Valorization of Fish Waste via Anaerobic Digestion: A Systematic Literature Review and Future Research Agenda

Abstract

Fish processing and aquaculture wastes are protein- and lipid-rich by-products that can be used to produce renewable energy. However, evidence on their anaerobic valorization is limited. This bibliometric and systematic review maps the field and synthesizes evidence on the anaerobic digestion and dark fermentation of fish-derived waste. Scopus records from 2000 to 2025 were screened according to the PRISMA 2020 guidelines. A total of 164 articles comprised the bibliometric corpus and 120 research articles informed the qualitative synthesis. The annual publication growth rate was 13.29%, with 65% of publications occurring between 2019 and 2025. Most described experiments employed laboratory-scale batch assays, mesophilic conditions and co-digestion. Methane yields from fish offal, silage and recirculating aquaculture system sludge ranged from 48 to 1174 mL CH₄/g VS. These differences reflect variations in feedstock composition and preparation, proportion of fish waste, selection of co-substrates and operating conditions. The process performance was mainly constrained by ammonia, volatile fatty acids and long-chain fatty acids. The modified Gompertz model predominated, whereas multi-step dynamic modeling remained rare. Microbial studies, primarily 16S rRNA gene surveys conducted in a batch-based manner, linked fish waste digestion to bacteria that degrade proteins and lipids, as well as hydrogenotrophic methanogens. However, community responses depended on the composition of the feedstock, inoculum and loading rate. Only three dark fermentation studies were identified, two of which used fish-derived substrates. Overall, progress toward industrial implementation requires fraction-specific characterization, validation in continuous systems, integration of hydrogen and methane production, dynamic modeling, multi-omics, digestate-safety assessment and integrated techno-economic and life cycle assessment based on pilot- and industrial-scale data. To assess potential database coverage bias, the search was repeated in Scopus and an equivalent search was run in Web of Science. Five additional eligible studies were identified.

1. Introduction

According to the latest report from the Food and Agriculture Organization of the United Nations (FAO), global fisheries and aquaculture production reached a record 235 million tons in 2024, including 195 million tons of aquatic animals [1]. The total aquaculture production reached 141 million tons, of which 103 million tons consisted of aquatic animals [1]. In 2022 aquaculture first surpassed capture fisheries as the principal producer of aquatic animals. By 2024 this sector accounted for 53% of global aquatic animal production, compared with approximately 92 million tons produced by capture fisheries [1]. Since capture fisheries production has remained within a relatively narrow range of 86 to 94 million tons per year since the late 1980s, nearly all subsequent growth in global aquatic animal production has come from aquaculture [1].

During fishing operations, a part of the catch is discarded as bycatch or unwanted harvest and initial on-board processing generates the first stream of residues. Approximately 35% of the global fisheries and aquaculture harvest is lost or wasted each year [1,2]. Fish processing plants located on land generate substantial amounts of by-products, including skin, heads, bones and viscera, which (depending on the species, processing method and final product) constitute from 30 to 70% of the mass of the original raw material [3,4,5]. Further losses occur during storage, transport, wholesale, retail and consumption, particularly because fish products are highly perishable and depend on effective cold-chain management [6]. As global output from aquaculture and capture fisheries increases, the scale of these losses is growing, making better waste management and circular resource use essential [7]. Despite growing interest in improving material circularity and the use of alternative organic residues for biofuel production [8,9], fish waste encompasses a broad and heterogeneous class of feedstocks that remains incompletely characterized for anaerobic digestion (AD) [3,10,11,12,13,14,15]. In this context, converting fish waste into biogas offers two potential benefits: (i) improved waste management and (ii) renewable energy recovery, potentially creating economic value under favorable conditions [2,4]. Although studies published since 2004 have investigated both mono-digestion and co-digestion of fish waste [3,4,10,11,12,16,17,18,19], several authors have explicitly described the available evidence as scarce or limited [3,10,11,12]. This fragmented evidence base provides a clear rationale for systematically mapping and synthesizing the field.

Bibliometric analysis (BA) enables the processing of voluminous scientific datasets, thereby identifying studies of high scientific impact and mapping the bibliometric and intellectual structure of a research field by analyzing the social and structural ties among its scholarly outputs [20,21,22,23]. In contrast, systematic literature review (SLR) offers a qualitative technique [24] that systematically summarizes and synthesizes existing knowledge within a given domain [25]. The combination of both methods constitutes a bibliometric–systematic literature review (B-SLR) [26], an approach that combines quantitative network metrics with qualitative content analysis while retaining the principles of replicability and transparency typical of SLR [24,27,28]. Consequently, it offsets the individual limitations of BA and SLR and supports the derivation of robust theoretical insights [24]. This integrated approach substantially expands the analytical horizon, allowing researchers to capture subtle thematic nuances that might otherwise remain unnoticed when applying BA or SLR in isolation [22,24]. Moreover, by combining quantitative metrics with in-depth qualitative interpretation, it becomes possible to examine a much larger number of publications, thereby ensuring that the identified trends, issues and research gaps are more closely linked to the current development trajectory of a given discipline [22]. Regular review articles, including bibliometric reviews based on data extracted from sources such as Web of Science or Scopus, help the scientific community synthesize established knowledge, identify research gaps and distinguish emerging research directions from recurring hypotheses [29,30]. The knowledge map presented in this paper acts as a radar in the publishing noise [22,31,32]. It aims to reveal real research gaps and guide future research efforts toward areas with the potential for novel contributions. This may reduce the risk of unnecessary duplication of research efforts [29,33].

Previous reviews have examined the use of fish- and fish-waste-based fertilizers in organic farming [34], treatments of fish industry waste, the environmental impacts of these treatments, valorization routes [35], biodiesel production from fish waste [36] and fish waste biorefinery approaches for high-value product recovery [37]. Sultan et al. [38] mapped the literature on the broadly defined field of fish waste management using bibliometric and content analyses, only briefly identifying biomethane generation as an emerging research topic. However, these studies did not jointly examine the bibliometric development of the field or the experimental evidence concerning fish-derived feedstocks, operating conditions, kinetic models and inhibition mechanisms in the context of gaseous biofuel production. This review aims to address this gap.

Against this background, the present review examines research on the anaerobic valorization of fish-derived waste published between 2000 and 2025. To characterize both the structure of the research field and the experimental evidence reported in individual studies, the review combines BA with SLR.

Accordingly, the following research questions were formulated to define the scope of the review: (Q1) What temporal, citation, geographical and institutional patterns characterize research on the anaerobic conversion of fish-derived wastes between 2000 and 2025? (Q2) Which categories of fish-derived waste and co-substrates have been investigated in mono-digestion and co-digestion systems? (Q3) What methane yields, optimization strategies, kinetic models and process limitations have been reported? (Q4) To what extent have dark fermentation and integrated hydrogen–methane systems been investigated for fish-derived wastes? (Q5) What evidence gaps remain and which research directions may support process scale-up and industrial implementation?

To the best of the authors’ knowledge, no previous review identified in the preliminary search combined bibliometric mapping with a systematic synthesis of experimental evidence on gaseous biofuel production from fish-derived wastes.

2. Materials and Methods

Scopus was selected as the primary bibliographic source for the integrated B-SLR because it provides structured, internally consistent metadata for citation, affiliation, keyword, and cited-reference analyses. The bibliometric and qualitative datasets used in the main analysis were derived exclusively from Scopus. It also supports reproducible, field-based searching and data export and offers broad coverage of engineering, environmental, agricultural, fisheries and energy literature relevant to the anaerobic conversion of fish-derived residues [39,40]. This decision prioritized the consistency and reproducibility of the bibliometric dataset. It was not based on the assumption that Scopus provides exhaustive thematic coverage. Since the coverage and indexing policies of bibliographic databases differ, studies that are only indexed in other sources may have been omitted. The implications of this restriction are discussed in Section 2.3.

2.1. Search Strategy, Eligibility Criteria and Study Selection

The review procedure followed the B-SLR framework proposed by Marzi et al. [24]. The search was designed to identify research articles published between 2000 and 2025 on the anaerobic processing of fish-derived waste. All records were retrieved from Scopus on 12 November 2025. This corpus merges descriptors related to fish waste with terms specific to AD and dark fermentation. Terms related to dark fermentation were included to capture studies investigating hydrogen production and integrated hydrogen to methane systems within the defined review scope (

Table A1. Final Scopus search query.

Final core corpus

TITLE-ABS-KEY(“fish waste*” OR “finfish waste*” OR “fish by-product*” OR “fish byproduct*” OR “fish offal” OR “fish viscera” OR “fish silage” OR “fish ensilage” OR “ensiled fish” OR “stick water” OR “stickwater” OR “press cake” OR “presscake” OR “surimi wash water” OR “fish processing waste*” OR “fish processing wastewater” OR “fish processing effluent” OR “fishmeal plant effluent” OR “fish cannery wastewater” OR “seafood waste” OR “seafood processing waste*” OR “seafood processing wastewater” OR “seafood processing effluent” OR “aquaculture sludge” OR “fish sludge” OR “aquaculture effluent” OR “aquaculture wastewater” OR “recirculating aquaculture system” OR “RAS sludge” OR “RAS backwash” OR “drum filter backwash” OR “rest raw material”) AND TITLE-ABS-KEY(“anaerobic digest*” OR “methane ferment*” OR biogas OR biomethane OR methanogenesis OR “biogas production” OR “biogas yield” OR “specific biogas production” OR “cumulative biogas production” OR “methane production” OR “methane yield” OR “specific methane production” OR “specific methane yield” OR “ultimate methane yield” OR “cumulative methane production” OR “biochemical methane potential” OR BMP OR codigestion OR “co-digestion” OR “co digestion” OR “anaerobic co-digestion” OR AcoD OR “two-stage anaerobic digestion” OR “two stage anaerobic digestion” OR “two-phase anaerobic digestion” OR “two phase anaerobic digestion” OR “anaerobic membrane bioreactor” OR AnMBR OR “dark ferment*” OR biohydrogen OR “bio-hydrogen” OR “bio h2” OR bioH2 OR “H2 production” OR “H2 yield” OR “CH4 production” OR “CH4 yield” OR “hydrogen production” OR “hydrogen yield” OR “hydrogen production rate” OR HPR OR “hydrogen evolution rate” OR HER OR “specific hydrogen production rate” OR SHPR OR “biochemical hydrogen potential” OR BHP OR “bioCH4” OR “bio-CH4”)

).

All searches were performed using the Scopus Advanced Search interface and the TITLE-ABS-KEY field. The complete Boolean search strategy is presented in Table A1. The third and final search iteration returned 286 records. Using the duplicates_check() function in the Bibliometrix R package (version 5.2.0), a DOI-based deduplication was performed and two duplicate records were identified. The manual inspection identified three additional duplicates, leaving 281 unique records for title-and-abstract screening. The study selection process was reported in accordance with PRISMA 2020 (Figure 1 and Figure 2) and the completed checklist is provided in Appendix C (

Table A5. PRISMA 2020 checklist.

Section and Topic	Item #	Checklist Item	Location Where Item is Reported
Title
Title	1	Identify the report as a systematic review.	Title page
Abstract
Abstract	2	See the PRISMA 2020 for Abstracts checklist.	Abstract
Introduction
Rationale	3	Describe the rationale for the review in the context of existing knowledge.	Section 1
Objectives	4	Provide an explicit statement of the objective(s) or question(s) the review addresses.	Section 1
Methods
Eligibility criteria	5	Specify the inclusion and exclusion criteria for the review and how studies were grouped for the syntheses.	Section 2.1; Supplementary Text S1
Information sources	6	Specify all databases, registers, websites, organisations, reference lists and other sources searched or consulted to identify studies. Specify the date when each source was last searched or consulted.	Section 2 and Section 2.1
Search strategy	7	Present the full search strategies for all databases, registers and websites, including any filters and limits used.	Section 2.1; Appendix A, Table A1 (Scopus) and Table A2 (Web of Science)
Selection process	8	Specify the methods used to decide whether a study met the inclusion criteria of the review, including how many reviewers screened each record and each report retrieved, whether they worked independently and if applicable, details of automation tools used in the process.	Section 2.1; Supplementary Text S1
Data collection process	9	Specify the methods used to collect data from reports, including how many reviewers collected data from each report, whether they worked independently, any processes for obtaining or confirming data from study investigators and if applicable, details of automation tools used in the process.	Supplementary Text S1
Data items	10a	List and define all outcomes for which data were sought. Specify whether all results that were compatible with each outcome domain in each study were sought (e.g., for all measures, time points, analyses) and if not, the methods used to decide which results to collect.	Supplementary Text S1
	10b	List and define all other variables for which data were sought (e.g., participant and intervention characteristics, funding sources). Describe any assumptions made about any missing or unclear information.	Supplementary Text S1
Study risk of bias assessment	11	Specify the methods used to assess risk of bias in the included studies, including details of the tool(s) used, how many reviewers assessed each study and whether they worked independently and if applicable, details of automation tools used in the process.	Supplementary Text S1 (methodological reporting quality appraisal; no formal risk-of-bias assessment)
Effect measures	12	Specify for each outcome the effect measure(s) (e.g., risk ratio, mean difference) used in the synthesis or presentation of results.	Supplementary Text S1
Synthesis methods	13a	Describe the processes used to decide which studies were eligible for each synthesis (e.g., tabulating the study intervention characteristics and comparing against the planned groups for each synthesis (item #5)).	Supplementary Text S1
	13b	Describe any methods required to prepare the data for presentation or synthesis, such as handling of missing summary statistics, or data conversions.	Supplementary Text S1
	13c	Describe any methods used to tabulate or visually display results of individual studies and syntheses.	Section 2.2; Supplementary Text S1
	13d	Describe any methods used to synthesize results and provide a rationale for the choice(s). If meta-analysis was performed, describe the model(s), method(s) to identify the presence and extent of statistical heterogeneity and software package(s) used.	Supplementary Text S1
	13e	Describe any methods used to explore possible causes of heterogeneity among study results (e.g., subgroup analysis, meta-regression).	Supplementary Text S1 (no subgroup meta-analysis or meta-regression)
	13f	Describe any sensitivity analyses conducted to assess robustness of the synthesized results.	Section 2.1 and Section 3.4
Reporting bias assessment	14	Describe any methods used to assess risk of bias due to missing results in a synthesis (arising from reporting biases).	Supplementary Text S1 (no formal reporting-bias assessment)
Certainty assessment	15	Describe any methods used to assess certainty (or confidence) in the body of evidence for an outcome.	Supplementary Text S1 (no formal certainty assessment)
Results
Study selection	16a	Describe the results of the search and selection process, from the number of records identified in the search to the number of studies included in the review, ideally using a flow diagram.	Section 2.1 and Figure 2 (primary Scopus search); Section 3.4 (database-coverage sensitivity analysis)
	16b	Cite studies that might appear to meet the inclusion criteria, but which were excluded and explain why they were excluded.	Section 2.1; Supplementary Text S1
Study characteristics	17	Cite each included study and present its characteristics.	Section 3.2.1, Section 3.2.2, Section 3.2.3, Section 3.2.4, Section 3.2.5 and Section 3.2.6; Table 2, Table 3, Table 4, Table A3 and Table A4
Risk of bias in studies	18	Present assessments of risk of bias for each included study.	Supplementary Text S1
Results of individual studies	19	For all outcomes, present, for each study: (a) summary statistics for each group (where appropriate) and (b) an effect estimate and its precision (e.g., confidence/credible interval), ideally using structured tables or plots.	Section 3.2.2, Section 3.2.3, Section 3.2.4, Section 3.2.5 and Section 3.2.6; Table 2, Table 3, Table 4 and Table A3
Results of syntheses	20a	For each synthesis, briefly summarise the characteristics and risk of bias among contributing studies.	Section 3.2.1, Section 3.2.2, Section 3.2.3, Section 3.2.4, Section 3.2.5 and Section 3.2.6; Table 3, Table 4 and Table A3
	20b	Present results of all statistical syntheses conducted. If meta-analysis was done, present for each the summary estimate and its precision (e.g., confidence/credible interval) and measures of statistical heterogeneity. If comparing groups, describe the direction of the effect.	Narrative synthesis: Section 3.2.1, Section 3.2.2, Section 3.2.3, Section 3.2.4, Section 3.2.5 and Section 3.2.6; Supplementary Text S1
	20c	Present results of all investigations of possible causes of heterogeneity among study results.	Possible sources of heterogeneity were explored narratively in Section 3.2.2, Section 3.2.3, Section 3.2.4, Section 3.2.5 and Section 3.2.6
	20d	Present results of all sensitivity analyses conducted to assess the robustness of the synthesized results.	Section 3.4
Reporting biases	21	Present assessments of risk of bias due to missing results (arising from reporting biases) for each synthesis assessed.	N/A
Certainty of evidence	22	Present assessments of certainty (or confidence) in the body of evidence for each outcome assessed.	N/A
Discussion
Discussion	23a	Provide a general interpretation of the results in the context of other evidence.	Section 3.1, Section 3.2, Section 3.3, Section 3.4 and Section 5
	23b	Discuss any limitations of the evidence included in the review.	Section 2.3; Section 3.2.2, Section 3.2.3 and Section 3.2.4; Section 5
	23c	Discuss any limitations of the review processes used.	Section 2.3
	23d	Discuss implications of the results for practice, policy and future research.	Section 4 and Section 5
Other information
Registration and protocol	24a	Provide registration information for the review, including register name and registration number, or state that the review was not registered.	Section 2.1; Section 2.3; Data Availability Statement
	24b	Indicate where the review protocol can be accessed, or state that a protocol was not prepared.	Retrospective OSF methodological record is identified Section 2.1; Supplementary Text S1; Data Availability Statement
	24c	Describe and explain any amendments to information provided at registration or in the protocol.	N/A
Support	25	Describe sources of financial or non-financial support for the review and the role of the funders or sponsors in the review.	Funding statement
Competing interests	26	Declare any competing interests of review authors.	Conflicts of Interest statement
Availability of data, code and other materials	27	Report which of the following are publicly available and where they can be found: template data collection forms; data extracted from included studies; data used for all analyses; analytic code; any other materials used in the review.	Data Availability Statement (the study selection datasets and bibliometric corpus are available via OSF)

