Nutrient composition of important fish species in Bangladesh and potential contribution to recommended nutrient intakes

Fish, in Bangladesh where malnutrition remains a signiﬁcant development challenge, is an irreplaceable animal-source food in the diet of millions. However, existing data on the nutrient composition of ﬁsh do not reﬂect the large diversity available and have focused on only a few select nutrients. The purpose of this study was to ﬁll the gaps in existing data on the nutrient proﬁles of common ﬁsh in Bangladesh by analysing the proximate, vitamin, mineral and fatty acid composition of 55 ﬁsh, shrimp and prawn species from inland capture, aquaculture and marine capture ﬁsheries. When comparing species, the composition of nutrients of public health signiﬁcance was diverse. Iron ranged from 0.34 to 19 mg/100 g, zinc from 0.6 to 4.7 mg/100 g, calcium from 8.6 to 1900 mg/100 g, vitamin A from 0 to 2503 m g/100 g and vitamin B12 from 0.50 to 14 m g/100 g. Several species were rich in essential fatty acids, particularly docosohexaenoic acid in capture ﬁsheries species (86–310 mg/100 g). The potential contribution of each species to recommended nutrient intakes (RNIs) for pregnant and lactating women (PLW) and infants was calculated. Seven species for PLW and six species for infants, all from inland capture, and all typically consumed whole with head and bones, could potentially contribute (cid:2) 25% of RNIs for three or more of these nutrients, simultaneously, from a standard portion. This illustrates the diversity in nutrient content of ﬁsh species and in particular the rich nutrient composition of small indigenous species, which should guide policy and programmes to improve food and nutrition security in Bangladesh.


Introduction
In Bangladesh, fish is an irreplaceable animal-source food in the diet of millions, both in terms of quantity -accounting for approximately 60% of animal protein intake at 18.1 kg consumed per person per year -and frequency of consumption, far exceeding that of any other animal-source food . The country possesses diverse and abundant aquatic resources with 267 freshwater fish species (Thilsted, 2010), and an annual production of 3.1 million tonnes . Bangladesh is also one of many developing countries to experience the proliferation of aquaculture, now the world's fastest growing food production sector, during a period of decline in capture fisheries .
While remaining largely successful in increasing supply to meet the demand of a growing population, the extent to which the growth of aquaculture has been able to mitigate reduction in dietary diversity and micronutrient intake from the diverse but waning capture fisheries sector, focusing on only a few select large species, is questionable . Despite improvement in some food and nutrition security indicators (JPGSPH and HKI, 2012), malnutrition, largely caused by inadequate micronutrient intake, remains widespread with 41% of children under five years suffering from stunted growth (NIPORT et al., 2013). The fisheries and aquaculture sector has been recognised as a key resource in tackling food and nutrition security issues and features prominently in the national development agenda (Government of the production, access and nutrient intakes, and in devising policies and programmes such as development of improved production technologies (Thilsted and Wahab, 2014a), to ensure that food supply optimally fulfils population nutrient requirements. However, despite the clear importance of fish in the Bangladeshi diet, existing composition data do not reflect the large diversity of species available for consumption and have only focused on a few select nutrients rather than comprehensive nutrient profiles. The recently published Food Composition Table for Bangladesh is a useful compilation of existing composition data on important foods (including a number of fish and fish products); however, the data come from a large number of sources including regional databases with varying sampling and analytical methods, some of which are now several decades old (INFS, 2013).
The primary objective of this study was to document comprehensive nutrient composition profiles of important fish, shrimp and prawn species in Bangladesh with a specific focus on small indigenous species (SIS). Species and nutrient components selected for analyses were chosen to 'fill the gaps' in existing data (Roos et al., 2002), using rigorous sampling and analytical methods, as well as to extend the data to include more species diversity. The secondary objective was to estimate the potential contribution of fish, shrimp and prawn species to recommended nutrient intakes (RNIs) during the first 1000 days of life, which means for women throughout pregnancy and lactation, and for infants from age 7 to 23 months. Specific nutrients considered are iron, zinc, calcium, iodine, vitamin A and vitamin B12, which are of known public health concern in Bangladesh ( Craviari et al., 2008;Fischer et al., 1999;ICDDRB et al., 2013). Data presented in this paper are the most comprehensive collection on the nutrient composition of important fish, shrimp and prawn species in Bangladesh, both in terms of the number of species and the nutrient components analysed, to date.

