Arnold H. Taylor

Extraction of a weak climatic signal by an ecosystem

The complexity of ecosystems can cause subtle and chaotic responses to changes in external forcing. Although ecosystems may not normally behave chaotically, sensitivity to external influences associated with nonlinearity can lead to amplification of climatic signals. Strong correlations between an El Niño index and rainfall and maize yield in Zimbabwe have been demonstrated; the correlation with maize yield was stronger than that with rainfall. A second example is the 100,000-year ice-age cycle, which may arise from a weak cycle in radiation through its influence on the concentration of atmospheric CO2 (ref. 5). Such integration of a weak climatic signal has yet to be demonstrated in a realistic theoretical system. Here we use a particular climatic phenomenon—the observed association between plankton populations around the UK and the position of the Gulf Stream—as a probe to demonstrate how a detailed marine ecosystem model extracts a weak signal that is spread across different meteorological variables. Biological systems may therefore respond to climatic signals other than those that dominate the driving variables.

A modelling investigation of the role of phytoplankton in the balance of carbon at the surface of the North Atlantic

A simple model is used to examine the way phytoplankton growth influences carbon dioxide concentration in the surface waters of the ocean. The model consists of a one-dimensional system of two layers, a mixed layer and a thermocline, in which there is a single population of phytoplankton dependent on a single nutrient (represented by nitrate). Beneath the thermocline is a deep layer containing constant nutrient concentration, TCO2, and alkalinity, but almost no phytoplankton. The mixed layer rises and falls seasonally, and the phytoplankton population increases or declines in response to the annual cycle of light and temperature; a diurnal cycle of irradiance is also included. The model calculates the changes in the total inorganic carbon content and the alkalinity resulting from phytoplankton growth and death and hence determines the annual cycle of pCO2. Although simple, the model reproduces the major features of the annual distributions of phytoplankton and pCO2 in the North Atlantic. A spring bloom occurs at nonequatorial latitudes, the bloom appearing progressively later as the site is moved northward. South of 60° a deep-chlorophyll maximum is present during the summer. At the equator it extends throughout the year and is accompanied by nutrient depletion within the pycnocline. Predicted annual rates of primary production are in agreement with those measured at ocean weather station India (59°N) and at Bermuda (32°N). The model results show the sharp drawdown of CO2 that is associated with the spring bloom. Phytoplankton have the greatest influence on pCO2 at the highest latitudes, the annual cycle of pCO2 at 35°N being dominated by the seasonal changes in temperature. The levels of pCO2 predicted are similar to Geochemical Ocean Sections Study values. The model reproduces the diurnal cycle observed during the U.K. Biogeochemical Ocean Flux Study (BOFS) Lagrangian experiment in 1989. Comparisons of temporal series of chlorophyll-a and pCO2 obtained from the model with values observed during areal surveys carried out at 47° and 60°N during the BOFS cruise indicate that some of the spatial variability and the differences between latitudes may result from differences in timing of up to 10 days. The model reveals the seasonal cycles of the vertical fluxes of carbon, especially the pronounced effect of the spring bloom. The downward flux of particulate carbon exceeds the upward flux of dissolved inorganic carbon at 60° and 47°N but not at the equator. When a northward transport of water was included in the model, the spring bloom was brought forward because of seeding from the south and was terminated sooner when the water arriving became depleted in nutrients. The most significant feature of the presence of advection was the increased upward flux of dissolved inorganic carbon caused by the sharper vertical gradient of TCO2. Experiments were performed in which the summer phytoplankton were replaced by coccolithophores. The partial pressure of dissolved CO2 was increased, reducing the influx from the atmosphere. Therefore the extra particulate carbon sinking out of the surface waters as coccoliths was supplied by an increased upward flux of dissolved inorganic carbon from the deep ocean.

Gulf Stream shifts following ENSO events

Over the past three decades the annual mean latitude of the Gulf Stream off the coast of the United States has been forecastable from the intensity of the North Atlantic Oscillation (NAO), the predictions accounting for more than half the variance. Here we show that much of the unexplained variance can be accounted for by the Southern Oscillation in the Pacific, the Gulf Stream being displaced northwards following El Niño-Southern Oscillation (ENSO) events. This provides a link between events in the equatorial Pacific and the circulation and weather conditions of the North Atlantic.

