Volume 169, Issue 3 p. 525-536
Free Access

Chlorophyll content and fluorescence responses cannot be used to gauge reliably phytoplankton biomass, nutrient status or growth rate

Mikaela Kruskopf

Mikaela Kruskopf

Institute of Environmental Sustainability, Wallace Building, University of Wales Swansea, Swansea SA2 8PP, UK

Search for more papers by this author
Kevin J. Flynn

Kevin J. Flynn

Institute of Environmental Sustainability, Wallace Building, University of Wales Swansea, Swansea SA2 8PP, UK

Search for more papers by this author
First published: 24 November 2005
Citations: 181
Author for correspondence: Kevin J. Flynn Tel: +44 1792 295726 Fax: +44 1792 295447 Email: [email protected]

Summary

  • To consider the relationship between chlorophyll a (Chl a) content and phytoplankton growth and nutrient status, four phytoplankton species were grown in nitrogen (N)-limited [and, for one species, phosphorus (P)-limited] culture and measurements were made of CNP biomass, in vivo and in vitro Chl a content, the ratio of variable to maximum fluorescence (FV/FM) and the performance index for photosynthesis, PIABS (a derivative of the O-J-I-P analysis of photosystem II functionality).

  • Interspecies differences plus the development of intraspecies differences during nutrient stress produced c. 10-fold variations in Chl : C. Estimates of C from in vivo Chl content were better than those from extracted Chl content, as the decline in Chl : C during nutrient stress was offset in part by increased Chl fluorescence.

  • F V/FM was not a robust indicator of nutrient status or relative growth rate. Responses of FV/FM in cells re-fed the limiting nutrient showed no consistent pattern with which to gauge nutrient status. PIABS showed some promise as an indicator of nutrient status and relative growth rate.

  • Chl a content and fluorescence parameters do not deserve the unquestioned status they usually enjoy as indicators of biomass and physiological status.

Introduction

The development of robust methods for the determination of the biomass, growth rate and nutrient status of phytoplankton is one of the ‘Holy Grails’ of limnological and oceanographic research. Because chlorophyll a (Chl a) is ubiquitous in phytoplankton, and methods for its measurement are highly sensitive, approaches based on analysis of chlorophyll content and photochemistry have found ready acceptance in field studies. Indeed, such has been the acceptance of these methods that in many reports the concentration of Chl a is de facto regarded as phytoplankton biomass. Recently, with the increasing ready availability of instruments designed to measure in vivo photochemistry, there has been an increase in the use of the ratio of variable to maximum fluorescence (FV/FM) as an indicator of algal nutrient status (Lippemeier et al., 2001, 2003; Young & Beardall, 2003a,b) and of the use of associated methods for estimation of primary production (Gilbert et al., 2000b). These instruments and methods are also being used to follow algal physiology in response to ultraviolet (UV) damage (Sobrino et al., 2004) and during trophic interactions (Juneau et al., 2003; Lurling & Verschoor, 2003).

The work described here originated as a comparison between two different types of instrument for monitoring algal nutrient status. While the pulse amplitude modulation (PAM) type of instrument and determinations of FV/FM dominate aquatic research (Antal et al., 2001; Fuchs et al., 2002; Kromkamp & Forster, 2003; Muller, 2004; Springer et al., 2005), in the terrestrial plant arena an alternative has been developed based around the O-J-I-P test inspired by Strasser and colleagues (Strasser & Strasser, 1995; Krüger et al., 1997; Strasser et al., 2000; Strauss et al., 2005). The latter approach requires a different type of instrument, capable of very high data collection rates (10 µs), to capture the initial kinetics of the Kautsky effect curve during a single high-intensity light flash of 1 or 2 s duration. The O-J-I-P analysis requires the measurement of the fluorescence signal at inflexion points (called O, J, I and P) of the Kautsky effect curve that are apparent when the curve is plotted on a log time axis. Standard PAM-type instruments do not record the fluorescence transients with sufficient resolution. As algal suspensions (especially in marine systems) are thin and the instruments capable of such rapid data collection have lacked photomultiplier tubes to provide the required sensitivity, there have hitherto been few applications of O-J-I-P analyses to algae (Strasser & Govindjee, 1992; Qiu et al., 2004). Here, we considered just one of the aggregate O-J-I-P outputs, the performance index PIABS, which is reported directly by the Hansatech Handy-Photosynthesis Efficiency Analyzer (PEA; Hansatech Instruments, Norfolk, UK) used in this study. PIABS is described further below; in essence it reports an index for the photosynthetic performance on a chlorophyll basis.

