Abstract

Artemia is an industrially important genus used in aquaculture as a nutritious diet for fish and as an aquatic model organism for toxicity tests. However, despite the significance of Artemia, genomic research remains incomplete and knowledge on its genomic characteristics is insufficient. In particular, Artemia franciscana of North America has been widely used in fisheries of other continents, resulting in invasion of native species. Therefore, studies on population genetics and molecular marker development as well as morphological analyses are required to investigate its population structure and to discriminate closely related species. Here, we used the Illumina Hi-Seq platform to estimate the genomic characteristics of A. franciscana through genome survey sequencing (GSS). Further, simple sequence repeat (SSR) loci were identified for microsatellite marker development. The predicted genome size was ∼867 Mb using K-mer (a sequence of k characters in a string) analysis (K = 17), and heterozygosity and duplication rates were 0.655 and 0.809%, respectively. A total of 421467 SSRs were identified from the genome survey assembly, most of which were dinucleotide motifs with a frequency of 77.22%. The present study will be a useful basis in genomic and genetic research for A. franciscana.

Introduction

The genus Artemia (Crustacea: Branchiopoda: Anostraca), known as brine shrimp, is an aquatic invertebrate living mainly in salt lakes. To date, seven species have been assigned to the genus Artemia excluding the parthenogenetic populations called Artemia parthenogenetica [1]. Artemia species are important in aquaculture industry because their dormant cysts are easily hatched, and the nauplii can be used as a nutrient-rich food for fish [2,3]. In addition, they are also widely used as aquatic model organisms for ecotoxicity tests, along with Daphnia [4–6]. However, despite the significance of Artemia in aquaculture, genomic research is still incomplete, and the genomic characteristics are less known, compared with Daphnia, because of the relatively large estimated genome size of 0.93–2.93 Gb [7–9].

Artemia franciscana, a native species in North America, has been extensively used and introduced to other continents for commercial purposes, thereby affecting the local population’s biodiversity [10,11]. Additionally, there were some incorrect identifications, especially A. franciscana as Artemia salina, in citing the species used as test organisms in literature [12]. These problems, including population structure changes and species misidentification, suggest the need for not only morphological analyses but also population genetic studies and additional marker development to discriminate closely related species.

Genome survey sequencing (GSS) using next-generation sequencing (NGS) is a time- and cost-effective way to evaluate genome information, such as genome size, heterozygosity level, and repeat content, and can be used to develop molecular markers [13,14]. Simple sequence repeats (SSRs) or microsatellites are short tandem repeats of one to six nucleotides that have been utilized as genetic markers because of their outstanding abundance and high variability [15]. In Artemia, several microsatellite markers have already been developed for use in population genetic studies [16,17], but they are limited because they are not based on genome-wide data.

In the present study, we aimed to estimate the genomic characteristics of A. franciscana through GSS and then identify SSRs from GSS for microsatellite marker development. The present study would be useful for population genetics and molecular species identification and as a framework for subsequent whole-genome sequencing of A. franciscana.

Materials and methods

Materials and DNA extraction

Nauplii of A. franciscana, originating from the Great Salt Lake (Utah, U.S.A.), were hatched on commercial cysts (INVE Technologies NV, Dendermonde, Belgium). Cultures were maintained in 30 g/l salt water at 25°C with aeration. Live Tetraselmis sp. was fed to A. franciscana during the culture period. One egg-bearing female individual was cultured separately and the progenies were used in the subsequent experiments. Genomic DNA was extracted from whole five adults using phenol/chloroform method. The quality and quantity of the DNA were checked using a BioAnalyzer (Agilent Technologies, Santa Clara, CA, U.S.A.) and Qubit fluorometer (Invitrogen, Life Technologies, Carlsbad, CA, U.S.A.).

Genome sequencing, assembly, and K-mer analysis

Genomic DNA was randomly sheared into 350-bp fragments using an ultrasonicator (Covaris, U.S.A.). A paired-end DNA library was prepared and sequenced with Illumina Hi-Seq 2000 platform. To ensure the quality of data, adaptors, poly(N) sequences, and low-quality reads were filtered out, and only clean reads were subjected to K-mer (a sequence of k characters in a string) analysis. K-mer analysis was performed using Jellyfish 2.1.4 [18] with a K-value of 17, 19 and 25. Based on the 17-mer distribution, GenomeScope [19] in R version 3.4.4 [20] was used to estimate genome size, heterozygosity rate, and repeat content. The de novo genome assembly was carried out using Maryland Super-Read Celera Assembler (MaSuRCA) version 3.3.4 [21].

