DEVELOPMENT OF AN AUTOMATIC WEIGHING PLATFORM FOR MONITORING BODYWEIGHT OF BROILER CHICKENS IN COMMERCIAL PRODUCTION

Danni ZHOU, Yi ZHOU, Pengguang HE, Lin YU, Jinming PAN, Lilong CHAI, Hongjian LIN

Front. Agr. Sci. Eng. ›› 2023, Vol. 10 ›› Issue (3) : 363-373.

PDF(6181 KB)
Front. Agr. Sci. Eng. All Journals
PDF(6181 KB)
Front. Agr. Sci. Eng. ›› 2023, Vol. 10 ›› Issue (3) : 363-373. DOI: 10.15302/J-FASE-2023510
RESEARCH ARTICLE
RESEARCH ARTICLE

DEVELOPMENT OF AN AUTOMATIC WEIGHING PLATFORM FOR MONITORING BODYWEIGHT OF BROILER CHICKENS IN COMMERCIAL PRODUCTION

Author information +
History +

Highlights

● An automatic weighing system for monitoring bodyweight of broilers was developed.

● The new system was compared to the established live-bird sales weighing system data and tested in various conditions.

● The system demonstrated superior accuracy and stability for commercial houses.

Abstract

Bodyweight is a key indicator of broiler production as it measures the production efficiency and indicates the health of a flock. Currently, broiler weight (i.e., bodyweight) is primarily weighed manually, which is time-consuming and labor-intensive, and tends to create stress in birds. This study aimed to develop an automatic and stress-free weighing platform for monitoring the weight of floor-reared broiler chickens in commercial production. The developed system consists of a weighing platform, a real-time communication terminal, computer software and a smart phone applet user-interface. The system collected weight data of chickens on the weighing platform at intervals of 6 s, followed by filtering of outliers and repeating readings. The performance and stability of this system was systematically evaluated under commercial production conditions. With the adoption of data preprocessing protocol, the average error of the new automatic weighing system was only 10.3 g, with an average accuracy 99.5% with the standard deviation of 2.3%. Further regression analysis showed a strong agreement between estimated weight and the standard weight obtained by the established live-bird sales system. The variance (an indicator of flock uniformity) of broiler weight estimated using automatic weighing platforms was in accordance with the standard weight. The weighing system demonstrated superior stability for different growth stages, rearing seasons, growth rate types (medium- and slow-growing chickens) and sexes. The system is applicable for daily weight monitoring in floor-reared broiler houses to improve feeding management, growth monitoring and finishing day prediction. Its application in commercial farms would improve the sustainability of poultry industry.

Graphical abstract

Keywords

automatic weighing / weight monitoring / floor housing / uniformity / precision poultry farming

Cite this article

Download citation ▾
Danni ZHOU, Yi ZHOU, Pengguang HE, Lin YU, Jinming PAN, Lilong CHAI, Hongjian LIN. DEVELOPMENT OF AN AUTOMATIC WEIGHING PLATFORM FOR MONITORING BODYWEIGHT OF BROILER CHICKENS IN COMMERCIAL PRODUCTION. Front. Agr. Sci. Eng., 2023, 10(3): 363‒373 https://doi.org/10.15302/J-FASE-2023510

1 INTRODUCTION

Kumquat (Fortunella spp.) is classified into the economically important true citrus fruit tree group together with Citrus and Poncirus, which belongs to the family of Rutaceae[1,2]. It is a common fruit crop and ornamental tree characterized by small, flavorful and brilliant fruit[3]. Given that the major edible part is the aromatic pericarp and rind, kumquat can provide more antioxidant and antimicrobial metabolites than Citrus spp., whose primary edible tissue is the juice vesicle[48]. Hence, it is also widely processed into succade (candied peel) and jam, as well as added to beverage, tea and cocktail as a natural flavor. Fortunella inherently possesses multiple elite agronomic traits among the citrus taxa, such as small tree size, cold and dry tolerance, short juvenility and citrus canker resistance[3,911]. More importantly, monoembryonic Hong Kong kumquat (Fortunella hindsii), which is known as a mini-citrus, has been developed as a model system for functional genomic study of citrus due to its short juvenility and sexual reproduction, which are rare features among the citrus taxa[12].
According to the classification of Swingle and Tanaka, Fortunella are classified based on their morphological and phenological characteristics: fewer locules (3–9) in each ovary and only two ovules in each locule, small tree size, and continuous flowering in summer[2,3,13]. Among Fortunella, Meiwa (F. crassifolia), Nagami (F. margarita) and Marumi (F. japonica) kumquat (the cultivated Fortunella spp.) are widely cultivated for fruit production in China, Japan, Indonesia and Malay Peninsula. Hong Kong kumquat, a wild species with the smallest hesperidium, is indigenous to southern China and mainly used for miniascape and medicine. Changshou kumquat (Fortunella obovata) is an ornamental cultivar in east Asia. Besides, calamondin (Citrus madurensis) is also regarded as a relative of Fortunella due to its analogous morphology[2,3,13,14].
Although phylogeny of citrus taxa has fascinated scientists for decades and is still a focus for research, Fortunella is considered a relative of Citrus and its phylogeny remains unresolved, possibly because few species (1–3) and samples (1–6) were used[1520]. Therefore, there are still some uncertainties concerning this genus. Tanaka[13] proposed that Fortunella should be further divided into subgenus Eufortunella (F. margarita, F. crassifolia and F. japonica) and Protocitrus (F. hindsii) due to the primitive morphological characters of F. hindsii, especially the fruit organ. Swingle speculated that F. crassifolia might originate from the hybridization between F. margarita and F. japonica considering its intermediate fruit shape (short oblong to round) between oval (F. margarita) and round (F. japonica), or a backcrossing with Citrus[2]. With the rapid development of molecular biology and genomics, numerous citrus genotypes have been demonstrated to originate from hybridization or introgression, such as sweet orange (C. sinensis), grapefruit (C. paradisi), lemon (C. limon) and lime (C. aurantiifolia)[2123], as well as calamondin (Citrus madurensis) and Changshou kumquat (F. obovata)[1517], which poses challenges to the classic taxonomy of Citrus spp. Specifically, no study has comprehensively demonstrated the phylogeny and classification of Fortunella based on a systematic collection of various germplasm.
In addition, little is known about the origin of cultivated Fortunella spp. and their relationship with the only wild Fortunella spp., Hong Kong kumquat. Fortunella is academically recognized to originate from China[1,2,11,13]. The history of kumquat cultivation in China can be traced back to Special Local Flora and Fauna in Linhai (Ying Shen, c. 250), and it was repeatedly mentioned in later Chinese literature, such as Guang Zhi (Yigong Guo, c. 270), Bei Hu Lu (Gonglu Duan, c. 870), Gui Tian Lu (Xiu Ouyang, 1067), Bian Min Tu Zuan (Fan Kuang, 1502) and Hua Li Bai Yong (Changzuo Weng, c. 1718)[2,3,9,11,13,14]. However, to the best of our knowledge, no primitive population of cultivated Fortunella spp. has been reported either in ancient literature or modern studies. Although, F. hindsii has been found to be widely distributed in the primitive forests of southern China from ancient to modern times[1,2,11,13,14]. More importantly, ancient Chinese scholars clearly distinguished cultivated Fortunella and F. hindsii in the rigorous pomology of Citrus Record (Yanzhi Han, 1178), herbology of Compendium of Materia Medica (Shizhen Li, 1578) and floriculture monographs of Flower Mirror (Haozi Chen, 1688), respectively. These facts suggest that cultivated Fortuenlla spp. were not selected in modern times, which leads to ongoing controversy about their evolutionary origin. One reasonable hypothesis is that cultivated Fortunella spp. originated from natural crossing or backcrossing between a primitive Fortunella spp. (probably F. hindsii) and Citrus spp.[18]. Another hypothesis is that cultivated Fortunella was directly domesticated from F. hindsii, because the main difference in phenotype between them is in the fruit: the fruit of F. hindsii are smaller, seedier and thin-rinded with a bitter and spicy taste, whereas fruit of cultivated Fortunella spp. are larger with thicker albedo and a sweet and palatable taste. According to the local chronicles of New Book for Southern Life (Yi Qian, 1016), Composition of Chicken’s Ribs (Chuo Zhuang, 1143) and New Anecdotes in Guangdong (Dajun Qu, 1687), Luo Fu kumquat (F. margarita) was first selected from wild kumquat by monks living on Mount Luofu in Guangdong Province, and served as a tribute to emperors in the period of the Tang Dynasty. However, there has been no molecular evidence supporting these hypotheses.
With the rapid improvement of the population genetic method based on germplasm collection and molecular data, a number of novel primitive species and unexpected centers of origin of modern cultivars have been discovered, not only for citrus species[24,25], but also for some rare landscape[26] and medicinal plants[27], providing instructive information for breeding improvement and genetic conservation. Therefore, this study aimed to determine the genetic nature of Fortunella with a systematic collection of various germplasm, and conducted comprehensive phylogenetic and population analyses based on the chloroplast loci, nuclear microsatellites (nSSR) and genomic single nucleotide polymorphism (SNP) data. The findings provide new insights into the phylogeny, classification and evolution of Fortunella, which may greatly facilitate further research related to this genus.