) [27].

Studies were eligible if they met the following criteria: (i) were original research articles published in English between 2000 and 2025, (ii) investigated solid or liquid residues from fish processing or fisheries, (iii) examined AD, anaerobic co-digestion or dark fermentation and (iv) reported at least one outcome relevant to the review questions. These outcomes included biogas, biomethane or biohydrogen production; operating conditions; kinetic modeling; digestate characteristics; process stability and inhibition.

Reviews, book chapters, conference papers, editorials, letters, technical notes and errata were excluded because the evidence synthesis focused on primary research and considering secondary publications as independent evidence units could result in duplicate reporting of information from primary studies. Records were also excluded if the investigated material, biological process, outcome or study design fell outside the scope of the review. Full-text availability was not treated as a substantive eligibility criterion. However, access to the complete report was required for data extraction and methodological reporting quality appraisal.

Two reviewers screened the titles and abstracts of all 281 records independently using the above-described eligibility criteria. Any disagreements or unclear classifications were resolved through discussion. Inter-rater agreement was evaluated using Krippendorff’s alpha, yielding an alpha value of 0.81 [24,41]. A total of 117 records were excluded: forty-eight because the investigated material or substrate was outside the scope of the review, forty-one because the biological process was ineligible, twenty-four because the reported outcomes were outside the scope of the review and four because of an ineligible study design. Consequently, 164 records were retained for BA. Full-text retrieval was attempted for all 164 reports, but 44 could not be obtained. The remaining 120 reports underwent a full-text eligibility assessment and a methodological reporting quality appraisal; none was excluded at this stage. Thus, the 164 eligible bibliographic records formed the bibliometric corpus, whereas the 120 retrievable full-text reports formed the qualitative evidence synthesis corpus (Figure 2). The studies included in the qualitative evidence synthesis are listed in

Table A4. Studies included in the systematic literature review.

No.	Feedstock	Process	Reference
1	Fish waste and sisal pulp	Batch anaerobic mono-digestion and co-digestion	[10]
2	Fish waste and bagasse	Batch anaerobic mono-digestion and co-digestion	[18]
3	Pig manure, tuna fish waste and biodiesel waste	Batch anaerobic co-digestion with feed optimization	[56]
4	Fruit and vegetable waste with fish waste, abattoir wastewater and waste activated sludge	Mesophilic anaerobic co-digestion in sequencing batch reactors	[55]
5	Tuna, sardines, mackerel and needlefish waste with gorse	Batch anaerobic mono-digestion and co-digestion	[11]
6	Seaweed, brown algae, green algae and fish viscera	Batch anaerobic mono-digestion	[16]
7	Salmon fish waste, fish sludge and Jerusalem artichoke	Batch anaerobic mono-digestion and co-digestion after oil extraction	[61]
8	Pig manure, fish waste and biodiesel waste	Continuous mesophilic anaerobic co-digestion	[81]
9	Fish waste, bread waste, brewery grain waste and fish waste silages	Ensiling followed by batch anaerobic digestion	[17]
10	Fish by-product, steam-exploded Salix and cow manure	Mesophilic semi-continuous anaerobic co-digestion	[84]
11	Fish waste and residual strawberry extrudate	Mesophilic batch anaerobic co-digestion	[82]
12	Fish waste silage and cow manure	Mesophilic semi-continuous anaerobic co-digestion	[68]
13	Saline sludge from brackish aquaculture recirculation system	Batch anaerobic digestion with compatible solute addition	[71]
14	Fish waste silage and cow manure	Mesophilic continuous anaerobic co-digestion with metagenomic analysis	[57]
15	Dairy cow slurry and fish silage	Farm-scale mesophilic anaerobic co-digestion with energy balance	[95]
16	Blackwater with Greenlandic halibut offal and shrimp offal	Mesophilic batch anaerobic co-digestion compared with aerobic storage	[92]
17	Strawberry extrudate, fish waste and crude glycerol	Mesophilic lab-scale anaerobic co-digestion	[89]
18	Microalgal bacterial flocs from pikeperch aquaculture wastewater	Batch anaerobic digestion with pretreatment screening	[138]
19	Saline aquaculture sludge from a recirculating aquaculture system	Mesophilic ASBR anaerobic digestion with carbohydrate addition and ultrasonication pretreatment	[101]
20	Primary and secondary rainbow trout sludge	Anaerobic acidogenic digestion with indigenous Alcaligenes faecalis	[139]
21	Marine fish waste sludge from a recirculating aquaculture system	Anaerobic biomethane conversion with a halotolerant microbial consortium	[72]
22	Manure, straw, bagasse, fish processing wastewater, alcohol waste, food waste and human excrement	Industrial anaerobic co-digestion for bioCNG production with efficiency modeling	[83]
23	Digestates from food waste, sewage sludge, animal manure, whey permeates and fish ensilage	Digestate soil application and metal leaching assessment	[140]
24	Round goby heads, intestines, skin and bone residues with sewage sludge	Batch anaerobic co-digestion	[98]
25	Fish sludge and dairy manure	Drying and anaerobic co-digestion followed by fertilizer and logistics assessment	[76]
26	Fish waste and organic material from stranded beach debris	BMP assays for anaerobic digestion potential and residue valorization assessment	[141]
27	Whey, manure and fish ensilage	Batch BMP anaerobic mono-digestion and co-digestion	[58]
28	Tilapia and African catfish RAS sludge	Sequential UASB-EGSB anaerobic digestion with aerobic and anaerobic batch controls	[59]
29	Fish waste silage and cow manure	Semi-continuous CSTR anaerobic co-digestion with HRT and microbial analysis	[69]
30	Wastewater sludge with garden grass or fish waste	Anaerobic co-digestion blend optimization	[142]
31	Goat dung, chicken dung, fish waste, rice waste, POME and sewage sludge	Mesophilic BMP anaerobic digestion with inoculum comparison	[143]
32	Fish processing waste and liquid fraction from hydrothermal carbonization of bamboo residues	Batch anaerobic co-digestion with HTC liquid fraction addition	[93]
33	Fish processing waste with bamboo hydrochar	Batch anaerobic digestion with hydrochar addition	[94]
34	Fish ensilage, soapstock, alkaline fish glycerine and ethyl monoesters	Batch BMP anaerobic co-digestion	[91]
35	Fish waste and vegetable waste	Batch anaerobic mono-digestion and co-digestion	[107]
36	Tilapia head, carcass, viscera, fin, skin, scale and mixed residues	Batch BMP anaerobic digestion of separate waste fractions	[144]
37	Dolphin fish offal and river tamarind	Batch BMP anaerobic co-digestion with mixing ratio optimization	[85]
38	Fish processing wastewater	Batch and continuous mesophilic anaerobic digestion in a SA-ABR with calcium, cobalt and iron supplementation	[78]
39	Trout aquaculture sludge	Mesophilic anaerobic digestion in conventional and hybrid fixed bed reactors	[134]
40	Fishery by-products, rice bran and sewage sludge	Acidogenic fermentation followed by batch anaerobic co-digestion	[108]
41	Fish waste, sewage sludge and grass	Lab-scale anaerobic co-digestion with microbial and metabolomic analysis	[119]
42	Anchovy processing waste after fish oil extraction	Batch BMP anaerobic digestion after limonene extraction	[109]
43	Salmon filleting co-streams and cattle manure	Modeled mesophilic anaerobic co-digestion for biogas and fertilizer production	[96]
44	Sisal waste with marine fish processing waste and other co-substrates	LCA-based anaerobic co-digestion scenario with bioenergy co-production	[127]
45	Viscera from Argentine hake, Brazilian flathead, Brazilian codling and stripped weakfish	Digestive proteinase extraction and characterization	[145]
46	Dewatered fish sludge and manure solids	Phosphorus fertilizer assessment after drying, composting, separation or pyrolysis	[75]
47	Salicornia europaea and Salicornia ramosissima plant material	Batch BMP anaerobic digestion of halophyte biomass	[146]
48	Fish processing effluent	Anaerobic treatment in ABR followed by anaerobic filter	[79]
49	Fish solid waste and plant waste from aquaponics	Onsite anaerobic digestion in UASB and plant waste digester	[74]
50	Fish waste powder	Mesophilic batch anaerobic digestion with microbial seed comparison	[110]
51	Sewage sludge, food waste and fish waste	Semi-continuous anaerobic co-digestion with repeated fish waste spikes	[118]
52	Cod processing fish waste and manure residues	LCA and cost scenario for anaerobic co-digestion with CHP energy recovery	[122]
53	Fish silage	Batch anaerobic digestion with FeCl3 addition and magnetic field exposure	[99]
54	Fish intestines	Feedstock characterization and methane content analysis for biogas production	[147]
55	Farmed rainbow trout by-products	Mesophilic batch anaerobic mono-digestion with kinetic modeling	[112]
56	Aquaculture solids from rainbow trout RAS	Continuous anaerobic digestion with iron supplementation in freshwater and saline conditions	[103]
57	Fish sludge from Atlantic salmon and rainbow trout RAS	Batch anaerobic mono-digestion with varying initial solids concentration	[102]
58	Anchovy fillet leftovers	LimoFish valorization with anaerobic digestion option and LCA	[126]
59	Fish waste and anaerobic sewage sludge	Batch anaerobic co-digestion with sewage sludge to fish waste ratio optimization	[67]
60	Fish sludge and dead fish from sea trout production	Consequential LCA with anaerobic digestion valorization	[125]
61	Pig slurry, orange pomace and fish waste	Batch anaerobic co-digestion with fish waste incorporation	[12]
62	Fish waste and water hyacinth	Batch anaerobic co-digestion with RSM optimization	[87]
63	Fish waste and water hyacinth	Mesophilic batch anaerobic co-digestion with substrate ratio, inoculum and dilution testing	[86]
64	Raw and cooked tuna viscera with fat waste, dairy wastewater and secondary sludge	Batch anaerobic mono-digestion with thermal pretreatment and co-digestion tests	[66]
65	African catfish RAS sludge and greenhouse plant residues	Batch, semi-continuous CSTR and UASB anaerobic digestion with mono-fermentation and co-fermentation	[73]
66	Aquaculture wastewater and digested cow manure	Thermophilic pilot-scale anaerobic co-digestion with fertilizer recovery	[97]
67	Food waste residue from MCCA production and RAS fish sludge	Thermophilic batch anaerobic digestion and co-digestion followed by plant growth testing	[77]
68	Mixed fish waste and vegetable market waste	Mesophilic batch anaerobic co-digestion with kinetic modeling	[90]
69	RAS fish sludge, food waste and fruit and vegetable waste	Mesophilic BMP, batch pilot and semi-continuous anaerobic co-digestion	[104]
70	Fishery processing wastewater	Bioelectrochemically improved anaerobic digestion in AD-BES system	[100]
71	Sardine processing waste	Single-stage and two-stage anaerobic digestion with microbial pretreatment	[111]
72	Fish sludge from aquaponics	Coupled aquaponics with onsite UASB anaerobic digestion for nutrient recovery	[148]
73	Fish waste, water hyacinth and cow dung	Mesophilic batch anaerobic mono-digestion and co-digestion with ratio optimization	[88]
74	European seabass fecal waste from different aquafeeds	Anoxic batch anaerobic digestion for organic acid production and nutrient solubilization	[120]
75	Potato waste, leftover cooked food and fish waste	Batch anaerobic mono-digestion in small-scale floating-drum digester	[136]
76	Fish waste and primary sludge	BMP anaerobic co-digestion with mixing ratio and kinetic analysis	[19]
77	Sarcodia residuals and shrimp pond bottom sludge	Anaerobic digestion within zero waste aquaculture valorization	[137]
78	Fish waste and cow dung	Anaerobic digestion for biogas and transesterification for biodiesel	[135]
79	Brackish aquaculture sludge from RAS	UASB anaerobic digestion	[149]
80	Freshwater RAS sludge	Mesophilic ASBR anaerobic digestion with saline adaptation	[150]
81	Brackish aquaculture sludge from three RAS	Sludge characterization and UASB anaerobic digestion for methane production	[151]
82	Raw chitin from crab shells	Fungal prehydrolysis followed by dark fermentation and anaerobic digestion	[65]
83	Salmon fish sludge from land-based RAS	Batch dark fermentation for hydrogen and volatile fatty acid production	[64]
84	Model fruit and vegetable waste with acid-hydrolyzed fish waste	Hyperthermophilic dark fermentation co-digestion for hydrogen production	[63]
85	Fish waste and cow dung	Techno-economic biorefinery model with anaerobic co-fermentation and fertilizer recovery	[124]
86	UASB sludge from marine RAS with Cryptocaryon irritans life stages	Parasite survival testing under anaerobic sludge conditions	[152]
87	Seafood processing wastewater	Pilot-scale AnMBR treatment in real-time mode	[80]
88	Fish sludge	Batch SMFD anaerobic treatment with electron shuttles	[117]
89	Fish waste	Techno-economic modeling of hydrolysis and anaerobic digestion valorization routes	[5]
90	Mechanical filter wash water sludge from Clarias gariepinus RAS	Two-stage sedimentation for sludge concentration and further biogas use	[153]
91	Fish silage and cattle manure	Model-based low-solids anaerobic co-digestion with effluent recirculation and separation	[70]
92	Fish sludge and aquaculture effluent water from salmon RAS	LCA of anaerobic digestion and microalgae cultivation for nutrient and energy recovery	[123]
93	RAS sludge from Atlantic salmon fed two diets	Batch anaerobic mono-digestion with fish diet comparison	[13]
94	Dairy manure and aquaculture sludge	Batch anaerobic co-digestion with mixing ratio and HRT optimization	[14]
95	Olive mill wastewater, cow dung, fruit and vegetable waste and fish waste	Mesophilic batch anaerobic co-digestion with kinetic modeling	[15]
96	Artisanal fish waste from Tumaco	Mesophilic batch anaerobic mono-digestion with TS concentration testing	[4]
97	Fish waste and fish crude oil waste from carp viscera	Mesophilic batch anaerobic mono-digestion with microbial community analysis	[3]
98	Tilapia and Clarias filleting by-products from RAS	LCA of RAS farming with by-products used for biogas production	[154]
99	Raw and deoiled tuna waste with onion stalks	Semi-continuous anaerobic co-digestion with fish oil extraction and fertilizer recovery	[2]
100	Nile perch waste and cow rumen cud	Anaerobic co-digestion for biogas production	[155]
101	Nile perch heads, skin, offal and skeleton	Anaerobic digestion with fish waste fraction comparison	[156]
102	Fish processing wastewater	Continuous upflow aged refuse packed bioreactor treatment with methane recovery	[157]
103	Fish sludge	Anaerobic digestion with iron-sludge addition and intermittent aeration	[158]
104	Dead fish waste	Repeated-batch anaerobic digestion start-up with feeding strategy comparison	[159]
105	Brackish aquaculture sludge	Mesophilic batch and continuous anaerobic digestion	[160]
106	Olive flounder and starry flounder fractions	Mesophilic BMP anaerobic mono-digestion	[161]
107	Waste activated sludge and fish waste	Mesophilic batch anaerobic co-digestion	[162]
108	Saline fish wastewater and cow manure	BMP anaerobic co-digestion with salinity toxicity assessment	[163]
109	Fish processing wastewater	Continuous mesophilic anaerobic digestion	[164]
110	Waste activated sludge and aquaculture sludge	Mesophilic batch anaerobic co-digestion	[165]
111	Sewage sludge and acid-fermented fish by-product broth	Anaerobic co-digestion	[166]
112	Saline tuna processing wastewater	Continuous anaerobic digestion	[167]
113	Saline fish evisceration wastewater	Batch anaerobic digestion at different feed-to-microorganism ratios and salinities	[168]
114	Fish waste and seagrass with macroalgae	Batch anaerobic mono-digestion and co-digestion	[169]
115	Tilapia and sturgeon processing fractions and water treatment sludge	Batch anaerobic mono-digestion with first-order kinetic modeling	[170]
116	Fish waste and strawberry extrudate	Anaerobic co-digestion at different mixing ratios	[171]
117	Aquaculture wastewater sediment	Anaerobic mono-digestion with biomethane yield modeling	[172]
118	Trout processing by-products	Anaerobic mono-digestion at different organic matter loads	[173]
119	Brackish aquaculture sludge	Anaerobic mono-digestion under different electron-acceptor conditions	[174]
120	Biologically pretreated Nile perch solid waste with fish scales as a biofilm carrier	Anaerobic digestion in packed-bed bioreactors with different biofilm carriers	[175]

