Functional Convergence and Taxonomic Divergence in the Anchoveta (Engraulis ringens) Microbiome

Abstract

Gut microbial community assembly involves a critical bioenergetic trade-off, yet the gut microbes with roles in influencing intestinal metabolic homeostasis remain poorly understood in pelagic ecosystems. A central unresolved question is whether microbiome structure is primarily governed by stochastic geographic drift or by deterministic metabolic filters imposed by diet. Here, we test the metabolic release hypothesis, which posits that access to high-quality prey physiologically “releases” the host from obligate dependence on diverse fermentative symbionts. By integrating δ15N analysis with 16S rRNA metabarcoding in the anchoveta from the South Pacific waters (Engraulis ringens), we reveal a profound, diet-induced restructuring of the gut ecosystem. We demonstrate that trophic ascent triggers a deterministic collapse in microbial alpha diversity (rs = −0.683), driven by the near-complete competitive exclusion of fermentative bacteria (rs = −0.874) and the resulting dominance of a specialized proteolytic core. Mechanistically, the bioavailability of zooplankton-derived protein favors efficient endogenous hydrolysis over costly microbial fermentation, rendering functional redundancy obsolete. Crucially, we find that while metabolic function converges, taxonomic identity remains geographically structured (r = 0.532), suggesting that local environments supply the specific taxa to fulfill universal metabolic roles. These findings establish a link between δ15N as a nutritional physiology proxy of anchoveta and its gut for microbial functional state, bridging the gap between nutritional physiology and ecosystem modeling to better inform the management of global forage fish stocks.

1. Introduction

Microbial community assembly in the fish gut represents an ecological battleground where deterministic and stochastic forces vie for supremacy. A central debate in the field is whether this structure is primarily governed by immediate environmental filters—such as diet and trophic niche—or by biogeographic history, which encompasses dispersal limitations and isolation [1,2,3]. Substantial evidence demonstrates that trophic plasticity and feeding habits exert robust selective pressure, dictating bacterial composition in response to specific resources [4,5,6]. Simultaneously, the aquatic environment functions as a dynamic reservoir, where habitat characteristics and water quality drive initial colonization and succession [7,8,9]. However, this environmental influence is modulated by host selection, a process in which intrinsic factors like genotype and physiology impose strict biological filters that favor specific symbionts over transient colonizers [10,11]. This interplay between deterministic selection and the ecological drift resulting from geographic barriers [12] underscores the critical need to integrate community ecology with biogeography to fully understand holobiont resilience in changing aquatic ecosystems.

The persistence of a diverse gut microbial community is governed by a strict bioenergetic trade-off, a dynamic underpinned by metabolic cost theory. Maintaining a complex microbiota imposes a significant energetic burden on the host; evolutionarily, this cost is only justified if symbionts confer critical metabolic functions that the host genome lacks, such as the fermentation of recalcitrant polysaccharides [1,2]. Physiological evidence corroborates this tension: while herbivorous fish rely obligatorily on high fermentative diversity to hydrolyze plant matter, carnivores and higher-trophic-level species employ efficient endogenous proteolytic enzymes, frequently resulting in less diverse microbiomes and reduced functional dependence on bacteria [13]. In this context, the anchoveta (Engraulis ringens) serves as a compelling study model due to its trophic plasticity and central role in upwelling system food webs. Related Engraulis species dynamically adjust their diet between phytoplankton and zooplankton based on resource availability [14], making it crucial to establish the metabolic cost and microbial dependence in E. ringens. Determining whether this keystone species maximizes energy assimilation through specific microbial alliances during phytoplankton consumption will reveal hidden mechanisms of ecological resilience, facilitating the integration of microbial digestive physiology into fisheries productivity models [15].

