Direct determination of Zn in individual Daphnia magna specimens by means of solid sampling high-resolution continuum source graphite furnace atomic absorption spectrometry

Abstract

In this work, the capabilities of solid sampling high-resolution continuum source graphite furnace atomic absorption spectrometry (GFAAS) for the direct determination of the Zn body burden in individual specimens of Daphnia magna (small aquatic invertebrates of approx. 2 mm length that are widely used in ecotoxicological research) were investigated. It was concluded that this technique offers interesting features for this type of application, permitting the fast (approx. 2 min for the measurement of a single sample) and practically contamination-free analysis of these invertebrates, as the only preparation required is drying. Moreover, one of the traditional disadvantages of line source-GFAAS, the narrow linear range achieved using a particular set of experimental conditions, can be overcome with this technique by the judicious selection of the detector pixels used for signal quantification, such that the linear range can be extended simply by reprocessing the data, without the need to carry out any additional measurement. This feature is particularly interesting for this application, since every Daphnia magna specimen can be monitored only once, and thus, there is no possibility to alter the conditions and repeat the analysis for samples displaying a Zn content outside the linear range. In this case, the use of central and/or side pixels permits achieving a working range between 5 and 400 ng for the 307.590 nm Zn line, which is fit-for-purpose. Analysis of two biological certified reference materials (SRM 1549 Non-fat milk powder and SRM 1577b Bovine liver) was also investigated, and accuracy could be demonstrated when carrying out the calibration versus aqueous standard solutions.

---

1. Introduction

In aquatic ecotoxicology, the measurement of whole body concentrations of metals in small organisms, such as Daphnia magna, is a prerequisite to a better understanding of uptake, storage and elimination processes and, thus, of the relation between these processes and toxicological effects of exposure to increased metal concentrations in the water.^1–10 This type of study is necessary to produce environmental regulations that are scientifically sound. Among the different elements that are investigated in this context, Zn is one of the most important, as this transition metal plays a significant role in many biochemical processes, but can also exert toxicity when individuals are exposed to elevated Zn concentrations.^5,9

When working with small species (e.g. D. magna adults of maximum about 300 μg dry weight), researchers often combine several individual specimens into a pooled sample, prior to acid digestion and analysis with atomic absorption spectrometry (AAS) or inductively coupled plasma mass spectrometry.^4,5,9 This is done to ensure that the final metal concentration in the digest is substantially higher than the limit of detection, as requested for reliable quantification. The value of the limit of detection in digested samples is often driven by contamination due to the acid digestion step (e.g., metal content in the acid(s), metal leaching from container wall, airborne contamination). This is especially true for Zn, owing to its ubiquity.^11–13 Unfortunately, combining several individuals into a pooled sample before digestion has the disadvantage that standard deviations among samples no longer represent inter-individual variation.¹⁴ Instead, they are lower by a factor equal to the square root of the number of individuals per pooled sample. Yet, this is often not accounted for in statistical comparisons between groups of samples (e.g., samples from a control group and a high Zn exposure group). Additionally, the need for multiple individuals per measurement increases the size (and costs) of the experimental design required to address the hypotheses of interest.

In this context, the use of direct solid sampling approaches can be beneficial, offering improved limits of detection because no sample dilution is carried out and the risk of contamination is greatly minimized. Some solid sampling techniques such as synchrotron radiation X-ray fluorescence^6,8 or laser ablation-ICPMS may allow to investigate the distribution of the target element in different animal compartments, while other techniques such as in-torch vaporization-ICP optical emission spectrometry have been proposed to determine total body burdens of small invertebrates.¹²

For the latter purpose, the use of solid sampling-graphite furnace AAS (SS-GFAAS) can be very suitable, owing to the potential of the technique to directly analyze sub-mg samples relying on straightforward calibration with aqueous standards.^15–17 This potential has been further enhanced with the arrival of high-resolution continuum source (HR-CS) GFAAS,^18,19 which offers improved capabilities for the correction of spectral interferences, as well as for expanding the linear range, and, to the best of the authors' knowledge, has not been investigated yet for the direct monitoring of Zn in solid samples.

The goal of the current paper is to evaluate the potential of solid sampling high-resolution continuum source GFAAS for the direct determination of Zn in individual D. magna specimens, weighing a few hundreds micrograms each. The potential of the technique to adjust the sensitivity and the linear range, without the need to carry out additional measurements using alternative experimental conditions, will be discussed in detail.

