Highlights

  • Integrated trophic positions (iTPs) of mesozooplankton communities were studied.

  • Diversity, body weight, and turnover of species control the iTP values.

  • The iTP index will bridge stable isotope ecology and food web science.

Introduction

The trophic structure of ecosystems has often been visualized as biomass pyramids since early ecological studies defined the concept of trophic level and food chain (Elton 1927; Lindeman 1942), which are now termed trophic position (TP) and food web, respectively. In ocean ecosystems, zooplankton plays a central role in transferring organic matter from phytoplankton to higher organisms such as fish and sea mammals. In the subarctic North Pacific, mesozooplankton (> 100 μm) is the dominant component of the ecosystem, accounting for 34% of total biomass (Ikeda and others 2008). Multiple feeding guilds of mesozooplankton cover a wide spectrum of TPs ranging from herbivores (TP = 2) to carnivores (TP ≥ 3) (Kiørboe 2011). Therefore, it is anticipated that functional diversity controls the trophic structure (for example, biomass pyramid) of mesozooplankton, which is crucial for understanding the entire ecosystem, including fish and sea mammals with higher TPs.

Compound-specific nitrogen isotope (δ15N) analysis of amino acids (CSIA-AA) has drawn attention as a precise tool to estimate the TP of organisms in a variety of ecosystems (after Chikaraishi and others 2007; McCarthy and others 2007; Popp and others 2007). One of the most advantageous features of CSIA-AA over conventional isotope analysis for bulk tissue of organisms is that the TPs of organisms can be estimated without baseline δ15N data, which fluctuate in space and time (Chikaraishi and others 2011). In the subtropical North Pacific, Hannides and others (2009) found that zooplankton TPs estimated from CSIA-AA varied from approximately 2.0–3.0, which supports the aforementioned hypothesis that the zooplankton community includes multiple feeding guilds, including herbivores, omnivores, and carnivores.

Using CSIA-AA, Ishikawa and others (2017) proposed a new index, the integrated TP (iTP), which can be easily determined from the TP value of the whole community in a given area or space, as follows:

$$ iTP = \mathop \sum \limits_{i = 1}^{n} \left( {TP_{i} \times \frac{{B_{i} }}{{B_{T} }}} \right), $$
(1)

where TPi is the TP of species i, and Bi and BT are the biomass of species i and the total biomass of the focal community per unit area or space, respectively. The iTP index can be used as a proxy for the energy requirement of the focal community because more primary production is needed to sustain a community biomass with a higher iTP value. Ishikawa and others (2017) found a negative correlation between iTP and biodiversity in stream macroinvertebrate communities. However, it is unclear whether this relationship is universal.

The aim of the present study was to provide the first results of iTP data for mesozooplankton communities in the ocean using CSIA-AA. The western North Pacific was investigated because it offers several oceanographic settings that affect local biogeochemistry. Among the investigated offshore sites, the northwestern subarctic gyre (Oyashio) supplies nutrient-rich seawater to local phytoplankton (Harrison and others 2004). In contrast, the investigated nearshore sites (off the Pacific coast of Japan) exhibit complex hydrography, which is affected by the western subtropical gyre (Kuroshio) and coastal waters (off Sagami Bay, Miyaguchi and others 2006), as well as Oyashio (off Joban/Kashima, Yokouchi and others 2000). Therefore, it is predicted that mesozooplankton communities differ among sites, which allow us to compare less diverse compositions dominated by large species at the offshore sites and more diverse compositions at the nearshore sites. Using this dataset, species composition, functional diversity, and mean body weight in the mesozooplankton communities were examined to elucidate factors controlling iTP variation. Moreover, the biomass pyramid was estimated using variations in observed iTP data.

Materials and Methods

Study Sites

Mesozooplankton samples were obtained from the Zooplankton Sample Collection of the Japan Fisheries Research and Education Agency stored in the Fisheries Resources Institute of Shiogama Field Station, which was formerly known as the Odate Collection (Odate 1994). Samples were collected from 14 sites in the western North Pacific using remodeled NORPAC and Marutoku plankton nets with a mesh size of 335 μm and 45 cm opening diameter on multiple research cruises at water depths of 0–150 m below sea level from April to August of 2000 to 2015 (Table S1). The study sites included Pacific sites (P sites, P1–P6), off Joban/Kashima sites (J sites, J1–J4), and off Sagami Bay sites (S sites, S1–S4) (Figure 1). The collected samples were fixed with 5% buffered formalin and kept at room temperature until further processing. Formalin fixation of animal tissue does not affect δ15N measurements of amino acids (Ogawa and others 2013).