2 METHODS

2.1 Plant materials

Thirty-eight Fortunella accessions including cultivars, landraces, residential garden plants and hybrids were sampled from Zhejiang, Hunan, Jiangxi, Fujian, Guangdong, Guangxi and the Citrus Research Institute of Chinese Academy of Agriculture Sciences (Chongqing, China), and 10 citrus accessions including pummelo (Citrus maxima), citron (C. medica), mandarin (C. reticulata), sweet orange, sour orange (C. aurantium), lemon, lime (C. aurantifolia), papeda (C. ichangensis), trifoliate orange (Poncirus trifoliata), Chinese box orange (Atalantia buxifolia) were sampled from the Institute of Citriculture of Huazhong Agriculture University (Wuhan, China). All the samples were prepared for chloroplast analysis and nSSR genotyping (Tab.1). For whole-genome resequencing, the Hong Kong kumquat accessions collected close to cultivated environment and used in the above experiment were excluded. For synonymous accessions belonging to cultivated Fortunella and showing genetic similarity higher than 95% in nSSR analysis, only one sample was retained and included. Finally, 15 cultivated Fortunella (CUL) and 15 wild Hong Kong kumquat (HK) accessions, representative of their respective populations, were prepared for next generation sequencing. The DNA was extracted from leaves following the method developed by Cheng et al.[28].
Tab.1 Details and utilization of the accession evaluated in this study
Accession code Common name Species Location Description Utilization
LF Nagami Fortunella margarita Taizhou, Zhejiang Germplasm 1, 2, 3
QZLF Nagami F. margarita Quzhou, Zhejiang Germplasm 1, 2
DGLF Nagami F. margarita Zhejiang Landrace 1, 2, 3
JZLF Nagami F. margarita Zhejiang Landrace 1, 2
WZLF Nagami F. margarita Wenzhou, Zhejiang Landrace 1, 2
LFHY Nagami F. margarita Taizhou, Zhejiang Landrace 1, 2
LW Narumi F. japonica Ningbo, Zhejiang Germplasm 1, 2, 3
GXCP Narumi F. japonica Liuzhou, Guangxi Landrace 1, 2, 3
LCHP Meiwa F. crassifolia Guilin, Guangxi Cultivar 1, 2
HPJG Meiwa F. crassifolia Guilin, Guangxi Landrace 1, 2, 3
LSHY Meiwa F. crassifolia Yongzhou, Hunan Germplasm 1, 2
YXBZQ Meiwa F. crassifolia Sanming, Fujian Landrace 1, 2, 3
JX4 Meiwa F. crassifolia Ji’an, Jiangxi Cultivar 1, 2, 3
LYJD Meiwa F. crassifolia Changsha, Hunan Germplasm 1, 2, 3
WZJD Meiwa F. crassifolia Wenzhou, Zhejiang Germplasm 1, 2, 3
LSJG Meiwa F. crassifolia Yongzhou, Hunan Landrace 1, 2, 3
LYJG Meiwa F. crassifolia Changsha, Hunan Cultivar 1, 2
NBJD Meiwa F. crassifolia Ningbo, Zhejiang Cultivar 1, 2
RAJG Meiwa F. japonica Liuzhou, Guangxi Landrace 1, 2, 3
YSJG Meiwa F. crassifolia Guilin, Guangxi Cultivar 1, 2, 3
G1 Meiwa F. crassifolia Guilin, Guangxi Landrace 1, 2
G2 Meiwa F. crassifolia Guilin, Guangxi Landrace 1, 2, 3
DJD Hong Kong F. hindsii Fujian Ornamental 1, 2
XJD Hong Kong F. hindsii Fujian Ornamental 1, 2
WTD Hong Kong F. hindsii Jieyang, Guangdong Wild 1, 2, 3
JD Hong Kong F. hindsii Fujian Ornamental 1, 2
DR01 Hong Kong F. hindsii Longyan, Fujian Wild 1, 2, 3
DB02 Hong Kong F. hindsii Ganzhou, Jiangxi Wild 1, 2, 3
CZ Hong Kong F. hindsii Chenzhou, Hunan Wild 1, 2, 3
MJS Hong Kong F. hindsii Ningbo, Zhejiang Wild 1, 2, 3
WGJ Wenguangju Hybrid / Ornamental 1, 2
JGZ / Hybrid / Rootstock 1, 2
SJJ Calamondin Citrus madurensis Fujian Cultivar 1, 2
CS Changshou F. obovata Fujian Ornamental 1, 2
HCYJG Meiwa F. crassifolia Ganzhou, Jiangxi Germplasm 1, 2, 3
XLF Nagami F. margarita Ningbo, Zhejiang Landrace 1, 2, 3
YXJG Meiwa F. crassifolia Sanming, Fujian Cultivar 1, 2
YCJD Meiwa F. crassifolia Quanzhou, Fujian Germplasm 1, 2
ML Lime C. aurantifolia / Cultivar 1, 2
XC Sweet orange C. sinensis / Cultivar 1, 2
ZHI Trifoliate orange Poncirus / Cultivar 1, 2
XY Foshou citron C. medica / Cultivar 1, 2
YCC Papeda C. ichangensis / Wild 1, 2
GXMY Pummelo C. maxima / Cultivar 1, 2
MSYJ Mandarin C. reticulata / Cultivar 1, 2
HKC Box orange Atalantia / Relatives 1, 2
NM Lemon C. limon / Cultivar 1, 2
SC Sour orange C. aurantium / Cultivar 1, 2
CFX Hong Kong F. hindsii Ganzhou, Jiangxi Wild 3
JIEX Hong Kong F. hindsii Jieyang, Guangdong Wild 3
DYS02 Hong Kong F. hindsii Sanming, Fujian Wild 3
LH08 Hong Kong F. hindsii Xiamen, Fujian Wild 3
DYT01 Hong Kong F. hindsii Longyan, Fujian Wild 3
ZX9 Hong Kong F. hindsii Wenzhou, Zhejiang Wild 3
LY43 Hong Kong F. hindsii Ningbo, Zhejiang Wild 3
HC27 Hong Kong F. hindsii Shaoguan, Guangdong Wild 3
XMS Hong Kong F. hindsii Guangzhou, Guangdong Wild 3
JLS Hong Kong F. hindsii Ganzhou, Jiangxi Wild 3

