Survey of vitamin D and 25-hydroxyvitamin D in traditional native Alaskan meats, fish, and oils

Greater consumption of traditional foods has been associated with improved vitamin D status in Arctic and subArctic populations, including Alaskan Native Americans. However, lack of vitamin D food composition data impairs epidemiological studies on health outcomes, and development of specific dietary recommendations. Vitamin D, including 25(OH)D3 was quantified in samples of native fish, fish eggs, meats (caribou, goose, whale, seal) and traditionally prepared whale and seal oil collected from Alaskan tribes. Vitamin D3, 25(OH)D3, and vitamin D2 were assayed in alkaline-saponified samples by UPLC-MS, after derivatization with 4-phenyl-1,2,4triazole-3,5-dione, with in-house control materials and/or NIST SRM 1546a Meat Homogenate included in each analytical batch. All but the land animals and bearded seal meat contained ≥2 μg vitamin D3/100 g, with>10 μg/100 g in steelhead trout; dried sheefish, whitefish, smelt; smoked/dried salmon; fermented sheefish eggs; whale and seal oils. Large between-sample differences in bearded seal oil suggested possible effects of season and/or maturity on vitamin D content. 25(OH)D3 was> 0.3 μg/100 g in many foods, notably smoked salmon, beluga whale skin/fat and oil and spotted seal (but not other seal) oil, with the highest levels in dried beluga whale meat, skin/fat, and oil (up to 1.2). Vitamin D2 was< 0.2 μg/100 g in all foods.


Introduction
Vitamin D is an essential nutrient for bone health and has been increasingly recognized as a factor in immune system modulation, musculoskeletal health, cancer, and other aspects of health and disease (Pludowski et al., 2013). Vitamin D 3 is readily formed in the epidermis with sun (ultraviolet light) exposure, from its precursor, 7-dehydrocholesterol (Wacker and Holick, 2015). However, many individuals and populations have suboptimal vitamin D levels, as measured by serum 25-hydroxyvitamin D 3 [25(OH)D 3 ] concentration (Holick, 2005;Ross et al., 2011), due to insufficient sun exposure or dietary intake. Populations at higher latitudes are particularly susceptible to deficiency (Barake et al., 2010;Chen et al., 2007;Webb et al., 1988), and in these groups, dietary sources of vitamin D become more significant. Studies have found higher serum 25(OH)D 3 in individuals than what would be predicted by ultraviolet light (UV) exposure and dietary intake determined by food intake surveys and food composition data, and part of the explanation might be underestimation of dietary intake due to a lack of reliable and complete data for vitamin D in foods, particularly for 25(OH)D 3 Taylor et al., 2016). Since 25(OH) D 3 has a higher biological activity than D 3 and is not routinely measured in foods or included in food composition databases, it may be an overlooked component of vitamin D intake (Cashman et al., 2012;Norman, 2008;Ovesen et al., 2003).
Suboptimal vitamin D status has been documented in a variety of Arctic and sub-Arctic populations, including Alaskan and Canadian Northwest Native Americans (Frost and Hill, 2008;Lebrun et al., 1993;Sharma et al., 2011;Singleton et al., 2015;Weiler et al., 2007). There has been a trend toward westernization of the diet of Native populations (Ballew et al., 2006;Johnson et al., 2009;Murphy et al., 1995;Nobmann et al., 2005), although greater consumption of traditional foods in North American Arctic and sub-Arctic populations has been associated with better vitamin D status (Andersen et al., 2013;Bersamin et al., 2007;Luick et al., 2014;Mansuri et al., 2016;Mohatt et al., 2007;Ryman et al., 2015;Sheehy et al., 2014). "Traditional" foods are defined as those "composed of items from the local, natural environment that are culturally acceptable" (Kuhnlein and Receveur, 1996). However, a lack of food composition data for traditional Native foods impairs both epidemiological studies on vitamin D status and health and the development of specific dietary recommendations for increasing vitamin D intake in these populations (Johnson et al., 2009 T food composition data, including sampling that is relevant to the population being studied (Ahuja et al., 2013). Food sampling for the USDA National Food and Nutrient Analysis Program (NFNAP) (Haytowitz et al., 2008;Haytowitz and Pehrsson, 2018) has provided analytical food composition data for the USDA National Nutrient Database for Standard Reference (SR) (USDA, 2016), including foods for the What We Eat In America (WWEA) National Health and Nutrition Examination Survey (NHANES) (U.S. Department of Agriculture, Food Surveys Research Group, 2017). Sampling is based on statistical models that account for market share and sources of variability in the food supply, to obtain representative population-wide average food composition data (Perry et al., 2002). However, these models cannot be translated to Native populations, since traditionally harvested and prepared foods must be obtained from remote locations, directly from the tribes who may not have excess to spare. Thus, efforts to develop robust vitamin D data for traditional foods need to be focused on those that are significant sources of the nutrient. There is a lack of information on vitamin D in these foods to serve as a basis to narrow the number of foods for further investigation.
