Effective population sizes of representative fish during the Quaternary period
In the present study, genome-wide heterozygosity varied among the fifteen representative fish species with high-quality genome assemblies (Fig. 2a). Interestingly, the mean heterozygous sites per nucleotide in the genome sequence of zebrafish was 1.1 × 10− 2. This finding is much higher than in other fish. The zebrafish variant was directly obtained from the Ensembl database, which integrated comprehensive results based on data from different individuals. Hence, this guaranteed the higher sequencing coverage and efficient variant data in zebrafish, possibly contributing to the higher heterozygosity.
The heterozygous rate per nucleotide in Mexican tetra was about 2 × 10− 4, representing the lowest among these examined fish species. The mean heterozygous rates observed among the three varieties of arowana were 4.5 × 10− 3, 2.7 × 10− 3 and 2.0 × 10− 3 (golden, green and red arowana), respectively. The rate was 2.0 × 10− 3 for Japanese flounder, which was greatly reduced compared with the heterozygosity in the half-smooth tongue sole (5.3 × 10− 3).
The historical Ne values of the fifteen representative fish species were inferred using the PSMC approach over a range from 10 kya up to 20 mya (Fig. 2b). The results allow us to track population changes during the Neogene period (2.58–23.03 mya) and the Quaternary period (following the Neogene Period spanning from 2.58 mya to present). Over these periods, the maximal Ne values among the fifteen examined fish species had experienced remarkable variations, ranging from 2,000,000 to 6,000,000 in blue tilapia, Nile tilapia and half-smooth tongue sole; the minimal Ne ranged between 1,000–6,000 (large yellow croaker and Mexican tetra), 50,000–10,000 (Atlantic herring, channel catfish, and spotted green pufferfish), and 100,000–200,000 (half-smooth tongue sole, green arowana, and golden arowana). Thus, the Ne values ranged in at least four orders of magnitude among various teleost species during the examined periods.
Multiple rounds of Ne increase and decline were observed in most of the examined fish species. These observations may potentially be related to the drastic variations in global climate that occurred during the Neogene and the Quaternary periods. Over several million years, climatic glaciation movements appeared in a relatively fixed order and the cycle repeated every 100,000 years or less [32]. Consequently, fish species were forced to adapt to new environmental circumstances during interglacial periods. Climatic glaciation movements may have contributed to the severe decline in population sizes, or even in some cases of extinction. During the mild interglacial periods, however, the population sizes recovered rapidly (see more details in Fig. 2b).
Ten of the fifteen examined fish species demonstrated drastic reductions in Ne at the beginning of the LGP. This either occurred just before the LGP or at the beginning of this period (Fig. 2b). For example, the green arowana reduced from approximately 180,000 individuals at the beginning of the LGP to 9,600 by the end. The stickleback experienced a dramatic five-fold reduction from approximately 1,540,000 to 31,000 individuals during the same period.
Fluctuations of Effective Population Expansions and Contractions
Eleven of the fifteen examined fish species in our analysis showed that Ne had changed dramatically over the observation time. Fluctuations of effective population expansions and contractions had considered to been a general feature of most of fishes.
More specifically, the Ne among three varieties of Asian arowana (Scleropages formosus) varied remarkably (Fig. 3). Before 2 mya, three varieties presented relatively similar Ne values. Since then, however, the green arowana had higher Ne values (100,000–240,000) than the other two varieties, reaching the first peak (240,000) at 600–800 kya; after that, it experienced a reduction to 80,000 until approximately 60 kya. The red and the golden arowana demonstrated slight variations with Ne values of 50,000–100,000 until approximately the beginning of the LGP, when the red arowana experienced an undulation of rising steadily first and then falling to 50,000 until 50 kya; however, the golden arowana experienced an opposite trend simultaneously from 70,000 decreasing to around 50,000 and then rising again to 130,000 until 50 kya. At the end of the examined period (until 10 kya), the Ne of golden arowana was approximately three-fold greater than the red arowana and approximately two-fold greater than the green arowana. In our previous study [33], it was predicted that Asian arowana and spotted gar diverged 384 mya. Since then, three varieties diverged approximately 1–4 mya to evolve as an independent lineage. Green arowana diverged from the other two varieties approximately 3.8 mya, and the latter two varieties diverged approximately 1.7 mya. The Ne variations could briefly reflect the evolutionary trajectories of this fish. Prior to 2 mya, three varieties of arowana had similar population sizes, with a slightly lower value in the red arowana. Subsequently, the green arowana diverged from the other two varieties and experienced a different evolutionary process, with a remarkable expansion in population (the first and second peaks). Although the red arowana diverged from the golden arowana approximately 1.7 mya, both varieties experienced similar changes in Ne from 1 mya.
