 =31.54, P< 0.001; Fig. 4B ). Fig. 4. Open in a new tab Relationship between trait score and true rank for individuals . For (A) males ( n =27), trait score was calculated as equal weighting of claw size and claw strength, and for (B) females ( n =32) as equal weighting of body length and claw strength. True rank was estimated from a Bradley–Terry model after 11 rounds of contests, and lower ranks indicate better performance. Shading represents standard error. When determining the actual contribution of each trait to dominance, claw size was more important in predicting contest rank in males (estimate=−4.21, t =−3.23, P =0.004) than claw strength (estimate=−1.75, t =−1.34, P =0.19). However, increases in both traits were still associated with better rankings ( R 2 =0.38, F 2,24 =7.46, P =0.003). For females, body length was a stronger predictor of final rank (estimate=−6.65, t =−5.83, P <0.001) than claw strength (estimate=−3.46, t =−3.03, P =0.005), with increases in both traits contributing to better overall rankings ( R 2 =0.57, F 2,29 =19.8, P< 0.001). Based on trait ranks alone, 12 males and 3 females needed to be selected to ensure inclusion of the top 3 (10%) individuals ( Fig. 5 ). As the number of rounds progressed, the number of individuals required decreased. After 2 rounds, 23 males and 19 females had to be selected to include the top 3, which decreased to 3 males and 7 females after 10 rounds. Fig. 5. Open in a new tab Rank of top three C. destructor in staged contests across rounds. For (A) males and (B) females, the number in data points represents the true rank of the individual (1, 2 and 3), assumed as their estimated rank after 11 rounds of contests. ‘Pre-contests’ represents rank based on trait scores. Contest ranks were estimated at each round using a Bradley–Terry model. DISCUSSION This study aimed to test whether approaches common in ecology could be used to test the efficacy of talent identification strategies using crayfish contests as a simplified and tractable system. Our results demonstrate that by combining trait-based predictions with repeated contest outcomes, one can effectively narrow the pool of individuals needed to identify the top performers. In both simulated and empirical analyses, the number of individuals required to capture the top 10% of dominant performers declined non-linearly across contest rounds. For example, in the full simulated population, selection based solely on pre-contest trait rankings required an average of 34 males and 35 females. After 10 rounds of contests, this number dropped to 21 males and 25 females and eventually plateaued at around 17 individuals per sex by round 14. Hybrid models that integrated trait and contest data were even more efficient. When contest participation was restricted to the top 40 individuals based on trait ranks, which mimics pre-selection processes in sports, efficiency improved further, requiring only 13 males and 14 females by round 20 to capture the top 10%. Empirical data based on male and female C. destructor supported these findings. For males, trait scores (claw size and strength) explained 35% of the variation in contest rankings, while for females, trait scores (body length and claw strength) explained 51%. In early rounds, identifying the top 3 performers required selecting 23 males and 19 females; by round 10, this dropped to just 3 males and 7 females. These findings highlight how trait-based pre-selection, refined by contest outcomes, can be used to efficiently identify high-performing individuals even under uncertainty. Our experiment and analyses serve as a proof of concept for the broader aims of applying ecological approaches to testing talent identification strategies. Our findings offer valuable insights that form the foundation from which to develop improved statistical models that better capture the relative and potentially interactive effects of traits on dominance. Future directions include staging contests in more complex environmental situations such as using multiple competitors and environmental variation, which in a sporting context