In general, foods naturally rich in vitamin D include fish and fish oils, and some meats and organ tissues. However, the level can vary widely among specific foods in these groups, and the content can vary based on location of sampling (Roseland et al., 2018). While some fish are high in vitamin D (e.g., salmon, trout), many others contain negligible vitamin D (Mattila et al., 1995;Roseland et al., 2018;Schmid and Walther, 2013). Thus, it is reasonable to expect that some Native Alaskan fish and meats might be high in vitamin D, including 25(OH) D 3 . Kuhlein at al. (2006) reported on the vitamin D 3 content of several traditional native North American meats and fish sampled from Canadian tribes, but did not include25(OH)D 3 . Foods which are potentially good sources of 25(OH)D 3 include some fish and organ meats (Roseland et al., 2018), pork fat (Clausen et al., 2003), and eggs from vitamin D supplemented hens (Exler et al., 2010). Furthermore, the vitamin D content of a particular species of fish or mammal can be influenced by diet or environment (Liu et al., 2013;Mattila et al., 1997;Rao and Raghuramulu, 1996;Roseland et al., 2018). Processing prior to consumption also can influence vitamin D; variation has been observed based on removal of the skin from meats and fish is removed (Pierens and Fraser, 2015), sun drying (Barnkob et al., 2016), and differences in cooking methods (Jakobsen and Knuthsen, 2014;Mattila et al., 1999;Roseland et al., 2018). Specific environmental conditions and methods of preserving (e.g. smoking, drying), storing, and cooking foods likely vary among different indigenous populations, which have localized food supplies and often unique techniques for handling and preparation and that deserve consideration when evaluating dietary intake in specific populations.
An effort to obtain food composition data specifically for Alaskan and American Indian Native (AIAN) traditional foods was initiated by the United States Department of Agriculture (USDA) Agricultural Research Service (ARS) Nutrient Data Laboratory (NDL) (Beltsville, MD), with funding from the National Institutes of Health Office of Research on Minority Health (ORMH) and the Indian Health Service (IHS) (Rockville, MD). The main focus was to obtain food composition data for key nutrients, for inclusion in the American Indian/Alaskan Native database in SR, to support efforts on health studies of Native Americans (Amy and Pehrsson, 2003;Johnson et al., 2009;Pehrsson et al., 2005a, b;Perry et al., 2002). Meanwhile, efforts were being made in general to add data on vitamin D to SR , including studies on analytical methods and quality control Byrdwell, 2009;Phillips et al., 2008), vitamin D content of fish and shellfish (Byrdwell, 2009;Byrdwell et al., 2013), milk (Patterson et al., 2010), eggs , and mushrooms (Phillips et al., 2011). A key part of this work was validating methods for vitamin D and 25(OH)D. Until recently, methods for vitamin D were generally not standardized and often gave disparate results between laboratories using similar methods Phillips et al., 2008). More recently, methodology for measurement of vitamin D and 25(OH)D by LC-MS has been validated (Byrdwell, 2009;Huang et al., 2009). The USDA Nutrient Data Laboratory (Beltsville, MD) has demonstrated that it is possible to obtain consistent results for vitamin D 3 and 25(OH)D 3 among laboratories . Matrixspecific control materials and certified reference materials are, however, critical to include to account for the many factors that impact accuracy and precision of vitamin D quantification in different types of foods (Roseland et al., 2018).
The objective of this work was to measure vitamin D 3 and 25(OH)D 3 in samples of traditional meats and fish obtained from Native Alaskan tribes, to screen for foods that could be a rich dietary source of vitamin D, as a basis for focusing more detailed food composition studies as well as to add to the dearth of published data for traditional foods, using validated analytical methodology and including traceable quality control materials.

