Characterisation and chemometric evaluation of 17 elements in ten seaweed species from Greenland

Several Greenland seaweed species have potential as foods or food ingredients, both for local consumption and export. However, knowledge regarding their content of beneficial and deleterious elements on a species specific and geographical basis is lacking. This study investigated the content of 17 elements (As, Ca, Cd, Cr, Cu, Fe, Hg, I, K, Mg, Mn, Na, Ni, P, Pb, Se and Zn) in 77 samples of ten species (Agarum clathratum, Alaria esculenta, Ascophyllum nodosum, Fucus distichus, Fucus vesiculosus, Hedophyllum nigripes, Laminaria solidungula, Palmaria palmata, Saccharina latissima and Saccharina longicruris). Element profiles differed between species but showed similar patterns within the same family. For five species, different thallus parts were investigated separately, and showed different element profiles. A geographic origin comparison of Fucus species indicated regional differences. The seaweeds investigated were especially good sources of macrominerals (K > Na > Ca > Mg) and trace minerals, such as Fe. Iodine contents were high, especially in macroalgae of the family Laminariaceae. None of the samples exceeded the EU maximum levels for Cd, Hg or Pb, but some exceeded the stricter French regulations, especially for Cd and I. In conclusion, these ten species are promising food items.


Introduction
Marine macroalgae, commonly known as seaweeds, are increasingly becoming popular as 2 food items in the Nordic countries [1], as well as in Greenland [2,3], where they have 3 been a part of the traditional Inuit diet [4,5]. Moreover, seaweeds have been identified 4 as a sustainable income source in the remote and sparsely populated areas of the 5 Northern Periphery and Arctic region of Northern Europe and Greenland -a region 6 with a low population density and pristine waters [6]. 7 Having detailed insight in the nutritional composition of macronutrients (lipids, 8 carbohydrates, proteins, etc.) and minor components, including essential and 9 non-essential elements, is important for both currently consumed seaweed species and 10 species of interest for future human consumption [7]. 11 Seaweeds have a highly variable nutritional composition [8,9] but are generally good 12 sources of minerals and iodine [8,10]. However, in some cases they are also known to Materials and methods 48 Samples and sampling locations 49 A total of 77 samples belonging to ten seaweed species were collected in the intertidal or 50 upper subtidal zone between June and September in 2017 and 2018 at low tide 51 conditions along the shore or by divers in West, South and East Greenland, see   Table 1 provides an overview of the number of species per location. The harvest sites 53 were chosen to represent different areas in Greenland. The species and number of 54 samples were as following: Agarum clathratum (3), Alaria esculenta (9), Ascophyllum 55 nodosum (7), Fucus distichus (8), Fucus spp. (7, specimens that were too small to be 56 distinguished as either F. distichus or F. vesiculosus), Fucus vesiculosus (15), 57 Hedophyllum nigripes (5), Laminaria solidungula (6), Palmaria palmata (2), Saccharina 58 latissima (10) and Saccharina longicruris (3). Sample pre-treatment 63 Samples were rinsed in clean seawater at the collection site, epibiota were carefully 64 removed, samples were frozen in clean food grade plastic bags at -20°C and transported 65 frozen to the laboratory in Denmark. Samples were freeze dried (Christ Beta 1-8, 66 Martin Christ Gefriertrocknungsanlagen GmbH, Osterode am Harz, Germany) and, for 67 compositional comparison of different thallus parts from five algal species, thereafter 68 manually divided into blade, midrib and stipe, see      Institute of Japan, Tsukuba, Japan)) were processed and analysed alongside the rest of 98 the samples. As internal standard, a mixture of Bi, In and Rh was prepared from single 99 element calibration standards (PlasmaCal, SCP Science). A calibration curve from 100 0 ng mL -1 to 400 ng mL -1 was prepared for all elements from single element calibration  were processed along with the samples. One in every ten samples was also determined 120 in duplicate. Tellurium (Te) (PlasmaCal, SCP Science) was used as the internal 121 standard and a calibration curve was prepared from ultrapure iodide (Iodide 1000 g L -1 122 Spectrascan SS11I, Ski, Norway) from 0 ng mL -1 to 100 ng mL -1 iodine. The samples 123 were analysed on an iCAP Q ICP-MS (Thermo Scientific, Bremen, Germany) equipped 124 with an ASX-520 AutoSampler (Cetac) running Qtegra version 2.10.3324.83 (64 bit) (Thermo Scientific).

