Phylogeography and connectivity of the Pseudocalanus (Copepoda: Calanoida) species complex in the eastern North Pacific and the Pacific Arctic Region

Abstract

The genus Pseudocalanus (Copepoda, Calanoida) is among the most numerically dominant copepods in eastern North Pacific and Pacific-Arctic waters. We compared population connectivity and phylogeography based on DNA sequence variation for a portion of the mitochondrial cytochrome oxidase I gene for four Pseudocalanus species with differing biogeographical ranges within these ocean regions. Genetic analyses were linked to characterization of biological and physical environmental variables for each sampled region. Haplotype diversity was higher for the temperate species (Pseudocalanus mimus and Pseudocalanus newmani) than for the Arctic species (Pseudocalanus acuspes and Pseudocalanus minutus). Genetic differentiation among populations at regional scales was observed for all species, except P. minutus. The program Migrate-N tested the likelihood of alternative models of directional gene flow between sampled populations in relation to oceanographic features. Model results estimated predominantly northward gene flow from the Gulf of Alaska to the Beaufort Sea for P. newmani. Model scenarios that allowed bidirectional gene flow between sampled populations gave the best Bayesian predictions for P. acuspes, P. mimus and P. minutus. Under current warming trends, biogeographical boundaries and barriers for Pseudocalanus species may shift, allowing habitat range expansion or contraction and resulting in altered population connectivity between Arctic and sub-Arctic populations.

INTRODUCTION

Phylogeography studies the patterns of genetic variation within and among species on a geographical scale. Specifically, it incorporates a species’ biogeographical past and how underlying forces, such as evolutionary and ecological processes, have structured contemporary geographical distributions (Avise, 2000; Knowles and Maddison, 2002). Portions of mitochondrial genes have frequently been used as genetic markers in phylogeographic studies due to high concentrations of mitochondrial DNA in eukaryotic organisms, its clonal maternal inheritance pattern and the detectable patterns of haplotype diversity within and between populations (Avise, 2000). The mitochondrial cytochrome oxidase I (COI) gene has proven to be a useful genetic marker for studies of marine planktonic copepods to discriminate and identify cryptic species (Bucklin et al., 1998, 2001; Goetze, 2003; Goetze and Ohman, 2010; Viñas et al., 2015), to understand the degree of population connectivity of cosmopolitan species (Goetze, 2005; Blanco-Bercial et al., 2011) and to measure gene flow between distinct geographic populations (Costa et al., 2014). Among other genetic markers used for population genetic and phylogeographic studies of marine copepods are the mitochondrial genes cytochrome B (Provan et al., 2009; Milligan et al., 2011) and 16S rRNA (Goetze, 2003; Nelson et al., 2009), nuclear microsatellites (Provan et al., 2009) and genomic single nucleotide polymorphisms (Brito and Edwards, 2009; Unal and Bucklin, 2010).

In general, holozooplankton are regarded as having very large population sizes and high rates of dispersal, and therefore high evolutionary potential (Peijnenburg and Goetze, 2013). The phylogeographic patterns expressed within holozooplankton populations can be affected by a number of physical mechanisms that create obstructions to gene flow. In addition to physical barriers (Blanco-Bercial et al., 2011), gene flow patterns may be determined by oceanographic features, such as gyre systems (Goetze, 2005) and physical characteristics of the water column including temperature and salinity (Yebra et al., 2011). Temperature and salinity may be the most important structuring factors for the North Pacific Ocean and the Pacific Arctic Region (PAR), which would strongly influence how species’ distributions will be altered with continued warming trends. Many zooplankton species are adapted to particular salinity and/or temperature ranges and may see either a range expansion or contraction with projected climate change scenarios. Therefore, it is necessary to understand the hydrographic controls of the study region in order to accurately model and interpret patterns of gene flow within and between zooplankton populations.

Over the past decade, there have been several studies that focused on the population genetics, connectivity and phylogeography of planktonic copepods of the North Pacific and PAR. The sibling species Calanus glacialis and Calanus marshallae, which present continuing difficulties in identification using morphological characters (Frost, 1974), have been examined using mt16S rRNA, which revealed genetic differences between populations of C. glacialis in the North Pacific and Arctic Ocean (Nelson et al., 2009). Studies have provided insights into identification of Calanus pacificus subspecies (Nuwer et al., 2008), and evolutionary processes shaping the contemporary phylogeny of the Neocalanus genus (Machida et al., 2006).

The copepod genus Pseudocalanus is comprised of seven species that co-occur as differing assemblages within their geographic ranges, which span the Arctic and temperate–boreal marine ecosystems of the Northern Hemisphere. They are herbivorous epipelagic filter–feeders that target a wide size range of food particles, such as diatoms, flagellates and coccolithophores (Poulet, 1973; Corkett and Mclaren, 1979; Cleary et al., 2015), and opportunistically feed on sea-ice algae in Arctic regions (Conover et al., 1986). Pseudocalanus are small-bodied neritic copepods with P. newmani at the smaller end and P. major at the larger end of the size spectrum (Frost, 1989). Nonetheless, prosome length alone is not a reliable taxonomic tool for species identification, due to overlapping size ranges and temperature-dependent size shifts. Pseudocalanus species display only very subtle morphological differences in the adult stage with diagnostic features dependent upon the shape of the urosomal segment containing the genital pore as well as the shape of the seminal receptacle itself (Frost, 1989). However, species of Pseudocalanus show typical levels of interspecific genetic divergence for COI sequences of copepods (10–23%) (Bucklin et al., 2003). These extremely subtle morphological differences have created immense difficulties in accurate species identification and have resulted in a general lack of detailed species-specific distribution data, with co-occurring species typically treated as a species complex and reported simply as Pseudocalanus spp.