BMP—Biochemical Methane Potential; RAS—Recirculating Aquaculture System; UASB-EGSB—Upflow Anaerobic Sludge Blanket-Expanded Granular Sludge Bed; CSTR—Continuous Stirred-Tank Reactor; HRT—Hydraulic Retention Time; POME—Palm Oil Mill Effluent; HTC—Hydrothermal Carbonization; LCA—Life Cycle Assessment; ABR—Anaerobic Baffled Reactor; UASB—Upflow Anaerobic Sludge Blanket; CHP—Combined Heat and Power; RSM—Response Surface Methodology; MCCA—Medium-Chain Carboxylic Acid; AD-BES—Anaerobic Digestion Bioelectrochemical System; ASBR—Anaerobic Sequencing Batch Reactor; SMFD—Simultaneous Methanogenesis, Feammox and Denitrification.

.

This review was not prospectively registered in a public repository. To improve transparency of the methods used, a retrospective record of the methods and the associated study-selection datasets were deposited in the Open Science Framework after the database search, screening, data extraction, BA and qualitative evidence synthesis were completed (https://doi.org/10.17605/OSF.IO/FGJMB).

Further details on eligibility criteria, study selection, data extraction, methodological reporting, quality appraisal and narrative synthesis are provided in Supplementary Text S1.

The primary dataset used in this review was obtained from Scopus and Web of Science on 12 November 2025. To evaluate potential bias in database coverage, the Scopus search was rerun and an equivalent search was conducted in the Web of Science Core Collection. The Scopus TITLE-ABS-KEY query (Table A1) was converted to the Web of Science Topic (

Table A2. Final Web of Science search query.

Final core corpus

TS=(“fish waste*” OR “finfish waste*” OR “fish by-product*” OR “fish byproduct*” OR “fish offal” OR “fish viscera” OR “fish silage” OR “fish ensilage” OR “ensiled fish” OR “stick water” OR “stickwater” OR “press cake” OR “presscake” OR “surimi wash water” OR “fish processing waste*” OR “fish processing wastewater” OR “fish processing effluent” OR “fishmeal plant effluent” OR “fish cannery wastewater” OR “seafood waste” OR “seafood processing waste*” OR “seafood processing wastewater” OR “seafood processing effluent” OR “aquaculture sludge” OR “fish sludge” OR “aquaculture effluent” OR “aquaculture wastewater” OR “recirculating aquaculture system” OR “RAS sludge” OR “RAS backwash” OR “drum filter backwash” OR “rest raw material”) AND TS=(“anaerobic digest*” OR “methane ferment*” OR biogas OR biomethane OR methanogenesis OR “biogas production” OR “biogas yield” OR “specific biogas production” OR “cumulative biogas production” OR “methane production” OR “methane yield” OR “specific methane production” OR “specific methane yield” OR “ultimate methane yield” OR “cumulative methane production” OR “biochemical methane potential” OR BMP OR codigestion OR “co-digestion” OR “co digestion” OR “anaerobic co-digestion” OR AcoD OR “two-stage anaerobic digestion” OR “two stage anaerobic digestion” OR “two-phase anaerobic digestion” OR “two phase anaerobic digestion” OR “anaerobic membrane bioreactor” OR AnMBR OR “dark ferment*” OR biohydrogen OR “bio-hydrogen” OR “bio h2” OR bioH2 OR “H2 production” OR “H2 yield” OR “CH4 production” OR “CH4 yield” OR “hydrogen production” OR “hydrogen yield” OR “hydrogen production rate” OR HPR OR “hydrogen evolution rate” OR HER OR “specific hydrogen production rate” OR SHPR OR “biochemical hydrogen potential” OR BHP OR “bioCH4” OR “bio-CH4”)

) field and the same restrictions were applied for publication year (2000–2025), language (English) and document type (article). The Web of Science search returned 271 records. A total of 212 records were found in both databases, while 59 were unique to Web of Science. These 59 records were screened using the original eligibility criteria. Of these, 54 were excluded because of an inappropriate material or substrate (n = 14), process (n = 22), outcome (n = 4) or study design (n = 14) and five met the inclusion criteria. These five studies were used only to assess the sensitivity of the review findings and were not added to the primary Scopus-based bibliometric (n = 164) or SLR (n = 120) datasets.

2.2. Bibliometric Analysis

After deduplication and application of the inclusion criteria, 164 unique records were imported into the Bibliometrix environment for the subsequent BA [31]. All procedures were run in R ver. 4.5.2 within RStudio 2025.09.2. The Completeness of Bibliographic Metadata report confirmed that all core descriptive fields such as abstract, affiliation, author, document type, source title, language, publication year, article title and total citations were fully populated. Missing values in the cited references, author keywords and DOI remained below 5%, which corresponds to a good data-quality status in the Bibliometrix taxonomy [31]. The normalized results table was exported as a CSV file and uploaded to the Datawrapper service, where an interactive choropleth visualizes publication counts by authors’ countries of affiliation. To unveil prevailing research themes and terminological linkages in the corpus, a keyword co-occurrence network was generated using VOSviewer ver. 1.6.20 [32].

Annual publication counts were examined using the Pettitt test, a non-parametric, rank-based procedure that detects a single abrupt shift in an ordered time series when the timing of the shift is unknown [42]. The null hypothesis assumes that the distribution of annual publication counts remains unchanged throughout the analyzed period. In contrast, the alternative hypothesis indicates a change at an unknown point. First, the Pettitt test was applied to the full 26-year series from 2000 to 2025. Then, it was reapplied to the pre- and post-change segments to identify additional breakpoints. Statistical significance was assessed at α = 0.05. Since the recursive application involved secondary, data dependent testing of a short, discrete series, the additional breakpoint resulting from this process was interpreted as exploratory. It was used to describe publication phases rather than to infer a causal change in research activity.

2.3. Methodological Limitations

Several limitations should be considered when interpreting the findings. First, the primary bibliometric and qualitative datasets only included English articles indexed in Scopus. Scopus was chosen because it offered the structured and consistent citation, affiliation, keyword, and cited-reference metadata required for the integrated bibliometric analysis. However, database size does not guarantee thematic exhaustiveness, so relevant studies that were only indexed in other databases or published in other languages may have been omitted. The supplementary coverage analysis identified five additional Web of Science only studies that were eligible. Including these studies would have increased the qualitative corpus from 120 to 125 reports (approximately 4%), but it would not have changed the main findings or identified evidence gaps. Nevertheless, the bibliometric maps and descriptive indicators remain specific to the Scopus-indexed, English-language literature and should not be interpreted as representing the entire relevant evidence base. Second, of the 164 reports that passed title-and-abstract screening, 44 could not be retrieved in full text. These reports did not contribute to the qualitative evidence synthesis and may have introduced availability bias, reducing the breadth of the synthesized evidence. Third, raw citation counts favor older publications because they have had more time to accumulate citations. Therefore, citations per year and normalized total citation scores were examined alongside raw citation counts. These indicators mitigate, but do not eliminate, differences arising from publication age, database coverage, document type, and citation practices. Finally, the review was not prospectively registered. Although a retrospective Open Science Framework methodological record was created after the database search, screening, data extraction and analyses were completed, it cannot demonstrate that all methodological decisions were specified in advance.

3. Results

3.1. Performance Analysis and Science Mapping

3.1.1. Publication and Citation Dynamics

A BA was performed to quantify the knowledge base on gaseous biofuels with particular emphasis on the AD processes used to valorize various feedstocks. The resulting corpus is well-defined and rapidly growing. Hence, systematic monitoring of research directions has become essential to prevent individual studies from getting lost in the mass of publications and to map persistent patterns as well as emerging trends over time. After searching, deduplication and screening, the final corpus comprised 164 peer-reviewed research articles published across 85 sources between 2000 and 2025 were retrieved. The average annual growth rate of publications is 13.29%, with a marked acceleration after 2015 (Figure 3). The annual publication output reached 18 articles in 2020, 2023 and 2024, peaking at 20 articles in 2025 as of 12 November 2025. However, the annualized mean citation rate did not increase correspondingly. The metric reached 12.27, 15.06 and 15.44 citations per article per year in 2004, 2008 and 2010, respectively. From 2012 through 2020, it fluctuated between 1.23 and 5.55 citations per article per year and subsequently declined steadily from 4.65 in 2020 to 0.95 in 2025. The early maxima should be interpreted cautiously because only one article was published in 2004 and one in 2010, whereas two articles were published in 2008. Consequently, the annual mean values for these years were highly sensitive to individual highly cited papers. By contrast, between 14 and 20 articles were published annually from 2022 to 2025. Furthermore, the lower citation rates observed for articles published in the most recent years may partly reflect residual citation-window bias. Despite annualization by citable age, newer publications have had less time to gain visibility and accumulate citations. Consequently, the observed pattern should not be interpreted as evidence of declining research quality or relevance.

The Pettitt test identified a statistically significant primary change point between 2011 and 2012 when applied to the 26 annual publication counts recorded between 2000 and 2025. The corresponding Pettitt test statistic was 168 and the approximate two-sided p-value was less than 0.001. Descriptively, mean annual publication output increased from one article per year from 2000–2011 to 10.86 articles per year from 2012–2025. Exploratory reapplication of the test to the 2012–2025 period suggested an additional breakpoint between 2018 and 2019. The Pettitt test statistic was 45 and the approximate two-sided p-value was 0.032. Mean annual output increased from 6.57 articles per year from 2012 to 2018 to 15.14 articles per year from 2019–2025. No additional statistically significant breakpoint was detected within the 2000–2011 segment. The Pettitt test statistic was 29 and the approximate two-sided p-value was 0.135. Based on these findings, three descriptive publication phases were distinguished: 2000–2011, 2012–2018 and 2019–2025. While these breakpoints indicate shifts in the central tendency of annual publication counts, they do not establish causal effects of particular scientific or policy events.

For each phase, a model denoted by $C\left(Y\right)$ was fitted to describe the dynamics of publication, where $Y$ is the within a phase year index taking consecutive integer values from 1 to n. For descriptive purposes, Phase 1 was represented by an exponential model, $C\left(Y\right)=0.277\mathit{exp}\left(0.172Y\right)$. Phase 2 displayed moderate variability without a pronounced trend and was described by a power model, $C\left(Y\right)=7.27Y^{-0.085}$. Phase 3 was represented by an exponential model, $C\left(Y\right)=9.930\mathit{exp}\left(0.100Y\right)$. Bibliometric data show that approximately 65% of all scientific articles were published between 2019 and 2025 (see Figure 3). The marked increase in publication output during Phase 3 indicates growing research interest in the valorization of fish waste through AD.