126
The limit of detection (LOD) and limit of quantification (LOQ) for both methods 127 were calculated using the standard deviation (SD) obtained from repeated analysis of 128 blank samples: 129 LOD respectively LOQ = SD of the blanks (ng/mL) * f * dilution factor 130 (mL)/sample amount (g), with f = 3 for LOD and f = 10 for LOQ.

152
Normality was assessed with the Shapiro-Wilk test. Since preconditions for 153 parametric tests were not met, Kendall's ranked correlation coefficient was used for 154 pairwise element correlations and the Kruskal-Wallis test was used to compare 155 differences between species and locations. Principal component analysis was carried out 156 using the prcomp function (stats package 3.4.4), with centring and scaling of samples. 157 A confidence level of 95% was used unless otherwise noted.

Results and discussion
159 Quality assurance 160 Quality assurance parameters are presented in Table B in S1 file.

161
For Na, I and P, LOD and LOQ were derived from the lowest accepted concentration 162 of the calibration curve since the blanks were below detection limit. Relative standard 163 deviations were in the range of 0.01% to 25%, with medians between 1.6% and 4.6% for 164 all elements except Hg and Se with relative standard deviations up to 55% and 31%, 165 respectively, and medians of 22% and 11%.

166
Assessment of normality revealed that element concentrations were not normally 167 distributed. Therefore, element concentrations are reported as median values with 168 median absolute deviation, and non-parametric tests were used for data analysis.

170
The full dataset of individual sample results is freely available online [28]. Samples with 171 different thallus parts examined separately are connected by their sampleID to the 172 respective powderIDs. In the present article, summarised results are presented and 173 discussed.

