Continuous plankton records stand the test of time: evaluation of flow rates, clogging and the continuity of the CPR time-series

Transcription

1 JOURNAL OF PLANKTON RESEARCH VOLUME 24 NUMBER 9 PAGES Continuous plankton records stand the test of time: evaluation of flow rates, clogging and the continuity of the CPR time-series EURGAIN H. JOHN, SONIA D. BATTEN, DARREN STEVENS, ANTHONY W. WALNE, TANYA JONAS AND GRAEME C. HAYS * SCHOOL OF BIOLOGICAL SCIENCES, UNIVERSITY OF WALES SWANSEA, SINGLETON PARK, SWANSEA SA2 8PP AND SIR ALISTER HARDY FOUNDATION FOR OCEAN SCIENCE, THE LABORATORY, CITADEL HILL, PLYMOUTH PL 2PB, UK * CORRESPONDING AUTHOR: g.hays@swan.ac.uk The Continuous Plankton Recorder (CPR) survey is one of the most extensive biological time-series in existence and has been in operation over major regions of the North Atlantic since 932. However, there is little information about the volume of water filtered through each sample, but rather a general assumption has persisted that each sample represents 3 m 3. Data from electromagnetic flowmeters, deployed on CPRs between 995 and 998, was examined. The mean volume filtered through samples was 3. m 3 and the effect of clogging on filtration efficiencies was not great. Consequently, even when the likely variations in flow due to clogging are taken into account, previously identified links between zooplankton abundance and climatic signals remain strong. INTRODUCTION Long term ecological time-series are becoming increasingly important research datasets with the expansion of studies documenting how biological systems have responded to climate change (Colebrook, 986; Forchhammer et al., 998; Reid et al., 998a; Post and Stenseth, 999; Beaugrand et al., 2; Forchhammer and Post, 2). However, funding constraints mean there are relatively few such time-series that span many decades, particularly in marine systems. Without question, one of the most extensive marine biological time-series is provided by the Continuous Plankton Recorder (CPR) survey. This survey, which is currently managed by the Sir Alister Hardy Foundation for Ocean Science, was instigated by Alister Hardy in the 93s following his development of the CPR (Hardy, 939). The CPR is an instrument that can be towed at high speed from ships-ofopportunity (i.e. non-research ships) over extended transects. The fundamental design of the CPR, and methods of analysis for zooplankton and phytoplankton, have remained unchanged since 948 and 958, respectively (Colebrook and Robinson, 986), with over 4 taxa being identified and counted (Warner and Hays, 994), enabling long term changes in plankton populations to be studied. As the CPR time-series has lengthened, so the value of the data has increased. Beginning in the 98s, CPR data began to reveal systematic changes in the zooplankton communities of the NE Atlantic, with these changes being linked to annual weather indices and with parallel trends also occurring at other trophic levels (Colebrook, 978; Aebischer et al., 99; Dickson et al., 988; Reid et al., 998b). More recently, long term trends in the abundance of certain key zooplankton taxa have been shown to be strongly correlated with the North Atlantic Oscillation, a widely used index for annual weather patterns in Northern Europe (Fromentin and Planque, 996; Planque and Taylor, 998). Despite this wide use of the CPR data, doubt has remained surrounding the continuity of the time-series. While the abundance of plankton in the samples is determined using standard microscopic analysis, the volume of water filtered in the collection of these organisms has not been routinely measured (Walne et al., 998). Therefore, it has been assumed that a constant volume of water is filtered through each sample. However, it is well known that when plankton is abundant, the filtering meshes of nets Oxford University Press 22