Sampling protocol
The sampling method was constrained by the nature of rural fish markets in this context, largely dependent on the activities of small-scale fishermen whose supply is unpredictable. As a result, single pooled samples of 54 fish, shrimp and prawn species commonly available during the monsoon season, were collected at local markets and fish landing sites in Mymensingh, Sylhet, Khulna and Cox's Bazar districts in Bangladesh as shown in Fig. 1, from July-September 2012. Additionally, one small fish species (Amblypharyngodon mola) was sampled both from the market (assumed to be from an inland capture source) and from a homestead pond in Dinajpur district (referred to as Mola (cultured) ). The number of fish collected for each sample was dependent on the average size of each fish species but was a total of approximately two kilograms for each sample. For small fish species (<500 g per fish), a single pooled sample of up to several hundred individual fish (to make a total sample of approximately two kilograms) was collected, for medium fish (500-750 g per fish), a single pooled sample of four individual fish, and for larger fish (>750 g per fish), a single pooled sample of two individual fish. Samples were packed in polyethylene bags at the collection site and transported in an insulated ice box lined with ice chips and away from direct sunlight, to a nearby laboratory facility.

Sample preparation
The identification details of each sample including common Bangla name, scientific name, location of sample collection and sample preparation details are shown in Table 1. In this paper, samples are referred to by the common Bangla name and are grouped according to the three dominant fish production sectors: inland capture, inland aquaculture and marine capture fisheries. Samples were cleaned by local fisher-folk to obtain raw, edible parts according to traditional practice. Depending on the fish species, edible parts may or may not include the head, viscera, scales, bones and other parts (Table 1). To avoid contamination of samples, non-metal equipment such as plastic cutting boards, buckets and strainers, and ceramic cutting knives were used to obtain raw edible parts. Fish samples were washed with deionised water after cleaning and before being packed in polyethylene bags and stored in a deep freezer at À18 8C. Frozen samples were transported in an insulated box, lined with dry ice to laboratories in New Zealand and Denmark for nutrient composition analysis. The temperature of fish samples was measured upon receipt at the testing facilities to ensure that the samples had remained frozen during transportation. Fish species were homogenised as per raw edible parts prior to analysis and subsamples of the homogenate were taken, with size appropriate for individual analytical tests (10-100 g). For several species, the homogenate included bones, and for others, bones were removed prior to homogenisation if they are typically discarded as plate waste, as shown in Table 1.

Analytical methods
The analytical methods for each nutrient component, and corresponding limits of quantitation (LOQ) and reproducibility are summarised in Table 2. 2.3.1. Analyses completed at AsureQuality Limited Laboratory, Auckland, New Zealand For proximate components (protein, fat, moisture, ash), vitamin B12 and folate, standard analytical methods as per the Association of Official Analytical Chemists (AOAC) were used, as listed in Table 2. Minerals (except iodine and selenium) were analysed using the inductively coupled plasma optical emission spectrometry (ICP-OES) method (APA et al., 2012). Iodine and selenium were analysed using the inductively coupled plasma mass spectrometry (ICP-MS) method (APA et al., 2012). Vitamin D and E were analysed using high performance liquid chromatography (HPLC) (Brubacher et al., 1985). Fatty acid composition was analysed using gas liquid chromatography (GLC) (Bannon et al., 1985).

Analyses completed at the National Food Institute, DTU, Denmark
Analyses of vitamin A, B12, D, E and folate in 29 species were carried out in Denmark where all tests were conducted in accordance with standard ISO17025 of the International Organization for Standardization, as summarised in Table 2 (ISO, 2005). Vitamin A, D and E were analysed using HPLC. Quantification of Table 1 Identification details of fish, shrimp and prawn samples and anatomical parts removed prior to analysis. vitamin A activity included all-trans-retinol, 13-cis-retinol, alltrans-3,4-dehydroretinol and 13-cis-3,4-dehydroretinol. Dehydroretinol is expected to demonstrate 40% of the biological activity of retinol (Shantz and Brinkman, 1950) and possibly up to 110% (Riabroy and Anumihardjo, 2011). For vitamin D, the CENmethod was modified to include quantitation of 25-hydroxy vitamin D3 (Jakobsen et al., 2007). Vitamin B12 was determined by microbiological assay using Lactobacillus delbrueckii as the test organism (Nord, 1960), and folate was determined using Lactobacillus casei as the test organism (CEN, 2003).