For methods of algal monitoring to be robust they should report unambiguous results when confronted with samples of unknown nutrient history and ideally of unknown phytoplankton composition. The aim of this work was to compare different methods of determining biomass, nutrient status and growth rate using chlorophyll-based methods. Four different algal species were grown to nutrient-limited low-growth-rate steady state and the real carbon (C) biomass, nutrient status and growth rates were compared with responses obtained from Chl-based methods.

Materials and Methods

General culture and growth conditions

The chlorophyte Dunaliella primalecta (axenic strain CCAP 11/34), the prymnesiophyte Emiliania huxleyi (axenic strain L; NIOZ, Texel, the Netherlands), the diatom Thalassiosira weissflogii (axenic strain CCAP 1085/1) and the dinoflagellate Scrippsiella trochoidea (CCAP 1134/5) were grown in stretch batch cultures (Page et al., 1999) at a dilution of 0.1 d−1 and a temperature of 18°C, with light supplied at 83 µmol m−2 s−1 photosynthetically active radiation (PAR) over the waveband 400–700 nm (in the centre of a water-filled flask; light sensor QSL-100; Ocean Systems, Biospherical Instruments Inc., San Diego, CA, USA) with a 12 h : 12 h light : dark cycle. The dilution rate was chosen to provide the required volume of growth medium for daily sampling from the culture volume employed, whilst at the same time giving a nutrient-limited final (quasi-steady-state) growth rate that was not so low as to lead to cell death in the stationary phase.

All media were based on sea water that had been aged for 2 months after lacing with T. weissflogii. This was then filtered (Whatman GF/C) and autoclaved (at 115°C followed by rapid cooling on ice to minimize precipitation) before aseptic addition of nutrient stocks via 0.22-µm Acrodisc filters (Pal-Gelman, Ann Arbor, MI, USA). Micronutrient additions were those of F/2 (Guillard & Lorenzen, 1972), with inclusion of silicon (Si) for T. weissflogii and HCO3 for E. huxleyi, or K medium (Keller et al., 1987) for S. trochoidea. The extra HCO3 was added for E. huxleyi as a safeguard against the possibility of this coccolithophorid exhausting the inorganic C supply, although the results of Clark & Flynn (2000) suggest that this is actually unlikely. Macronutrients were supplied to produce phosphorus (P)-replete cultures with 100 µm nitrate and 34 µm phosphate, or P-depleted (‘–P’) cultures for S. trochoidea experiments with 200 µm nitrate and 3.4 µm phosphate. Experiments were conducted in 5-l Erlenmeyer flasks, containing 2 l of culture, of which 200 ml was an inoculum at the commencement of each experimental cycle. Cultures were diluted (0.1 d−1) with fresh media daily, immediately after routine sampling at 1 h into the light period, compensating for the 200-ml volume of culture removed during sampling. Thus the culture volume was maintained at 2 l. At stabilization of cell numbers (i.e. growth rate ≈ dilution rate), 200 ml of cell suspension was used to inoculate the next cycle. The data presented came from at least two sequential experimental cycles.

Sampling procedure

Daily samples (200 ml) were collected aseptically for various measurements, as follows. Cell number and biovolume were measured on live cells in five replicates using an Elzone 282PC particle analyser (Particle Data, Europe, Luxembourg) with a 76-µm orifice, calibrated with latex beads. Occasionally cells were also counted and measured, after Lugol fixation in a haemocytometer by microscopy; close agreement was obtained between these methods, and therefore the Elzone was used for routine measurements. Samples for determination of cellular C and nitrogen (N) content were collected under low vacuum (< 100 mmHg) onto pre-ashed glassfibre filters (13-mm-diameter Gelman Type AE), and analysed using a PDZ Europa ANCA-GSL preparation module (PDZ Europa, Crewe, UK) coupled to a Europa 20–20 mass spectrometer, with isoleucine as the standard. Filtrate collected during filtration was kept frozen (−20°C) in polypropylene scintillation vials until analysis for nitrate and phosphate. Nitrate was analysed by cadmium (Cd) reduction to nitrite (Alpkem RFA/2 Cd coil; supplied by Advanced Medical Supplies Ltd, Basingstoke, UK) followed by the colorimetric (NED/SAD) assay of Strickland & Parsons (1972). Phosphate was determined using the molybdate reaction method (Strickland & Parsons, 1972). The cellular P : C ratio was estimated assuming that all nutrient-P removed from the medium entered particulate cell-P. In vitro (extracted) Chl concentrations were determined in dimethylformamide extracts (1 h at 6°C) using a Cecil Instruments CE 6601 Scanning multimode UV spectrophotometer (Cecil Instruments, Cambridge, UK), and the formula of Strickland & Parsons (1972), as well as by in vivo fluorescence using a Turner Aquafluor fluorometer (model 8000-001; Turner, Sunnyvale, CA, USA). Chl a from spinach (Sigma) was used to validate these methods. In vivo Chl content was also determined using the total Chl function of the PAM instrument (see below).