Building on the premise of energetic cost, we propose the metabolic release hypothesis as a unifying framework to explain microbiome dynamics in pelagic fish. This hypothesis posits that the anchoveta’s trophic ascent toward zooplankton—a protein-rich, highly digestible prey [14,16]—physiologically releases the host from its obligate dependence on fermentative microbial consortia. Unlike herbivory, which demands diverse and metabolically costly microbial machinery to hydrolyze complex carbohydrates, predation on high-quality prey allows endogenous proteolytic enzymes to assume the primary digestive role [11]. Consequently, we predict a structural simplification of the microbiome during zooplanktivorous feeding, challenging the classical ecological notion that a broader diet necessarily increases associated biodiversity. Rather than diversification, access to high-quality resources should precipitate a controlled collapse of alpha diversity by eliminating functionally redundant symbionts [3]. This ecological pruning signals a transition toward functional specialization, where the remnant microbiota likely pivots from digestive competition to immunological or pathogen exclusion roles, thereby optimizing the anchoveta’s energy budget in a variable environment [15].

We propose reconceptualizing the use of stable nitrogen isotopes δ¹⁵N in the microbial ecology of pelagic fish. Beyond its traditional role as a discrete indicator of trophic position (who eats whom) [17], δ¹⁵N functions as a continuous, robust proxy for the biochemical complexity of the digestive substrate available to the microbiota [18]. This isotopic value integrates the history of nutrient assimilation, signaling the specific metabolic gradient encountered by symbionts—ranging from the high fermentative demand imposed by phytoplankton-derived complex carbohydrates to the readily available amino acids characteristic of zooplanktivorous diets [19]. Consequently, this chemical gradient acts as a determinant stabilizing filter within the digestive tract. The selective pressure exerted by substrate quality, quantitatively inferred via δ¹⁵N, is of sufficient magnitude to override stochastic signals from local geographic variability or passive dispersal [8,10]. Thus, nutritional physiology—dictated by trophic level—homogenizes the functional structure of the anchoveta microbiome, relegating geographic location to a secondary driver of community assembly [20].

We present this study as a functional dissection of the microbial ecology of the gut anchoveta (Engraulis ringens), designed to unravel the mechanisms governing bacterial community assembly in dynamic pelagic systems. We postulate that trophic ascent is the primary determinant of gut microbiome architecture. Specifically, the shift toward a zooplankton-rich diet—evidenced by elevated values δ¹⁵N [14]—imposes strong stabilizing selection that overrides geographic stochasticity [21]. Mechanistically, we propose that this metabolic filter decouples the host from the fermentative microbial dependence required at basal trophic levels. This decoupling precipitates a deterministic collapse of alpha diversity, driving the community toward a core functionally optimized for proteolysis, independent of host genotype or latitudinal location. Under this conceptual framework, our objective was to evaluate the role of δ¹⁵N values as the governing driver of gut microbiome complexity in E. ringens. To this end, we analyzed the covariation among isotopic signals of δ¹⁵N, nutrient bioavailability, and bacterial diversity to determine whether dietary specialization imposes a functional convergence that supersedes local phylogeographic signals. Therefore, we tested three specific predictions: (1) Specialization on zooplankton will induce a metabolic release, where the abundance of bioavailable substrates favors endogenous enzymatic digestion over microbial fermentation, thereby eliminating functional redundancy and simplifying the community. (2) The latitudinal divergence observed between the catch zones stems not from geographic isolation, but from distinct local oceanographic forcing on prey fields. (3) Fermentative and proteolytic guilds of gut microbiome are fundamentally plastic and predictable; specifically, environmental conditions that increase the relative abundance of zooplankton will consistently drive microbiome simplification.

2. Materials and Methods

2.1. Sampling Collection

We analyzed 8 individuals, averaging 11.8 cm in total length, captured during a fisheries-independent research survey in 2020 aboard the B/C “Abate Molina” from the Instituto de Fomento Pesquero (IFOP) (

Table 1. Summary of anchoveta stable isotopes and microbiome metrics. Catch zone (Zone), bulk tissue stable isotope signatures (δ¹³C, δ¹⁵N) and elemental C:N ratios, alongside gut/intestine microbial alpha diversity indices (ASV richness and Shannon index) for each individual anchoveta specimen analyzed. ASV: Amplicon Sequence Variant.

