Members of the genus Artemia (Crustacea, Anostraca), also known as brine shrimp, are typical inhabitants of extreme environments that have low species diversity and simple trophic structures 1. These hypersaline environments vary considerably in terms of ionic composition, climatic conditions and altitude. As a general rule, chloride-rich lakes are the most adequate for Artemia development 2; however, some strains require carbonate- (Mono lake Artemia, USA) or sulfate-rich waters (Tso Kar Lake Artemia, Tibet) for survival 1345. Conversely, some other ions may be deleterious to Artemia; potassium could be very toxic because of its occurrence with sodium 67. Since water composition is important for Artemia survival, ecosystems where it occurs were classified in three categories based on their anionic composition: chloride-, sulfate- and carbonate-rich 2.

Artemia persimilis (Piccinelli and Prosdocimi 1968) and A. franciscana (Kellogg 1906) occur in waters of the American continent; the latter is the most cosmopolitan 89. However, A. persimilis was reported in Sardinia, Italy 10. Molecular (RAPD-randomly amplified polymorphic DNA) 11 and morphometric characters 12 indicate that A. franciscana occurs in the Colombian Caribbean.

To characterize Artemia strains the aquaculture industry employ an array of evaluation tools 1314: i) Cyst and nauplii biometry: cyst biometry assists in the determination of number of cysts.g^-1; generally, 1 g from strains that produce small cysts contains more cysts.g^-1, thus usually produce more nauplii.g^-1. Similarly, naupliar biometry is also an essential tool for quality evaluation. Shorter naupliar length is important, particularly, to feed fish larvae which contrary to crustaceans, has to engulf prey in a single bite; ii) Cyst hatching characteristics: it can be affected by environmental factors, genotypical conditions and/or improper processing/storage. An acceptable cyst product should contain minimal quantities of impurities (i.e. sand, salt crystals, etc.) 13. Hatching efficiency (HE) and hatching percentage (H%) vary greatly among commercial batches and account for most of the price difference 13. However, HE may be a better criterion than H% since HE considers impurities content (i.e. empty cyst shells). Hatching values for a commercial product may be as low as 100,000 nauplii.g^-1, but might yield ideally near 300,000 nauplii.g^-1 (with H% > 90). Hatching synchrony must be high (T_s = 12–16 h), and the last nauplii should hatch within 8 h after T₉₀ 13. When T_s is low (T₀–T₁₀₀ > 10 h), first hatched nauplii will have consumed most of their energy reserves by the time that the last nauplii hatched and harvesting has ended 13; and iii) Essential fatty acid (EFA) profiles: particular attention has been given by marine larvaeculture production facilities to search for EFA rich Artemia strains, i.e. arachidonic acid:20:4(n-6) (ArA), eicosapentaenoic acid: 20:5(n-3) (EPA), and docosahexaenoic acid: 22:6(n-3) (DHA) 15. Since ArA and DHA are vital in marine fish nutrition 161718 great effort has been devoted to incorporate high levels of ArA, DHA, and high ratios of DHA/EPA/ArA in live food. ArA is generally conserved during periods of starvation in marine fishes 19, and serves as the preferred precursor for eicosanoid biosynthesis 20. Artemia fed with enriched n-3 and n-6 highly unsaturated fatty acids (HUFA) in turn results in better larval growth and survival of several marine species fed with it 182122232425. The DHA/EPA ratio is very variable in non-enriched Artemia, with values often lower than 1. Through the addition of DHA rich emulsions the DHA/EPA ratio increases up to 7 15. Enrichment success is strain dependent (i.e. particular Chinese strains), and linked to variations in DHA catabolism. Further, EFA nutritional requirements may vary between species and developmental stages 18. In white bass larvae, the optimal DHA/ArA and EPA/ArA dietary ratios have been established at 2:1 and 1:1, respectively 26. This contrasts with flat fish larvae (i.e. turbot and Atlantic halibut) which require much higher ratios of over 10:1 2728. However, high ArA levels have been implicated in the malpigmentation of various flatfish species 29. An additional consideration is the possibility of DHA deficiency in neural tissues (i.e. vision) in larvae fed fish oil-based diet, as has been observed in Atlantic halibut larvae and juvenile herring eyes 3031.

