Adaïmé Marc-Élie, Kong Shu, Punyasena Surangi W
Department of Plant Biology, University of Illinois Urbana-Champaign, Urbana, IL 61801, USA.
Faculty of Science and Technology, University of Macau, Macau 999078, China.
PNAS Nexus. 2023 Dec 13;3(1):pgad419. doi: 10.1093/pnasnexus/pgad419. eCollection 2024 Jan.
The phylogenetic interpretation of pollen morphology is limited by our inability to recognize the evolutionary history embedded in pollen features. Deep learning offers tools for connecting morphology to phylogeny. Using neural networks, we developed an explicitly phylogenetic toolkit for analyzing the overall shape, internal structure, and texture of a pollen grain. Our analysis pipeline determines whether testing specimens are from known species based on uncertainty estimates. Features from specimens with uncertain taxonomy are passed to a multilayer perceptron network trained to transform these features into predicted phylogenetic distances from known taxa. We used these predicted distances to place specimens in a phylogeny using Bayesian inference. We trained and evaluated our models using optical superresolution micrographs of 30 extant species. We then used trained models to place nine fossil specimens within the phylogeny. In doing so, we demonstrate that the phylogenetic history encoded in pollen morphology can be recognized by neural networks and that deep-learned features can be used in phylogenetic placement. Our approach makes extinction and speciation events that would otherwise be masked by the limited taxonomic resolution of the fossil pollen record visible to palynological analysis.
花粉形态的系统发育解释受到限制,因为我们无法识别花粉特征中所蕴含的进化历史。深度学习提供了将形态学与系统发育联系起来的工具。我们利用神经网络开发了一个明确的系统发育工具包,用于分析花粉粒的整体形状、内部结构和纹理。我们的分析流程基于不确定性估计来确定测试标本是否来自已知物种。分类不确定的标本的特征会被传递到一个多层感知器网络,该网络经过训练,可将这些特征转化为与已知分类单元的预测系统发育距离。我们利用这些预测距离,通过贝叶斯推断将标本置于系统发育树中。我们使用30个现存物种的光学超分辨率显微照片对模型进行训练和评估。然后,我们利用训练好的模型将九个化石标本置于系统发育树中。通过这样做,我们证明了神经网络能够识别花粉形态中编码的系统发育历史,并且深度学习特征可用于系统发育定位。我们的方法使得古孢粉学分析能够发现那些原本会被化石花粉记录有限的分类分辨率所掩盖的灭绝和物种形成事件。