Figure 1
figure 1

Study sites in the western North Pacific. P1–P6 (red), J1–J4 (green), and S1–S4 (blue) are the Pacific sites, Joban/Kashima sites, and Sagami sites, respectively.

Full size image

The mean temperature (Locarnini and others 2018) and the nitrate concentration (Garcia and others 2018) data were obtained from the World Ocean Atlas (grid resolution 1°, averaged values of 0–150 m below sea level during the sampled years of 2005–2017). The mean bottom depth data were obtained from the global digital elevation model (ETOPO2) (NOAA National Geophysical Data Center 2006) for the P sites and from the Japan Oceanographic Data Center (JODC) (500-m Gridded Bathymetry Data, https://www.jodc.go.jp/jodcweb/JDOSS/infoJEGG.html) for the J and S sites. The chlorophyll a concentration data were obtained from the Ocean Colour Climate Change Initiative (CCI) project (Sathyendranath and others 2020). The dominant phytoplankton types were obtained from Hirata and others (2011) (Table 1).

Table 1 Oceanographic Settings of the Study Sites
Full size table

Collection and Processing of Samples

In the laboratory, each formalin sample was gently stirred, homogenized, and then aliquoted using the Motoda splitting box (Motoda 1959), with the exception of large species, such as Chaetognatha, Hydrozoa, and Neocalanus, which were manually picked up, counted, and weighed. The number of individuals (N) in the aliquot was counted to ensure that there were > 200 individuals. In addition, all individuals were examined under a binocular microscope, which identified 155 taxa from 14 communities. Taxonomy followed the World Register of Marine Species (http://www.marinespecies.org). If applicable, different growth stages were distinguished (that is, calyptopis, copepodite, egg, furcilia, juvenile, metanauplius, and nauplius; Table S1) because different growth stages of the same species may belong to different feeding guilds. We counted N and measured the wet weight (WW) in grams of each species (and each growth stage if applicable) per aliquot, and then multiplied the numbers by the split ratio specific to each site (from 1/30 to 1/4, Table S1). At each site, the body weight (M) of each species or growth stage was calculated as the WW (as biomass B) divided by N (that is, M = B/N), although the estimated M values might be too rough for species that potentially included multiple growth stages.

Estimation of Species Diversity and Functional Feeding Guild

The mean body weight of the community (MT) was calculated as the total biomass of the community (BT) (g WW haul−1) divided by the total number of individuals (NT) (that is, MT = BT /NT). There was a significant correlation between the number of species (S) and NT (R2 = 0.41, p = 0.008), and between S and BT (R2 = 0.28, p = 0.03) in the dataset. To separate the net effects of diversity on iTP from those of NT and BT, the Shannon diversity index (H') and the biomass-based evenness index (Pielou’s J') (0 ≤ J' ≤ 1) were calculated using NT, BT, and S. The H' index was calculated for each site as

$$ H^{\prime} = - \mathop \sum \limits_{i = 1}^{S} \left( {p_{i} \times \ln p_{i} } \right), $$
(2)

where pi is the number of individuals of species i divided by that of the local community (0 ≤ pi ≤ 1) (Shannon and Weaver 1963).

To assess the skewness of biomass distribution in the focal community, biomass-based H' (Hb') was calculated as follows (Wilhm 1968):

$$ H^{\prime}_{b} = - \mathop \sum \limits_{i = 1}^{S} \left( {b_{i} \times \ln b_{i} } \right), $$
(3)

where bi is the proportional biomass of species i relative to the local community (0 ≤ bi ≤ 1). When bi is equal among all species, Hb' takes the maximum value as follows:

$$ H^{\prime}_{{{\text{b}}\,\,{\text{max}}}} = - \mathop \sum \limits_{i = 1}^{S} \left( {\frac{1}{S} \times \ln \frac{1}{S}} \right) = - S \times \left( {\frac{1}{S} \times \ln \frac{1}{S}} \right) = \ln S. $$
(4)

Therefore, the biomass-based J' index (0 ≤ J' ≤ 1) was derived as follows (Pielou 1966):

$$ J^{\prime} = \frac{{H^{\prime}_{{\text{b}}} }}{{H^{\prime}_{{{\text{b}}\,\,\max }} }} = \frac{{H^{\prime}_{{\text{b}}} }}{\ln S}. $$
(5)

Values of BT, MT, NT, H', and J' for each site were used as ecological variables in later analyses. To test the hypothesis that the functional feeding guild of mesozooplankton species in a community controls iTP, each species was categorized as herbivores, carnivores, or others (including omnivores, detritivores, or parasites) based on information available from the literature (Table S2). Then, the proportion of biomass of each feeding guild in a community was calculated to determine whether the iTP variation can be explained by the dominance of specific feeding guild(s).