The increase observed around 2011 temporally coincided with several European and international initiatives concerning waste management, bioeconomy and marine-resource governance. Although the bibliometric data do not establish a causal relationship, these policy developments provide a relevant context for interpreting the growth of the field. Momentum after 2012 coincided with the EU’s framing of the bio- and circular economy as cornerstones of the broader energy- and resource-transition agenda [43,44]. A further policy stimulus arose from the 2013 reform of the Common Fisheries Policy, which revised the framework Regulation 1380/2013 (introducing, inter alia, the landing obligation), updated the market Regulation 1379/2013 and established the European Maritime and Fisheries Fund through Regulation 508/2014 [45,46]. Entering into force on 1 January 2014, these provisions formed part of the policy context associated with the second bibliometric phase (2012–2018) (Figure 3). The notion of a blue economy had already gained wider international endorsement at the Rio+20 Conference in 2012, where it was adopted by numerous coastal states [47]. Legislative momentum strengthened further in 2018 with the adoption of Directive (EU) 2018/851 amending the Waste Framework Directive and its recitals stress that improved waste governance and the reuse of materials enhance environmental protection and energy efficiency. In the same year the EU updated its Bioeconomy Strategy, explicitly prioritizing a sustainable, circular and climate-neutral economic model [48]. Collectively, these initiatives provide the policy backdrop to the post-2018 surge in scholarly output. Ferreira et al. [49] corroborate this trajectory, showing an exponential rise in publications assessing bioeconomic impacts from 2017 onwards. Sultan et al. [38] reported that the mean publication year for the keywords “fish waste(s)” and “anaerobic digestion” was 2014 and 2016, respectively, and these averages reflect the temporal concentration of the literature rather than a peak in popularity.

3.1.2. Country and Institutional Footprint

An analysis of the scholarly output dedicated to the utilization of fish waste and its derivatives shows that authors are affiliated with as many as 51 countries; nevertheless, a pronounced geographical imbalance emerges (Figure 4). Researchers based in Norway contribute the greatest number of papers, trailed by their counterparts in China, Italy, the Republic of Korea and Israel. Supplementary Table S1 provides the complete country-level distribution of publications based on author affiliation data. Within the corpus, 63.6% of records originate from high-income economies, whereas 35.1% come from low- and middle-income economies; affiliations that could not be unambiguously classified represent only 1.3%. The most prolific institutions are Ben-Gurion University of the Negev and Norwegian University of Life Sciences, each with 14 articles, followed by Wageningen University & Research (11), Moi University (8) and the Norwegian Institute of Bioeconomy Research (6). Of the top ten leading aquaculture producers listed by the Food and Agriculture Organization for 2024, only four (China, India, the Republic of Korea and Norway) also appeared among the ten most active countries in fish waste research, giving limited overlap between the two rankings (Jaccard index = 0.25) [50]. A comparison of their respective positions further discloses a weak inverse association (Spearman’s ρ = −0.18). Although modest, this coefficient is sensitive to the small sample size and the analytical assumptions adopted. These results indicate that national aquaculture production alone does not explain the geographical distribution of research activity. Other factors may contribute to this pattern, but their effects were not evaluated in the present analysis.

3.1.3. High-Impact Works and Journals Overview

In a bibliometric survey of 164 articles published in 85 journals, the five most prolific periodicals contributed 48 papers, which represent 29% of the corpus but account for 46% of all citations. The remaining 71 journals published only isolated or scant papers, 61 titles (72%) published a single article and a further 10 published two papers, underscoring the marked fragmentation of the field. Ranking the sources by article number reveals a clear hierarchy; with Bioresource Technology at the top of the list (16 publications; h-index = 12; g-index = 16), followed by Journal of Environmental Management (9 publications; h-index = 7; g-index = 9), Waste and Biomass Valorization (9 publications; h-index = 5; g-index = 8), Aquacultural Engineering (7 publications; h-index = 5; g-index = 7) and Waste Management (7 publications; h-index = 6; g-index = 7). The dominance of Bioresource Technology is conspicuous—its output is nearly twice that of the joint second-placed journals and each of its papers has been cited an average of 73 times, demonstrating its scientific impact. By contrast, Waste and Biomass Valorization, although relatively productive, has a moderate impact presumably due to its younger vintage or lower visibility. Citation-based comparisons between journals should be interpreted cautiously because citation counts depend on publication age, article type and the time available for citations to accumulate.

Table 1 lists the ten most-cited papers worldwide and the five references cited most frequently within the focal set of articles on AD of fish waste. Four of the top ten global papers were published in Bioresource Technology, granting this journal most citations within the elite cohort and cementing its central role in the discourse. At the local level, Standard Methods for the Examination of Water and Wastewater and the biochemical methane potential (BMP) protocol proposed by Angelidaki et al. [51] rank among the most frequently cited references, confirming that rigorous analytical procedures underpin research on fish waste valorization [52,53]. The review by Chen et al. [54], which examines inhibitors of AD, occupies the second position and thus highlights that toxicity and process stability remain significant technological challenges. The only experimental article to feature in both the global and local rankings is Bücker et al. [3], underscoring its dual influence on the broader literature and on the 164-paper corpus analyzed here.

3.1.4. Keyword Co-Occurrence Mapping

Keyword co-occurrence analysis was carried out to delineate the field’s principal research themes and to visualize terminological relationships within the collected literature corpus. VOSviewer (developed in 2010 by Nees Jan van Eck and Ludo Waltman at Leiden University) served as the mapping tool of choice [32]. The analysis relied on author keywords. During preliminary filtering, full counting was applied with a minimum frequency of two, reducing the initial set of 556 unique terms to 79. The network itself was generated using fractional counting so that each keyword had a proportional contribution and records containing very long keyword lists did not dominate the map. Thematic clusters were retrieved using the built-in clustering algorithm (resolution = 1.10 and minimum cluster size = 8), a tool that revealed several well-defined groups while suppressing singletons and other hard-to-interpret microclusters. In comparison, the default resolution of 1.00 with no size threshold generated eight clusters, including single-member ones. Each cluster was then labeled based on its most salient keywords to reflect the underlying research strand. The choice of parameters followed recommendations for transparent bibliometric configurations [20,21,22,24]. The results are presented as (i) a network map depicting total link strength and cluster colors (Figure 5) and (ii) an overlay visualization that displays the mean publication year of each keyword, thereby signaling emerging topics (Figure 6).

Table 1. The ten globally most cited articles in the bibliometric corpus and the five references most frequently cited by articles in that corpus (local citations), 2000–2025. Source: Scopus and the Bibliometrix R package.

Rank	Article	Literature	Journal	Total Citations	Citations per Year	Normalized Total Citations	Open Access Article (Yes/No)
1	Improvement of fruit and vegetable waste anaerobic digestion performance and stability with co-substrates addition	[55]	Journal of Environmental Management	306	18.00	2.76	No
2	Anaerobic batch co-digestion of sisal pulp and fish wastes	[10]	Bioresource Technology	270	12.27	1.00	No
3	A methodology for optimising feed composition for anaerobic co-digestion of agro-industrial wastes	[56]	Bioresource Technology	247	15.44	1.00	No
4	Ensiling of fish industry waste for biogas production: A lab scale evaluation of biochemical methane potential (BMP) and kinetics	[17]	Bioresource Technology	235	18.08	3.25	No
5	A metagenomic study of the microbial communities in four parallel biogas reactors	[57]	Biotechnology for Biofuels (Biotechnology for Biofuels and Bioproducts)	131	10.92	3.76	Yes
6	Synergistic effects of anaerobic co-digestion of whey, manure and fish ensilage	[58]	Bioresource Technology	102	12.75	2.59	No
7	Nutrient mineralization and organic matter reduction performance of RAS-based sludge in sequential UASB-EGSB reactors	[59]	Aquacultural Engineering	96	12.00	2.44	No
8	Fish waste: An efficient alternative to biogas and methane production in an anaerobic mono-digestion system	[3]	Renewable Energy	82	13.67	2.94	No
9	Nitrogen and carbon balance in a novel near-zero water exchange saline recirculating aquaculture system	[60]	Aquaculture	77	8.56	2.57	No
10	Improved utilization of fish waste by anaerobic digestion following omega-3 fatty acids extraction	[61]	Journal of Environmental Management	75	5.36	1.99	No
Top five references most frequently cited by research articles in the corpus, 2000–2025
Rank	Document title	Literature	Document type	Publisher		Local citations
1	Standard Methods for the Examination of Water and Wastewater	[52,53]	Technical standards/reference book	APHA Press, Washington DC		17
2	Inhibition of anaerobic digestion process: A review	[54]	Review article	Bioresource Technology		16
3	Fish waste: An efficient alternative to biogas and methane production in an anaerobic mono-digestion system	[3]	Research article	Renewable Energy		10
4	Defining the biomethane potential (BMP) of solid organic wastes and energy crops: a proposed protocol for batch assays	[51]	Research article	Water Science and Technology		9
5	Co-digestion of waste organic solids: batch studies	[62]	Research article	Bioresource Technology		8

The local citation ranking was derived from the reference lists of the 164 research articles in the bibliometric corpus, therefore the ranked references were not restricted by the document type eligibility criteria applied to the full research studies synthesis and may include books, standards and reviews.

Table 1. The ten globally most cited articles in the bibliometric corpus and the five references most frequently cited by articles in that corpus (local citations), 2000–2025. Source: Scopus and the Bibliometrix R package.

Rank	Article	Literature	Journal	Total Citations	Citations per Year	Normalized Total Citations	Open Access Article (Yes/No)
1	Improvement of fruit and vegetable waste anaerobic digestion performance and stability with co-substrates addition	[55]	Journal of Environmental Management	306	18.00	2.76	No
2	Anaerobic batch co-digestion of sisal pulp and fish wastes	[10]	Bioresource Technology	270	12.27	1.00	No
3	A methodology for optimising feed composition for anaerobic co-digestion of agro-industrial wastes	[56]	Bioresource Technology	247	15.44	1.00	No
4	Ensiling of fish industry waste for biogas production: A lab scale evaluation of biochemical methane potential (BMP) and kinetics	[17]	Bioresource Technology	235	18.08	3.25	No
5	A metagenomic study of the microbial communities in four parallel biogas reactors	[57]	Biotechnology for Biofuels (Biotechnology for Biofuels and Bioproducts)	131	10.92	3.76	Yes
6	Synergistic effects of anaerobic co-digestion of whey, manure and fish ensilage	[58]	Bioresource Technology	102	12.75	2.59	No
7	Nutrient mineralization and organic matter reduction performance of RAS-based sludge in sequential UASB-EGSB reactors	[59]	Aquacultural Engineering	96	12.00	2.44	No
8	Fish waste: An efficient alternative to biogas and methane production in an anaerobic mono-digestion system	[3]	Renewable Energy	82	13.67	2.94	No
9	Nitrogen and carbon balance in a novel near-zero water exchange saline recirculating aquaculture system	[60]	Aquaculture	77	8.56	2.57	No
10	Improved utilization of fish waste by anaerobic digestion following omega-3 fatty acids extraction	[61]	Journal of Environmental Management	75	5.36	1.99	No
Top five references most frequently cited by research articles in the corpus, 2000–2025
Rank	Document title	Literature	Document type	Publisher		Local citations
1	Standard Methods for the Examination of Water and Wastewater	[52,53]	Technical standards/reference book	APHA Press, Washington DC		17
2	Inhibition of anaerobic digestion process: A review	[54]	Review article	Bioresource Technology		16
3	Fish waste: An efficient alternative to biogas and methane production in an anaerobic mono-digestion system	[3]	Research article	Renewable Energy		10
4	Defining the biomethane potential (BMP) of solid organic wastes and energy crops: a proposed protocol for batch assays	[51]	Research article	Water Science and Technology		9
5	Co-digestion of waste organic solids: batch studies	[62]	Research article	Bioresource Technology		8

The local citation ranking was derived from the reference lists of the 164 research articles in the bibliometric corpus, therefore the ranked references were not restricted by the document type eligibility criteria applied to the full research studies synthesis and may include books, standards and reviews.

The resulting network comprised 79 keywords connected by 279 links, with a cumulative total link strength of 424, and was partitioned into five thematic clusters. The red cluster focuses on the AD of aquaculture-derived residues, such as aquaculture sludge, waste and fish offal and links AD to the concepts of the circular economy and sustainability, including resource recovery, nutrient recycling, life cycle assessment and biomethane. The green cluster groups advanced wastewater treatment and process intensification topics relevant to fish farming and seafood processing effluents. These themes include anaerobic membrane bioreactors (AnMBRs), denitrification, nitrogen removal and pilot-scale operation. Notably, the term “dark fermentation” only occurs in this cluster, consistent with its role as the hydrogen producing acidogenic stage in a two-stage system. The blue cluster reflects a biorefinery-oriented valorization pathway in which fish waste is stabilized by ensiling and converted into valuable intermediates, such as volatile fatty acids (VFAs). Nutrient recovery is represented by struvite precipitation under ammonium-related constraints. The yellow cluster includes co-digestion, BMP tests and kinetic, optimization studies targeting methane yield improvement and process stabilization through careful co-substrate selection and modeling. The purple cluster addresses system-level integration in aquaculture, including salinity problems, upflow anaerobic sludge blanket (UASB) reactors and recirculating aquaculture systems (RASs), linked to energy and biogas production. Because the maps are sensitive to underlying keywords and inclusion thresholds, assigning “dark fermentation” only to the green cluster suggests limited thematic connectivity within this dataset. Reporting the term’s frequency and examining closely related synonyms (e.g., biohydrogen) through an overlay or sensitivity analysis can strengthen this interpretation.

Full-text screening identified only three studies that explicitly investigated dark fermentative hydrogen production from fish-derived or related seafood materials. Saidi et al. [63] and An-Stepec et al. [64] used fish-derived substrates, whereas Gorrasi et al. [65] examined commercial raw chitin obtained from crab shells.

3.2. Fish Waste and By-Products

3.2.1. Classification of Fish-Derived Feedstocks, Co-Substrates and Inoculum

In research papers devoted to the AD of fish waste, several clearly defined substrates or fractions recur. The most prevalent are soft slaughter offal viscera, gills, heads, tails, vertebral columns, skin and scales—collectively referred to as offal [10,16,18,56,66,67]. A second group represents processing by-products generated after preliminary treatment, including solid residues after enzymatic oil or protein extraction [61], whole-fish silage from mortalities (acidified or dried) [58,68,69,70] and fish crude oil extraction waste, which is a solid fraction obtained during crude oil separation with centrifuges [3].