2. Experimental

2.1. Instrumentation

All the experiments in this work were carried out using a high-resolution continuum source atomic absorption spectrometer (HR-CS AAS), ContrAA 700, commercially available from Analytik Jena AG (Jena, Germany). The optical system comprises: a xenon short-arc lamp (GLE, Berlin, Germany) operating in “hot-spot” mode as the radiation source, a high-resolution double echelle monochromator (DEMON) and a linear CCD array detector with 588 pixels, 200 of which are used for analytical purposes (monitoring of the analytical signal and BG correction), while the rest are used for internal functions, such as correcting for fluctuations in the lamp intensity. At 307.590 nm, the atomic line for Zn used in this work, the spectral bandwidth per pixel was 1.6 pm. More details on this instrument can be found elsewhere.^19,20

The HR-CS AAS instrument is also equipped with a transversely heated graphite tube atomizer, pyrolytic graphite tubes for solid sampling (without dosing hole) and an automated solid sampling accessory (SSA 600), which incorporates a microbalance with a readability of 1 μg. This microbalance is recalibrated on a monthly basis with a 2.0 gram weight provided by the manufacturer (Sartorius, Goettingen, Germany) for this purpose. The typical uncertainty (standard deviation) of these measurements (n = 10) is 1 μg or lower. The samples were introduced using solid sampling graphite platforms.

2.2. Samples and standards

2.2.1. Standards and reagents. Purified water was obtained from a Milli-Q system (Millipore, Billerica, USA). Zn, Rh and Pd solutions were prepared from commercially available 1 g · L⁻¹ single-element standards (Merck, Darmstadt, Germany), by appropriate dilution with 0.14 mol · L⁻¹ HNO₃. The thiourea solution was prepared by dissolution of 1 gram of the solid reagent (Ultrapure quality, Scharlau, Barcelona, Spain) in 100 mL of milli-Q water. Colloidal Pd was synthesized as described elsewhere.²¹ 14 mol · L⁻¹ HNO₃ was purchased from Merck.

2.2.2. Samples. For this study, Daphnia magna individuals were taken from a monoclonal in-house laboratory population, which is continuously maintained in biologically filtered and aerated Ghent (Belgium) tap water (pH 8.0, hardness 180–200 mg · L⁻¹ as CaCO₃, 2–3 mg · L⁻¹ dissolved organic carbon (DOC), 5–7 μg · L⁻¹ Zn), as described elsewhere.⁵ Forty-two individuals 21 days old were randomly sampled from this population to form the base-level population for determination of Zn content. Twenty-four other individuals were exposed for 24 h in the same tap water, but supplemented with 2.5 mg Zn L⁻¹, added as ZnCl₂. During this exposure, the D. magna individuals were not fed with algae. This set of individuals will further be referred to as the Zn-exposed population.

Two biological reference materials produced by NIST (National Institute of Standards and Technology, Gaithersburg, USA) and certified for their Zn content were analyzed in this work for validation purposes. The materials investigated were SRM 1549 Non-fat milk powder and SRM 1577b Bovine liver. Both were available as powders.

2.3. Procedure for the direct determination of Zn by means of solid sampling HR-CS GFAAS

The samples were directly analyzed without any prior preparation step, except for drying. Before analysis, the samples were dried until constant weight at 40 °C. The solid sampling device used allows for automatic weighing and transport of the samples into the furnace.²² The empty platform was first transported to the microbalance using a pair of tweezers. After taring, an individual D. magna specimen (or a suitable amount of the sample when aiming at analysis of the reference materials) was placed onto the platform and weighed. Finally, the platform was transferred into the graphite furnace and subsequently subjected to the temperature program. All these operations were fully controlled from the computer, except for the deposition of the sample onto the platform, which was carried out manually. Calibration was performed using 10 μL of aqueous solutions of the appropriate concentrations, added with a micropipette onto the sampling platform.

The operating conditions are summarized in Table 1. Integrated absorbance (A_int) was selected as the measurement mode in all cases. For every measurement, the values obtained for the four central detector pixels (pixels 99, 100, 101 and 102), corresponding to a spectral interval of 6.4 pm, were summed. In the case of samples exceeding the linear range of the calibration curve, the signals for the samples, as well as for the standards, were reprocessed using only the pixels 99 and 102.