SSR detection and primer design

Genome-wide SSR identification was conducted using QDD version 3.1.2 pipeline [22]. First, the assembled sequences were used to extract microsatellite sequences with di- to hexanucleotides motifs. Next, sequences were compared using all-against-all BLAST to detect unique singleton sequences. In the last step, primer pairs were designed using two iterative methods for each sequence.

Results and discussion

Genome sequencing data statistics

In the present study, a total of 22.8 Gb of raw data for A. franciscana were generated by Illumina paired-end library (Table 1). Quality value (Q) is regarded to be correctly sequenced when Q20 and Q30 values are at least 90 and 85%, respectively [23]. The Q20 and Q30 values for the present study were 99.3 and 96.0%, respectively (Table 1); hence, the sequencing accuracy of A. franciscana was high. Additionally, the GC content of the raw data was 35.8% (Table 1).

Table 1
Statistics for the GSS data of A. franciscana
Lib IDRaw data (bp)Q20 (%)Q30 (%)GC content (%)
PE350 22814108862 99.3 96.0 35.8 
Lib IDRaw data (bp)Q20 (%)Q30 (%)GC content (%)
PE350 22814108862 99.3 96.0 35.8 

Genome size prediction

In the present study, Illumina paired-end data was used for K-mer analysis using a K-value of 17. The predicted genome size was approximately 867 Mb (Figure 1). In a previous study, the genome size of A. fransciscana based on flow cytometry was 930 Mb [9]. Our estimation and previous results, measured using different methods, were quite similar, with a difference of 63 Mb. These estimates are smaller than 1.47 Gb in A. salina and 2.93 Gb in tetraploid parthenogenetic population [7,8]. In addition, the heterozygosity and duplication rates were calculated to be 0.655 and 0.809%, respectively (Figure 1).

Distribution of K-mer (K = 17)

Figure 1
Distribution of K-mer (K = 17)

Blue bars represent the observed K-mer distribution; black line represents the modeled distribution without the error K-mers (red line), up to a maximum K-mer coverage specified in the model (yellow line). Len, estimated total genome length; Uniq, unique portion of the genome (not repetitive); Het, heterozygosity rate; Kcov, mean K-mer coverage for heterozygous bases; Err, error rate; Dup, duplication rate.

Figure 1
Distribution of K-mer (K = 17)

Blue bars represent the observed K-mer distribution; black line represents the modeled distribution without the error K-mers (red line), up to a maximum K-mer coverage specified in the model (yellow line). Len, estimated total genome length; Uniq, unique portion of the genome (not repetitive); Het, heterozygosity rate; Kcov, mean K-mer coverage for heterozygous bases; Err, error rate; Dup, duplication rate.

Genome assembly results

The results of preliminary genome assembly of A. franciscana are shown in Table 2. We obtained 122231 contigs with a total length of 841603395 bp. The maximum and N50 contig lengths were 1508123 and 14130 bp, respectively. The GC content of contigs was 35.50% (Table 2). Further assembly generated 46193 scaffolds with a total length of 938041450 bp. The maximum and N50 scaffold lengths were 2555521 and 67542 bp, respectively. The GC content of the scaffolds was 31.85% (Table 2).

Table 2
Statistics of the assembly in A. fransciscana
Total length (bp)Total numberMax length (bp)N50 length (bp)GC content (%)
Contig 841603395 122231 1508123 14130 35.50 
Scaffold 938041450 46193 2555521 67542 31.85 
Total length (bp)Total numberMax length (bp)N50 length (bp)GC content (%)
Contig 841603395 122231 1508123 14130 35.50 
Scaffold 938041450 46193 2555521 67542 31.85 

These genome survey data provide useful information for genomic research of A. franciscana and related species. However, further study combined with more advanced NGS technologies using PacBio long read sequencing and high-throughput chromosome conformation capture (Hi-C) method are necessary to improve whole genome sequencing and assembly data. If so, A. franciscana could be used in a wider range of research fields (e.g. comparative genomics) as a reference genome.

