This study shows that conjoined fish twins can have a joined cardiovascular system, and that one twin can support its conjoined sibling with both oxygen and nutrients. The fact that these blood vessel connections are visible also in yolk sac larvae shows that twinning in fish can be a valuable model in which to investigate cardiovascular morphogenesis in conjoined twins. This may contribute to fill in gaps in knowledge on this subject in humans, and support the development of best practice protocols for the treatment of human conjoined twins.

The yolk sac in fish is transparent, and the cardiovascular system and blood stream can easily be studied in vivo over time in the same individual through the yolk sac larvae stage. The time from hatching until the yolk sac is absorbed in Atlantic salmon is approximately 400 day degrees, meaning 67 days at 6 °C. Indeed, the heart rate in conjoined Oreochromis mossambicus twins have been successfully studied from day 2 to 8 after hatching [15]. Several different techniques can be applied to induce twinning in fish eggs, such as elevated water temperature [16, 17], hypoxia [18], irradiation, gynogenetic inbreeding [19], induction of mutations [20], and centrifugation (reviewed in [1]). That we have observed only one pair of conjoined twins that have survived the first feeding period in our research facility during the last 35 years, and that we found 0.4 % eyed eggs with four eyes in a random population of Atlantic salmon, all hatching to become conjoined twins, supports the notion that conjoined fish twins are rare and very seldom survive the transition from yolk sac absorbtion to external feeding. Previous studies have shown the occurrence of conjoined twins at hatch to be 0.2 % in seahorses (Hippocampus guttulatus) [21], 0.07 % in pink salmon (Oncorhynchus gorbuscha) [22], 0.6 % in Arctic char [23], 0.1 % in Nile tilapia (Oreochromis niloticus) and blue tilapia (Oreochromis aureus) [24], 0.3 % in Porichthys notatus [25], and 0.05 % in coho salmon (Oncorhynchus kisutch) [26]. Although the above mentioned literature shows that twinning is not a typical observation in fish fry, there are reports on massive outbreaks; for instance there is a report on an epidemia where only 10.000 out of 153.000 eggs put down survived, and the number of conjoined twins was observed to run into the thousands [16]. This was attributed to a temperature rise in the rearing water, and/or iron contamination. Also, a twinning rate ranging between 0.5 and 4 % was reported in a study where chum salmon (Oncorhynchus keta) eggs were incubated at 18 °C, in contrast to no twins at 8 °C [17]. It is possible that we could have kept some of the specimens studied herein alive through first feeding if special care was taken, and the rearing environment adjusted as such.

Conjoined twins of brook trout (Salvelinus fontinalis) have shown to develop situs inversus viscerum – symmetry reversal of the viscera [27]. Indeed, in humans, vascular anomalies such as presence of anomalous vessels have been reported in a case study on a cadaver with situs inversus [28]. However, whether the development of anomalous vessels that connect the cardiovascular systems of conjoined salmonid twins is related to the phenomena of situs inversus is not known. Situs inversus in fish conjoined twins may mean that the v. vitellina hepaticas are closely located and can grow into each other and fuse.

The twins investigated in the present study had intact circulatory systems, inter-connected by arteries and veins. This is the first time that this phenomena has been recorded in fish, most probably since conjoined twins rarely survive past the first feeding period, after the yolk sac is absorbed. There are a few records of larger specimens of conjoined twins, always pairs with one normal individual and one parasitic individual [36]. The smallest of the twins studied in the present study, with a blocked mouth and degraded and non-functional gills can also be regarded as parasitic.