The third group includes sludge and solids from RASs operated with freshwater or slightly brackish water, as well as sediments removed from UASB reactors or from hybrid coupled aquaponics AD units [71,72,73,74,75,76,77]. Finally, wastewater from filleting, cooking and frying operations and fecal slurries originating from tank-based grow-out systems constitute the fourth class [74,78,79,80]. Throughout the literature reviewed, researchers consistently use a set of abbreviations to refer to specific types of fish waste. The most frequently used abbreviations are FW (fish waste), FCOW (fish crude oil waste), FPW/FWP (fish processing waste), FS (fish sludge), FPH (fish protein hydrolysate), FWS (fish waste silage), RFV (raw fish viscera), CFV (cooked fish viscera), MFPW (marine fish processing wastewater), SPW (seafood processing wastewater) and AS (aquaculture sludge).

In most studies, anaerobic reactors have been inoculated with anaerobic sludge from full-scale digesters. The most commonly reported inoculum was anaerobic sludge or digestate obtained from full-scale mesophilic digesters at wastewater treatment plants [11,18,67,77]. Another frequently used inoculum source is granular sludge from internal-circulation brewery or winery reactors [11,56,81,82]. Other specialized inoculum mentioned in the literature included sisal wastewater sludge from a Tanzanian fiber mill and the digestate from agricultural biogas plants treating a mixture of cattle manure and food waste [10]. Several laboratory investigations have prepared composite inoculum by mixing food waste sludge with brewery granules or combining digestates from household waste and cattle manure processing to broaden the microbial spectrum and shorten the start-up period [73].

The co-substrates used in the anaerobic co-digestion of fish waste can be grouped into four categories. The first group consists of lignocellulosic agricultural and forestry biomass, including sugarcane bagasse [18], sisal pulp [10], cassava residues [83], steam-exploded willow (Salix) wood chips [84], gorse shrubs [11], Leucaena foliage [85], water hyacinth biomass [86,87,88] and Jerusalem artichoke biomass [61]. The second group consists of starch- or sugar-rich food processing residues, such as waste bread and brewers spent grain [17], orange pomace [12], strawberry pomace [82,89] and fruit and vegetable waste [55,90] and residues from greenhouse cultivation [73]. The third group contains nitrogen- or lipid-rich inputs, including pig manure [56,81], cattle manure [58,68,69], poultry litter, sewage sludge, crude or deoiled biodiesel glycerol [56,89], whey permeate, alkaline fish soapstock [91] and domestic blackwater [92]. The fourth group represents unconventional additives, such as bamboo-derived hydrochar and the liquid fraction generated during bamboo hydrothermal carbonization [93,94].

3.2.2. Biogas and Biomethane Yield from Fish Waste

The literature concerning fish waste valorization indicates a pronounced predominance of laboratory-scale investigations, most of which were carried out in batch mode, then substantially fewer studies addressed bench and pilot scale systems, whereas only two papers reported demonstration scale trials [95,96] and one article described full-scale operation [83]. Most research has focused on co-digestion and the experiments were mainly performed at mesophilic temperatures. The reported biogas and methane yields were within a wide range as summarized in

Table A3. Summary of biogas and methane yields from fish waste feedstocks.

Feedstock	Inoculum Source	Digestion Mode	Test Type and Reactor Configuration	Temperature	Reported Biogas/Methane Outcome (as Reported)	References
Fish waste (offal)	Municipal anaerobic digester sludge	Mono-digestion	Batch (500 mL)	Mesophilic 37 ± 1 °C	433.4 mL biogas g−1 VS (73.34% CH4)	[18]
Fish waste (offal) (75%) with bagasse (25%)	Municipal anaerobic digester sludge	Co-digestion	Batch (500 mL)	Mesophilic 37 ± 1 °C	409.5 mL biogas g−1 VS (78.46% CH4)	[18]
Fish processing waste (offal)	Sisal wastewater sludge	Mono-digestion	Batch (600 mL)	Mesophilic-low 27 ± 1 °C	0.39 m3 CH4 kg−1 VS	[10]
Fish waste (offal) (33%) with sisal pulp (67%)	Sisal wastewater sludge	Co-digestion	Batch (600 mL)	Mesophilic-low 27 ± 1 °C	0.62 m3 CH4 kg−1 VS	[10]
Fish waste (10%) with fruit and vegetable waste (90%)	Mesophilic sludge from an FVW digester	Co-digestion	Batch (ASBR) (2 L)	Mesophilic 35 °C	0.436 L biogas g−1 VS removed	[55]
Fish waste (offal) (5%) with biodiesel waste (11%) and pig manure (84%)	Granular biomass from a pilot hybrid reactor treating wine waste and from an IC reactor treating brewery wastewater	Co-digestion	Batch (500 mL)	Mesophilic 35 °C	320.6 L CH4 kg−1 COD (62.8 L CH4 kg−1 wet wt)	[56]
Fish viscera	Digested sludge from biogas plant	Mono-digestion	Batch (285 mL)	Mesophilic 35 °C	127 ± 20 mL CH4 g−1 VS	[16]
Fish waste (tuna, sardine, needle fish)	Municipal WWTP sludge	Mono-digestion	Batch (in glass vials)	Mesophilic 37 °C	0.47 g COD-CH4 g−1 COD (0.26 L CH4 g−1 VS)	[11]
Fish waste (mackerel)	Municipal WWTP sludge	Mono-digestion	Batch (in glass vials)	Mesophilic 37 °C	0.59 g COD-CH4 g−1 COD (0.35 L CH4 g−1 VS)	[11]
Fish waste (10%) with pig manure (90%)	Granular sludge from brewery IC reactor	Co-digestion	CSTR (9.2 L)	Mesophilic 35 °C	0.43 L biogas L−1 d−1 (57.0% CH4)	[81]
Fish waste (5%) with pig manure (95%)	Granular sludge from brewery IC reactor	Co-digestion	CSTR (9.2 L)	Mesophilic 35 °C	0.59 L biogas L−1 d−1 (59.0% CH4)	[81]
Salmon fish waste heads	Full-scale food waste digester effluent	Mono-digestion	Batch (300 mL)	Mesophilic 37 °C	828 ± 15 m3 CH4 t−1 VS (294 m3 CH4 t−1 wet)	[61]
Fish sludge (after ω-3 and protein extraction)	Full-scale food waste digester effluent	Mono-digestion	Batch (300 mL)	Mesophilic 37 °C	742 ± 17 m3 CH4 t−1 VS (234 m3 CH4 t−1 wet)	[61]
Fish waste with bread waste silage—ensiled mix	Swine manure digester sludge	Mono-digestion	Batch (2.0 L)	Mesophilic 36.5 °C	482 mL CH4 g−1 VS	[17]
Pacific saury and mackerel and cuttlefish waste	Swine manure digester sludge	Mono-digestion	Batch (1.3 L)	Mesophilic 36.5 °C	435–543 mL CH4 g−1 VS	[17]
Fish canning waste (20%) with strawberry extrudate (80%)	Granular brewery sludge, sewage-sludge hydrolytic biomass	Co-digestion	Batch (1 L)	Mesophilic 35 °C	120 mL CH4 g−1 VS	[82]
Fish by-product (20%) with steam exploded Salix (40%) and cow manure (40%)	Farm digester slurry already adapted to fish silage	Co-digestion	CSTR (10 L)	Mesophilic 37 ± 1 °C	191.3 mL CH4 g−1 VS	[84]
Fish waste silage with cow manure	Pilot-plant culture	Co-digestion	S-CSTR (8 L)	Mesophilic 37 °C	0.400 L CH4 g−1 VS per day	[68]
Fish silage with farm slurry	Same farm digestate	Co-digestion	CSTR (320 m3)	Mesophilic 37 °C	742.0 L CH4 kg−1 VS degraded	[95]
Fish waste silage (3–16%) with cow manure	Digestate from a lab CSTR treating the same mix	Co-digestion	CSTR (8 L)	Mesophilic 37 °C	0.20–0.30 L CH4 g−1 VS (the higher at 16% fish waste silage)	[57]
Fish waste (5%) with strawberry extrudate (54%) and crude glycerol (41%)	Mesophilic digester sludge (municipal)	Co-digestion	Batch (1 L)	Mesophilic 35 °C	308 L CH4 kg−1 VS	[89]
Greenlandic halibut offal with blackwater	Digested sewage sludge	Co-digestion	Batch (1 L)	Mesophilic 37 °C	619.7 mL CH4 g−1 VS	[92]
Shrimp offal with blackwater	Digested sewage sludge	Co-digestion	Batch (1 L)	Mesophilic 37 °C	346.4 mL CH4 g−1 VS	[92]
Saline sludge from recirculating aquaculture	Adapted mixed sludge in ASBR	Mono-digestion	Anaerobic sequencing batch reactor (3 L)	Mesophilic 35 ± 1 °C	0.08 g COD L−1 day−1	[101]
Marine fish waste sludge	Halotolerant consortium enriched from RAS sludge	Mono-digestion	Batch (500 mL)	Mesophilic 26–35°C	428 mmol CH4 kg−1 COD day−1	[72]
Mixed municipal, industrial and agricultural wastes including fish	Full scale plant mixed sludge	Co-digestion	CSTR (n.s.)	Mesophilic (n.s.)	3 695.78 Nm3 CH4 d−1 (1035 day)	[83]
Fish viscera round goby intestines	Waste water treatment sewage sludge	Mono-digestion	Batch (100 mL)	Mesophilic 23 and 35 °C	887 L CH4 kg−1 VS at 35 °C and 853 L CH4 kg−1 VS at 23 °C	[98]
Fish waste mixed fraction (heads, skin, bone, intestines)	Waste water treatment sewage sludge	Mono-digestion	Batch (100 mL)	Mesophilic 23 and 35 °C	660 L CH4 kg−1 VS the average of both temperatures	[98]
Fish ensilage	Diluted mixed sludge from food waste and cow manure digester	Mono-digestion	Batch (555 mL)	Mesophilic 37 °C	691 mL CH4 g−1 VS	[58]
Fish ensilage (85%) with cow manure (15%)	Diluted mixed sludge from food waste and cow manure digester	Co-digestion	Batch (555 mL)	Mesophilic 37 °C	729 mL CH4 g−1 VS	[58]
Salmon fish waste silage (13%) with cow manure	Slurry from pilot CSTR treating	Co-digestion	CSTR (8 L)	Mesophilic 37 °C	0.253–0.312 L CH4 g−1 VS	[69]
Fish processing waste	Food and kitchen waste digestate sludge	Mono-digestion	Batch (500 mL)	Mesophilic 37 °C	178 mL biogas g−1 VS and 97 mL CH4 g−1 VS (54.2% CH4)	[94]
Fish processing waste with bamboo hydrochar (1:2)	Food and kitchen waste digestate sludge	Mono-digestion	Batch (500 mL)	Mesophilic 37 °C	292 mL biogas g−1 VS and 219 mL CH4 g−1 VS (74.9% CH4)	[94]
Fish processing waste	Food and kitchen waste digestate sludge	Mono-digestion	Batch (500 mL)	Mesophilic 37 °C	142 mL CH4 g−1 VS (61% CH4)	[93]
Fish processing waste (75%) with liquid fraction of hydrochar (25%)	Food and kitchen waste digestate sludge	Co-digestion	Batch (500 mL)	Mesophilic 37 °C	133 mL CH4 g−1 VS	[93]
Fish waste mixed with vegetable market waste (1:1)	Adapted food waste digester sludge	Co-digestion	Batch (125 mL)	Mesophilic (n.s.)	463 mL CH4 g−1 VSfed	[107]
Fish ensilage with soap stock	Degassed sludge	Co-digestion	Batch (500 mL)	Mesophilic 39 ± 1 °C	775 mL biogas g−1 TS (61% CH4)	[91]
Fish offal (50%) with river tamarind (50%)	Lab anaerobic sludge	Co-digestion	Batch (50 mL)	Mesophilic (n.s.)	330 NmL CH4 g−1 oDM (76% CH4)	[85]
Concentrated effluent from a trout farm after microfiltration and settling	Recirculated mesophilic digestate from the pilot plant	Mono-digestion	CSTR (280 L)	Mesophilic 38 °C	648.8 NL CH4 kg−1 VS	[134]
Fish processing wastewater	Seed sludge from a mesophilic digester and granular sludge from a slaughterhouse UASB reactor	Mono-digestion	Self-agitated anaerobic baffled reactor (SA-ABR) (15.4 L)	Mesophilic 35 ± 1 °C	1.35 L biogas L−1 reactor d−1 (75% CH4)	[78]
Acidogenic fermented fish by-product with rice bran (30%) with sewage sludge (70%)	Lab anaerobic sludge	Co-digestion	Batch (10 L)	Mesophilic 35 ± 1 °C	0.57 m3 CH4 kg−1 VS	[108]
Salmon backbones with cattle manure (5:1)	Cattle manure digestate circulated at the farm	Co-digestion	CSTR (2300 m3)	Mesophilic 38 ± 1 °C	88 Nm3 biogas t−1 (65% CH4) per wet mixture	[96]
Anchovy sludge after limonene oil extraction	Mesophilic agro-industrial digestate	Mono-digestion	Batch (1.1 L)	Mesophilic 35 ± 0.5 °C	0.28 m3 CH4 kg−1 VS	[109]
Squid guts (10%) with mixed dead fish (80%) and soy (10%) and sesame	Sludge from a full-scale digester that treats livestock manure, food waste and food wastewater	Mono-digestion	Batch (1.2 L)	Mesophilic 37 °C	350.5 mL CH4 g−1 VS	[110]
Fish waste slurry (1% vol) with sludge and food waste (5:5)	Mixed sludge digestate	Co-digestion	Batch (80 mL)	Mesophilic 37 °C	0.27 L CH4 g−1 COD added	[118]
Fish waste slurry (5% vol) with sludge and food waste (5:5)	Mixed sludge digestate	Co-digestion	Batch (80 mL)	Mesophilic 37 °C	0.32 L CH4 g−1 COD added	[118]
Fish sludge (catfish RAS)	Granular sludge developing in UASB	Mono-digestion	UASB (1300 L)	Mesophilic 26–35 °C	0.93 m3 biogas kg−1 VS (74.5% CH4)	[74]
Fish silage	Mesophilic digestate from biogas plant	Mono-digestion	Batch (500 mL)	Mesophilic 40 °C	683 mL CH4 g−1 VS	[99]
Thickened RAS fish sludge (3.5% TS)	Dairy manure digester effluent	Mono-digestion	Batch (300 mL)	Mesophilic 35 °C	519 mL CH4 g−1 VS	[102]
Mixed by-products of farmed rainbow trout (offal)	Municipal WWTP sludge	Mono-digestion	CSTR (1 L)	Mesophilic 37 °C	206.68 NmL CH4 g−1	[112]
Saline RAS solids	Full-scale mesophilic digester sludge	Mono-digestion	Batch (1 L)	Mesophilic low 28 °C	0.08–0.25 NL CH4 g−1 VS	[103]
Fish waste (25 vol%) with pig slurry (80%) and orange pomace pulp (20%)	Mesophilic agro-industrial digestate	Co-digestion	Batch (350 mL)	Mesophilic 37 ± 0.5 °C	627.23 mL CH4 g−1 VS added	[12]
Fish waste (50 vol%) with pig slurry (80%) and orange pomace pulp (20%)	Mesophilic agro-industrial digestate	Co-digestion	Batch (350 mL)	Mesophilic 37 ± 0.5 °C	669.68 mL CH4 g−1 VS added	[12]
African catfish RAS sludge	Digested sewage sludge	Mono-digestion	Batch (32.7 L)	Mesophilic 38 °C	229 NL CH4 kg−1 VS	[73]
African catfish RAS sludge with cucumber residues (25%)	Agricultural biogas plant digestate	Co-digestion	CSTR (10 L)	Mesophilic 38 °C	381 NL CH4 kg−1 VS	[73]
Raw tuna viscera (thermal pretreatment)	Acidogenic reactor sludge	Mono-digestion	Batch (50 mL)	Mesophilic 37 °C	0.27 g COD-CH4 g−1 CODadded	[66]
Raw tuna viscera (cooked) (50%) with fat waste (50%)	Acidogenic reactor sludge	Co-digestion	Batch (50 mL)	Mesophilic 37 °C	0.87 g COD-CH4 g−1 CODadded	[66]
Aquaculture wastewater with cow manure	Partially digested cow manure	Co-digestion	CSTR (45 L)	Thermophilic 54 °C	0.45 m3 CH4 kg−1 VS	[97]
Mixed fish waste	Vegetable market waste digester sludge	Mono-digestion	Batch (650 mL)	Mesophilic 37 ± 2 °C	0.289 L biogas g−1 VS	[90]
Mixed fish waste with vegetable market waste (1:3)	Vegetable market waste digester sludge	Co-digestion	Batch (650 mL)	Mesophilic 37 ± 2 °C	0.489 L biogas g−1 VS	[90]
Fish sludge (20 vol%) with residue biomass coming from MCCA production	Full scale biogas plant sludge	Co-digestion	Batch (400 mL)	Thermophilic 50 °C	375–445 L CH4 kg−1 VS	[77]
Fishery processing industrial wastewater	Adapted anaerobic community in AD-BES	Mono-digestion	AD-BES (7.5 L)	Mesophilic 35 °C	200–600 L CH4 kg−1 COD	[100]
Recirculating aquaculture fish sludge (63%) with food (18%), fruit and vegetable waste (19%)	Dedicated mixed food waste digester slurry	Co-digestion	Batch (35 L)	Mesophilic 37 °C	401 mL CH4 g−1 VS	[104]
Recirculating aquaculture fish sludge	Dedicated mixed food waste digester slurry	Mono-digestion	Batch (35 L)	Mesophilic 37 °C	48.94 mL CH4 g−1 VS	[104]
Sardine processing waste with no pretreatment	Municipal MWWTP sludge	Mono-digestion	Batch (1 L)	Mesophilic 37 °C	1174 mL CH4 g−1 VS	[111]
Sardine processing waste with S. cerevisiae pretreatment	Municipal MWWTP sludge	Mono-digestion	Batch (1 L)	Mesophilic 37 °C	821.5 mL CH4 g−1 VS	[111]
Sardine processing waste with Bacillus sp. pretreatment	Municipal MWWTP sludge	Mono-digestion	Batch (1 L)	Mesophilic 37 °C	260 mL CH4 g−1 VS	[111]
Fish waste (gills and viscera)	Cow dung	Mono-digestion	Batch (500 mL)	Mesophilic 37 °C	970 mL biogas g−1 VS including 610 mL CH4 g−1 VS	[88]
Fish waste (gills and viscera) with water hyacinth (1:1)	Cow dung	Co-digestion	Batch (500 mL)	Mesophilic 37 °C	1655 mL biogas g−1 VS including 890 mL CH4 g−1 VS	[88]
Fish waste (offal)	Cow dung	Co-digestion	Anaerobic baffled biodigester (n.s.)	Mesophilic 27 °C	69% CH4 of biogas	[135]
Fish waste (offal) (12%) with primary sludge (88%)	Preincubated anaerobic sludge from municipal PS digesters	Co-digestion	Batch (285 mL)	Mesophilic 35 °C	459 mL CH4 g−1 VS added	[19]
Fish waste	Cow manure slurry	Mono-digestion	Batch (230 L)	Mesophilic low 21 °C	248.5 L biogas kg−1 TS (74% CH4)	[136]
Shrimp pond bottom sludge with Sarcodia residuals (1:1)	Preincubated anaerobic inoculum	Co-digestion	Batch (750 L)	Mesophilic 37 °C	478.06 L CH4 kg−1 VS d−1 (62.9% CH4)	[137]
Fish waste (offal) (33%) with olive mill wastewater (33%) and fruit, vegetable waste (33%)	Cow dung	Co-digestion	Batch (400 mL)	Mesophilic 37 °C	132.2 NmL CH4 g−1 VS	[15]
Fish processing wastewater	Mixed halotolerant consortium and manure sludge	Mono-digestion	SA-ABR (10.4 L)	Mesophilic 35 ± 1 °C	0.39–0.45 L biogas g−1 COD (74% CH4)	[78]
Fish waste (offal)	Acclimated biogas plant sludge	Mono-digestion	Batch (2 L)	Mesophilic 35 °C	540.5 ml CH4 g−1 VS	[3]
Fish crude oil waste	Acclimated biogas plant sludge	Mono-digestion	Batch (2 L)	Mesophilic 35 °C	426.3 ml CH4 g−1 VS	[3]
Fish waste (offal)	UASB sludge from slaughterhouse wastewater	Mono-digestion	Batch (250 mL)	Mesophilic 37 ± 2 °C	464.5 mL CH4 g−1 VS	[4]
Raw tuna waste (30%) with onion stalks (70%)	Full-scale agro-waste digester sludge	Co-digestion	Semi continuous reactor (2 L)	Mesophilic 35 °C	0.24 NL CH4 g−1 VS loaded	[2]