Sample collection
The foods sampled included raw, preserved (dried, frozen, fermented, canned, smoked) and prepared (boiled or baked) fish, land and marine mammals, and marine mammal oils. Samples were collected from Alaskan Native tribes, according to the sampling framework described previously (Perry et al., 2002), from the locations shown in Fig. 1. The foods sampled were selected based on published research, informal surveys and interviews with tribal council, elders, and members (e.g., at clinical setting), collaboration (e.g., food frequency questionnaires from other government or academic studies and focus groups. Handling of Alaskan native endangered species samples was authorized under permit 782-1694 from the National Marine Mammal Laboratory of the Alaska Fisheries Science Center of the National Oceanic and Atmospheric Agency (Seattle, WA). All food collection was approved by the councils of each tribe, and where presentations and supportive documents were required they were submitted to the tribe.
Samples were obtained within and among villages and tribes, from festivals or markets, homes, senior centers (food parties) and schools, as summarized in Table 1. Multiple samples of each food were obtained when possible, although the number of samples was necessarily limited by the availability of extra food for this study. Prepared foods (smoked, boiled, dried, etc.) had been processed by traditional methods, but details of these procedures were not always available. The food samples were, as much as possible, weighed on site (as the amount collected was often considered a common portion), frozen, and then packed on dry ice or blue ice packs and shipped via overnight express to the Food Analysis Laboratory Control Center (FALCC) in the Biochemistry Department at Virginia Tech (Blacksburg, VA). Upon receipt at the FALCC, samples were inspected to confirm integrity and then stored frozen at < −15 ± 5°C until prepared for analysis.

Sample preparation
Samples were thawed in the refrigerator (4 ± 2°C) prior to homogenization (meats and fish overnight, fish eggs 4-6 h, and oils until completely liquefied). Canned fish samples were homogenized using a Robot Coupe ® industrial stainless steel food processor (RSI 6 V Blixer, Robot Coupe USA, Jackson, MS). Other fish, meat, and fish egg samples were homogenized in the same manner, except with the addition of liquid nitrogen to aid the homogenization process. Oils were thawed at 4 ± 2°C until liquefied, then stirred in a stainless steel bowl with a stainless steel spoon to homogenize. After homogenization, subsamples were dispensed in 15-30 g aliquots among 30-mL or 60-mL glass jars with Teflon™-lined lids (Qorpak ® , Bridgeville, PA; #GLC-07098 and GLC-08640, respectively). The jars were sealed under nitrogen, surrounded with aluminum foil, and stored at < −55°C until analyzed. Subsamples of each composite were batched with control and/or reference materials (Section 2.5) and shipped on dry ice via overnight express delivery to the laboratory, which confirmed receipt of the intact frozen samples, with dry ice remaining and stored them at −70°C until analyzed.

Vitamin D analysis
Vitamin D (D 3 , 25(OH)D 3 , and D 2 ) were analyzed by ultra-high performance liquid chromatography tandem mass spectrometry (UPLC-MS) at a commercial laboratory, which also participated in the inter-laboratory study on the analysis of vitamin D and 25(OH)D in food matrices .

Moisture and total fat analyses
Moisture (AOAC, 2005a) and total fat (AOAC, 2005b) were also measured at the laboratory, along with control materials, to provide relevant proximate composition data.

Quality control
Samples of a Pork/Egg control material ("Pork/Egg CC, a mixture of bratwurst sausage and cooked egg yolks), and/or NIST SRM® 1546a Meat Homogenate [National Institute of Standards and Technology (NIST), Gaithersburg, MD] that were part of the inter-laboratory study , and/or an in-house Salmon control material ("Salmon D CC", canned salmon) previously reported (Phillips et al., 2008) were analyzed with each batch of samples. Some of the Native  (Table 1) shown as stars (www.mappery.com/Alaska-Road-Map-3).  Alaskan samples were analyzed in replicate, in independent analytical runs as a measure of day-to-day precision. Precision was estimated by percent relative standard deviations (RSD) for the samples assayed in replicate and for the control materials assayed over multiple runs. Acceptability of the RSD in each case was evaluated using the Horwitz ratio (HorRat), the assayed RSD/expected RSD, where the expected RSD was calculated as the assayed mean(μg/100 g)/(100/1,000,000) −0.1505 (Horwitz and Albert, 2006).