In 2016, Liu et al. [34] completed the genome sequencing project of channel catfish (from an American population), and generated a draft reference genome assembly with a total length of 783 Mb and a scaffold N50 of 7.7 Mb. Subsequently, we reported another high-quality genome assembly of channel catfish (from a Chinese population; Chen et al., 2016), with approximately 845 Mb in total length and a scaffold N50 of 7.7 Mb (see Table 1). Although the Chinese population has been recognized as the introduction from America, genetic characteristics of the Chinese population seem to exhibit significant differences from the American counterpart after 30 years of artificial breeding in China [35]. These differences were also witnessed from the present prediction of the temporal dynamics in the Ne. The similar trend of Ne was observed from 10 to 100 kya; prior to that, the tendencies of Ne for both populations experienced dramatic differences (see Fig. 3). The heterozygosity in the Chinese population was 0.2%, which was higher than the American population (0.155%). These results confirmed the observed differences at a genomic level between the two populations.
Table 1
Summary of the examined teleost species and their genomes
Common name
|
Scientific name
|
Assembly version
|
Genome size (bp)
|
Gene number
|
Zebrafish
|
Danio rerio
|
GRCz11
|
1,373,471,384
|
25,709
|
Mexican tetra
|
Astyanax mexicanus
|
GCA_000372685.1
|
1,191,242,572
|
23,041
|
Atlantic herring
|
Clupea harengus
|
GCA_000966335.1
|
807,711,962
|
23,095
|
Japanese flounder
|
Paralichthys olivaceus
|
GCA_001904815.2
|
545,775,252
|
21,787
|
Tongue sole
|
Cynoglossus semilaevis
|
GCA_000523025.1
|
470,199,494
|
21,381
|
Three-spined stickleback
|
Gasterosteus aculeatus
|
BROAD S1
|
461,533,448
|
20,772
|
Channel catfish (BGI)1
|
Ictalurus punctatus
|
PRJNA319455 43
|
845,391,728
|
21,937
|
Large yellow croaker
|
Larimichthys crocea
|
GCA_000972845.1
|
678,938,134
|
24,418
|
Asian seabass
|
Lates calcarifer
|
GCA_001640805.1
|
668,481,366
|
24,348
|
Nile tilapia
|
Oreochromis niloticus
|
GCA_000188235.1
|
927,383,394
|
21,437
|
Asia arowana2
|
Scleropages formosus
|
GCA_001624265.1(Golden)
|
777,359,276
|
22,016
|
|
|
GCA_001624245.1(Green)
|
746,544,453
|
21,524
|
|
|
GCA_001624255.1 (Red)
|
738,407,480
|
21,256
|
Spotted green pufferfish
|
Tetraodon nigroviridis
|
TETRAODON 8.0
|
358,618,246
|
19,583
|
Fugu
|
Takifugu rubripes
|
FUGU5
|
358,618,246
|
18,505
|
Grass carp3
|
Ctenopharyngodon idella
|
---
|
1,076,149,922
|
32,785
|
Atlantic salmon
|
Salmo salar
|
GCA_000233375.4
|
2,966,890,203
|
43,899
|
Atlantic cod
|
Gadus morhua
|
gadMor1
|
832,114,588
|
20,083
|
Medaka
|
Oryzias latipes
|
ASM223467v1
|
734,057,086
|
19,669
|
Amazon molly
|
Poecilia formosa
|
GCA_000485575.1
|
748,923,461
|
23,613
|
Southern platyfish
|
Xiphophorus maculatus
|
GCA_000241075.1
|
729,662,853
|
20,367
|
Spotted gar
|
Lepisosteus oculatus
|
GCA_000242695.1
|
945,878,036
|
18,328
|
Coelacanth
|
Latimeria chalumnae
|
GCA_000225785.1
|
2,860,591,921
|
19,555
|
Tropical clawed frog4
|
Xenopus tropicalis
|
GCA_000004195.1
|
1,511,735,326
|
18,442
|
1 The high-quality genome assembly was available in GigaScience Database at http://dx/doi.org/10.5524/100212. |
2 The genome information of golden variety of Asian arowana was used for the divergence time evaluation. |
