The reduction of plankton biomass induced by mesoscale stirring: A modeling study in the Benguela upwelling

Highlights

• We have studied the plankton dynamics in the Benguela upwelling region.

• We consider a simple NPZ model coupled with different velocity fields.

• There is a reduction of biological production when lateral stirring is increased.

• 2D offshore advection seems to be the dominant process involved.

Abstract

Recent studies, both based on remote sensed data and coupled models, showed a reduction of biological productivity due to vigorous horizontal stirring in upwelling areas. In order to better understand this phenomenon, we consider a system of oceanic flow from the Benguela area coupled with a simple biogeochemical model of Nutrient-Phyto-Zooplankton (NPZ) type. For the flow three different surface velocity fields are considered: one derived from satellite altimetry data, and the other two from a regional numerical model at two different spatial resolutions. We compute horizontal particle dispersion in terms of Lyapunov exponents, and analyzed their correlations with phytoplankton concentrations. Our modeling approach confirms that in the south Benguela there is a reduction of biological activity when stirring is increased. Two-dimensional offshore advection and latitudinal difference in primary production, also mediated by the flow, seem to be the dominant processes involved. We estimate that mesoscale processes are responsible for 30–50% of the offshore fluxes of biological tracers. In the northern area, other factors not taken into account in our simulation are influencing the ecosystem. We suggest explanations for these results in the context of studies performed in other eastern boundary upwelling areas.

Introduction

Marine ecosystems of the eastern boundary upwelling zones are well-known for their major contribution to the world ocean productivity. They are characterized by wind-driven upwelling of cold nutrient-rich waters along the coast that supports elevated plankton and pelagic fish production (Mackas et al., 2006). Variability is introduced by strong advection along the shore, physical forcings by local and large scales winds, and high submeso- and mesoscale activities over the continental shelf and beyond, linking the coastal domain with the open ocean.

The Benguela Upwelling System (BUS) is one of the four major Eastern Boundary Upwelling Systems (EBUS) of the world. The coastal area of the Benguela ecosystem extends from southern Angola (around $17°\mathrm{S}$) along the west coast of Namibia and South Africa ($36°\mathrm{S}$). It is surrounded by two boundary currents, the warm Angola current in the north and the temperate Agulhas current in the south. The BUS can itself be subdivided into two subdomains by the powerful Luderitz upwelling cell (Hutchings et al., 2009). Most of the biogeochemical activity occur within the upwelling front and the coast, although it can be extended further offshore toward the open ocean by the numerous filamental structures developed in offshore (Monteiro, 2009). In the BUS, as in the other major upwelling areas, high mesoscale activity due to eddies and filaments impacts strongly a marine planktonic ecosystem over the continental shelf and beyond (Brink and Cowles, 1991, Martin, 2003, Sandulescu et al., 2008, Rossi et al., 2009).

The purpose of this study is to analyze the impact of horizontal stirring on phytoplankton dynamics in the BUS within an idealized two dimensional modeling framework. Based on satellite data of the ocean surface, Rossi et al., 2008, Rossi et al., 2009 recently suggested that mesoscale activity has a negative effect on chlorophyll standing stocks in the four EBUS. This was obtained by correlating remote sensed chlorophyll data with a Lagrangian measurement of lateral stirring in the surface ocean (see Methods section). This result was unexpected since mesoscale physical structures, particularly mesoscale eddies, have been related to higher planktonic production and stocks in the open ocean (McGillicuddy et al., 2007) as well as off a major EBUS (Correa-Ramirez et al., 2007). A more recent and thorough study performed by Gruber et al. (2011) in the California and the Canary current systems extended the initial results from Rossi et al., 2008, Rossi et al., 2009. Based on satellite derived estimates of net primary production, of upwelling strength and of Eddy Kinetic Energy (EKE) as a measure of the intensity of mesoscale activity, they confirmed the suppressive effect of mesoscale structures on biological production in upwelling areas. Investigating the mechanism behind this observation by means of on 3D eddy-resolving coupled models, Gruber et al. (2011) showed that mesoscale eddies tend to export offshore and downward a certain pool of nutrients not being effectively used by the biology in the coastal areas. The process they called “nutrients leakage” is also having a negative feedback by diminishing the pool of deep nutrients available in the surface waters being re-upwelled continuously.

In our work, we focused on the Benguela area, being the most contrasting area of all EBUS in terms of stirring intensity (Rossi et al., 2009). Although the mechanisms studied by Gruber et al. (2011) seem to involve 3D dynamics, the initial observation of this suppressive effect was essentially based on two-dimensional (2D) datasets (Rossi et al., 2008). In this work we use 2D numerical analysis in a semi-realistic framework to better understand the effects of a 2D turbulent flow on biological dynamics, apart from the complex 3D bio-physical processes. The choice of this simple horizontal numerical approach is indeed supported by other theoretical 2D studies that also displayed a negative correlation between stirring and biomass (Tél et al., 2005, MacKiver and Neufeld, 2009, Neufeld and Hernández-García, 2009). Meanwhile, since biological productivity in upwelling areas relies on the (wind-driven) vertical uplift of nutrients, we introduced in our model a nutrient source term with an intensity and spatial distribution corresponding to the upwelling characteristics. Instead of the commonly used EKE, which is an Eulerian diagnostic tool, we used here a Lagrangian measurement of mesoscale stirring that has been demonstrated as a powerful tool to study patchy chlorophyll distributions influenced by dynamical structures at mesoscale, such as upwelling filaments (Calil and Richards, 2010). The Lagrangian perspective provides a complementary insight to transport phenomena in the ocean with respect to the Eulerian one. In particular, the concept of Lagrangian Coherent Structure may give a global idea of transport in a given area, separating regions with different dynamical behavior, and signaling avenues and barriers to transport, which are of great relevance for the marine biological dynamics. While the Eulerian approach describes the characteristics of the velocity field, the Lagrangian one addresses the effects of this field on transported substances, which is clearly more directly related to the biological dynamics. For example the work by Hernández-Carrasco et al. (2012) describes currents in the world ocean having the same level of Eddy Kinetic Energy but having two different stirring characteristics, as quantified by Lagrangian tools. Further discussions comparing Lagrangian and Eulerian diagnostics can be found, for example, in d'Ovidio et al. (2009) and the above cited Hernández-Carrasco et al. (2012). To consider velocity fields with different characteristics and to test the effect of the spatial resolution, different flow fields are used, one derived from satellite and two produced by numerical simulations at two different spatial resolutions. Our modeled chlorophyll a concentrations are compared with observed distributions of chlorophyll a (a metric for phytoplankton) obtained from the SeaWiFS satellite sensor.

This paper is organized as follows. Section 2 is a brief description of the different datasets used in this study. Section 3 depicts the methodology, which includes the computation of the finite-size Lyapunov exponents, and the numerical plankton-flow 2D coupled model. Then, our results are analyzed and discussed in Section 4. Finally in Section 5, we summed-up our main findings.