b i o s y s t e m s e n g i n e e r i n g 2 2 7 ( 2 0 2 3 ) 1 9 5 e2 1 6

Available online at [Link]

ScienceDirect

journal homepage: [Link]/locate/issn/15375110

Research Paper

Precision feeding system for largemouth bass

(Micropterus salmoides) based on multi-factor
comprehensive control

Xiaojuan Liu a, Kaihua Du a, Chuyi Zhang a, Yi Luo a, Zongyao Sha b,

Chunfang Wang a,*
a
College of Fisheries, Huazhong Agricultural University, Engineering Research Centre of Green Development for
Conventional Aquatic Biological Industry in the Yangtze River Economic Belt, Ministry of Education, Hubei Provincial
Engineering Laboratory for Pond Aquaculture, Wuhan 430070, China
b
School of Remote Sensing and Information Engineering, Wuhan University, 430079, China

article info
In order to solve a series of problems in the current feeding process, such as inaccuracy,
Article history: unevenness and high labour costs, a precision feeding system for largemouth bass was
Received 8 October 2022 developed by integrating feeding management model, water quality monitoring system,
Received in revised form fish feeding activity sensor, automatic feeding machine and software platform. Based on
6 February 2023 the analysis of the growth data of largemouth bass and the bioenergetics principle, the
Accepted 14 February 2023 feeding management model for predicting the growth, feed requirements and waste
Published online 27 February 2023 output of largemouth bass was constructed and validated under different culture modes.
The real-time DO parameters were used to fine-tune the feed requirements predicted by
Keywords: the model. The feed requirements were then further precisely adjusted by combining the
Largemouth bass (Micropterus intensity of fish feeding activities. The web platform client used supported the access of
salmoides) any browser and provided a visual interface for software users. The experimental results
model showed that the feeding management model provides reliable estimates of growth, feed
precision feeding system requirements and waste output for largemouth bass. The function of dynamically
multi-factor comprehensive control adjusting the model and feeding parameters through feedback and production data pro-
vided by cloud serves or users improved the application effectiveness and scope of the
system. The system comprehensively considered multiple factors such as fish biological
characteristics, physiological stages, feed nutrients composition, environmental factors
and fish appetite, and incorporated relevant technologies or equipment to provide an
effective tool for users to shift from experience-driven to knowledge-driven approaches to
better optimise production and feeding, thereby improving the economic benefits and
environmental sustainability of aquaculture.
© 2023 IAgrE. Published by Elsevier Ltd. All rights reserved.

* Corresponding author.
E-mail address: cfwang@[Link] (C. Wang).
[Link]
1537-5110/© 2023 IAgrE. Published by Elsevier Ltd. All rights reserved.
196 b i o s y s t e m s e n g i n e e r i n g 2 2 7 ( 2 0 2 3 ) 1 9 5 e2 1 6

Nomenclature UN Urinary nitrogen loss

ZN Branchial nitrogen loss
DO Dissolved oxygen
DN Digestible nitrogen
W0 initial body mass
RN Retained nitrogen
Wn Final body mass
CP Crude protein
Wd Body mass of dead fish
CL Crude lipid
T Water temperature
CHO Carbohydrate
BM Body mass
ADC Apparent digestibility coefficient
DEreq Digestible energy requirement
TSW Total solid waste
DE Digestible energy
SWN Solid nitrogen wastes
RE Retained energy
SWP Solid phosphorus wastes
HeE Basal metabolism energy
DWN Dissolved nitrogen wastes
UE þ ZE Urinary and branchial energy
DWP Dissolved phosphorus wastes
HiE Heat increment of feeding
DM Dry matter
GE Gross energy
RAS Recirculating aquaculture system
BP Body protein
JAS Juanyang aquaculture system
BL Body lipid
LRAS Land-based recirculating aquaculture systems
BA Body ash
TGC Thermal-unit growth coefficient
BPh Body phosphorus
FCR Feed conversion ratio
ME Metabolic energy
FI Feed intake
FE Faecal energy
RSS Residual sum of squares
IE Intake of energy
PFR Predicted feeding requirement
N Nitrogen
OFI Observed feed intake
P Phosphorus

251e258) and rainbow trout (Oncorhynchus mykiss) (Lagardere

1. Introduction et al., 2004). However, hydrophones and cameras have the
disadvantages due to environmental noises and water
Fishery and aquaculture products are important sources of turbidity leading to inaccurate feeding (Zhou et al., 2017). Un-
protein, providing food and income for millions of people derfeeding may lead to slower growth and lower economic
worldwide. Both a greater demand for fish and declining returns, whilst overfeeding results in feed waste, economic
capture fisheries has led to the dramatically increased global losses, increased waste output, reduced DO and endangered
aquaculture production over the past 30 years. Total aqua- fish health (Abidi & Khan, 2014). Since the appropriate feeding
culture fish production has reached a record of 82 million strategies or methods are crucial to increase production,
tonnes in 2018 (FAO, 2021), and by 2030, nearly two-thirds of improving animal welfare and reducing environmental pollu-
global fish consumption is expected to come from aquaculture tion (Føre et al., 2011; Watanabe, 2002; Wu et al., 2015), there is
(FAO, 2016; Kobayashi et al., 2015). Given the fact that feed is an urgent need to develop a commercially available, reliable
the highest cost item in aquaculture, accounting for more and robust feeding control methods or systems to help fish
than 50% of the total operating cost (Davis & Hardy, 2022), feed farmers and managers achieve precise feeding operations.
feeding has become the main factor determining the profit- In recent years, many studies have proposed various
ability of aquaculture. feeding control methods and techniques to achieve the pre-
Fish feed requirements are affected by numerous factors, cision feeding of aquatic animals. Mathematical modelling is
such as physiology, nutrition, environment, biology and considered a powerful approach to describe and solve prac-
feeding methods (Sun et al., 2016). Due to the complexity of the tical problems and also an effective way to address the chal-
culture environment and diversity of culture modes, it is more lenges faced by aquaculture (Buterbaugh & Willoughby, 1967;
difficult to determine the feeding demand of fish than that of Cho, 1992; Duman et al., 2007; Ren et al., 2020). Several models
livestock. Although changes of fish feeding status can be have been proposed to estimate the growth, feed re-
observed intuitively, feeding methods are often highly quirements and waste output of aquatic animals (Iwama and
dependent on farmer's subjective experience, and cannot be Tautz, 1981; Liu et al., 2018; Zhou et al., 2005). Amongst them,
easily quantified using unified or scientific standards. This bioenergetics-based models offer a simple way to describe or
often results in high labour cost and unreasonable feeding. In predict the partition of dietary energy-yielding nutrients (Cho
recent years, some new feeding appetite sensors including & Bureau, 1998). Bioenergetic models are very useful and
hydrophones, and cameras (acoustic cameras, underwater practical, as they allow prediction of fish feed requirement
cameras, and land-based cameras) have been used or pro- based on expected growth performance, dietary feed nutri-
posed for shrimp farming (Reis et al., 2022; Smith and Tabrett, tional composition or digestible energy (DE) content, and ex-
2013) and certain fish species, such as Atlantic Salmon (Salmo pected body energy deposition (Davis & Hardy, 2022). Dietary
salar) (Liu et al., 2014), tilapia (Oreochromis mossambicus) (Zhou t DE content is considered to be a major factor controlling fish
al., 2018), turbot (Scophthalmus maximus) (Lagarder, 2000, pp. feed intake. When given diets with balanced nutrition and
 b i o s y s t e m s e n g i n e e r i n g 2 2 7 ( 2 0 2 3 ) 1 9 5 e2 1 6 197

different digestible energy levels, fish will adjust their overall technologies are vital for developing intelligent feeding con-
feed intake to maintain the required level of energy intake and trol systems with efficient monitoring and feedback tools.
ensure growth potential (Houlihan et al., 2008; Yamamoto Although the above devices (or sensors) can provide simpler
et al., 2000). However, early bioenergetic models were not and more versatile methods for determining or adjusting the
sufficiently flexible to be applied to the various conditions feed requirements of fish, these methods completely ignore
encountered in aquaculture (Bureau & Hua, 2008). More the biological characteristics of the fish, which can result in
recently, great progress has been made in the study of bio- providing feed amounts that are not optimal to meet the
energetic models, and models have been developed for gilt- growth potential and nutritional requirements of fish. As all
head seabream (Sparus aurata) (Libralato and Solidoro, 2008), methods have advantages and disadvantages relative to the
rainbow trout (O. mykiss) (Bureau et al., 2003), tilapia (O. mos- specific technology used (Føre et al., 2018) but applications
sambicus) (Chowdhury et al., 2013) and Gibel carp (Carassius that allow mathematical models to be effectively combined
auratus gibelio) (Liu et al., 2018). Using bioenergetic principles, with real-time data from multiple sensors or technologies will
the growth, energy consumption, nitrogen and phosphorus potentially provide better feed demand estimate than those
output of fish fed with different feed composition at the obtained using sensors or models alone.
different growth stages can be predicted by factorial models The aim of this paper is to report on the development of a
(Chowdhury et al., 2013; Liu et al., 2018). Bioenergetic models precision feeding system for largemouth bass (Micropterus
can accurately describe the energy budget and its relationship salmoides) based on integrated multi-factor comprehensive
with the main influencing factors, such as temperature and control, which can effectively combine multiple methods
body mass, and determine the feeding amount according to such as nutritional model, water quality monitoring system
the nutritional requirements of fish and dietary nutritional and fish feeding activity sensor through step-by-step regula-
composition, thus increasing the nutritional utilisation effi- tion. The built-in nutritional model can predict fish growth,
ciency and improving the growth and health of fish. feed requirements and waste production, and the water
However, in practice, the culture modes of aquatic animals quality monitoring and fish feeding activity sensors can be
and the factors affecting fish feeding are complex and diverse. used to adjust fish feeding amount, feeding speed and feeding
Relying on only a set of established models with fixed pa- duration in real time to provide more accurate estimation of
rameters to predict growth performance and feed require- feeding parameters, thus achieving correct and automatic
ment across all culture modes may not achieve the expected feeding of fish under “limited” human supervision, providing
predictions, which has led to models being limited to a rela- users with an effective tool to better optimise production and
tively narrow range (Davis & Hardy, 2022). Since all current feeding, improve economic benefits and environmental sus-
models can only predict feeding requirements for a day or for tainability. The system consists of the following modules: (1) a
a period, and the appropriate feeding frequency for fish is bioenergy-based nutritional model to predict fish growth, feed
usually 2e3 times/day or more (Mabroke et al., 2021; Wu et al., requirements and waste output; (2) a water quality monitoring
2021), the determination of the optimal feeding amount per system for real-time monitoring of water temperature and DO
time is a key problem that the model cannot solve. In addition, parameters and transmitting them to the cloud server; (3) a
existing models required manual input of equations and the fish feeding activity sensor that can get the fish feeding in-
execution of calculations based on these equation (Zhou et al., tensity based on the water flow under the current feeding
2005). The calculation steps, analysis process, and results state of the fish group; (4) an automatic feeding machine can
presented are complex and cumbersome, making it difficult to automatically adjust the feeding amount, feeding speed and
scale up and utilise on a large scale. Therefore, allowing the feeding duration according to the instructions sent by the
model to adjust parameters based on actual production data cloud server, and upload the feeding information to a cloud
and developing an online serve platform to implement the server after feeding; (5) the cloud server can receive and store
automatic prediction of the model are necessary steps to- the data obtained from all devices and platform clients,
wards expanding applications of the model. analyse and process these data through embedded models,
Fish feeding and appetite are also influenced by various and send feeding instructions to the feeding machine; (6) a
environmental factors, among which water temperature and client application is developed to connect the end users and
DO levels are the most important factors determining feed the cloud server, which visualises the remotely sensed data
consumption at each feeding (Buentello et al., 2000). In recent (temperature, DO, and fish feeding activities), analytical re-
years, much research has been conducted on the monitoring sults from the nutrition models, and running status of feeding
of environmental parameters in aquaculture, and several machines.
water quality monitoring systems for aquaculture ponds
based on Internet of Things (IoT) technology have been
developed (Huan et al., 2020; Schmidt et al., 2018). These sys- 2. Materials and methods
tems enable remote real-time collection and data storage of
information from multi-sensor processors such as water 2.1. Model construction
temperature and DO, and some of them have been used to
adjust the feeding rate (Wu et al., 2015). Furthermore, in order 2.1.1. Growth model
to observe the fish feeding behaviour more intuitively or A database of largemouth bass was created using production
directly in real time, some sensors and technologies, also have records from 8 separate fish lots for fish raised in a recircu-
been developed to determine feed demand by analysing fish lating aquaculture system at Huazhong Agricultural Univer-
feeding behaviour (Zhou et al., 2017). These devices or sity (Wuhan, China) (Table 1) and 77 studies published
198 b i o s y s t e m s e n g i n e e r i n g 2 2 7 ( 2 0 2 3 ) 1 9 5 e2 1 6

