Laura G. Peteiro

Evaluation of self‐thinning models and estimation methods in multilayered sessile animal populations

Self-thinning (ST) models have been widely used in the last decades to describe population dynamics under intraspecific competition in plant and animal communities. Nevertheless, their applicability in animal populations is subjected to the appropriate inclusion of space occupancy and energy requirements. Specifically, the disposition of gregarious sessile animals in complex matrices hampers the application of classical ST models. This paper reviews the self-thinning models, regression methods (central tendency and frontier techniques) and discrimination criteria currently applied for gregarious sessile species through application to the analysis of mussel populations (Mytilus galloprovincialis) grown in suspended culture. In addition, we propose to model the temporal evolution of site occupancy in the stochastic frontier function (SFF). Our results confirm that the number of layers should be included in the classical bidimensional ST model for the analysis of multilayered populations. The estimated parameters obtained by the different fitting techniques depended on the measurement method of the variables in the model. This, together with the proximity between the space and food self-thinning theoretical exponents (SST and FST, respectively) highlights the difficulty in discriminating the competition limiting factor (space/food) from the self-thinning exponent. On the other hand, the SFF provided congruent results for biomass and individual mass analysis, in contrast to the lack of robustness observed for the central tendency regression methods. Furthermore, the SFF approach allowed a dynamic interpretation of the ST process providing insight into the temporal evolution of site occupancy. These results highlight the suitability of the stochastic frontier approach in the analysis of self-thinning dynamics in sessile animal populations.

Independent estimates of marine population connectivity are more concordant when accounting for uncertainties in larval origins

Marine larval dispersal is a complex biophysical process that depends on the effects of species biology and oceanography, leading to logistical difficulties in estimating connectivity among populations of marine animals with biphasic life cycles. To address this challenge, the application of multiple methodological approaches has been advocated, in order to increase confidence in estimates of population connectivity. However, studies seldom account for sources of uncertainty associated with each method, which undermines a direct comparative approach. In the present study we explicitly account for the statistical uncertainty in observed connectivity matrices derived from elemental chemistry of larval mussel shells, and compare these to predictions from a biophysical model of dispersal. To do this we manipulate the observed connectivity matrix by applying different confidence levels to the assignment of recruits to source populations, while concurrently modelling the intrinsic misclassification rate of larvae to known sources. We demonstrate that the correlation between the observed and modelled matrices increases as the number of observed recruits classified as unknowns approximates the observed larval misclassification rate. Using this approach, we show that unprecedented levels of concordance in connectivity estimates (r = 0.96) can be achieved, and at spatial scales (20–40 km) that are ecologically relevant.