would resemble multi-player game situations and different weather and field conditions. Additionally, we made simplifying assumptions such as using only a limited set of traits to predict dominance outcomes. It would be very interesting to determine whether additional traits can improve contest outcomes and how different traits are ranked. A trait-ranking approach would test whether different traits contributed equally and additively to performance and could thereby facilitate predictions of individual fitness in particular natural environments, or game performance in a sport context. Future experiments can incorporate these more refined trait characteristics into statistical models and test their predictive power across multiple independent groups. Ecologically, extending this work into mesocosm experiments where success can be defined by survival, growth and population sizes will further increase its relevance to nature and enhance analogies to the complexity of sports. Such ecological systems would offer a closer parallel to the multi-agent, dynamic conditions of team sports, where individuals interact in complex, unpredictable environments. Identifying true potential in youth athletes requires navigating a complex landscape where age, size and developmental maturity can obscure underlying future quality, but ecological model systems may also help to address this issue. In youth soccer, players are typically grouped into annual age cohorts that include substantial variation in both chronological age and physical maturity, which makes the identification of future potential even more complex ( Carling et al., 2009 ; Gil et al., 2014 ). For example, in an under-13 elite soccer squad, a player born in January can be 10% older than a player born in December. This discrepancy confers significant advantages in physical performance and coordination, leading to a widespread bias where older, more mature players within cohorts are often mistaken for more promising individuals ( Gastin and Bennett, 2014 ; Helsen et al., 1998 ; Philippaerts et al., 2006 ). The stronger and more physically developed individuals are better at the time of selection but this may have little relevance to their actual relative performance at maturity if later-maturing individuals are also provided with equivalent opportunities. Consequently, soccer academies around the world are disproportionately filled with players born in the early months of the year ( Figueiredo et al., 2009 ) – up to 60% of players in elite academies may be born in the first 3 months of a cohort year ( Hunter et al., 2023 ). This phenomenon, known as the relative age effect, represents not only a form of structural discrimination but also a missed opportunity for club success, as many fail to identify younger players with long-term potential. Bio-banding is one attempt to mediate the effects of age-biases in team sports where players are placed into ‘bands’ or groups based on stages of physical development rather than age for specific competitions and training ( Bradley et al., 2019 ; Lüdin et al., 2022 ; Malina et al., 2019 ). Despite the positive attitude towards these measures, there is limited evidence that selection processes or outcomes have changed as a result of bio-banding, because this banding usually occurs after players have been selected for talent squads ( Bradley et al., 2019 ; Lüdin et al., 2022 ; Malina et al., 2019 ). An alternative approach to addressing biases in youth selection is to develop and use age- and size-corrected assessments of individual players, particularly for skill-based traits because these are less sensitive to rapid changes in growth and physical development ( Hunter et al., 2022 ). Incorporating age- and size-corrected traits into sequential selection strategies and then testing their efficacy will be complicated, but analyses such as ours based on ecological model systems such as crayfish may provide novel and easily manipulated systems. Outcomes of competitive interactions in these ecological systems depend on multiple overlapping variables such as