Fluorescence analysis

Fluorescence analysis was undertaken using the Walz PHYTO-PAM-ED pulse amplitude modulated fluorescence system with Phyto-Win software version 1.45 (hereafter PAM; Heinz Walz GmbH, Effeltrich, Germany) and the Hansatech Handy-Photosynthesis Efficiency Analyzer (hereafter PEA; Hansatech Instruments). While the PAM contains a photomultiplier sensor giving high sensitivity, the PEA was originally designed for analysis of higher plant leaves. For the work conducted here, Hansatech provided a prototype PEA sensor with enhanced sensitivity when working with algal suspensions. Although both instruments report the derived parameter FV/FM, the method for measuring minimal fluorescence (F0) differs. The PEA computes F0 by extrapolation of the initial few microseconds of the rise of the Kautsky effect curve back to t0, while the PAM measures F0 as the fluorescence signal in the presence of a very low level of actinic illumination.

Fluorescence analyses were conducted on fresh samples taken during daily sampling, and also on samples that had been collected and held in separate vessels and spiked by addition of additional nutrients. For the latter, an extra 100 µm nitrate and 6.25 µm phosphate were added to whatever nutrient concentrations were present at the time of daily sampling; the fluorescence response was monitored at 2, 4, 6, 23 and 48 h after spiking, with the sample being held at the same temperature and irradiance conditions as used for the main culture.

All samples were dark-adapted for 15 min before measurement of the fluorescence response. The PAM sample was stirred during measurements. The PEA takes its measurements for just 2 s; the sample within the sealed darkened sample head was agitated by inversion of the sample chamber before analysis. For both instruments, the minimal (F0) and maximal (FM) fluorescence, and the potential maximum photosystem II (PSII) quantum yield [(FM − F0)/FM = FV/FM], were measured. With PAM, rapid light curves (RLC; so-called ‘short-term’ photosynthesis–irradiance curves) were measured yielding EK (light saturation parameter) and ETRmax (maximum electron transport rate) values. These were measured with light over the waveband 400–700 nm increasing from 16 to 764 µmol photons m−2 s−1, and then decreasing back down to 16 µmol photons m−2 s−1, with exposure at each photon flux density (PFD) of 20 s; this period was chosen after initial experiments indicated that exposure for this period allowed the fluorescence response to stabilize. The up-PFD and down-PFD curves were almost identical and the RLCs were fitted through all 20 data points. The total in vivo Chl fluorescence signal was also recorded from the PAM.

With the PEA, the Kautsky effect curve was captured and the performance index (PIABS) reported directly by the instrument. PIABS is described here using the same terminology as that used in Strauss et al. (2005), which should be consulted for further information.

image( Eqn 1)

The first term accounts for the contribution made by light absorption, with reference to the fraction of reaction centre Chl (ChlRC) per total Chl (ChlRC + Chlantenna); this is described by γ. The second term describes the contribution made by the light reactions; ϕPo is the maximum quantum yield of primary photochemistry. The last term describes the contribution made by the dark reaction; ψo is the efficiency of the dark redox reactions. Operationally, within the PEA software, the components of the equation for PIABS are derived as follows.

image( Eqn 2)

(F50µs, F300µs and F2ms, the fluorescence values at 50 µs, 300 µs and 2 ms after the dark-acclimated sample has been subjected to saturating illumination.)

Results

Figure 1 shows, for each of the P-replete experiments, changes in C-biomass, cell count, cellular N : C and P : C, Chl a, PIABS and FV/FM. Missing data points reflect instrument or other logistic failings, or (for PIABS) are explained in the next paragraph. These data show the expected growth dynamics and changes in elemental composition accompanying batch culture dynamics into a quasi-steady-state growth rate of 0.1 d−1 (i.e. the dilution rate).

Details are in the caption following the image

Changes in carbon (C) biomass, cell number, algal nitrogen : carbon (N : C) and nitrogen : phosphorus (N : P) ratios, extracted (in vitro) chlorophyll a (Chl a) content, the performance index (PIABS) and the ratio of variable to maximum fluorescence (FV/FM) measured by pulse amplitude modulation (PAM) and photosynthesis efficiency analyser (PEA) instruments. Measurements were made during stretch-batch growth of Dunaliella primalecta, Emiliania huxleyi, Thalassiosira weissflogii and Scrippsiella trochoidea in N-limited culture. PIABS values are given only when PAM and PEA FV/FM values are similar, indicating that sufficient Chl was present to allow the proper functioning of the PEA instrument.