Calculation of iTP and Biomass Pyramids

The difference in δ15N between ‘trophic’ amino acids, such as glutamic acid (Glu), which greatly fractionates 15N against 14N at each trophic step, and ‘source’ amino acids, such as phenylalanine (Phe), which hardly fractionates 15N at each trophic step, is proportional to the TP (Chikaraishi and others 2009):

$$ {\text{TP = }}\frac{{\delta^{15} N_{Glu} {-} \delta^{15} N_{Phe} { } - { 3}{\text{.4}}}}{{TDF_{Glu} - TDF_{Phe} }}{ + 1}, $$
(6)

where δ15NGlu and δ15NPhe are δ15N of Glu and Phe in the focal mesozooplankton community, respectively, 3.4 is the initial difference (‰) between δ15NGlu and δ15NPhe found in phytoplankton, and TDFGlu and TDFPhe are the trophic discrimination factors (‰) of δ15NGlu (8.0‰ ± 1.2‰) and δ15NPhe (0.4‰ ± 0.5‰), respectively (Chikaraishi and others 2009). In this study, the TP of the composite mesozooplankton biomass is defined as iTP.

To illustrate the biomass pyramid for each mesozooplankton community using its iTP value, we assumed that (1) complex food webs can be disassembled and reassembled to a simple food chain (food network unfolding; Higashi and others 1989; Kato and others 2018); (2) mesozooplankton are either herbivores (TP = 2), carnivores (TP = 3), or omnivores (2 < TP < 3) (Hannides and others 2009); and (3) the grazing food chain is the primary trophic pathway. Assumptions (2) and (3) further lead to another assumption that (4) the omnivore is divided into fractions of herbivores and carnivores, depending on contributions from the omnivore’s diets with TP = 1 and TP = 2 to its biomass (for example, an omnivore with TP = 2.5 is equal to 50% herbivore and 50% carnivore). The iTP value was deconvoluted to biomass at TP = 2 (B2) and biomass at TP = 3 (B3) by rewriting Eq. 1 as follows:

$$ iTP{ } = \frac{{2 \times B_{2} + 3 \times { }B_{3} }}{{B_{T} }} $$
(7)
$$ B_{T} = B_{2} + B_{3} $$
(8)

It should be noted that contributions from the microbial loop were not considered.

Statistical Analysis

To test differences in species composition between mesozooplankton communities, we employed nonmetric multidimensional scaling (NMDS) based on the Bray–Curtis dissimilarity using species abundance data (that is, the number of individuals N) for each site. Taxonomic vectors (that is, taxonomic groups) and ecological vectors (for example, BT, MT, NT, H', and J') controlling species composition were plotted on the NMDS space. Analysis of variance (ANOVA) was applied to compare the difference in taxonomic and ecological factors between regions (that is, offshore P sites vs. nearshore J and S sites). ANOVA was also used to examine the significance of differences in iTP values between periods of bloom (April to June, n = 5) and post-bloom (July to August, n = 9) of phytoplankton in the western North Pacific (Ikeda and others 2008). To test the mean difference between regions (offshore P sites vs. nearshore J and S sites) taking a covariate into account, analysis of covariance (ANCOVA) was applied to (1) the relationship between species biomass (response variable) and species body weight (covariate) in different regions; and (2) the relationships between iTP (response variable) and H', J', MT, and the Copepoda proportion (covariates) in different regions. All statistical analyses and graphing were performed at a significance level of α = 0.05 using R 3.5.3 (R Development Core Team 2019). Main data are summarized in Table S4.

