Three investigations were conducted around the Pacific Ocean in the CRD, CS, and TS, as shown in Figure 1 with current data from Hu et al. (2015). Two of the investigated regions (CRD and CS) were classified as the North Pacific Tropical Water and the TS were classified as Subtropical area (Suga et al., 2000). Basic hydrographic data (i.e., salinity, temperature, depth, and chlorophyll a in vivo fluorescence) and water samples were collected using a conductivity-temperature-depth (CTD) rosette system, with Sea-Bird oxygen sensors (Sea Bird Electronics Incorporation, Bellevue, WA, United States).

Sampling plans included: (1) Tintinnid samples in the CRD were conducted at 5 sites from June 25 to July 22 in 2010 (8°30′–10°8′ N, 86°52′–92°59′ W, Figure 1). At each site, 7 layers were sampled from 0 to 200 m: surface layer, surface mixed layer, the layer above Chl a max, Chl a max layer, layer below Chl a max, interface layer, and deep control layer. (2) Tintinnid samples in the CS were collected at 5 sites from November 30 to December 5 in 2012 (2°–6°12′ N, 124°50′–128° E, Figure 1). The sampling depths ranged from 0 to 200 m or 2 m up to the bottom: 5, 30, 50, 75, 100, 150, and 200 m. (3) Tintinnid samples in the TS were collected at 5 sites in December 2012 (29°0.1′–31°0.2′ N, 129°0.6′–130°0.1′ E, Figure 1). The sampling depths ranged from 0 to 200 m or 2 m above the bottom: 5, 30, 50, 100, and 200 m.

Sampling methods were similar in the three investigated regions. At each site, a 10, 20, or 30 L water sample was collected using Niskin bottles at each layer, and then filtered slowly and gently through a net (mesh size 20 μm). The concentrated samples (∼150 ml) were fixed with formalin solution to 5% final concentration and stored in adiaphanous plastic bottles in the lab. Tintinnid enumeration was conducted according to the method of Utermöhl (1958). Subsamples of 20 ml from well-mixed concentrated samples were pipetted into a sedimentation chamber and settled for 24 h, and subsequently counted under an Olympus IX 71 inverted microscope (200× or 400×) with a photographic measurement system.

Tintinnid species identifications were made on the basis of lorica morphology and dimensions according to literature (Kofoid and Campbell, 1929, 1939; Zhang et al., 2012; Santoferrara et al., 2017). Here we employed an approach (Dolan et al., 2006) to identification: any intermediate forms or slight variants were pooled with the morphologically closest and most numerous species. To measure the lorica oral diameter (LOD) and lorica length (LL), 10 to 20 individuals of each species were randomly picked and measured, and lorica volume (LV) was calculated. Species with rare occurrences were assigned the average dimensions reported in literatures as mentioned above.

Environmental factors were different in the three investigated regions (Figure 2). The temperature ranged from 11.97 to 28.51 (18.59 ± 5.96)°C, 13.55 to 29.18 (24.42 ± 4.94)°C, and 12.75 to 24.40 (21.91 ± 3.27)°C in the CRD, CS, and TS, respectively. Salinity fell in the ranges of 33.45–34.93 (34.51 ± 0.54), 31.55–35.13 (34.14 ± 1.48), and 34.14–34.86 (34.44 ± 0.17) in the CRD, CS, and TS, respectively. Chl a in vivo fluorescence ranged from 0.10 to 1.22 (0.53 ± 0.28), 0.04–1.01 (0.22 ± 0.22), and 0.05–0.84 (0.37 ± 0.24) in the CRD, CS, and TS, respectively. Chl a values were highest in the CRD. With consideration of water depths and cluster analysis, four water masses were distinguished by the temperature-salinity diagram in each of the three sampling regions (Figure 2A).

In total, 46 species of tintinnids were found across the three investigated regions (Table 1), with a total of 37, 38, and 34 species in the CRD, CS, and TS, respectively. Total tintinnid abundances and biomass ranged 533–93,624 ind m^–3 (2.2–372.9 μg C m^–3 for biomass) in the CRD, 100–35,600 ind m^–3 (0.2–238.8 μg C m^–3) in the CS, and 2,200–69,400 ind m^–3 (13.1–263.7 μg C m^–3) in TS, respectively. Tintinnid assemblages in the CS had the largest species number, but were lowest in both abundances and biomass (Figure 2B). Log-transformed abundances followed a normal distribution. ANOVA with main effects (regions and seasons) without interaction showed clear effects on tintinnid assemblages for regions (p < 0.05) but not for seasons. Margalef’s species richness index (D) ranged 0.4–1.7 in the CRD, 0–2 in the CS, and 1.3–2.9 in the TS, respectively. Shannon diversity index (H’) ranged 1.2–2.6 in the CRD, 0–2.6 in the CS, and 2–2.8 in the TS, respectively. Pielou’s evenness index (J) ranged 0.6–1 in the CRD, 0.8–1 in the CS, and 0.7–0.9 in the TS, respectively.