age, size, aggression and ecological context ( Graham and Angilletta, 2020 ; Gruber et al., 2016 ; Walter et al., 2011 ). Therefore, ecological systems offer a unique opportunity to examine how performance traits interact under realistic but experimentally tractable settings. For example, researchers can disentangle the confounding effects of developmental timing and physical growth using experimental manipulations, and then test different strategies for assessing relative performance under controlled conditions. Insights from such experiments could inform more equitable, evidence-based approaches to youth talent identification, ultimately helping clubs recognise and develop players based on true potential rather than temporary physical advantages. While this study has focused on the value of ecological perspectives and systems for improving talent identification frameworks for sport, the conceptual exchange need not be unidirectional. The analytical tools and decision-making frameworks developed for identifying top performers in sport may hold reciprocal value for ecological applications, particularly in conservation programmes involving animal translocations or captive breeding for reintroduction. These initiatives often hinge on selecting individuals with the greatest likelihood of long-term survival and reproductive success in the wild ( Berger-Tal et al., 2020 ; Seddon et al., 2007 ). For instance, Biggins et al. (2011) examined the reintroduction of black-footed ferrets ( Mustela nigripes ) into a previously unoccupied prairie dog colony and found that wild-born individuals, compared with captive-born counterparts, exhibited significantly higher short-term survival. The captive-born ferrets showed greater nightly movements that were correlated with lower survival ( Biggins et al., 2011 ). These findings underscore the influence of behavioural traits, such as boldness, shelter use and competitive ability, on reintroduction outcomes. In this context, a refined understanding of trait-to-performance relationships and its connection with success, as found in talent identification strategies in sport, could significantly enhance selection accuracy for specific favourable traits. By applying such frameworks, ecologists may improve the efficacy of translocation and reintroduction efforts by identifying and prioritising individuals with favourable trait combinations most aligned with post-release success, especially during the high-risk early stages. Beyond conservation, this approach is also relevant to understanding how whole-organism performance translates into competitive success and ultimately fitness. By explicitly linking morphological and behavioural traits to performance outcomes and success, this framework provides a tool to revisit fundamental questions within the morphology–performance–fitness paradigm ( Arnold, 1983 ). More broadly, this cross-disciplinary approach highlights a fertile area for innovation, where sports science and ecology co-develop tools for managing individual variation in complex, high-stakes environments. The cross-disciplinary approach used in this study shows promise not only for improving fairness and accuracy in youth sport selection but also for ecological applications such as translocation and reintroduction programmes, where choosing the right individuals may be critical. Supplementary Material Supplementary information jexbio-229-251395-s1.pdf (1,020.4KB, pdf) DOI: 10.1242/jexbio.251395_sup1 Acknowledgements We thank members of the Wilson Performance Lab who helped with data collection. Footnotes Author contributions Conceptualization: L.A.W., M.S.C., S.P.L., F.S., R.S.W.; Data curation: L.A.W., R.S.W.; Formal analysis: L.A.W., M.S.C., R.S.W.; Funding acquisition: R.S.W.; Investigation: L.A.W., M.S.C., R.S.W.; Methodology: L.A.W., M.S.C., S.P.L., F.S., R.S.W.; Project administration: R.S.W.; Resources: R.S.W.; Software: R.S.W.; Supervision: R.S.W.; Validation: L.A.W., M.S.C., S.P.L., F.S., R.S.W.; Visualization: L.A.W., M.S.C.; Writing – original draft: L.A.W., M.S.C., S.P.L., F.S., R.S.W.; Writing – review & editing: L.A.W., M.S.C., S.P.L., F.S., R.S.W. Funding This research was supported by Hidden Vale Conservation Research grants and University of Queensland funding. Open Access funding provided by University of Queensland. Deposited in PMC for immediate release. Data and resource availability Data and code available from the corresponding author upon request. All other relevant data and details of resources can be found within the article. ECR Spotlight This article has an associated ECR Spotlight interview with Lana Waller . Special Issue This article is part of the special issue ‘The Integrative Biology of Exercise’, guest edited by Erika Eliason, Christopher Guglielmo, Natalie Holt and Monica Daley. See related articles at https://journals.biologists.com/jeb/issue/229/7 . References Abalos, J., Pérez i de Lanuza, G., Bartolomé, A., Liehrmann, O., Aubret, F. and Font, E. (2024). Behavioral threat and appeasement signals take precedence over static colors in lizard contests. Behav. Ecol. 35, arae045. 10.1093/beheco/arae045 [ DOI ] [ Google Scholar ] Ali, A. (2011). Measuring soccer skill performance: a review. Scand. J. Med. Sci. Sports 21, 170-183. 10.1111/j.1600-0838.2010.01256.x [ DOI ] [ PubMed ] [ Google Scholar ] Arnold, S. J. (1983). Morphology, performance and fitness. Am. Zool. 23, 347-361. 10.1093/icb/23.2.347 [ DOI ] [ Google Scholar ] Berger-Tal, O., Blumstein, D. and Swaisgood, R. (2020). Conservation translocations: a review of common difficulties and promising directions. Anim. Conserv. 23, 121-131. 10.1111/acv.12534 [ DOI ] [ Google Scholar ] Bergkamp, T. L. G., Niessen, A. S. M., den Hartigh, R. J. R., Frencken, W. G. P. and Meijer, R. R. (2019). Methodological issues in soccer talent identification research. Sports Med. 49, 1317-1335. 10.1007/s40279-019-01113-w [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Biggins, D. E., Godbey, J. L., Horton, B. M. and Livieri, T. M. (2011). Movements and survival of black-footed ferrets associated with an experimental translocation in South Dakota. J. Mammal. 92, 742-750. 10.1644/10-MAMM-S-152.1 [ DOI ] [ Google Scholar ] Bradley, B., Johnson, D., Hill, M., McGee, D., Kana-Ah, A., Sharpin, C., Sharp, P., Kelly, A., Cumming, S. P. and Malina, R. M. (2019). Bio-banding in academy football: player's perceptions of a maturity matched tournament. Ann. Hum. Biol. 46, 400-408. 10.1080/03014460.2019.1640284 [ DOI ] [ PubMed ] [ Google Scholar ] Bradley, R. A. and Terry, M. E. (1952). Rank analysis of incomplete block designs: the method of paired comparisons. Biometrika 39, 324-345. [ Google Scholar ] Briffa, M. and Sneddon, L. U. (2007). Physiological constraints on contest behaviour. Funct. Ecol. 21, 627-637. 10.1111/j.1365-2435.2006.01188.x [ DOI ] [ Google Scholar ] Bywater, C. L., Angilletta, M. J. and Wilson, R. S. (2008). Weapon size is a reliable indicator of strength and social dominance in female slender crayfish ( Cherax dispar ). Funct. Ecol. 22, 311-316. 10.1111/j.1365-2435.2008.01379.x [ DOI ] [ Google Scholar ] Camata, T. V., Hunter, A. H., Smith, N. M. A., Crowther, M. S., Costa, M. A., Moura, F. A. and Wilson, R. S. (2025). Influence of age and body size on the dribbling performance of young elite soccer players: a cross-sectional descriptive study. J. Funct. Morphol. Kinesiol. 10, 118. 10.3390/jfmk10020118 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Carling, C., Le Gall, F., Reilly, T. and Williams, A. M. (2009). Do anthropometric and fitness characteristics vary according to birth date distribution in elite youth academy soccer players? Scand. J. Med. Sci. Sports 19, 3-9. 10.1111/j.1600-0838.2008.00867.x [ DOI ] [ PubMed ] [ Google Scholar ] Christensen, M. K. (2009). “An eye for talent”: talent identification and the “practical sense” of top-level soccer coaches. Sociol Sport J. 26, 365-382. 10.1123/ssj.26.3.365 [ DOI ] [ Google Scholar ] Clutton-Brock, T. H. and Albon, S. D. (1979). The roaring of reed deer and the evolution of honest advertisement. Behaviour 69, 145-170. 10.1163/156853979X00449 [ DOI ] [ Google Scholar ] Dawkins, R. and Krebs, J. R. (1979). Arms races between and within species. Proc. R. Soc. Lond. B Biol. Sci. 205, 489-511. 