Note: Utilization 1, chloroplast sequencing; 2, nSSR; and 3, resequencing.

2.2 Chloroplast loci and nuclear SSR analysis

Five chloroplast loci, including two intergenic spacers (trnK-matK and trnQ-psbK), two introns (rpl16 and rps16) and one coding sequence (matK) were amplified using the primers (Table S1) designed based on the sweet orange chloroplast genome[29]. Polymerase chain reaction amplification and amplicon sequencing followed a workflow previously described by Yang et al.[30]. Raw sequence data were imported into MEGA 7.0 and trimmed for multiple alignment[31]. Finally, a matrix of 4413 bp sequence by 48 samples was obtained for the following analysis. The five chloroplast regions of each sample were linked up for the construction of a phylogenetic tree using the maximum parsimony algorithm built in MEGA 7.0 with 1000 bootstrap replicates. The raw tree was annotated by iTOL[32]. The nucleotide polymorphism was calculated by DNASP 6.0 software[33]. The haplotype network was constructed using NETWORK 10.0.1[34].
Forty-six nSSRs (Table S2) were selected from the Sweet Orange Genome data set[22] and genotyping was performed following the protocol described by Ruiz et al.[35]. The polymorphism bands were recorded as the format of Genalex 6.5 for genetic similarity, diversity and principal coordinate analysis[36]. The genetic similarity matrix was transformed to the format of MEGA 7.0 for the construction of the phylogenic dendrogram (algorithm with 1000 bootstrap replicates). By using Genalex 6.5, the data set was transformed to the format of Structure 2.3.4[37] for genetic structure analysis. The K value was tested from 2 to 10 with three replicates and then the best K was estimated by Structure Harvester[38].

2.3 Resequencing work flow and population genomic analysis

Ten microgramme of high-quality genomic DNA of each sample was prepared for the construction of an NGS library. The paired-end sequencing libraries with an average insert size of ~300 bp were constructed and then sequenced by using Illumina Hiseq 2500 platform with an average depth of about thirtyfold genome coverage. The raw paired-end reads were removed with the adapter and quality filtered using Trimmomatic 0.33[39] with an option of SLIDINGWINDOW:4:15 MINLEN:36 HEADCROP:5. The clean reads were mapped to the mini-citrus reference genome V1.0[12] using BWA (0.7.12)[40] with default parameters. The SAM (sequence alignment map) files were transformed to BAM (binary alignment map) files[41] using SAMtools (1.3.1) with the q parameter set to 30, and then sorted and duplication removed with the default parameters. AddOrReplaceReadGroups procedure in Picard was performed to add a Read Group to BAM files. Population-based SNP calling was performed using SAMtools and the raw SNPs were flittered with the criteria of QUAL < 30.0 || MQ < 40.0 || DP < 5.0 using the Bcftools tool. The SNPs were annotated by using SnpEff (4.3T)[42].
Principal component analysis was performed by using GCTA (1.92.3)[43]. The population structure was estimated using ADMIXTURE (1.3.0)[44] and the K value was tested from 2 to 6. All the Fst indexes were calculated using VCFtools[45], and the high differentiation genome region was screened by a criterion of mean Fst > 0.3 for a 10-kb window according to the statistical distribution of global Fst. The linkage disequilibrium (LD) decay was calculated using PLINK (1.90)[46]. The Pi values were calculated by using Variscan (2.0.3)[47] with the parameters WidthSW set to 20,000 and JumpSW set to 10,000. Population demography analysis was performed using the pairwise sequentially Markovian coalescent model[48]; the paired-end clean reads were transformed to psmcfa format using the fq2psmcfa script. The mean generation time was set at 4 years for CUL and 2 years for HK. The mutation rate was assumed as 2.2 × 10−8 substitutions per site per generation as described by Wang et al.[24].