Fish ID	Zone	δ13C	δ15N	C:N Ratio	ASV Richness	Shannon Index
Fish-1	Pisagua	−17.58	11.30	6.81	289	5.14
Fish-2	Pisagua	−19.28	10.60	7.50	890	5.69
Fish-3	Pisagua	−19.66	10.60	4.91	1033	6.24
Fish-9	Pta Farellones	−17.74	12.70	5.79	158	3.54
Fish-11	Mejillones	−17.86	12.00	5.53	415	5.10
Fish-12	Mejillones	−18.05	11.60	4.58	228	4.82
Fish-17	Arica	−16.58	14.50	4.33	55	3.84
Fish-18	Arica	−15.76	11.90	3.61	87	3.67

). The survey collected specimens for this study in four regions: Arica (n = 2; ~12.9 cm), Pisagua (n = 3; ~11.4 cm), Mejillones (n = 2; ~11.6 cm) and Punta Farellones (n = 1; 11.2 cm) in South Pacific waters (Figure 1). All fish were captured from a depth range of 5 to 50 m using a midwater trawl. After capture, specimens were frozen at −20 °C. Coinciding with each fish sample, we collected 20 L of surface water and zooplankton using a Bongo plankton net with a ring diameter of 57 cm, length of 2.6 m, and mesh of 180 µm. The water was stored in total darkness, while the zooplankton was filtered, separated, dehydrated, and then stored under vacuum.

2.2. Lab Workup

2.2.1. Stable Isotopes

In the laboratory, we filtered seawater samples and divided them into two fiberglass filters (GFF Whatman^TM 1.5 µm, 4.7 cm (Whatman, Marlborough, MA, USA)). The filter for δ¹³C was placed in a desiccator with 20 mL of 37% HCl to remove inorganic carbon, and the filter for δ¹⁵N was stored in aluminum foil at −80 °C [22]. The zooplankton was separated by size fraction into mesozooplankton (0.2–2 mm) and macrozooplankton (>2.0 mm). All tissues were analyzed for δ¹⁵N and δ¹³C at the Universidad Andres Bello, Stable Isotope Facility (LAI) using a continuous flow mode (CF) “Nu-Instruments” mass spectrometer, coupled with a Eurovector EA-3024 elemental analyzer (Eurovector, Pavia, Italy). Isotope ratios are reported in δ notation, using standard Pee Dee Belemnite for δ¹³C and Atmospheric Nitrogen for δ¹⁵N. Thus, δ¹³C or δ¹⁵N = [(R sample/R standard) − 1], where R is ¹³C/¹⁴C or ¹⁵N/¹⁴N, respectively. This analysis had an accuracy of ±0.1‰ for δ¹⁵N and δ¹³C.

2.2.2. DNA Extraction, 16S Amplicon-Based Sequencing, and Taxonomic Assignment

DNA Extraction: DNA was extracted from 25–30 mg of stomach-intestine tissue using the PowerSoilTM DNA Isolation Kit (MoBio Laboratories, Solana Beach, CA, USA), following the manufacturer’s protocol. Homogenization was performed on a TissueLyser II (Qiagen, Hilden, Germany) with two 1-min cycles at 30 Hz, separated by a 1-min pause. DNA concentration was quantified using a Qubit 4 fluorometer with the QubitTM dsDNA BR Assay (Thermo Fisher Scientific, Waltham, MA, USA). Extracts were stored at −20 °C until further processing.