Results

The offshore P sites were characterized by low surface water temperatures and high nitrate concentrations and were dominated by large size phytoplankton groups (Table 1). By contrast, the nearshore J and S sites had high surface water temperatures and low nitrate concentrations, where small size phytoplankton groups were dominant (Table 1). In all the study sites, copepods dominated (39–99.5%) the whole community biomass. Among the subclass Copepoda, species composition of mesozooplankton communities differed greatly among study sites (Figure 2). In the P sites, Neocalanus spp. (family Calanidae) accounted for > 70% of total mesozooplankton biomass. Biodiversity in the offshore P sites was significantly lower (n = 6, H' = 0.84–2.51, J' = 0.04–0.21) than that in the nearshore J and S sites (n = 8, H' = 1.77–3.59, J' = 0.35–0.54), where Calanus spp. (family Calanidae) were dominant, with a biomass proportion of 10–55% (ANOVA, H' vs. region: F = 14.7, p = 0.002; J' vs. region: F = 59.0, p < 0.001). The contribution of Malacostraca was small in the offshore P sites (< 4%) and substantial in the nearshore J and S sites (0–35%). Hydrozoa made a major contribution to the communities in the J sites (12–21%), whereas the contribution of Chaetognatha was considerable in the S sites (10–24%).

Figure 2
figure 2

Biomass proportion of major taxonomic groups in mesozooplankton communities in the study sites; blue: Calanus (Calanidae); sky blue: Paracalanidae; light green: Metridinidae; green: Hydrozoa; pink: Malacostraca; red: Neocalanus (Calanidae); brown: Eucalanidae; purple: Other Copepoda; orange: Chaetognatha; black: Others.

Full size image

The NMDS results indicated that the community composition of the offshore P sites was significantly different from that of the nearshore J and S sites (Figure 3). The distance between study sites on the NMDS space corresponded with the geographic distance between sites: the community compositions of the P and S sites had less variation within their respective local areas. Although the J sites were geographically close to one another, their community compositions differed. The numbers of individuals of Calanus spp., Neocalanus spp., Paracalanidae, other Copepoda, Chaetognatha, Malacostraca, and others were significant (p < 0.05) taxonomic vectors (Figure 3a). Similarly, the ecological vectors N, BT, MT, H', J', δ15NGlu, δ15NPhe, and iTP were significantly (p < 0.05) related to species composition (Figure 3b). The δ15NPhe values of mesozooplankton communities tended to be low in the offshore P sites and high in the nearshore J and S sites (Figure 3b). The mean δ15NGlu, δ15NPhe, and iTP values ranged from 11.31 to 17.31 ‰, from − 3.33 to 5.15 ‰, and from 1.86 to 2.97, respectively (Table 2).

Figure 3
figure 3

Plot of nonmetric multidimensional scaling (NMDS) for mesozooplankton communities among study sites (P1–P6 in red, J1–J4 in green, and S1–S4 in blue). Significant (a) taxonomic and (b) ecological factors (p < 0.05) controlling community composition are shown as gray vectors. The direction and length of vectors correspond with the relevance and strength of the variable on the NMDS space, respectively. CA Calanus, NC Neocalanus, PC Paracalanidae, OC other Copepoda, OT others, CH Chaetognatha, MA Malacostraca, NT total number of individuals, BT total community biomass, MT mean body weight (BT/NT), H′ Shannon diversity index, J′ Pielou’s evenness index.

Full size image
Table 2 δ15N Values for Glutamic Acid (δ15NGlu) and Phenylalanine (δ15NPhe) and iTP Values of Mesozooplankton Communities
Full size table

The offshore P sites had a large number of species with body weight M of 100 to 102 (Figure 4a), whereas the nearshore J and S sites were characterized by a large number of species with an M of 10–2 to 10–1 (Figure 4b, c). The relationship between M and species biomass B was significant for the three sites, with a steeper slope at the P sites where the large weight species (M > 100 and B > 103, mainly Neocalanus spp.) dominated the total mesozooplankton biomass (ANCOVA, F = 11.9, p < 0.001, slope: 0.89 for offshore P sites and 0.70 for nearshore J and S sites) (Figure 4d).

Figure 4
figure 4

Distribution of mesozooplankton body weight in communities of the a Pacific, b Joban/Kashima, and c Sagami sites. Bars and curves represent histograms and probability density functions, respectively. d Plot of log10 (biomass, B) versus log10 (body weight, M) for each species. B = M × N (the number of individuals). The plot colors red, green, and blue represent Pacific, Joban/Kashima, and Sagami sites, respectively.