SSR loci identification

From the genome survey assembly of A. franciscana with a total length of ∼938 Mb, a total of 421467 repeat motifs were identified. The types of motifs contained 77.22% (325450) dinucleotide, 20.38% (85912) trinucleotide, 2.11% (8877) tetranucleotide, 0.20% (838) pentanucleotide, and 0.09% (390) hexanucleotide (Table 3). The percentage of dinucleotide repeats was the highest, and as the repeat motif length increased, the number of loci decreased, similar to other studies [13,24,25]. It has been suggested that longer repeats have higher mutation rates, causing instability [26,27] and shorter persistence times because of their downward mutation bias towards a reduction in repeat number [28].

Table 3
Distribution pattern of SSR motifs
Repeat motifNumber of repeatsTotal
567891011–20>20
Dinucleotide (325450) 
AC/GT 31132 11605 6082 3178 2092 1243 2171 244 57747 
AG/CT 35678 11966 7985 5895 3480 2493 11440 7824 86761 
AT/AT 79855 25592 14230 10458 6305 4932 19250 12887 173509 
CG/CG 6515 796 100 22     7433 
Trinucleotide (85912) 
AAT/ATT 21685 7466 3070 1400 510 233 1033 466 35863 
ACT/AGT 17268 6614 2388 1297 385 172 586 78 28788 
AAG/CTT 9019 1737 467 126 40 36 59 12 11496 
AAC/GTT 3136 807 203 42     4188 
ATC/GAT 1617 337 120 57     2131 
AGG/CCT 1355 142 24      1521 
ACC/GGT 767 139 24      930 
AGC/GCT 555 51 42 15     663 
CCG/CGG 176 83       259 
ACG/CGT 73        73 
Tetranucleotide (8877) 
AAAT/ATTT 1844 233 54 12     2143 
AATT/AATT 720 238 12 33     1003 
AATC/GATT 746 134 36     925 
AAAG/CTTT 564 156 48 15 12  26  821 
AGAT/ATCT 430 252 79 15 15 24  821 
ACAG/CTGT 552 175  26   765 
AATG/CATT 477 84 30    15  606 
AAGT/ACTT 344 58 35      437 
AAAC/GTTT 302 108 17      427 
ACAT/ATGT 202 116 57    18 12 405 
Others 380 60 66   12  524 
Pentanucleotide (838) 
AATAT/ATATT 68 40 12      120 
AAAAT/ATTTT 78 15 12      105 
ACTAT/ATAGT 42 24 15   12  102 
AAATT/AATTT 81 15       96 
AATTC/GAATT 66        66 
Others 259 57  15  12  349 
Hexanucleotide (390) 
AGAGCC/GGCTCT   24 37    61  
AATATT/AATATT 15 20       35 
AACAAT/ATTGTT 30        30 
AATACT/AGTATT 27        27 
AAATAT/ATATTT 24        24 
Others 126 36  27  15  213 
Total 216208 69156 35211 22631 12900 9165 34673 21523 421467 
Repeat motifNumber of repeatsTotal
567891011–20>20
Dinucleotide (325450) 
AC/GT 31132 11605 6082 3178 2092 1243 2171 244 57747 
AG/CT 35678 11966 7985 5895 3480 2493 11440 7824 86761 
AT/AT 79855 25592 14230 10458 6305 4932 19250 12887 173509 
CG/CG 6515 796 100 22     7433 
Trinucleotide (85912) 
AAT/ATT 21685 7466 3070 1400 510 233 1033 466 35863 
ACT/AGT 17268 6614 2388 1297 385 172 586 78 28788 
AAG/CTT 9019 1737 467 126 40 36 59 12 11496 
AAC/GTT 3136 807 203 42     4188 
ATC/GAT 1617 337 120 57     2131 
AGG/CCT 1355 142 24      1521 
ACC/GGT 767 139 24      930 
AGC/GCT 555 51 42 15     663 
CCG/CGG 176 83       259 
ACG/CGT 73        73 
Tetranucleotide (8877) 
AAAT/ATTT 1844 233 54 12     2143 
AATT/AATT 720 238 12 33     1003 
AATC/GATT 746 134 36     925 
AAAG/CTTT 564 156 48 15 12  26  821 
AGAT/ATCT 430 252 79 15 15 24  821 
ACAG/CTGT 552 175  26   765 
AATG/CATT 477 84 30    15  606 
AAGT/ACTT 344 58 35      437 
AAAC/GTTT 302 108 17      427 
ACAT/ATGT 202 116 57    18 12 405 
Others 380 60 66   12  524 
Pentanucleotide (838) 
AATAT/ATATT 68 40 12      120 
AAAAT/ATTTT 78 15 12      105 
ACTAT/ATAGT 42 24 15   12  102 
AAATT/AATTT 81 15       96 
AATTC/GAATT 66        66 
Others 259 57  15  12  349 
Hexanucleotide (390) 
AGAGCC/GGCTCT   24 37    61  
AATATT/AATATT 15 20       35 
AACAAT/ATTGTT 30        30 
AATACT/AGTATT 27        27 
AAATAT/ATATTT 24        24 
Others 126 36  27  15  213 
Total 216208 69156 35211 22631 12900 9165 34673 21523 421467 