16S rRNA gene amplification and sequencing: Prokaryotic amplicon libraries were constructed by Genoma Mayor SpA (Santiago, Chile) (http://www.genomamayor.com/ (accessed on 1 March 2021)). The V4 hypervariable region of the 16S rRNA gene was amplified using the primer pair 515F-Y (5′-GTGYCAGCMGCCGCGGTAA) and 806RB (5′-GGACTACNVGGGTWTCTAAT), modified with Illumina-specific adapters and barcodes according to the Earth Microbiome Project protocols (https://earthmicrobiome.ucsd.edu/protocols-and-standards/16s/ (accessed on 14 April 2021)). PCR amplification was performed using 2X KAPA HiFi HotStart Ready Mix (KAPA Biosystems, Cape Town, South Africa). Products were purified with AMPure XP beads (Beckman Coulter, CA, USA), quantified via the Quant-iT PicoGreen dsDNA Assay Kit (Invitrogen, Waltham, MA, USA) on a HOEFER DQ300 fluorometer, and sized using a DNF-900 Kit (Advanced Analytical Technologies, Inc., Ames, IA, USA) on a Fragment Bioanalyzer. Libraries were pooled and prepared following the Illumina Denature and Dilute Libraries Guide, then sequenced on an Illumina HiSeq 2500 platform (2 × 250 bp) (Illumina, San Diego, CA, USA) targeting 100,000 reads per library. Raw sequence data are available in the NCBI SRA database under BioProject ID PRJNA1043743.

Bioinformatics: Sequence data were processed using the DADA2 v1.26.0 R package [23]. Low-quality sequences were filtered and trimmed, and paired-end reads were merged and denoised to generate Amplicon Sequence Variants (ASVs). Taxonomy was assigned to each ASV using the Ribosomal Database Project (RDP) Naive Bayesian Classifier, trained specifically for the V4 region of the 16S rRNA gene [24].

2.3. Data Analysis

To test our first prediction, we calculated the alpha-Shannon microbial diversity (H’) as follows:

$$H^{\prime }=-\sum pi LN pi$$

where pi represents the relative abundance of each amplicon sequence variant (ASV) within an individual anchoveta gut.

For our second prediction, ASVs were categorized by putative metabolic function. We distinguished between a proteolytic microbiome, associated with zooplankton digestion in marine fish (e.g., Vibrio, Photobacterium, Pseudoalteromonas, Shewanella), and a fermentative microbiome, linked to the digestion of detritus or complex carbohydrates (e.g., Clostridia, Bacteroides, Cetobacterium, and certain Enterobacteriaceae). We removed chloroplasts and cyanobacteria from the analysis. To mitigate compositional bias, we quantified the balance between these functional groups using a log-ratio metric, termed the Y-score:

$$\mathrm{Y}\mathrm{s}\mathrm{c}\mathrm{o}\mathrm{r}\mathrm{e}=LN({\displaystyle \frac{proteolytic abundance}{fermentative abundance}})$$

A positive Y-score indicates a microbiome dominated by proteolytic bacteria, while a negative value signifies dominance by fermentative taxa; a score of zero reflects functional equilibrium. The relationships between these microbial metrics (H’ and Y-score) and trophic indicators were assessed using Spearman’s rank correlation (rs), chosen for its robustness with small sample sizes and its ability to detect monotonic relationships without assuming linearity. To test our third prediction regarding functional convergence versus geographic influence, we employed Mantel tests. We hypothesized that microbial community dissimilarity would correlate more strongly with trophic distance δ¹⁵N than with geographic distance. Consequently, we correlated the microbiome dissimilarity matrix (Bray–Curtis) against the δ¹⁵N Euclidean distance matrix and, separately, against the geographic distance matrix (km) between capture zones. Finally, to visualize these relationships, we performed a Principal Coordinate Analysis (PCoA) based on Bray–Curtis distances, overlaying vectors for the fishing zone and δ¹⁵N. Statistical analyses were performed using Python 3.x [25] with the following packages: “pandas” for data manipulation [26], “NumPy” for numerical computing [27], “SciPy” for statistical tests, and stats models for post hoc analyses [27].

3. Results

We identified a total of 2669 prokaryotic ASV across 93 orders in anchoveta stomach samples collected from the study area. The microbial community was dominated by the orders: Alteromonadales, Cellvibrionales, Chitinophagales, Oxyphotobacteria, Flavobacteriales, Lactobacillales, Rhodobacterales, Synechococcales, Verrucomicrobiales, and Vibrionales. Analysis of alpha diversity revealed spatial heterogeneity. Mean Shannon diversity indices were comparable between the Pisagua (5.7 ± 0.55) and Mejillones (4.96 ± 0.19) fishing zones. However, diversity in these areas was markedly higher than in the Arica (3.76 ± 0.12) and Punta Farellones (3.54) zones (Table 1).