The inadequacy of the PEA for analysis of low-Chl a samples is clear from a comparison of the FV/FM plots between PEA and PAM. We assumed that when these two estimates converged there was sufficient signal for the proper working of the PEA. Accordingly, PIABS values from the PEA are only shown where PEA and PAM values for FV/FM are similar. While all four organisms displayed broadly similar patterns of elevated FV/FM on recovery from N-stress and falling FV/FM on developing N-stress (as indicated by declining N : C), the magnitude of the change varied greatly. Only for E. huxleyi were changes in FV/FM pronounced. In contrast, PIABS values showed clear changes, declining with N-stress.

The P-depleted culture of S. trochoidea (Fig. 2) also showed no clear changes in FV/FM over the batch growth cycle (again disregarding the early PEA-derived values), although here PIABS also showed no depression as growth halted by nutrient (here P) limitation. Indeed, PIABS increased.

Details are in the caption following the image

Changes in carbon (C) biomass, cell number, algal nitrogen : carbon (N : C) and nitrogen : phosphorus (N : P) ratios, extracted (in vitro) chlorophyll a (Chl a) content, the performance index (PIABS) and the ratio of variable to maximum fluorescence (FV/FM) measured by pulse amplitude modulation (PAM) and photosynthesis efficiency analyser (PEA) instruments during stretch-batch growth of phosphorus (P)-limited Scrippsiella trochoidea. PIABS values are given only when PAM and PEA FV/FM values are similar, indicating that sufficient Chl was present to allow the proper functioning of the PEA instrument.

The RLCs generated by the PAM showed no change in α (the initial slope of the Chl-specific photosynthesis–irradiance (PE) curve) with declining nutrient status (Fig. 3; cf. 1, 2). All but D. primalecta, however, showed a decline in ETRmax of c. 40% over this period, with pro rata changes in EK (not shown). Values for EK were in the range 500–1200 µmol m−2 s−1. However, the curve-fitting algorithm supplied with the instrument did not appear to give optimal fits.

Details are in the caption following the image

Changes in the normalized values of the initial slope of the rapid light curve (RLC), α, and the plateau value, ETRmax (maximum electron transport rate). Measurements were made during the final batch growth cycles of nitrogen (N)-limited Dunaliella primalecta, Emiliania huxleyi, Thalassiosira weissflogii and Scrippsiella trochoidea (Fig. 1) and P-limited S. trochoidea (S. trochoidea-P; Fig. 2).

The Autofluor- and PAM-derived values for Chl a were well correlated (Fig. 4a). However, the in vivo Autofluor Chl a signal overestimated the in vitro Chl a content by an increasing amount as the cultures became nutrient-stressed (i.e. at high values of Chl ml−1; Fig. 4b). Only for T. weissflogii was this deviation minor.

Details are in the caption following the image

Relationships between in vivo chlorophyll a (Chl a) content determined by Turner Autofluor and pulse amplitude modulation (PAM) (a), and Autofluor and extracted (in vitro) Chl a content (b). All data from the experiments described in Fig. 1 (nitrogen-limited) and Fig. 2 (phosphorus-limited Scrippsiella trochoidea) were used. Deviation from the 1 : 1 line in (b) is associated with enhanced in vivo fluorescence in cells following nutrient exhaustion.

Figure 5a shows the correlation between C-biomass and Chl a concentration, not only within a species but also between species. Like most of the correlations (Table 1), the statistical significance of the correlation does not detract from the poor predictive power of the relationship. The deviation within species reflects the decline in Chl a content as nutrient stress develops (see also 1, 2, 4). The situation is better between C-biomass and in vivo (Autofluor) Chl (Fig. 5b), reflecting the increasing fluorescence signal per unit of Chl a seen in Fig. 4b. For each species, values of Chl : C varied by a factor of at least twofold, declining with N : C (Fig. 5c). However, there was an order of magnitude spread in values of Chl : C over the entire data set. In all of these instances, the higher Chl a content of the chlorophyte D. primalecta is evident; the data for the prymnesiophyte and dinoflagellate are most comparable in their spread and extent. Chl : C covaried with growth rate (Fig. 5d), although the spread across all species was again great and the relationship was nonsignificant for S. trochoidea-P (Table 1).

Details are in the caption following the image

Relationships between carbon (C) biomass and extracted (in vitro) chlorophyll a (Chl a) content (a), between C-biomass and in vivo (Autofluor) Chl a content (b), between algal nitrogen : carbon (N : C) and in vitro Chl : C (c), and between growth rate (µ) and in vitro Chl : C (d). All data from the experiments described in Fig. 1 (N-limited) and Fig. 2 (phosphorus-limited Scrippsiella trochoidea) were used.