Presentation of results
All proximate components and minerals were analysed in duplicate and presented here as the mean, reported to the same number of significant figures as per original analytical results. For some samples a result of 'none detected' is given when a quantifiable result was found for one replicate but the corresponding duplicate returned a result below the LOQ. All minerals (except sulphur) were reported in metric units per kg of raw, edible parts but are presented here as metric units per 100 g raw, edible parts for ease of use. Energy was calculated using Atwater factors from assayed proximate components (Merill and Watt, 1973). Due to resource limitations, vitamins and fatty acids were analysed singly and presented here as per analytical results.
Vitamin A components are presented as mg/100 g of 13-cisretinol, 13-cis-3,4-dehydroretinol, all-trans-retinol, all-trans-3,4dehydroretinol and b-carotene and then total vitamin A in retinol activity equivalents (mg RAE/100 g) has been calculated according to the following conversion factors: 1 mg all-transretinol = 1 mg RAE, 1 mg 13-cis-retinol = 0.75 mg RAE (Ames et al., 1955), 1 mg all-trans-3,4-dehydroretinol = 0.4 mg RAE, 1 mg 13cis-3,4-dehydroretinol = 0.4 mg RAE (Shantz and Brinkman, 1950), 1 mg b-carotene = 0.08 mg RAE (Ottin et al., 2006). Species analysed for vitamin D and E at AsureQuality were reported in International Units per 100 g of raw edible parts (IU/100 g) and were converted to International System of Units (SI) units (mg/ 100 g) using the following conversion factors: vitamin D2 (mg/ 100 g) = vitamin D2 (IU/100 g) Â 0.025, vitamin D3 (mg/ 100 g) = vitamin D3 (IU/100 g) Â 0.025 and vitamin E (tocopherol) (mg/100 g) = vitamin E (tocopherol) (IU/100 g) Â 0.67 (FAO/ INFOODS, 2012). Fatty acid components are presented here as per analytical results and total n-6 polyunsaturated fatty acids (PUFA) and n-3 PUFA were calculated from the fatty acid profile. All results are presented as per 100 g raw, edible parts. The composition of nutrients of public health significance, vitamins and fatty acids, in relation to RNI's have been discussed in the results section. 2.4.1. Previously published data on nutrient composition of fish species Data on the mineral content of 13 species and vitamin A content of 20 species had previously been published using similar sampling methods and therefore these analyses were not repeated but have been included in the presentation of results here for completeness. In this pre-existing data, minerals (except selenium) were analysed by atomic absorption spectrometry (AAS), selenium was analysed using inductively coupled plasma atomic emission spectrometry (ICP-AES), and vitamin A was analysed using HPLC (Roos, 2001). Due to slight differences in methodology for mineral analysis in previous data and newly presented data, care should be taken in making comparisons across species.

Statistical analyses of results
Descriptive statistics of the data are presented including the range and mean, rounded to the same number of significant figures as original analytical results. Pearson's correlation coefficients were calculated using STATA (version 12.1, StataCorp, College Station, TX, USA), to describe the linear dependence of fat, moisture and energy for all 55 species; and ash and various minerals for 41 species for which all mineral compositions were analysed.

Calculation of potential contribution to recommended nutrient intakes
The potential contribution of each species to RNIs of nutrients of interest during the first 1000 days was calculated first by assigning an average RNI target for each nutrient as shown in Table 5, for pregnant and lactating women (PLW) to account for variations in requirements throughout the three trimesters of pregnancy and first 12 months of lactation, and for infants to account for variations in requirements throughout the period from age 7 to 23 months (FAO/WHO, 2004); then by calculating the contribution from a standard portion of each species (50 g/day for PLW and 25 g/day for infants) as a percentage of the average RNI. The nutrients of interest considered here are iron, zinc, calcium, iodine, vitamin A and vitamin B12. The RNIs for iron and zinc further vary according to estimated overall dietary bioavailability which is dependent on a number of factors including the presence of animal-flesh foods, phytates and other factors; and are therefore provided according to four and three dietary bioavailability categories, respectively. The typical Bangladeshi diet based on polished rice, fish and vegetables is assumed to fit best with criteria used to define the '10% bioavailability' category for iron, and 'moderate bioavailability' category for zinc (FAO and WHO, 2004).

Proximate composition
The energy, protein, fat, moisture and ash composition of all 55 species are shown in Table 3. The total energy content varied greatly with a range of 267-1020 kJ/100 g which is related to variation in fat content in the different species, as evidenced by a correlation coefficient of 0.98. The total protein content in fish species ranged from 11.9 to 20.6 g/100 g and can be assumed to be of high dietary quality, being an animal-source protein (WHO, 2007). The fat content ranged from 0.3 to 18.3 g/100 g. Fat generally varies much more widely than other proximate components of fish, and usually reflects differences in the way fat is stored in particular species but may also be affected by seasonal/lifecycle variations and the diet/food availability of the species at the time of sampling (Ababouch, 2005). For example, bottom dwelling species such as the indigenous major carps are typically lean fish, storing fat in the liver (Ababouch, 2005), whereas, migratory fish such as Ilish have a higher content of dark muscle which tends to be rich in fat (Alam et al., 2012). The moisture content of fish species ranged from 60.2 to 85.4 g/100 g and, as expected was negatively correlated with fat and energy content (correlation coefficient of À0.91 and À0.95 respectively). Ash content ranged from 0.7 to 5.3 g/100 g and is positively correlated with mineral content, particularly calcium, phosphorus, magnesium and zinc, with correlation coefficients of 0.98, 0.95, 0.85, and 0.74 respectively. The large variation in ash content is likely related to inclusion of bones as edible parts in some species, which would lead to higher ash content in these.