Our objective was to characterize Artemia franciscana biotopes in Colombia and to evaluate the viability of some strains for commercial exploitation. The data presented here may be of importance for the aquaculture industry to find new Artemia sources.

Along the Colombian Caribbean coast, nine potential Artemia habitats were explored with variable quantities (from very few cysts in the sediment to several pounds dispersed along the pond edges) of cysts and/or biomass and variable surface area (2.5 Tayrona to 4000 ha Manaure).

There were significant differences (Table 1) in cyst diameter (P = 0.00001), grouping strains in the following manner according to this parameter: 1) San Francisco Bay (SFB-Control ARC1258, USA), 2) Galerazamba and Tayrona, 3) Kangarú, 4) Manaure, and 5) Salina Cero and Pozos Colorados. Chorion thickness from Tayrona was the thinnest, followed by Salina Cero, Galerazamba, Manaure, SFB, Kangarú, and Pozos Colorados.

There were significant differences (Table 1) in naupliar size (P = 0.0001) where strains grouped as follows (smallest to largest): 1) Galerazamba, 2) Manaure, 3) SFB, Kangarú, and Salina Cero, 4) Pozos Colorados, and 5) Tayrona.

Overall, Manaure: had a high number of cysts.g^-1 (Table 2) and HE, but a low H% and T_s. Galerazamba: had a low number of cysts.g^-1 compared to other commercial cyst types, and a low H%, contrasting with a high HE and T_s. Salina Cero: had a high number of cysts.g^-1, but a low H%, HE and T_s. SFB (control): had a high number of cysts.g^-1, HE and H%, but a low T_s.

From the three (Cl^-, SO₄^2- and CO₃ ^2-) characteristic anions used to classify hypersaline ecosystems 2, Cl^- was the most abundant anion (Table 3) in all locations evaluated. The physicochemical parameters monitored (Table 4) presented some tendencies inherent to each site. Salinity in Pozos Colorados and Salina Cero had a tendency to maintain low salinities (rarely crystallizing), contrasting with Manaure that presented salinities close to crystallization in the evaporation portion of the salt production circuit. Similarly, pH in Galerazamba, Salina Cero and Manaure was towards the low end pH for Artemia production; while for Tayrona and Pozos Colorados it was towards the ideal pH (8.0 to 8.5). Percent O₂ saturation was overall normal in most sites, with the exception of Tayrona which was rather low in some months. Water temperature was at the upper limit in most sites, and extremely high only in Pozos Colorados. Nitrite was overall low in all sites, contrasting with high nitrate concentration in all sites. Phosphate was also low, except in Pozos Colorados where it was too high. Primary production (chlorophyll a) was rather at the low end for hypersaline ecosystems. Precipitation was high in the southern sites explored (as expected) and low at northernmost locations (dessert-like sites).

Cyst samples from locations (Table 5) where enough cysts were collected to perform FAME analysis (FAME was actually done on freshly hatched nauplii from cysts), exhibited high EPA and ArA levels comparable to control sample (SFB-ARC1258). In contrast, most cysts collected (including SFB) at different locations, and during different months, presented low DHA levels (Manaure was the only exception with high DHA levels). Some variations in EPA and ArA levels were observed in all sites, contrasting with much lower DHA levels which remained constant for all locations, except for Manaure which exhibited variable DHA levels. DHA/EPA ratio was overall very low for all sites compared to SFB cysts. All strains had a low DHA/ArA ratio, but a high EPA/ArA ratio, SFB included.

Vanhaecke and Sorgeloos 32 reported cyst diameter as small as 224 μm for the San Francisco Bay strain (California, USA), while Abatzopoulos et al. 33 reported cyst diameters as large as 330 μm for the bisexual species A. tibetiana, surpassing even the well known large cyst diameters of the polyploid parthenogenetic strains with a typical diameter near 280 μm. Cysts from Great Salt Lake (GSL-Utah, USA) have a larger cyst diameter (244.2 – 252.5 μm) compared with those from SFB (California, USA) (223.9 – 228.7 μm) 32; cyst diameters in the Colombian Caribbean are more similar to cysts from GSL than SFB (control). Thus, a cyst diameter grouping in the following order is possible for strains (smallest to largest): 1) San Francisco Bay (SFB-Control ARC1258, USA), 2) Galerazamba and Tayrona, 3) Kangarú, 4) Manaure, and 5) Salina Cero and Pozos Colorados. However, within the same species, strains present different cyst diameter as well as different chorion thickness. The chorion thickness for GSL ranges from 4.7 to 5.7 μm and for SFB it is 7.1 to 8.3 μm 32. In the case of cyst samples examined in this study, Tayrona were the thinnest, followed by Salina Cero, Galerazamba, Manaure, SFB (USA), Kangarú and Pozos Colorados (see Table 1).