3 RESULTS

3.1 Phylogenetic analysis of Fortunella based on chloroplast loci

Among the 38 Fortunella accessions, 25 polymorphic sites (Np) and 11 chloroplast haplotypes (Nh) were identified from five chloroplast loci with a haplotype diversity (Hd) of 0.69, a nucleotide diversity of (Pi) 7.3 × 10−4 and an average number of nucleotide difference (Nk) of 3.14. The locus trnK-matK was the most polymorphic one (with 12 Np and 5 Nh), suggesting the application potential of this locus for germplasm barcoding in the future; whereas trnQ-psbK was the most conserved locus with only one polymorphic site. With the data of 10 citrus accessions added to the above data set, the Np and Nh value increased markedly to 156 and 11, respectively, resulting a Hd of 0.80, a Pi of 3.0 × 10−3 and a Nk of 12.77 (Table S3). These results indicated higher chloroplast conservativeness in Fortunella than in Citrus. Eight SNPs showed diagnostic value for Fortunella/Citrus, which may serve as useful markers for offspring identification in cytomixis or crossing breeding between these two genera (Table S4).
Distance-based clustering by neighbor joining revealed six main clades (Fig.1) among the 48 accessions, with Chinese box orange in clade I (black; located at the basal), citron in clade II (yellow), papeda, wild mandarin and lime in clade III (light green), Poncirus in clade IV (dark green), pummelo, lemon sweet orange and sour orange in clade V (olive), and all the 38 Fortunella accessions in clade VI (red). Fortunella spp. were clearly separated from citron, mandarin, pummelo and papeda, indicating an independent phylogeny of Fortunella in the true citrus fruit tree group. The overall tree topology indicates that Fortunella has a closer phylogenic relationship with Citrus than with Poncirus. Within the Fortunella clade, no obvious hierarchical structure was observed and all the accessions clustered with very low genetic differences to each other, indicating the monophyletic origin of the Fortunella lineage. All the four known hybrid accessions, SJJ (calamondin), CS (Changshou kumquat, F. obovata), WGJ (Wenguangju) and JGZ (a rootstock), were clustered within the Fortunella clade, indicating that their female parent should be Fortunella.
Fig.1 Phylogenetic tree and haplotype network of Fortunella based on five chloroplast loci. (a) Phylogenetic tree of 38 Fortunella and 10 citrus accessions. Clades of the tree are highlighted by different colors. Clade I, black, Chinese box orange (Atalantia buxifolia); Clade II, yellow, citron (Citrus medica); Clade III, light green, lime (C. aurantiifolia), papeda (C. ichangensis) and wild mandarin (C. reticulata); Clade IV, dark green, trifoliate orange (Poncirus trifoliata); Clade V, olive, sweet orange (C. sinensis), lemon (C. limon), pummelo (C. maxima) and sour orange (C. aurantium); and Clade VI, red, kumquat (Fortunella spp.). (b) Haplotype network of 38 Fortunella and 10 citrus accessions. Each dot on the network presents a type of haplotype. The genera are: blue, Poncirus; yellow, Citrus; black, Atalantia; and red, Fortunella. (c) Composition of the 11 Fortunella haplotypes. Species are highlighted as: purple, Hong Kong kumquat (F. hindsii); blue, Meiwa kumquat (F. crassifolia); salmon, Nagami kumquat (F. margarita); green, Marumi kumquat (F. japonica); and gray, hybrid kumquat.

Full size|PPT slide

To further clarify the cytoplasmic evolution within Fortunella, a haplotype network was constructed using the median-joining algorithm (Fig.1). Notably, 11 haplotypes of Fortunella fell into a single branch, which is divergent from the eight Citrus haplotypes, demonstrating the independent chloroplast origin of Fortunella. The 11 haplotypes could be divided into two distinct groups: CUL with five haplotypes (H_1 to H_5) and HK with six haplotypes (H_6 to H_11), suggesting a further dichotomous differentiation after the origination of the common Fortunella ancestor. Among the 11 haplotypes, H_1 is the most frequent one shared by 21 accessions, followed by H_7 (four accessions), H_6 (two accessions), H_3 (two accessions) and H_4 (two accessions) (Table S5). Three hybrid accessions (Wenguangju, JGZ and Changshou) shared haplotype H_7 with one HK accession, whereas the most valuable kumquat hybrid cultivar, calamondin, shared haplotype H_1 with other 20 edible CUL accessions, indicating that their female parents are different.