Mineral composition
The iron, zinc, calcium, iodine, selenium, phosphorus, magnesium, sodium, potassium, manganese, sulphur and copper composition for all species are shown in Table 4.

Iron
Iron content varied considerably with a range from 0.34 to 19 mg/100 g and a mean value of 2.6 mg/100 g. Three species of fish and one species of prawn were identified that would meet !25% of the RNI for PLW and infants: Chapila, Darkina, Mola and Najari Icha (Table 5). These results show a greater range in iron content compared to a values reported in the global FAO/INFOODS database on fish and shellfish (excluding molluscs) (FAO/INFOODS, 2013). Of interest is that iron content of cultured Mola (19 mg/ 100 g) is much higher than previously reported values for capture Mola (5.7 mg/100 g). This may be partly attributable to sampling variability, methodological differences in analysis of iron content, or may reflect real differences in the accumulation of iron in this species based on differing environmental conditions. The true nature and magnitude of these differences should be further investigated. Overall, the data presented here indicate that several species (all from inland capture fisheries) may contribute significantly to dietary iron intakes in Bangladesh which is of high bioavailability as an animal-source food (FAO and WHO, 2004). This may have important policy implications given the public health significance of iron deficiency in Bangladesh, with prevalence recently estimated at 10.7% in preschool aged children and 7.1% in adult women (ICDDRB et al., 2013), and the well documented negative effects of deficiency on physical and cognitive development, pregnancy outcomes, morbidity and mortality.

Zinc
Zinc concentration varied considerably from 0.6 to 4.7 mg/ 100 g with a mean content of 1.9 mg/100 g. These results are within the range of fish and seafood reported elsewhere (FAO/ INFOODS, 2013). Four species were identified that would meet !25% of the RNI for PLW; Chela, Darkina, cultured Mola and Rani, and two species: Chela and cultured Mola, which would meet !25% of the RNI for infants, from a standard portion (Table 5). A further seven species of fish and one species of prawn (all of which are capture species) would meet 20-25% of RNIs for PLW (Dhela, Ekthute, Kachki, Kata Phasa, Mola, Najari Icha, Tengra, and Tit Punti) and a further six species of fish and one species of prawn would meet 20-25% of RNIs for infants (Darkina, Dhela, Ekthute, Mola, Najari Icha, Rani and Tit Punti). In light of recent estimates of a national prevalence of zinc deficiency in 57.3% of women and 44.6% of pre-school aged children in Bangladesh (ICDDRB et al., 2013), several SIS and prawn species could contribute significantly to dietary zinc intake, also taking into consideration that zinc in animal-source foods is highly bioavailable (FAO and WHO, 2004).

Calcium
Calcium content ranged considerably from 8.6 to 1900 mg/ 100 g with a mean content of 600 mg/100 g. These results are within the range of fish and seafood reported elsewhere (FAO/ INFOODS, 2013). As would be expected, calcium content was much higher in species in which bones are commonly consumed and included in the edible parts. Fourteen species were identified that would meet !50% of the RNI for PLW, and 18 species that would meet !50% of the RNI for infants (Table 5). Calcium deficiency nationally has not been evaluated, however, it has been implicated in the development of rickets, estimated to affect 550,000 children in 2008 (Craviari et al., 2008;Fischer et al., 1999;ICDDRB, 2009), and in a study in two rural subdistricts of Bangladesh, it was estimated no women or young children had diets adequate in calcium, attributable to low food intake and low dietary diversity (Arsenault et al., 2013). In developed countries, dairy products tend to be the primary source of dietary calcium; however, this is not the case in Bangladesh where frequency of dairy consumption is very low JPGSPH and HKI, 2012). The data presented here further support the conclusion that in Bangladesh, SIS eaten whole, with bones are a significant source of highly bioavailable dietary calcium (Larsen et al., 2000;Roos et al., 2007a,b).

