The Southern Polar Front as a key to mesoplankton migratory behavior

Diel and seasonal vertical migrations of zooplankton represent a widespread phenomenon occurring in marine and freshwater environments. Diel migrations are panoceanic, while seasonal migrations usually occur in temperate and polar areas. This paper describes differences in the diel and seasonal vertical migrations in the Drake Passage north and south of the Polar Front (PF). We analyzed material of 85 stations collected in spring of 2008 and 2010 (October–November) and in summer of 2010 and 2011 (January) within the 0–300 m depth range during various time of a day. At each station we sampled the upper mixed (UL), the middle (ML), and the deeper layers (DL) bounded by hydrological gradients. Diel migrations were significantly different south and north of the PF in terms of total abundance, biomass, diversity and individual taxa density. In both seasons, mesoplankton dielly migrated between the ML/DL and the UL north of the PF and between layers below 300 m and the DL and ML south of the PF. Deeper range of diel migrations south of the PF was coupled with a general mesoplankton descent in summer period compared to spring. Conversely, north of the PF, mesoplankton ascended to upper layers in summer, which was mirrored in lesser depths of diel migrations. The differences in the plankton distribution on both sides of the PF are likely associated with variations of vertical distribution of phytoplankton. Some abundant taxa such as Aetideus sp. and Oithona plumifera showed both common (nighttime ascend) and inverted (nighttime descend) vertical migrations depending on season and position related to the PF.

In this study we focused on the impact of the PF on mesoplankton vertical migrations. The following hypothesis was tested: the PF influences seasonal and diel vertical migrations of the mesoplankton within the upper 300 m. We analyzed and compared diel dynamics of abundance, biomass and diversity of the mesoplankton on both sides of the PF on the basis of samples collected in the Drake Passage during four cruises at various time of day and at different depth ranges (

Results
Variability of community characteristics along the day. Spring samples. North of the PF, abundance and biomass showed similar trends in the UL: values were maximal around the midnight with certain decrease towards the midday. South of the PF, trends of abundance in the UL were opposite (Fig. 2). Trends were similar in the ML on the both sides of the PF and opposite in the DL north and south of the PF. Both abundance and biomass increased around midnight in the DL south of the PF and decreased north of the PF (Fig. 2). Similar trends were recorded in 2008 (Fig. 3).
Diel dynamics of diversity were similar on both sides of the PF. Around midnight, diversity was maximal in the UL (Figs. 2, 3) and minimal or nearly constant in the ML and in the DL (Fig. 2). In 2008 diversity was nearly constant (Fig. 3).
Overall, the extrema of trends were located between the Time of Day values of -1 and + 2 north of the PF and between 0 and + 5 south of the PF (Figs. 2, 3).
Summer samples. Trends in abundances and biomass were similar on both sides of the PF in the UL and ML and opposite in the DL: minimal around the midnight north of the PF and maximal around the midnight south of the PF (Fig. 4).
Diel dynamics of the diversity (species number and ES100) in summer were similar to those in spring with the midnight peak in upper layers and reverse pattern in middle and deeper layers (Fig. 4).
Extrema of polynomial trends were located in the same periods as in the spring time: between − 1 and + 2 north of the PF and between 0 and + 5 south of the PF (Fig. 4).  www.nature.com/scientificreports/ the spring season north of the PF. In summer, negative correlation of species number and daily position was observed ( Table 1). Abundance of many individual taxa was robustly correlated with daily position of samples. These were different larvae, Polychaeta, Pteropoda, Ostracoda etc. Dominant copepod species of the genera Aetideus, Candacia, Calanus, Calanoides, Clausocalanus, Ctenocalanus, Euchaeta, Pareuchaeta, Rhincalanus, Scaphocalanus, Scolecithricella, Pleuromamma, Metridia, Oithona and Oncaea also showed statistically significant diel dynamics (Table 1).
Day-night comparison. Integral community characteristics. Spring. Comparison of averaged night/ day ratios of total abundance, biomass and ES100 by layers is shown in Figs. 5 and 6. North of the PF, abundances in the UL were maximal at night (mean value-2,141 ind. m −3 ) and fell to 1731 ind. m −3 in the daytime; in the DL the dynamics were opposite (e.g. 110 vs. 235 ind. m −3 ). South of the PF, diel dynamics were different from that north of the PF: in the UL, the abundance maximum was observed during daytime (1619 ind. m −3 vs. 1561 ind. m −3 ); in the DL the mean night and day abundances were, respectively, 175 ind. m −3 and 51 ind. m −3 (Fig. 5).