Pseudocalanus newmani and Pseudocalanus mimus are considered as temperate species, while Pseudocalanus acuspes and Pseudocalanus minutus are Arctic species (Frost, 1989). Within the northern Gulf of Alaska (GoA), the predominant species are P. mimus and P. newmani with P. minutus present in low numbers in both the shelf region and within Prince William Sound (PWS; Napp et al., 2005). Pseudocalanus mimus is the most abundant Pseudocalanus species in the eastern North Pacific (Napp et al., 2005), and is also found to numerically dominate the outer domains of the Bering Sea (Bailey et al., 2015). Pseudocalanus acuspes, P. minutus and P. newmani numerically dominate the species complex in the shallow Chukchi (Lane et al., 2008; Hopcroft and Kosobokova, 2010; Hopcroft et al., 2010; Questel et al., 2013) and Beaufort Seas (Horner and Murphy, 1985; Darnis et al., 2008; Smoot, 2015). In the Pacific, the geographical distribution of P. acuspes is primarily restricted to the PAR and extends south into the Bering Sea (Bailey et al., 2015), an ecosystem heavily influenced by seasonal ice cover.

Molecular protocols have been developed to discriminate Pseudocalanus species based on DNA sequence variation of the COI gene (Bucklin et al., 1995, 2001, 2003), which allows reliable species discrimination and identification. Subsequent studies using this and other genetic markers have allowed researchers to gain better insights into species distribution and abundance (Bucklin et al., 2001, 2015; McGillicuddy & Bucklin, 2002; Grabbert et al., 2010; Bailey et al., 2015; Erikson, 2015), confirmation of presence/absence within a region (Aarbakke et al., 2011; Holmborn et al., 2011), demographic inferences (Aarbakke et al., 2014) and population genetic differentiation (Unal et al., 2006).

This study undertakes a comparative phylogeographic analyses of temperate and Arctic Pseudocalanus species and characterizes patterns and pathways of connectivity among populations of four species that are broadly sympatric and numerically dominant in the copepod assemblages of the eastern North Pacific and the PAR (Coyle and Pinchuk, 2003; Llinás et al., 2009; Questel et al., 2013; Ershova et al., 2015). Using COI sequence variation, we infer barriers to gene flow and provide a basis for predicting how the species’ geographic distributions and ranges may respond to climate change.

METHOD

Sample collection

Zooplankton samples were collected in 2013 as part of various oceanographic programs conducted in the northern GoA and the PAR (Fig. 1). Samples from the GoA and two fjord systems within PWS (Icy Bay and Columbia Glacier) were collected as part of the Seward Line Research Program (https://www.sfos.uaf.edu/sewardline). Samples from the PAR were collected in the Beaufort Sea during the Transboundary program (Smoot, 2015), and in the Chukchi Sea by the Chukchi Sea Environmental Studies Program (CSESP; Day et al., 2013). All samples were collected down to a maximum of 100 m using 150-µm mesh nets and preserved in 95% nondenatured ethanol following the protocols in Bucklin (2000).

Fig. 1.

Averaged current flow fields for the eastern North Pacific Ocean and the PAR. Modified, with consent, from Danielson et al. (2011). Stars represent sampling locations. BG, Beaufort Gyre; SCC, Siberian Coastal Current; BSeaW, Bering Sea Water; BSW, Bering Shelf Water; ANS, Aleutian North Slope; BS, Bering Slope; AW, Anadyr Water; ACC, Alaska Coastal Current; ACW, Alaska Coastal Water; AS, Alaskan Stream; PWS, Prince William Sound.

Pathways of transport

The GoA is a semi-enclosed sub-Arctic basin in the North Pacific Ocean that is heavily influenced by the Alaska Coastal Current (ACC), a nutrient-poor, buoyancy-driven current bound to the coastal regions of Alaska (Stabeno et al., 1995, 2004; Weingartner et al., 2005). PWS, a sub-Arctic embayment, is connected to the GoA through two main pathways: Hinchinbrook Entrance on the westernmost side and Montague Strait on the eastern side, through which the ACC enters and exits. The ACC then continues westward through the Aleutian Islands and into the Bering Sea, where it flows along the continental shelf break region (Stabeno et al., 1995). From there, currents flow in a strong northward direction through Bering Strait and across the Chukchi Sea, a shallow shelf ecosystem, in a complicated mixture of water masses (Coachman and Aagard, 1988; Weingartner et al., 1998, 2013). The majority of water flowing over the northeastern Chukchi Sea shelf exits through Barrow Canyon or turns eastward at Point Barrow and flows into the Beaufort Sea (Pickart, 2004). The anticyclonic Beaufort Gyre persists over the Canadian Basin and sets up a countercurrent to the water masses entering the region from the Chukchi Sea. Additionally, water masses situated below the upper 50 m of the water column over the Beaufort Sea's continental slope reverses flow and aids in the transport of Pacific water, and zooplankton, eastward within the Beaufort Undercurrent (Aagaard, 1984; Pickart, 2004). Hence, the dominant northward-flowing currents play an important role in determining the degree of connectivity and the extent of penetration of Pacific copepods into the PAR.