10.1098/rspb.1979.0081 [ DOI ] [ PubMed ] [ Google Scholar ] Den Hartigh, R. J. R., Niessen, A. S. M., Frencken, W. G. P. and Meijer, R. R. (2018). Selection procedures in sports: Improving predictions of athletes' future performance. Eur. J. Sport Sci. 18, 1191-1198. 10.1080/17461391.2018.1480662 [ DOI ] [ PubMed ] [ Google Scholar ] Figueiredo, A. J., Gonçalves, C. E., Coelho e Silva, M. J. and Malina, R. M. (2009). Characteristics of youth soccer players who drop out, persist or move up. J. Sports Sci. 27, 883-891. 10.1080/02640410902946469 [ DOI ] [ PubMed ] [ Google Scholar ] Gastin, P. B. and Bennett, G. (2014). Late maturers at a performance disadvantage to their more mature peers in junior Australian football. J. Sports Sci. 32, 563-571. 10.1080/02640414.2013.843016 [ DOI ] [ PubMed ] [ Google Scholar ] Gil, S. M., Badiola, A., Bidaurrazaga-Letona, I., Zabala-Lili, J., Gravina, L., Santos-Concejero, J., Lekue, J. A. and Granados, C. (2014). Relationship between the relative age effect and anthropometry, maturity and performance in young soccer players. J. Sports Sci. 32, 479-486. 10.1080/02640414.2013.832355 [ DOI ] [ PubMed ] [ Google Scholar ] Graham, Z. A. and Angilletta, M. J. (2020). Claw size predicts dominance within and between invasive species of crayfish. Anim. Behav. 166, 153-161. 10.1016/j.anbehav.2020.06.021 [ DOI ] [ Google Scholar ] Graham, Z. A., Padilla-Perez, D. J. and Angilletta, M. J. (2020). Virile crayfish escalate aggression according to body size instead of weapon size. Anim. Behav. 163, 9-15. 10.1016/j.anbehav.2020.02.014 [ DOI ] [ Google Scholar ] Green, S. J., Brookson, C. B., Hardy, N. A. and Crowder, L. B. (2022). Trait-based approaches to global change ecology: moving from description to prediction. Proc. R. Soc. Lond. B Biol. Sci. 289, 20220071. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Gruber, C., Tulonen, J., Kortet, R. and Hirvonen, H. (2016). Resource availability and predation risk influence contest behavior and dominance hierarchies in crayfish. Behav. Ecol. Sociobiol. 70, 1305-1317. 10.1007/s00265-016-2139-6 [ DOI ] [ Google Scholar ] Hardy, I. C. and Briffa, M. (2013). Animal Contests : Cambridge University Press. [ Google Scholar ] Heilmann, F., Weinberg, H. and Wollny, R. (2022). The impact of practicing open- vs. closed-skill sports on executive functions-a meta-analytic and systematic review with a focus on characteristics of sports. Brain Sci. 12, 1071. 10.3390/brainsci12081071 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Helsen, W. F., Starkes, J. L. and Van Winckel, J. (1998). The influence of relative age on success and dropout in male soccer players. Am. J. Hum. Biol. 10, 791-798. 10.1002/(SICI)1520-6300(1998)10:6<791::AID-AJHB10>3.0.CO;2-1 [ DOI ] [ PubMed ] [ Google Scholar ] Hendry, A. P. (2023). Prediction in ecology and evolution. Bioscience 73, 785-799. 10.1093/biosci/biad083 [ DOI ] [ Google Scholar ] Hunter, A. H., Smith, N. M., Camata, T. V., Crowther, M. S., Mather, A., Souza, N. M., Ramos-Silva, L. F., Pazetto, N. F., Moura, F. A. and Wilson, R. S. (2022). Age-and size-corrected kicking speed and accuracy in elite junior soccer players. Sci. Med. Footb. 6, 29-39. 10.1080/24733938.2021.1899274 [ DOI ] [ PubMed ] [ Google Scholar ] Hunter, A. H., Smith, N. M. A., Camata, T. V., Crowther, M. S., Mather, A., Moura, F. A., Santiago, P. R. P. and Wilson, R. S. (2023). Are soccer players born later in the year more technically skilled than those born earlier in the year. Int. J. Sports Sci. Coach 18, 2015-2024. 10.1177/17479541221117630 [ DOI ] [ Google Scholar ] Husak, J. F., Fox, S. F., Lovern, M. B. and Bussche, R. A. V. D. (2006). Faster lizards sire more offspring: sexual selection on whole-animal performance. Evolution 60, 2122-2130. 10.1111/j.0014-3820.2006.tb01849.x [ DOI ] [ PubMed ] [ Google Scholar ] Lailvaux, S. P. and Husak, J. F. (2014). The life history of whole-organism performance. Q Rev. Biol. 89, 285-318. 