. Reported gas yields were retained on the normalization bases used in the original studies. COD- and VS-normalized yields were not directly compared because conversion between these bases requires study-specific COD/VS data. Regarding thermophilic experimental conditions, Cichy et al. [77] investigated the co-digestion of fish sludge with residual biomass originating from medium-chain carboxylic acid (MCCA) production at 50 °C, achieving specific methane yields of 375–445 L CH₄/kg VS. Similarly, Siddique et al. [97] reported a methane yield of 450 L CH₄/kg VS in a pilot-scale installation operated at 54 °C using a mixture of aquaculture wastewater and cow manure. Of all the groups of aquaculture waste, fish offal remains the most explored material. However, as the scale of the process increases beyond laboratory batch tests, attention shifts to feedstocks that are easier to pump and have a more homogenous composition, specifically fish processing wastewater, acidified fish silage and sludge from RASs.

Regarding the whole fish offal, this substrate displays the highest and most variable yield ranging from 120 to 890 mL CH₄/g VS [3,10,16,17,18,82,94,98]. This range includes both mono-digestion and co-digestion experiments with different proportions of fish-derived material. The lowest value was obtained from co-digestion of fish-canning waste (20%) with strawberry extrudate (80%) [82], whereas the highest yield was recorded by Gruduls et al. [98] in mono-digestion of round goby viscera. Ensilage waste gave methane yields of 200–729 mL CH₄/g VS [17,58,68,91,99]. Methane yields from fish processing wastewater, reported on a COD basis in the original studies, ranged from 290 to 600 mL CH₄/g COD and were not directly compared with VS-normalized yields [78,100]. BMP values for sludge originating from RASs vary widely from 48 to 690 mL CH₄/g VS [73,101,102,103]. For instance, Netshivhumbe et al. [104] obtained only 48 mL CH₄/g VS from mono-digestion of RAS fish sludge, however, co-digestion of this material with food waste (18%) and fruit and vegetable waste (19%) increased methane production to 401 mL CH₄/g VS. Within the subgroup of RAS-derived sludge studies, the highest methane yield was reported by Zhu et al. [74] for mono-digestion of salmon RAS sludge.

3.2.3. Kinetic and Mechanistic Modeling of Gas Production During Anaerobic Digestion of Fish Waste

A subset of the included studies evaluated empirical kinetic models describing cumulative methane production or apparent rate-limiting behavior. To avoid ambiguity between gas yield and substrate concentration, the five formulations are written below using a single notation. The identified models are the modified Gompertz model (1), the first order model (2), the modified logistic model (3), the transference model (4) and the modified Chen and Hashimoto gas yield formulation (5) [105,106].

$$S\left(t\right)=S_0\mathit{exp}\left\{-\mathit{exp}\left[{\displaystyle \frac{R_{max}\mathrm{e}(\lambda -t)}{S_0}}+1\right]\right\}$$

(1)

$$S\left(t\right)=S_0\left[1-\mathrm{e}\mathrm{x}\mathrm{p}\left(-k_{\mathrm{h}}t\right)\right]$$

(2)

$$S\left(t\right)={\displaystyle \frac{S_0}{1+\mathit{exp}\left[\frac{4R_{max}\left(\lambda -t\right)}{S_0}+2\right]}}$$

(3)

$$S\left(t\right)=S_0\left\{1-\mathit{exp}\left[-\left({\displaystyle \frac{R_{max}\left(t-\lambda \right)}{S_0}}\right)\right]\right\},\hspace{1em}\hspace{1em}t\ge \lambda$$

(4)

$$S\left(\theta \right)=S_0\left(1-{\displaystyle \frac{K_{CH}}{{\mu }_m\theta +K_{CH}-1}}\right),\hspace{1em}\hspace{1em}\theta \ge {\displaystyle \frac{1}{{\mu }_m}}$$

(5)

where: $S\left(t\right)$ is the cumulative specific methane yield at a digestion time $t$ days (mL CH₄/g VS_added); $S\left(\theta \right)$ is the specific methane yield predicted by the modified Chen–Hashimoto model at hydraulic retention time $\theta$ (mL CH₄/g VS_added); $S_0$ is the fitted asymptotic (ultimate) specific methane yield (mL CH₄/g VS_added); $R_{max}$ is the maximum specific methane production rate (mL CH₄/g VS_added/d); $e$ is Euler’s number, exp(1) = 2.7183; $\lambda$ is the lag phase (d); $t$ is the AD time in batch tests (d); $k_{\mathrm{h}}$ is the apparent first-order hydrolysis/disintegration rate constant (d⁻¹); $K_{CH}$ is Chen and Hashimoto kinetic constant (dimensionless); ${\mu }_m$ is the maximum specific growth rate of methanogenic microorganisms (d⁻¹); $\theta$ is the hydraulic retention time (d).

The modified Gompertz equation was the most frequently applied cumulative model and frequently produced high coefficients of determination within the individual datasets examined [12,16,17,88,91,107,108,109,110,111]. In the study conducted by Essalhi and Bengueddour [112], both the modified Gompertz and the logistic model provided an equally good fit, whereas in the investigations of Rajendiran et al. [90] and Noh et al. [19] the logistic model showed a slight advantage.

The first-order model, which assumes hydrolysis to be a rate-limiting stage of AD, was described in eight papers [12,17,82,90,91,107,110,112], but it was reported as optimal only in the study of Serrano et al. [82], possibly because no alternative equations were evaluated in that study. Two less common approaches were also explored: the modified Chen–Hashimoto gas yield equation, applied by Kafle et al. [17], and the transference model examined by Essalhi and Bengueddour [112], both of which were outperformed by the modified Gompertz and logistic formulations. Of the limited comparisons available, the modified Chen–Hashimoto and transference models produced less favorable fit statistics than the modified Gompertz and logistic formulations. However, the small number of comparisons does not allow us to draw a general conclusion about their suitability for all fish substrates.

In the study by Gurung et al. [16], the modified Gompertz equation applied to the AD of fish viscera gave a coefficient of determination of 0.95. In that study, the fitted asymptotic yield of the BMP was 163 mL CH₄/g VS and a lag phase of 12.6 days was observed. As reported by Kafle et al. [17], for most substrates, the model’s prediction error remained below 5% for mixtures of mackerel and cuttlefish waste, however, it was greater for the substrates digested alone. In the study conducted by Paone et al. [109], a coefficient of determination of 0.997 was calculated for anchovy waste, which displayed a BMP of 268.7 mL CH₄/g VS and a lag phase of 3.145 d. In turn, Ziagova et al. [111] calculated a coefficient of determination between 0.93 and 0.98 in the experiments with sardine processing waste. Across the reviewed studies, the modified Gompertz model frequently achieved high R² values. However, comparisons between studies should be interpreted cautiously because different datasets, feedstocks, experimental conditions and goodness-of-fit indicators were used. Moreover, R² alone does not evaluate residual structure, parameter identifiability, predictive performance, or the biological plausibility of fitted parameters. Therefore, future model comparisons should also consider complementary validation criteria, such as residual sum of squares, root mean square error (RMSE), Akaike information criterion (AIC), Bayesian information criterion (BIC), residual diagnostics and parameter uncertainty.

The logistic model has become a credible competitor in recent years. Essalhi and Bengueddour [112] obtained R² = 0.9870 for the logistic model and 0.9889 for the Gompertz model, both exceeding the transference and first-order models (each R² = 0.9825) when applied to fish offal mixtures. Rajendiran et al. [90] reported satisfactory fits for all three kinetic equations applied for co-digestion of fish and vegetable wastes, with the highest coefficients of determination exceeding 0.906. The first-order equation used chiefly to characterize hydrolysis typically delivers lower coefficients of determination. In the work of Kafle et al. [17] the R² was found to be from 0.647 to 0.826 and consistently underpredicted biomethane yield. Likewise, Sarker [91] observed relatively low methane yields from fish ensilage, however, Essalhi and Bengueddour [112] reported prediction errors of 6.95% for the first-order model and 6.97% for the transference model, which indicated a limited application of both models for lipid-rich materials.

Unlike the empirical models fitted to cumulative biogas and biomethane production data, Xu et al. [70] employed a mechanistic modeling approach for continuous mesophilic low-solids co-digestion of fish silage and cattle manure at a mass-based ratio of 3:17. The authors used the reduced, mass-based ADM1-R3 (multi-step dynamic model) variant developed by Weinrich and Nelles [113]. In this reduced model, the degradation of particulate carbohydrates, proteins and lipids to acetate is represented by three first-order sum reactions and acetogenesis is assumed to occur instantaneously. Consequently, valerate, butyrate and propionate are not represented as separate intermediates but are included in the overall acetate concentration. Xu et al. [70] further adapted ADM1-R3 to the protein- and sulfur-rich fish silage by modifying the assumed protein composition to account for sulfur and, by incorporating hydrogen sulfide formation during protein hydrolysis, its acid–base dissociation in the liquid phase, liquid–gas mass transfer and dynamic pH control. The modified model was coupled with path optimization to evaluate seven process configurations combining effluent recirculation with different component-separation technologies. The model results indicated that the lower methane yield in the base case was associated with incomplete lipid conversion and acetate accumulation. In contrast, effluent recirculation combined with component separation primarily enhanced the concentration of methanogens and the conversion of proteins and lipids rather than the hydrolysis rates. The simulated methane yield increased from 0.204 kg CH₄/kg VS in the base case to 0.290 kg CH₄/kg VS in the optimized configurations [70]. Although the optimized system was considered scalable, Xu et al. [70] recommended further assessment of thermophilic operation to enhance hydrolysis rates. They also recommended determining the maximum feasible fish silage share under effluent recirculation and periodically reassessing the economic assumptions. More broadly, the scarcity of thermophilic studies identified in this review suggests that the performance and process stability of thermophilic digestion of protein- and lipid-rich fish substrates are not well understood.