Table 1 Instrumental parameters used in the determination of Zn by means of HR CS GFAAS

Wavelength/nm	307.590 nm
Number of detector pixels summed	4 (≈ 6.4 pm)^a
Sample mass/mg	0.200–0.500 (NIST SRM1549)
	0.400–2.200 (NIST SRM 1577b)
	0.130–0.320 (Daphnia magna samples)
Tube lifetime	≈ 500 cycles

---

Temperature program

---

Step	Temperature/°C	Ramp/ °C·s^−1	Hold time/s	Ar gas flow/L·min^−1
a Except for samples exceeding the linear range for which only two pixels (99 and 102) were used. See section 3.2. for details. b 2 s integration used for Daphnia magna samples.
Drying	200	10	35	2.0
Pyrolysis	800	20	25	2.0
Auto zero	800	0	5	0
Atomization	1900	1500	4^b	0
Cleaning	2450	1000	1	2.0

---

3. Results and discussion

3.1. Optimization of the conditions for Zn determination: temperature program and possible use of modifiers

Zn has been determined a number of times using line source SS-GFAAS, although most often in environmental and industrial samples.¹⁸ Only a few papers have reported procedures for the analysis of biological/organic samples,^23,24 sometimes evaluating the suitability of the technique for controlling the homogeneity of reference materials.^25,26 For this type of samples, the slurry approach has also been investigated.^13,27,28

A review of all this literature shows that different chemical modifiers have been proposed for this element, including platinum group metals^13,25 and permanent modifiers.²⁸ Still, very often, the use of a modifier was not considered as necessary.^23,24,26,29 In the current work, the potential benefit of using Pd (as nitrate or in colloidal form),²¹ but also of Rh or thiourea (as an example of a reducing organic modifier) were evaluated. In principle, from the point of view of thermal stabilization, the use of Pd shows some improvement, allowing to raise the pyrolysis temperature up to 1000 °C (see Fig. 1). However, in the absence of a chemical modifier, it is already feasible to use a pyrolysis temperature up to 800–900 °C, which is typically more than enough for efficient matrix removal when analyzing biological samples.³⁰ Very similar blank levels (A_int≈0.001 s under optimum conditions) were found in all cases, but the use of modifiers seems to result in a (very moderate) decrease of sensitivity. Therefore, for the sake of simplicity, no modifier was used in further experiments and a pyrolysis temperature of 800 °C was selected. Under these conditions, the optimum atomization temperature (guaranteeing not only sufficient sensitivity, but also a well-defined signal with minimum tailing) was found to be in the range 1800–2000 °C. Thus, a value of 1900 °C was selected.

Fig. 1 Comparison of the pyrolysis curves obtained for 100 ng of Zn when monitoring the 307.590 nm atomic line in the presence of various chemical modifiers. An atomization temperature of 1900 °C was used.

3.2. Optimization of the conditions for Zn determination: possibilities of CS-HR GFAAS for the adjustment of the sensitivity and the expansion of the linear range

One of the main problems associated with solid sampling-GFAAS derives from the limited linear range of atomic absorption lines (roughly 10–20 times over the limit of quantification). This factor, in combination with the practical difficulties for diluting solid samples, makes it often problematic to determine the analyte at the level of interest. This problem is usually solved by using alternative working conditions,¹⁵ such as less sensitive lines, maintaining the Ar flow during atomization and/or using the 3-field mode Zeeman-effect background correction,^31,32 if available. These solutions help to adjust the sensitivity to the required level, although it is necessary to stress that the linear range for a particular set of experimental conditions is still rather limited.

This situation can be particularly problematic in this kind of study. As the goal is to determine the total content of Zn in individual D. magna specimens, it would not be possible to repeat the measurement for any of these specimens using alternative conditions when the value obtained falls outside the linear range. Every specimen can only be measured once. Thus, it is necessary to ensure that absolute Zn amounts ranging from a few ng to a few hundreds ng (the expected levels according to the mass of D. magna individuals—between 130 and 320 μg dry weight—and the information previously available^3,5) can be measured using a single set of experimental conditions.