Table 1. Correlations for data shown in 4-7
Figure Correlation Dunaliella primalecta Emiliania huxleyi Thalassiosira weissflogii Scrippsiella trochoidea S. trochoidea-P Total
4a Af Chl vs PAM Chl 0.94 0.98 0.98 0.96 0.98 0.96
4b Af Chl vs in vitro Chl 0.96 0.95 0.97 0.79 0.94 0.93
5a C vs in vitro Chl 0.95 0.92 0.85 0.77 0.89 0.61
5b C vs Af Chl 0.98 0.99 0.93 0.98 0.93 0.72
5c N : C vs Chl : C 0.48* 0.77 0.68 0.89 0.84 0.65
5d µ vs Chl : C 0.62 0.75 0.88 0.86 0.44 0.48
6a µnorm vs N : C 0.73 0.90 0.91 0.97 0.53 0.77
6b µnorm vs P : C 0.77 0.79 0.17
7a µnorm vs FV/FM 0.64 0.82 0.83 0.11 0.22 0.40
7b µnorm vs (in vitro/in vivo) Chl 0.36 0.67 0.74 0.85 0.88 0.61
7c µnorm vs PIABS 0.66* 0.82 0.91 0.68* −0.66* 0.62
  • All correlations are significant at 99% except those indicated with *, which are significant at 95%, or with †, which are nonsignificant.
  • Total excluding the S. trochoidea-P series.
  • Af, Turner Autofluor fluorometer; C, carbon; Chl, chlorophyll; FV/FM, the ratio of variable to maximum fluorescence; N, nitrogen; P, phosphorus; PAM, pulse amplitude modulation; PIABS, the performance index for photosynthesis; µnorm, normalized C-specific growth rate.

The data presented in 6, 7 are plotted against the C-specific growth rates normalized to the maximum value observed during the experiments (µnorm). The uppermost (extreme) data points for N : C (Fig. 6a) and P : C (Fig. 6a; S. trochoidea-P) vs µnorm describe quota-like relationships for each species, with a decline in growth rate associated with a decline in quota. There was also a decline in the P quota (P : C) in N-limited S. trochoidea (Fig. 6b).

Details are in the caption following the image

Relationships for Scrippsiella trochoidea between normalized carbon (C)-specific growth rate and cellular nitrogen : carbon (N : C) ratio (a), and cellular phosphorus : carbon (P : C) ratio (b). The original data are described in Fig. 1 (N-limited) and Fig. 2 (P-limited).

Details are in the caption following the image

Relationships between normalized carbon (C)-specific growth rate and the pulse amplitude modulation (PAM)-derived ratio of variable to maximum fluorescence (FV/FM) (a), the ratio of in vitro (extracted) to in vivo (Autofluor) chlorophyll a (Chl a) content (b), and the photosynthesis efficiency analyser (PEA)-derived performance index (PIABS) (c). All data from the experiments described in Fig. 1 (nitrogen-limited) and Fig. 2 (phosphorus-limited Scrippsiella trochoidea) were used.

Except for E. huxleyi, there was no strong relationship (indicated by a steep slope in the plots) between FV/FM and normalized growth rates (µnorm) (Fig. 7a), or between FV/FM and N : C (not shown, but compare 6, 7). There is thus no evidence that FV/FM could be used reliably to determine the nutrient status or relative growth rate of a mixed or poorly described phytoplankton population. The ratio of in vitro Chl to in vivo Chl (i.e. extracted : Autofluor-derived Chl a) varied with µnorm (Fig. 7b), except for D. primalecta (Table 1). The value of PIABS varied positively with µnorm (Fig. 7c), although the direction of the relationship between PIABS and µnorm was opposite (i.e. negatively) for P-limited S. trochoidea (Table 1). The overall spread of PIABS values would complicate an assignment of relative growth rate in a population of mixed composition.

There is no evidence, from Fig. 8, that the responses in FV/FM to nutrient spiking of cells could be used to determine nutrient status. For most samples (except for N-deprived D. primalecta) there was an initial depression of FV/FM on addition of nutrients; this was most obvious over the first 6 h after the spike. Changes in FV/FM after 24 h may reflect exhaustion of the nutrients in the spike, especially in the later samples, which would have been at a higher cell density and of lower nutrient status (Fig. 1).

Details are in the caption following the image

Changes in the pulse amplitude modulation (PAM)-derived ratio of variable to maximum fluorescence (FV/FM) after spiking with NO3[final additional concentration of 100 µm nitrogen (N)] plus PO43–[final concentration of 6.25 µm phosphorus (P)]. Measurements were made during the final batch growth cycles of N-limited Dunaliella primalecta, Emiliania huxleyi, Thalassiosira weissflogii and Scrippsiella trochoidea (Fig. 1) and P-limited S. trochoidea (S. trochoidea-P; Fig. 2). The initial value is indicated by an open circle; different line types indicate different series of incubations.