Of the dinucleotides, the most frequent motif was AT/AT (53.31%), followed by AG/CT (26.66%), AC/GT (17.74%), and CG/CG (2.28%). Of the trinucleotides, the most frequent motif was AAT/ATT (41.74%), followed by ACT/AGT (33.51%) and AAG/CTT (13.38%). ACG/CGT (0.08%) was the least frequent trinucleotides motif. The most abundant motifs among the tetra-, penta- and hexanucleotides were AAAT/ATTT (24.14%), AATAT/ATATT (14.32%) and AGAGCC/GGCTCT (15.64%), respectively (Table 3).

Overall, the motifs including A or T were more abundant than those including C or G, consistent with the findings of Daphnia pulex genome-wide SSR study [29]. These results might be due to the high slippage rate of A/T motifs, addition of 3′ poly(A) tail by retrotransposon elements or transition of methylated C to T residues at CpG sites [14,30–32]. These data for SSRs in A. franciscana will be used as valuable references for the development of microsatellite markers, although further validation studies using various Artemia population are needed.

Conclusion

In the present study, the genome of A. franciscana was analyzed and assembled, and SSR loci were identified from the GSS data. The K-mer analysis (K = 17) estimated the genome size of A. franciscana to be ∼867 Mb, and the heterozygosity and duplication rates were 0.655 and 0.809%, respectively. Genome assembly results showed that contig N50 was 14130 bp, with a total length of 841603395 bp, whereas scaffold N50 was 67542 bp, with a total length of 938041450 bp. A total of 421467 SSRs were identified, of which dinucleotide motifs were the most abundant and hexanucleotide motifs were the least abundant. The present study will be a useful foundation for genomic and genetic studies on A. franciscana.

Data Availability

The A. franciscana genome project has been registered in NCBI under the BioProject number PRJNA449186. The whole-genome sequence has been deposited in the Sequence Read Archive (SRA) database under accession numbers SRS3156165 and SAMN08892388.

Competing Interests

The authors declare that there are no competing interests associated with the manuscript.

Funding

This work was a part of the project titled ‘Development of potential antibiotic compounds using polar organism resources’ [grant numbers 15250103, KOPRI Grant PM20030]; and ‘Ecosystem Structure and Function of Marine Protected Area (MPA) in Antarctica’ [grant numbers 20170336, KOPRI Grant PM19060] funded by the Ministry of Oceans and Fisheries, Korea.

Author Contribution

H.P. conceived the study. E.J., S.J.L., E.k.C., J.K., S.G.L., J.H.L., and J.-H.K. performed the genome sequencing, assembly, and analysis. E.J., S.J.L., and H.P. mainly wrote the manuscript. All authors contributed in writing and editing the manuscript, collating supplementary information, and creating the figures.