In this particular case, the most sensitive line for Zn (213.856 nm) is too sensitive for the application intended. This line has been used for Zn determinations below the ng level.¹⁸ Moreover, that line shows a pronounced overlap with a non-resonance Fe line (213.859 nm) that could create difficulties for samples with a high iron level.²⁴ The alternative less sensitive (but still resonance) Zn line of 307.590 nm is more appropriate for the purpose intended here and was selected for further work. This line has been preferred in line source-SS-GFAAS papers to determine Zn at the ng level.^23–25

It is necessary to stress that the use of HR-CS GFAAS results in some fundamental differences in the way the atomic absorption signal is processed in comparison with traditional line source AAS devices. A 3D signal is now obtained adding a new dimension (wavelength) owing to the high spectral resolution of the instrument. The actual resolution of the instrument depends on the wavelength monitored. For the Zn 307.590 nm atomic line, the wavelength interval covered by each one of the 200 detector pixels is approximately 1.6 pm. Thus, the spectral environment that is simultaneously monitored is approximately 0.32 nm (±0.16 nm around the center pixel).

The final sensitivity obtained depends on the number of detector pixels used, a setting that can be freely selected by the analyst. Fig. 2 shows a typical line profile for Zn. As can be seen, in this case, the whole atomic absorption profile is practically covered by 6–7 of these pixels. Obviously, the more pixels that are selected, the higher the sensitivity and the lower the characteristic mass. However, the inclusion of pixels corresponding to the wings of the atomic line would improve the sensitivity only moderately, causing at the same time an increase in the noise level. It has already been reported that best limit of detections are usually obtained using three pixels only (the central pixel plus the adjacent ones).^33,34 However, in this case there is not a real “central line”, as both pixels 100 and 101 offer a very similar sensitivity. Thus, this aspect was investigated in more detail.

Fig. 2 Absorption line profiles obtained for 100 ng of Zn at 307.590 nm using the conditions shown in Table 1. The profile has been segmented such that the portion of the signal monitored by every detector pixel can be appreciated.

Table 2 shows the characteristic masses and the limits of detection obtained for different numbers of pixels monitored. As can be seen, the best limit of detection is obtained when using the four central pixels. Adding more pixels hardly improves the sensitivity and, as expected, deteriorates the limits of detection. Thus, selection of the four central pixels (approximately 6.4 pm of spectral bandwidth in total) is recommended and these four pixels were monitored in further work.

Table 2 Evolution of the figures of merit observed for Zn monitoring at the analytical line of 307.590 nm using HR-CS GFAAS as a function of the number of pixels selected for quantification. Every pixel covers a spectral interval of 1.6 pm. The furnace conditions used are shown in Table 1

Pixels monitored	Linearity/ng	R^2	Sensitivity/s ng^−1	mo/ng	LOD/ng
101	10 to 150	0.996	6.39 10^−4	7.1	2.9
100 + 101	5 to 150	0.997	1.28 10^−3	3.5	1.7
99 + 100 + 101 + 102	5 to 150	0.997	1.79 10^−3	2.5	1.4
98 + 99 + 100 + 101 + 102 + 103	5 to 150	0.998	1.89 10^−3	2.3	1.9
99 + 102	25 to 400	0.999	5.10 10^−4	8.3	5.9

---

By summing the signals detected with these four pixels, good linearity was obtained from 5 to 150 ng of Zn. This aspect is illustrated in Fig. 3a, where a typical calibration curve obtained with these four central pixels is shown. However, for higher masses of Zn (see points highlighted in grey), linearity is lost and the curve is better fitted to a polynomial equation. The use of the Dixon-Q test on the residuals of the calibration points confirmed that, indeed, the residual for 200 ng Zn was the first one to be significantly different from the previous at the 95% confidence level (q = 0.701; q_critical = 0.576).

Fig. 3 Linearity observed when monitoring the Zn 307.590 nm atomic line as a function of the detector pixels selected. A) Calibration curve using the four most sensitive central pixels (99, 100, 101 and 102). B) Calibration curve using only two side pixels (99 and 102). In both cases, the points highlighted in grey fall outside the linear range and are best fitted using a polynomial equation.