Full size image

Among the trophic guilds across all the sites, herbivore biomass was dominant (51–99%) (Figure 5). The proportion of herbivore biomass in the P sites (87–99% of all trophic guilds) was characterized primarily by the biomass of Neocalanus spp. (> 88% of all herbivores), which had larger body weights (M = 4.5 ± 2.7 mg WW, mean ± SD) than other mesozooplankton (typically M < 1 mg WW) (Table S1). In the nearshore J and S sites, herbivore biomass accounted for 51–75% of all trophic guilds, which was characterized by small-sized Calanus spp. (M = 0.6 ± 0.2 mg WW, 21–75% of all herbivores) and Eucalanus spp. (MT = 0.7 ± 0.8 mg WW, 0.4–29% of all herbivores) (Table S1). The proportion of carnivores in the mesozooplankton biomass was relatively high in the nearshore J and S communities (1.8–37% of all trophic guilds) compared to the offshore P sites. Hydrozoa and Chaetognatha accounted for 55–94% and 49–66% of all carnivores in the nearshore J and S sites, respectively. The differences in the total community biomass (g WW) and proportions (%) of herbivores, carnivores, and others between regions were all significant (ANOVA, biomass vs. region: F = 9.2, p = 0.01; herbivores vs. region: F = 40.1, p < 0.001; carnivores vs. region: F = 10.4, p = 0.007; and others vs. region: F = 16.7, p = 0.002) (Figure 5).

Figure 5
figure 5

Correlation matrix plot for integrated trophic position (iTP), total community biomass (g WW haul−1), biomass proportion (%) of three feeding guilds (herbivores, carnivores, and others), and sampling sites (P: Pacific sites P1–P6 in red; J: Joban/Kashima sites J1–J4 in green; and S: Sagami sites S1–S4 in blue). Scatter plots, correlation coefficients, and data distribution are displayed on the bottom left, top right, and center, respectively.

Full size image

The four ANCOVAs indicated that iTP values were significantly higher in the offshore P sites with covariates H' (t = 3.2, p = 0.009), J' (t = 2.4, p = 0.04), mean body weight (MT) (t = 4.5, p < 0.001), and Copepoda proportion (t = 2.7, p = 0.02) than the nearshore J and S sites (Figure 6). The interaction terms between the four covariates and regions were all insignificant (F < 1.8, p > 0.2), indicating that slopes of regression lines for the four covariates and iTP were not significantly different between regions. There was no significant difference in iTP values between periods of bloom and post-bloom of phytoplankton (ANOVA, F = 0.002, p = 0.96). According to Eqs. (7) and (8), the proportions of biomass of TP = 2 (B2/BT) and TP = 3 (B3/BT) were 35% ± 20% (range, 3–62%) and 65% ± 20% (range, 38–97%), respectively, for the offshore P sites (n = 6) and 80% ± 14% (range, 58–97%) and 20% ± 14% (range, 3–42%), respectively, for the nearshore J and S sites (n = 6, excluding sites J3 and S3 where iTP was lower than 2) (Figure 7).

Figure 6
figure 6

Plot of iTP versus a Shannon diversity index H'; b Pielou’s evenness index J'; c log10 (mean body weight, MT); and d proportion of Copepoda biomass relative to total community biomass BT (%). The color and size of plots denote the study site (P sites in red, J sites in green, and S sites in blue) and community biomass, respectively.

Full size image
Figure 7
figure 7

Schematics representing the integrated trophic position (iTP) concept and the body weight hypothesis. CSIA-AA, compound-specific nitrogen isotope (δ15N) analysis of amino acids. The biomass pyramids were estimated using Eqs. (7) and (8).

Full size image

Discussion

In the present study, both the Shannon diversity index H' and Pielou’s evenness index J' in offshore P sites were significantly lower than those in nearshore J and S sites. Because J' decreases with the decrease in evenness of biomass among species, the results suggest that a small number of species that are dominant in terms of biomass decrease the J' index and increase the iTP index for mesozooplankton communities. This is also supported by the skewing of body weight distribution toward larger weights (up to about 102 g WW) in the offshore P sites, where the mesozooplankton communities had higher iTP values than the nearshore J and S sites. The results are consistent with the previous finding that biodiversity in stream macroinvertebrate communities is negatively correlated with iTP (Ishikawa and others 2017). However, the iTP values of stream macroinvertebrates were negatively correlated with H', but not J', suggesting that biomass-dominant species in stream food webs has intermediate TP values (Ishikawa and others 2017).

