Identification of Species by Combining Molecular and Morphological Data Using Convolutional Neural Networks

Systematic Biology, Год журнала: 2021, Номер 71(3), С. 690 - 705

Опубликована: Сен. 11, 2021

Integrative taxonomy is central to modern and systematic biology, including behavior, niche preference, distribution, morphological analysis, DNA barcoding. However, decades of use demonstrate that these methods can face challenges when used in isolation, for instance, potential misidentifications due phenotypic plasticity methods, incorrect identifications because introgression, incomplete lineage sorting, horizontal gene transfer Although researchers have advocated the integrative taxonomy, few detailed algorithms been proposed. Here, we develop a convolutional neural network method (morphology-molecule [MMNet]) integrates molecular data species identification. The newly proposed (MMNet) worked better than four currently available alternative tested with 10 independent sets representing varying genetic diversity from different taxa. High accuracies were achieved all groups, beetles (98.1% 123 species), butterflies (98.8% 24 fishes (96.3% 214 moths (96.4% 150 total species). Further, MMNet demonstrated high degree accuracy ($>$98%) closely related same genus. average two modest subgenomic (single nucleotide polymorphism) sets, comprising eight putative subspecies respectively, 90%. Additional tests show success rate identification under this most strongly depends on amount training data, robust sequence length image size. Analyses contribution types (image vs. gene) indicate both are important model, contribute slightly more. approaches developed here serve as foundation future integration multimodal information such image, audio, video, 3D scanning, biosensor characterize organisms more comprehensively basis improved investigation, monitoring, conservation biodiversity. [Convolutional network; deep learning; taxonomy; single polymorphism; identification.].

Язык: Английский

---

Species Diagnosis and DNA Taxonomy

Methods in molecular biology, Год журнала: 2024, Номер unknown, С. 33 - 52

Опубликована: Янв. 1, 2024

Язык: Английский

---

Issues and Perspectives in Species Delimitation using Phenotypic Data: Atlantean Evolution in Darwin’s Finches

Systematic Biology, Год журнала: 2017, Номер 67(2), С. 181 - 194

Опубликована: Авг. 31, 2017

Progress in the development and use of methods for species delimitation employing phenotypic data lags behind conceptual practical advances molecular genetic approaches. The basic evolutionary model underlying to delimit assumes random mating quantitative polygenic traits, so that distributions within a should be approximately normal individuals same sex age. Accordingly, two or more distinct traits suggest existence multiple species. In light this model, we show analytical approaches employed taxonomic studies using are often compromised by three issues: 1) reliance on graphical analyses convey little information phenotype frequencies; 2) exclusion characters potentially important following reduction dimensionality; 3) measures central tendency evaluate distinctiveness. We outline overcome these issues based statistical developments related mixture models (NMMs) illustrate them empirically with reanalysis morphological recently used claim there no morphologically Darwin's ground-finches (Geospiza). found negligible support relative hypotheses recognizing Although limits among merit further assessments additional sources information, our results bear implications other areas inquiry including speciation research: because have likely speciated not trapped process "Sisyphean" evolution as argued, they remain useful understand forces involved speciation. Our work underscores importance grounded appropriate delimitation. discuss how NMMs offer new perspectives kind inferences available systematists, significant repercussions ideas about structure biodiversity.

Язык: Английский

---

Two DNA barcodes and morphology for multi-method species delimitation in Bonnetina tarantulas (Araneae: Theraphosidae)

Molecular Phylogenetics and Evolution, Год журнала: 2016, Номер 101, С. 176 - 193

Опубликована: Май 3, 2016

Язык: Английский

---

Integrating coalescent species delimitation with analysis of host specificity reveals extensive cryptic diversity despite minimal mitochondrial divergence in the malaria parasite genus Leucocytozoon

BMC Evolutionary Biology, Год журнала: 2018, Номер 18(1)

Опубликована: Авг. 30, 2018

Coalescent methods that use multi-locus sequence data are powerful tools for identifying putatively reproductively isolated lineages, though this approach has rarely been used the study of microbial groups likely to harbor many unrecognized species. Among symbionts, integrating genetic species delimitation with trait could indicate reproductive isolation, such as host specificity data, despite its potential inform limits. Here we test ability an integrative combining and delimit within avian malaria parasite genus Leucocytozoon in central Alaska. We sequenced seven nuclear loci 69 samples multiple (GMYC BPP models), tested differences infection patterns among putative based on 406 individual infections, characterized morphology. found cryptic morphology masked a highly diverse assemblage, most recovering support at least 21 separate occur sympatrically have divergent patterns. Reproductive isolation appears evolved low mtDNA divergence, one instance two cytb haplotypes differed by single base pair (~ 0.2% divergence) were supported However, there was no consistent association between divergence three pairs observed idiosyncratic ecological haplotype be either conspecific, specificity, or specialization. Integrating non-traditional types provide novel view diversity parasites missed previously using barcodes. Species show is species-rich Alaska, extraordinary species-level worldwide. will important understanding evolutionary history symbionts moving forward.

Язык: Английский

---