These working conditions are expected to be suitable to cover most of the samples to be analyzed. However, as discussed before, there could be some individual specimens that, because of a higher Zn accumulation or because they show a higher mass, could be outside this range. One interesting advantage of HR-CS-GFAAS is that, independently of the number of pixels used for quantification, all of the 200 analytical pixels are always recorded. Thus, it is always possible to reprocess the analytical data obtained using other pixels. It has already been demonstrated by Heitman et al. that the use of less sensitive “side pixels” only would enable an extension of the linear range.³³ In this case, as shown in Fig. 3b, using only the signals of the pixels 99 and 102 is a simple way to achieve linearity up to 400 ng Zn, which should be enough for the application intended.

Thus, the following strategy is proposed: i) always using the experimental conditions discussed in sections 3.1 and shown in Table 1, and ii) summing the signal for the pixels 99, 100, 101 and 102, as the standard procedure for most samples. However, if for some samples the integrated absorbance values are higher than that corresponding to 150 ng, the data (for the samples and the standards) will be reprocessed by summing the signals for detector pixels 99 and 102 only. In this way, a working range of almost two orders of magnitude (between 5 and 400 ng) can be covered without any modification of the experimental settings (temperature program, Ar flow, etc).

3.3. Analysis of solid samples: analysis of certified reference materials

Once the optimum conditions for analysis were established using solutions, the performance with solid samples was investigated using two different certified reference materials of biological origin. The materials selected were NIST SRM 1549 Non-fat milk powder and NIST SRM 1577b Bovine liver. By using these materials and properly selecting the sample mass, we could mimic similar conditions as those expected for the D. magna individuals. For instance, SRM 1549 has a Zn level similar to that expected from the D. magna base level population. Thus, by using sample masses between 0.2 and 0.5 mg, the conditions for the determination of this population of D. magna could be simulated closely. On the other hand, the Zn content of NIST SRM 1577b is higher and more similar to what could be expected from the D. magna population exposed to higher levels of Zn.

The analysis of both samples demonstrated that it is possible to directly transfer the conditions previously optimized with solutions. The signal profiles obtained for both samples are well defined and no indications of spectral interferences were observed (see Fig. 4b and 5b). Even though in all the cases the signal profiles follow a similar trend, it is also clear that the matrix components have some small effect on the atomization mechanism, as could be anticipated. For instance, the signal for NIST SRM 1549 appears at almost the same time as the signal for an aqueous standard, but is slightly narrower and taller (see Fig. 6a). On the contrary, the signal for NIST SRM 1577b is a bit delayed and exhibits more tailing (see Fig. 6b). Thus, the use of peak height for quantification, in spite of providing a two-fold improvement in terms of sensitivity, is not recommended, as it would result in overestimated values for NIST SRM 1549 (approx. 15% biased high) and underestimated values for NIST SRM 1577b (approx. 25% biased low).^35,36

Fig. 4 Comparison of the time- and wavelength-resolved absorbance spectrum obtained upon the atomization of a similar mass of Zn (approximately 15 ng) from a) an aqueous solution, b) SRM 1549 non-fat milk powder, c) a Daphnia magna (base level) specimen, under the conditions indicated in Table 1.

Fig. 5 Comparison of the time- and wavelength-resolved absorbance spectrum obtained upon the atomization of a similar mass of Zn (approximately 150 ng) from a) an aqueous solution, b) SRM 1549 SRM 1577b bovine liver, c) a Daphnia magna (Zn-exposed population) specimen, under the conditions indicated in Table 1.

Fig. 6 a) Comparison of the time-resolved absorbance signal measured at 307.590 nm (sum of 4 central pixels) obtained upon the atomization of a similar mass of Zn from an aqueous solution, directly from SRM 1549 Non-fat milk powder and directly from a Daphnia magna (base level) specimen, under the conditions indicated in Table 1; b) Comparison of the time-resolved absorbance signal measured at 307.590 nm (sum of 4 central pixels) obtained upon the atomization of a similar mass of Zn from an aqueous solution, directly from SRM 1577b bovine liver and directly from a Daphnia magna (Zn-exposed population) specimen, under the conditions indicated in Table 1.

Nevertheless, when using integrated absorbance (A_int), similar sensitivity values were obtained for all the materials and also for aqueous solutions, as illustrated in Fig. 6. Therefore, direct analysis of the solid samples is possible and calibration versus a curve constructed with aqueous standards seems feasible, provided that integrated absorbance is always employed and that the conditions shown in Table 1 are used.