174
Species comparison 175 Median element contents are presented in Table 2. The most abundant cations were 176 K>Na>Ca>Mg (3.79 g kg -1 to 108 g kg -1 freeze dried weight). Schiener and 177 colleagues [29] found the same sequence in an investigation that included among other 178 seaweeds species, S. latissima and A. esculenta. These four light metal cations are also 179 the most abundant cations in seawater, with Na>Mg>Ca>K [30]. In seaweeds, these 180 cations are gradually replaced by heavier divalent metal ions such as Cu from the 181 seawater [15], during the continued growth of the algae [31]. Laminariaceae. We therefore theorize that the four members of Laminariaceae studied 200 here also possess iodoperoxidases, leading to the high observed accumulation of iodine. 201 Riget, Johansen and Asmund [33] reported concentrations of selected elements (As, 202 Cd, Cr, Cu, Fe, Pb, Zn) in A. nodosum and F. vesiculosus collected in the Nuuk area 203 between 1980 and 1990. The major differences between their findings, and the findings 204 of this study were increased Fe and Zn concentrations for both species. For A. nodosum, 205 they reported Fe concentrations of 16 -43 mg kg -1 , in this study we found 140 mg kg -1 , 206 and Zn concentrations were reported as 6.6 -10.7 mg kg -1 , while in this study we found 207 58.1 mg kg -1 . For F. vesiculosus, Fe concentrations of 33 -77 mg kg -1 were reported by 208 Riget, Johansen and Asmund [33], while in this study we found 133 mg kg -1 and they 209 reported Zn concentrations of 7.2 -10.2 mg kg -1 compared to the 50.6 mg kg -1 in this 210 study. The most likely explanation could be the difference in sampling: Riget, Johansen 211 and Asmund [33] collected five samples of growing tips, while in this study the entire 212 thallus was analysed, and samples were pooled so this study only reports one result per 213 species from Nuuk. Another explanation could be the increased human and industrial 214 activity in the area since their study -Nuuk has nearly doubled in size, from around   The presence of overall tendencies in element content or fingerprint per algal family 224 were assessed by PCA, as presented in Fig 3. Both Hg and Se were excluded from the 225 analysis due to the very low observed concentrations, which could not be quantified for 226 the majority of samples. Fucaceae (A. nodosum, F. distichus and F. vesiculosus), Laminariaceae (H. nigripes, L. solidungula, S. latissima and S. longicruris), which had a 229 higher content of especially iodine, but also K and P. Alariaceae (A. esculenta) could 230 not be distinguished from the other families with this method, and for Agaraceae (A. 231 clathratum) and Palminariales (P. palmata), the low sample number of three and two 232 samples, respectively, precluded analysis in this manner. To the best of the authors 233 knowledge, this is the first PCA presented in the literature of this specific combination 234 of species. It is interesting to note that Laminariaceae, known for their high contents of 235 iodine, could be distinguished from Fucaceae based mainly on their iodine content. 236 We also used PCA to investigate the influence of nearby human settlement size, 237 based on the content of elements associated with anthropogenic contamination (Cd, Cr, 238 Cu, Pb and Zn). However, there was no clear correlation evident.  Concentrations of As were higher in stipes than in blades for S. latissima (Fig 4,   245 panel A), and similarly, K concentrations in S. latissima and S. longicruris (Fig 4, 246 panel E). A possible explanation for this could be that metal(loid)s (such as As, Cd, Hg, 247 K and Pb) are stored associated with biopolymers [15], and these biopolymers are 248 differently distributed throughout the thallus. Research into the properties of alginate, 249 with respect to divalent metal ions, from Laminara digitata and Laminaria hyperborea 250 in the 1960ies also showed differences between stipe and other (nondisclosed) parts of 251 macroalgae [36] and is supported by observations by S. Wegeberg & O. Geertz-Hansen 252 (unpublished data). 253 Pétursdóttir and colleagues [37] reported similar concentrations for total As in stipes 254 of Icelandic A. esculenta (53 mg kg -1 ± 3 mg kg -1 their study, 45 mg kg -1 ± 6 mg kg -1 our 255 study), however they reported much higher concentrations in the midrib (43 mg kg -1 ± 256 4 mg kg -1 their study, 23 mg kg -1 ± 7 mg kg -1 our study) and the blade (93 mg kg -1 ± 257 4 mg kg -1 their study, 31 mg kg -1 ± 16 mg kg -1 our study) than we found.