2 JOURNAL OF PLANKTON RESEARCH VOLUME 24 NUMBER 9 PAGES can become clogged with material, which then reduces the subsequent flow rate (Brander et al., 993; Walne et al., 998). Hence if the extent of clogging has varied systematically throughout the CPR time-series, artefacts may have been introduced into the time-series. Recently, electromagnetic flowmeters have been developed which can measure the flow rate through CPRs (Walne et al., 998). Here we examine data from the first 4 years of these flowmeter deployments and quantify the relationship between the volume of water filtered per sample and the extent of clogging. We then use this relationship on historically collected samples to predict the respective volumes of water filtered in the collection of each sample, and in this way we objectively reconstruct the CPR time-series. Hence we establish the true continuity of the CPR time-series, and validate the trends in zooplankton abundance that have previously been identified. METHOD Background to CPR survey To date, over 2 samples have been analysed by the CPR survey, representing extensive spatial and temporal coverage. CPRs are towed at a depth of 6 7 m (Hays and Warner, 993). Water enters through an inlet aperture of.6 cm 2 and passes through a 27 µm silk filtering mesh [for a full description on how CPRs sample see (Warner and Hays, 994)]. Samples corresponding to nautical miles (8.5 km) of tow (equivalent to 3 m 3 of water filtered assuming % filtration efficiency) then undergo standard analysis: phytoplankton abundance is first estimated from a visual assessment of the coloration and accorded a greenness index (from to 6.5) by reference to standard colour charts, then the organisms collected on the silk are identified and counted (Colebrook and Robinson, 986; Warner and Hays, 994). The level of identification varies: some organisms are identified to species, others to genus or class only. Electromagnetic flowmeters The electronic flowmeter designed for the CPR is described by Walne et al. (Walne et al., 998). Flow rate (recorded in l min ) is stored on electronic loggers and then downloaded onto a computer on return of the CPR to the laboratory. Time and position of deployment and retrieval are recorded so that the volume of water filtered can be determined for each sample (Walne et al., 998). Preliminary screening of the raw data was carried out and tows with large variances in their flow rate (due to poor flowmeter performance) were discarded from analysis. The volume (m 3 ) filtered per sample was then calculated from the corresponding average flow rate. Fig.. Map of tow routes with operational flowmeters, , illustrating the three spatial scales used to subdivide the data. The four regions (divided by dotted lines) were separated according to oceanographic area: R, North Sea; R2, Northern Atlantic; R3, Central Atlantic; R4, Irish Sea. 942

3 E. H. JOHN ET AL. CONTINUOUS PLANKTON RECORDS RESULTS Assessment of flow data collected on operational CPR tows Flowmeters were deployed on CPRs between 995 and 999 on 5 routes covering several areas of the North Sea and North Atlantic (Figure ). In total, they were deployed on 286 tows, generating flow data for nearly 2 samples. The expected flow rate to produce a total filtered volume of 3 m 3 per sample is generally ~7 8 l min and many examples of flow rates that lay close to this range of values were recorded (Figure 2). Quantification of the relationship between the volume filtered and extent of clogging The mean volume filtered per sample was 3. m 3 (SD = ±.8 m 3, Figure 3). As an initial examination of the importance of clogging, we examined the mean volume filtered at different levels of mesh coloration, total phytoplankton numbers and total small (<2 mm) zooplankton Flow (L min - ) Time (h) Fig. 2. Two examples of raw flow rate data recorded on CPRs. These tows were on route R in the southern North Sea during December 997 (solid line) and September 998 (heavy solid line). Number of samples Volume filtered (m 3 ) Fig. 3. Frequency histogram of the recorded volumes filtered (m 3 ) per sample. Mean volume filtered (m 3 ) Mean volume filtered (m 3 ) Mean volume filtered (m 3 ) R 2 = R 2 =.86 Colour index Total phytoplankton (numbers per CPR sample).5 R 2 = Total traverse zooplankton (numbers per CPR sample) Fig. 4. The relationship between the mean volume filtered per sample (±SE bars) and sample colour (greenness index, a), total phytoplankton abundance (b) and for total traverse (i.e. small) zooplankton (c). numbers (Figure 4). The mean volume filtered decreased with increased coloration of the mesh and increased plankton numbers. However, even with the highest plankton numbers recorded, maximal clogging only decreased the volume filtered per sample by ~2%. It was readily apparent that the observed variation in volumes filtered per sample was relatively trivial when compared with seasonal differences in plankton numbers. We derived empirical relationships between the volume filtered and plankton numbers per sample over three spatial scales (shown in Figure ): first, for the individual tow routes, then for data grouped into four geographical a b c 943