Molecular analysis

Adult female Pseudocalanus were picked from preserved zooplankton samples using a Leica MZ16 or M205C dissecting microscope. Key morphological characteristics for species identification, as detailed by Frost (1989), were examined using a compound microscope. Disproportional effort was expended looking for the rarer species within each habitat. Copepods were then washed in sterile MilliQ water to remove traces of ethanol prior to DNA extraction. Total genomic DNA was extracted using a DNeasy Blood and Tissue Kit (QIAGEN), with a final elution volume of 200 µL in AE Buffer.

PCR amplification of a 710-base pair (bp) fragment of the COI gene was achieved using 5 μL of 5 × Green GoTaq^® Flexi Buffer, 2.5 μL of 25 mM MgCl₂, 0.7 μL of 10 mM dNTPs, 1 μL of each forward and reverse primer (10 μM), 0.15 units of GoTaq^® Flexi DNA Polymerase (Promega), 11.8 µL MilliQ water and 3 μL of DNA template, for a total reaction volume of 25 µL. The PCR protocol used was as follows: 94°C for 3 min, 35 cycles of 94°C for 40 s, 60°C for 40 s and 69°C for 50 s, and 1 cycle of 69°C for 7 min. The two primers used were PseudoF: 5′-TTCGAATAGAGYTAGGHMVAGY-3′ (forward) and the Folmer et al. (1994) primer HCO-2198: 5′-TAAACTTCAGGGTGACCAAAAAATCA-3′ (reverse). The forward primer, PseudoF, was designed from COI sequences obtained using the primer set LCO-1490 (5′-GGTCAACAAATCATAAAGATATTGG-3′) and HCO-2198 (Folmer et al., 1994).

PCR products were checked by electrophoresis at 100 V for 50 min on a 1% agarose/TBE gel stained with Gel Red (Biotium). Cytochrome oxidase I bands were visualized under a UV light using a UVP ChemiDoc-It² imager. PCR products from successful amplifications were purified using 2 µL ExoSAP-IT for every 5 µL PCR product and incubated at 95°C for 15 min.

DNA sequencing used the same primers as for PCR amplification and the Big Dye Terminator Ver. 3.1 kit (Applied Biosystems, Inc., ABI). The cycle sequencing protocol used was modified from Glenn and Schable (2005) and was as follows: 95°C for 1 min, 50 cycles of 96°C for 10 s, 50°C for 5 s and 60°C for 4 min, and 1 cycle of 72°C for 2 min. Sequence reactions were cleaned using the CCDB Sephadex clean-up protocol (www.dnabarcoding.org) and run on an ABI 3130 Genetic Analyzer capillary DNA sequencer. Sequences were manually checked for accurate base calling and contigs generated using the DNA sequence assembly program Sequencher Ver. 5.2.4 (Gene Codes Corp.).

The COI sequences were aligned by CLUSTAL-W (Thompson et al., 1994) using the Molecular Evolutionary Genetics Analysis (MEGA Ver. 6) software package (Tamura et al., 2013). Primers were trimmed from the ends of sequences for an initial aligned length of ~535 bp. Species’ identities for COI sequences were verified based on BLAST searches through the NCBI GenBank database (Altschul et al., 1997).

Statistical analysis

Nucleotide diversity (π) and haplotype diversity (H_d) for the COI gene were calculated using the software DnaSP Ver. 5 (Librado and Rozas, 2009). Maximum Parsimony gene trees were analyzed using the best-fit nucleotide substitution model (Tamura model) (Tamura, 1992) as determined by MEGA Ver. 6. The significance of the substitution model was estimated through 10 000 coalescence simulations under a bootstrap test of 1000 replicates. Haplotype networks were determined using Haploviewer (Center for Integrative Bioinformatics; available at http://www.cibiv.at/~%20greg/haploviewer).

A hierarchical Analysis of MOlecular VAriance (AMOVA) was used to examine population genetic structure of each species using the software Arelquin 3.5 (Excoffier and Lischer, 2010). Samples were grouped according to their respective region (i.e. ocean basin; Table I). The significance of the variance partitions, among-regions (Φ_CT), among-samples, within-regions (Φ_SC) and within-samples (Φ_ST), was determined based on 10 100 permutations. F_ST distances, reported as Φ_ST (Tamura substitution model), were calculated for all pairs of each Pseudocalanus species and tested for significance under 10 100 permutations and with α = 0.05, after sequential Bonferroni correction (Holm, 1979). Samples with four or fewer individuals were removed from the analysis while all negative Φ_ST values obtained were assumed to be zero.