3.2.4. Stability and Inhibition Mechanisms in the Anaerobic Digestion of Fish Waste

A review of research from the past 25 years indicates that the AD of fish waste is governed by three dominant inhibitory mechanisms. Owing to its high protein and lipid content, the substrate undergoes proteolysis and lipolysis, which trigger distinct process limitations. First, protein hydrolysis followed by amino-acid deamination generates ammonium ions and free ammonia, whose toxicity can impair methanogenesis. Second, whenever the rate of acidogenesis exceeds the capacity of methanogens to assimilate intermediate products, VFAs accumulate and pH declines. Third, lipolysis yields long-chain fatty acids (LCFAs) that are adsorbed onto the biomass surface, restricting substrate and electron transport and thereby imposing an additional burden on the microbial community.

The hydrolysis of the protein fraction in fish waste releases peptides and amino acids. These are subsequently degraded and deaminated during acidogenesis, which leads to the formation of total ammonia nitrogen (TAN). In an aqueous solution, TAN exists as both ionized ammonium (NH₄⁺) and un-ionized free ammonia (NH₃) [114]. The relative proportions of these forms depend mainly on pH and temperature. Bücker et al. [3] emphasized that, while nitrogen is indispensable for microbial growth, its excess may lead to free ammonia build-up and inhibition of AD. However, the AD remains stable even at elevated potentially toxic NH₄⁺-N levels due to a gradual acclimation of microbiota [84]. Bermúdez-Penabad et al. [66] reported that increasing the fish wastes total solids (TS) content lowered the BMP, coinciding with concurrent accumulation of ammonia and VFAs. Collectively, these studies indicate that high substrate loading rates or low inoculum to substrate ratios foster ammonia accumulation, inhibit methanogenesis and prolong the lag phase.

The readily fermentable fractions of fish waste biopolymers, namely soluble proteins, amino acids and sugars, are rapidly transformed into VFAs during hydrolysis and acidogenesis. When VFA production outpaces methanogenic uptake capacity at high loading rates or short retention times, VFAs accumulate and pH declines. Xu et al. [18] demonstrated that maintaining a sufficiently high inoculum-to-substrate ratio (I/S) markedly mitigates the risk of acidification, while co-digestion with bagasse stabilizes the process more effectively than the mono-digestion of fish waste. Kinetic modeling by Silva and Fragoso [12] further showed that addition of 50% fish waste containing 6% lipids (TS basis) triples the lag phase, because the adsorbed lipids impede lipase activation and mass transfer. In the experiments performed by Bücker et al. [3] protein lipid fish waste exhibited the highest BMP and the residual lipids were gradually β-oxidized to acetate and hydrogen in syntrophic association with methanogens. However, as observed by Tian et al. [115], free ammonia accumulation favors the accumulation of VFAs and hydrogen, which, in turn, suppresses LCFA β-oxidation and ultimately exacerbates overall process inhibition. Excess ammonia increases the partial pressure of hydrogen and the concentration of VFAs, raising the redox potential to a level at which LCFA oxidation and the associated cooperation of syntrophic bacteria with methanogens become thermodynamically unfavorable.

During lipid hydrolysis of fish waste, LCFAs are released and then adsorbed onto the cell surface, which hinders substrate transport and inhibits the activity of syntrophic bacteria and methanogens. Silva and Fragoso [12] demonstrated that the concurrent hydrolysis of proteins and lipids leads to the accumulation of both ammonium ions and LCFAs, markedly inhibiting the digestion process. The adsorbed fatty acids impede mass transfer and promote sludge flotation. Yang et al. [116] summarized evidence indicating that combined LCFA and ammonia stress may suppress β-oxidation and alter the abundance of microorganisms associated with lipid degradation. These findings indicate that free ammonia further increases the LCFA toxicity, preventing full utilization of lipid-rich fish waste.

In addition to substrate composition, AD of fish waste is considerably influenced by operating parameters, including excessive organic loading rate and low inoculum-to-substrate (I/S) ratio. Therefore, maintaining adequate hydraulic retention time (HRT) and gradually increasing the organic loading rate (OLR) helps prevent reactor overload and potential inhibition. The low carbon-to-nitrogen ratio of fish-derived substrates may promote the ammonia accumulation and reduce digestion performance, particularly at high solids concentrations (see

Table 2. Physicochemical characteristics of fish waste reported in anaerobic digestion studies.

Substrate	Process	Total C [%]	Total N [%]	C/N Ratio	TS [%]	VS [%]	Ref.
Guts, digestive tracts, viscera	batch BMP	50.59 b	8.84 b	5.7	25.2 e	88.9 f	[4]
Common carp viscera	batch BMP	52.6 a	8.1 a	6.5	29.1 e	89 f	[3]
Fish head, internal organs, fish fat, bladder	batch	54.2 c	5.06 c	10.72	43.5 e	39.2 e	[94]
Fish gills and viscera	batch	n.a.	n.a.	8.69	25.9 e	20.7 e	[88]
Viscera, gills and scales of bighead fish	batch	n.a.	n.a.	8	38.1 e	93.7 f	[18]
Offal, scales, gills and washing water	batch	51 c	5.85 c	9	32.2 e	55.3 f	[10]
Fish by-product (oil fraction removed)	semi-continuous	53.6 c	9.8 c	5.5	25.6 e	96 f	[84]
Fish wastes	ASBR	48.2 b	5.4 b	8.8	19.6 e	60.2 f	[55]
Aquaculture sludge	batch BMP	52.51 d	7.27 d	8.43 g	n.a.	67.3 f	[14]

(%wt) ^a—percentage by weight; (%TS) ^b—percentage of total solids; (%d.b.) ^c—percentage on a dry basis; (%daf) ^d—percentage on a dry ash-free basis; (%w.b.) ^e—percentage on a wet basis; (%VS/TS) ^f—volatile solids expressed as a percentage of total solids; (molar basis) ^g—carbon-to-nitrogen ratio calculated on a molar basis; n.a.—not available.

) [4]. This imbalance has been addressed mainly through co-digestion with carbon-rich materials. Onion stalks were used to balance deoiled fish waste in a semi-continuous system [2], whereas batch studies evaluated sugarcane bagasse and water hyacinth as complementary plant-based substrates [18,88]. Dairy manure was also combined with aquaculture sludge, with the 10:90 manure-to-sludge mixture providing the highest specific methane production among the tested proportions [14]. These mixtures supplied additional carbon and diluted nitrogen-rich fish material, thereby improving the nutrient balance and process stability [2,18,88]. Nevertheless, successful mono-digestion of fish waste with a low C/N ratio has also been reported, indicating that the lipid content, inoculum characteristics and operating conditions can be equally important [3,14].

3.2.5. Microbial Communities During Anaerobic Digestion of Fish Waste

Several studies have investigated microbial communities of fish-derived materials during AD (

Table 3. Microbial analyses in anaerobic digestion studies involving fish waste.

Substrate	Process and Microbial Sampling	Microbial Analysis	Observation	Ref.
Fish waste and fish crude oil extraction waste from common carp viscera	Batch; 35 °C End of digestion after 14 and 17 days, respectively	16S rRNA gene sequencing for Bacteria and Archaea and ITS1/2 sequencing for Fungi using Illumina MiSeq	Clostridia-dominated fish waste at 67.5%, whereas Gammaproteobacteria-dominated fish crude oil waste at approximately 40%. Ascomycota was the most abundant fungal phylum in both substrates.	[3]
Fish sludge from a commercial tilapia farm	Batch; 35 °C End of the 40-day experiment	16S rRNA gene amplicon sequencing of the V3 and V4 regions for Bacteria and the V4 and V5 regions for Archaea using Illumina MiSeq	Methanobacterium, Methanosarcina and Methanothrix were the dominant archaeal groups. Electron shuttles increased the total abundance of iron-reducing bacteria from 8.2% to 13.4%.	[117]
Fish waste and primary sludge	Batch; 35 °C End of the 53-day digestion	16S rRNA gene amplicon sequencing for Bacteria and Archaea using Illumina iSeq 100	Increasing fish waste shifted the community toward Methanospirillum and Syntrophomonas, which increased from 20.5% to 85.2% and from 0.6% to 7.4%, respectively, whereas Candidatus Cloacamonas decreased from 19.2% to 4.9%.	[19]
Fish silage	Batch; 40 °C Days 1 and 65	16S rRNA gene amplicon sequencing of the V4 region for Bacteria and Archaea using Illumina MiSeq	Fish silage digestion was dominated by hydrogenotrophic Methanobacterium, representing 87.2–94.8% of Archaea. The bacterial community included protein-degrading Lutispora and Proteiniboraceae as well as long-chain-fatty-acid-degrading Cloacimonadota W27.	[99]
Fish waste with primary sludge, secondary sludge and food waste	Semi-continuous; 37 °C During stabilization and 15 spike cycles	16S rRNA gene sequencing of bacterial V4 and archaeal V5 and V6 regions using Illumina iSeq 100 and qPCR for total Archaea	A 1% addition of fish waste had little effect on the bacterial community, whereas repeated 5% additions of this material increased Proteiniphilum, Sedimentibacter and Guggenheimella. Methanospirillum became dominant after the eighth spike, reaching 68% of abundance by the fifteenth spike.	[118]
Fish waste powder	Batch; 37 °C Initial and endpoint samples	16S rRNA gene sequencing of bacterial V4 and archaeal V5 and V6 regions using Illumina iSeq 100 and microbial network analysis	Fish waste degradation increased Proteiniphilum, Aminobacterium, dgA-11 gut group and Syntrophomonas abundance. The seeds from a digester that co-digested livestock manure, food waste and food wastewater retained Methanosaeta as the dominant methanogen by the end of the process.	[110]
Fish waste, wastewater sludge and grass	Batch; 35 °C Days 21 and 28	Sequencing of reverse-transcribed bacterial and archaeal 16S rRNA V4 and V5 regions and untargeted metabolomics using LC-HRMS	Increasing the fish waste proportion reduced bacterial and archaeal diversity. Clostridiales represented more than 90% of the bacterial community during the fish waste mono-digestion.	[119]
Fish waste silage and cow manure	Semi-continuous; 37 °C End of each hydraulic retention period	qPCR targeting SAOB, as well as the Methanosarcinaceae, Methanomicrobiales and Methanosaetaceae families	Increased fish waste loading led to higher levels of ammonium tolerant, SAOB and Methanomicrobiales. The loss of Tepidanaerobacter acetatoxydans at 16% and 19% fish waste coincided with process failure.	[69]

ITS—internal transcribed spacer; qPCR—quantitative polymerase chain reaction; LC-HRMS—liquid chromatography coupled to high-resolution mass spectrometry; SAOB—syntrophic acetate-oxidizing bacteria.

) [3,19,69,72,99,110,117,118,119,120]. High-throughput 16S rRNA gene amplicon sequencing was the predominant analytical approach [3,19,99,110,117,118,119,120]. The DNA-based community profiles were obtained during the mono-digestion of fish waste and fish crude oil extraction waste [3], co-digestion of fish waste with primary sludge [19], batch digestion of fish silage under different magnetic field conditions [99], mono-digestion of fish waste with different microbial inoculum [110], biological treatment of fish sludge in a simultaneous methanogenesis Feammox and denitrification system [117], co-digestion of fish waste with primary sludge, secondary sludge and food waste [118] and AD of aquaculture fecal waste generated by fish farming [120]. The RNA-derived 16S rRNA amplicon sequencing was combined with untargeted metabolomics to investigate metabolically active communities and metabolite degradation in batch digesters fed binary mixtures of sewage sludge with either fish waste or grass [119]. The targeted qPCR was used to quantify the total archaea community during the co-digestion of fish waste with sewage sludge and food waste [118] and selected syntrophic acetate-oxidizing bacteria and methanogenic groups during co-digestion of fish waste silage with cow manure [69]. Additional approaches included fungal ITS1/2 amplicon sequencing [3], microbial network analysis [110,118] and enrichment, isolation and molecular characterization of a biomethane-producing consortium from saline fish sludge [72].

Microbial analysis was performed on samples collected at the end of the digestion period following mono-digestion of fish waste and fish crude oil extraction waste [3], co-digestion of fish waste with primary sludge [19] and treatment of fish sludge in a simultaneous methanogenesis, Feammox and denitrification system [117]. Initial and endpoint samples were compared during fish silage digestion under different magnetic field conditions [99], fish waste digestion with different microbial inoculum [110] and AD of aquaculture fecal waste generated by fish fed different diets [120]. Samples collected at multiple digestion times or operational stages were analyzed in batch digesters fed binary mixtures of sewage sludge with either fish waste or grass [119], semi-continuous digesters subjected to repeated fish waste additions [118] and reactors co-digesting fish waste silage with cow manure [69].

Overall, these studies showed that microbial community composition or activity varied with the type and proportion of fish-derived substrates [3,19,119], inoculum source and acclimation [72,110], fish waste loading [69,118] and the application of magnetic fields or electron shuttles [99,117]. The observed changes were associated with substrate hydrolysis or degradation [99,110,117], syntrophic acetate oxidation [69], methanogenesis [19,72,99,110,117] and process stability under increased or sudden fish waste loading [69,118]. Most evidence was obtained from batch experiments [3,19,99,110,117,119,120], with the remaining research involving laboratory scale semi-continuous reactors [69,118] and one study involving a characterized enrichment culture [72]. Future work was directed toward continuous or full-scale operation [19,72], pretreatment and co-digestion with a wider range of organic wastes [19] and activated-carbon reuse, long-term performance, process optimization, scalability and multi-omics analysis of electron-shuttle-assisted systems [117]. Further mechanistic research was recommended to combine enzyme analysis with magnetic stimulation [99], validate microbe metabolite associations using microbial cultures or labeled substrates [119] and determine the degradability and fermentability of individual carbohydrate types in aquaculture feces [120].

3.2.6. Dark Fermentation of Fish-Derived and Related Seafood Residues Highlighted by Bibliometric Analysis

BA identified dark fermentation as an infrequently represented and weakly connected research topic (Figure 5). Full-text verification confirmed that all three of the identified studies investigated fermentative biohydrogen production experimentally (

Table 4. Substrates, operating conditions and principal findings of dark fermentation studies identified through BA.