4 JOURNAL OF PLANKTON RESEARCH VOLUME 24 NUMBER 9 PAGES areas which corresponded roughly to oceanographic areas, and finally for the entire dataset. When stepwise multiple regression analyses were performed, the amount of variation in the volume filtered per sample that could be explained by plankton numbers decreased from a maximum of 95% to a minimum of 2% as the spatial scale increased. To produce a single generic equation relating the volume filtered per sample to plankton numbers on the filtering meshes, we ran a stepwise multiple regression using the 9 phytoplankton and nine zooplankton taxa that had the greatest overall abundance, and therefore might be expected to have the most significant effect on flow. In addition, the remaining plankton taxa were grouped into general taxonomic categories (e.g. total centric diatoms, total dinoflagellates, total copepods) and also included in analyses. This analysis showed that in the entire dataset, 5% of the variation in the volume filtered per sample could be explained by the following six variables: numbers of total phytoplankton, Thalassiosira sp., Noctiluca scintillans, total centric diatoms, Nitzschia seriata and total Rhizosolenia sp. In order to verify that a linear relationship was the most appropriate type to use for this particular dataset, the analyses were repeated using a number of data transformations. However, these did not improve the goodness of fit. Use of empirically derived relationship to predict the volumes filtered for historically collected samples Using the various equations (one generic and four regional) relating the volume filtered to the extent of clogging, we recalculated plankton abundances for samples collected between 995 and 998, and then compared them with the absolute abundances derived from the measured flow as well as abundances calculated by assuming 3 m 3 were filtered per sample. The coefficients of determination (r 2 ) between these different abundance measurements indicated that while the assumption of 3 m 3 per sample is reasonably sound, using the volumes filtered predicted from the extent of clogging gave improved estimates of absolute abundance (Table I). This is exemplified by an r 2 of.99 between actual and recalculated abundance of Calanus finmarchicus (Figure 5). The generic equation to predict the volume filtered per sample was applied to the historical dataset (948 present, representing nearly 73 samples) and the resulting predicted plankton abundance values compared with those calculated when assuming that the volume filtered remained constant at 3 m 3. Annual means of C. finmarchicus, for , did not differ significantly (P >.5 in a paired sample t-test) when using a constant value or the predicted values for the volume filtered. The relationship between the predicted abundance of C. finmarchicus and the NAO index (Fromentin and Planque, 996) was strong for (Figure 6). DISCUSSION The importance of long term monitoring of biological systems to study and detect responses to environmental change has long been established. A prerequisite of long term studies is that data collection is carried out in a consistent manner to establish the continuity of such timeseries. However, in reality, the logistical difficulties involved often mean that consistency is not always achieved. For example, data from the California Cooperative Oceanic Fisheries Investigations (CalCOFI) have provided hydrographic measurements in the southern California bight since 949. However, methods for zooplankton sampling have been changed twice (Ohman and Smith, 995), potentially disrupting the continuity of the time-series. Under international programmes such as the Joint Global Ocean Flux Study ( JGOFS), which involves Table I: Coefficients of determination (r 2 ) between absolute plankton abundances and those calculated a,b,c Taxa Abundance a Abundance b Abundance c Total copepods Calanus finmarchicus Calanus helgolandicus Total zooplankton a Calculated using a constant volume filtered of 3 m 3 per sample. b Sample volumes predicted using generic equations. c Sample volumes predicted using regional regression equations. 944