Table I:

Sampling locations and numbers of individuals sequenced for Pseudocalanus spp. collected during 2013 from the eastern North Pacific and PAR

Basin	Cruise ID	Station	Bottom depth (m)	Date	Lat (°N)	Long (°W)	P. acuspes	P. minutus	P. mimus	P. newmani
Beaufort	TB13	A1-50	50	21 August 2013	70.043	−141.1363	12
		A2-200	207	19 August 2013	70.512	−142.1004	42	13	1	13
		A2-1000	997	18 August 2013	70.628	−142.2088		2	1	10
		A6-50	50	14 August 2013	70.672	−146.1369	6	3		2
		A6-200	200	14 August 2013	70.889	−146.0859	18	21		8
		A6-1000	1004	17 August 2013	71.014	−146.1102	38	13	1	3
Chukchi	WWW1304	BF007	43	25 September 2013	71.241	−163.4092	39	9	2	9
		KF007	40	21 September 2013	70.772	−165.6299	62	14		9
		KF011	40	23 September 2013	70.895	−166.0141	11	5		4
		KF017	41	22 September 2013	71.021	−165.6390	48	10		8
		SF020	38	30 September 2013	71.994	−164.1493	11
		TF004	42	3 October 2013	71.247	−164.1828	6			4
PWS	TXF13	CG	193	7 May 2013	60.984	−147.0793	1	28	52	36
		Icy Bay	123	8 May 2013	60.241	−148.3303	1	33	33	63
GoA	TXF13	GAK1	271	5 May 2013	59.841	−149.4697	1	37	17	13
		GAK4	201	5 May 2013	59.402	−149.0572	1		12	13
		GAK8	289	4 May 2013	58.787	−148.4752			10	13
Total							297	188	129	208

Gene flow between populations of each Pseudocalanus species from the GoA, PWS and PAR was modeled using the coalescent-based program Migrate-N Ver. 3.6.11 (Beerli, 2012). Migrate-N uses ratios of maximum likelihood or Bayesian inference to estimate migration rates and effective population size (N_E) under the assumption of asymmetrical migration rates at different subpopulation sizes (Beerli and Felsenstein, 2001; Beerli, 2004, 2006). Custom migration models were constructed for each species of Pseudocalanus and structured based on the number of sequences obtained per area, as well as regional geography and hydrography.

For P. acuspes, Migrate-N model scenarios tested were as follows: South-to-North, North-to-South and Full (Fig. 2). The South-to-North model tested gene flow from the Chukchi Sea into the Beaufort Sea; the North-to-South scenario tested the reverse flow. The Full model scenario allowed for bidirectional gene flow between both the Chukchi and Beaufort Seas. Model scenarios tested for P. minutus and P. newmani were: All-North, Part-North, South and Full (Fig. 2). The All-North scenario tested gene flow in a northward flow pattern, with restrictions placed on exchange from more northern populations to southern ones (e.g. Chukchi Sea flowing into the GoA). The Part-North scenario modeled gene flow with unidirectional flow from the GoA into the Chukchi Sea, with bidirectional flow between the Chukchi and Beaufort Seas, as well as between the GoA and PWS populations. The South model was set up to test gene flow from the Beaufort Sea, into the Chukchi Sea and subsequently into the GoA, with bidirectional gene flow between the GoA and PWS populations. The Full model scenario allowed gene flow between all populations, but with restrictions due to geographical constraints. Those restrictions isolated the Beaufort Sea from the GoA and PWS populations, and the Chukchi Sea from the PWS population. Lastly, model scenarios tested for P. mimus were: Out-PWS, In-PWS and Full (Fig. 2). The Out-PWS model tested gene flow between Icy Bay and Columbia Glacier—and subsequently out of PWS and into the GoA—whereas the In-PWS model tested gene flow from the GoA into Icy Bay and Columbia Glacier and between the two fjord systems.

Fig. 2.

Conceptual representation of each Migrate-N model scenario tested for P. minutus and P. newmani (A), and P. acuspes and P. mimus (B) in the eastern North Pacific and PAR.

Parameters for each Migrate-N model run were kept at the default settings with the following exceptions: (i) parameter start settings for theta (θ) and migration rates (M) used the Mode values from the posterior distributions of an initial run's F_ST-based θ and M; (ii) the SLICE sampler method was used for Bayes-proposals for both θ and M and (iii) long-chain values (1–3) were tested for optimal posterior distributions. We report the Bayes factor predictions for custom model scenarios for each species.

RESULTS

A total of 822 COI sequences were obtained for the four Pseudocalanus species collected from the GoA, PWS and the Chukchi and Beaufort Seas (Table I; GenBank accession nos KU141424–KU142246). The aligned sequence length used for the analyses ranged from 518 to 536 bp (Table II). Pseudocalanus minutus and P. newmani were found throughout the entire study region, whereas P. acuspes was restricted to the PAR and P. mimus was confined to the eastern North Pacific sampling location. Few individuals of P. acuspes and P. mimus were sequenced from the eastern North Pacific and the PAR, respectively (Table I), despite considerable effort to find them within our collections. The presence of P. mimus in the Arctic Ocean has also been confirmed through sequencing the 28S ribosomal RNA gene (GenBank accession no: EF460783).

Table II:

Summary of intraspecific variation for COI sequences for Pseudocalanus from the eastern North Pacific and PAR during 2013

Affinity	Species	N	BP	H	π	SD	Hd	SD
Arctic	P. acuspes	297	518	59	3.3 × 10−3	3.6 × 10−4	0.661	0.032
	P. minutus	188	535	13	1.6 × 10−3	1.3 × 10−4	0.606	0.027
Temperate	P. mimus	129	536	52	7.8 × 10−3	4.3 × 10−4	0.914	0.020
	P. newmani	208	529	54	6.9 × 10−3	6.6 × 10−4	0.807	0.020

N, number of individuals sequenced; BP, base pair sequence length; H, number of haplotypes; H_d, haplotype diversity; SD, standard deviation; π, nucleotide diversity.