For all three investigated regions, maxima of tintinnid species richness and abundances always occurred in the upper two water masses (Figure 2B), although richness and abundance maxima were not always co-located, and high values were more likely to occur at the interface of different water masses. In the CRD, tintinnid abundances peaked in water mass CRDII at 20 m in St. CRD2 in relatively strong temperature gradients at interfaces between water masses, while biomass peaked in St. CRD5 due to the dominance shift from small individuals Ascampbelliella armilla to a larger species Rhabdonella elegans. In the CS, tintinnid species richness peaked in water mass CSII at St. CS4 and changed largely with low total abundances in the thermohaline intrusion in the upper 100 m in St. CS2 at the entrance of the CS. In the TS, tintinnid species richness and abundances peaked in water mass TSII especially in the intersection of different water masses where isotherms sloped sharply, whereas the least individuals were found in water mass IV. Tintinnid abundances in water mass TSI were consistently high while those in water mass TSIII were similarly low. When water mass intrusions were dramatic, tintinnid abundance decreased but species richness remained high.

For tintinnid assemblages in the three investigated regions, similar taxonomic compositions were revealed in terms of species, abundant species, genera, LOD size classes, and biogeographic patterns (Figure 3 and Tables 2, 3). A total of 29 species (63.0% of total species richness, Figure 3A) occurred in all three regions, i.e., core species, which significantly comprised 97.1% of the total abundances. These core species also occupied 77.8% of the total genera (21 genera out of 27, occupying 99.3% of total abundances, Figure 3B), and 81.3% of size-classes (13 out of 16, occupying 99.9% of total abundances, Figure 3C). In total, 5 common species and 12 endemic species were found in this study, although all their abundances were very low (<2,000 ind m^–3, except Steenstrupiella gracilis). The overwhelming majority of core species indicated the numerically abundant forms remained more or less the same set of species in the three regions.

Although abundant species were different (Table 2 and Figure 3E), all of them occurred as core species in the three regions. There were 7, 5, and 6 abundant species observed in samples of the CRD, CS, and TS, respectively. In all three regions, Proplectella perpusilla (P. perpusilla) was the numerically dominant species. There were 3 dominant species (P. perpusilla, Eutintinnus lususundae, and Parundella aculeate) in common between the TS and CS, and 2 dominant species (Dadayiella ganymedes and P. perpusilla) in common between the TS and CRD. However, tintinnids in the CRD had no dominant species in common with the CS, which was consistent with the largest distance between the two investigated regions. Abundances of all the dominant species were found significantly positively correlated to total abundances (p < 0.05).

Similar compositions of tintinnid assemblages were revealed among the three regions in terms of LOD size classes (Table 2 and Figure 3C). The LOD ranges and number of size-classes of all and co-habitated species were similar in the three regions (Table 3), and LOD size-class 26–30 μm contained the most redundant species and the largest proportion of total abundances (Figure 3C). Notably, tintinnid populations in the CRD were clearly distinguished from those in the CS and TS in LOD size classes for abundant species. Three out of 4 LOD size classes were the same in the CS and TS, while only one was the same as in the CRD. Redundants were most numerous in the size classes of the first dominant species in the CS and TS, but not in the CRD.

Three biogeographic distribution patterns were identified: cosmopolitan, warm-water, and neritic which occurred only in one site in the CRD nearest to the coast. All the core species were found as two biogeographic patterns: cosmopolitan (16 species) and warm-water (12 species). The two types occupied most of the species number with similar proportions, while abundances of the cosmopolitan type were higher than those of warm water in the TS and CS (Figure 3D). Endemic species of the three regions were revealed related to biogeographic types, i.e., endemic species were neritic, cosmopolitan, and warm-water species in the CRD, CS, and TS, respectively, with the addition of warm-water species in each area. The endemic species of the CRD were restricted to water masses CRDIII and CRDIV in the intermediate layer, while endemic species of the TS were restricted to water mass TSI and TSIII in the upper layer.