North-South shifts of the Gulf Stream : Ocean-atmosphere interactions in the North Atlantic

Year-to-year changes in the latitude of the north wall of the Gulf Stream are very similar to those seen in the abundances of zooplankton observed by the Continuous Plankton Recorder Survey around the British Isles and also to those in the abundance of zooplankton in Lake Windermere. These connections must reflect changing weather patterns across the North Atlantic. The index of Gulf Stream position was constructed from the north wall data by principal components analysis. The first principal component, the index used, has eigenvector coefficients that all have the same sign, and is a measure of the latitude of the whole of the north wall. However, the component may represent the occurrences of meanders that are extensive in space and time rather than displacements of the Gulf Stream as a whole. This principal component has been used to calculate weighted averages of monthly mean sea-level pressure and of monthly mean numbers of cyclone tracks in order to show the changes in weather patterns associated with displacements of the north wall. Northward displacements of the north wall were accompanied by significantly reduced cyclone numbers in the northernmost regions of the Atlantic (annually and in the autumn) and, in spring, summer, and autumn, a region of reduced atmospheric pressure in the central Atlantic area 40°–60°N, 30°–50°W (locally significant). There was some tendency (not significant) for storm tracks to be deflected around the south side of this region. The pattern in winter is less clear and shows no statistical significance. Changes in the vicinity of the British Isles were generally too small to be statistically significant but were generally consistent with a lower frequency of storms in spring and autumn. As the biological changes appear to be caused by variations in the onset of thermal stratification during the spring they may be the result of relatively small changes in the atmospheric forcing. The atmospheric changes show no indications of the sources of the Gulf Stream displacements, the anomaly winds opposing the displacements. This may be because meanders of the Gulf Stream are not simply related to any single atmospheric variable. The clearest and most statistically significant meteorological signals were all well downstream from the north wall. Although the displacements of the north wall are caused by changing weather patterns over the North Atlantic, the Gulf Stream is also a region of strong heat transfers from the ocean to the atmosphere. Sawyers criterion indicates that distortions of this heat source could cause noticeable disturbances to the atmospheric circulation over the North Atlantic. A numerical model based on the analytical model of Smagorinsky is used to investigate the perturbations of the zonal circulation that might be caused by displacements of this local heat source. The predictions are in agreement with the changes seen in the central Atlantic during summer, spring, and autumn (but not those during winter). In the region where the model predicts atmospheric pressure reductions should occur, there are no positive correlation coefficients between the position of the north wall and the surface atmospheric pressure but a significant excess of negative correlation coefficients compared with chance, and northward shifts of the Gulf Stream were accompanied by significant reductions in atmospheric pressure. It is therefore possible that displacements of the north wall could influence weather patterns further east. The model predicts that any changes over the European continental shelf will be weak. An accurate description of the dynamics of the Gulf Stream may be an important requirement of coupled ocean–atmosphere models.

Estimating pCO2 from sea surface temperatures in the Atlantic gyres

A simple one-dimensional model, validated with observations from ship of opportunity programs, was run at different locations in the North and South Atlantic gyres to produce seasonal partial pressure of CO2 (pCO2)–sea surface temperature (SST) relationships. The pCO2–SST relationships obtained at different locations in the North Atlantic gyre can be approximated by two regression lines, one from February to July and another from August to January. An algorithm including SST, latitude, longitude and atmospheric pCO2 was constructed for each period. The robustness of these relationships was tested along several transects in the North Atlantic gyre and found to be in good agreement with the observations. The same approach was used in the South Atlantic gyre, but more observations are required in this region. In both gyres, the pCO2–SST relationships are close to 4%/1C, which is higher than the pCO2– SST relationships deduced from a CO2 climatology. r 2002 Elsevier Science Ltd. All rights reserved.

The influence of the spring phytoplankton bloom on carbon dioxide and oxygen concentrations in the surface waters of the northeast Atlantic during 1989

Abstract By representing growth, decay and vertical mixing during the spring phytoplankton bloom as a series of rate constants, a simple model is constructed that predicts phytoplankton abundance, carbon dioxide concentration and oxygen saturation as continuous functions of latitude and time. The predictions are compared with surface distributions of the partial pressure of carbon dioxide (pCO2), total dissolved inorganic carbon (TCO2), chlorophyll and oxygen saturation mapped during May 1989 on a series of surveys between 47 and 60°N in the eastern North Atlantic near 20°W. Altoough the observations were strongly variable on spatial scales of less than 100 km, the systematic changes revealed in the transects are quantitatively described by the theoretical expressions. Total vertical fluxes of carbon can be calculated from the model, and these can be integrated temporally and spatially. During the course of the spring bloom approximately 5 g m−2 of carbon entered the ocean surface from the atmosphere in the northeast Atlantic and the potential net loss of carbon to the deep ocean was about 16 g m−2.

Diurnal variations of convective mixing and the spring bloom of phytoplankton

Abstract Turbulent stirring of the surface mixed layer extends to shallower depths during the day than in the night because the increased buoyancy resulting from solar heating inhibits the mixing associated with wind action and surface heat loss. Calculations using a simple model in which a mixed layer of constant depth is more strongly coupled to deeper layers at night than during the day indicate that the reduction of mixing by day may be critical to the onset of the spring phytoplankton bloom, for this occurs at a time when there is increasing solar warming in the day and yet considerable heat loss and wind stirring at night. If the Kraus-Turner model of the mixed layer is used to estimate the mixing rates occurring during darkness, the values obtained agree with those that give realistic simulations of the spring bloom. Diurnal observations of chlorophyll a , p CO 2 and oxygen saturation made at 60°N during the Lagrangian experiment carried out in 1989 as part of the U.K. Biogeochemical Ocean Flux Study can be modelled more successfully if the day-night changes in vertical mixing are included in the same manner as the single layer model. The calculations indicate that these changes may shift the timing of the bloom by about 1 week and may account for the depth of penetration of some spring blooms. This process needs to be considered when modelling the coupling between climate and phytoplankton.