In all, 178 unique haplotypes were identified among the four species of Pseudocalanus over the five sampling regions (Table II). Nucleotide diversity (π) and haplotype diversity (H_d) were lower for the Arctic species P. acuspes and P. minutus (3.6 × 10⁻⁴ and 1.3 × 10⁻⁴; 0.66 and 0.61, respectively), and higher for the temperate species P. mimus and P. newmani (7.8 × 10⁻⁴ and 6.9 × 10⁻⁴; 0.91 and 0.81, respectively; Table II). There were numerous haplotypes for all species, with a range of frequencies for P. mimus, and one (in the case of P. acuspes) or two (P. minutus and P. newmani) major haplotypes, with several less frequent or unique haplotypes (Fig. 3). Of the 188 P. minutus sequenced, only 13 haplotypes were detected; however, haplotype diversity was almost equal to that expressed for P. acuspes. Fifty-two unique haplotypes were identified for P. mimus, with only one observed at all five sampling locations, which occurred most frequently (50%) in the Columbia Glacier sample. The two dominant haplotypes for each of P. minutus and P. newmani were found to be relatively abundant in all five sampling locations, with fairly equal representation exhibited by each haplotype.

Fig. 3.

Cytochrome oxidase I haplotype networks for P. acuspes, P. minutus, P. mimus and P. newmani. Each circle represents a unique haplotype; sizes are scaled to the number of individuals expressing that particular haplotype. Each node represents a single bp mutation.

Pairwise Φ_ST values for samples collected across the four regions showed comparisons between the Chukchi and Beaufort Seas to be significant for P. acuspes (Table III). Significant values were also observed for P. newmani when Beaufort Sea samples were compared against samples from the GoA and PWS (Table IV). No significant values were observed between samples of P. minutus (Table V), while P. mimus showed only one significant comparison between one GoA sample and Columbia Glacier (Table VI). Analyses of molecular variance showed significant differentiation of regional populations for P. acuspes, P. mimus and P. newmani (Table VII); no significant differentiation at this scale was found for P. minutus. Overall, the greatest amount of variance was explained by individuals within samples for all four species, yet none of the variance components (i.e. variance among regions, among samples within a region or individuals within samples) were statistically significant for P. minutus.

Table III:

Pairwise Φ_ST distances between samples of P. acuspes from the eastern North Pacific and PAR during 2013

P. acuspes	Beaufort Sea					Chukchi Sea
Samples	A1-50	A2-200	A6-50	A6-200	A6-1000	BF007	KF007	KF011	KF017	SF020	TF004
A1-50		0.215	0.902	0.001	0.001	0.000	0.000	0.000	0.000	0.000	0.061
A2-200	0.016		0.242	0.130	0.008	0.000	0.000	0.051	0.000	0.047	0.264
A6-50	0.000	0.024		0.004	0.017	0.000	0.000	0.000	0.000	0.001	0.034
A6-200	0.113	0.023	0.199		0.907	0.282	−0.003	0.464	0.139	0.336	0.057
A6-1000	0.168	0.051	0.247	0.000		0.410	−0.002	0.507	0.271	0.323	0.062
BF007	0.247	0.092	0.375	0.004	0.000		0.657	0.580	0.253	0.412	0.007
KF007	0.310	0.127	0.414	0.024	0.013	0.000		0.377	0.118	0.436	0.007
KF011	0.142	0.049	0.294	0.000	0.000	0.000	0.005		0.457	1.000	0.096
KF017	0.249	0.105	0.343	0.014	0.005	0.005	0.007	0.000		0.582	0.018
SF020	0.137	0.054	0.305	0.006	0.000	0.000	0.000	0.000	0.000		0.088
TF004	0.103	0.019	0.220	0.138	0.161	0.290	0.320	0.170	0.236	0.216

Φ_ST values are below and P values are above the diagonal. Bold numbers indicate significant values after sequential Bonferroni correction (α = 0.05).

Table IV:

Pairwise Φ_ST distances between samples of P. newmani from the eastern North Pacific and PAR during 2013

P. newmani	Beaufort Sea			Chukchi Sea			PWS		GoA
Stations	A2-200	A2-1000	A6-200	BF007	KF007	KF017	CG	IB	GAK1	GAK4	GAK8
A2-200		0.721	0.079	0.011	0.001	0.006	0.000	0.000	0.000	0.000	0.000
A2-1000	0.000		0.056	0.005	0.000	0.004	0.000	0.000	0.000	0.000	0.000
A6-200	0.140	0.171		0.544	0.287	0.512	0.558	0.013	0.112	0.129	0.116
BF007	0.239	0.296	0.000		0.201	0.574	0.429	0.056	0.345	0.132	0.051
KF007	0.423	0.480	0.087	0.066		0.900	0.742	0.476	0.668	0.991	0.786
KF017	0.314	0.370	0.000	0.000	0.000		0.926	0.394	0.607	0.761	0.484
CG	0.347	0.393	0.000	0.000	0.000	0.000		0.061	0.563	0.447	0.436
IB	0.553	0.605	0.153	0.071	0.000	0.000	0.020		0.970	0.225	0.447
GAK1	0.449	0.508	0.105	0.028	0.000	0.000	0.000	0.000		0.695	0.698
GAK4	0.433	0.486	0.097	0.052	0.000	0.000	0.000	0.013	0.000		1.000
GAK8	0.488	0.552	0.154	0.119	0.000	0.000	0.000	0.000	0.000	0.000

Φ_ST values are below and P values are above the diagonal. Bold numbers indicate significant values after sequential Bonferroni correction (α = 0.05).