Table 1 e Water temperature ( C), interval time (days), initial body mass (g), final body mass (g), total feeding amount (g) and
survival rate (%) of largemouth bass in different tanks.
Tank Time (d) Temperature ( C) Initial body Final body Feeding amount Survival rate (%)
mass (g) mass (g) (g fish1)
1 209 25 ±2 2.40 ± 0.28 298.42 ± 27.46 309.95 95.00
2 209 25 ±2 2.56 ± 0.31 290.17 ± 30.22 305.56 90.00
3 209 25 ±2 2.35 ± 0.32 293.39 ± 22.34 308.95 100.00
4 209 25 ±2 1.82 ± 0.22 309.42 ± 28.32 318.40 89.29
5 209 25 ±2 1.60 ± 0.15 255.26 ± 19.23 268.35 92.00
6 209 25 ±2 2.63 ± 0.30 291.13 ± 24.13 308.42 95.00
7 153 25 ±2 68.11 ± 8.72 435.42 ± 37.46 448.50 93.33
8 153 25 ±2 60.11 ± 6.75 420.40 ± 38.11 439.50 93.33

between year 2012 and 2022 (Table 2). The growth trajectory of observations from 9 studies published between 2012 and 2022
largemouth bass was described using the thermal-unit growth (Table 2). The energy of HiE (kJ fish1) (Eq. (6)) and UE þ ZE (kJ
coefficient (TGC) (Eq. (1)), where 1-b of the TGC model was fish1) (Eq. (7)) were assessed using the equations reported by
calibrated by minimising the residual sum of squares (RSS) Kaushik (1998).
(Vittinghoff et al., 2005, p. 340) (Eq. (3))
DEreq ¼ RE þ HeE þ HiE þ ðUE þ ZEÞ (4)
Wn ð1bÞ  W0 ð1bÞ
TGC ¼ 100  Xn (1)  
Ti HeE ¼ aT2 þ bT þ c  Wd (5)
i¼1

 P  1 HiE ¼ ME þ HeE (6)

TGC ni¼1 Ti ð1bÞ
Wn ¼ W0 ð1bÞ þ (2)
100
UE þ ZE ¼ 24:9  ðUN þ ZNÞ (7)
X 2 2
RSS ¼ yi  Yi (3) Where, (aT þ bT þ c) is the relationship between temperature
i (T) and HeE, d is the body mass exponent. ME is Metabolic
Where, n (¼ 1, 2, …) is the day number recorded from W ¼ W0, energy, UN and ZN are urinary and branchial nitrogen loss,
W0 and Wn are initial body mass (g) and final body mass (g), respectively. Since UN þ ZN can be calculated by (DN - RN)
respectively; T is average daily temperature ( C). Yi and yi are (Kaushik, 1998), the non-faecal energy losses (UE þ ZE, kJ
observed body mass and predicted body mass by Eq. (2). fish1) can be re-arrangement as: 24.9  (DN - RN), where, RN
and DN are retained nitrogen and digestible nitrogen,
2.1.2. Body composition, energy requirement and feed respectively. DE of feed was estimated from the contents of
requirement nutrients in the feed, including crude protein (CP, %), crude
The data set on body composition was obtained from 56 lipid (CL, %) and carbohydrate (CHO, %), and their apparent
studies published between 2012 and 2022, including 313 sets of digestibility coefficient (ADC) (Bureau et al., 2003).
observations with fish mass from 6.00g to 201.23g (Table 2).
DEfeed ¼ 23:6  CP  ADCCP þ39:5  CL  ADCCL þ 17:2
Data were selected for model analysis consistent with the (8)
CHO  ADCCHO
official analysis method of the Associations of Analytical
Chemists (AOAC, 2005). Isometric regression models Finally, the feed requirement was determined as follows:
(y ¼ aW þ b) and allometric regression models (y ¼ aWb)
DEreq
were used to describe the relationship between body mass (BM) Feedreq ¼ (9)
DEfeed
and body composition (moisture, protein, ash, lipid and gross
energy). The relationships between body phosphorus (BPh) and
2.1.3. Waste output model
BM were explored by a similar method using 69 observations
Total aquaculture waste (TW) includes solid and dissolved
from 9 studies published between 2017 and 2022 (Table 2).
waste that can be calculated using the approach based on
DEreq was estimated by the sum of retained energy (RE),
bioenergetics and nutrient mass balance recommended by
basal metabolism energy (HeE), urinary and branchial energy
Chowdhury et al. (2013). When the uneaten feed is minimal in
(UE þ ZE) and heat increment of feeding (HiE) (Cho & Bureau,
fish cultured, waste includes solid and dissolved waste and
1998) (Eq. (4)). HeE was estimated by oxygen consumption rate,
can be estimated according to Eqs. (10)e(12).
and the oxyecalorific coefficient is 13.6 kJ g1 [O2] consumed.
HeE (kJ fish1) was estimated by the equation that as a function TSW ¼ DMintake  ð1  ADCDM Þ (10)
of temperature and metabolic body mass (Kaushik, 1998) (Eq.
(5)). The dataset of oxygen consumption of largemouth bass SWN or SWP ¼ Nintake or Pintake  ð1  ADCN or ADCP Þ (11)
was from Yang et al. (2019) and Cai et al. (2012). Retained en-
ergy (RE) was calculated by the differences in gross energy (GE)
of final body mass and initial body mass. The data set of the DWN or DWP ¼ ðNintake or Pintake  ADCN or ADCP Þ  Nretained or Pretained
energy budget of largemouth bass was obtained from 69
(12)
 b i o s y s t e m s e n g i n e e r i n g 2 2 7 ( 2 0 2 3 ) 1 9 5 e2 1 6 199

Where, TSW, SWN, SWP, DWN and DWP are total solid waste,
Table 2 e List of studies used for analysing body chemical
solid nitrogen wastes, solid phosphorus wastes, dissolved
composition and body phosphorus composition of
largemouth bass. nitrogen and dissolved phosphorus wastes, respectively; DM
is dry matter.
Parameter Published studiesa
Growth Chen et al. (2012); Xiao (2012); Zhu et al. 2.1.4. Model validation
(2012); Yun et al. (2013); Yuan et al.
The growth and feed requirement models for largemouth bass
(2014); Gou et al. (2015); Frederick et al.
(2016); Huang et al. (2017); Lin et al.
were validated using production data from three different
(2017); Ding (2018); Jiang et al. (2018); Li cultured modes; a temperature-controlled recirculating aqua-
et al. (2018); Ren et al. (2018); Song et al. culture system (RAS), a Juanyang aquaculture system (JAS) (He
(2018); Gong et al. (2019); Guo et al. (2019); & Hou, 2021) at Huazhong Agriculture University (Wuhan,
Li et al. (2019 a,b); Shi et al. (2019); Yin et China) and the land-based recirculating aquaculture systems
al. (2019); Yu et al. (2019); Zhang et al.
(LRAS) in Wuhan Wuqi Oriental Aquaculture Co., Ltd (Wuhan,
(2019); Ding et al. (2020); Cai et al. (2020);
China). A total of 12 series of production data were collected for
Chen et al. (2020); Guo et al. (2020); He
(2020); He et al. (2020a,b); Li et al. (2020 largemouth bass with initial body mass of 2.0e9.7 g and final
a,b); Ren et al. (2020); Romano et al. body mass of 255e480g, and the culture period was 153e260 d.
(2020); Wang et al. (2020 a,b,c,d; Wang All the fish cultured in three modes were fed for satiation twice
(2020); Xie et al. (2020); Xu et al. (2021); a day with three commercial feeds (Table 3), and recorded the
Zhang et al. (2020); Zhou et al. (2020); daily feed intake and mortality. The fish raised in different
Chen et al. (2021); Li et al. (2021 a,b,c); Du
tanks were weighed every two weeks for their average body
et al. (2021); Liu et al. (2021); Lu et al.
mass. The revised TGC growth model was validated by
(2021); Miller et al. (2021); Ni et al. (2021);
Rahman et al. (2021); Rossi et al. (2021); comparing the observed body mass with the model-predicted
Wang et al. (2021 a,b); Wu et al. (2021 body mass, and the feed requirement model was validated
a,b); Yin et al. (2021 a,b); Yusuf et al. using the observed feed conversion ratio (FCR) with the model-
(2021); Fan et al. (2022); Gu et al. (2022); predicted FCR. The FCR was calculated as follows:
He et al. (2022); Liang et al. (2022); Liu
et al. (2022); Ma et al. (2022); Romano FI
FCR ¼ (13)
et al. (2022); Xi et al. (2022); Xie et al. Wn  W0 þ Wd
(2022); Yang et al. (2022 a,b); Yu et al.
Where, FI and Wd are feed intake and body mass of dead fish,
(2022); Zhang et al. (2022 a,b); Zhao et al.
(2022 a,b); Zhong et al. (2022).
respectively. Wn is the final body mass and W0 is the initial
Body Chen et al. (2012); Zhou et al. (2012); Zhu body mass.
composition et al. (2012); Yun et al. (2013); Gou et al.
(2015); Frederick et al. (2016); Huang et al. 2.2. Overall design of the fish precision feeding system
(2017);Ding (2018); Jiang et al. (2018); Li
et al. (2018); Lin et al. (2018); Song et al. The precision feeding system includes a data acquisition de-
(2018); Guo et al. (2019); Li et al. (2019 a);
vice, automatic feeding machine, platform client and cloud
Shi et al. (2019); Yin et al. (2019); Zhang
et al. (2019); Ding et al. (2020); Guo et al.
server (Fig. 1). The platform client is used to input and store
(2020); He (2020); He et al. (2020 b); Li fish basic culture information, set feeding frequency and
et al. (2020 a,b); Ren et al. (2020); Wang feeding time. Basic culture information includes fish species,
et al. (2020 a,b); Wang (2020); Du et al. initial stocking size, initial fish biomass, daily mortality and
(2021); Li et al. (2021 a); Miller et al. (2021); feed nutrient composition. Data acquisition devices, including
Yadav et al. (2020); Yang et al. (2020);
water quality monitoring sensors (DO and temperature
Zhang et al. (2020); Zhong et al. (2020);
Zhou et al. (2020); Rossi et al. (2021);
Wang et al. (2021 a,b); Wu et al. (2021
a,b); Xu et al. (2021); Yin et al. (2021 a,b); Table 3 e Feed composition and digestible energy in three
Yusuf et al. (2021); Chen et al. (2022); Cui different cultured modes of largemouth bass.
et al. (2022); Feng et al. (2022); Gu et al.
(2022); He et al. (2022); Liu et al. (2022);
Composition of feed Feed 1a Feed 2b Feed 3c
Romano et al. (2022); Xie et al. (2022); Crude Protein% 49.58 50.1 50.78
Yang et al. (2022 a); Zhang et al. (2022 a); Crude Lipid% 12.6 10.5 11.3
Zhu et al. (2022 a,b) Crude Ash% 10.6 12.4 11.6
Body Huang et al. (2017); Li et al. (2020 b); Ren Phosphorus% 1.46 1.59 1.48
phosphorus et al. (2020); Wang (2020); Wang et al. Gross energy, kJ g1 19.2 18.4 18.8
(2021 a); Wu et al. (2021 a, b); Cui et al. Digestible dry matter% 62.4 60.7 61.6
(2022); Liu et al. (2022) Digestible crude protein% 44.6 45.1 45.7
Energy budget Gou et al. (2015); Ding (2018); Li et al. Digestible Phosphorus% 0.99 1.08 1.01
(2019 a); He (2020); He et al. (2020 b); Digestible energy, kJ g1 16.5 15.8 16.2
Wang (2020); Wang et al. (2020 a); Xiao a
Feed from Hanchuan Hengxing Feed Co., Ltd (Hubei, China).
(2012); Zhu et al. (2022 b) b
Feed from Jingmen Wulong Feed Co., Ltd (Hubei, China).
a c
All published studies in the table are listed in Appendix A. Feed from Tianmen Yuehai Feed Co., Ltd (Hubei, China).
200 b i o s y s t e m s e n g i n e e r i n g 2 2 7 ( 2 0 2 3 ) 1 9 5 e2 1 6

Fig. 1 e Schematic diagram of the fish precision feeding system based on multi-factor comprehensive control.