Discussion

Biomass determinations

The concept that Chl is algal biomass is so entrenched that it is not uncommon for parameters that would arguably be better expressed on a C-basis to be related to, or ‘truthed’ to, Chl rather than to real biomass (e.g. Llewellyn et al., 2005). The relationship between C-biomass and in vivo Chl fluorescence is better than that with in vitro (extracted) Chl (Fig. 5; Table 1), because the increased fluorescence per unit of Chl in nutrient-deprived cells (Fig. 4b) compensates for the decline in Chl : C coincident with nutrient stress. It is also possible that in nature declining Chl : C with increasing nutrient stress may be compensated by increasing Chl : C with photoacclimation in response to self-shading within developing blooms. However, this would still not counter species differences in the mean value of Chl : C.

The fact remains that Chl : C is highly variable and in ignorance of the conversion factor much effort is arguably misdirected, if not wasted, in measuring physiological parameters against Chl. This applies to field and laboratory studies both of the photoautotrophs and of associated predator–prey interactions. The problem, demonstrated here between species (Fig. 5) and with and without nutrient limitation (cf. Fig. 1), is all too obvious to ecosystem modellers who have to either assume a fixed ratio, or model changes in Chl : C with photoacclimation and nutrient status (Flynn, 2003). Photoacclimation would only exacerbate the problem (Anning et al., 2000) shown here with nutrient stress. Further, only the relationship between N-stress, light and Chl : C has been considered worthy of attention in models. The impact of P-stress has been ignored, reflecting the lack of experimental data upon which to base models. It would also be of value to explicitly model the interaction between physiological status and the in vivo Chl signal, if only empirically rather than mechanistically.

Nutrient status and growth rate determinations

While phytoplankton growth rates may be adversely affected by factors other than nutrient status (most obviously light and temperature), declining nutrient status invariably impacts on growth rate. Determining either or both of these features, growth rate and nutrient status, is of importance. Few studies of algal physiology report a sufficient range of data to make them of value for modelling. The data given here describe dynamic growth for four different species suitable for model testing and validation. Included are values for N : C and P : C; it is of interest that not only does the relationship between the C-quota for the limiting nutrient and growth rate describe the expected relationship (Fig. 6), but the quota for the nonlimiting nutrient does the same (Fig. 6b and see also Fig. 1). That is to say that, when either N or P is limiting, the C-specific cell content of the other element declines, although not in proportion; cellular N : P is not constant but varies by the best part of an order of magnitude. The implication is that multiple nutrient stresses do not interact via a threshold mechanism (as usually assumed) but through an interactive term (Flynn, 2001).

Not only does Chl give a poor index for C-biomass, but the value of indicators of photosynthetic activity (RLC parameters, Fig. 3; FV/FM, Fig. 6b) do not give robust indicators of nutrient status or relative growth rate. The decline in FV/FM with growth rate in T. weissflogii (Fig. 1) was similar to that reported by Kolber et al. (1988) and Parkhill et al. (2001) for Thalassiosira pseudonana. A similar (and more useful) trend was seen here for E. huxleyi (Fig. 7a). In D. primalecta, FV/FM increased until nitrate exhaustion, after which it declined (Fig. 1), while in S. trochoidea FV/FM remained stable (0.65–0.7) throughout the growth cycle, both under N (Fig. 1) and under P (Fig. 2) limitation. In consequence there were distinct differences between the species in the relationship between FV/FM and normalized growth rates (Fig. 7a) but, especially when these trends were considered in total, there was no useful relationship that could safely be exploited in an analysis of mixed or uncharacterized populations.

The maximum quantum yield of PSII (indicated by FV/FM) in phytoplankton has been widely accepted to be influenced by nutrient stress, and the depression of FV/FM has consequently been used as an indicator of nutrient stress or imbalance (Cleveland & Perry, 1987; Kolber et al., 1988; Lippemeier et al., 2001, 2003; Young & Beardall, 2003a, 2003b). However, contrasting results indicating high and constant FV/FM values have also been reported (Cullen et al., 1992; MacIntyre et al., 1997; Parkhill et al., 2001). Parkhill et al. (2001) suggested that FV/FM is not a good measure of nutrient limitation under balanced growth conditions and remains constant and independent of nutrient-limited growth rate under different irradiances. However, they did show that under nutrient-starved conditions FV/FM declines, reflecting the extreme degree of nutrient stress. In nature, the development of nutrient stress may be moderated by local nutrient regeneration and also by decreases in illumination (e.g. through self-shading) which effectively enhances relative if not absolute C-stress, thus countering nutrient (e.g. N or P) stress.