263
The differences between the study of Pétursdóttir and colleagues [37] and ours could 264 be related to the time of sampling: they sampled during late winter, while the seaweeds 265 in our study were collected during June to September. Previous studies have shown a 266 seasonal change in nutritional composition of seaweeds (e.g. [29,38]). Another possible 267 explanation is the small sample size of both studies: two samples of Pétursdóttir and 268 colleagues [37], whereas the present study reports on three stipe and five blade samples. 269 Ronan and colleagues [39] reported that total arsenic concentrations of both A. 270 nodosum and L. digitata increased with the age of the thallus part, which is a probable 271  (Fig 4., panel C). This could be due to iron accumulating differently in older compared 281 to younger macroalgae, or thallus parts, which have been shown to grow at different 282 rates by Buggeln and colleagues [40]. 283 Differences in element concentrations can also depend on where the sample is taken 284 on the blade. This is due to the localization of meristem and thus the allocation of 285 nutrients for growth, e.g., close to the stipe and hence close to the meristem or distally 286 (S. Wegeberg and O. Geertz-Hansen (unpublished data on biopolymers), [29]). For L. 287 solidungula, the blade generation is also significant. In the present study, neither the 288 localisation on the blade nor the blade generation were investigated. are grouped into the same quadrant as those from the dump in Sisimiut. By analysing 303 many elements, it is thus possible to distinguish between locations even at small sample 304 sizes per location. Analysis of location differences based on a single element through 305 Kruskal-Wallis testing revealed statistically significant differences (p<0.05) for the 306 following elements: As, Cd, Cu, Fe, Mg, P, Pb and Zn. This is also reflected in the 307 PCA, where these eight elements have the strongest influence on the PCs, as evidenced 308 by the length of the arrows representing the loadings. The observed natural variation at a given sampling site is due to different factors, 310 both abiotic (e.g. salinity) and biotic (e.g. fouling). These factors lead to metabolic 311 changes which affect growth rates and element uptake [41]. Brinza and colleagues [42] 312 found that Zn uptake rates differed greatly between Danish and Irish F. vesiculosus. 313 They explained the greater Zn uptake rates in the specimens from Denmark with . This is also reflected in the small angles between the loadings for K-P and Cr-Fe 322 shown in figure 3, a sign of correlation. Elements are ordered alphabetically for ease of reading. Hg and Se were excluded from the analysis due to the low number of quantifiable samples. Only statistically significant correlations (p < 0.05) are shown.
Mg 2+ , Na + and K + are some of the most common cations in seawater [30,43]. 323 Seaweed acquires these light metal ions from seawater, and they are, together with 324 Ca 2+ , indeed reported as the main cations in seaweed biomass [15]. 325 An explanation for the strong correlation observed between Cr and Fe can be that 326 when Fe partially replaces Ca in the alginate matrix of the cell walls, it creates 327 favourable binding sites for Cr, as reported by Nayak and colleagues [44]. 328 A significant correlation between As-P (with a Kendalls tau coefficient of 0.32 in the 329 present study), was also reported by Taylor and Jackson [45]. They reported a similar 330 ratio of As-P in brown algae (0.015) as found in this study (0.025), which is slightly 331 lower than the ratio found in seawater of 0.033 (As = 0.002 mg L -1 ; P = 0.06 332 mg L -1 [43]). They argued that this similarity in ratio could be due to these two 333 elements being taken up by the same mechanism. However, we conclude that the 334 difference in ratios between seawater and in the seaweed suggests that seaweed is indeed 335 able to differentiate between As and P. Arsenates and phosphates have similar chemical 336 properties, which contributes to the toxicity of arsenates [46]. In marine algae, As(V) 337 enters cells through phosphate transporters, while As(III) enters through the plasma 338 membrane via aquaglyceroporins and hexose permeases [46,47]. Most of the arsenic 339 taken up by macroalgae is stored as arsenosugars which are considered as less toxic to 340 humans than inorganic arsenic [45].  This is supported by Desideri and colleagues [49], who reported element correlations 349 as Pearson's coefficients on a total of 14 samples from a mixture of edible macroalgae 350 and microalgae, some of the latter even being freshwater organisms. They did not 351 indicate the threshold for significance, but element pairs significantly correlated in our 352 study and those with high correlation coefficients in their study (> 0.7) were As-I, 353 Cu-Mn and Ni-Mn. 354 We theorize that the shared correlations are due to chemical similarity of the 355 element pairs, however an in-depth investigation is beyond the scope of this study.

356
Nutritional and food safety aspects 357 Table 3 summarises current European and Nordic guidelines on recommended daily 358 intake levels, upper daily intake levels, maximum levels in the EU and France and 359 toxicological guideline values for the elements investigated in our study. Since average 360 seaweed consumption data for Europe has not been documented, we based our intake 361 scenario on a typical seaweed serving size in a Danish restaurant. From the four 362 collected responses, the portion size of a seaweed salad ranged from 20 g to 50 g, with a 363 median of 33 g. To assess the nutritional benefits and exposure to toxic elements, we 364 calculated element concentrations found in a 33 g single-seaweed species salad, prepared 365 from fresh seaweed (Table 4). Our estimated serving size is comparable with Sá 366 Monteiro and colleagues [13], who estimated an intake of 5 g freeze dried weight per 367 week, which corresponds to about 30 g fresh seaweed at an estimated moisture content 368 of 80% (based on moisture contents reported by Holdt and Kraan [8]).