Iodine
Iodine was below detectable limits in eight of the 55 species, and ranged up to 120 mg/100 g, with a mean of 22 mg/100 g. Only one species of prawn (Najari Icha) would contribute to !25% of the RNI, and one species of fish (Darkina) would contribute !20% of the RNI, for PLW and infants (Table 5). The iodine content of foods tends to be largely dependent on environmental conditions. Marine fish and seafood tend to be rich dietary sources with a mean composition of 83 mg/100 g in marine fish reported elsewhere (FAO/WHO, 2004); however, this was not particularly evident in the marine species analysed here, with a range of only 6.9-41 mg/ 100 g. This is the first study in which the iodine content of fish, shrimp and prawn in Bangladesh was analysed. The composition of iodine in inland capture species reported here was within the range of fish and seafood reported elsewhere (FAO/INFOODS, 2013), but most species are unlikely to be a significant source of dietary iodine.
3.2.5. Selenium, phosphorus, magnesium, sodium, potassium, manganese, sulphur, copper and chromium The contents of these minerals were analysed for data completeness but they are not associated with significant public health concerns currently, and therefore, their nutritional significance is not discussed here. Selenium content of foods varies significantly according to surrounding environmental conditions. The selenium content in species analysed here showed a wide range from 5 to 110 mg/100 g, consistent with data reported elsewhere (FAO/INFOODS, 2013). Phosphorus content ranged from 110 to 1000 mg/100 g, with higher composition in fish species with bones included in edible parts, also consistent with values reported elsewhere (FAO/INFOODS, 2013). The ranges of magnesium (21-57 mg/100 g), sodium (26-110 mg/100 g) and potassium (58-350 mg/100 g) content were broadly consistent with ranges for other fish and seafood reported elsewhere (FAO/INFOODS, 2013). Manganese content ranged from 0.010 to 2.8 mg/100 g and is higher than results reported elsewhere (FAO/INFOODS, 2013), which may be related to water pollution (Tö rnqvist et al., 2011). Sulphur content ranged from 160 to 300 mg/100 g and is higher than results reported in the FAO/INFOODS global database, although consistent with results reported elsewhere in the literature (Vlieg et al., 1991). Copper content ranged from 0 to 1.2 mg/100 g with highest values found in shrimp and prawn, far  Roos (2001).
n = 1 pooled sample. Table 5 Potential contribution of fish, shrimp and prawn species in a standard portion a , to average daily RNI b,c (%) for PLW d and infants (7-23 months).
a Standard portion is assumed to be 50 g/day for PLW and 25 g/day for infants. b RNI, recommended nutrient intake. c See section 2.5 for explanation of calculation of average daily RNI. d PLW, pregnant and lactating women. e mg RAE, retinol activity equivalent.
-, nutrient composition not analysed, therefore unknown contribution to RNI. f Shaded species are those that could potentially contribute to !25% of daily RNIs for PLW and/or infants for 3 or more nutrients of public health significance, if provided in a 50 g or 25 g serve, respectively. exceeding that in fish species, with 0.49 and 1.2 mg/100 g found in Harina Chingri and Najari Icha, respectively; although largely consistent with results reported for fish and seafood elsewhere (FAO/INFOODS, 2013). Chromium was undetectable in almost all species, with the exception of cultured Mola and Najari Icha which had very low concentrations of 0.027 and 0.022 mg/100 g, respectively, also consistent with data reported elsewhere (FAO/ INFOODS, 2013).

Vitamin composition
The vitamin A, B12, D, E and folate composition of fish and shrimp species is shown in Table 6.

Vitamin A
In addition to vitamin A content of 20 species originally presented by Roos (2001), data on a further 28 species (and cultured Mola) are presented in Table 6. Total vitamin A was undetected in 11 species and ranged up to 2503 mg RAE/100 g. As expected, cultured Mola fish had significant concentrations of retinol and dehydroretinol as had been identified previously in capture Mola (Roos, 2001). Three species (all SIS): Mola, Dhela and Darkina were identified that could potentially contribute !25% of RNI for PLW and infants in a standard portion. The data presented here support previous studies in Bangladesh which have identified that some SIS such as Mola have potential to play a significant role in food-based strategies to address vitamin A deficiency (Roos et al., 2007a).