Our results clearly support the contention that FV/FM is not a robust indicator of nutrient stress and/or relative growth rate. It may work for some species (of those we tested, most obviously E. huxleyi) but for general screening it cannot be trusted. Any nutrient-specific responses (as seen by Lippemeier et al., 2001) may also be masked by species-specific differences. We also looked for a systematic pattern in the diel variation of FV/FM with changes in nutrient status (not shown). While we did not see such a pattern, the variation in FV/FM in samples taken at different times during the day was greater in nutrient-replete cells. Serodio et al. (2005) report short-term variation in the form of RLC which they ascribe at least in part to endogenous diel rhythms. They also comment on diel variations in FV/FM; while we observed the latter, all our RLC were obtained at the same time within the light : dark cycle. Villareal & Morton (2002) and Villareal (2004) also reported taxon-specific differences in diel changes in fluorescence signals. Diel variations would thus introduce even further uncertainty to the interpretation of FV/FM values in relation to the nutrient status of the cells in field samples.

Our attempts to test the usefulness of short-term spiking responses of the photosystems, as indicated by FV/FM (Fig. 8), also failed to indicate a consistent pattern. The concept behind this test was an extension of the traditional nutrient-spiking experiments in which changes in C-fixation are monitored. The hypothesis was that FV/FM would respond more rapidly and hence facilitate development of a short-term spiking incubation of an hour or so. The main response of FV/FM to nutrient re-feeding was a rapid decline (1–6 h after spiking) followed by a slower recovery over the following 48 h. The exception to this was D. primalecta, in which the spiking caused a different type of response at the beginning of the experiment compared with the last 6 d of each experimental cycle. During the first few days, the nutrient spike caused an increase in FV/FM, this response moderating and finally changing to a delayed decrease in FV/FM during the latter (nutrient-limited) phase of the each experiment cycle. Young & Beardall (2003b) demonstrated a significant decrease in FV/FM in Dunaliella tertiolecta in response to N starvation, with a rapid recovery of FV/FM after resupply of nitrate; this is similar to our results under zero or minimal nutrient stress.

Similar nutrient re-addition (spiking) studies have been conducted by Lippemeier et al. (2001), who recorded a strong decrease, rather than an increase, in the photochemical efficiency of light-adapted cells after phosphate and silicate re-addition, while a nitrate re-addition gave a strong delayed increase in fluorescence response. In our study, initially P-limited S. trochoidea showed a similar response to spiking as did N-limited cells. The change in the photochemical efficiency of the PSII reaction centres was observed 1 h after spiking, with the decrease continuing for 6 h after spiking on most occasions (Fig. 8). However, in the latter part of the batch growth cycle where P-stress was greater, recovery to pre-spiking FV/FM values was not so apparent as with N-limited cells. Thus, protracted P limitation seemed to have a longer lasting effect on the quantum efficiency of PSII than did N limitation. Lippemeier et al. (2003) showed a strong increase in efficiency in re-fed P-limited Alexandrium minutum cells over the 3 d following spiking, these changes in the photochemical efficiency being first observed c. 2 h after the P spike.

We can conclude that the short-term response of FV/FM to nutrient addition, like the simple immediate value of FV/FM mentioned previously, while varying with the nutrient status of the cells, does not demonstrate a consistent quantitative variation that could be applied to all phytoplankton. Accordingly, while FV/FM values may be of utility when studying single species, or blooms dominated by one species, they should not otherwise be used to determine either nutrient status or relative growth rates of phytoplankton. We thus support the view of Parkhill et al. (2001) that this methodology cannot give a robust index of nutrient status.

Several years after Parkhill et al. (2001) gave their warning, the use of FV/FM is widespread, aided no doubt by the sensitivity and simplicity of the measurement. Thus, for example, Springer et al. (2005) included FV/FM in their monitoring of the growth of a dinoflagellate bloom, commenting that the stability in this index ‘provided little evidence of photo-physiological stress as would have been expected under nutrient-limiting conditions’. In that instance, there was no reason to suspect nutrient limitation, so any such shift in FV/FM could have been ascribed to some other factor, such as viral attack (Juneau et al., 2003) or photodamage (Vaillancourt et al., 2003; Sobrino et al., 2004). There are other problems with the measurement of FV/FM in natural waters in the presence of Chl-breakdown products produced by predators (Fuchs et al., 2002).