Table V:

Pairwise Φ_ST distances between samples of P. minutus from the eastern North Pacific and PAR during 2013

P. minutus	Beaufort Sea			Chukchi Sea				PWS		GoA
Samples	A2-200	A6-200	A6-1000	BF007	KF007	KF011	KF017	CG	IB	GAK1
A2-200		0.418	0.694	0.614	0.604	0.243	0.747	0.317	0.218	0.736
A6-200	0.000		0.375	0.698	0.292	0.716	0.214	0.101	0.032	0.092
A6-1000	0.000	0.000		0.610	0.833	0.335	0.908	0.801	0.732	0.801
BF007	0.000	0.010	0.000		0.524	0.494	0.614	0.432	0.221	0.435
KF007	0.000	0.000	0.000	0.000		0.212	0.822	0.845	0.740	0.744
KF011	0.041	0.000	0.090	0.000	0.142		0.156	0.146	0.054	0.133
KF017	0.000	0.033	0.000	0.000	0.000	0.156		0.738	0.893	0.733
CG	0.000	0.041	0.000	0.000	0.000	0.111	0.000		0.694	0.506
IB	0.020	0.085	0.000	0.031	0.000	0.228	0.000	0.000		0.449
GAK1	0.000	0.039	0.000	0.000	0.000	0.120	0.000	0.000	0.000

Table VI:

Pairwise Φ_ST distances between samples of P. mimus from the eastern North Pacific and PAR during 2013

P. mimus	PWS		GoA
Stations	CG	IB	GAK1	GAK4	GAK8
CG		0.067	0.001	0.844	0.856
IB	0.022		0.006	0.553	0.409
GAK1	0.140	0.105		0.027	0.017
GAK4	0.000	0.000	0.141		0.978
GAK8	0.000	0.000	0.178	0.000

Φ_ST values are below and P values are above the diagonal. Bold numbers indicate significant values after sequential Bonferroni correction (α = 0.05).

Table VII:

Analysis of MOlecular VAriance (AMOVA) for Pseudocalanus species from the eastern North Pacific and PAR during 2013

	Source of variation	DF	Sum of squares	Variance components	Percent of variation	Fixation indices	P value
Arctic	P. acuspes
	Among regions	1	8.672	0.0460	5.45	ΦCT = 0.0545	0.0306 ± 0.0018
	Among samples	9	15.363	0.0395	4.68	ΦSC = 0.0495	0.0008 ± 0.0003
	within regions
	Within samples	282	213.791	0.7581	89.86	ΦST = 0.1014	0.0000 ± 0.0000
	Total	292	237.826	0.8437
	P. minutus
	Among regions	3	1.619	0.0048	1.17	ΦCT = 0.0117	0.2354 ± 0.0038
	Among samples	6	2.125	−0.0035	−0.85	ΦSC = −0.0086	0.4866 ± 0.0054
	within regions
	Within samples	174	70.555	0.4055	99.68	ΦST = 0.0032	0.4172 ± 0.0047
	Total	183	74.299	0.4068
Temperate	P. mimus
	Among regions	1	3.074	−0.0355	−1.65	ΦCT = −0.0165	0.6045 ± 0.0049
	Among samples	3	14.349	0.1245	5.80	ΦSC = 0.0571	0.0063 ± 0.0008
	within regions
	Within samples	119	244.690	2.0562	95.85	ΦST = 0.0415	0.0046 ± 0.0006
	Total	123	262.113	2.1453
	P. newmani
	Among regions	3	65.476	0.446	21.9	ΦCT = 0.2190	0.0210 ± 0.0014
	Among samples	7	18.015	0.0672	3.30	ΦSC = 0.0423	0.1536 ± 0.0036
	within regions
	Within samples	184	280.231	1.5230	74.79	ΦST = 0.2521	0.0000 ± 0.0000
	Total	194	363.723	2.0363

Samples are grouped by region (Chukchi Sea, Beaufort Sea, GoA, Icy Bay and Columbia Glacier). Bold numbers indicate significant values (P = 0.05). DF, degrees of freedom.

Low numbers of P. acuspes were found in the GoA and PWS, which constrained the Migrate-N models of gene flow to the PAR. Similarly, low numbers of P. mimus were found in the PAR, which constrained the model scenarios of gene flow to the eastern North Pacific. Marginal likelihood outputs predicted the Full model to be the best pathway for gene flow between sampled populations of P. acuspes and P. mimus (Table VIII). These results indicated strong bidirectional gene flow between the Chukchi and Beaufort Seas for P. acuspes and between the GoA and PWS for P. mimus. Marginal likelihood outputs predicted the Part-North model, simulating gene flow from the GoA north through the Chukchi Sea and into the Beaufort Sea, to be the best fit for sampled populations of P. newmani (Table VIII). All model results for P. acuspes, P. mimus and P. newmani gave good Bayesian unimodal posterior distributions. Marginal likelihood outputs predicted the Full model, which allowed bidirectional gene flow between sampled populations with geographical restrictions, to be the best fit for P. minutus (Table VIII). However, Bayesian posterior distributions were not unimodal, but instead showed noisy distributions for all migration rates tested. Models were further tested using Metropolis-Hastings as the Bayes-proposals and longer-running models, with long-chain increments set to 10 000. These parameter changes did not result in cleaner or more unimodal posterior distributions.