369
In general, all investigated seaweed species are good sources of essential minerals and 370 trace elements. One portion of a single-seaweed salad contributes with between 1% to 371 55% of the recommended intake for a specific element. For example, one portion of S. Abbreviations: Recommended daily intake (RI), upper daily intake (UI), lower confidence limit of the benchmark dose (BMDL), tolerable weekly intake (TWI), tolerable daily intake (TDI), body weight (bw), wet weight (ww), dry weight (dw). The French regulations apply to seaweed in vegetable or condiment form. * Food supplements consisting exclusively or mainly of dried seaweed, products derived from seaweed, or of dried bivalve molluscs. ** No recommendation given due to lack of sufficient evidence [62]. *** Lower value for men and women post menopause, higher value for women. **** Food supplements. ***** Lower value for women and higher for men.

372
latissima salad contains 647 µg Fe, corresponding to a daily recommended intake of 5% 373 (for women) to 7% (for men and women post menopause).

374
However, iodine levels were high: P. palmata was the only seaweed for which iodine 375 exposure did not exceed the recommended upper daily intake of 600 µg for adults, which 376 is in accordance with other studies [10,32]. However, it has been shown that iodine 377 concentrations of e.g. S. latissima can greatly be reduced by soaking in warm 378 freshwater [63] or blanching in hot freshwater [64]. A recent study in Ammassalik (East 379 Greenland) by Andersen and colleagues [4] showed that consumption of locally 380 Table 4. Calculated median element content for a single-seaweed salad. Where applicable, percentage of recommended daily intake is indicated in parentheses. Elements exceeding recommended upper intake levels are marked in bold font.

391
They concluded that a single meal containing seaweed only had a temporary effect on 392 the thyroid, even at high iodine concentrations in the food.

393
Furthermore, it is important to note that, while K is an important constituent of the 394 human diet, for patients on a low potassium diet (2 g day -1 to 3 g day -1 ), the 395 consumption of one seaweed salad prepared from H. nigripes, L. solidungula or S. 396 longicruris would contribute with over 0.5 g K, which is up to 25% of the recommended 397 daily intake for these patients.

398
None of the individual samples exceeded the EU maximum levels for Hg of 3 mg kg -1 399 wet weight, see also table 3, for individual sample results see [28]. Many samples (48) 400 exceeded the maximum levels for Cd according to French regulations. However, the 401 French Agency for Food, Environmental and Occupational Health & Safety (ANSES) is 402 currently evaluating whether these maximum levels will be maintained or increased to 403 the European level, which was not exceeded by any sample. Only two samples exceeded 404 the French limit for Pb, but none exceeded the European limit for Pb. Many samples of 405 Laminariaceae (H. nigripes, L. solidungula, S. latissima, S. longicruris) exceeded the 406 French regulation maximum level for iodine.

407
The content of total arsenic is listed for future reference, since the content of 408 inorganic arsenic, for which there exist toxicological guideline values, was not quantified 409 in this study.

411
In this study, 77 samples of ten Greenland seaweed species were collected and analysed 412 for the content of 17 elements.

413
The element profiles varied between species, and species from the same family accordance with other studies. The results from the thallus part analysis of stipe, rib or 417 blade can be used to select or discard specific seaweed parts, depending on desired high 418 or low concentrations of specific elements.

419
Elements associated with anthropogenic contamination showed no clear trend with 420 human settlement size. Broad geographic differentiation, based on element profile, was 421 possible for Fucus species. However, the geographic identification was obfuscated in the 422 case of Sisimiut and Ilulissat, for samples collected close to waste discharge. The strong 423 influence of human waste on the elemental profile means one should refrain from 424 harvesting close and downstream to waste discharge into the sea, even though current 425 European limits for toxic elements were not exceeded.

426
Iodine contents were very high in some species of the Laminariaceae, which limits 427 consumption of untreated raw macroalgae according to recommendations on daily 428 intake. However, studies on washing and blanching treatments of seaweeds from other 429 areas show that these treatments are very effective in iodine reduction, while 430 maintaining a good nutritional profile. Recent studies in Greenland furthermore suggest 431 that bioavailability of iodine from seaweed might be as low as 50%, and that intake of a 432 single meal containing seaweed only had a temporary effect on the thyroid.  Furthermore, a more detailed investigation of seaweeds from different areas and 437 substratum will help to elucidate geographic differences.   Table B in S1 file Limit of detection (LOD), limit of quantification (LOQ), 442 and reference material analysis. 443