3 The genome assembly of grass carp was downloaded from the Grass Carp Genome project at http://www.ncgr.ac.cn/grasscarp/. |
4 The outgroup for construction of the divergence time tree (see Fig. 1). |
Nile tilapia had a stable Ne (around 1,000,000) until approximately 1 mya, when it increased to a peak of 2,500,000. Subsequently, tilapia remained stable for 600,000 years until approximately 220 kya, when it experienced a drastic reduction in Ne to 200,000 within the LGP (see Fig. 3). Recently, a high-quality genome assembly of another tilapia species, blue tilapia (Oreochromis aureus), was reported by us [36], which had a significant variation in population size compared to the Nile tilapia (Fig. 4). More specifically, from 6 to 2 mya the Ne of the blue tilapia presented an approximately 6-fold increase (at the peak) around 2–3 mya, when it reached 5,700,000 at 2 mya (see Fig. 4). During the following 1 million years, the Ne of blue tilapia declined rapidly. Since then, the population size of blue tilapia has never recovered, but it was steadily declining to less than 200,000. The two sequenced tilapia species diverged approximately 23.2 mya, suggesting that they experienced different evolutionary processes during the examined period. The population of blue tilapia flourished until the junction of the Neogene and Quaternary periods. However, blue tilapia experienced one cycle of population expansion and contraction (see Fig. 4). Possibly due to the decline in temperature within the LGP and the low tolerance of tilapias to low temperature, these two species had low Ne values during the LGP.
The effective population size of two sequenced flatfishes, half-smooth tongue sole and Japanese flounder, also were fluctuating. Before the LGP, the two flatfish species had similar trends of the Ne, with a steady increase over time. Both flatfishes experienced a sharp decline since the beginning of LGP. The Ne of Japanese flounder reached its lowest level around 100 kya, which was 1 kya earlier than the half-smooth tongue sole. The most significant variance between the two flatfishes occurred 60 kya, when the population of the half-smooth tongue sole was maintained at the lowest level while the population of the Japanese flounder rapidly increased to high levels (Fig. 3).
The population sizes varied significantly in Atlantic herring and spotted green pufferfish (Fig. 3) over time, especially within the LGP. The Ne values of these two fish reached peaks at 0.2–1 mya. Interestingly, the increase in populations was observed in these species during the LGP. Medaka (Fig. 3) demonstrated Ne changes within a shorter period than the other species. It is proposed that the low coverage of the SNP data set (Supplementary Table S1) may contribute to this phenomenon.
Other teleost species
Four fishes with a shorter generation time, including stickleback, zebrafish, Mexican tetra and bule tilapia mentioned above, presented similar population trends (Fig. 4). Each species showed a population rise at first and then a decline with differing times to reach peak size. The peak population of stickleback reached 1,550,000 at the beginning of the LGP; in contrast, zebrafish and Mexican tetra trended to reach peaks approximately 0.3 mya.
However, fugu, comparable with medaka, also shown Ne changes within a shorter period than the other species. Its Ne presented a rise during the LGP (Fig. 4).