sensor) and fish feeding activity sensor, collect the real-time 2.2.2. Fish feeding activity sensor
information of water temperature, DO and fish feeding A fish feeding activity sensor was used to send water move-
behaviour of the culture ponds or tanks, and send the infor- ment of the fish in the non-feeding state and normal feeding
mation to the cloud server. The cloud serves to automatically state to the cloud server in real time through the ESP8266-wifi
analyse the acquired information and adjust the current module (Espressif Systems company, Shanghai, China). The
feeding amount according to the culture background, feed water movement data was obtained by fixing the floating body
nutritional composition, water quality (water temperature in the feeding area. Fish feeding behaviour causes water to
and DO) and fish feeding activity intensity and then send move, driving the floating body to drift, sensors inside the
signal to continue or stop the feeding. floating body monitored the water movement which was
uploaded to the cloud server. The current water movement of
2.2.1. Water quality monitoring system the fish in non-feeding state was set as the initial threshold,
The real-time water quality monitoring system includes and the current water movement in normal feeding state was
sensor module, analogue-to-digital (AD) convert module and set as an extra or activation threshold. The built models on the
GPRS wireless communication module. The structural dia- cloud server analysed the relationship between fish feeding
gram of the water quality monitoring system was shown in activities intensity and the thresholds during fish feeding, and
Fig. 2. Temperature sensor and DO sensor was used to obtain sent the different instructions to the single-chip (MCU) which
the water temperature and DO content in the farming ponds directly or indirectly through a relay adjusted the feeding
or tanks. The AD convert module is responsible for converting amount, feeding speed and feeding duration of the automatic
electrical signals into digital signals. A wireless communica- feeding machine (Fig. 3). The relay module was used to turn
tion module USR-GPRS-780, developed by Youren communi- on/off the power supply to start or close the automatic feed
cation company, Jinan, China, was used to create the machine by the different instructions from MCU. After feeding
transmission channel between the cloud server and base was completed, feedbacks from the feeding machines were
station for data transmission and reception, thus realising 24- sent back to the cloud server by the MCU and the ESP8266-wifi
h continuous collection, processing and uploading of aqua- module stored and analysed the data for adjusting the next
culture water temperature and DO parameters. feeding operation.

Fig. 2 e Flow chart of the water quality monitoring system.

 b i o s y s t e m s e n g i n e e r i n g 2 2 7 ( 2 0 2 3 ) 1 9 5 e2 1 6 201

Fig. 3 e Flow chart of feeding activity sensor regulating feed.

2.2.3. Automatic feeding machine quality parameters (water temperature and DO), fish activity
The automatic feeding machine includes a feed bin, an intensity and other necessary data for the implementation of
unloading unit, a spreading unit, and a single-chip feeding the software package, and provide data services for other
controller (Fig. 1). The blanking unit comprises an auger and systems. The middle layer is the business logic layer, which is
the first motor. The spreading unit comprises a centrifugal developed based on Java Server Pages (JSPs). It is responsible
spreading tray and a second motor. The feeding controller can for performing algorithm data analysis and processing the
control the operation of the blanking unit and the spreading business logic of the software system, and providing various
unit according to the feeding instruction sent by the cloud back-end services (e.g., mass calculation, feed requirement
server to control the feeding amount, feeding speed and calculation, waste output calculation, actual feeding amount
feeding duration. The auger is located at the outlet of the feed calculation, etc.) in response to requests from the UI layer. The
bin for conveying the feed from the feed bin to the spreading top layer is the user interface layer, which is based on HTML5/
unit. A fixed amount of feed drops from the bin with each CSS technology and can support the access of any modern
rotation of the auger. The first motor adopts a 24V closed-loop browser, providing users with a visual interface for interactive
subtraction DC motor, which is connected to the feeding operation and satisfying simultaneous access and operation
controller and the auger, respectively. The feeding controller by multiple users. The Apache Tomcat (v8.0) used in this study
adjusts the power of the first motor according to the feeding is based on Java technology and can be used as both applica-
instruction sent by the cloud server, and then adjusts the tion server and web server. Its advantage lies mainly in its
running speed of the auger to achieve precise regulation of the open-source code, which support cross-platform applications,
feeding speed. The centrifugal spreading tray is arranged so the developed software packages can run on almost all
below the auger and connected with the second motor that platforms, including Windows and Linux operating systems.
distributes feed into the culture ponds or tanks. The second
motor is a 24V DC motor, which is connected to the feeding
controller and the centrifugal spreading tray, respectively.
According to the feeding instruction sent by the cloud server,
the feeding controller can adjust the spreading distance or
range up to 10 m by controlling the speed of the second motor.
Besides, the automatic feeding machine includes a laser
sensor located at the top of the feed bin, which is used to
determine the drop distance of the feed in the bin after each
feeding, and send the drop distance to the cloud server. The
cloud server calculates and stores the actual feeding amount
each time based on the feed drop distance and the cross-
sectional area of the feed bin. The feeding machine is
charged either by solar battery or a plug-in electrical outlet.

2.2.4. Development of platform client

The platform structure includes 3 layers of applications: user
interface layer (UI), business logic layer (BLL) and data access
layer (DAL) (Fig. 4). The bottom layer takes the database as the
basic access layer, and its main function is to store and access Fig. 4 e Conceptual framework of the platform client in the
all fish production data, model-related parameters, water fish precision feeding system.
202 b i o s y s t e m s e n g i n e e r i n g 2 2 7 ( 2 0 2 3 ) 1 9 5 e2 1 6

equation was higher than in the allometric equation (Fig. 6)

3. Results (Table 4). However, simple linear isometric equations for other
body chemistry composition, including body lipid (BL), ash
3.1. Bioenergetics-based combined nutrition model (BA), and phosphorus (BPh) provided more reliable estimates
than the allometric equation (Fig. 6). Additionally, gross en-
3.1.1. Determination of growth models ergy also increased as isometric fashion, with slightly lower
The TGC values of largemouth bass with different body mass RSS values in the isometric equation than in the allometric
displayed certain growth laws. When the mass of the large- equation (Fig. 6) (Table 4).
mouth bass was in the range 2e500g, there were two break- The proportion of retained energy (RE), faecal energy (FE),
points in the piecewise linear plateau model (Fig. 5). metabolic energy (ME) and urinary and branchial energy
Depending on the location of these points (the intersection of (UE þ ZE) in intake of energy (IE) were 37.52%, 14.62%, 41.57%
the two lines), the growth cycle of largemouth bass was and 6.39%, respectively (Fig. 7A). The energy requirement of
divided into three main growth stages: <16.21g (stage 1), largemouth bass for HeE increased first and then decreased
16.21e183.68g (stage2), and >183.68g (stage 3). The optimal with increasing temperature from 15  C to 33  C, and reached
body exponents (1-b) were different of each sub-model at the maximum at 31  C with a body mass exponent 0.7827. By
different growth stages. By minimising the residual sum of analysing the oxygen consumption data of largemouth bass
squares (RSS), the (1-b) of largemouth bass at the three growth with different body masses at different water temperatures
stages were 0.124 (stage1), 0.403(stage2), 0.993(stage 3), with (13  Ce33  C), the HeE can be determined as HeE ¼
corresponding TGC values of 0.026, 0.188 and 9.181, respec- (0.0781 T2þ4.8805T-33.7243)  (BM0.7827) (Fig. 7B). The urinary
tively (Eq. 14). and branchial energy (UE þ ZE) was closely correlated with
8 digestible energy intake (DEintake) and at rates of 7.49% DEintake
>
>  P  1
>
> 0:026  T 0:124 and 8.08% (HeE þ HiE þ RE), respectively. Both the HeE and HiE
>
> W00:124 þ ; W0 < 16:21
>
> 100
>
> were significantly correlated with ME, and by the equation
>
>
>
< P  1 ME ¼ HiE þ HeE, the HiE can be estimated by 47.64%HeE
0:188  T 0:403
Wn ¼ W00:403 þ ; 16:21  W0 < 183:68 (14) (Fig. 7C). Based on the formula for each energy, the digestible
>
> 100
>
> energy, oxygen requirements and feed requirements of
>
>
>
>  P  1
>
> 9:181  T 0:993 largemouth bass with different body masses can be calcu-
>
> 0:993
þ ; W0  183:68
>
: W 0 lated (Table 5).
100

Where, n (¼ 1, 2, …) is the day number recorded from W ¼ W0, 3.1.3. Estimation of waste output
W0 and Wn are initial body mass (g) and final body mass (g), The outputs of total solid waste and dissolved nitrogen and
respectively; T is average daily temperature ( C). phosphorus wastes of largemouth bass fed with three
different feeds were shown in Table 6. The feed1 used in
3.1.2. Estimation of body composition, energy requirement mode1 had the highest digestible dry matter and was pre-
and feed requirement dicted to have the lowest total waste output (307 kg t1 and
The body chemical composition (%) of largemouth bass from 336 kg t1 of feed fed and fish produced. The total waste per kg
6.0g to 201.2g was relatively stable with increasing body mass feed per tonne of fish produced in the three feeds ranged from
(Fig. 6A). The RSS of body protein (BP) in the isometric 58.3 to 62.9 g for N and from 10.2 to 12.2 g for P.

Fig. 5 e Piecewise linear regression of the thermal-unit growth coefficient (TGC) as the function of body mass (BM): (a)
BM < 100g, (b)BM > 15g. Dotted lines indicate the body mass at break point.
 b i o s y s t e m s e n g i n e e r i n g 2 2 7 ( 2 0 2 3 ) 1 9 5 e2 1 6 203

Fig. 6 e Whole body chemical composition (%) of largemouth bass in relation to body mass(A); Isometric and allometric
relationships between whole body mass and body protein (B1 and B2), lipid (B3 and B4), ash (B5 and B6), phosphorus (B7 and
B8) and gross energy (C1 and C2) of largemouth bass.

3.1.4. Model validation Agricultural University and commercial farms. The results
The growth model and feed requirement models were vali- showed that the observed live body mass was significantly
dated at the aquaculture research station of Huazhong correlated with the predicted values by the model in all three
aquaculture modes (P < 0.05). The R2 values were higher in
RAS (from 0.997 to 1.00) (Fig. 8A) and JAS mode (from 0.984 to
0.988) (Fig. 8B) than in LRAS mode (from 0.922 to 0.931)
Table 4 e Residual sum squares (RSS) of the isometric and
(Fig. 8C). In addition, the observed feed conversion ratio (FCR)
allometric regression analysis of body.
significant correlated with the predicted values in RAS and JAS
Predicts Number Isometric RSS Allometric RSS mode (P < 0.05) (Fig. 8A, B). But in LRAS mode, only the
Body protein 313 109 106 observed FCR was significantly correlated with the predicted
Body lipid 307 602 637 FCR in tank L2 (P < 0.05) (Fig. 8C).
Body ash 277 38 49
Gross energy 307 950 781 994 589
3.2. Implementation of the precision feeding system
Body phosphorus 69 0.44 0.46

Mass with body protein, lipid, ash, phosphorus and gross energy as The whole feeding process for the fish precision feeding
predictors.
system based on a multi-factor comprehensive control is a
204 b i o s y s t e m s e n g i n e e r i n g 2 2 7 ( 2 0 2 3 ) 1 9 5 e2 1 6

Fig. 7 e The energy budget of largemouth bass. A: the relationship between retained energy (RE), metabolic energy (ME),
faecal energy (FE), urinary and branchial energy (UE þ ZE) and intake of energy (IE), and IE¼FE þ RE þ MEþ(UE þ ZE). B:
Multiple regression model of basal metabolism (HeE) of largemouth bass in relation to body mass (BM, g) and temperature
( C) when fish reared at water temperature between 13 and 33  C; C: the relationship between UE þ ZE, heat increment of
feeding (HiE), HeE and other energy.

step-by-step regulation process of general control, fine- background (e.g. fish species, initial stocking size, initial fish
adjustment and precision-adjustment. The general princi- biomass and daily mortality), feed nutritional composition
ple is to obtain the feed requirement based on the and real-time water temperature, the precision feeding sys-
bioenergetics-based nutrition model, carry out fine adjust- tem obtained a reasonable feed requirement based on the
ment to adjust the feed requirement given by the nutrition fish bioenergetics models. After receiving the signal from the
model according to the real-time DO level, and carry out cloud server, the cloud server provides real-time feedback on
precise adjustment to precisely regulate feeding amount, the current DO level through the DO sensor, fine-tunes the
feeding speed and feeding duration according to feeding feed requirements according to the current DO level, and
activity intensity. The flow chart of the fish precision feeding sends the first feeding instruction to the automatic feeding
system based on multi-factor comprehensive control is machine at a set feeding time to start the feeding machine.
shown in Fig. 9. According to the data of fish culture At the same time, the fish feeding activity information is
 b i o s y s t e m s e n g i n e e r i n g 2 2 7 ( 2 0 2 3 ) 1 9 5 e2 1 6 205