5 E. H. JOHN ET AL. CONTINUOUS PLANKTON RECORDS Log recalculated abundance Log actual abundance Fig. 5. The relationship between the actual abundances [log (n + )] of C. finmarchicus (calculated using measured flow data) and (i) those calculated assuming a constant volume of 3 m 3 (, dotted trendline, R 2 =.94, n = 672) and (ii) those calculated using the generic relationship between clogging and volume filtered per sample (, solid trendline, R 2 =.98, n = 672). Log abundance NAO Index Fig. 6. Relationship between abundance [log (n + )] of C. finmarchicus (calculated using the generic relationship between clogging and volume filtered per sample) and the NAO Index for the period (R 2 =.48). The NAO index was calculated as the winter difference of normalized sea level pressure between Lisbon, Portugal and Stykkisholmur, Iceland (Hurrell, 995). over 2 countries with six time-series stations spanning three oceans, standardized protocols are followed (Knap et al., 996) to enable spatial and temporal inter-comparisons of biological and hydrographic measurements. To maintain the true continuity of the CPR survey, methods for the collection and analysis of samples have remained unchanged since 948, with the exception of a modification to the tail fin (between 975 and 986) for improving the stability of the instrument at high speeds (Hays and Warner, 993). One anomaly in the sampling protocol has been the absence of information on the volume of water passing through each sample, due to lack of flow measurements on CPRs. The importance of determining filtration efficiencies of plankton nets to enable the calculation of absolute plankton abundances has been recognized for decades (McQueen and Yan, 993). However, flowmeters have not been used on CPRs historically because meters that are small enough to fit into the narrow water tunnel of the instruments, and that are able to record and store data over long periods, were not available. It is only relatively recently that electromagnetic (EM) flowmeters have been developed for the CPR that do not impede the flow of water through the instrument, and are able to log results over extensive towing (Walne et al., 998). There were several clear findings from our results. First, the flow recorded by EM flowmeters during the period , over an extensive spatial scale, had a mean of 3. m 3, close to the previously assumed value of 3 m 3. Second, the effect of clogging is present (Figure 4) but the absolute effects on water flow are relatively small. There is very little information available regarding the clogging and filtration coefficients of high-speed plankton samplers (Hays, 994); however, samplers often filter more than the theoretical quantity of water that would be expected from the inlet area, leading to filtration efficiencies in excess of % (LeFèvre, 973). Two factors that may mitigate the extent of clogging in CPRs are the relatively small size of the inlet aperture and the movement of the mesh. Reduction in the size of the inlet aperture decreases the amount of water entering, which prevents clogging under normal working conditions, hence improving the reliability of quantitative data from samplers (LeFèvre, 973). Furthermore, as the CPR silk is continuously moving during towing, the effect of clogging is less than expected when compared with other samplers (e.g. U-Tow) (Hays et al., 998). Third, analysis of the dataset on three spatial scales indicated that it was possible to establish much stronger relationships between clogging and flow for individual tow routes compared with the entire dataset. This is presumably due to spatial variations in the plankton taxa sampled in different areas. For instance, centric diatoms and dinoflagellates were most abundant in the North Sea and Atlantic regions, whilst the diatom Thalassiosira sp. was more abundant in the North and South Atlantic, with the dinoflagellate N. scintillans accounting for nearly 2% of the total plankton in the Irish Sea samples. Finally, the variation between the volumes filtered per sample (coefficient of variation =.26) was relatively trivial compared with the variation in taxon abundance (e.g. C. finmarchicus coefficient of variation = 9.8) and hence the implications of variable flow rates for recalculating absolute plankton abundances were relatively minor, especially since CPR data are usually presented as monthly or annual means. As a consequence of these 945