Vitamin B12
The vitamin B12 content in fish species ranged from 0.50 to 14 mg/100 g (n = 49). The highest concentration, 14 mg/100 g, was found in Majhari Thai Pangas (juvenile Thai Pangas), however, this was not maintained in the adult Thai Pangas with a concentration of only 1.5 mg/100 g. Very limited data on vitamin B12 in fish and seafood are available for comparison in the literature. In the Australian food composition database, vitamin B12 content of fish and seafood ranges from 0.2 to 15.2 mg/100 g which is consistent with results reported here (FSANZ, 2010). For PLW and infants, 13 and 21 species respectively, were identified that would potentially contribute !100% of the daily RNI in a standard portion. Care should be taken however, when comparing results of vitamin B12 in species analysed by different laboratories due to differences in analytical methods. This is the first analysis of vitamin B12 composition of fish species in Bangladesh, and is of particular public health significance given the recent estimate of a national prevalence of vitamin B12 deficiency in 22% of adult women and the clear negative implications of deficiency on cognitive development and function (de Benoist, 2008). As dietary sources of vitamin B12 are exclusively animal-source foods, of which, in Bangladesh, fish is the most significant, increased consumption of fish is likely to be an appropriate food-based strategy to prevent and fight vitamin B12 deficiency.

Vitamin D
Vitamin D3 was undetected in five species and ranged up to 34 mg/100 g (n = 49). Very limited data on vitamin D in fish and seafood are available for comparison in the literature. The range reported here is greater than the range of Vitamin D3 in Australian fish and seafood at 0-20 mg/100 g (FSANZ, 2010), and similar to the range of vitamin D3 reported for selected fish and seafood in the United States at 0-33 mg/100 g (Byrdwell et al., 2013). Considering that the RNI of total vitamin D is 5 mg/day for PLW and infants, it is likely that several species could contribute significantly to dietary vitamin D intakes. Although the same analytical methods were used, comparisons between species of low vitamin D3 content (<0.1 mg/100 g) analysed in different laboratories should be made with caution due to differences in the LOQ in analysis by the two laboratories. For example, 14 species were identified with concentrations of vitamin D3 by analysis at DTU which would not have returned detectable concentrations by analysis at AsureQuality (LOQ of 0.05 mg/100 g at DTU compared to 0.5 mg/ 100 g at AsureQuality). Of the species analysed for vitamin D2 (n = 20) only five species were found to have detectable concentrations ranging from 0.39 to 2.9 mg/100 g. Vitamin D2 is however, generally only considered to be found in plant-source foods, specifically yeasts and fungi. There is evidence, however, that it is found in microalgae and zooplankton, and if this forms part of the diet of fish, may account for its presence (Rao and Raghuramulu, 1996). No species were found to have detectable concentrations of 25-hydroxyvitamin D3 (n = 29). This is the first time that vitamin D content in fish in Bangladesh has been evaluated. The data presented here indicate that some species may contribute significantly to dietary vitamin D intakes in Bangladesh and it is recommended that further analysis of both vitamin D2 and D3, using standard analytical methods be conducted.

Vitamin E
Vitamin E in the form of a-tocopherol, d-tocopherol and gtocopherol was analysed in 20 species at AsureQuality. The form of vitamin E with highest biological activity, a-tocopherol, was analysed in an additional 25 species at DTU (Table 6). Across all 45 species, a-tocopherol was undetected in four species and ranged up to 1.9 mg/100 g. Very limited data on a-tocopherol in fish and seafood are available for comparison in the literature. In the Australian food composition database, a-tocopherol content of fish and seafood ranges from 0.1 to 4.2 mg/100 g which is broadly consistent with results reported here (FSANZ, 2010). It is worth pointing out, however, that although the same method of analysis was used by the two laboratories, differences in the LOQ mean that samples tested at AsureQuality were less likely to return detectable concentrations of a-tocopherol compared to those tested at DTU (LOQ of 0.07 and 0.02 mg/100 g at AsureQuality and DTU, respectively). No species analysed for other vitamin E components were found to have detectable concentrations of d-tocopherol and only two species were found to have detectable concentrations of g-tocopherol which were Tara Baim and Shing with 0.01 and 0.04 IU/100 g, respectively (0.007 and 0.03 mg/100 g, respectively). This is the first time the vitamin E content of fish species in Bangladesh has been analysed. Considering that the daily RNI for infants ranges from 2.7 to 5.0 mg a-tocopherol equivalents/day, (no recommendation for PLW), the data presented here indicate that some fish are a potentially important source of vitamin E, particularly in the form of a-tocopherol and it is therefore recommended that further analysis, using standard methods be conducted.