3.2 Genetic analysis of Fortunella based on nSSR markers

To further dissect the population structure of Fortunella, 47 nSSR loci were amplified and analyzed among the 38 kumquat and 10 citrus accessions. Two hundred and four alleles were detected among the 38 kumquat accessions. On average, the number of alleles (Na) and effective number of alleles (Ne) was 4.34 and 2.27, respectively. The allele number varied between 2 (for locus C13, D04B, A03, A21, A24, A18, E27, E28 and E30) and 8 (E6) (Table S6). The average Shannon’s information index (I) and expected heterozygosity (He) was calculated as 0.92 and 0.49, respectively. E1 was the most informative locus with an I value of 1.58, and E30 was the least informative one (I = 0.39). Most of the loci (28 out of 47) showed He values higher than 0.5. These results indicated that Fortunella has higher nuclear diversity than chloroplast diversity, and this data set is more powerful for the dissection of the population structure. With the addition of the 10 citrus accessions, 325 alleles were detected. On average, the Na and Ne was 6.91 and 2.94, respectively. The Na varied between 2 (E27) and 15 (B26) (Table S7). The I and He was calculated as 1.27 and 0.48, respectively. B26 was the most informative locus with a high I value of 1.99, and E27 was the least informative one (I = 0.26).
The principal coordinate analysis based on Nei’s genetic distance revealed the genetic divergence between Fortunella and Citrus accessions as well as within Fortunella (Fig.2). The first two principal coordinates accounted for 26.3% and 11.9% of the total genetic variance, respectively. There were two distinct groups (in green and orange dashed areas) on the positive X and Y axis formed by the 34 Fortunella accessions, and both were significantly differentiated from the 10 citrus accessions (in a gray dashed area); whereas the four Fortunella hybrid accessions formed a group intermediate between Fortunella and Citrus.
Fig.2 Genetic structure of 38 kumquat and 10 citrus accessions based on 47 nSSR loci. (a)Principal component analysis of 38 kumquat and 10 citrus accessions. Accessions are presented by different colors (as in Fig. 1(a)). Clade I, black, Chinese box orange (Atalantia buxifolia); Clade II, yellow, citron (Citrus medica); Clade III, light green, lime (C. aurantiifolia), papeda (C. ichangensis) and wild mandarin (C. reticulata); Clade IV, dark green, trifoliate orange (Poncirus trifoliata); Clade V, olive, sweet orange (C. sinensis), lemon (C. limon), pummelo (C. maxima) and sour orange (C. aurantium); Clade VI, red, kumquat (Fortunella spp.). (b) Phylogenetic tree and population structure of 38 kumquat and 10 citrus accessions based on nSSR data. On the left side, phylogenetic tree was constructed using distance-based UPGMA method; clades of known kumquat hybrids (WGJ, CS, JGZ and Calamondin), cultivated Fortunella spp. (F. margarita, F. crassifolia and F. japonica) and Hong Kong kumquat (F. hindsii) accessions are shown in pink, orange and green, respectively. On the right side, each accession is represented by a horizontal stacked bar of genetic components with the proportion shown in color for K = 2 estimated by STRUCTURE.

Full size|PPT slide

The phylogenic dendrogram constructed based on the genetic similarity matrix included the hybrid accessions (Fig.2). The overall tree topology was consistent with that of the phylogenetic tree for chloroplast and that in previous studies based on nSSR markers, indicating that Fortunella is closer to Citrus but distant from Poncirus. The Fortunella accessions were well organized into three clusters (hybrids, CUL and HK), which was consistent with the principal coordinate analysis presented in Fig.2. The first cluster included four hybrid kumquat accessions with a genetic similarity (GS) of 80.7%; the second cluster comprised 26 CUL accessions with a GS of 83.4%; and the third cluster had eight HK accessions with a GS of 76.8%. In the CUL clade, hierarchical structures were discovered, which basically corresponded to F. margarita, F. japonica and F. crassifolia.
The above genotyping data were further used to investigate the genetic structure of the 34 true Fortunella accessions, with the exclusion of the four hybrids. Evanno’s test indicated a sharp signal at K = 2 (ΔK = 530.0), implying that two gene pools (in red and purple bars) were involved in the evolution of modern Fortunella (Fig.2 and Fig. S1). All the 26 CUL accessions only showed genetic components derived from the red ancestor. Seven out of the eight Hong Kong kumquat accessions showed single genetic components derived from the purple ancestor, while the remaining one (WTD) exhibited a mixture of genetic components, with 79.8% of HK and 20.2% of CUL.
The nSSR analysis combined with chloroplast analysis demonstrated the independent phylogeny of Fortunella among citrus taxa and indicated the monophyletic origin of all Fortunella spp.. Furthermore, these results also implied the subdivision of Fortunella into two lineages corresponding to CUL and HK.

3.3 Comparative genomic analysis between cultivated Fortunella and wild Hong Kong kumquat populations

Given the high morphological similarity[2,13] and obviously different fruit phenotype (Fig.3 between cultivated Fortunella and wild Hong Kong kumquat, a final data set consisting of 5,104,141 high-quality SNPs (Table S8) genotyped from 15 CUL accessions from the main production areas (population CUL) and 15 HK accessions from primitive forests was obtained by whole-genome sequencing to reveal the genetic relationship between these two populations (Tab.1).
Fig.3 Phenotype and population structure of cultivated Fortunella spp. (CUL) and wild Hong Kong kumquat (HK). (a) Fruit phenotypes of the four Fortunella spp. (b) Cross and longitudinal sections of CUL and HK; CUL has larger fruit organ with thickened and sweet albedo, whereas HK has smaller fruit with thin and acerb peel. (c) Population structure among 15 CUL and 15 HK accessions based on whole-genomic 5,104,141 SNPs. Each accession is represented by a vertical stacked column of genetic components with the proportion shown in color for K = 2, 3 and 4 estimated by ADMIXTURE.

Full size|PPT slide

To estimate the most likely ancestral model between CUL and HK, we predefined the number of ancestral numbers (K) from two to six, and evaluated the confidence by cross-validation (CV) (Fig. S2). The minimum CV error (0.43) was observed at K = 2, which clearly confirmed that Fortunella comprises the two populations (Fig.3). Intriguingly, three HK accessions (DB02, JLS and HC27) from Jiulianshan (Ganzhou, Jiangxi Province) showed admixed genetic background, suggesting they were subjected to recent introgressions from CUL or shared ancestral variations with CUL. In addition, to determine the genetic structure of CUL, the genetic structures at K = 3 and 4 (CV error = 0.49 and 0.50, respectively) were also plotted. Unexpectedly, at K = 3, two subgroups were identified in HK, with some accessions admixed between them. At K = 4, CUL diverged into two subpopulations: F. margarita and F. crassifolia. It is also out of expectation that all the F. japonica accessions showed an admixture background, indicating that F. japonica instead of F. crassifolia has a hybrid background. This result challenges the hypothesis proposed by Swingle that F. crassifolia is a hybrid of F. margarita and F. japonica.
In the genomic diversity analysis, the segregating sites, mutation number (Eta), singleton number (Eta_E) of HK were obviously higher than those of CUL (Tab.2), indicating a higher level of allelic variation in HK. The Pi and Theta of HK (0.23 and 0.26) were nearly twofold those of CUL (0.12 and 0.10), indicating a higher general genomic diversity of HK than CUL. Notably, the neutral test statistics (Tajima’s D, Fu & Li D* and Fu & Li F*) of both CUL and HK were deviated from zero, but distributed in opposite polarities, indicating that directional selection might have occurred in their evolution history but in opposite directions (domestication and natural selection). The lineage disequilibrium (LD) strength of the two populations was further compared (Fig. S3). The LD of CUL (orange) decayed to half at ~20 kb, while that of HK (green) decayed to half at ~10 kb, indicating the LD strength of CUL is generally higher than that of HK. Collectively, according to the lower genetic diversity and stronger LD strength of CUL, it can be speculated that artificial selection might have been involved in its origin; while for HK, given its higher genetic diversity and weaker LD strength, it can be inferred that natural selection might have been the key driving force for its evolution.
Tab.2 Genomic diversity of cultivated Fortunella (CUL) and wild Hong Kong kumquat (HK) populations
Statistics CUL HK
Number of segregating sites 3,737,798 9,140,932
Total number of mutations 3,762,401 9,269,923
Number of singletons 706,304 3,862,794
Pi 0.12 0.23
Theta 0.10 0.26
Tajima_D 0.61 −0.46
FuLi_Dstar 0.75 −0.46
FuLi_Fstar 0.82 −0.53