To explore the overall variability in community composition, we performed PCoA which revealed a clear separation between samples in the 0–50 m depth range and deeper depths (Figures 4A–C), and depth explained even more of the variance in terms of LOD size class. Furthermore, to explore the relationships between abiotic factors and tintinnids, we conducted RDA analysis and found clear distinctions among samples from different water depths, especially in terms of LOD size-class (Figures 4E,E1, r² = 0.59, p < 0.01). Samples above 100 m were clearly distinguished from those below 100 m (Figures 4E1,F1). Water depth was negatively correlated tintinnid species richness and abundance (p < 0.05). Clear distinctions were also revealed by RDA analysis among tintinnid assemblages in the three regions in terms of species and genera (Figures 4D,F,D1,F1), while tintinnid assemblages in the TS were aggregated and close to those in the CS in terms of LOD size-class (Figure 4E). The latitudinal decline of tintinnid species richness (Dolan et al., 2016) was evident in the case of the CRD (r = −0.359, p < 0.05) and the CS (r = −0.302, p < 0.05), but not in the TS. Temperature and Chl a were positively correlated with tintinnid species richness and abundances (p < 0.01).

Species abundance distribution was fitted to the geometric, log-normal, log-series, and mZSM distributions by the AIC statistic (Table 4 and Figure 5). The log-normal distribution provided the best match to the observed pattern in the CRD and TS while the geometric distribution gave the best fit in the CS both in terms of species and LOD size-class. The AIC values of the geometric distribution in the CS were very close to those of log-normal distributions. The geometric series, describing a sequential monopolization of resources, described well the tintinnid assemblages in the CS, which were warmest, with largest species richness but least abundance, and highly dominated by Proplectella perpusilla and Eutintinnus lususundae. The log-normal distribution, thought to result from complex species interactions, provided the best fit for the CRD and TS. For most sites, the observed distribution most closely matched log-series, coherent with the neutral theory of random colonization from a large species pool.

We hypothesized that the greater similarity between tintinnid populations in the CS and TS (relative to CRD) was the result of geographical proximity and circulation patterns that link these regions. Lagrangian simulations (Supplementary Figure 1) revealed that it took the particles ≥710 days to migrate from the CS to CRD and ≥450 days to travel from the CRD to the CS, which approaches the cycle time of Pacific Equatorial Countercurrent. Thus, the result supported our hypothesis that tintinnid population could migrate across the Pacific Ocean near the equator, and they are more likely moving from the CS to CRD. Transit time from the CS to TS was ≥160 days (≥420 days from TS to CS), while it required ≥1,670 days for ciliates to be advected from the TS to CRD and 1,870 days from CRD to TS. With consideration of the Kuroshio Current and Mindanao Current, we deduced that tintinnids should migrate from someplace between the two currents, and migrate to the TS along with Kuroshio Current and the CS along with the Mindanao Current and then reach the CRD along with the Pacific Equatorial Countercurrent.

What are the basin-scale relationships among the tintinnid assemblages as evidenced by the data from three different regions in this study? Based on the distance and ocean currents, we may have three possible scenarios. The first one is that tintinnid populations might migrate from the CRD to CS along with the North Equatorial Current and then to the TS along with the Kuroshio Current. Alternatively, tintinnid populations in the CS and TS might derive from the same origin or share a similar community structure, whereas tintinnids migration from the CRD to the other two regions is questionable. The third scenario is that there is no similarity among those tintinnid populations in the three regions. Comparisons of taxonomic compositions in terms of species, abundant species, LOD size classes, and genera supported the second scenario. Kim et al. (2012b) revealed tintinnids as bioindicators that more sensitively detected water mass extension and intrusion than physical properties. El-Serehy et al. (2014) reported the Suez Canal as a selective barrier and/or as a link in the migration of tintinnid protozoa. Li et al. (2016) excavated the interactions between different tintinnids in different water masses.

While our simple Lagrangian model allowed us to quantify advection patterns that potentially link these study regions, several points could be included in future-focused models: (1) the growth, death, and predation and life cycle of tintinnids could be explicitly models to assay community turnover and thus the succession of tintinnid communities during transit; (2) species growth responses to environmental variables could be included to differentiate between the importance local growth and advective processes in maintaining species patterns and determine the differences between “invasion” verse “migration” for the advected species competing with local species; and (3) the effects of chaotic advection by oceanic currents on biodiversity patterns of rare protist species could be investigated (mentioned by Martin et al., 2020).