Diel dynamics of biomass were similar to that of abundances on both sides of the PF and in all layers except that the highest biomass was observed in the ML instead of the UL south of the PF (Fig. 5). Diversity (ES100) values were generally higher in the ML and DL than in the UL on both sides of the PF, and the night/day differences were not significant (Fig. 5).
Summer. North of the PF, maximal zooplankton abundances were recorded in the UL and the values were higher than in spring (3,878 ind. m −3 at night and 4,717 ind. m −3 at day). In the DL abundances were lower than in spring (90 ind. m −3 at night and 99 ind. m −3 at day) (Fig. 6). South of the PF, distribution of abundances was different from that in spring: they were highest in the UL at night and in the ML at day (Fig. 6). North of the PF, biomass was maximal in the UL during daytime, while in the ML the night values were higher. South of the PF, biomass maximum was observed at night in the DL (Fig. 6). The diversity distribution south of the PF was also shifted to the DL, similar to the biomass dynamics (Fig. 6).
The overall differences in night and day distribution of integral parameters were lower in summer compared to spring (Table 2).
Species distribution. Distribution of some dominant taxa differed significantly in spring and summer samples (Table 2) including copepods, which showed correlation with diel position (see Section 3.1.3.). During the spring, some shallow-water species were clearly more abundant in the upper layers at night and in the deeper layers in the daytime (e.g. Clausocalanus sp.) (Fig. 7). In summer, the night ascent was less prominent or even  www.nature.com/scientificreports/ absent among these species (Fig. 7). Mesopelagic species were more abundant at night in the ML and DL and nearly absent in the UL (e.g. Aetideus sp., Euchaeta marina, Pleuromamma robusta) (Fig. 7). Most taxa occurred deeper south of the PF than north of the PF, and deeper in summer than in spring ( Table 2). Copepods Aetideus spp., Rhincalanus gigas, Oithona plumifera and Oncaea sp. in summer migrated inversely (midnight decent- Table 3, Fig. 7). Correlation of abundances of Aetideus and Oithona species with abundances of possible predators revealed negative correlation values with hydromedusae, siphonophores and predatory copepods (Euchaeta, Pareuchaeta and Heterorhabdus genera) (Table 4).
Overall, the UL was more enriched in taxa at night than in the daytime (46-75% vs. 25-54% of recorded taxa). The ML was still more enriched at night, but the night/day differences were less significant (52-64% at night vs 36-48% in the daytime). In the DL the proportions varied depending on the zone and/or the season (Table 3).

Discussion
Vertical migrations of Subantactric and Antarctic zooplankton were previously studied in the Drake Passage and adjacent Southern Ocean both in diel and seasonal aspects 16,[19][20][21]28,29 . One of the common ways to describe the magnitude of diel migrations is a night/day ratio of mesozooplankton characteristics 20,29,30 . As an example, the mean night/day ratios of the mesoplankton abundance and biomass were reported to be 1.06 and 1.17, respectively, in the upper 100 m of the Amundsen Sea 29 . In waters around South Georgia the night/day rations varied from 0.61 to 1.69 and from 0.73 to 1.45 for abundance and biomass, respectively 20 . The ratios were obtained from the station pools not divided by any of the hydrological fronts 20,29 . However, the fronts may drastically influence mesoplankton distribution [19][20][21][22] , so we hypothesized that the observed variations are linked to the position of main hydrological fronts. Indeed, in our dataset, the ratio depended on the position respective to   Table 2). Additionally to the mesoplankton abundance and biomass, the similar night/day differences were also demonstrated for the diversity values, which was never reported before. The ES-100 changes indicate that the vertical migration involves many mesoplankton species in a similar way (including the rare ones), as this index includes both number of species and the evenness of their abundances 31,32 .
The diel vertical migrations are thought to be triggered by the light intensity changes during the day 9,16,20,29,30 . The upward migrations were previously recorded around 7:00-9:00 p.m., with the following morning descent at 5:00-6:00 a.m. 16 . The difference in time is explained by changes in dusk and dawn time linked to different latitude or date 15,16,33 . In our samples, according to the distribution of integral parameters along the day, the nighttime extremum (maximum or minimum, depending on the PF position and the depth range) fell within the 11:00 p.m. and 3:00 a.m. north from the PF and within 1:00 a.m. and 4:00 a.m. south from the PF. The exact time of mesoplankton ascent and descent is hard to assess in our dataset.