The fully developed conjoined twin specimen dissected in the present study originated from a first-generation population of Atlantic salmon (?) x Arctic char (?) hybrids that had been held at a water temperature of 10 °C during egg incubation. In Atlantic salmon, 10 °C during egg incubation is high enough to induce developmental anomalies such as aplasia of septum transversum, and 8 °C is recommended as a maximum for commercial farming [29]. Earlier studies have shown that elevated temperature may [17], or may not [23] increase the occurrence of conjoined twins in salmonids. Further, heat shock treatment for 3–4 min applied 27 min after fertilization increased the twinning rate by three or four times over that of un-shocked eggs in Nile and blue tilapia [24]. It has been suggested – in salmonids – that elevated temperature results in the accumulation of unbroken cortical vesicles that disturbs the very early stages of development [17]. There are two specific causes of twinning in fish: i) early splitting of the blastodisc due to reduced cell adhesion during the early cell cycles [20]; (ii) effects on microtubule rearrangements leading to the aberrant transport of the dorsal determinant [30].

Although several studies of fish conjoined twins have performed detailed studies on tissue organization and histology [46, 21, 26], the present study is the first on the circulatory system in fully developed fish conjoined twins. Despite the fact that the small twin in the present study expressed anomalies in most body components and tissues, the digestive tract appeared normal both with regard to gross morphology and histology. Similarly, a parasitic salmonid twin that had grown into the body wall of a normal ‘host’ twin had a normal digestive tract [3]. Here, the other body parts of the parasitic twin were largely degenerated. Indeed, developmental fate maps show that the endoderm is derived from cells that are more vegetally located in the blastula and which may join the dorsal axis at a later developmental stage [31, 32], when cells may have had a chance of converging into a single tissue (even if induced in separate locations) during gastrulation. This could explain why endodermally-derived tissues are not affected during twinning.

The specimen that was dissected in the present study had a joined circulatory system but with separate hearts. An early study on the circulatory system in conjoined human twins [7] reported several different circulatory system phenotypes in twins with a joint circulatory system, some with one heart, or with two separate hearts, and some with ‘separate’ hearts joint through a common atrium or a transverse sac. A more recent study on the cardiovascular system of five cases of conjoined human twins [9] concluded that cardiac morphogenesis in conjoined twins appears to depend on the site of the conjoined fusion. In humans, the arrangement of the cardiovascular system is largely dependent on the external morphological type [9, 33], where 75 % of thoracopagus – the most common morphological type and joined by the thorax – conjoined twins have a fused heart [34]. All the 5 types of conjoined twins defined herein have similar human phenotypes [35]. The yolk sac larvae conjoined twins in the present study showed a large variation in where along the body axis they were fused, and some only shared a common yolk sac. How twinning type relates to the arrangement of the cardiovascular system in conjoined fish twins needs to be studied in more detail in order to evaluate the suitability of fish conjoined twins as model organisms to study cardiovascular morphogenesis in conjoined twins. The different level of attachment in the herein observed conjoined twins may depend on the stage of development when the secondary fusion occurs. The attachment point will most probably move caudally with increasing time between primary fission and secondary fusion [36], meaning that the types i to iv classified herein may reflect increasing time between primary fission and secondary fusion. Unfortunately we did not genotype these types. However, the genotyping of the type v confirms that these originated from a primary fission of the early cleaving blastoderm. In the herein type v, the embryos do not fuse macroscopically, but develops into two separate individuals that undergoes ‘secondary fusion’ after the yolk sac is absorbed. However, their cardiovascular systems may be fused. In Atlantic salmon, the vitelline vein begins to spread over the yolk sac at the end of somitogenesis [36], making a fusion of the cardiovascular system after somitogenesis is completed possible in twins that share a common yolk sac. Whether this path of abnormal development (type v herein) is developmentally similar to the normal development of un-conjoined monozygotic twins in humans, is unclear. Suggesting this, is a record of 144 and 146 mm long twin embryos of Shortnose spurdog (Squalus megalops), only conjoined by a 36 mm long and 1.5 mm thick cord of embryonic tissue [37]. The point of attachment to the yolk sac in sharks is indeed much smaller compared to that in salmonids.

The small parasitic and handicapped twin studied herein would have died without the support of oxygen and nutrients from its bigger host twin. Death of the small twin would most probably be detrimental for the large twin. Hence, the development of a joined circulatory system may have been a mechanism to support life, or it may just be a developmental anomaly.