Accuracy was validated by results for three commercially available relevant food matrix reference materials assayed in a previous interlaboratory study on quantitation of vitamin D and 25(OH)D : NIST SRM ® 1546a Meat Homogenate, NIST SRM ® 1577c Freeze-Dried Bovine Liver, and NIST SRM ® 1845a Whole Egg Powder), analyzed in select batches during the period of this study. The accuracy of 25(OH)D 3 measurements was established in a previous report on eggs, which included recovery tests and the Pork/Egg CC , and also relative to consensus values for NIST SRM ® 1546a Meat Homogenate, SRM ® 1845a Whole Egg Powder, and SRM ® 1577c Freeze-Dried Bovine Liver, obtained in the previous interlaboratory study .

Quality control data
The assayed vitamin D 3 concentration in all certified reference materials was within the certified mean ± uncertainty (Table 2). There are no food matrix reference materials with certified value for 25(OH) D 3 (although, more recently NIST has reported a reference value for 25(OH)D 3 in SRM ® 1549a Whole Milk Powder). Therefore, we relied on data for the Pork/Egg CC, SRM ® 1546a Meat Homogenate, SRM ® 1845a Whole Egg Powder, and SRM ® 1577c Freeze-Dried Bovine Liver previously characterized , also validated by previously reported recovery studies for the Pork/Egg CC , to validate the accuracy of results during the period of this study. The assayed 25(OH)D 3 was within the previously established tolerance limits for all of these materials (Table 2).
Because the Native food samples were analyzed over multiple assay batches, it was important to define and demonstrate the inter-assay precision, to establish uncertainty in mean values and enable comparison of vitamin D content among independent samples of a particular food or among different foods. Fig. 2 illustrates the precision of vitamin D 3 and 25(OH)D 3 analysis for the Salmon and Pork/Egg control materials; both had excellent precision over multiple analytical runs, with HorRat < |1.0|. The HorRat is a measure of the analytical precision demonstrated, relative to what is expected based on the concentration of the nutrient in the matrix, and should be between |1.0| and |2.0| (Horwitz and Albert, 2006). The HorRats for replicate analyses of samples are given in italics after the RSDs, in Table 3, and ranged from 0.05 to 1.5 for vitamin D 3 (mean, 0.6; median, 0.7), and from 0.06 to 1.4 for 25(OH)D 3 (mean, 0.6; median, 0.5). These data demonstrate excellent precision of analyses across the range of sample matrices and analyte concentrations, and that data for different foods assayed in different batches can be compared with confidence that differences are not due to run-to-run analytical variability.

Vitamin D content of foods
Vitamin D 2 was < 0.1 μg/100 g in all cases. Although some vitamin D 2 could be present in marine mammals and fish, due to diet (Roseland et al., 2018), it was not detected in samples analyzed in this study. Vitamin D 3 was ≥2 μg/100 g [≥13% of the DRI of 15 μg (600 IU) set by IOM (2010)] in all of the foods except the land animals, bearded seal meat, and mature (but not young) bearded seal oil. More than 10 μg (400 IU) vitamin D 3 per 100 g was present in steelhead trout; dried sheefish, whitefish, and smelt; fermented sheefish eggs; salmon (including smoked salmon); and beluga whale and seal oils [except, interestingly, only in young bearded seal (20 μg/100 g) but not mature bearded seal oil (0.29 μg/100 g)]. Many of the foods were significant sources of 25(OH)D 3 . The biological activity of 25(OH)D 3 has been hypothesized to be potentially 1.7-5 times that of D 3 (Bischoff-Ferrari et al., 2012;Jakobsen, 2007;Ovesen et al., 2003;Taylor et al., 2016), at least in part due to greater absorption relative to vitamin D 3 (Borel et al., 2015). Using a potency of 5x (to account for foods which cumulatively might contribute significant vitamin D activity), a 25(OH)D 3 level ≥0.3 μg would represent at least 10% of the DRI for vitamin D. Foods containing ≥0.3 μg 25(OH)D 3 per 100 g were dried and smoked salmon; beluga whale dried meat, boiled skin and fat, and oil; and spotted seal oil. The highest levels of 25(OH)D 3 (0.7-1.2 μg/100 g) were in beluga whale skin and fat, dried meat, and oil, and in smoked red salmon. All of the land animals had vitamin D 3 contents < 0.1 and 25(OH)D 3 contents < 0.2 μg/100 g. There was no correlation between vitamin D 3 and 25(OH)D 3 concentrations within any of the groups of all samples, fish and marine mammal products only, or marine mammal products only (ρ = 0.177, ρ = 0.127, and ρ = −0.278, respectively),  K.M. Phillips et al. Journal of Food Composition and Analysis 74 (2018) 114-128 but the relationship was generally positive among samples of fish (ρ = 0.759).