Table VIII:

Bayesian predictions for custom migration models using Migrate-N for Pseudocalanus from the eastern North Pacific and PAR during 2013

	P. acuspes	Full	South-to-North	North-to-South
Arctic	Bezier ImL	−1841.328	−1979.752	−1965.854
	LBF (Besier)	0.000	−138.423	−124.526
	Model Probability	1.000	0.000	0.000
	Choice	1 (best)	3	2
	P. minutus	Full	All-North	Part-North	South
	Bezier ImL	−972.779	−983.676	−978.970	−1145.543
	LBF (Besier)	0.000	−10.897	−6.190	−172.764
	Model Probability	0.998	0.000	0.002	0.000
	Choice	1 (best)	3	2	4
	P. mimus	Full	Out-PWS	In-PWS
Temperate	Bezier ImL	−1491.687	−1565.433	−1589.876
	LBF (Besier)	0.000	−73.746	−98.189
	Model Probability	1.000	0.000	0.000
	Choice	1 (best)	2	3
	P. newmani	Full	All-North	Part-North	South
	Bezier ImL	−1679.050	−1674.089	−1669.113	−1806.162
	LBF (Besier)	−9.936	−4.976	0.000	−137.048
	Model Probability	0.000	0.007	0.993	0.000
	Choice	3	2	1 (best)	4

Bold values indicate best model choice.

DISCUSSION

An important component of forecasting how marine ecosystems will respond to climate change is understanding how individual species’ distributions and abundances will change with warming trends. Multiyear studies of the mesozooplankton communities from the Chukchi Sea have revealed differing responses on a species level in warm years versus cold years (Matsuno et al., 2011; Questel et al., 2013; Ershova et al., 2015). Specifically, Pseudocalanus spp., which numerically dominate the copepod assemblages, are least abundant in colder years (Questel et al., 2013; Ershova et al., 2015). Overall, the temperate P. newmani is the most prevalent of the four Pseudocalanus species across the Chukchi Sea (Matsuno et al., 2011; Questel et al., 2013; Ershova et al., 2015) exhibiting broader spatial distributions in warmer years and, consequently, pushing faunal barriers for Arctic species northward (Ershova et al., 2015). Interestingly, this exemplifies how even very closely related species, with similar niches and geographical distributions, elicit very different responses to changing environmental conditions.

Results from this study suggest P. newmani will likely exhibit greater resilience to climate change, as it was found in high abundances throughout the entire study region and displayed a dominant northward gene flow pattern, concurrent with local hydrography. Conversely, it is plausible that P. acuspes will experience, and perhaps already has done so, range contraction, where it would be restricted to the colder Arctic environment. Pseudocalanus acuspes was very rare in the GoA and PWS regions, being restricted to cold glacial fjords, and thus showed very low levels of population connectivity between the eastern North Pacific and PAR. Overall, results from our Migrate-N model simulations indicate high levels of connectivity between established populations of each Pseudocalanus species in the Arctic and eastern North Pacific. Shifting species boundaries will increase the need to understand how climate effects will cascade through marine ecosystems. In particular, northward movement of boreal generalist species has great potential to alter Arctic food webs (Kortsch et al., 2015).

Results from the Migrate-N model scenarios indicated that there is a strong degree of population connectivity between North Pacific and Arctic populations of each of the four species, with bidirectional gene flow occurring between geographically adjacent populations. However, posterior distributions were quite noisy and resolution could not be improved by changing model parameters for P. minutus. These results could indicate that either this species has extremely high rates of contemporary gene flow or that it recently went through a population bottleneck, for which single locus mitochondrial data would not be informative enough on their own to definitively model gene flow. In the North Atlantic, P. minutus had the lowest level of genetic structuring of the Pseudocalanus species studied, with most of the variance explained by within-population comparisons (Aarbakke et al., 2014). Bayesian skyline tests indicated that these populations of P. minutus underwent population expansion during the current interglacial 25 000–10 000 YBP (Aarbakke et al., 2014). Pseudocalanus minutus was found to be evenly distributed across the Bering Sea, with highest abundances occurring over the shelf region (Bailey et al., 2015). Nucleotide and haplotype diversities expressed within that region were comparable to those found in our study 2.01 × 10⁻³ and 0.670 (Bailey et al., 2015) versus 1.6 × 10⁻³ and 0.606, respectively.

Pseudocalanus acuspes appeared to be the most abundant of the four species in the Arctic populations. Despite sharing one highly frequent haplotype, populations of this species showed significant differentiation between the two regions in this study. Our results differ from a prior study, which showed no significant differences among populations (Sevigny et al., 1989). This discrepancy is most likely the result of the reduced ability of allozymes to resolve population structure, as well as analyzing individual P. acuspes from a relict population (Bedford Basin) in the North Atlantic that has been genetically isolated since the last glacial maximum. The prospect of a relict population was also observed in the Gulf of Finland, where mitochondrial sequence data revealed a high degree of variance among samples of P. acuspes, yet populations exhibited extremely low mitochondrial diversity (H_d = 0.024) (Aarbakke et al., 2014), suggesting that these populations started to diverge 200 000–50 000 YBP.