Table 5 e Energy requirement, oxygen requirement and expected feed conversion ratio of largemouth bass growing from
1g to 500g based on assumption that the fish are reared at 25  C.
BM REa HeEb HiEc UE þ ZEd DEe Oxygenf FCRg Feed
1 1 1
g fish MJ kg g kg [body mass gain] %BM d1
1 7.18 2.14 1.02 0.84 11.17 232 0.74 4.8
5 7.18 2.54 1.21 0.88 11.81 276 0.75 3.9
10 7.18 2.65 1.26 0.90 11.99 288 0.77 3.5
25 7.18 2.84 1.35 0.92 12.29 308 0.81 2.9
50 7.18 3.29 1.56 0.97 13.01 357 0.85 2.6
100 7.18 3.61 1.72 1.01 13.52 392 0.88 2.0
200 7.18 3.93 1.87 1.05 14.03 426 0.95 1.5
300 7.18 4.33 2.06 1.10 14.67 470 1.10 1.1
500 7.18 5.46 2.60 1.23 16.47 592 1.39 0.9
a
RE (kJ fish1) ¼ 7.18  (body mass at T (nþ1))-7.18  (body mass at Tn).
b
HeE¼(-0.0769 T2þ4.8105T-33.2077)  (BM0.78).
c
HiE ¼ 0.4756  HeE.
d
UE þ ZE ¼ 0.0808  (RE þ HeE þ HiE).
e
DE requirement ¼ RE þ HeE þ HiEþ(UE þ ZE).
f
Oxygen requirement ¼ (HeE þ HiE)/oxyecalorific coefficient (13.6 kJ g1 [O2] consumed).
g
Expected FCR ¼ mass of feed fed/body mass gain of fish.

transmitted to the cloud server. Based on the analysis of the client of the precision feeding system. Users can select the
relationship between the real-time feeding activity intensity species of fish and input the parameters such as stocking fish
of the fish group and using the set threshold, the cloud number, initial body mass, culture period, feed nutritional
server sends the second feeding command to the feeding composition, water temperature (or by water quality moni-
machine and further adjusts the feeding amount and feeding toring sensor) and current mortality to determine the current
speed until the feeding is completed. In addition, the cloud body mass, fish biomass and feed requirements (Fig. 9). The
server of the system also includes a correction unit, which is system also allows users to modify some parameters by
used to calculate the actual feeding amount each time ac- inputting their own production data to better match the
cording to the dispersed distance of feed after each feeding, model to different cultured modes and environments.
and then saves and corrects the current feed requirement Through the online platform client, after selecting the fish
based on the actual feed feeding amount and environmental species, users can input culture period (e.g. day, week or
parameters. month) and water temperature, and then the growth model
parameter TGC (labelled as empirical TGC) and corresponding
3.3. General control unit based on bioenergetics-based 1-b will be automatically adjusted. Using the production data
nutrition model of the tank 1 in LRAS to correct the parameters, the growth
and FCR prediction results for other tanks in LRAS were shown
The control unit incorporated all the functions of the above in Fig. 10. The model accuracy using empirical TGC from tank
models, such as growth prediction, feed requirement predic- 1 in LRAS was significantly higher than the model using
tion and nitrogen and phosphorus waste output prediction, theoretical TGC (labelled as theoretical TGC, based on the
and all these models have been implanted into the platform analysis of experimental data) (Figs. 8C and 10).

3.4. Fine-adjustment of feed requirement by dissolved

oxygen
Table 6 e Output of solid and dissolved wastes fed with
three feed compositions showed in Table 3.
Considering the influence of environmental factors, espe-
Parameters Feed composition cially DO on fish feeding, an optimisation model of DO is
in three modes established by analysing the DO level fed back by the system,
Feed 1 Feed 2 Feed 3 the predicted feeding requirement (PFR) by the model and the
TSW kg 1
t [feed fed] 307 323 314 observed feed intake (OFI). As shown in Fig. 11, When the DO
kg t1 [fish produced] 336 369 351 level was less than 5 mg l1, the ratio between the OFI and
Solid N waste kg t1 [fish produced] 11.4 11.9 11.8 PFR was a significant linear correlation with the DO level
Solid P waste kg t1 [fish produced] 5.7 6.4 5.8 (P < 0.001), which indicates that the lower DO reduces the
Dissolved waste
actual feed intake of fish, resulting in excessive feed
Dissolved kg t1 [fish produced] 46.9 51.0 50.3
requirement predicted by the model. Therefore, based on the
N waste
Dissolved kg t1 [fish produced] 4.5 5.9 4.8 DO levels collected automatically by the system, combined
P waste with the DO optimisation model, the feed requirement can be
Total N waste kg t1 [fish produced] 58.3 62.9 62.2 fine-tuned to obtain more accurate feed demand prediction
Total P waste kg t1 [fish produced] 10.1 12.2 10.6 results.
206 b i o s y s t e m s e n g i n e e r i n g 2 2 7 ( 2 0 2 3 ) 1 9 5 e2 1 6

Fig. 8 e The relationship between observed and predicted body mass (BM), and observed and predicted feed conversion ratio
(FCR) of largemouth bass at different tanks. A: Fish were cultured in indoor recirculating aquaculture system (RAS); B: The
fish were cultured in JuanYang aquaculture system (JAS); C: The fish was cultured in land-based recirculating aquaculture
systems (LRAS). The blue and red shaded areas were 95% confidence intervals.
 b i o s y s t e m s e n g i n e e r i n g 2 2 7 ( 2 0 2 3 ) 1 9 5 e2 1 6 207

Fig. 9 e The feeding operation flow of fish precision feeding system based on multi factor comprehensive control.

3.5. Precision-adjustment of feed requirement by fish determine the fish appetite (Fig. 12b). If the fish do not stop
feeding activity sensor feeding in advance, the power supply of the feeding machine
is turned off after feeding the required amount of feed.
In the actual feeding process, according to the real-time Conversely, if the fish stop feeding in advance, the current
feeding activity intensity of fish and the initial threshold, the feeding amount and DO level for the secondary correction of
motor power of the blanking unit of the feeding machine was the feed requirement is feedback.
adjusted to change the feeding speed for increasing or
reducing feeding, thus avoiding excessive dissolution of feed
caused by too fast feeding speed and injury of fish competing 4. Discussion
for food caused by too slow feeding speed. When the fish
feeding activity intensity is greater than the activation The strong demand for precise, digitised and intelligent con-
threshold determined by the built models, an instruction to trol systems in aquaculture and the rapid development trend
increase feeding is followed. When fish feeding activity in- of 5G technology could provide a good opportunity for the
tensity is between the thresholds, an instruction to reduce developing a precision aquaculture feeding system. The
feeding is triggered. The feeding machines are instructed to aquaculture environment is complex and challenging, with a
stop feeding when the fish feeding activity intensity is close to variety of culture modes for raising aquatic animals, including
the initial threshold (Fig. 9). The dynamic diagram of fish tanks, ponds, cages, runways, etc (Vis et al., 2020). Due to the
feeding activity intensity is shown in Fig. 12. The initial high turbidity of the water and the scattered distribution of
feeding intensity is output was once per second (Fig. 12a). In the animals, it is difficult to accurately estimate the fish
order to further improve the accuracy and stability, this study populations and feed requirements (Davis & Hardy, 2022).
used the average of the feeding intensity every 30 s to Mathematical models have been proved to be effective in
208 b i o s y s t e m s e n g i n e e r i n g 2 2 7 ( 2 0 2 3 ) 1 9 5 e2 1 6

Fig. 10 e The relationship between observed and predicted body mass (BM) (a, b, c and d), and the relationship between
observed and predicted feed conversion ratio (FCR) (A, B, C and D) of largemouth bass at different tanks. The fish was
cultured in land-based recirculating aquaculture systems (LRAS). The body mass and FCR was predicted using empirical
TGC value and 1-b. The blue and red shaded areas were 95% confidence intervals.

predicting growth, optimal feed ration, FCR and waste output predicting different culture modes and environments can be
of fish fed diets of varying nutritional composition and reared quite different, so it is important to optimise the model to
in different environments (Dumas et al., 2007; Liu et al., 2018; improve its applicability in the actual cultured process. In the
Sun et al., 2016). In this study, by analysing the growth tra- current study, the parameters modified based on the analysis
jectory of largemouth bass, and using piecewise analysis, of production data from one culture tank were used to predict
growth models of different growth stages were established the fish mass and FCR in other culture tanks under the same
according to the location of the growth allometric point, thus culture mode. The prediction accuracy of the model with
improving the prediction of the growth models. Based on the modified parameters was significantly higher than that of the
principles of bioenergetics and animal nutrition, the feed re- model with fixed parameters, and there was also a significant
quirements and waste output of largemouth bass under correlation between the predicted and observed values, indi-
different culture modes were also estimated. The validation cating that it is feasible to use the models with optimised
results under different culture modes illustrated that the parameters according to the previous production data to
model has good prediction effect in a certain application predict the current production and feed requirement on the
range, especially under the same culture mode and environ- premise of the same culture mode and environment. To
ment. However, our research has also shown that even for the improve the practicability of this function, the function
same species, the accuracy of the fixed-parameter models in module has been implanted into the platform client and
labelled as empirical TGC. when using the prediction function
of the platform, users can select the empirical TGC values
generated from their own farm data to improve the accuracy
of the model.
Water temperature and DO are the primary factors deter-
mining the growth and feed intake of fish (Yang et al., 2015;
Davis & Hardy, 2022). Since fish are variable temperature an-
imals, the water temperature affects the fish appetite as well
as altering fish energy metabolism, especially the basic
metabolism, which influences the growth performance and
feed utilisation efficiency of fish (Favero et al., 2022; Liu et al.,
2018). Fish farming is usually carried out in open production
systems, where DO levels can be regulated to an appropriate
level (4.5 mg l1) by aerator without affecting fish growth
Fig. 11 e Linear programming model for dissolved oxygen and feed intake, but culture water temperature varies with the
(DO). natural ambient temperature (Vis et al., 2020). Therefore, the
 b i o s y s t e m s e n g i n e e r i n g 2 2 7 ( 2 0 2 3 ) 1 9 5 e2 1 6 209

Fig. 12 e The dynamic change diagram of fish feeding activity intensity. (a): Feeding activity intensity value output per
second. (b): Average feeding activity intensity output every 30 s.

TGC growth model, which considers water temperature vari- spreading distance (range up to 10 m) by controlling the speed
ation, was used instead of other commonly used growth of the second motor of automatic feeding machine according
models, and the effect of water temperature on basal meta- to the feeding instructions sent by the cloud server to promote
bolic energy was also considered when predicting the energy the feed evenly distributed on the water surface of the feeding
requirement of fish. However, in practical aquaculture, DO area, thus ensuring that each individual fish has the oppor-
levels are often affected by various external factors, such as tunity to be adequately fed, and avoiding fish gathering and
extreme weather and water eutrophication (Coffin et al., 2018; rushing for food, which consumes physical energy and causes
Irby et al., 2018; Rixen et al., 2010), so it is clearly necessary to feed to sink to the bottom and be wasted.
adjust feed requirements according to DO levels. Based on the The general trend in the development of intelligent feeding
feeding requirement predicted by the nutrition model and the control is to combine various methods or technologies to
DO level obtained from the water quality monitoring system, obtain more reliable and precise indicators for feeding
the DO level is significantly correlated with the ratio between decision-making (Zhou et al., 2017). Although mathematical
the predicted and the observed value of the feed requirement, models can explain the continuous effects of environmental,
indicating that this fine-tuning can effectively improve the physiological and nutritional factors on growth trajectory and
prediction accuracy of feeding requirements. feed intake of fish, and estimate the feed requirement to meet
To minimise the economic inefficiency and pollution load growth of fish (Chowdhury et al., 2013), there are too many
of the culture system, feeds should preferably be presented in factors affecting fish feed intake, especially under abnormal
a form that promotes rapid and complete consumption (Davis culture conditions (e.g., fish disease, extreme weather, low DO
& Hardy, 2022). The appropriate feeding method should take and high ammonia levels in water), resulting in the actual feed
into account the feeding situation of the fish group and grasp intake of fish often cannot be in accordance with the expected
the feeding rhythm (Millot and Be gout, 2009). Usually, the feed requirements (Davis & Hardy, 2022). Moreover, the dis-
feeding rhythm is “slow at both ends and fast in the middle” tribution of each feeding amount in a day, and “how and for
(Zhou et al., 2017). This is because at first, the fish group has how long” are also important issues when using mathemat-
not yet fully entered the optimal feeding state. With the in- ical models to estimate the feed intake of fish. Compared to
crease of feeding time, the number of fish feeding continues to bioenergetics-based nutritional models, fish feeding behav-
increase until some fish reach or close to satiety, and the iour sensor provides an automatic and relatively simple
number of fish feeding gradually decreases. The self- method for analysing fish appetite and feeding behaviour (Li
developed fish feeding activity sensor based on current et al., 2020; Zhou et al., 2018). However, due to the strong
water movement provides a new approach to assess fish sense of imitation and competition in the feeding activities of
appetite automatically, consistently, and independent of vis- fish groups (especially some predatory fish), ignoring the
ibility conditions. By setting up a computer program, the biological characteristics, the physiological and nutrients
automatic feeding machine can change the feeding speed and needs of fish, feeding relying solely on fish appetite may also
feeding duration depending on the fish feeding activity in- lead to excessive feeding of fish, especially under commercial
tensity obtained from the sensor, which can effectively reduce conditions. It is easy to overfeed if the application of feed is too
the amount of uneaten feed and the residence time of feed in aggressive and exceeds the feeding intensity, resulting in the
the water. Also, the feeding controller can adjust the uneaten feed being waste (Davis & Hardy, 2022).
210 b i o s y s t e m s e n g i n e e r i n g 2 2 7 ( 2 0 2 3 ) 1 9 5 e2 1 6