The results obtained for both reference materials are shown in Table 3. As can be seen, an excellent agreement with the corresponding reference values could be achieved, proving the accuracy of the method. The precision obtained was approximately 11 and 8% RSD, respectively. Obviously, this is higher than the typical values for solution analysis (2–3% would be normal for this instrument), as a consequence of the level of heterogeneity of the solid samples.³⁷

Table 3 Results obtained for the direct analysis of two reference materials using the procedure described in section 2.3. The furnace conditions used are shown in Table 1

Sample	Sample amount/mg	Zn amount/ng	Replicates	Average/μg g^−1	Standard deviation/μg g^−1	Reference value/μg g^−1
a Using only pixels 99 and 102 for signal integration.
SRM 1549 Non-fat milk powder	0.2–0.5	10–25	25	45.8	5.3	46.1 ± 2.2
SRM 1577b Bovine liver	0.4–1.2	50–150	25	118.4	9.1	127 ± 16
SRM 1577b Bovine liver^a	1.3–2.2	160–280	25	120.4	8.0	127 ± 16

---

Finally, the analysis of the sample NIST SRM 1577b was also repeated using higher masses such that the integrated atomic absorption values would fall outside the calibration curve if the four central detector pixels are used. Thus, these data were reprocessed using only pixels 99 and 102, as discussed in the previous section. The usefulness of this approach could be validated in this way since a similar level of accuracy and precision, in comparison with the standard approach of using four detector pixels, was obtained, as shown in Table 3.

3.4. Analysis of solid samples: analysis of D. magna specimens

After validation of the method using reference materials, the analysis of the samples of interest was undertaken. As indicated before, two different populations of D. magna were subjected to analysis: the base level population (forty-two specimens), and the Zn-exposed population (twenty-four specimens), which received an additional Zn supplement during culturing. More details of these populations are provided in section 2.2.2.

Analysis of the samples was carried out as described in section 2.3. For four samples of the second population the linear range obtained with four detector pixels was exceeded and the approach discussed in section 3.2., and based on using side pixels only, was followed to overcome this issue. For the rest, analysis of discrete D. magna samples proved very straightforward, as the specimens have a size very suitable for manipulation and deposition onto the graphite platform: they are approx. 2 mm long (see Fig. 7), with a weight ranging from 0.13 to 0.32 mg. The only aspect relevant to report is that it is recommended to deposit 5 μL of milli-Q water on top of the samples before analysis in order to prevent them from being ejected from the furnace during the drying step.

Fig. 7 Photograph of a Daphnia magna specimen.

Regarding the signals obtained, examples for both populations can be seen in Fig. 4c and 5c. The spectrum is a bit more complex as some additional lines appear. This is more evident in Fig. 4c than in Fig. 5c, simply because the scale is different as the Zn level is much higher for the latter. A more detailed view of the spectrum is shown in Fig. 8. As can be seen (Fig. 8c), a line appeared to the left of the Zn line, which can be attributed to Fe (307.572 nm). Moreover, there are some structures, possibly owing to molecular lines. However, these other lines pose little problems for Zn monitoring owing to the high resolution of the instrument and, also, to the fact that these other lines are well separated in time, as they correspond to more refractory species. To further prove this point, Fig. 8c′ shows the profile obtained during the first 2 s of atomization of a D. magna specimen, which is as clean as the one obtained for the aqueous standard or for the reference material. Therefore, the subtraction of these molecular structures (e.g., by using least-squares background correction, as it is sometimes performed for HR-CS AAS)¹⁹ was not considered as necessary, and integration of the signals for D. magna samples during the first two seconds of atomization only was preferred instead.

Fig. 8 Wavelength-resolved time-integrated absorbance spectrum obtained upon the atomization of a similar mass of Zn (15 ng) from a) an aqueous solution, b) SRM 1549 non-fat milk powder, and a Daphnia magna (base level) specimen either during the whole atomization step (c), during the first two seconds of the atomization step (c′), or during the last two seconds of the atomization step (c′′).

The spectral profiles are very similar to those obtained for solutions (see Fig. 6), except for the fact that they tend to appear slightly earlier. Anyway, the use of integrated absorbance should guarantee a good degree of accuracy, as demonstrated with the reference materials.