10.1086/678567 [ DOI ] [ PubMed ] [ Google Scholar ] Lailvaux, S. P. and Husak, J. F. (2017). Predicting life-history trade-offs with whole-organism performance. Integr. Comp. Biol. 57, 325-332. 10.1093/icb/icx073 [ DOI ] [ PubMed ] [ Google Scholar ] Lailvaux, S. P. and Irschick, D. J. (2006). A functional perspective on sexual selection: insights and future prospects. Anim. Behav. 72, 263-273. 10.1016/j.anbehav.2006.02.003 [ DOI ] [ Google Scholar ] Lüdin, D., Donath, L., Cobley, S. and Romann, M. (2022). Effect of bio-banding on physiological and technical-tactical key performance indicators in youth elite soccer. Eur. J. Sport Sci. 22, 1659-1667. 10.1080/17461391.2021.1974100 [ DOI ] [ PubMed ] [ Google Scholar ] Malina, R. M., Cumming, S. P., Rogol, A. D., Coelho-e-Silva, M. J., Figueiredo, A. J., Konarski, J. M. and Kozieł, S. M. (2019). Bio-banding in youth sports: background, concept, and application. Sports Med. 49, 1671-1685. 10.1007/s40279-019-01166-x [ DOI ] [ PubMed ] [ Google Scholar ] Maynard Smith, J. and Harper, D. (2003). Animal Signals : Oxford University Press. [ Google Scholar ] McCalman, W., Crowley-McHattan, Z. J., Fransen, J. and Bennett, K. J. (2022). Skill assessments in youth soccer: a scoping review. J. Sports Sci. 40, 667-695. 10.1080/02640414.2021.2013617 [ DOI ] [ PubMed ] [ Google Scholar ] Meylan, C., Cronin, J., Oliver, J. and Hughes, M. (2010). Talent identification in soccer: the role of maturity status on physical, physiological and technical characteristics. Int. J. Sports Sci. Coach 5, 571-592. 10.1260/1747-9541.5.4.571 [ DOI ] [ Google Scholar ] Milner-Gulland, E. J. and Shea, K. (2017). Embracing uncertainty in applied ecology. J. Appl. Ecol. 54, 2063-2068. 10.1111/1365-2664.12887 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Mouquet, N., Lagadeuc, Y., Devictor, V., Doyen, L., Duputié, A., Eveillard, D., Faure, D., Garnier, E., Gimenez, O., Huneman, P.et al. (2015). REVIEW: predictive ecology in a changing world. J. Appl. Ecol. 52, 1293-1310. 10.1111/1365-2664.12482 [ DOI ] [ Google Scholar ] Philippaerts, R. M., Vaeyens, R., Janssens, M., Van Renterghem, B., Matthys, D., Craen, R., Bourgois, J., Vrijens, J., Beunen, G. and Malina, R. M. (2006). The relationship between peak height velocity and physical performance in youth soccer players. J. Sports Sci. 24, 221-230. 10.1080/02640410500189371 [ DOI ] [ PubMed ] [ Google Scholar ] Phillips, E., Davids, K., Renshaw, I. and Portus, M. (2010). Expert performance in sport and the dynamics of talent development. Sports Med. 40, 271-283. 10.2165/11319430-000000000-00000 [ DOI ] [ PubMed ] [ Google Scholar ] Reilly, T., Williams, A. M., Nevill, A. and Franks, A. (2000). A multidisciplinary approach to talent identification in soccer. J. Sports Sci. 18, 695-702. 10.1080/02640410050120078 [ DOI ] [ PubMed ] [ Google Scholar ] Schneider, C. A., Rasband, W. S. and Eliceiri, K. W. (2012). NIH Image to ImageJ: 25 years of image analysis. Nat. Methods 9, 671-675. 10.1038/nmeth.2089 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Seddon, P. J., Armstrong, D. P. and Maloney, R. F. (2007). Developing the science of reintroduction biology. Conserv. Biol. 21, 303-312. 10.1111/j.1523-1739.2006.00627.x [ DOI ] [ PubMed ] [ Google Scholar ] Stearns, S. C. (1998). The Evolution of Life Histories : Oxford University Press. [ Google Scholar ] Tibbetts, E. A. and Dale, J. (2004). A socially enforced signal of quality in a paper wasp. Nature 432, 218-222. 10.1038/nature02949 [ DOI ] [ PubMed ] [ Google Scholar ] Tibbetts, E. A. and Lindsay, R. (2008). Visual signals of status and rival assessment in Polistes dominulus paper wasps. Biol. Lett. 4, 237-239. 10.1098/rsbl.2008.0048 [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Travassos, B., Araújo, D., Davids, K., O'Hara, K., Leitão, J. and Cortinhas, A. (2013). Expertise effects on decision-making in sport are constrained by requisite response behaviours – A meta-analysis. Psychol. Sport Exerc. 14, 211-219. 10.1016/j.psychsport.2012.11.002 [ DOI ] [ Google Scholar ] Turner, H. and Firth, D. (2012). Bradley–Terry models in r: the bradleyterry2 package. J. Stat. Softw. 48, 1-21. 