How to effectively couple variety prediction and control through feedback and cultured data provided by the cloud
methods to achieve fish intelligent and precision feeding has serves and users improves the accuracy, precision and
been a long-term exploration by many researchers (Føre et al., repeatability of the system further enhancing the possible
2018; Zhou et al., 2017). This study coupled the bioenergetics- application and scope of the system.
based nutritional model with some mechanical devices, such Due to the complexity of fish culture environments and
as water quality sensors (water temperature and DO), fish culture modes, different technologies or methods have their
feeding activity sensor and automatic feeding machines, and own advantages and disadvantages in determining feeding
used IoT technology to implant all functional modules into the amount. In this study, the precision feeding system compre-
software platform to form a precision feeding system, which hensively considers multiple factors such as fish biological
realizes centralized control, remotely management and characteristics, physiological stages, feed nutrient composi-
dynamically adjustment parameters through computer pro- tions, environmental factors and fish appetite, and combined
gram settings and cloud servers, and presents visual charts or with relevant technologies or equipment. This is an important
tables for users through the platform client. The precision attempt to synthesise multiple methods for determining the
feeding system proposed in this study integrates the predicted feeding requirements and provide a commercially available,
values of mathematical models, environmental parameters, reliable and robust tool to assist feeding decision-making for
and real-time fish feeding activity intensity to comprehen- fish farmers and managers, contributing to the shift of aqua-
sively regulate the feeding level of fish. The feedback mech- culture from experience-driven to knowledge-driven methods
anism in the system proposed is used to detect the fish feeding for increasing productivity, improving fish culture and
behaviour until feeding is completed, which allows the fish to enhancing environmental sustainability.
be satiation without overfeeding and consequent feed waste.

Declaration of competing interest

5. Conclusions
The authors declare that they have no known competing
This study developed a precision feeding system for large- financial interests or personal relationships that could have
mouth bass (M. salmoides) based on multi-factor comprehen- appeared to influence the work reported in this paper.
sive control, which integrates mathematical models,
environmental parameters and real-time fish feeding behav-
iour and automatic feeding machine. By using step-by-step
Acknowledgments
adjustment of the feeding amount and controlling the
feeding speed and feeding duration of the automatic feeding
Thanks to anonymous reviewers for their constructive com-
machine, the system can achieve precise and automatic
ments to help improve the paper. This work was financially
feeding operation, as well as providing the potential for
supported by the National Key Research and Development
remote control and the centralised management of culture
Program of China, China [grant number: 2019YFD0900303 and
ponds or tanks. The bioenergetics-based nutritional model
2019YFD0900200] and Fundamental Research Funds for the
proposed in this study is based on feed nutritional composi-
Central Universities (grant number: 2662020SCPY005).
tion, body mass gain, body composition and energy require-
ment and this has taken into consideration biological Appendix
characteristics, growth patterns, and the influence of water
temperature on growth, feed utilisation and energy meta- References cited in Table 2: List of studies used for analysing
bolism. The similarities in body mass and FCR between the body chemical composition and body phosphorus composi-
observed values in three different cultured modes and pre- tion of largemouth bass.
dicted values by the models or optimised models suggests that
the nutritional model constructed in this study can provide
reliable estimates of largemouth bass growth, feed require- references
ment and N, P waste output when fish reared at different feed
compositions, temperature and cultured modes. The DO level
is significantly correlated with the ratio between the predicted Abidi, S. F., & Khan, M. A. (2014). Evaluation of feeding rate based
value and the observed value of the feed requirement, indi- on growth, feed conversion, protein gain and carcass quality
cating that using the real-time DO parameter from the online of fingerling Indian major carp, Catla catla (Hamilton).
Aquaculture Research, 45(3), 439e447. [Link]
water quality monitoring system to fine-tune the feed
j.1365-2109.2012.03245.x
requirement can effectively improve the prediction accuracy AOAC. (2005). Official methods of analysis (18th ed.). Washington,
of feed requirements. In this study, the fish feeding activity DC, USA: AOAC International.
sensor was combined with automatic feeding machine to Buentello, J. A., Gatlin, D. M., & Neill, W. H. (2000). Effects of water
adjust the feeding pattern, which can further improve the temperature and dissolved oxygen on daily feed consumption,
accuracy of feeding, and the developed platform client visu- feed utilization and growth of channel catfish (Ictalurus
alises all functions in the precision feeding system, providing punctatus). Aquaculture, 182, 339e352. [Link]
S0044-8486(99)00274-4
a more convenient operation interface and intuitive results
Bureau, D. P., Gunther, S. J., & Cho, C. Y. (2003). Chemical
presentation, which can be accessed in any web browser. This composition and preliminary theoretical estimates of waste
dynamically adjusts the model and feeding parameters
 b i o s y s t e m s e n g i n e e r i n g 2 2 7 ( 2 0 2 3 ) 1 9 5 e2 1 6 211

outputs of rainbow trout reared in commercial cage culture Cui, Z., Zhang, J., Ren, X., & Wang, Y. (2022). Replacing dietary fish
operations in Ontario. North American Journal of Aquaculture, meal improves ecosystem services of largemouth bass
65(1), 33e38. [Link] (Micropterus salmoides) farming. Aquaculture, 550, Article
065<0033:CCAPTE>[Link];2 737830. [Link]
Bureau, D. P., & Hua, K. (2008). Models of nutrient utilization by Davis, D. A., & Hardy, R. H. (2022). Feeding and fish husbandry. In
fish and potential applications for fish culture operations. In Fish nutrition. San Diego, USA: Academic Press.
J. France, & E. Kebreab (Eds.), Mathematical modelling in animal Ding, G., Li, S., Wang, A., & Chen, N. (2020). Effect of chicken
nutrition (pp. 442e461). Wallingford, UK: CBA international. haemoglobin powder on growth, feed utilization, immunity
Buterbaugh, G. L., & Willoughby, H. (1967). A feeding guide for and haematological index of largemouth bass (Micropterus
brook, brown, and rainbow trout. The Progressive Fish-Culturist, salmoides). Aquaculture and Fisheries, 5(4), 187e192. https://
29(4), 210e215. [Link] [Link]/10.1016/[Link].2019.04.003
AFGFBB][Link];2. Ding, G. (2018). Study on the application of animal protein and
Cai, L., Bai, J. J., Li, S. M., Jiang, P., & Chen, K. C. (2012). plant protein in largemouth bass (Microptrrus salmoides) diets
Determination on the low lethal temperature tolerance, (in Chinese). Shanghai: Shanghai Ocean University. The
oxygen consumption rate and asphyxiation point of juvenile Thesis for the Doctorate.
largemouth bass. Journal of Shanghai Ocean University, 21(6), Dumas, A., France, J., & Bureau, D. P. (2007). Evidence of three
971e975. [Link] (in growth stanzas in rainbow trout (Oncorhynchus mykiss) across
Chinese). life stages and adaptation of the thermal-unit growth
Cai, Z. N., Qian, X. Q., & Xie, S. Q. (2020). Optimal dietary protein coefficient. Aquaculture, 267(1), 139e146. [Link]
concentrations for largemouth bass (Micropterus salmoides) of 10.1016/[Link].2007.01.041
different sizes (10e500g). Aquaculture International, 28(2), Du, R. Y., Zhang, H. Q., Chen, J. X., Zhu, J., He, J. Y., Luo, L.,
831e840. [Link] Lin, S. M., & Chen, Y. J. (2021). Effects of dietary Bacillus
Chen, W., Chang, K., Chen, J., Zhao, X., & Gao, S. (2021). Dietary subtilis DSM 32315 supplementation on the growth, immunity
sodium butyrate supplementation attenuates intestinal and intestinal morphology, microbiota and inflammatory
inflammatory response and improves gut microbiota response of juvenile largemouth bass. Micropterus salmoides.
composition in largemouth bass (Micropterus salmoides) fed Aquaculture Nutrition, 27(6), 2119e2131. [Link]
with a high soybean meal diet. Fish Physiology and anu.13347
Biochemistry, 47(6), 1805e1819. [Link] Fan, Z., Wu, D., Li, J., Zhang, Y., Cui, Z., Li, T., Zheng, X., Liu, H.,
021-01004-W Wang, L., & Li, H. (2022). Assessment of fish protein
Chen, S., Jiang, X., Liu, N., Ren, M., Wang, Z., Li, M., Chen, N., & hydrolysates in juvenile largemouth Bass (Micropterus
Li, S. (2022). Effects of dietary berberine hydrochloride salmoides) diets: Effect on growth, intestinal antioxidant
inclusion on growth, antioxidant capacity, glucose status, immunity, and microflora. Frontiers in Nutrition, 9.
metabolism and intestinal microbiome of largemouth bass [Link]
(Micropterus salmoides). Aquaculture, 552, Article 738023. https:// FAO. (2016). The state of world fisheries and aquaculture 2016. Rome,
[Link]/10.1016/[Link].2022.738023 Italy: FAO. [Link]
Chen, Y. J., Liu, Y. J., Yang, H. J., Yuan, Y., Liu, F. J., Tian, L. X., production.
Liang, G. Y., & Yuan, R. M. (2012). Effect of dietary oxidized fish FAO. (2021). The state of world fisheries and aquaculture 2021. Rome,
oil on growth performance, body composition, antioxidant Italy [Link]
defence mechanism and liver histology of juvenile production.
largemouth bass. Micropterus salmoides. Aquaculture Nutrition, Favero, G. C., dos Santos, F. A. C., da Costa Júlio, G. S., Batista, F. S.,
18(3), 321e331. [Link] Bonifa  cio, C. T., Torres, I. F. A., Paranhos, C. O., & Luz, R. K.
2095.2011.00900.x (2022). Effects of water temperature and feeding time on
Chen, Y., Sun, Z., Liang, Z., Xie, Y., Su, J., Luo, Q., Zhu, J., Liu, Q., growth performance and physiological parameters of Piaractus
Han, T., & Wang, A. (2020). Effects of dietary fish oil brachypomus juveniles. Aquaculture, 548, Article 737716. https://
replacement by soybean oil and l-carnitine supplementation [Link]/10.1016/[Link].2021.737716
on growth performance, fatty acid composition, lipid Feng, Z., Zhong, Y., He, G., Sun, H., Chen, Y., Zhou, W., & Lin, S.
metabolism and liver health of juvenile largemouth bass, (2022). Yeast culture improved the growth performance, liver
Micropterus salmoides. Aquaculture, 516, Article 734596. function, intestinal barrier and microbiota of juvenile
[Link] largemouth bass (Micropterus salmoides) fed high-starch diet.
Cho, C. Y. (1992). Feeding systems for rainbow trout and other Fish & Shellfish Immunology, 120, 706e715. [Link]
salmonids with reference to current estimates of energy and 10.1016/[Link].2021.12.034
protein requirements. Aquaculture, 100(1e3), 107e123. https:// Føre, M., Alfredsen, J. A., & Gronningsater, A. (2011).
[Link]/10.1016/0044-8486(92)90353-M Development of two telemetry-based systems for
Cho, C. Y., & Bureau, D. P. (1998). Development of bioenergetic monitoring the feeding behaviour of Atlantic salmon (Salmo
models and the Fish-PrFEQ software to estimate production, salar L.) in aquaculture sea-cages. Computers and Electronics in
feeding ration and waste output in aquaculture. Aquatic Living Agriculture, 76(2), 240e251. [Link]
Resources, 11(4), 199e210. [Link] [Link].2011.02.003
7440(98)89002-5 Føre, M., Frank, K., Norton, T., Svendsen, E., Alfredsen, J. A.,
Chowdhury, M. A. K., Siddiqui, S., Hua, K., & Bureau, D. P. (2013). Dempster, T., Eguiraun, T., Watson, W., Stahl, A., Sunde, L. M.,
Bioenergetics-based factorial model to determine feed Schellewald, C., Skøien, K. R., Alver, M. O., & Berckmans, D.
requirement and waste output of tilapia produced under (2018). Precision fish farming: A new framework to improve
commercial conditions. Aquaculture, 410e411(2), 138e147. production in aquaculture. Biosystems Engineering, 173,
[Link] 176e193. [Link]
Coffin, M. R., Courtenay, S. C., Pater, C. C., & van den Heuvel, M. R. Frederick, C. A., Coyle, S. D., Durborow, R. M., Bright, L. A., &
(2018). An empirical model using dissolved oxygen as an Tidwell, J. H. (2016). Effect of taurine supplementation on
indicator for eutrophication at a regional scale. Marine Pollution growth response and body composition of largemouth bass.
Bulletin, 133, 261e270. [Link] North American Journal of Aquaculture, 78(2), 107e112. https://
[Link].2018.05.041 [Link]/10.1080/15222055.2015.1084070
212 b i o s y s t e m s e n g i n e e r i n g 2 2 7 ( 2 0 2 3 ) 1 9 5 e2 1 6