Folate
Folate content was analysed in 49 species and is shown in Table 6. Folate content was below detectable limits in 17 species and ranged up to 18 mg/100 g, consistent with results reported elsewhere (FSANZ, 2010). Comparisons between species of low folate content (<8 mg/100 g) analysed in different laboratories should be made with caution due to differences in analytical methods and LOQs (LOQ of 8 and 0.2 mg/100 g at AsureQuality and DTU, respectively). For example, 20 species were identified by analysis at DTU with concentrations of folate that would not have returned detectable concentrations had they been analysed at AsureQuality. This is the first time folate has been analysed in fish species in Bangladesh. Considering that the RNI for PLW ranges from 500 to 600 mg dietary folate equivalents (DFE)/day and for infants, 80-150 mg DFE/day, the results indicate that all species analysed would generally be considered low dietary sources of folate, and therefore unlikely to contribute significantly to dietary folate intake in Bangladesh.

Fatty acid composition
All samples with a total fat content of >6 g/100 g were analysed further for composition of 38 fatty acids and the results are shown in Table 7 (in addition to juvenile Thai Pangas which had a total fat content of 1.4 g/100 g but was analysed for the purpose of comparison with its adult counterpart). Although it is recognised that fish species with fat content <6 g/100 g may well be good sources of fatty acids, due to resource constraints in this study, species with higher total fat content were prioritised for analyses. Total PUFA, monounsaturated fatty acid (MUFA) and saturated fatty acid (SFA) contents ranged from 0.5 to 3.6 g/100 g, 0.4-7.7 g/ 100 g and 0.5-8.9 g/100 g, respectively. Ilish was the most significant source of PUFA and SFA, whereas Thai Pangas was the most significant source of MUFA. The total n-3 PUFA content ranged from 211 to 2034 mg/100 g, with the most significant sources being Ilish and Parse. Total n-6 PUFA content ranged from 178 to 2157 mg/100 g, with the most significant sources being Thai Pangas and Jat Punti. The ratio of n-6:n-3 PUFA was highest in Thai Pangas, which is also the only farmed and omnivorous fish analysed here (except for its juvenile counterpart). This may reflect differences in the diet and or environmental conditions of farmed versus capture fish among other factors (Li et al., 2011), although this would require further investigation. A more balanced n-6:n-3 PUFA ratio is more desirable in prevention of cardiovascular and other chronic diseases (Simopoulos, 2008), however this evidence relates to higher disease risk with low n-3 intake, rather than high n-6 intake or a high n-6:n-3 PUFA ratio and as such, no dietary recommendation for such a ratio exists (FAO, 2010). The percentage contribution to daily average nutrient requirement of docosahexaenoic acid (DHA) for PLW and infants (7-23 months) from a standard portion of fish is shown in Fig. 2, and clearly  :0  mg  1550  710  100  580  270  150  630  430  620  52  C15:0  mg  64  36  56  100  36  98  310  160  29  nd  C16:0  mg  5780  2310  1290  3270  3270  2270  5530  2690  5180  310  C17:0  mg  36  28  88  180  55  120  160  230  31  22  C18:0  mg  1320  620  570  1120  1050  750  610  100  1560  130  C20:0  mg  25  16  28  41  39  43  26  36  35  nd  C22:0  mg  19  18  16  17  11  27  18  19  19  nd  C24:0  mg  20  22  19  42  28  41  21  42  nd  nd  C14:1  mg  11  nd  28  100  14  98  63  71  11  18  C15:1  mg  nd  nd  15  38  nd  39  21  38  nd  nd  C16:1  mg  2110  820  190  580  390  490  1930  760  160  51  C17:1  mg  10  nd  22  72  27  62  150  92  16  nd  C18:1n-6  mg  nd  nd  nd  nd  16  13  53  12  nd  nd  C18:1n-7  mg  700  280  130  310  230  270  470  370  210  48  C18:1n-9  mg  3730  1240  1790  3680  4890  2850  1490  2040  7010  180  C20:1n-9  mg  250  65  32  52  120  83  30  61  210  nd  C20:1n-11,13  mg  nd  nd  34  15  11  41  22  120  14  nd  C22:1n-9  mg  24  nd  nd  nd  47  nd  nd  nd  23  demonstrates that all species, except adult and juvenile Thai Pangas would contribute !20% of daily requirements of DHA for both PLW and infants. This is of particular interest given the growing body of literature on the role of fatty acids in growth and development during the first 1000 days, and specifically, the role of DHA in normal retinal and brain development (FAO, 2010). The data presented here indicate that important fish species in Bangladesh, particularly indigenous species are a good dietary source of fatty acids and should be considered in food-based approaches to optimise growth and development during the first 1000 days. It is also recommended that the fatty acid composition of further species be analysed.