10.18637/jss.v048.i09 [ DOI ] [ Google Scholar ] Vaeyens, R., Lenoir, M., Williams, A. M. and Philippaerts, R. M. (2007). Mechanisms underpinning successful decision making in skilled youth soccer players: An analysis of visual search behaviors. J. Mot. Behav. 39, 395-408. 10.3200/JMBR.39.5.395-408 [ DOI ] [ PubMed ] [ Google Scholar ] Violle, C., Navas, M.-L., Vile, D., Kazakou, E., Fortunel, C., Hummel, I. and Garnier, E. (2007). Let the concept of trait be functional!. Oikos 116, 882-892. 10.1111/j.0030-1299.2007.15559.x [ DOI ] [ Google Scholar ] Walter, G. M., Van Uitregt, V. O. and Wilson, R. S. (2011). Social control of unreliable signals of strength in male but not female crayfish, Cherax destructor . J. Exp. Biol. 214, 3294-3299. 10.1242/jeb.056754 [ DOI ] [ PubMed ] [ Google Scholar ] Wickham, H. (2016). ggplot2: Elegant Graphics for Data Analysis . New York: Springer-Verlag. [ Google Scholar ] Williams, A. M. and Reilly, T. (2000). Talent identification and development in soccer. J. Sports Sci. 18, 657-667. 10.1080/02640410050120041 [ DOI ] [ PubMed ] [ Google Scholar ] Wilson, R. S., Angilletta, M. J.Jr, James, R. S., Navas, C. and Seebacher, F. (2007). Dishonest signals of strength in male slender crayfish ( Cherax dispar ) during agonistic encounters. Am. Nat. 170, 284-291. 10.1086/519399 [ DOI ] [ PubMed ] [ Google Scholar ] Wilson, R. S., David, G. K., Murphy, S. C., Angilletta, M. J., Niehaus, A. C., Hunter, A. H. and Smith, M. D. (2017). Skill not athleticism predicts individual variation in match performance of soccer players. Proc. R. Soc. Lond. B Biol. Sci. 284, 20170953. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Wilson, R. S., Hunter, A. H., Camata, T. V., Foster, C. S. P., Sparkes, G. R., Moura, F. A., Santiago, P. R. P. and Smith, N. M. A. (2021). Simple and reliable protocol for identifying talented junior players in team sports using small-sided games. Scand. J. Med. Sci. Sports 31, 1647-1656. 10.1111/sms.13969 [ DOI ] [ PubMed ] [ Google Scholar ] Wilson, R. S., Hunter, A. H., Camata, T. V., Foster, C. S., Sparkes, G. R., Santiago, P. R. and Smith, N. M. (2025a). Dribbling and passing performances predict individual success in small-sided soccer games. Int. J. Sports Sci. Coach 0, 17479541251346902. 10.1177/17479541251346902 [ DOI ] [ Google Scholar ] Wilson, R. S., Sparkes, G. R., Waller, L., Hunter, A. H., Santiago, P. R. P. and Crowther, M. S. (2025b). Quantitative talent identification reimagined: sequential testing reduces decision uncertainty. Appl. Sci. 15, 9707. 10.3390/app15179707 [ DOI ] [ Google Scholar ] Wood, S. N. (2011). Fast stable restricted maximum likelihood and marginal likelihood estimation of semiparametric generalized linear models. J R Stat Soc Ser B Stat Methodol 73, 3-36. 10.1111/j.1467-9868.2010.00749.x [ DOI ] [ Google Scholar ] Wood, S. N. (2017). Generalized Additive Models: An Introduction with R , 2nd Edn: Chapman and Hall/CRC. [ Google Scholar ] Associated Data This section collects any data citations, data availability statements, or supplementary materials included in this article. Supplementary Materials Supplementary information jexbio-229-251395-s1.pdf (1,020.4KB, pdf) DOI: 10.1242/jexbio.251395_sup1 Articles from The Journal of Experimental Biology are provided here courtesy of Company of Biologists ACTIONS View on publisher site PDF (467.3 KB) Cite Collections Permalink PERMALINK Copy RESOURCES Similar articles Cited by other articles Links to NCBI Databases Cite Copy Download .nbib .nbib Format: AMA APA MLA NLM Add to Collections Create a new collection Add to an existing collection Name your collection * Choose a collection Unable to load your collection due to an error Please try again Add Cancel Follow NCBI NCBI on X (formerly known as Twitter) NCBI on Facebook NCBI on LinkedIn NCBI on GitHub NCBI RSS feed Connect with NLM NLM on X (formerly known as Twitter) NLM on Facebook NLM on YouTube National Library of Medicine 8600 Rockville Pike Bethesda, MD 20894 Web Policies FOIA HHS Vulnerability Disclosure Help Accessibility Careers NLM NIH HHS USA.gov Back to Top