Gong, Y., Yang, F., Hu, J., Liu, C., Liu, H., Han, D., Jin, J., Yang, Y., Irby, I. D., Friedrichs, M. A., Da, F., & Hinson, K. E. (2018). The
Zhu, X., Yi, J., & Xie, S. (2019). Effects of dietary yeast competing impacts of climate change and nutrient reductions
hydrolysate on the growth, antioxidant response, immune on dissolved oxygen in Chesapeake Bay. Biogeosciences, 15(9),
response and disease resistance of largemouth bass 2649e2668. [Link]
(Micropterus salmoides). Fish & Shellfish Immunology, 94, 548e557. Iwama, G. K., & Tautz, A. F. (1981). A simple growth model for
[Link] salmonids in hatcheries. Canadian Journal of Fisheries and
Gou, S., Chen, N., Xu, X., Liu, Z., & Yin, J. (2015). Effects of dietary Aquatic Sciences, 38(6), 649e656. [Link]
digestible starch levels on growth performance, body Jiang, Y., Zhao, P. F., Lin, S. M., Tang, R. J., Chen, Y. J., & Luo, L.
composition, and non-specific immunological index of (2018). Partial substitution of soybean meal with fermented
largemouth bass (Micropterus salmoides). Journal of Fisheries of soybean residue in diets for juvenile largemouth bass.
China, 39(10), 1499e1510. [Link] Micropterus salmoides. Aquaculture Nutrition, 24(4), 1213e1222.
jfc.20150309746 [Link]
Gu, J., Liang, H., Ge, X., Xia, D., Pan, L., Mi, H., & Ren, M. (2022). A Kaushik, S. J. (1998). Nutritional bioenergetics and estimation of
study of the potential effect of yellow mealworm (Tenebrio waste production in non-salmonids. Aquatic Living Resources,
molitor) substitution for fish meal on growth, immune and 11(4), 211e217. [Link]
antioxidant capacity in juvenile largemouth bass (Micropterus Kobayashi, M., Msangi, S., Batka, M., Vannuccini, S., Dey, M. M., &
salmoides). Fish & Shellfish Immunology, 120, 214e221. https:// Anderson, J. L. (2015). Fish to 2030: The role and opportunity
[Link]/10.1016/[Link].2021.11.024 for aquaculture. Aquaculture Economics and Management, 19(3),
Guo, J. L., Kuang, W. M., Zhong, Y. F., Zhou, Y. L., Chen, Y. J., & 282e300. [Link]
Lin, S. M. (2020). Effects of supplemental dietary bile acids on Lagardere, J. P. (2000). Feeding sounds of turbot (Scophthalmus
growth, liver function and immunity of juvenile largemouth maximus) and their potential use in the control of food supply in
bass (Micropterus salmoides) fed high-starch diet. Fish & Shellfish aquaculture I. Spectrum analysis of the feeding sounds.
Immunology, 97, 602e607. [Link] Aquaculture. [Link]
[Link].2019.12.087 Lagardere, J. P., Mallekh, R., & Mariani, A. (2004). Acoustic
Guo, J. L., Zhou, Y. L., Zhao, H., Chen, W. Y., Chen, Y. J., & Lin, S. M. characteristics of two feeding modes used by brown trout
(2019). Effect of dietary lipid level on growth, lipid metabolism (Salmo trutta), rainbow trout (Oncorhynchus mykiss) and turbot
and oxidative status of largemouth bass, Micropterus (Scophthalmus maximus). Aquaculture, 240, 607e616. https://
salmoides. Aquaculture, 506, 394e400. [Link] [Link]/10.1016/[Link].2004.01.033
[Link].2019.04.007 Liang, C., Zhao, X., Jiao, L., Shen, Y., Luo, J., Zhu, T., Zhao, W.,
He, M. (2020). Study on the effects of fermented soybean meal in fish Gen, Z., Zhou, Q., & Jin, M. (2022). Effects of different lipid
meal replacement and the improvement of glutamine and sources on growth performance, fatty acids composition in
clostridium butyrium on its effectiveness in the diet of largemouth tissue and expression of genes related to lipid metabolism
bass (Micropterus samoides) (in Chinese). Shanghai: Shanghai in largemouth bass (Micropterus salmoides). Aquaculture
Ocean University. [Link] Reports, 23, Article 101013. [Link]
[Link].2020.000339. The Thesis for the Doctorate. [Link].2022.101013
He, X. G., & Hou, J. (2021). Research progress on pond Juanyang Libralato, S., & Solidoro, C. (2008). A bioenergetic growth model
mode, 2021 Journal of Huazhong Agriculture University, 40(3), for comparing Sparus aurata's feeding experiments. Ecological
21e29 (in Chinese). Modelling, 214(2e4), 325e337. [Link]
He, M., Li, X., Poolsawat, L., Guo, Z., Yao, W., Zhang, C., & Leng, X. [Link].2008.02.024
(2020a). Effects of fish meal replaced by fermented soybean Li, S., Dai, M., Qiu, H., & Chen, N. (2021a). Effects of fishmeal
meal on growth performance, intestinal histology and replacement with composite mixture of shrimp hydrolysate
microbiota of largemouth bass (Micropterus salmoides). and plant proteins on growth performance, feed utilization,
Aquaculture Nutrition, 26(4), 1058e1071. [Link] and target of rapamycin pathway in largemouth bass,
anu.13064 Micropterus salmoides. Aquaculture, 533, Article 736185.
He, G., Sun, H., Liao, R., Wei, Y., Zhang, T., Chen, Y., & Lin, S. [Link]
(2022). Effects of herbal extracts (Foeniculum vulgare and Li, S., Ding, G., Song, F., Sang, C., Wang, A., & Chen, N. (2020a).
Artemisia annua) on growth, liver antioxidant capacity, Comparison of dehulled, fermented and enzyme-treated
intestinal morphology and microorganism of juvenile soybean meal in diets for largemouth bass, Micropterus
largemouth bass, Micropterus salmoides. Aquaculture salmoides: Effects on growth performance, feed utilization,
Reports, 23, Article 101081. [Link] immune response and intestinal morphology. Animal Feed
[Link].2022.101081 Science and Technology, 267, Article 114548. [Link]
He, M., Yu, Y., Li, X., Poolsawat, L., Yang, P., Bian, Y., Guo, Z., & 10.1016/[Link].2020.114548
Leng, X. (2020b). An evaluation of replacing fish meal with Li, S., Ding, G., Wang, A., Sang, C., & Chen, N. (2019a).
fermented soybean meal in the diets of largemouth bass Replacement of fishmeal by chicken plasma powder in diets
(Micropterus salmoides): Growth, nutrition utilization and for largemouth bass (Micropterus salmoides): Effects on growth
intestinal histology. Aquaculture Research, 51(10), 4302e4314. performance, feed utilization and health status. Aquaculture
[Link] Nutrition, 25(6), 1431e1439. [Link]
Houlihan, D., Boujard, T., & Jobling, M. (Eds.). (2008). Food intake in Li, S., Lian, X., Chen, N., Wang, M., & Sang, C. (2018). Effects of
fish. John Wiley & Sons. dietary vitamin E level on growth performance, feed
Huang, D., Wu, Y., Lin, Y., Chen, J., Karrow, N., Ren, X., & Wang, Y. utilization, antioxidant capacity and nonspecific immunity of
(2017). Dietary protein and lipid requirements for juvenile largemouth bass, Micropterus salmoides. Aquaculture nutrition,
largemouth bass, Micropterus salmoides. Journal of the World 24(6), 1679e1688. [Link]
Aquaculture Society, 48(5), 782e790. [Link] Lin, S. M., Jiang, Y., Chen, Y. J., Luo, L., Doolgindachbaporn, S., &
jwas.12417 Yuangsoi, B. (2017). Effects of Astragalus polysaccharides (APS)
Huan, J., Li, H., Wu, F., & Cao, W. (2020). Design of water quality and chitooligosaccharides (COS) on growth, immune response
monitoring system for aquaculture ponds based on NB-IoT. and disease resistance of juvenile largemouth bass,
Aquacultural Engineering, 90, Article 102088. [Link] Micropterus salmoides. Fish & Shellfish Immunology, 70, 40e47.
10.1016/[Link].2020.102088 [Link]
 b i o s y s t e m s e n g i n e e r i n g 2 2 7 ( 2 0 2 3 ) 1 9 5 e2 1 6 213