Plankton carried by the current often have distinct environmental niches and remain recognizable until they die and disintegrate, which make them reliable indicators of water mass movements (Kim et al., 2012a,b). The reliability of an indicator species is determined by the higher frequency of occurrence in a particular water mass and by identifiable morphology. The indicator species must be resistant to gradual changes in water properties (Schwenke, 1971), but should not be sufficiently capable of survival over a wide range of the change (Raymont, 1980; Kato and Taniguchi, 1993). Thus, with hard lorica, tintinnids are a reliable indicator of water mass movements (e.g., Kato and Taniguchi, 1993; Kim et al., 2012a,b; El-Serehy et al., 2014; Li et al., 2016), and information from tintinnid biological indicators was suggested to support physical oceanographic data to confirm ambiguous water mass properties (Kim et al., 2012a).

The hard loricae outside the tintinnids can remain even after their death making tintinnids even more effective as bioindicator of large-scale currents. Empty loricae of tintinnids have been used as indicators for tracing the movement of water masses and currents because their sinking rate, as well as decomposition rate, can be assumed to be very slow (Taniguchi, 1983; Kato and Taniguchi, 1993; Pierce and Turner, 1993; Suzuki and Taniguchi, 1995). Empty loricae were transported long distances by currents before settling to the sediments (Echols and Fowler, 1973). On the other hand, when loricae were found empty or damaged, this might indicate long-distance transport. Conversely, when loricae occupied by a living ciliate were observed, it was likely that the species lived either at or near the sampling site (Kim et al., 2012b). In this regard, the proportion of empty or damaged loricae was suggested as a useful property in tintinnid investigation in the open sea. However, there are limitations for tintinnids usage as bioindicators: the absence of standardization of the sampling methods, fixation method and analytical protocols, discrepancy of different identification schemes in the literature, as well as several highly polymorphic genera and the fact that criteria for the delimitation of the species may vary among the observers (Gómez, 2007).

Gómez (2007) recorded 42 tintinnid species with low density (> 10 ind l^–1 from Figures) in the CS and vicinity, while abundance was about 20–30 cells l^–1 in the south of the Kuroshio Current. This study revealed similar results in the CS (Taniguchi, 1977) and in the CRD (Freibott et al., 2016), and higher abundance in the TS than those in the south of Kuroshio Current (Gómez, 2007). The assemblage of abundant species in the CRD appeared distinct from those abundant species of the CS and TS in terms of LOD size classes, presumably reflecting exploitation of different sizes of prey items. Many factors influencing tintinnid abundance have been discussed, including temperature, salinity, Chl a (e.g., Wang et al., 2018), water masses, latitude (Dolan et al., 2016), copepod nauplii and bottom depth (Santoferrara et al., 2011), mutualistic diatoms (Vincent et al., 2018), food conditions (Kazama and Urabe, 2016), etc. Most of these factors were assessed at micro-scales, while this study provided the possibility and practicability of quantifying tintinnid variation at the meso-scale. As Woods (1999) mentioned, we still cannot explain quantitatively how the relative abundance of plankton species varies geographically.

Redundant species were more common in the CS in low latitude than other regions, while warm-water species decreased with increasing latitude, providing a possible explanation of the latitudinal decline of tintinnid species richness. Redundants were most numerous in the size classes of the dominant species and can likely replace a dominant species, a result that is consistent with Dolan et al. (2016).

Species abundance distributions describe community structure and are a key component of biodiversity theory and research (Antão et al., 2021) and have been conducted on tintinnids in several previous studies (Dolan et al., 2016; Zhang et al., 2017). It was generally accepted that most distributions of species abundance in large assemblages are log-normally distributed (Gaston and Blackburn, 2000; Magurran and Henderson, 2003). This study showed that a log-normal distribution provided the best fit for tintinnid abundance distribution in the CRD and TS although a geometric distribution was the best fit in the CS both in terms of species and LOD size-class. Geometric distributions of LOD size-class abundance are most simply attributed to the availability of prey concentration and size given the close relationship between LOD size and prey exploited by tintinnids (Dolan, 2010). Geometric distribution mainly occurs in species-poor and often harsh environments (Magurran, 1988; Fattorini, 2005), or in the very early stages of succession (Whittaker, 1972; Matthews and Whittaker, 2015). The potential of the elucidation of macroscale species abundance has far been an inaccessible but critical property of biodiversity (Fukaya et al., 2020). Several theories have been involved in the SAD, including at different scales—the Neutral Theory of Biodiversity (NTB; Hubbell, 2001) and the Maximum Entropy Theory of Ecology (METE; Harte et al., 2008), and supporting or disproving these and other theories will require more studies that quantify the species distributions of plankton in the open seas.