Figure 7 illustrates a hypothesis to explain the iTP values of mesozooplankton, the body weight hypothesis, which assumes a positive relationship between the TP and body weight of mesozooplankton. The observed variation in iTP values is well explained by the biomass proportion of copepods. In particular, the dominance of Neocalanus spp. was related to the high iTP values of mesozooplankton communities of the offshore P sites, whereas Calanus, Paracalanidae, Eucalanidae, and Metridinidae were responsible for the low iTP values of the nearshore J and S sites. It is reported that the largest Neocalanus species (N. cristatus), which is supposed to be a herbivore, actually feeds on sinking aggregates, including microzooplankton and detritus (Dagg 1993), suggesting that the TP value of N. cristatus is nearly 3. Our previous results using CSIA-AA reported that Neocalanus, with a larger body weight, have higher TP values (2.6–2.7), whereas other Calanidae, such as Paracalanus, with a smaller body weight, have lower TP values (2.2–2.3) (Matsubayashi and others 2020; but see also Hannides and others 2009 for zooplankton TP unrelated to body weight in the subtropical Pacific). These pieces of evidence suggest that body weight is the first-order determinant of the iTP values of mesozooplankton communities in the ocean.

The body weight hypothesis predicts that the iTP value of the whole animal community of a given food web, including large consumers such as fish and mammals, would be positively correlated with the mean body weight of the entire community. Previous studies reported that the correlation between TP and body size for fish is significant (for example, Arim and others 2010; Ishikawa and others 2021) or not significant (for example, Layman and others 2005). The former pattern was supported by inter-species variations in mesozooplankton TP and body weight as mentioned in the previous paragraph. Although the mesozooplankton community is a subset of marine food webs, mesozooplankton accounts for the greatest biomass up to 34% in the subarctic North Pacific ecosystem (Ikeda and others 2008). The mesozooplankton production relative to the total biological production in marine food webs would be higher than their biomass proportion, given that mesozooplankton turns over faster than large consumers do. Therefore, it is suggested that our finding be scaled up to the entire marine ecosystems, which will be further discussed later.

Chaetognatha accounted for more than 10% of mesozooplankton biomass of the S sites. However, the iTP values in the S sites were lower than those in the P and J sites, even though Chaetognatha are considered predators (Nagasawa and Marumo 1972) and should reflect a higher TP. Similarly, mesozooplankton communities of the J1, J2, and J3 sites, with low iTP values, were characterized by considerable proportions of Hydrozoa that feed on aggregate particles and/or small zooplankton (Purcell 1981). This is probably because (1) their ontogenetic life stage plays a role in the measured iTP and/or (2) the feeding modes of these organisms are more omnivorous than previously thought. In contrast, the mesozooplankton communities of the P sites, with high iTP values, were dominated by Neocalanus, which feeds mainly on particles, including both phytoplankton and small zooplankton (Takahashi and others 2008). These results suggest that the feeding guilds of dominant species do not necessarily correspond with the functionally expected iTP values of the communities to which they belong, probably due to the opportunistic or ontogenetic feeding strategy of mesozooplankton (for example, Takahashi and others 2008).

Our samples did not contain zooplankton smaller than 335 μm, which may have greatly contributed to the community biomass, particularly in the nearshore J and S sites. It is expected that small zooplankton would decrease the iTP values, because their TP values are considered to be around 2 (Basedow and others 2016). Therefore, the negative relationship between iTP and biodiversity would be even more intensified by considering small zooplankton. Furthermore, Gutiérrez-Rodríguez and others (2014) and Decíma and others (2017) found that the TDFGlu value between protistan consumers and their food is more variable and generally smaller than the typical TDFGlu value (8.0‰ ± 1.2‰), which was used to calculate the iTP value in this study. Protistan consumers showed the δ15N value of alanine (δ15NAla) decoupled with δ15NGlu, which offers a greater and more constant TDF for Ala than TDFGlu (Decíma and others 2017). However, there was a significantly positive correlation between the δ15NGlu and δ15NAla values for our mesozooplankton samples (R2 = 0.59, p < 0.001; Figure S1). These results are consistent with little or no contribution of protists to the samples (that is, two protists out of 155 taxa, Table S1), as the sampling net was too coarse to capture these species.