The biometric analysis grouped strains according to naupliar size (Instar I) as follows (smallest to largest): 1) Galerazamba, 2) Manaure, 3) SFB, Kangarú, and Salina Cero, 4) Pozos Colorados, and 5) Tayrona. However, it is remarkable that the naupliar length measured by Vanhaecke and Sorgeloos 32 for the Galerazamba strain collected in 1977 greatly differed from that length in our study (480 ± 31.1 vs. 390.3 ± 24.5 μm); maybe because of some physicochemical effects (i.e. salinity) on the strain manifested over time, and/or food conditions and/or cysts harvested from different pond sizes (widely separated harvest sites), which have been reported 32 to also affect cyst size and chorion thickness. The same authors reported that SFB cysts produced at 180 mg.l^-1 in vitro are significantly smaller than cysts produced at lower salinities. Biometrical studies performed on several strains from different geographical origins concluded that Artemia biometrical parameters were mainly strain specific 32. These authors revealed that the length of Instar-I nauplii (both bisexual and parthenogenetic) may vary between 430 and 520 μm. Moreover, the North American Instar I nauplii (bisexual) tend to be at the lower previously given range. Thus, naupliar length for SFB (California) is between 428 ± 28.8 to 431 ± 23.7 μm, and for GSL strain (Utah) is between 486 ± 30.6 to 489 ± 29.2 μm 32. Naupliar size is non-critical for the feeding of crustacean larvae, which can capture and manipulate nutritional particles with their feeding appendages 18. In contrast, prey size is very critical for fish larvae, which do not have feeding appendages and must engulf particles. The correlation between naupliar size and fish larvae mortality indicates that 20% of the larvae die of hunger when being offered nauplii greater than 480 μm in the first stages of feeding 34. Thus, depending on the developmental stage of the cultured fish larvae, selecting an appropriate naupliar size as live feed is critical.

Overall, samples from Manaure, Galerazamba and Salina Cero, according to the cyst quality test, exhibited a relatively high number of cysts.g^-1. H% might have been affected (low in all samples evaluated) because the cyst processing method used (1/3 HP air blower with no heating element or temperature control) could not maintain a constant drying temperature/airflow, or some impurities were still present in samples. H% is dependent on degree of diapause termination, cysts energy content and amount of dead/non-viable/abortic embryos due to improper processing and/or storage 13. Furthermore, HE reflects three factors: 1) H%, 2) presence of other components (i.e. empty shells, salt, sand, cysts water content), and 3) individual cyst weight. The low hatching synchrony of cysts from Manaure and Salina Cero could be attributed to environmental factors (i.e. raining after cysts were dehydrated, salinity, etc.), and/or as mentioned before improper processing. Significant interactions have been reported among some physicochemical-biotic factors (salinity, percent O₂ saturation and chlorophyll a) and Artemia cyst production 35, factors which might affect consequently cyst quality.

The ionic analysis of all locations registered a Cl^- anion predominance (Table 3), as expected 36, because of their thalassohaline origin. All hypersaline environments analyzed are suitable habitats for A. franciscana development 2. Colombian hypersaline ecosystems sampled are similar to other American Cl^- dominant hypersaline biotopes such as Leslie saltworks (California – USA) 37, La Sal del Rey (Texas – USA) 38 and GSL (Utah – USA) 39.