Our analysis revealed a strong, statistically significant negative correlation between stable nitrogen isotopes and the alpha diversity of the anchoveta gut microbiome (rs = −0.587, p = 0.027) (Figure 2). Anchoveta with lower isotopic values, indicative of a mixed phytoplankton-zooplankton diet (see Supplementary Materials), harbored highly diverse microbial communities (H’~6.0). Conversely, trophic ascent toward a strictly zooplanktivorous diet (δ¹⁵N > 14‰) was associated with a marked collapse in microbial diversity (H’ < 5.0) (Table 1). These data support the existence of a diet-dependent metabolic filter: the availability of highly digestible substrates in zooplankton favors endogenous enzymatic digestion over microbial fermentation. This shift reduces the necessity for functional redundancy, precipitating an ecological simplification of the microbiota toward a specialized core, a pattern consistent with metabolic release theory in dynamic pelagic systems. Thus, dietary specialization on zooplankton (high δ¹⁵N) drives a significant reduction in gut microbiome diversity. This validates our prediction that trophic ascent induces a metabolic release, wherein endogenous digestion displaces microbial fermentation, eliminating redundancy and simplifying community structure.

Our functional analysis of the anchoveta gut microbiome reveals a profound restructuring of the bacterial community, predictably driven by the host’s trophic position (

Table 2. Metabolic guild abundance and derived functional index. The abundance (sequence counts) of proteolytic and fermentative bacterial guilds detected in each anchoveta specimen. The Yscore quantifies the functional balance of the gut community, calculated as the natural logarithm of the ratio of proteolytic to fermentative abundance.

Fish ID	Proteolytic Abundance	Fermentative Abundance	Yscore
Fish-1	9821	6549	0.405
Fish-2	16,278	11,141	0.379
Fish-3	17,260	1988	2.161
Fish-9	17,812	6433	3.291
Fish-11	18,656	6354	1.077
Fish-12	10,778	991	2.387
Fish-17	1055	83	2.542
Fish-18	9600	393	3.196

). By integrating δ¹⁵N data with bacterial functional profiles, we demonstrate that dietary variation acts as a determinant ecological filter shaping microbial guild composition. The data show an overall dominance of the proteolytic guild (>90% relative abundance), consistent with the physiology of a carnivorous fish. However, a critical dynamic exists within the minority component: a strong, statistically significant negative correlation (rs = −0.874, p < 0.05) (Figure 3) links trophic level to the abundance of fermentative bacteria. In individuals at lower trophic levels (δ¹⁵N~10.5), fermentative bacteria maintain a modest but detectable presence (~10%), likely sustained by the ingestion of phytoplankton-derived carbohydrates. As the anchoveta ascends trophically (δ¹⁵N > 12.5), driven by oceanographic conditions favoring zooplankton consumption, we observe a virtual collapse of the fermentative guild. This is illustrated by a linear increase in the proteolytic-to-fermentative log-ratio (Figure 3A), indicating that specialization on a protein-rich diet actively eliminates the ecological niche for fermenters. This pattern supports the hypothesis that regional differences in anchoveta microbiomes stem not from geographic isolation, but from local functional responses to distinct prey fields (phytoplankton vs. zooplankton), validating the substrate-driven assembly model. We conclude that microbiome assembly in anchoveta is a functionally deterministic process: trophic ascent forces a metabolic specialization that maximizes the proteolytic guild while competitively excluding fermenters, confirming that observed microbiome variability is a plastic response to local prey availability.