There was a wide range in vitamin D content among different samples for some foods, for which multiple samples were possible to collect [halibut (n = 4), dried and smoked salmon (n = 4), beluga whale skin and fat (n = 2), and young bearded seal oil (n = 2)]. There was a striking difference in vitamin D 3 content among samples of beluga whale skin and fat (2.97 and 10.0 μg/100 g), although 25(OH)D 3 content (1.2 μg/100 g) did not vary. Both D 3 and 25(OH)D 3 varied among samples of dried and smoked salmon (32.6-53.6 and 0.081-0.836 μg/100 g, respectively) and were not related to moisture content (39.8-67.8 μg D 3 and 0.112-1.06 μg 25(OH)D 3 per 100 g dry weight, respectively).
There are some publications that discuss possible reasons for variability in vitamin D between species and within different samples of food species (Kenny et al., 2004;Mattila et al., 1997;Pierens and Fraser, 2015;Rao and Raghuramulu, 1996;Roseland et al., 2018), including variation in diet. Reports on vitamin D metabolism and content of marine mammals are more limited (Keiver et al., 1988;Kenny et al., 2004;Routti et al., 2010;Wilske and Arnbom, 1996). Seals belong to the suborder Pinnipedia, comprising flipper-footed marine animals [seals (Phocidfamily), sea lions (Otariid family), and walruses (Odobenid family)]; whales belong to the order Cetacea, that also includes dolphins and porpoises (Marine Mammal Center, 2016). Pinnepeds spend some time on land, and thus (in contrast to whales and fish), have the potential for cutaneous synthesis of vitamin D in addition to dietary sources. Wilske and Arnbom (1996) studied elephant seals and found two annual peaks in serum 25(OH)D 3 , occurring after periods of intensive exposure to UV radiation. Because the hunting season in far northern latitudes would be in the late spring-early fall, vitamin D stores in seals could be expected to build up over the summer and be higher in animals caught in late summer or fall. Thus, the low vitamin D Fig. 2. Quality control data for vitamin D 3 and 25-hydroxyvitamin D [25(OH)D 3 3] in two matrices, a composite of bratwurst sausage and egg yolks (Pork/Egg CC) and canned sockeye salmon (Salmon D CC). SD = standard deviation; RSD = relative standard deviation. HorRat = ratio of assayed to expected RSD, with expected RSD = assayed mean/((100/1,000,000) −0.1505 ), according to Horwitz and Albert (2006). content of oil from the mature bearded seal but high content in the oil from young bearded seal (Table 3) might be due to the time of harvest (May for the mature seal, August for the young seals; see Table 1), rather than age of the animal. Kenny et al. (2004) found significant differences in vitamin D3 content of blubber among cetaceans and pinnepeds that was correlated with diet, with those primarily consuming invertebrates (e.g., bowhead whale, Pacific walrus) having lower vitamin D 3 levels than those consuming primarily fish (e.g., beluga whale, ringed seal). However, in the present study, the vitamin D content of beluga whale oil, while high (mean, 7.11 μg/100 g), was substantially lower than that of ringed seal oil (29.0 μg/100 g). One limitation of the present study (necessary due to the factors stated in Section 2.1) and others is that there are few samples per type of animal and lack of details on age, sex, time of harvest, and diet, so there are not enough data to draw conclusions about the relationship between vitamin D content and time of harvest, maturity of animal or other variables. These areas are worthy of further study.