In nature Artemia is found at salinity levels between 60 to 220 g.l^-1 (depending on the strain and/or species) and in neutral to alkaline waters, at temperatures generally below 34°C, and at rather low O₂ levels. The low salinities measured for several months at Pozos Colorados and Salina Cero could hinder overtime Artemia production by favoring nauplii production (tolerant to low salinities and pHs) and affecting adults survival 35. Similarly, the low end pH registered in Galerazamba, Salina Cero and Manaure could affect Artemia biomass, and even cyst production overtime. As pH decreases below 7.0 naupliar growth decreases and in adults the overall appearance deteriorates 40. The same authors concluded that the optimum pH for Artemia growth was from 8.0 to 8.5. In the case of cyst, hatching efficiency is greatly compromised at pHs below 8.0 41. The low O₂ levels registered at the Tayrona site could be attributed to the high accumulation of organic matter in the pond from surrounding vegetation. Water temperature was at the upper limit in most sites, with extreme temperatures at Pozos Colorados due to the lack of water circulation and small ponds size (< 0.5 ha). The nutrients (N:P, ideally 15:1) ratio was generally maintained within expected limits, except in Pozos Colorados where it was too high (1:1). The low primary production (chlorophyll a) determined in all sites was characteristic of hyperhaline environments. The most photosynthetically productive hypersaline environments 36 are the hypohalines and mesohalines. Furthermore, primary productivity in Salina Cero and Pozos Colorados was the highest among all sites, because of their low salinities. It is widely accepted that salinities higher than 50 g.l^-1 hinder considerably primary productivity in hypersaline ecosystems perhaps because of an ionic complex formation of the dissolved macronutrients or because of a generic biologic phenomenon of a drastic specific reduction of microalgae, also occurring at higher salinities 42. The high precipitation in the southern sites explored (Salina Cero, Galerazamba, Pozos Colorados and Tayrona) affected constantly Artemia cyst and biomass production by decreasing salinity and affecting light intensity.

The high EPA and ArA levels (Table 5) for Colombian cysts, determined by EFA analysis, compared to the control (SFB) have great potential for the aquaculture industry. In contrast, the low DHA level content in cysts from the Colombian sites (except Manaure) might be of concern if fed to marine larvae without further DHA enrichment, since DHA deficiency affects neural tissues development 3031, survival and growth 18, particularly at the larval stage. The very low DHA/EPA ratios (<0.5) for all Colombian sites were as expected with values often lower than 1, but this ratios could be increased up to 7 15 through the addition of DHA rich emulsions. All strains had a low DHA/ArA ratio, contrasting with a high EPA/ArA ratio (close to dietary ratio 2:1 and 1:1 for marine fish-white bass larvae, respectively 26). Artemia ArA storage/usage mechanism might be similar to that of other marine organisms (marine fishes) which conserve ArA even during periods of starvation 19. The observed fluctuations in ArA and EPA levels according to FAME analysis between, and even within strains (collected in different months), may be due to year-round variations in the biochemical composition of the primary producers available to adult Artemia 15.

The Colombian A. franciscana habitats analyzed are of marine (thalassohaline) origin; thus, all locations were expected to be Cl^- rich.

The relatively small nauplii are appropriate for marine larvaeculture. In contrast, the strains from Tayrona, Kangarú, Salina Cero, and Pozos Colorados may be of use but limited to Artemia small biomass production quantities, because of the small surface area of their respective locations; Artemia could be exploited at these locations for local aquaculture applications.

In general, cyst quality evaluation for Manaure, Salina Cero and Galerazamba strains determined that cysts from these three locations could improve their quality by concentrating efforts on cyst processing techniques. Further, cyst quality might have been affected by interactions among some physicochemical-biotic factors and Artemia cyst production in their ecosystem which could be improved by managing some of these key physicochemical-biotic factors and/or infrastructure management (e.g. brine concentration in the different basins and nutrients).

EFA profiles demonstrated that Tayrona, Galerazamba, Manaure and Salina Cero strains are suitable for marine aquaculture because of their high EPA/ArA ratio, but might have to be fortified with DHA rich emulsions depending on the nutritional requirements of the species to be cultured, because of their overall low DHA content.

WNC co-designed and carried out the experiment, participated in data collection, performed data analyses, and wrote manuscript.

GCD, LCH, OCR, JGL and IMM participated in data collection.

OCR evaluated cysts quality.

PS co-designed the experiment, FAME analysis performed at Artemia Reference Center.