Pseudocalanus mimus and P. newmani were found throughout the samples collected from the eastern North Pacific. However, the lack of P. mimus within the Chukchi and Beaufort Seas was most likely a result of the abnormally cold temperatures experienced in the PAR for the 2013 open water season (Weingartner et al., 2014), which may have impeded the survival and reproduction of this species on the Bering Sea shelf. Nucleotide and haplotype diversities were high for all populations of P. mimus and P. newmani, complementing the patterns observed for the Bering Sea populations (Bailey et al., 2015). Aarbakke et al. (2014) and Sevigny et al. (1989) both observed high levels of sequence diversity for P. newmani, with approximations that populations in the North Atlantic have remained stable in size for over 250 000 years (Aarbakke et al., 2014).

Estimates of gene flow for the Full model scenarios for both P. minutus and P. mimus indicated that population connectivity is bidirectional between the two sampled fjord systems of PWS, as well as within the northern GoA. This flow scenario was also incorporated into the accepted Part-North model for P. newmani. These patterns complement the regional hydrographic flow, where the ACC from the GoA enters PWS through Hinchinbrook Entrance and exits through Montague Strait. During the summer months, hydrographic flow decreases in intensity, causing flow to reverse direction (Halverson et al., 2013).

The portion of the Part-North model scenario for P. newmani simulating northward gene flow from the GoA through the Chukchi Sea and into to the Beaufort Sea is also consistent with the dominant northward flow of Pacific water through the Bering Sea and across the Chukchi Sea. However, this same model scenario was not the best choice for P. minutus. Instead, the model scenario that allowed for southward gene flow from the Chukchi Sea to the GoA gave the best Bayesian predictions. Explanations for these results could be attributed to the periodic southward flow of water through the Bering Strait during the winter months (Woodgate et al., 2005), as well as the low diversity within P. minutus populations, which seemingly resulted in multimodal posterior distributions in the Migrate-N model simulations. Therefore, these results should be considered with caution for P. minutus. Model scenarios for P. acuspes, P. minutus and P. newmani supported bidirectional gene flow between the Beaufort and Chukchi Seas, which can be accredited to the juxtaposition of the anticyclonic Beaufort Gyre and the eastward flowing Chukchi Sea water masses setting up a countercurrent flow system in the region.

The moderate levels of haplotype diversity among Arctic species observed in this study may be a direct reflection of population size, where Arctic species have smaller population sizes and ranges than temperate species, thus resulting in lower haplotype diversities. This phenomenon has been observed in other marine copepods (e.g. Calanus finmarchicus and Nannocalanus minor), which display low mitochondrial diversity and small effective population sizes, most likely resulting from a population contraction during the last glacial maximum (Bucklin and Wiebe, 1998). The low nucleotide diversity, low numbers of alleles over a large number of individuals and moderate H_d recorded for P. minutus in this study are likely indications that the evolutionary history of this species entailed genetic isolation or a population bottleneck associated with the closing of the Bering Strait during the Pleistocene Ice Ages (1.6 MYA to 10 000 YBP) due to lower sea level exposing the Bering Land Bridge (Sancetta, 1983).

The phylogeographic analyses presented here confirm the biogeographic distribution of four sibling species of Pseudocalanus, which live sympatrically in the eastern North Pacific and PAR (Frost, 1989). Estimates of population connectivity based on COI sequence variation indicated that Pseudocalanus species inhabiting the eastern North Pacific and PAR show high levels of gene flow. Similarly, Aarbakke et al. (2011) observed a high degree of connectivity among populations of P. moultoni in the North Atlantic/Arctic sector. Strong gene flow despite large geographic distances has been observed among the cosmopolitan copepod Clausocalanus spp. (Blanco-Bercial et al., 2011) and Calanus sinicus (Huang et al., 2014).

Our analyses also revealed gene flow patterns for P. acuspes, P. newmani and P. mimus that agree with local hydrographic flow. For instance, gene flow models indicate a northward flow for populations of P. newmani in the GoA through the Chukchi Sea and into the Beaufort Sea, reflecting the dominant northward flow of water masses through the Bering Strait and into the PAR. The temperate species P. newmani and the Arctic species P. minutus were abundant in all sampling regions, yet genetic structuring was stronger for P. newmani compared to P. minutus. Pseudocalanus mimus (temperate species) and P. acuspes (Arctic species) were found in extremely low abundances in Arctic and temperate samples, respectively. These patterns indicate that physical environmental conditions in the eastern North Pacific and PAR, including hydrography and ocean currents, serve as both pathways of exchange and barriers to gene flow for planktonic marine copepods.

FUNDING

This work was primarily supported by the Chukchi Sea Environmental Studies Program (CSESP) funded by ConocoPhillips, Anchorage, AK; Shell Exploration and Production Company, Anchorage, AK and Statoil USA E + P, Inc., Anchorage, AK, with added contributions from UAF through the Northern Gulf of Alaska Applied Research Award. We would also like to thank BOEM for supporting the Transboundary Program as well as NPRB, EVOS and AOOS for supporting the Seward Line cruises, both of which provided samples for this study. Research reported in this publication was supported by an Institutional Development Award (IDeA) from the National Institute of General Medical Sciences of the National Institutes of Health to J.M.Q. (grant number P20GM103395).