6 JOURNAL OF PLANKTON RESEARCH VOLUME 24 NUMBER 9 PAGES relatively small variations in flow, the link between the NAO index and abundance of C. finmarchicus previously identified assuming a constant volume of 3 m 3 per sample (Fromentin and Planque, 996; Planque and Taylor, 998) was found to remain valid even when clogging-induced flow variations were taken into account. The future of the CPR survey as a flagship of planktonic time-series in northern Europe is optimistic, and as the time-series lengthens not only does it become easier to establish the cause of long term patterns are but the strength of the earlier data is also enhanced. Data collected is currently available from The Sir Alister Hardy Foundation of Ocean Science ( sahfos) for use by the scientific community. Any uncertainty over the robustness of CPR data, as far as clogging is concerned, can now be cleared, as the results presented here confirm the reliability of past datasets. For most practical purposes, the previous assumption that each sample represents 3 m 3 is likely to be sound. ACKNOWLEDGEMENTS This work was funded by the Natural Environment Research Council (UK) under its Marine Productivity Thematic Programme (contract GST/2/2753). REFERENCES Aebischer, N. J., Coulson, J. C. and Colebrook, J. M. (99) Parallel longterm trends across four marine trophic levels and weather. Nature, 347, Beaugrand, G., Ibanez, F. and Reid, P. C. (2) Spatial, seasonal and long-term fluctuations of plankton in relation to hydroclimatic features in the English Channel, Celtic Sea and Bay of Biscay. Mar. Ecol. Prog. Ser., 2, Brander, K. M., Milligan, S. P. and Nichols, J. H. (993) Flume tank experiments to estimate the volume filtered by high-speed plankton samplers and the assess the effect of net clogging. J. Plankton Res., 5, Colebrook, J. M. (978) Continuous Plankton Records: zooplankton and environment, north-east Atlantic and North Sea, Oceanol. Acta,, Colebrook, J. M. (986) Environmental influences on long-term variability in marine plankton. Hydrobiologia, 42, Colebrook, J. M. and Robinson, G. A. (986) The Continuous Plankton Recorder Survey. Tech.Ser.I.O.C., 3, Dickson, R. R., Kelly, P. M., Colebrook, J. M., Wooster, W. S. and Cushing, D. H. (988) North winds and production in the eastern north Atlantic. J. Plankton Res.,, Forchhammer, M. C. and Post, E. (2) Climatic signatures in ecology. Trends Ecol. Evol., 5, Forchhammer, M. C., Post, E. and Stenseth, N. C. (998) Breeding phenology and climate. Nature, 39, Fromentin, J.-M. and Planque, B. (996) Calanus and environment in the eastern North Atlantic. II. Influence of the North Atlantic Oscillation on C. finmarchicus and C. helgolandicus. Mar. Ecol. Prog. Ser., 34, 8. Hardy, A. C. (939) Ecological investigations with the Continuous Plankton Recorder: object, plan and methods. Hull Bull. Mar. Ecol.,, 57. Hays, G. C. (994) Mesh selection and filtration efficiency of the Continuous Plankton Recorder. J. Plankton Res., 6, Hays, G. C. and Warner, A. J. (993) Consistency of towing speed and sampling depth for the continuous plankton recorder. J. Mar. Biol. Assoc. UK, 73, Hays, G. C., Walne, A. W. and Quartley, C. P. (998) The U-Tow: a system for sampling mesozooplankton over extended spatial areas. J. Plankton Res., 2, Hurrell, J. W. (995) Decadal trends in the North Atlantic Oscillation: regional temperatures and precipitation. Science, 269, Knap, A., Michaels, A., Close, A., Ducklow, H. and Dickson, A. (eds) (996) Protocols for the Joint Global Ocean Flux Study (JGOFS) Core Measurements. JGOFS Report No. 9, Reprint of the IOC Manuals and Guides No. 29, UNESCO 994. LeFèvre, J. (973) Clogging and filtration coefficient in a high-speed plankton sampler. Mar. Biol., 2, McQueen, D. J. and Yan, N. D. (993) Metering filtration efficiency of freshwater zooplankton hauls: reminders from the past. J. Plankton Res., 5, Ohman, M. D. and Smith, P. E. (995) A comparison of zooplankton sampling methods in the CalCOFI time series. Cal. Coop. Ocean Fish., 36, Planque, B. and Taylor, A. (998) Long-term changes in zooplankton and the climate of the North Atlantic. ICES J. Mar. Sci., 55, Post, E. and Stenseth, N. C. (999) Climatic variability, plant phenology, and northern ungulates. Ecology, 8, Reid, P. C., Planque, B. and Edwards, M. (998a) Is observed variability in the long-term results of the CPR survey a response to climate change? Fish. Oceanogr., 7, Reid, P. C., Edwards, M., Hunt, H. G. and Warner, A. J. (998b) Phytoplankton change in the North Atlantic. Nature, 39, 546. Walne, A. W., Hays, G. C. and Adams, P. R. (998) Measuring the filtration efficiency of the Continuous Plankton Recorder. J. Plankton Res., 2, Warner, A. J. and Hays, G. C. (994) Sampling by the Continuous Plankton Recorder survey. Prog. Oceanogr., 34, Received on February 5, 2; accepted on December 6, 2 946