Lin, S. M., Shi, C. M., Mu, M. M., Chen, Y. J., & Luo, L. (2018). Effect Miller, K. K., Rossi, W., Jr., & Habte-Tsion, H. M. (2021).
of high dietary starch levels on growth, hepatic glucose Assessment of total dietary phosphorus requirement of
metabolism, oxidative status and immune response of juvenile largemouth bass, Micropterus salmoides, using
juvenile largemouth bass, Micropterus salmoides. Fish & soybean meal-based diets: Effects on production
Shellfish Immunology, 78, 121e126. [Link] performance, tissue mineralization, physiological
[Link].2018.04.046 parameters in plasma and intestine and expression of head-
Li, S., Sang, C., Wang, A., Zhang, J., & Chen, N. (2019b). Effects of kidney genes. Aquaculture Nutrition, 27(1), 116e128. https://
dietary carbohydrate sources on growth performance, [Link]/10.1111/anu.13169
glycogen accumulation, insulin signaling pathway and Millot, S., & Be gout, M. L. (2009). Individual fish rhythm directs
hepatic glucose metabolism in largemouth bass, Micropterus group feeding: A case study with sea bass juveniles
salmoides. Aquaculture, 513, Article 734391. [Link] (Dicentrarchus labrax) under self-demand feeding conditions.
10.1016/[Link].2019.734391 Aquatic Living Resources, 22(3), 363e370. [Link]
Liu, Y., Lei, M., Victor, H., Wang, Z., Yu, C., Zhang, G., & Wang, Y. alr/2009048
(2022). The optimal feeding frequency for largemouth bass Ni, M., Liu, M., Lou, J., Mi, G., Yuan, J., & Gu, Z. (2021). Stocking
(Micropterus salmoides) reared in pond and in-pond-raceway. density alters growth performance, serum biochemistry,
Aquaculture, 548, Article 737464. [Link] digestive enzymes, immune response, and muscle quality of
[Link].2021.737464 largemouth bass (Micropterus salmoides) in in-pond raceway
Liu, Z. Y., Li, X., Fan, L. Z., Lu, H. D., Liu, L., & Liu, Y. (2014). system. Fish Physiology and Biochemistry, 47(4), 1243e1255.
Measuring feeding activity of fish in RAS using computer [Link]
vision. Aquacultural Engineering, 60, 20e27. [Link] Rahman, M. M., Li, X., Sharifuzzaman, S. M., He, M., Poolsawat, L.,
10.1016/[Link].2014.03.005 Yang, H., & Leng, X. (2021). Dietary threonine requirement of
Liu, Y., Lu, Q., Xi, L., Gong, Y., Su, J., Han, D., Zhang, Z., Liu, H., juvenile largemouth bass, Micropterus salmoides. Aquaculture,
Jin, J., Yang, Y., Zhu, X., & Xie, S. (2021). Effects of replacement 543, Article 736884. [Link]
of dietary fishmeal by cottonseed protein concentrate on [Link].2021.736884
growth performance, liver health, and intestinal histology of Reis, J., Peixoto, S., Soares, R., Rhodes, M., Ching, C., & Davis, D. A.
largemouth bass (Micropterus salmoides). Frontiers in Physiology, (2022). Passive acoustic monitoring as a tool to assess feed
2308. [Link] response and growth of shrimp in ponds and research
Liu, X., Sha, Z., Wang, C., Li, D., & Bureau, D. P. (2018). A web- systems. Aquaculture, 546, Article 737326. [Link]
based combined nutritional model to precisely predict growth, 10.1016/[Link].2021.737326
feed requirement and waste output of Gibel carp (Carassius Ren, X., Huang, D., Wu, Y. B., Jiang, D. L., Li, P., Chen, J. M., &
auratus gibelio) in aquaculture operations. Aquaculture, 492, Wang, Y. (2020). Gamma ray irradiation improves feather meal
335e348. [Link] as a fish meal alternate in largemouth bass Micropterus
Li, D., Wang, Z., Wu, S., Miao, Z., Du, L., & Duan, Y. (2020). salmoides diet. Animal Feed Science and Technology, 269, Article
Automatic recognition methods of fish feeding behavior in 114647. [Link]
aquaculture: A review. Aquaculture, 528, Article 735508. https:// Ren, J. S., Jin, X., Yang, T., Kooijman, S. A., & Shan, X. (2020). A
[Link]/10.1016/[Link].2020.735508 dynamic energy budget model for small yellow croaker
Li, X., Zheng, S., Cheng, K., Ma, X., & Wu, G. (2021b). Use of Larimichthys polyactis: Parameterisation and application in its
alternative protein sources for fishmeal replacement in the main geographic distribution waters. Ecological Modelling, 427,
diet of largemouth bass (Micropterus salmoides). Part II: Effects Article 109051. [Link]
of supplementation with methionine or taurine on growth, [Link].2020.109051
feed utilization, and health. Amino Acids, 53(1), 49e62. https:// Ren, X., Wang, Y., Chen, J. M., Wu, Y. B., Huang, D., Jiang, D. L., &
[Link]/10.1007/S00726-020-02922-4 Li, P. (2018). Replacement of fishmeal with a blend of poultry
Li, X., Zheng, S., Ma, X., Cheng, K., & Wu, G. (2020b). Effects of byproduct meal and soybean meal in diets for largemouth
dietary starch and lipid levels on the protein retention and bass. Micropterus salmoides. Journal of the World Aquaculture
growth of largemouth bass (Micropterus salmoides). Amino Acids, Society, 49(1), 155e164. [Link]
52(6), 999e1016. [Link] Rixen, T., Baum, A., Sepryani, H., Pohlmann, T., Jose, C., &
Li, X., Zheng, S., Ma, X., Cheng, K., & Wu, G. (2021c). Use of Samiaji, J. (2010). Dissolved oxygen and its response to
alternative protein sources for fishmeal replacement in the eutrophication in a tropical black water river. Journal of
diet of largemouth bass (Micropterus salmoides). Part I: Effects of Environmental Management, 91(8), 1730e1737. [Link]
poultry by-product meal and soybean meal on growth, feed 10.1016/[Link].2010.03.009
utilization, and health. Amino Acids, 53(1), 33e47. https:// Romano, N., Egnew, N., Quintero, H., Kelly, A., & Sinha, A. K.
[Link]/10.1007/s00726-020-02920-6 (2020). The effects of water hardness on the growth, metabolic
Lu, Q., Xi, L., Liu, Y., Gong, Y., Su, J., Han, D., Yang, Y., Jin, J., indicators and stress resistance of largemouth bass
Liu, H., Zhu, X., & Xie, S. (2021). Effects of dietary inclusion of Micropterus salmoides. Aquaculture, 527, Article 735469.
Clostridium autoethanogenum protein on the growth [Link]
performance and liver health of largemouth bass (Micropterus Romano, N., Fischer, H., Rubio-Benito, M. M., Overtuf, K.,
salmoides). Frontiers in Marine Science. [Link] Sinha, A. K., & Kumar, V. (2022). Different dietary
fmars.2021.764964 combinations of high/low starch and fat with or without bile
Mabroke, R. S., Zidan, A. E. N. F., Tahoun, A. A., Mola, H. R., Abo- acid supplementation on growth, liver histopathology, gene
State, H., & Suloma, A. (2021). Feeding frequency affect feed expression and fatty acid composition of largemouth bass,
utilization of tilapia under biofloc system condition during Micropterus salmoides. Comparative Biochemistry and Physiology
nursery phase. Aquaculture Reports, 19, Article 100625. https:// Part A: Molecular & Integrative Physiology, 266, Article 111157.
[Link]/10.1016/[Link].2021.100625 [Link]
Ma, S., Liang, X., Chen, P., Wang, J., Gu, X., Qin, Y., Blecker, C., & Rossi, W., Jr., Allen, K. M., Habte-Tsion, H. M., & Meesala, K. M.
Xue, M. (2022). A new single-cell protein from Clostridium (2021). Supplementation of glycine, prebiotic, and nucleotides
autoethanogenum as a functional protein for largemouth bass in soybean meal-based diets for largemouth bass (Micropterus
(Micropterus salmoides). Animal Nutrition, 10, 99e110. https:// salmoides): Effects on production performance, whole-body
[Link]/10.1016/[Link].2022.04.005 nutrient composition and retention, and intestinal
214 b i o s y s t e m s e n g i n e e r i n g 2 2 7 ( 2 0 2 3 ) 1 9 5 e2 1 6

histopathology. Aquaculture, 532, Article 736031. https:// metabolism through modulating nutrient-related pathways in
[Link]/10.1007/s10695-019-00636-3 largemouth bass (Micropterus salmoides). Aquaculture Reports,
Schmidt, W., Raymond, D., Parish, D., Ashton, I. G., Miller, P. I., 20, Article 100642. [Link]
Campos, C. J., & Shutler, J. D. (2018). Design and operation of a Watanabe, T. (2002). Strategies for further development of aquatic
low-cost and compact autonomous buoy system for use in feeds. Fisheries Science, 68(2), 242e252. [Link]
coastal aquaculture and water quality monitoring. j.1444-2906.2002.00418.x
Aquacultural Engineering, 80, 28e36. [Link] Wu, B., Huang, L., Chen, J., Zhang, Y., Chen, X., Wu, C., Deng, X.,
[Link].2017.12.002 Gao, J., & He, J. (2021). Effects of feeding frequency on growth
Shi, C. M., Zhao, H., Zhai, X. L., Chen, Y. J., & Lin, S. M. (2019). performance, feed intake, metabolism and expression of fgf21
Linseed oil can decrease liver fat deposition and improve in grass carp (Ctenopharyngodon idellus). Aquaculture, 545,
antioxidant ability of juvenile largemouth bass, Micropterus 737196. [Link]
salmoides. Fish Physiology and Biochemistry, 45(5), 1513e1521. Wu, T. H., Huang, Y. I., & Chen, J. M. (2015). Development of an
[Link] adaptive neural-based fuzzy inference system for feeding
Smith, D. V., & Tabrett, S. (2013). The use of passive acoustics to decision-making assessment in silver perch (Bidyanus
measure feed consumption by Penaeus monodon (giant tiger bidyanus) culture. Aquacultural Engineering, 66, 41e51. https://
prawn) in cultured systems. Aquacultural Engineering, 57, [Link]/10.1016/[Link].2015.02.001
38e47. [Link] Wu, Y., Ma, H., Wang, X., & Ren, X. (2021a). Taurine
Song, M. Q., Shi, C. M., Lin, S. M., Chen, Y. J., Shen, H. M., & Luo, L. supplementation increases the potential of fishmeal
(2018). Effect of starch sources on growth, hepatic glucose replacement by soybean meal in diets for largemouth bass
metabolism and antioxidant capacity in juvenile largemouth Micropterus salmoides. Aquaculture Nutrition, 27(3), 691e699.
bass, Micropterus salmoides. Aquaculture, 490, 355e361. [Link]
[Link] Wu, Y., Wang, Y., Ren, X., Huang, D., Si, G., & Chen, J. (2021b).
Sun, M., Hassan, S. G., & Li, D. (2016). Models for estimating feed Replacement of fish meal with gamma-ray irradiated soybean
intake in aquaculture: A review. Computers and Electronics in meal in the diets of largemouth bass. Micropterus salmoides.
Agriculture, 127, 425e438. [Link] Aquaculture Nutrition, 27(4), 977e985. [Link]
[Link].2016.06.024 anu.13239
Vis, H. V. D., Kolarevic, J., Stien, L. H., Kristiansen, T. S., Xiao, W. W. (2012). Effect of dietary lipid and protein levels on growth,
Gerritzen, M., Braak, K. V. D., Abbink, W., Sæther, B. S., & body composition, non-specific immunity and hematological
Noble, C. (2020). Welfare of farmed fish in different production parameters of largemouth bass (Micropterus salmoides) (in
systems and operations. The Welfare of Fish, 20, 323e361. Chinese). Shanghai: Shanghai Ocean University, Master of
[Link] degree.
Vittinghoff, E., Glidden, D. V., Shiboski, S. C., & McCulloch, C. E. Xie, X., Wang, J., Guan, Y., Xing, S., Liang, X., Xue, M., Wang, J.,
(2005). Regression methods in biostatistics: Linear, logistic, survival, Chang, Y., & Leclercq, E. (2022). Cottonseed protein
and repeated measures models. New York, USA: Springer. concentrate as fishmeal alternative for largemouth bass
Wang, P. (2020). Study on the requirement of phosphorus and iron for (Micropterus salmoides) supplemented a yeast-based
Micropterus salmoides (in Chinese). Shanghai: Shanghai Ocean paraprobiotic: Effects on growth performance, gut health and
University. [Link] microbiome. Aquaculture, 551, Article 737898. [Link]
Master of degree. 10.1016/[Link].2022.737898
Wang, L., Cui, Z., Ren, X., Li, P., & Wang, Y. (2021a). Growth Xie, S., Yin, P., Tian, L., Liu, Y., & Niu, J. (2020). Lipid metabolism
performance, feed cost and environmental impact of and plasma metabolomics of juvenile largemouth bass
largemouth bass Micropterus salmoides fed low fish meal diets. Micropterus salmoides were affected by dietary oxidized fish oil.
Aquaculture Reports, 20, Article 100757. [Link] Aquaculture, 522, Article 735158. [Link]
[Link].2021.100757 [Link].2020.735158
Wang, A., Li, S., Liu, Y., Han, Z., & Chen, N. (2020a). The inclusion Xi, L., Lu, Q., Liu, Y., Su, J., Chen, W., Gong, Y., Han, D., Yang, Y.,
of fenugreek seed extract aggravated hepatic glycogen Zhang, Z., Jin, J., Liu, H., Zhu, X., & Xie, S. (2022). Effects of fish
accumulation through reducing the expression of genes meal replacement with Chlorella meal on growth performance,
involved in insulin pathway and glycolysis in largemouth pigmentation, and liver health of largemouth bass (Micropterus
bass, Micropterus salmoides. Aquaculture, 528, Article 735567. salmoides). Animal Nutrition, 10, 26e40. [Link]
[Link] [Link].2022.03.003
Wang, Y., Ni, J., Nie, Z., Gao, J., Sun, Y., Shao, N., Li, Q., Hu, J., & Xu, X., Li, X., Xu, Z., Yao, W., & Leng, X. (2021). Dietary Azomite, a
Xu, G. (2020c). Effects of stocking density on growth, serum natural trace mineral complex, improved the growth,
parameters, antioxidant status, liver and intestine histology immunity response, intestine health and resistance against
and gene expression of largemouth bass (Micropterus bacterial infection in largemouth bass (Micropterus salmoides).
salmoides) farmed in the in-pond raceway system. Aquaculture Fish & Shellfish Immunology, 108, 53e62. [Link]
Research, 51(12), 5228e5240. [Link] [Link].2020.11.016
Wang, Y., Xie, S., Nie, Z., Li, Q., Sun, Y., Shao, N., Gao, J., Hu, J., Yadav, A. K., Rossi, W., Jr., Habte-Tsion, H. M., & Kumar, V. (2020).
Xu, P., & Xu, G. (2020d). Optimum feeding frequency of Impacts of dietary eicosapentaenoic acid (EPA) and
juvenile largemouth bass (Micropterus salmoides) reared in in- docosahexaenoic acid (DHA) level and ratio on the growth,
pond raceway recirculating culture system. Fish Physiology and fatty acids composition and hepatic-antioxidant status of
Biochemistry, 46(6), 2197e2212. [Link] largemouth bass (Micropterus salmoides). Aquaculture, 529,
020-00866-w Article 735683. [Link]
Wang, W., Xu, Y., Chi, S., Yang, P., Mai, K., & Song, F. (2020b). [Link].2020.735683
Dietary lysine regulates body growth performance via the Yamamoto, T., Shima, T., Unuma, T., Shiraishi, M., Akiyama, T., &
nutrient-sensing signaling pathways in largemouth bass Tabata, M. (2000). Voluntary intake of diets with varying
(Micropterus salmoides). Frontiers in Marine Science, 7, Article digestible energy contents and energy sources, by juvenile
595682. [Link] rainbow trout Oncorhynchus mykiss, using self-feeders.
Wang, W., Yang, P., He, C., Chi, S., Li, S., Mai, K., & Song, F. (2021b). Fisheries Science, 66(3), 528e534. [Link]
Effects of dietary methionine on growth performance and 2906.2000.00083.x
 b i o s y s t e m s e n g i n e e r i n g 2 2 7 ( 2 0 2 3 ) 1 9 5 e2 1 6 215