The results obtained are shown in Table 4. For the base level population, the mean Zn burden was 80.1 μg g⁻¹ dry weight with a standard deviation of only 9.5 μg g⁻¹ dry weight (RSD = 11.8%). This mean value is within the range of Zn burdens recently reported for this monoclonal population of D. magna under non-Zn-stressed conditions, i.e. 62.3–155 μg Zn g⁻¹ dry weight.^3,5 This is a rather homogenous population, since all but four samples fall into the 70 to 100 μg g⁻¹ range. The population was found to be symmetrical and the Kolgomorov–Smirnov test confirmed a Gaussian distribution (test value = 0.076; critical value = 0.137).

Table 4 Results obtained upon analysis of both Daphnia magna populations by means of solid sampling HR-CS GFAAS

	Base level population	Zn-exposed population
Average/μg g^−1	80.1	576.8
Median/μg g^−1	80.8	554.0
Standard deviation/μg g^−1	9.5	143.1
RSD/%	11.8	24.8
Kurtosis	1.14	−0.36
Asymmetry coefficient	−0.18	0.59
Minimum value/μg g^−1	52.8	359.0
Maximum value/μg g^−1	102.0	886.0
Number of samples	42	24

---

On the other hand, the population cultured in a Zn-enriched medium showed a substantially higher Zn content, with a mean Zn body burden of 576.8 μg g⁻¹. This population was observed to be less symmetrical than the previous one (values lower than the mean were found more often), but still the Kolgomorov–Smirnov test did not indicate a significant deviation from normality (test value = 0.112; critical value = 0.177). The most relevant aspect is that the variation among different specimens was significantly larger for this population (RSD = 24.8%), and the difference between the minimum (359 μg g⁻¹) and the maximum (886 μg g⁻¹) values was almost a factor of three. Obviously, there are significant between-individual differences in Zn accumulation, as this high level of variation cannot be explained assuming a different base Zn level alone. The expression of a higher between-individual variation under stressful conditions, (such as the high Zn exposure level used here) is well-known in ecotoxicology.³⁸

4. Conclusions

The use of solid sampling high-resolution continuum source GFAAS makes the direct analysis of individual specimens of D. magna (or of similar invertebrates) feasible, in a simple and straightforward way. The potential of the technique for reprocessing the data after the measurements with the number of detector pixels of choice helps ensuring that the integrated absorbance values for all samples always fall inside the linear range of the calibration curve.

The analysis of every individual only takes two minutes, which, together with the feasibility to carry out the calibration versus aqueous standards, makes the method very attractive for ecotoxicological studies, particularly for an element as contamination-prone as Zn.

Acknowledgements

This study was financially supported by the Spanish Ministry of Education and Science (Project CTQ2006-03649/BQU), the Government of Aragón (DGA research project PM013/2007) and the FWO-Vlaanderen (G.0046.06). Jorge Briceño acknowledges the University of Carabobo for his doctoral grant. Karel De Schamphelaere is a senior research assistant of the Flemish Science Foundation (FWO-Vlaanderen, Belgium).