Nutritional significance
In some Native Arctic populations, large quantities of wild caught foods, including salmon and other fish, contribute significantly to the diet (Ballew et al., 2006). Numerous researchers have reported improved vitamin D status and health outcomes in Native Alaskan and other Arctic populations consuming a diet rich in traditional foods (Andersen et al., 2013;Bersamin et al., 2007;Johnson et al., 2009; Sheehy et al., 2014;Singleton et al., 2015). However, the conclusions about specific nutrient and health outcomes rest on the quality and completeness of food composition data, which are lacking . Sharma et al. (2011) reviewed the association between lower risk of vitamin D deficiency and consumption of traditional foods in Arctic populations and suggested increasing the intake of fatty fish and marine mammals. However, in the present study, the vitamin D 3 and fat contents among fish and marine mammal samples were not K.M. Phillips et al. Journal of Food Composition and Analysis 74 (2018) 114-128 highly correlated (ρ = 0.154; see Fig. 4), confounding the contention that "fatty fish" are the best sources of vitamin D, but consistent with a similar observation by Lu et al. (2007). Additionally, within fish and marine mammals, certain species appear to be far better sources of vitamin D than others (Figs. 3 and 4). Of the fish, halibut and whitefish were relatively lower in vitamin D (< 10 μg D 3 and/or < 0.2 μg 25(OH)D 3 per 100 g) than the other species [salmon, steelhead trout, sheefish, and dried fish (salmon, whitefish, sheefish, smelt)] that contained as much as 53.6 μg D 3 and nearly 1 μg 25(OH)D 3 per 100 g, equivalent to, per 85 g serving, more than 300% of the 15 μg DRI specified by the Institute of Medicine (2011). Among different marine mammals and also within species, there were large inter-sample differences in some cases. Probably one of the most significant sources of vitamin D that would be unique to the Native diet is seal oil. In many traditional Native Alaskan households (e.g. Inuit and Yup'ik), it is used as a dipping sauce for bread and meats/fish, added to soups or stews, and in some traditional medicinal preparations (Magdanz and Wolfe, 1988 and Table 1). Among all samples analyzed, the ringed seal oil was one of the highest in vitamin D, with 29.0 μg D 3 /100 g, equivalent to 4.35 μg per one 15 g (∼1 tablespoon) serving (29% of the DRI). The content was substantially higher than beluga whale oil, although the latter was also a good source of vitamin D [7.11 μg D 3 and 0.739 μg 25(OH)D 3 100 g]. However, there was one sample of bearded seal oil with < 0.5 μg D 3 / 100 g). Given the limited number of samples, it is possible that differences are not necessarily inter-species necessarily, but due to variation among individuals within a species, as discussed in Section 3.2.
The results of this survey study suggest the importance of collecting detailed information on intake of wild caught foods, including season, maturity, and preparation, and correlating those intakes with the corresponding food composition data reflecting factors that influence vitamin D content in individual animals, in epidemiological studies seeking to correlate nutrient intake with health outcomes in Native populations. For example, the average vitamin D 3 content among all seal oil samples was 17.0 μg/100 g, but the range was 0.37-29 (< 1 to 4.35 per 15 g serving), and 25(OH)D 3 was < 0.1 μg/100 g in all samples but one, which contained a meaningful amount (0.628 μg/100 g) (section 3.3). Considering its varied and widespread use and that it would be consumed in variable amounts and up to more than tens of grams per day, just monitoring intake of "seal oil" and an average value for vitamin D in seal oils could over-or underestimate vitamin D intake by more than 50% of the DRI if 30 g were consumed, based on the range found in samples in this study.
One point worth noting is the risk of pesticide residues (e.g., organohalogens) and heavy metals (e.g., mercury) that can accumulate in sea mammals, and in the case of organohalogens, concentrate in the fat (for example, see Brown et al., 2016;Gmelch et al., 2017;Laird et al., 2013;Pedro et al., 2017;Reiner et al., 2016). Data on potential toxic components in the marine mammal oils and fats consumed in particular populations would be beneficial to determine prior to recommending massive increases in consumption. Additionally, there have been cases of botulism from improper storage and handling of seal oil and dried meats (Bendinger, 2014; Centers for Disease Control and Prevention and Alaska Department of Health and Social Services, 2016; Dankmeyer, 2016) so attention should be paid to education in handling for food safety. Table 4 summarizes vitamin D data for Arctic fish and marine mammal products from the literature compared to values from the present study of Alaskan Native foods. Differences in vitamin D 3 were Fig. 4. Relationship between vitamin D 3 and total fat contents in Native Alaskan fish and marine mammal foods (excluding oils). K.M. Phillips et al. Journal of Food Composition and Analysis 74 (2018) 114-128 notable among studies/samples for seal oils and blubber (1.6-160 μg/ 100 g for ringed seal and < 0.1-29.2 μg/100 g for bearded seal) and whale oil and blubber (< 1.6-74.4 μg/100 g). It is difficult to determine whether the variability is due to actual within-species variation due to environment or maturity, for example (as discussed for variation among samples in this study, section 3.2 and in Roseland et al., 2018) or analytical differences, due to lack of reference samples in common that would quantify the contribution of any analytical variability, as discussed in other publications (Phillips et al., 2006;Phillips and Rasor, 2016;Roseland et al., 2016). The existing literature do not include data on 25(OH)D 3 , and some of the foods can be significant contributors (> 0.6 μg/100 g) (see Fig. 3). Thus, 25(OH)D 3 should be included in food composition databases used to assess vitamin D intake in native populations.