Note: Pi, nucleotide diversity statistic; Theta, Watterson’s estimator Theta; Tajima_D, Tajima’s D test statistic; FuLi_Dstar, Fu & Li’s D* statistic; FuLi_Fstar, Fu & Li’s F* statistic.

3.4 Genetic differentiation and demographic history analyses between cultivated Fortunella and wild Hong Kong kumquat populations

To investigate the level of genetic differentiation between CUL and HK, the Fst between CUL and each geographic group of HK was calculated (Tab.3). The Fst between CUL and HK was 0.364, which is a relatively high level of genetic differentiation for perennial tree species[49,50]. Although the three CUL species showed close genetic relationship in the chloroplast and nSSR analysis, the genetic differentiation level between each pair is higher than that of the any pair of pummelo, citron, mandarin and papeda[23,24,51], indicating they should be designated to three different species. Among the three CUL species, the highest Fst was detected between F. margarita and F. crassifolia, which again supports the hybrid origin of F. japonica.
Tab.3 Genomic divergence between cultivated Fortunella (CUL) and wild Hong Kong kumquat (HK) populations
Comparison set Fst
CUL vs HK 0.364
CUL vs HK (Jiulianshan) 0.361
CUL vs HK (Guangdong coast) 0.438
CUL vs HK (Fujian coast) 0.456
CUL vs HK (Zhejiang coast) 0.469
F. margarita vs F. japonica 0.338
F. margarita vs F. crassifolia 0.345
F. japonica vs F. crassifolia 0.327

Note: The resequencing data (15 cultivated Fortunella and 15 wild Hong Kong kumquat accessions) are archived in NCBI under BioProject PRJNA736109. The matchup between accession code in paper and data code in archive please see Table S9.

To trace the potential domestication clues between CUL and HK, the highly differentiated genomic regions were screened by the criterion of Fst > 0.3 according to the statistical distribution of global Fst (median = 0.3) (Fig. S4). In total, 138 blocks on 79 contigs were identified (Table S10), which contained 747 protein coding genes (supplementary data set). Gene ontology analysis of these genes (Fig. S5 and supplementary data set) showed that acylglycerol acyltransferase activities (GO:0019432 and GO:0046463) and glyceride biosynthesis processes (GO:0019432 and GO:0046463) were highly enriched, which might be related to the high drought and cold tolerance of CUL[3,11,52]. We further manually annotated the 747 genes and their adjacent regions, and 36 genes involved in the tricarboxylic acid cycle were identified (Table S11).
To trace back the demographic history of CUL and HK, the pairwise sequentially Markovian coalescent model was used to infer the fluctuations in the effective population size (Ne) over time. As shown in Fig.4, obviously asynchronous Ne curves were detected for CUL and HK. CUL first exhibited a decline in Ne (known as a bottleneck) during ~0.7–1.2 mya, which might be associated with climatic variations in the Quaternary glacial period (QGP; ~0.02–3.0 mya); whereas HK later experienced a similar bottleneck during ~0.3–0.6 mya. These results suggested that a niche or geographic isolation between CUL and HK had been established during or before QGP. Therefore, the earlier bottleneck of CUL implied its higher latitude or altitude distribution than HK. As shown in the population structure analysis, there was very limited gene flow between CUL and HK, suggesting that the distribution of CUL during QGP was likely to be of higher latitude. After the bottleneck, both CUL and HK underwent Ne fluctuation, which still showed asynchronous trends, suggesting their different spatiotemporal distributions during the interglaciation.
Fig.4 Demographic history and speciation hypothesis of Fortunella. (a) Demographic history of cultivated Fortunella (CUL) and wild Hong Kong kumquat (HK) populations. Effective population size of the CUL (orange curve) and HK (green curve) were reconstructed by using the pairwise sequentially Markovian coalescent model. Quaternary glacial period is marked by blue background. Obvious population bottlenecks are marked by gray background. (b) Speciation hypothesis of Fortunella spp. Blue, orange and green dots represent the common ancestor of Fortunella, cultivated Fortunella spp. (F. margartita, F. crassifolia and F. japonica) and Hong Kong kumquat (F. hindsii) populations, respecitvely. The peaked line represents geographical barrier; straight arrow represents natural selection; dotted arrow indicates artificial selection. The northern population experienced earlier and severer climatic changes during Quaternary glacial period (QGP) compared to the southern population. During QGP, the northern and southern populations were gradually isolated from each other by geographical barrier (probably Nanling Mountains) and underwent adaptive evolution separately. With the southward migration of humans, modern cultivated Fortunella spp. were selected from the northern population.