We used a PCoA based on Bray–Curtis distances to evaluate the microbial community structure of the anchoveta in relation to geographic location and δ¹⁵N values. Visually, Figure 4 reveals strong spatial structuring, with samples clearly segregated by their location of origin (Arica, Mejillones, Pisagua, Punta Farellones). This observation is robustly validated by a Mantel test, which confirmed a significant correlation between microbiome dissimilarity and geographic distance (r = 0.532, p = 0.002). This result indicates that local environmental conditions or regional microbial biogeography exert a primary influence on the taxonomic identity of gut symbionts. In contrast, the influence of δ¹⁵N on community structure was statistically non-significant (r = 0.272, p = 0.124), despite an apparent visual association in the Arica vector. This indicates that while diet modulates alpha diversity and functional potential (as demonstrated by the proteolytic/fermentative ratio), it does not drive a strict taxonomic convergence across distant regions. Different anchoveta populations likely assemble their microbiomes using locally available bacterial taxa to fulfill similar metabolic roles, thereby maintaining high regional beta diversity. Hence, the taxonomic identity of the anchoveta microbiome is primarily structured by local geography, not by δ¹⁵N. This finding adds a crucial nuance to the predictable plasticity hypothesis: while microbiome function converges metabolically (e.g., proteolysis) in response to zooplankton consumption, the specific species composition remains divergent and site-dependent. This refutes the idea that dietary forcing homogenizes microbial taxonomy at a regional scale, highlighting the importance of priority effects or local biogeography in community assembly.

4. Discussion

This study functionally dissects the assembly rules governing the gut microbiome of the anchoveta (Engraulis ringens), demonstrating that trophic ascent acts as a potent metabolic filter that predictably simplifies bacterial community structure [13]. We reveal a deterministic collapse of microbial alpha diversity and a near-complete competitive exclusion of fermentative bacteria with increasing reliance on a zooplankton diet. Our data provide robust mechanistic support for the concept of metabolic release in pelagic fishes [1,5]. The significant negative correlation between δ¹⁵N and microbial diversity validates our central prediction: as anchoveta ascend trophically, the shift toward a highly digestible, protein- and lipid-rich zooplankton diet [14] reduces the energetic necessity for microbial pre-digestion. This shift favors endogenous enzymatic processing, eliminating the niche space required for the functional redundancy observed in lower-trophic-level feeders [13]. This simplification is driven by a stark restructuring of functional guilds. While the community remains dominated by proteolytic bacteria—consistent with carnivorous physiology [28]—the near-perfect negative correlation between δ¹⁵N and fermentative bacteria offers novel insight. This indicates that the high load of bioavailable protein actively eliminates the ecological niche for carbohydrate fermenters, driving the community toward a core optimized for proteolysis. The significant increase in the proteolytic-to-fermentative log-ratio provides a quantitative metric for this specialization, confirming patterns observed in other fish species where diet dictates gut microbiota composition [4,29].

Our results paint a nuanced picture of microbiome assembly, challenging the prediction of complete taxonomic convergence. The collapse of fermentative bacteria with increasing δ15N confirms that assembly is functionally plastic and predictable: environmental conditions favoring zooplankton consistently drive functional simplification, prevailing over priority effects [6]. However, the significant correlation between microbiome dissimilarity and geographic distance refutes the idea of taxonomic homogenization. While diet forces function to converge (e.g., toward proteolysis), the specific bacterial species fulfilling this role vary geographically. This highlights the crucial role of local microbial biogeography: different anchoveta populations recruit locally available taxa (e.g., from the water column; [30]) to fulfill the same core metabolic demand. Thus, while the host determines what metabolic roles are required, the environment dictates who fills them.

The ability to link the fish’s trophic position (via δ¹⁵N) to the functional simplicity of its microbiome is a major step toward predictive ecology. For the anchoveta fishery, our model suggests that tracking δ¹⁵N may provide a proxy for monitoring the functional state of the stock’s health. While our reliance on 16S rRNA gene sequencing provides strong correlative support for these functional conclusions, direct validation is needed. Future research should prioritize three key areas: use metagenomic sequencing to directly measure functional gene capacity (e.g., specific protease pathways) in order to conclusively confirm the competitive exclusion mechanism, conduct controlled feeding experiments across varying diets to measure the causality and metabolic cost of maintaining a diverse versus simplified microbiome, and combine stable isotopes, DNA metabarcoding, and bio-logging [18] to resolve the fine-scale temporal dynamics of this diet–microbiome coupling in the wild.