The scarcity of data on vitamin D in Native foods makes it difficult to draw definitive conclusions about vitamin D intake and health in Native populations. Yet, hypotheses about diet and health, and epidemiological associations between food components and health outcomes, depend on complete and valid food composition data that are representative of the food supply of the population being studied. Without such data, it is not possible to reliably estimate nutrient intake or possible covariation of different components. For example, Adler et al. (1994) reported decreased prevalence of impaired glucose tolerance and diabetes associated with daily seal oil or salmon consumption by Alaska Natives, and attributed the effect to omega-3 fatty acids.
However, in the present study, both of these foods were among the highest in vitamin D 3 (32.6-53.6 and 16.1-29.9 μg/100 g (bearded, ring, and spotted seal), respectively) and 25(OH)D 3 [0.081-0.836 and 0.628 (spotted seal), respectively]. Whereas this does not prove cause and effect, possible associations between co-varying nutrients, particularly omega-3 fatty acids and vitamin D, may be overlooked when food composition data are lacking.
Incomplete data on vitamin D in foods also means potential misestimation of vitamin D intake, particularly underestimation of the contribution of 25(OH)D 3 . Currently, SR does not contain data for 25(OH)D 3 , although it is planned to be included in future releases . Data for vitamin D 3 for only some of the samples were incorporated into SR beginning with release 19 (USDA, 2006) as part of the American Indian and Alaskan Native (AIAN) food composition database (Pehrsson et al., 2005a), due to the limited resources and quality of analytical methods available at the time. Many other food composition databases that have been used for dietary intake studies (e.g., Jakobsen and Saxholt, 2009;Lamberg-Allardt, 2006;Mileševića et al., 2018;Neufingerl et al., 2016;Ovesen et al., 2003) also have not contained data for 25(OH)D 3 . Recently, improved analytical methodology, including for 25(OH)D 3 , has allowed USDA to analyze more foods for both vitamin D 3 and 25(OH)D 3 . Currently, the USDA is transitioning to a larger, more comprehensive food composition database, that will eventually include 25(OH)D 3 values. . NDB numbers from SR are given in parentheses for the conventional foods. Error bars show the range for traditional foods for which more than one sample was analyzed, and the standard error for values from SR. * cooked; ** canned; ***data are for cuts with similar fat content and are the mean and range for NDB numbers 10983,10984,10985,10986,10987,10988,10990,10991,10992,10993. Serving sizes are the Reference Amounts Customarily Consumed Per Eating Occasion (U.S. Food and Drug Administration, 2017 for fish and fish eggs; U.S. Food and Drug Administration, 2012) for meats and animal oils and fats: uncooked meats and fish, 110 g; cooked meats and fish, 85 g; smoked salmon, 55 g; dried fish, 30 g; fish eggs, 15 g; oils and fats, 15 g. (No data for 25(OH)D 3 available in SR as of 2017).

Limitations and future research
The number of samples analyzed was necessarily limited by the fact that the foods given for analysis were from the Native hunted food supply, and intensive sampling would impact tribal food security.
Nonetheless, the wide range of vitamin D content among the marine mammal products, likely reflecting influences of species, maturity, environment, and season and the seasonality of food harvesting and consumption in these populations, suggests the importance of population-specific sampling of the food supply that reflects these variables. a Skin and fat. K.M. Phillips et al. Journal of Food Composition and Analysis 74 (2018) 114-128 The preliminary data from this study has pointed towards more intensive sampling of marine mammals and oils as most critical in obtaining accurate estimates of vitamin D intake, along with addressing other challenges that have been described for nutritional epidemiology in Native populations (Amy and Pehrsson, 2003;Fialkowski et al., 2010;Pehrsson et al., 2005a, b;Ryman et al., 2015;Sheehy et al., 2013;Standlee-Strom, 2016).

Declarations of interest
None.