Abbreviations

     
  • GSS

    genome survey sequencing

  •  
  • K-mer

    A sequence of k characters in a string

  •  
  • NGS

    next-generation sequencing

  •  
  • SSR

    simple sequence repeat

References

1.
Asem
A.
,
Rastegar-Pouyani
N.
and
De Los Ríos-Escalante
P.
(
2010
)
The genus Artemia leach, 1819 (Crustacea: Branchiopoda). I. True and false taxonomical descriptions
.
Latin Am. J. Aquatic Res.
38
,
501
506
2.
Bengtson
D.A.
,
Léger
P.
and
Sorgeloos
P.
(
1991
)
Use of Artemia as a food source for aquaculture
.
Artemia Biol.
11
,
255
285
3.
Kolkovski
S.
,
Curnow
J.
and
King
J.
(
2004
)
Intensive rearing system for fish larvae research II: Artemia hatching and enriching system
.
Aquacult. Eng.
31
,
309
317
4.
Nunes
B.S.
,
Carvalho
F.D.
,
Guilhermino
L.M.
and
Van Stappen
G.
(
2006
)
Use of the genus Artemia in ecotoxicity testing
.
Environ. Pollut.
144
,
453
462
[PubMed]
5.
Yu
J.
and
Lu
Y.
(
2018
)
Artemia spp. model-a well-established method for rapidly assessing the toxicity on an environmental perspective
.
Med. Res. Arch.
6
,
1
15
6.
Libralato
G.
,
Prato
E.
,
Migliore
L.
,
Cicero
A.
and
Manfra
L.
(
2016
)
A review of toxicity testing protocols and endpoints with Artemia spp
.
Ecol. Indic.
69
,
35
49
7.
Vaughn
J.
(
1977
)
DNA reassociation kinetic analysis of the brine shrimp, Artemia salina
.
Biochem. Biophys. Res. Commun.
79
,
525
531
[PubMed]
8.
Rheinsmith
E.
,
Hinegardner
R.
and
Bachmann
K.
(
1974
)
Nuclear DNA amounts in Crustacea
.
Comp. Biochem. Physiol. Part B
48
,
343
348
9.
De Vos
S.
,
Bossier
P.
,
Van Stappen
G.
,
Vercauteren
I.
,
Sorgeloos
P.
and
Vuylsteke
M.
(
2013
)
A first AFLP-based genetic linkage map for brine shrimp Artemia franciscana and its application in mapping the sex locus
.
PLoS ONE
8
,
e57585
[PubMed]
10.
Muñoz
J.
,
Gomez
A.
,
Green
A.J.
,
Figuerola
J.
,
Amat
F.
and
Rico
C.
(
2008
)
Phylogeography and local endemism of the native Mediterranean brine shrimp Artemia salina (Branchiopoda: Anostraca)
.
Mol. Ecol.
17
,
3160
3177
[PubMed]
11.
Naceur
H.B.
,
Jenhanive
A.B.R.
and
Romdhane
M.S.
(
2009
)
New distribution record of the brine shrimp Artemia (Crustacea, Branchiopoda, Anostraca) in Tunisia
.
Check List
5
,
281
288
12.
Ruebhart
D.R.
,
Cock
I.E.
and
Shaw
G.R.
(
2008
)
Brine shrimp bioassay: importance of correct taxonomic identification of Artemia (Anostraca) species
.
Environ. Toxicol.
23
,
555
560
13.
Xu
S-Y
,
Song
N.
,
Xiao
S.-J.
and
Gao
T.-X.
(
2020
)
Whole genome survey analysis and microsatellite motif identification of Sebastiscus marmoratus
.
Biosci. Rep.
40
,
BSR20192252
14.
Li
J.
,
Li
S.
,
Kong
L.
,
Wang
L.
,
Wei
A.
and
Liu
Y.
(
2020
)
Genome survey of Zanthoxylum bungeanum and development of genomic-SSR markers in congeneric species
.
Biosci. Rep.
40
,
BSR20201101
15.
Buschiazzo
E.
and
Gemmell
N.J.
(
2006
)
The rise, fall and renaissance of microsatellites in eukaryotic genomes
.
Bioessays
28
,
1040
1050
[PubMed]
16.
Muñoz
J.
,
Green
A.J.
,
Figuerola
J.
,
Amat
F.
and
Rico
C.
(
2009
)
Characterization of polymorphic microsatellite markers in the brine shrimp Artemia (Branchiopoda, Anostraca)
.
Mol. Ecol. Resour.
9
,
547
550
[PubMed]
17.
Eimanifar
A.
,
Asem
A.
,
Wang
P.-Z.
,
Li
W.
and
Wink
M.
(
2020
)