Latitudinal and seasonal variations in carbon dioxide and oxygen in the northeast Atlantic and the effects on Emiliania huxleyi and other phytoplankton

Several cruise programs, such as Transient Tracers in the Ocean (TTO), the Biogeochemical Ocean Flux Study (BOFS), and the Joint Global Ocean Flux Study (JGOFS), have measured physical, chemical, and biological variables in the northeast Atlantic, with the aim of understanding the seasonal variation in oceanic components such as carbon dioxide and oxygen and the role of different factors (e.g., the marine biota) in determining these seasonal variations. For this paper, data from the different cruise programs have been collated, and then three-dimensional (3-D) interpolated surfaces have been plotted in order to illustrate the seasonal and latitudinal changes in these different components. These data-generated plots are then compared with plots generated from the results of the simulation model of Taylor et al. (1991). The model-derived and data-derived plots are shown to be similar, and the model is subsequently used to explain the interaction of the processes underlying the observed variation in pCO2. It is argued that more data needs to be collected at the time of the spring bloom north of 50°N. The plot of carbon dioxide concentration is discussed in relation to the effects of carbon dioxide concentration on phytoplankton growth, as proposed by Riebesell et al.(1993). In addition, it is shown that the distribution of blooms of Emiliania huxleyi in the NE Atlantic (55°–63°N but not farther south) is not coincident with areas of low [CO2], contrary to the hypothesis that Emiliania huxleyi has evolved to be a successful competitor at low [C02].

A simple model of interannual displacements of the Gulf Stream

Interannual variations in the latitude of the north wall of the Gulf Stream 1966–1999 can be hindcast by a simple model forced solely by monthly values of the North Atlantic Oscillation (NAO) index. The model is a modified version of the one-dimensional model developed by D. Behringer et al. and utilizes the observed tendency for the centers of the Azores high and the Iceland low to move north or south as the NAO strengthens or weakens. Only one parameter, the active upper layer thickness, is changed from the original values. Although the model incorporates the thermal feedback on the wind stress used by Behringer et al., the results are almost unchanged without it. In generating the Gulf Stream position the model averages the NAO index over several months and introduces a time lag of at least a year. These delays are associated with the time taken for the lateral advection of heat. The model results are also compared with estimates of the latitude of the Gulf Stream made by A. Gangopadhyay et al. 1977–1988 and by T.M. Joyce et al. 1955–1997, and with the series of Gulf Stream transports 1954–1997 constructed by R.G. Curry and M.S. McCartney. The model reproduces the observed weakness of the seasonal variations in comparison to interannual variations. The NAO data are used to hindcast the annual latitude of the Gulf Stream for the period 1825–1999. The results show that since 1970 the Gulf Stream may have been more consistently farther south than during any period of the past 170 years.

1 North Atlantic climatic signals and the plankton of the European Continental Shelf

Abstract Climatic variability on the European Continental Shelf is dominated by events over the North Atlantic Ocean, and in particular by the North Atlantic Oscillation (NAO). The NAO is essentially a winter phenomenon, and its effects will be felt most strongly by populations for which winter conditions are critical. One example is the copepod Calanus finmarchicus , whose northern North Sea populations overwinter at depth in the North Atlantic. Its annual abundance in this region is strongly dependent on water transports at the end of the winter, and hence on the NAO index. Variations in the NAO give rise to changes in the circulation of the North Atlantic Ocean, with additional perturbations arising from El Nino - Southern Oscillation (ENSO) events in the Pacific, and these changes can be delayed by several years because of the adjustment time of the ocean circulation. One measure of the circulation is the latitude of the north wall of the Gulf Stream (GSNW index). Interannual variations in the plankton of the Shelf Seas show strong correlations with the fluctuations of the GSNW index, which are the result of Atlantic-wide atmospheric processes. These associations imply that the interannual variations are climatically induced rather than due to natural fluctuations of the marine ecosystem, and that the zooplankton populations have not been significantly affected by anthropogenic processes such as nutrient enrichment or fishing pressure. While the GSNW index represents a response to atmospheric changes over two or more years, the zooplankton populations correlated with it have generation times of a few weeks. The simplest explanation for the associations between the zooplankton and the GSNW index is that the plankton are responding to weather patterns propagating downstream from the Gulf Stream system. It seems that these meteorological processes operate in the spring. Although it has been suggested that there was a regime shift in the North Sea in the late 1980s, examination of the time-series by the cumulative sum (CUSUM) technique shows that any changes in the zooplankton of the central and northern North Sea are consistent with the background climatic variability. The abundance of total copepods increased during this period but this change does not represent a dramatic change in ecosystem processes. It is possible some change may have occurred at the end of the time-series in the years 1997 and 1998.