This study examined temporal patterns in collective employee turnover over a 75 month interval. Time series models were fit to subgroups of functional and dysfunctional turnover. Dysfunctional turnover was defined as voluntary separation among high and average performers and functional turnover was defined as voluntary separation of low performers. Results provided support for the hypothesis that temporal patterns of functional and dysfunctional turnover differ. Patterns among high and average performers were similar, such that employee turnover across several global regions increased during or near July. In contrast, employee turnover among low performers tended to spike during or soon after October. Forecast (prediction) accuracy of turnover differed across groups based on individual performance level. Specifically, turnover among low and average performers was forecast with greater accuracy than overall aggregated turnover or turnover among high performers, the latter being the most difficult to forecast. After time-dependent variation (autocorrelation) was removed from global turnover among high, average, and low performers, these series were cross-correlated with similarly cleaned organizational performance outcomes (i.e., net sales, operating income, diluted net earnings per share). Results from these analyses indicated that organizational performance had a lagged negative relationship with turnover among high performers. The dynamic nature of the turnover and performance variables examined underscores the importance of considering employee turnover as a continuous process. As such, employee turnover should be proactively managed over time.

Les écosystèmes procurent de nombreux services essentiels aux sociétés humaines à travers les effets positifs de la biodiversité des communautés d'espèces. Par conséquent, identifier le rôle des organismes vivants et les facteurs influençant la diversité de leurs fonctions (ou diversité fonctionnelle), est indispensable pour prédire efficacement l'évolution des écosystèmes soumis aux pressions locales et globales.Cette thèse s'articule donc autour de deux axes: (i) établir un cadre méthodologique pour décrire la structure fonctionnelle locale et régionale des communautés, et (ii) améliorer la connaissance de l'impact des poissons sur la dégradation de la matière organique.Dans un premier temps, nous avons consolidé les outils méthodologiques permettant (i) d'améliorer la fiabilité des dendrogrammes fonctionnels, (ii) l'étude comparative des principaux indices de diversité fonctionnelle à l'échelle locale, et (iii) le développement d'une nouvelle décomposition de la diversité fonctionnelle en composantes locale (α), régionale (γ) et turnover (β). Appliqué aux communautés ichthyologiques échantillonnés le long d'un gradient de salinité, dans le système lagunaire de Patos-Mirim (Brésil), ce socle méthodologique nous a permis de révéler une structure fonctionnelle stable le long du gradient, en dépit d'une forte variabilité en composition d'espèces, ces communautés étant structurées majoritairement par un filtre environnemental agissant sur les capacités de locomotion des poissons.Dans un second temps, nous avons étudié l'impact des communautés ichthyologiques sur le cycle des nutriments. Plus précisément, nous avons estimé le potentiel de dégradation de la matière organique de plusieurs espèces de poissons, en étudiant la diversité fonctionnelle et génétique de leur flore bactérienne intestinale. Nous avons montré que les communautés ichthyologiques pouvaient influencer le recyclage des nutriments de façon non négligeable en raison d'un important potentiel de dégradation commun à la plupart des espèces étudiées, ce potentiel étant peu affecté par la diversité génétique ou les facteurs environnementaux. / Ecosystems provide many services essential to Human societies through the positive effects of biodiversity exhibited by species communities. Therefore, identifying the role of living organisms and the factors influencing the diversity of their functions (i.e. functional diversity) is fundamental to accurately predict the evolution of ecosystems undergoing local and global pressures.This thesis is organized around two axes: (i) establishing a methodological framework to describe the functional structure of local and regional communities, and (ii) improving our knowledge of the impact of fish on the degradation of organic matter.First, we have consolidated the methodological tools through (i) the improvement of functional dendrograms reliability, (ii) the comparative study of the main indices estimating local functional diversity, and (iii) the development of a new decomposition of functional diversity into local (α) and regional (γ) components, and functional turnover (β). Applied to fish assemblages sampled along a salinity gradient in Patos-Mirim lagoons complex (Brazil), this methodological framework allowed us to reveal a steady functional structure, despite a high variability in species composition, these communities being primarily structured by environmental filtering acting on fish locomotion abilities.   In a second step, we studied the impact of fish communities on nutrient cycling. More specifically, we estimated degradation of organic matter potential of several fish species by studying the genetic and functional diversity of their intestinal bacterial flora. We showed that the fish community could significantly influence nutrient cycling through an important degradation potential, common to most species studied, which is weakly affected by genetic diversity or environmental factors.

Text in English / Research on employee turnover dates back many decades. However, this research traditionally focused on either understanding turnover intentions or the factors preceding turnover, such as job satisfaction. Despite the challenge in SA being huge and organisations struggling to keep their talent, the research on employee turnover is quite limited. Understanding the different variables that influence employee turnover and providing practical solutions on how to mitigate turnover would be valuable to many organisations. The aim of this research project was to understand the role of personality and organisational climate in employee turnover. In addition, a comprehensive model of employee turnover was developed and tested. The Five Factor Model was used to conceptualise personality, while (due to the limited nature of existing models) a new model was designed to conceptualise organisational climate. A sample of 1 536 people was drawn from a large retail organisation in SA, with 807 stayers and 729 leavers. Biographical, personality and organisational climate information was collected over a two-year period for both samples. Results of the study were mixed; there were no significant differences in the two samples regarding the big five personality factors, except when nine bipolar scales were used. On these scales, leavers were found to be more assertive, persuasive and optimistic than stayers. All five personality factors moderated HR policies & procedures in determining turnover. There were differences between the stayers and leavers samples with regard to age, gender, tenure and absenteeism. Organisational climate was a key determinant of whether people left or stayed and organisations had more than one climate. Personality, organisational climate and absenteeism accounted for 29% of turnover. The proposed model of employee turnover met most of the requirements of goof fit measures when using Structural Equation Modelling (SEM). / Psychology / D.Litt. et Phil.

Text in English / Research on employee turnover dates back many decades. However, this research traditionally focused on either understanding turnover intentions or the factors preceding turnover, such as job satisfaction. Despite the challenge in SA being huge and organisations struggling to keep their talent, the research on employee turnover is quite limited. Understanding the different variables that influence employee turnover and providing practical solutions on how to mitigate turnover would be valuable to many organisations. The aim of this research project was to understand the role of personality and organisational climate in employee turnover. In addition, a comprehensive model of employee turnover was developed and tested. The Five Factor Model was used to conceptualise personality, while (due to the limited nature of existing models) a new model was designed to conceptualise organisational climate. A sample of 1 536 people was drawn from a large retail organisation in SA, with 807 stayers and 729 leavers. Biographical, personality and organisational climate information was collected over a two-year period for both samples. Results of the study were mixed; there were no significant differences in the two samples regarding the big five personality factors, except when nine bipolar scales were used. On these scales, leavers were found to be more assertive, persuasive and optimistic than stayers. All five personality factors moderated HR policies & procedures in determining turnover. There were differences between the stayers and leavers samples with regard to age, gender, tenure and absenteeism. Organisational climate was a key determinant of whether people left or stayed and organisations had more than one climate. Personality, organisational climate and absenteeism accounted for 29% of turnover. The proposed model of employee turnover met most of the requirements of goof fit measures when using Structural Equation Modelling (SEM). / Psychology / D.Litt. et Phil.