Full size|PPT slide

4 DISCUSSION

4.1 Phylogeny and classification of Fortunella

To the best of our knowledge, this is the first comprehensive study focusing on the phylogeny of Fortunella. According to the high chloroplast conservation and low haplotype diversity of Fortunella, as well as the paralleling phylogeny of Fortunella to Citrus and the intrinsically different flowering seasons of Fortunella (from summer to autumn) and Citrus spp. (in spring)[2,13], it can be speculated that after differentiation from the common ancestor of the true citrus fruit tree group, the Fortunella lineage underwent a relatively independent evolutionary trajectory, which was in agreement with the previous phylogenic studies of Citrus spp. based on chloroplast and nSSR data[15,19,20,53]. Thus, it can be confirmed that the crossing event between Fortunella and Citrus was not involved in the origin of CUL. Since CUL and HK accessions were closely clustered in the same clade without obvious hierarchical structure in haplotype network of chloroplast, it seems not reasonable to further classify Fortunella genus into subgenus Protocitrus and Eufortunella as proposed by Tanaka[13]. The genetic structure analysis demonstrated that F. japonica instead of F. crassifolia has a hybridization genetic background, rejecting the hypothesis that F. crassifolia is a natural hybrid between F. margarita and F. japonica[2]. Each pair of cultivated Fortunella species showed a relatively high level of genetic differentiation, indicating that each of them are justifiably ranked as a species and rejecting the concept of a F. margarita complex[18]. According to the results of the present study, especially the asynchronous demographic changes between HK and CUL, we could modify the hypothesis proposed by Yasuda et al.[18] as follows. F. hindsii is a surviving ancestor for other Fortunella spp. and modern cultivated Fortunella might have derived from numerous mutations and selections involving F. hindsii or other extinct Fortunella ancestors. However, it remains unclear whether there is a direct domestication relationship between CUL and HK. Wild collection with larger scale, fine annotation of the Fortunella genomes, genetic mapping of key genes involved in the different fruit phenotypes between CUL and HK, and related gene function researches may comprehensively provide further answer to this question.

4.2 Geographic origin of Fortunella

Although Fortunella is considered to have originate in China[2,3,9,11,13,14], no solid molecular evidence has been reported. Here, the demographic history analysis suggested that the ancient distribution of CUL should be closer to the north than HK. The current distribution of wild HK is mainly in mountainous and coast area of southern China[54]. These facts provide the first molecular evidence for the continental origin of cultivated Fortunella, which still needs fossil evidence for validation. Furthermore, because Nanling Mountains (24°–26° N, 110°–115° E) is the northern border of wild F. hindsii distribution and has been proven to be the centers of origin of citrus species, such as C. ichangensis and C. reticulata[24,25], we speculate that Nanling might be the main geographic barrier for the gene flow between primitive CUL and HK during QGP. Since the admixed genetic background of HK-Jiulianshan (belonging to Nanling) population has been detected, further wild investigation and germplasm collection in this area is necessary to determine whether the genetic introgression is caused by natural pollination from the kumquat gardens nearby, or primitive populations of F. margarita, F. crassifolia and F. japonica still survive in the glacial refuge in Nanling.

4.3 Hypothesis for the speciation and evolution of Fortunella

Based on the results of this work and previous studies[12,23,55], we propose a new hypothesis about the evolutionary history of Fortunella (Fig.4). After differentiation from the Citrus lineage (~5−6 mya), the ancestor of Fortunella evolved into an independent lineage widely distributed in central and southern China. Along with the progression of QGP, the northern and southern populations of Fortunella were gradually isolated from each other (possibly by Nanling mountains). The northern Fortunella population was confronted with earlier and more severe natural selection (cold and dry), and thus experienced an earlier QGP bottleneck, which resulted in adaptive evolution such as thickened albedo with enrichment of sugar and secondary metabolites to protect the seeds from freezing. However, the southern population encountered moderate and later natural selection, and thus experienced later bottleneck and maintained the phenotype of primitive fruit. Along with the southward migration of humans[56,57], a few individuals of the northern population were selected and cultivated, and thus survive till the present as Luo Fu or Nagami (F. margarita), Jin Dan or Meiwa (F. crassifolia) and Luo Wen or Marumi (F. japonica). However, the southern population mainly underwent continuous natural selection and was discovered successively by ancient Chinese horticulturalists and modern western scholars, and named as Shan Jin Gan and Hong Kong kumquat, respectively.

5 CONCLUSIONS

In this work, by phylogenetic analysis based on chloroplast and nSSR data and population genomic analysis based on SNP data, we provide some new insights into the phylogeny, classification, and historical demography of Fortunella. First, Fortunella has an independent phylogeny among the true citrus fruit trees, and comprises two main populations corresponding to cultivated Fortunella spp. and Hong Kong kumquat. F. japonica instead of F. crassifolia has a hybrid origin. Artificial selection might involve in the evolution of cultivated Fortunella spp. instead of crossing between Fortunella and Citrus. A new hypothesis about the speciation of Fortunella has been proposed based on the results of the present study. Future research could focus on the domestication relationship between F. hindsii and cultivated Fortunella. These germplasms, data, results and perspectives would not only serve as useful resources for genetic improvement of kumquat and citrus, but also contribute to further evolutionary studies of citrus taxa in the future.