Species size, edible parts and biodiversity
When making comparisons between species on their overall nutritional value, it is important to consider the size of the fish, and implications for what is typically considered 'edible' parts. SIS are typically consumed whole with head, bones and in some cases, viscera, in stark contrast to large species, where edible parts typically include body tissue only. The nutritional consequences of this can be seen when anatomical components of fish are analysed separately. For example, in Mola, the eyes and viscera have an extremely high concentration of vitamin A compared to the body tissue (Roos et al., 2002). Although separate anatomical parts have not been analysed here, this trend can be seen when considering the edible parts of SIS compared to that of large indigenous and aquaculture species (Table 1), and potential contributions to RNIs (Table 5). When considering iron, zinc, calcium, vitamin A and vitamin B12 requirements, there are seven species that would contribute to !25% of RNI for PLW and six species that would contribute to !25% of RNI for infants, for three or more micronutrients simultaneously, when consumed in a 50 g or 25 g portion, respectively. These species are all SIS (except for Najara Icha which is a prawn) and edible parts include head and bones. This underlines the importance of considering typical consumption patterns, often related to size of the individual fish, shrimp or prawn, in design of programmes that aim to influence production and or consumption of fish. One promising example of this, now gaining momentum, is the inclusion of nutrient-rich Mola in pond polyculture systems with carps, which is being promoted throughout rural Bangladesh (Thilsted and Wahab, 2014b,c). Table 5 also calls attention to the variation in potential nutrient contributions of different species. For example, some species contribute significantly to iron and calcium RNIs and less so to vitamin A, whereas, others contribute more to zinc and vitamin A RNIs and less so to iron RNIs. No single species is of resounding superior nutritional value than any other single species, across all nutrients, which emphasises the importance of biodiversity in fish consumption for meeting population nutrient needs.

Limitations
Nutrient composition of foods, including fish, is known to vary seasonally, depending on the stage of the life cycle, food availability, and changes in the wider environment. It was however, outside the scope of this study to attempt to sample to account for these variations. Furthermore, a relatively small size of pooled samples was used, and in the case of analysis of fatty acids, the use of single replicates. In the local context and considering resource constraints, it was also not possible to obtain larger more representative sample sizes. Another limitation of this study is the use of different methods for analysis of vitamin B12 and folate between species analysed by different laboratories, although the methods are generally considered similar (Indyk and Woollard, 2013;Indyke et al., 2002). Therefore, while it is recognised that these are limitations of the study, given the lack of existing data on nutrient composition of fish species in Bangladesh, the results are still of significant value, providing time and location specific estimates for comparison with future analyses.

Conclusions
Several species have been identified that would contribute significantly to the RNIs of multiple nutrients of public health significance. When considering the role of fish in food and nutrition security in recent decades, research, funding and interventions have largely focused on the development of aquaculture, particularly of large carps and introduced species, with an assumed benefit for nutrition-related outcomes, although this linkage is dubious. The data presented here show that from a nutritional perspective, species from inland capture fisheries, particularly small indigenous species (SIS), hold the potential to provide a much greater contribution to micronutrient intakes of vulnerable groups in the population compared to common aquaculture species. This is likely partially due to the way in which small fish are consumed, namely, whole with head and bones. Further still, given the large range in nutrient composition of the different species reported here, diversity in fish consumption, particularly of SIS, is likely to promote a more all-inclusive nutrient intake. This supports the compelling argument that to effectively target malnutrition, resources should be directed towards ensuring a more balanced approach of both sustainable capture fisheries management and aquaculture, including the development of innovative aquaculture technologies which include nutrient-rich species, in particular SIS. This paper significantly expands the current knowledge on the nutritional value of the large diversity of fish species in Bangladesh, and demonstrates that many species, particularly SIS and those from inland capture fisheries, have the potential to contribute significantly to RNIs for a variety of nutrients. In future studies, it would be useful to determine the real contribution of different species to nutrient intakes of vulnerable groups based on consumption, to better inform programmes targeting improved access, availability and consumption of nutritious foods. Fig. 2. Potential contribution to daily nutrient requirement a of docosahexaenoic acid (DHA) from a standard serve b of fish for PLW c and infants (7-23 months). (a) Daily average nutrient requirement of DHA for PLW is 200 mg/day and the adequate intake for infants is 10-12 mg/kg/day (FAO, 2010). For infants an average figure of 110 mg/day is used. This was calculated by taking the midpoint within the maximum range of adequate intakes throughout the age period (10 mg/kg/day for a 7 month old of 7.6 kg and 12 mg/kg/day for a 23 month old of 12.0 kg) where weight is estimated at the 50th percentile according to WHO growth standards (WHO, 2006). (b) Standard serve of fish for PLW is 50 g/day and for infants is 25 g/day. (c) PLW, pregnant and lactating women.