The range and direction of diel vertical migrations in our samples depended not only on the position respective the PF (north or south of that) but also on the season (spring or summer). Seasonal changes in vertical migrations of zooplankton were regularly observed in temperate and polar seas 16,21,28,30,[33][34][35] . The seasonal variation is often attributed to the overall decrease of diel light/dark cycles in summer 30,34,35 . In particular, the midnight sun period can stop any vertical movement of certain copepod species, as it was shown for Calanus species 35,36 . In winter, during the polar night, the mesoplankton migrates dielly despite the lack of obvious light/dark cycles 37 . In the Southern Ocean, an overall downward migration of zooplankton was observed in winter, as it was shown by summer/winter comparison 16,21,28 . Cisewski et al. 16 distinguished two periods of migration patterns: from February to October (late summer to early spring), when the migrations are driven by the day/night rhythm; and from October to January, when the most of the zooplankton rises to the uppermost waters (< 50 m) for feeding and reproduction. Our samplings, taken in October-November (spring) and January (summer), therefore, fall into the second migration period described by 16 .
Differences between the spring and summer samples were significant and concerned integral community parameters and distribution of certain species. The seasonal shifts in mesoplankton vertical migrations depended on the position respective to the PF. Spring/summer vertical distribution of the abundance and biomass demonstrated the overall summer ascent north of the PF and summer descent south of the PF (Figs. 5, 6). In particular, north of the PF the mesoplankton was mainly concentrated in the upper layers by the summertime. South of the PF, the nighttime maxima of the abundance and biomass descended to the depth of ~ 150-300 m by summer (possibly even deeper, to the depths unsampled in this study) (Figs. 5, 6). Seasonal zooplankton migrations are linked to the food source distribution (i.e. primary production rate, often expressed in Chl a values) which concentrates in the upper mixed layer during spring and summer 16 . In the Drake Passage, the maximum surface chlorophyll values were observed around November-December north of the PF and in December-January south of the PF 38,39 . Concentration of the mesoplankton in the UL north of the PF in spring and summer is thus explainable (Figs. 5, 6). However, the summer mesoplankton descent from the most productive UL to the deeper layers south of the PF is unexpected. A possible explanation may be in vertical distribution of the phytoplankton: e.g., in the Eastern Atlantic Sector of the Southern Ocean Chl a was concentrated in the UL north of the PF, but evenly distributed within the upper 100 m south of the PF 40 . In the Drake Passage south from the PF, the Chl a vertical distribution demonstrated deep maxima below the UL (pers. comm. by Dr. A. Demidov, unpublished data). The dynamics of individual species distribution expectably demonstrated similar migration patterns as the integral community parameters. Most recognized taxa migrated normally with the nighttime surface ascent, confirming previous studies 2, 5,16,21,22,28 . However, several taxa demonstrated the reverse migration pattern with the night descent to the deeper waters. Previously the negative pattern was shown for species of Oithona similis and Oncaea curvata copepods near the Antarctic coast (south off ACC) 13 .