Substrate	Inoculum	Pretreatment	Reactor	Main Operating Conditions	Reported Hydrogen Outcome	Principal Finding	Ref.
Commercial crab-shell chitin (5 g/L) with yeast extract (5 g/L)	F210 bacterial consortium (20% v/v) and methanogenic digestate for the second stage	Aerobic fungal prehydrolysis with L. muscarium (2 weeks)	Serum bottles (20 mL) in batch DF followed by AD	DF: 37 °C, 120 rpm, 30 days; AD: up to 83 days	147 mL H2/L	Although sequential recovery from hydrogen to methane was feasible, shorter pretreatment and further optimization of the hydrogen production stage were required. However, the study did not include a control in which chitin was subjected to dark fermentation without fungal prehydrolysis.	[65]
MFVW (100 mL) in seawater without AHFW; C/N 47	T. maritima DSM 3109 (10% v/v)	n.a.	STR (1.1 L) in batch	80 °C, pH 7.0, 150 rpm with NH4Cl and cysteine HCl added	109 mmol H2/L and 3.24 mol H2/mol hexose	MFVW as control sample provided the baseline H2 production used to evaluate the effect of adding AHFW.	[63]
MFVW (300 mL) and AHFW prepared from whole sardines (400 mL) in seawater; C/N 22		Grinding and two-step thermal-acid hydrolysis of sardines		80 °C, pH 7.0, 150 rpm without NH4Cl and cysteine HCl added	285 mmol H2/L and 3.86 mol H2/mol hexose	The highest hydrogen production and maximum productivity were achieved with a C/N ratio of 22 effectively doubling the peak hydrogen productivity.
Dewatered salmon RAS sludge from freshwater, brackish-water and seawater stages (5–50% w/v)	Native sludge microbiota and no external inoculum	Only dewatering	Serum bottles (110 mL) in batch	37 °C, dark, no mixing or pH control, 30 days	24.5 ± 17.5 mL H2/g dry sludge	Untreated freshwater sludge showed the highest hydrogen potential, whereas no detectable hydrogen was produced by seawater sludge and methane was not detected.	[64]

MFVW—model fruit and vegetable waste; AHFW—acid hydrolyzed fish wastes; n.a.—not available.

). However, only two of the studies used fish-derived materials. Saidi et al. [63] examined acid-hydrolyzed sardine material (Sardina pilchardus) and An-Stepec et al. [64] used Atlantic salmon aquaculture sludge. Gorrasi et al. [65], in contrast, investigated commercial chitin obtained from crab shells and should therefore be considered evidence concerning a related seafood-derived material rather than fish waste in the strict sense.

During dark fermentation, acidogenic microorganisms convert organic matter into hydrogen, carbon dioxide and soluble metabolites, particularly VFAs. Hydrogen must be recovered before methanogenesis occurs, or when methanogenic activity is suppressed, because otherwise it can be consumed during methane formation. Of the three studies, only Gorrasi et al. [65] investigated the sequential production of hydrogen and methane. Saidi et al. [63] focused on biohydrogen production by a defined hyperthermophilic culture. An-Stepec et al. [64] examined hydrogen and fermentation metabolites produced by the native microbial community of fish sludge.

All three studies evaluated hydrogen production under laboratory-scale batch conditions. However, they differed substantially in substrate preparation, reactor configuration and the microbial systems employed. Taken together, these studies demonstrate the feasibility of producing hydrogen through fermentation from fish- and seafood-derived materials. However, differences in feedstock composition and experimental design limit direct comparisons of process performance. Therefore, further laboratory-scale batch studies are still required to systematically address feedstock variability and process kinetics, while continuous or semi-continuous operations and scale-up remain to be demonstrated.

3.3. Economic Feasibility and Life Cycle Assessment

A techno-economic assessment (TEA) integrates technical process performance with economic parameters to evaluate the feasibility of implementing a technology [121]. In turn, life cycle assessment (LCA) provides a standardized framework for quantifying the environmental performance of a process or system across defined life cycle stages. These stages include goal and scope definition, inventory analysis, impact assessment and interpretation [121]. Several studies have investigated the techno-economic and environmental aspects of fish waste valorization through AD.

Substantial techno-economic or economic feasibility modeling was performed for alternative fish waste biorefinery pathways [5] and salmon processing co-streams [96], while more limited quantitative economic assessments concerned artisanal fish waste [4], modeled fish silage digestion [70], fish waste valorization within a cold supply chain [122] and RAS fish sludge management [123]. Although a complete discounted cash flow techno-economic analysis was not performed, experimental BMP data for artisanal fish waste were combined with community-scale energy and economic projections [4]. Fish silage digestion was evaluated entirely through modeling using a modified ADM1-R3 reduced model, process simulation and optimization to compare the effluent recirculation and separation configurations and their profitability [70]. Alternative fish waste biorefinery pathways were modeled in Aspen Plus and compared using economic indicators, including net present value and sensitivity analysis [5]. Four salmon co-stream processing concepts were modeled by combining capital and operating cost estimates, return on investment and Monte Carlo analysis to compare direct biogas production with the recovery of fish oil and proteins followed by AD of the residual fraction [96]. Experimental data on methane production from aquaculture solids were additionally extrapolated to potential heat and electricity recovery at different farm scales, although a complete economic assessment was not performed [103]. From an economic perspective, the modeled scenario combining AD with combined heat and power using fish waste and pig manure was estimated to achieve an internal rate of return of 14.38% within the Latvian cold supply chain [122], while a modeled biorefinery co-fermenting fish waste with cow dung had a specific capital investment cost of €659 per ton of treated substrate [124]. Importantly, the pathway maximizing energy recovery was not necessarily the most economically attractive, as direct biogas production was energetically favorable, whereas prior recovery of fish oil and proteins resulted in the highest profitability and return on investment [96].

Formal LCA was applied to fish waste valorization within a Latvian cold supply chain [122], fish sludge management in a modeled Norwegian RAS [123], trout production including AD of fish sludge and dead fish [125], anchovy processing residues [126] and anaerobic co-digestion systems in which marine fish processing waste was used as a co-substrate [127]. Environmental LCA and life cycle costing were jointly applied to a modeled fish waste anaerobic digestion scenario within the Latvian cold supply chain [122]. Fish sludge treatment scenarios involving transport, mono-digestion, co-digestion and nutrient recovery were compared using formal LCA together with a preliminary economic assessment for a modeled reference Norwegian RAS [123]. A consequential LCA of trout production incorporated AD of fish sludge and dead fish and accounted for uncertainty through Monte Carlo simulation [125]. Prospective LCA was used to extrapolate the LimoFish valorization of anchovy filleting residues from laboratory to industrial scale, including AD of the residual solid fraction [126]. An attributional LCA evaluated marine fish processing waste as a co-substrate for AD of sisal waste within a broader Tanzanian sisal production system [127]. For the Industry Average Site scenario, co-digestion of sisal waste with marine fish processing waste shifted the climate change balance from 41,049 to −4,458 kg CO₂-eq per tonne of export fiber when the fish processing waste was assumed to have no current beneficial reuse [127]. Direct comparison of these studies is limited by differences in functional units, system boundaries, feedstock definitions, process scale, geographical energy systems and treatment of co-products [122,123,125,126,127].

Overall, integrated environmental and economic assessments applied to the same fish waste AD system remain scarce, with combined LCA and life cycle costing reported for a fish cold-supply-chain scenario [122] and formal LCA accompanied by only a preliminary economic assessment for RAS fish sludge management [123]. Most assessments focused on mixed fish processing co-streams, aquaculture sludge and fish silage. Evaluations of individual fish waste fractions, such as viscera or heads, were rare, as were assessments of pilot- and industrial-scale operations that included pretreatment or prior recovery of valuable compounds, as well as integrated biorefinery pathways. The economic and environmental outcomes were strongly influenced by process configuration, scale, energy system assumptions, market conditions and co-product utilization across the reviewed studies [5,70,96,122,123,124,126,127]. Therefore, future assessments should prioritize integrated TEA with LCA frameworks that are validated with primary data from pilot- and industrial-scale systems and that are supported by transparent sensitivity analyses.

3.4. Database Coverage Sensitivity Analysis

A supplementary database comparison identified 59 unique Web of Science records. Five of these records met the eligibility criteria: Martinez et al. [128], Mannacharaju et al. [129], Essalhi et al. [130], da Borso et al. [131] and Dey and Thomson [132]. These five studies were not incorporated into the primary Scopus-based bibliometric (n = 164) or qualitative (n = 120) datasets but were considered as a sensitivity set. Of the studies included in the primary review, 131 bibliometric records (80%) and 102 qualitative reports (85%) were also indexed in Web of Science. Including the five eligible Web of Science only studies would expand the qualitative corpus from 120 to 125 reports (approximately 4%). Although these studies provided additional evidence on the anaerobic digestion of fish-derived waste, they did not contradict the main findings or alter the identified research gaps. The qualitative conclusions were therefore considered robust to database choice; however, the bibliometric maps and performance indicators remain specific to the Scopus corpus.

4. Future Research Directions

Based on the analysis of the available literature, several directions for future research can be identified. Future studies should determine synergistic combinations of fish waste fractions and characterize individual fractions in a manner that accurately reflects the composition and variability of industrial waste streams [11,19].

The effects of physical, chemical and biological pretreatment methods, particularly enzymatic pretreatment, should be investigated using both raw fish waste and residual streams generated during biorefinery processing. In addition to biomethane yields, these studies should consider energy and material balances, chemical consumption, process water requirements and the recovery of value-added compounds when assessing technical and economic feasibility [2,19,66,96].

Different combinations of fish waste fractions and suitable co-substrates should be evaluated in semi-continuous and continuous digestion systems to reduce process inhibition, enable higher organic loading rates and maximize methane production. Particular attention should be paid to long-term process stability and the accumulation of ammonia, VFAs and LCFAs, which may not be adequately represented by short-term batch assays [10,11,19,56,90,92].

The production of hydrogen and methane should be investigated both in separate single-stage processes and in integrated two-stage systems combining dark fermentation with AD for fish waste. In such systems, dark fermentation could be used to recover hydrogen and generate VFAs, whereas the remaining organic matter and fermentation metabolites could subsequently be converted into methane. Dark fermentation of fish waste appears to remain comparatively underexplored within the literature included in the bibliometric dataset, indicating a potentially important research gap in the sequential recovery of hydrogen and methane.

Future studies should combine time-resolved microbial community profiling with activity based and multi-omics analyses to identify the functional groups and metabolic pathways involved in protein and lipid degradation during fish waste digestion. Integrating these data with process measurements would clarify how the increasing fish waste loads and the accumulation of ammonia, LCFAs and sulfide affect microbial activity and bioreactor stability [69,99,117,118,119,120].

Because most previous studies relied on empirical equations fitted to cumulative biogas or biomethane production data, future modeling should move towards multi-step dynamic approaches, including ADM1-type frameworks capable of representing protein and lipid degradation, ammonia inhibition, acid base equilibria and changes in microbial populations [91,133].

Further research should evaluate digestate nutrient content, agronomic performance and environmental safety, including salinity and potential contaminants, before its agricultural use or application in land reclamation [2,85,90,92,96].

A stepwise biorefinery approach should also be developed in which high-value components [37] are recovered before the remaining organic stream is subjected to AD. Research should determine whether the extraction of these compounds improves the overall economic and environmental performance of fish waste valorization without substantially reducing the energy potential of the residual stream.

Taken together, these research directions could support a transition from reliance on short-term BMP assays to integrated fish waste valorization systems grounded in mechanistic understanding, environmental assessment and validation under industrially relevant conditions.

5. Conclusions

This bibliometric and systematic review mapped the development of research on gaseous biofuel production from fish waste and synthesized experimental evidence published from 2000 to 2025. The analysis revealed a rapidly expanding, yet fragmented, field. The bibliometric corpus included 164 articles and 120 full research papers were used for qualitative synthesis. The annual publication growth rate was 13.29%, with approximately 65% of articles published between 2019 and 2025. However, research activity was concentrated on high-income economies and showed limited overlap with leading aquaculture-producing countries. Additionally, it was dispersed across 85 journals. These patterns suggest that the increase in scientific interest has not yet resulted in a geographically balanced or consolidated evidence base.

The reviewed studies covered four broad groups of feedstocks: (i) fish offal, (ii) processed residues and silages, (iii) solids and sludge from recirculating aquaculture systems and (iv) wastewater or fecal slurries. Most of the evidence came from laboratory-scale, batch assays conducted under mesophilic conditions. These assays often used co-digestion. However, thermophilic, continuous, demonstration-scale and full-scale studies were scarce. The reported methane yields varied substantially due to differences in feedstock composition and preparation, proportion of fish waste, selection of co-substrates, inoculum, operating conditions and normalization basis. These differences limited direct comparisons among studies. Co-digestion was widely used to improve the nutrient balance and dilute inhibitory compounds; however, its benefits were mixture specific. Successful mono-digestion was also reported in several studies. The process instability was primarily associated with the effects of ammonia, VFAs and LCFAs. The accumulation of these compounds depended not only on the protein and lipid content but also the inoculum-to-substrate ratio, the organic loading rate, the hydraulic retention time and microbial adaptation.

The modified Gompertz equation was the most frequently applied empirical model. However, the logistic model provided comparable or occasionally better fits. Nevertheless, model comparisons relied mainly on coefficients of determination and rarely considered residual diagnostics, parameter uncertainty, predictive performance or biological plausibility. Only one study applied an adapted ADM1-R3 framework, revealing an obvious discrepancy between empirical curve fitting and multi-step dynamic modeling. Microbial studies, primarily batch-based 16S rRNA gene surveys, linked community responses to feedstock composition, inoculum source, fish waste loading and process interventions. These studies also identified taxa that are potentially involved in protein and lipid degradation, syntrophic acetate oxidation and methanogenesis. Nevertheless, time-resolved and activity-based evidence from continuous systems was limited. Dark fermentation was the least developed pathway. Only three experimental studies were identified of those only two used fish-derived substrates. No fully integrated, two-stage hydrogen–methane process using strictly fish-derived feedstock was identified in the reviewed corpus.

Overall, the available evidence suggests that AD is a promising method for recovering renewable energy from fish-derived residues. However, it does not yet provide a sufficient basis for widespread industrial implementation. Integrated TEA and LCA studies were also scarce and relied mainly on modeled scenarios rather than primary data from pilot- or industrial-scale systems. These conclusions apply to English-language, Scopus indexed literature. Progress toward industrial implementation requires feedstock characterization specific to each fraction, validation of selected mixtures and pretreatments in semi-continuous and continuous reactors and assessment of thermophilic operation and long-term inhibition. Integrating multi-step dynamic modeling with time-resolved microbial and multi-omics analyses, energy and material balances, digestate safety assessments and the recovery of higher-value compounds would facilitate the transition from short-term BMP assays to integrated, industrially relevant fish waste valorization systems.

The database coverage sensitivity analysis identified five additional Web of Science studies that are eligible. Including these studies would increase the size of the qualitative corpus by approximately 4% but would not significantly alter the main interpretations or evidence gaps. However, the bibliometric patterns should be interpreted as specific to the Scopus corpus.