Using ISSR genomic fingerprinting to study the genetic differentiation of Artemia Leach, 1819 (Crustacea: Anostraca) from Iran and neighbor regions with the focus on the invasive American Artemia franciscana
.
Diversity
12
,
132
18.
Marçais
G.
and
Kingsford
C.
(
2011
)
A fast, lock-free approach for efficient parallel counting of occurrences of k-mers
.
Bioinformatics
27
,
764
770
[PubMed]
19.
Vurture
G.W.
,
Sedlazeck
F.J.
,
Nattestad
M.
,
Underwood
C.J.
,
Fang
H.
,
Gurtowski
J.
et al.
(
2017
)
GenomeScope: fast reference-free genome profiling from short reads
.
Bioinformatics
33
,
2202
2204
[PubMed]
20.
R Core Team
(
2017
)
R: a language and environment for statistical computing
.
R Foundation for Statistical Computing. Vienna, Austria
21.
Zimin
A.V.
,
Marçais
G.
,
Puiu
D.
,
Roberts
M.
,
Salzberg
S.L.
and
Yorke
J.A.
(
2013
)
The MaSuRCA genome assembler
.
Bioinformatics
29
,
2669
2677
[PubMed]
22.
Meglécz
E.
,
Pech
N.
,
Gilles
A.
,
Dubut
V.
,
Hingamp
P.
,
Trilles
A.
et al.
(
2014
)
QDD version 3.1: A user‐friendly computer program for microsatellite selection and primer design revisited: experimental validation of variables determining genotyping success rate
.
Mol. Ecol. Resour.
14
,
1302
1313
[PubMed]
23.
Li
G.-Q.
,
Song
L.-X.
,
Jin
C.-Q.
,
Li
M.
,
Gong
S.-P.
and
Wang
Y.-F.
(
2019
)
Genome survey and SSR analysis of Apocynum venetum
.
Biosci. Rep.
39
,
BSR20190146
24.
Chen
M.
,
Tan
Z.
,
Zeng
G.
and
Peng
J.
(
2010
)
Comprehensive analysis of simple sequence repeats in pre-miRNAs
.
Mol. Biol. Evol.
27
,
2227
2232
[PubMed]
25.
Katti
M.V.
,
Ranjekar
P.K.
and
Gupta
V.S.
(
2001
)
Differential distribution of simple sequence repeats in eukaryotic genome sequences
.
Mol. Biol. Evol.
18
,
1161
1167
[PubMed]
26.
Kruglyak
S.
,
Durrett
R.T.
,
Schug
M.D.
and
Aquadro
C.F.
(
1998
)
Equilibrium distributions of microsatellite repeat length resulting from a balance between slippage events and point mutations
.
Proc. Natl. Acad. Sci. U.S.A.
95
,
10774
10778
27.
Wierdl
M.
,
Dominska
M.
and
Petes
T.D.
(
1997
)
Microsatellite instability in yeast: dependence on the length of the microsatellite
.
Genetics
146
,
769
779
[PubMed]
28.
Harr
B.
and
Schlötterer
C.
(
2000
)
Long microsatellite alleles in Drosophila melanogaster have a downward mutation bias and short persistence times, which cause their genome-wide underrepresentation
.
Genetics
155
,
1213
1220
[PubMed]
29.
Sung
W.
,
Tucker
A.
,
Bergeron
R.D.
,
Lynch
M.
and
Thomas
W.K.
(
2010
)
Simple sequence repeat variation in the Daphnia pulex genome
.
BMC Genomics
11
,
691
[PubMed]
30.
Kelkar
Y.D.
,
Tyekucheva
S.
,
Chiaromonte
F.
and
Makova
K.D.
(
2008
)
The genome-wide determinants of human and chimpanzee microsatellite evolution
.
Genome Res.
18
,
30
38
[PubMed]
31.
Tóth
G.
,
Gáspári
Z.
and
Jurka
J.
(
2000
)
Microsatellites in different eukaryotic genomes: survey and analysis
.
Genome Res.
10
,
967
981
[PubMed]
32.
Zhou
Y.
,
Bizzaro
J.W.
and
Marx
K.A.
(
2004
)
Homopolymer tract length dependent enrichments in functional regions of 27 eukaryotes and their novel dependence on the organism DNA (G+ C)% composition
.
BMC Genomics
5
,
95
[PubMed]

Author notes

*

These authors contributed equally to this work.

This is an open access article published by Portland Press Limited on behalf of the Biochemical Society and distributed under the Creative Commons Attribution License 4.0 (CC BY).