Given the physiological acclimation exhibited by cells in response to both nutrient status and irradiance, which is simulated in photoacclimation models (Flynn et al., 2001), one may expect the photosynthetic machinery to adjust to retain optimal performance. Shifts in FV/FM reflect an inability (perhaps a short-term inability) to adjust to physiological imbalances, but the most likely factors to impact on FV/FM may be expected to be those directly affecting photochemical competence such as photodamage (Vaillancourt et al., 2003). While variation in FV/FM is most certainly an indicator that something is affecting photochemistry, other methods are needed to identify that trigger; nutrient stress should not be assumed. This view runs contrary to that expounded by Yentsch et al. (2004), that FV/FM can indeed be used as an index for nutrient stress.

The value of in vitro : in vivo Chl (Fig. 7b) appears to be at least as powerful as FV/FM (Fig. 7a) for predicting relative growth rate (Table 1). Flynn et al. (1994) measured in vitro : in vivo Chl in a field situation and related variations in the ratio to nutrient status as determined by the ratio of intracellular concentrations of free glutamine : glutamate (Gln : Glu). Like the in vivo Chl fluorescence signal, Gln : Glu gives a more immediate indicator of physiological stress than does whole-cell N : C. However, both in vitro : in vivo Chl and Glu : Gln require sample processing; a pure fluorescence approach is more attractive. The value of the performance index PIABS, provided by the PEA instrument, is a pure fluorescence signal that gave consistent qualitative (though not quantitative) changes with changes in growth rate (Fig. 7c). Interestingly, the direction of that signal was also different between N-limited and P-limited growth for S. trochoidea; whether this result is typical for algae needs further investigation.

Comparison between PAM and PEA

While the two fluorescence analytical instruments used in this study both determine FV/FM, they are very different having been developed for different applications. While PAM instruments have been widely used in research on algal suspensions, the PEA system has not. Parkhill et al. (2001) conducted a detailed comparison of the traditional 3-(3′,4′-dichlorphenyl)-1,1-dimethylurea (DCMU) methodology and PAM methods for measuring FV/FM, concluding that they gave similar results. However, there are differences in the results reported by different instruments (Suggett et al., 2003); Kromkamp & Forster (2003) suggest that different terminologies be used to alert scientists to such differences. There are methodological problems that affect all such methods, such as the adequacy of the period of dark adaptation (Honeywill et al., 2002). Differences between the PAM and PEA FV/FM values (Fig. 1) are primarily associated with inadequate Chl in low-biomass suspensions; determination of the cause of any other differences requires examination with a PEA system with enhanced sensitivity.

The existing PEA instrument is very portable, is quick and easy to use, and with the modified sensor would be quite suitable for field work in high-density situations (> 0.3 mg Chl l−1) such as harmful algal blooms. The work described here suggests that the development of a photo-multiplier tube-equipped PEA-type system may be of use in marine algal research and potentially also in the monitoring of algal blooms. The differences in the shape of the fast fluorescence rise curve (the initial few seconds of the Kautsky effect curve), and derivations thereof, indicate changes in photosynthetic pathways such as the absorption and trapping of light energy, and the transfer of light energy to the electron transport chain. These energy fluxes reflect changes in the functioning of the PSII system in response to physiological stress (Strasser et al., 1995), providing more points of reference in addition to the changes in FV/FM reported by PAM-type instruments. Indeed, FV/FM alone can be remarkably insensitive to stress (Tóth et al., 2005). The results from the PIABS analysis (Fig. 7c) give reason to be optimistic that a more detailed examination of the fluorescence signal, using an O-J-I-P approach, will be more successful, mirroring its utility for higher plant research (e.g. Strasser et al., 2000; Strauss et al., 2005).

Conclusions

There is no doubt that methods of chlorophyll and photochemistry analysis are of immense value. The problem comes when the data derived from these measurements are renamed as something that they are not. Chl a concentrations are just that; they are not biomass measurements. By the same token, FV/FM is not an index for anything other than the efficiency of photosystem II (and arguably a rather crude one at that). A lack of change in FV/FM can likewise not be used to infer the absence of stress. Gordillo et al. (2001) and Glud et al. (2002) report nonlinearities in comparisons between fluorescence-derived primary production and that derived by 14C and O2 methods. The EK values reported by RLC are not the same as those reported by traditional methods, as the shapes of these short-term PE curves are not the same at high values of E (Gilbert et al., 2000a). In short, scientists should describe the methods they employ and not disguise them by the use of more general terms such as biomass, nutrient status and production. Such disguises also hide potential problems in interpretation of the data.

Acknowledgements

This research was supported by the Natural Environment Research Council (UK) and equipment funds from University of Wales Swansea. We thank Hansatech Instruments for supply of a prototype liquid-sample head for the PEA analyser. Comments made by anonymous reviewers and by Professor Reto Strasser are acknowledged with gratitude.