References

[1]
Hao C G. How to monitor the growth of broiler breeders. Poultry Husbandry and Disease Control, 2018, (9): 28−30 (in Chinese)
[2]
Li Z X. The significance and measures to improve the uniformity of broiler slaughter weight. Poultry Husbandry and Disease Control, 2019, (6): 22−24 (in Chinese)
[3]
Ji H Y. Reasons and solutions for slow growth of broiler chickens. Modern Animal Husbandry Science & Technology, 2020, (5): 24−25 (in Chinese)
[4]
Kristensen H H, Cornou C. Automatic detection of deviations in activity levels in groups of broiler chickens—A pilot study. Biosystems Engineering, 2011, 109(4): 369–376
CrossRef Google scholar
[5]
Flees J, Greene E, Ganguly B, Dridi S. Phytogenic feed- and water-additives improve feed efficiency in broilers via modulation of (an)orexigenic hypothalamic neuropeptide expression. Neuropeptides, 2020, 81: 102005
CrossRef Google scholar
[6]
Xin H, Liu K. Precision livestock farming in egg production. Animal Frontiers, 2017, 7(1): 24–31
CrossRef Google scholar
[7]
De Wet L, Vranken E, Chedad A, Aerts J M, Ceunen J, Berckmans D. Computer-assisted image analysis to quantify daily growth rates of broiler chickens. British Poultry Science, 2003, 44(4): 524–532
CrossRef Google scholar
[8]
Nielsen B L, Litherland M, Nøddegaard F. Effects of qualitative and quantitative feed restriction on the activity of broiler chickens. Applied Animal Behaviour Science, 2003, 83(4): 309–323
CrossRef Google scholar
[9]
Kristensen H H, Aerts J M, Leroy T, Wathes C M, Berckmans D. Modelling the dynamic activity of broiler chickens in response to step-wise changes in light intensity. Applied Animal Behaviour Science, 2006, 101(1−2): 125−143
[10]
Dawkins M S, Cain R, Merelie K, Roberts S J. In search of the behavioural correlates of optical flow patterns in the automated assessment of broiler chicken welfare. Applied Animal Behaviour Science, 2013, 145(1−2): 44−50
[11]
Wolff I, Klein S, Rauch E, Erhard M, Mönch J, Härtle S, Schmidt P, Louton H. Harvesting-induced stress in broilers: comparison of a manual and a mechanical harvesting method under field conditions. Applied Animal Behaviour Science, 2019, 221: 104877
CrossRef Google scholar
[12]
Aerts J M, Van Buggenhout S, Vranken E, Lippens M, Buyse J, Decuypere E, Berckmans D. Active control of the growth trajectory of broiler chickens based on online animal responses. Poultry Science, 2003, 82(12): 1853–1862
CrossRef Google scholar
[13]
Aerts J M, Lippens M, De Groote G, Buyse J, Decuypere E, Vranken E, Berckmans D. Recursive prediction of broiler growth response to feed intake by using a time-variant parameter estimation method. Poultry Science, 2003, 82(1): 40–49
CrossRef Google scholar
[14]
Cangar O, Aerts J M, Vranken E, Berckmans D. End-weight prediction in broiler growth. British Poultry Science, 2006, 47(3): 330–335
CrossRef Google scholar
[15]
Lott B D, Reece F N, Mcnaughton J L. An automated weighing system for use in poultry research. Poultry Science, 1982, 61(2): 236–238
CrossRef Google scholar
[16]
Turner M J B, Gurney P, Crowther J S W, Sharp J R. An automatic weighing system for poultry. Journal of Agricultural Engineering Research, 1984, 29(1): 17–24
CrossRef Google scholar
[17]
Doyle I, Leeson S. Automatic weighing of poultry reared on a litter floor. Canadian Journal of Animal Science, 1989, 69(4): 1075–1081
CrossRef Google scholar
[18]
Mollah M B R, Hasan M A, Salam M A, Ali M A. Digital image analysis to estimate the live weight of broiler. Computers and Electronics in Agriculture, 2010, 72(1): 48–52
CrossRef Google scholar
[19]
Mortensen A K, Lisouski P, Ahrendt P. Weight prediction of broiler chickens using 3D computer vision. Computers and Electronics in Agriculture, 2016, 123: 319–326
CrossRef Google scholar
[20]
Fontana I, Tullo E, Butterworth A, Guarino M. An innovative approach to predict the growth in intensive poultry farming. Computers and Electronics in Agriculture, 2015, 119: 178–183
CrossRef Google scholar
[21]
Wang K. Perching behavior of chickens and development of an automated weighing system for weight monitoring of group housed chickens. Dissertation for the Doctoral Degree. Hangzhou: Zhejiang University, 2020 (in Chinese)
[22]
Seber R T, De Alencar Nääs I, De Moura D J, Da Silva Lima N D. Classifier’s performance for detecting the pecking pattern of broilers during feeding. AgriEngineering, 2022, 4(3): 789–800
CrossRef Google scholar
[23]
Doornweerd J E, Kootstra G, Veerkamp R F, de Klerk B, Fodor I, van der Sluis M, Bouwman A C, Ellen E D. Passive radio frequency identification and video tracking for the determination of location and movement of broilers. Poultry Science, 2023, 102(3): 102412
CrossRef Google scholar
[24]
England A D, Gharib-Naseri K, Kheravii S K, Wu S B. Rearing broilers as mixed or single-sex: relevance to performance, coefficient of variation, and flock uniformity. Poultry Science, 2022, 101(12): 102176
CrossRef Google scholar
[25]
Chedad A, Vranken E, Aerts J M, Berckmans D. Behaviour of chickens towards automatic weighing systems. IFAC Proceedings Volumes, 2000, 33(29): 207–212
[26]
Chedad A, Aerts J M, Vranken E, Lippens M, Zoons J, Berckmans D. Do heavy broiler chickens visit automatic weighing systems less than lighter birds. British Poultry Science, 2003, 44(5): 663–668
CrossRef Google scholar
[27]
Amraei S, Abdanan Mehdizadeh S, Salari S. Broiler weight estimation based on machine vision and artificial neural network. British Poultry Science, 2017, 58(2): 200–205
CrossRef Google scholar
[28]
Amraei S, Mehdizadeh S A, Nääs I A. Development of a transfer function for weight prediction of live broiler chicken using machine vision. Engenharia Agrícola, 2018, 38(5): 776–782
CrossRef Google scholar
[29]
Ma W, Li Q, Li J, Ding L, Yu Q. A method for weighing broiler chickens using improved amplitude-limiting filtering algorithm and BP neural networks. Information Processing in Agriculture, 2021, 8(2): 299–309
CrossRef Google scholar
[30]
Liu D, Vranken E, Van Den Berg G, Carpentier L, Peña Fernández A, He D, Norton T. Separate weighing of male and female broiler breeders by electronic platform weigher using camera technologies. Computers and Electronics in Agriculture, 2021, 182: 106009
CrossRef Google scholar
[31]
Peng Y, Zeng Z, Lv E, He X, Zeng B, Wu F, Guo J, Li Z. A Real-Time automated system for monitoring individual feed intake and body weight of group-housed young chickens. Applied Sciences, 2022, 12(23): 12339
CrossRef Google scholar
[32]
Wang K, Pan J, Rao X, Yang Y, Wang F, Zheng R, Ying Y. An image-assisted rod-platform weighing system for weight information sampling of broilers. Transactions of the ASABE, 2018, 61(2): 631–640
CrossRef Google scholar

Acknowledgements

This research was funded by Zhejiang Provincial Key R&D Program (2021C02026) and China Agriculture Research System (CARS-40).

Compliance with ethics guidelines

Danni Zhou, Yi Zhou, Pengguang He, Lin Yu, Jinming Pan, Lilong Chai, and Hongjian Lin declare that they have no conflicts of interest or financial conflicts to disclose. All applicable institutional and national guidelines for the care and use of animals were followed.

RIGHTS & PERMISSIONS

The Author(s) 2023. Published by Higher Education Press. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0)
AI Summary AI Mindmap
PDF(6181 KB)

Supplementary files

FASE-21436-OF-ZCQ_suppl_1 (887 KB)

Accesses

Citations

Detail

Sections
Recommended

/