Yang, H., Bian, Y., Huang, L., Lan, Q., Ma, L., Li, X., & Leng, X. Yun, B., Xue, M., Wang, J., Fan, Z. Y., Wu, X. F., Zheng, Y. H., &
(2022a). Effects of replacing fish meal with fermented soybean Qin, Y. C. (2013). Effects of lipid sources and lipid peroxidation on
meal on the growth performance, intestinal microbiota, feed intake, growth, and tissue fatty acid compositions of
morphology and disease resistance of largemouth bass largemouth bass (Micropterus salmoides). Aquaculture International,
(Micropterus salmoides). Aquaculture Reports, 22, Article 100954. 21(1), 97e110. [Link]
[Link] Yusuf, A., Huang, X., Chen, N., Li, S., Apraku, A., Wang, W., &
Yang, K., Fan, Q., Zhang, L., Li, B., Gao, Y., Zeng, K., Wang, Q., David, M. A. (2021). Growth and metabolic responses of
Zhu, S., & Fang, G. (2015). Effect of dissolved oxygen levels on juvenile largemouth bass (Micropterus salmoides) to dietary
growth performance, energy budget and antioxidant vitamin c supplementation levels. Aquaculture, 534, Article
responses of yellow catfish, Pelteobagrus fulvidraco 736243. [Link]
(Richardson). Aquaculture Research, 46(8), 2025e2033. https:// Zhang, Q., Liang, H., Longshaw, M., Wang, J., Ge, X., Zhu, J., Li, S., &
[Link]/10.1111/are.12359 Ren, M. (2022a). Effects of replacing fishmeal with methanotroph
Yang, P., Yang, W., He, M., Li, X., & Leng, X. J. (2020). Dietary (Methylococcus capsulatus, Bath) bacteria meal (FeedKind®) on
synbiotics improved the growth, feed utilization and intestinal growth and intestinal health status of juvenile largemouth bass
structure of largemouth bass (Micropterus salmoides) juvenile. (Micropterus salmoides). Fish & Shellfish Immunology, 122, 298e305.
Aquaculture Nutrition, 26(2), 590e600. [Link] [Link]
anu.13020 Zhang, W., Tan, B., Liu, K., Dong, X., Yang, Q., Chi, S., Liu, H.,
Yang, P., Yao, W., Wang, Y., Li, M., Li, X., & Leng, X. (2022b). Zhang, S., & Wang, H. (2019). Effects of different dietary lipids
Dietary effects of fish meal substitution with Clostridium on growth, body composition and lipid metabolism-related
autoethanogenum on flesh quality and metabolomics of enzymes and genes in juvenile largemouth bass. Micropterus
largemouth bass (Micropterus salmoides). Aquaculture Reports, salmoides. Aquaculture Nutrition, 25(6), 1318e1326. https://
23, Article 101012. [Link] [Link]/10.1111/anu.12952
Yang, S. Q., Zheng, H. W., Sun, Y., & Zhu, A. Y. (2019). Effects of Zhang, Y. M., Xie, S. W., Guo, T. Y., Liu, Z. L., Fang, H. H., Zheng, L.,
ammonia exposure, temperature, and body weight on oxygen Xie, J. J., Tian, L. X., Liu, Y. J., & Jin, N. (2020). High dietary
consumption rate and asphyxiation point of juvenile starch inclusion impairs growth and antioxidant status, and
largemouth bass Micropterus Salmoides. Oceanologia et alters liver organization and intestinal microbiota in
Limnologia Sinica, 50(6), 1328e1333. [Link] largemouth bass. Micropterus salmoides. Aquaculture Nutrition,
hyhz20190600115 (in Chineses). 26(5), 1806e1821. [Link]
Yin, P., Xie, S., Liu, Z., Huo, Y., Guo, T., Fang, H., Zhang, Y., Liu, Y., Zhang, Y., Zhang, Y., Ke, T., Shi, B., Huang, L., Dong, Z., Guo, M.,
Niu, J., & Tian, L. (2019). Effects of dietary oxidized fish oil on Mugeni, C. S., Zhu, A., Wang, L., Xu, J., & Storebakken, T.
growth performance, antioxidant defense system, (2022b). Hydrolysates of whole forage-fish and Pacific krill are
apoptosis and mitochondrial function of useful to reduce fish meal in practical diets for largemouth
juvenile largemouth bass (Micropterus salmoides). Aquaculture, bass (Micropterus salmoides), and dietary fish hydrolysate
500, 347e358. [Link] suppresses expressions of intestinal oligopeptide transporter
[Link].2018.09.009 and taurine transporter genes. Aquaculture Reports, 25, Article
Yin, P., Xie, S., Zhuang, Z., Fang, H., Tian, L., Liu, Y., & Niu, J. 101203. [Link]
(2021a). Chlorogenic acid improves health in juvenile Zhao, L., Liao, L., Tang, X., Liang, J., Liu, Q., Luo, W., Adam, A. A.,
largemouth bass (Micropterus salmoides) fed high-fat diets: Luo, J., Li, Z., Yang, S., & Rahimnejad, S. (2022a). High-
Involvement of lipid metabolism, antioxidant ability, carbohydrate diet altered conversion of metabolites, and
inflammatory response, and intestinal integrity. Aquaculture, deteriorated health in juvenile largemouth bass. Aquaculture,
545, Article 737169. [Link] 549, Article 737816. [Link]
[Link].2021.737169 [Link].2021.737816
Yin, P., Xie, S., Zhuang, Z., He, X., Tang, X., Tian, L., Liu, Y., & Zhao, Y., Yang, C., Zhu, X. X., Feng, L., Liu, Y., Jiang, W. D., Wu, P.,
Niu, J. (2021b). Dietary supplementation of bile acid attenuate Huang, X. L., Chen, D. F., Yang, S. Y., Luo, W., Zhang, J. X.,
adverse effects of high-fat diet on growth performance, Li, S. W., Diao, H., Wei, X. L., Zhou, M. J., Zhou, X. Q., & Jiang, J.
antioxidant ability, lipid accumulation and intestinal health in (2022b). Dietary methionine hydroxy analogue
juvenile largemouth bass (Micropterus salmoides). Aquaculture, supplementation benefits on growth, intestinal antioxidant
531, Article 735864. [Link] status and microbiota in juvenile largemouth bass
[Link].2020.735864 Micropterus salmoides. Aquaculture, 556, Article 738279.
Yuan, Y., Chen, Y. J., Liu, Y. J., Yang, H. J., Liang, G. Y., & Tian, L. X. [Link]
(2014). Dietary high level of vitamin premix can eliminate Zhong, L., Liu, H., Zhang, H., Zhang, W., Li, M., Huang, Y., Yao, J.,
oxidized fish oileinduced oxidative damage and loss of Huang, X., Geng, Y., Chen, D., Ouyang, P., Yang, S., Luo, W., &
reducing capacity in juvenile largemouth bass (Micropterus Yin, L. (2022). High starch in diet leads to disruption of hepatic
salmoides). Aquaculture Nutrition, 20(2), 109e117. https:// glycogen metabolism and liver fibrosis in largemouth Bass
[Link]/10.1111/anu.12057 (Micropterus salmoides), which is mediated by the PI3K/Akt
Yu, H. H., Liang, X. F., Chen, P., Wu, X. F., Zheng, Y. H., Luo, L., signaling pathway. Frontiers in Physiology, 1023. [Link]
Qin, Y. C., Long, X. C., & Xue, M. (2019). Dietary 10.3389/fphys.2022.880513
supplementation of Grobiotic®-A increases short-term Zhong, Y. F., Shi, C. M., Zhou, Y. L., Chen, Y. J., Lin, S. M., & Tang, R. J.
inflammatory responses and improves long-term growth (2020). Optimum dietary fiber level could improve growth,
performance and liver health in largemouth bass (Micropterus plasma biochemical indexes and liver function of largemouth
salmoides). Aquaculture, 500, 327e337. [Link] bass, Micropterus salmoides. Aquaculture, 518, Article 734661.
[Link].2018.10.033 [Link]
Yu, H., Liang, H., Ge, X., Zhu, J., Wang, Y., Ren, M., & Chen, X. Zhou, H., Chen, N., Qiu, X., Zhao, M., & Jin, L. (2012). Arginine
(2022). Dietary chlorella (Chlorella vulgaris) supplementation requirement and effect of arginine intake on immunity in
effectively improves body colour, alleviates muscle largemouth bass. Micropterus salmoides. Aquaculture Nutrition,
inflammation and inhibits apoptosis in largemouth bass 18(1), 107e116. [Link]
(Micropterus salmoides). Fish & Shellfish Immunology, 127, Zhou, Y. L., Guo, J. L., Tang, R. J., Ma, H. J., Chen, Y. J., & Lin, S. M.
140e147. [Link] (2020). High dietary lipid level alters the growth, hepatic
216 b i o s y s t e m s e n g i n e e r i n g 2 2 7 ( 2 0 2 3 ) 1 9 5 e2 1 6

metabolism enzyme, and anti-oxidative capacity in juvenile Zhu, Y., Chen, Y., Liu, Y., Yang, H., Liang, G., & Tian, L. (2012).
largemouth bass Micropterus salmoides. Fish Physiology and Effect of dietary selenium level on growth performance, body
Biochemistry, 46(1), 125e134. [Link] composition and hepatic glutathione peroxidase activities of
019-00705-7 largemouth bass Micropterus salmoide. Aquaculture Research,
Zhou, C., Lin, K., Xu, D., Chen, L., Guo, Q., Sun, C., & Yang, X. 43(11), 1660e1668. [Link]
(2018). Near infrared computer vision and neuro-fuzzy model- 2109.2011.02972.x
based feeding decision system for fish in aquaculture. Zhu, J., Du, R. Y., Liu, Q. Q., Luo, L., Lin, S. M., Zhang, H. Q., &
Computers and Electronics in Agriculture, 146, 114e124. https:// Chen, Y. J. (2022a). Transcriptome analysis provides insights
[Link]/10.1016/[Link].2018.02.006 into the molecular mechanism of hepatocyte apoptosis in
Zhou, Z., Xie, S., Lei, W., Zhu, X., & Yang, Y. (2005). A bioenergetic response to feeding restriction in juvenile largemouth bass
model to estimate feed requirement of Gibel carp, Carassius Micropterus salmoides. Aquaculture, 548, Article 737550.
auratus gibelio. Aquaculture, 248(1e4), 287e297. [Link] [Link]
10.1016/[Link].2005.03.011 Zhu, S., Gao, W., Wen, Z., Chi, S., Shi, Y., Hu, W., & Tan, B. (2022b).
Zhou, C., Xu, D. M., Lin, K., Sun, C. H., & Yang, X. T. (2017). Partial substitution of fish meal by Clostridium autoethanogenum
Intelligent feeding control methods in aquaculture with an protein in the diets of juvenile largemouth bass (Micropterus
emphasis on fish: A review. Reviews in Aquaculture, 10(4), salmoides). Aquaculture Reports, 22, Article 100938. https://
975e993. [Link] [Link]/10.1016/[Link].2021.100938