Notes and references

1. C. Barata, D. J. Baird and S. J. Markich, Aquat. Toxicol., 1998, 42, 115–137.
2. X. J. Liu, I. H. Ni and W. X. Wang, Water Res., 2002, 36, 4563–4569.
3. K. A. C. De Schamphelaere, M. Canli, V. Van Lierde, I. Forrez, F. Vanhaecke and C. R. Janssen, Aquat. Toxicol., 2004, 70, 233–244.
4. M. Canli, Comp. Biochem. Physiol., Part C, 2005, 141, 110–116.
5. B. T. A. Muyssen, K. A. C. De Schamphelaere and C. R. Janssen, Aquat. Toxicol., 2006, 77, 393–401.
6. B. De Samber, R. Evens, K. De Schamphelaere, G. Silversmit, B. Masschaele, T. Schoonjans, B. Vekemans, C. R. Janssen, L. Van Hoorebeke, I. Szalóki, F. Vanhaecke, G. Falkenberg and L. Vincze, J. Anal. At. Spectrom., 2008, 23, 829–839.
7. I. Komjarova and R. Blust, Aquat. Toxicol., 2008, 90, 138–144.
8. B. P. Jackson, H. E. Pace, A. Lanzirotti, R. Smith and J. F. Ranville, Anal. Bioanal. Chem., 2009, 394, 911–917.
9. I. Komjarova and R. Blust, Aquat. Toxicol., 2009, 94, 81–86.
10. M. B. Vandegehuchte, F. Lemière and C. R. Janssen, Comp. Biochem. Physiol., Part C, 2009, 150, 343–348.
11. L. I. L. Balcaen, K. A. C. De Schamphelaere, C. R. Janssen, L. Moens and F. Vanhaecke, Anal. Bioanal. Chem., 2008, 390, 555–569.
12. A. T. Smith, H. R. Badiei, J. C. Evans and V. Karanassios, Anal. Bioanal. Chem., 2004, 380, 212–217.
13. E. Muñoz-Delgado, J. C. Morote-García, R. Romero-Romero, I. López-García and M. Hernández-Córdoba, Anal. Biochem., 2006, 348, 64–68.
14. Y. Saavedra, A. González and J. Blanco, Int. J. Environ. Anal. Chem., 2009, 89, 559–568.
15. M. A. Belarra, M. Resano, F. Vanhaecke and L. Moens, TrAC, Trends Anal. Chem., 2002, 21, 828–839.
16. M. Resano, F. Vanhaecke and M. T. C. de Loos-Vollebregt, J. Anal. At. Spectrom., 2008, 23, 1450–1475.
17. M. G. R. Vale, N. Oleszczuk and W. N. L. dos Santos, Appl. Spectrosc. Rev., 2006, 41, 377–400.
18. B. Welz, M. G. R. Vale, D. L. G. Borges and U. Heitmann, Anal. Bioanal. Chem., 2007, 389, 2085–2095.
19. B. Welz, D. L. G. Borges, F. G. Lepri, M. G. R. Vale and U. Heitmann, Spectrochim. Acta, Part B, 2007, 62, 873–883.
20. U. Heitmann, M. Schütz, H. Becker-Roβ and S. Florek, Spectrochim. Acta, Part B, 1996, 51, 1095–1105.
21. A. B. Volynsky and V. Krivan, J. Anal. At. Spectrom., 1996, 11, 159–164.
22. K. C. Friese and V. Krivan, Spectrochim. Acta, Part B, 1998, 53, 1069–1078.
23. M. A. Belarra, C. Crespo, M. P. Martínez-Garbayo and M. Resano, Spectrochim. Acta, Part B, 2003, 58, 1847–1858.
24. M. A. Belarra, M. P. Martínez-Garbayo, J. M. Anzano and J. R. Castillo, Anal. Sci., 1996, 12, 483–488.
25. C. S. Nomura, C. S. Silva, A. R. A. Nogueira and P. V. Oliveira, Spectrochim. Acta, Part B, 2005, 60, 673–680.
26. G. N. Kramer, K. H. Grobecker and J. Pauwels, Fresenius' J. Anal. Chem., 1995, 352, 125–130.
27. N. Campillo, P. Viñas, I. López-García and M. Hernández-Córdoba, Talanta, 1998, 46, 615–622.
28. O. Acar, Anal. Chim. Acta, 2005, 545, 244–251.
29. K. C. Friese, V. Krivan and O. Schuierer, Spectrochim. Acta, Part B, 1996, 51, 1223–1233.
30. M. Resano, J. Briceño and M. A. Belarra, J. Anal. At. Spectrom., 2009, 24, 1343–1354.
31. H. Gleisner, K. Eichardt and B. Welz, Spectrochim. Acta, Part B, 2003, 58, 1663–1678.
32. M. Resano, M. Aramendía, E. García-Ruiz, C. Crespo and M. A. Belarra, Anal. Chim. Acta, 2006, 571, 142–149.
33. U. Heitmann, B. Welz, D. L. G. Borges and F. G. Lepri, Spectrochim. Acta, Part B, 2007, 62, 1222–1230.
34. M. Resano, J. Briceño and M. A. Belarra, Spectrochim. Acta, Part B, 2009, 64, 520–529.
35. B. Welz, Spectrochim. Acta, Part B, 1992, 47, 1043–1044.
36. U. Kale and E. Voigtman, Spectrochim. Acta, Part B, 1995, 50, 1531–1541.
37. M. A. Belarra, M. Resano and J. R. Castillo, J. Anal. At. Spectrom., 1998, 13, 489–494.
38. B. T. A. Muyssen, C. R. Janssen and B. T. A. Bossuyt, Aquat. Toxicol., 2002, 56, 69–79.

---