The iTP values of the western North Pacific mesozooplankton communities (range, 1.86–2.97; mean ± SD, 2.33 ± 0.34; n = 14) were more variable than those of the stream macroinvertebrate communities (range, 2.39–2.79; mean ± SD, 2.60 ± 0.13; n = 15) (Ishikawa and others 2017). It should be noted that the iTP values lower than 2 found in J3 (1.86) and S3 (1.91) sites might be due to an artifact resulting from the presence of undigested phytoplankton (TP = 1) in the intestine of zooplankton. The biomass pyramids were inverted (that is, B2 < B3) in the offshore P sites (Figure 7), suggesting that ecological efficiency (production at TP = 3 divided by production at TP = 2) is not interchangeable with B3/B2. This statement conflicts with our previous study (Ishikawa 2018), which argues that iTP can be linked to ecological efficiency. In the subarctic North Pacific, where our P sites were located, seasonal blooms of phytoplankton are rapidly consumed by microzooplankton, which in turn are consumed by large Copepoda, such as Neocalanus (Landry and others 1993). Neocalanus species live for more than 1 year (Kobari and Ikeda 1999; Tsuda and others 1999), whereas smaller species, such as Paracalanus, have life cycle of less than 1 year (Liang and Uye 1996). Therefore, the produced organic matter is accumulated at a higher TP after the phytoplankton bloom because the turnover time increases as TP and body weight increase, which allows Neocalanus to play an important role as a food reservoir for their predators, such as fish, during low production periods (Ikeda and others 2008). The inverted biomass pyramids of the P sites are consistent with this scenario, where species with large body weights and long periods of turnover dominate the total mesozooplankton biomass, which increase the iTP values of these communities. Overall, the compatibility between iTP and ecological efficiency would be strongly dependent on the range of species weight and turnover of the focal community.

The iTP index can be also linked to food chain length (FCL), which is a conventional proxy for the biomass pyramid (Tunney and others 2012; Ward and McCann 2017). It is expected that the correlation between iTP and FCL would be positive when the biomass pyramid is bottom-heavy. This is because the biomass is exponentially decayed as TP increases in the bottom-heavy pyramid, where consumers at low TP such as mesozooplankton mainly control the iTP value, and the contribution from high TP consumers to iTP is negligible. In this setting, both FCL and iTP can be functions of the ecological efficiency, which is assumed to be constant across all trophic pathways (for example, 10 ± 5.8%, Pauly and Christensen 1995). In contrast, the top-heavy pyramid holds high TP consumers with significantly large biomass, which contributes to increasing the iTP value of the whole community. Therefore, it would not be feasible for iTP to predict FCL under this scenario. In other words, by examining the relationship between iTP and FCL in the future studies, we may be able to estimate the topology of biomass pyramids in nature.

Factors controlling the iTP variations found in both stream macroinvertebrates and marine mesozooplankton remain unclear. The correlation between biodiversity and iTP can be rather positive in a simple food chain where only producers and herbivores exist, thus the addition of species should essentially increase the iTP value (Ishikawa and others 2017). Furthermore, it is expected that the decline in the iTP value converges to somewhere around 2, even if the analysis is extended to more diverse communities (for example, the North Pacific subtropical gyre where zooplankton diversity is high; McGowan and Walker 1979). Future studies applying the iTP index to a variety of ocean regions as well as other systems, such as terrestrial ecosystems, will contribute to resolving the long-standing debate about the relationship between biodiversity and multitrophic ecosystem functioning. In particular, high-frequency monitoring of iTP, in collaboration with other methodologies, such as DNA metabarcoding (or ‘environmental DNA,’ for example, Hirai and others 2020), will open up a new frontier of cost-effective and simultaneous assessment of biodiversity and ecosystem functioning in the ocean. The results are also expected to have significant implications for factors controlling the structures of global biomass pyramids (Hatton and others 2015). Long-term monitoring results have shown that the average TP value of fisheries has been decreasing, whereas that of human beings has been increasing, for the past half-century, which implies a severe crisis in food production with the growing human population (Pauly and others 1998; Bonhommeau and others 2013). Given that a higher iTP value of the focal community requires more primary production to sustain its biomass (Pauly and Christensen 1995), we conclude that the iTP index will provide a unique signature of carrying capacity under ongoing biodiversity loss in the ocean, where sustainable food production is desired.