Seasonal shifts (reflected mainly in summer ascent) were previously reported for several species of copepods in the Southern Ocean, including Eucalanus longiceps, R. gigas, Neocalanus tonsus, Calanoides acutus, Calanus simillimus and Calanus propinquus 21 . Our data confirmed the spring/summer shifts for most of the species with possible exception of C. acutus. E. longiceps and N. tonsus were not statistically represented in our samples. According to our data, R. gigas, O. plumifera, Oncaea sp. and Aetideus spp. copepods changed direction of their www.nature.com/scientificreports/ www.nature.com/scientificreports/ diel migrations from normal to inversed depending on season and position respective to the PF. Inversed migrations are generally explained by a possible avoidance of nocturnal predators with normal migration patterns 5,15,30 , which is in accordance with our dataset: Aetideus sp. and O. plumifera show certain negative correlations with predators, including cnidarians and Euchaetidae copepods. These predatory mesoplankton organisms are listed as consumers of copepods, possibly including Oithona species 41 . We suggest that the predators influence migratory behavior of the taxa and may switch direction of migrations on both sides of the Polar Front. The observed details of the vertical migrations show fine adaptive adjustments of taxa to local factors depending on season (spring or summer), hydrological setting (north or south of the PF), and predators. The PF has a great impact on seasonal and vertical migrations of mesoplankton. Vertical dynamics differ on both sides the PF in seasonal and diurnal aspects. North of the PF the mesoplankton concentrates in the UL both in spring and in summer, while south of the PF the mesoplankton concentrates in the UL in spring and descends in deeper layer in summer. In spring north of the PF, most of the taxa dielly migrate within the upper 300 m, ascending from the DL to the UL at night, while south of the PF vertical migrations encompass deeper layers from below 300 m (unsampled in this study) to the DL. In summer and north of the PF, migrations of mesoplankton are concentrated in the UL and large-scale diel migrations are insignificant, which mirrors feeding and reproduction in phytoplankton rich strata. Conversely, south of the PF the mesoplankton is concentrated below the UL: a possible result of an even vertical distribution of phytoplankton with no prominent surface maximum. The described trends of the seasonal and diel migrations are shown in a simplified scheme representing a balance-like swing of mesoplankton maxima on northern and southern sides of the PF (Fig. 8). Individual taxa such as Aetideus sp. and O. plumifera showed both common (nighttime ascend) and inverted (nighttime descend) vertical migrations depending on season and position related to the PF. According to 42 , the SF and the SAF act as more important boundaries for zooplankton communities than the PF. Although, we do not have a representative data set to compare the areas south from the SF and north from the SAF, we suggest that the differences in migration patterns in those areas may differ significantly from the described in this study.

Material and methods
The material was based on four expeditions to the Drake Passage during early spring and summer. A total of 41 stations were sampled in October-November 2008 (RV "Akademik Sergey Vavilov", 25-th expedition), 15 stations in January 2010 (RV "Akademik Ioffe", 20-th expedition), 12 stations in November 2010 (RV "Akademik Sergey www.nature.com/scientificreports/ The three sampled strata were separated by vertical gradients of temperature and salinity. The uppermost mixed layer was well-defined and bounded from below by seasonal halo-and thermoclines; two deeper layers were separated from each other by the extrema of temperature and salinity profiles indicated by CTD-sensor at the same stations prior to biological sampling. Actual sampling depths ranged along the transect, the upper (UL) and middle layers (ML) typically represented the epipelagic, while the deep layer (DL) occurred mainly in the upper mesopelagic ( Fig. 1, Supplementary 1). The net was equipped with the closing device; vertical towing speed was 1 m/sec. The hydrological setting was previously published by [43][44][45] .
Daily position of each sample was calculated individually. We used the local astronomical midnight as a zero-point, which was assessed on the basis of local time and coordinates (https ://www.esrl.noaa.gov/gmd/ grad/solca lc/sunri se.html). Daily position of each sample was further calculated as a difference between the astronomical midnight and sampling time and expressed in decimals (positive for a.m. time; negative for p.m. time-Supplementary 1).
Zooplankton samples were fixed with 4% formalin and later sorted in laboratory. All organisms were identified to the lowest possible taxonomical level. For each taxon, the numbers of specimens in the sample and individual sizes (length) were recorded with a precision of 0.1 mm. On the basis of this primary dataset, the individual weights, and biomass were calculated. Wet weight w of taxa represented by i specimens was estimated as w = Σ (k * l i 3 ), where l i is length of an individual specimen, k is a species-dependent coefficient; tables of these coefficients have been published by 46,47 . Abundance and biomass were normalized to ind. m −3 and g ww m −3 , respectively, and presented in Supplementary 2.
Total abundance, biomass, species number and Hurlbert rarefaction index for 100 individuals (ES100) were used as integral community parameters. In order to avoid bias linked to unrepresentative sampling of larger organisms (jellyfishes, euphausiids) and Protozoa, we excluded these groups from abundance and biomass matrices. Correlation between Time of Day (daily position) and taxa abundances and integral community parameters was calculated using Spearman ranked correlation (modulus values of daily positions were used) 31 . To assess diel trends of different parameters, we plotted 2-nd order polynomial trendlines. Difference between day-time and night-time distribution was estimated using the night/day values ratios and the Student t-test. Night-time and day-time stations were identified as stations taken during ± 4 h from astronomical midnight and ± 4 h from astronomical midday, respectively. Horizontal box charts with standard errors were plotted for integral community parameters and for certain taxa to visualize differences between the depth layers and day/night distribution.