Lemmon, A. | Search bioRxiv

bioRxiv2018

Golden Orbweavers Ignore Biological Rules: Phylogenomic and Comparative Analyses Unravel a Complex Evolution of Sexual Size Dimorphism

Instances of sexual size dimorphism (SSD) provide the context for rigorous tests of biological rules of size evolution, such as Copes Rule (phyletic size increase), Renschs Rule (allometric patterns of male and female size), as well as male and female body size optima. In certain spider groups, such as the golden orbweavers (Nephilidae), extreme female-biased SSD (eSSD, female:male body length [≥] 2) is the norm. Nephilid genera construct webs of exaggerated proportions which can be aerial, arboricolous, or intermediate (hybrid). First, we established the backbone phylogeny of Nephilidae using 367 Anchored Hybrid Enrichment (AHE) markers, then combined these data with classical markers for a reference species-level phylogeny. Second, we used the phylogeny to test Cope and Renschs Rules, sex specific size optima, and the coevolution of web size, type, and features with female and male body size and their ratio, SSD. Male, but not female, size increases significantly over time, and refutes Copes Rule. Allometric analyses reject the converse, Renschs Rule. Male and female body sizes are uncorrelated. Female size evolution is random, but males evolve towards an optimum size (3.2-4.9 mm). Overall, female body size correlates positively with absolute web size. However, intermediate sized females build the largest webs (of the hybrid type), giant female Nephila and Trichonephila build smaller webs (of the aerial type), and the smallest females build the smallest webs (of the arboricolous type). We propose taxonomic changes based on the criteria of clade age, monophyly and exclusivity, classification information content, diagnosability, and arachnological community practice. We resurrect the family Nephilidae Simon 1894 that contains Clitaetra Simon 1889, the Cretaceous Geratonephila Poinar & Buckley 2012, Herennia Thorell 1877, Indoetra Kuntner 2006, new rank, Nephila Leach 1815, Nephilengys L. Koch 1872, Nephilingis Kuntner 2013, and Trichonephila Dahl 1911, new rank. We propose the new clade Orbipurae to contain Araneidae Clerck 1757, Phonognathidae Simon 1894, new rank, and Nephilidae. Nephilid female gigantism is a phylogenetically-ancient phenotype (over 100 ma), as is eSSD, though their magnitudes vary by lineage and, to some extent, biogeographically.

evolutionary biology↗

bioRxiv2017

Natural Language Processing for Classification of Acute, Communicable Findings on Unstructured Head CT Reports: Comparison of Neural Network and Non-Neural Machine Learning Techniques

Background and PurposeTo evaluate the accuracy of non-neural and neural network models to classify five categories (classes) of acute and communicable findings on unstructured head computed tomography (CT) reports.\n\nMaterials and MethodsThree radiologists annotated 1,400 head CT reports for language indicating the presence or absence of acute communicable findings (hemorrhage, stroke, hydrocephalus, and mass effect). This set was used to train, develop, and evaluate a non-neural classifier, support vector machine (SVM), in comparisons to two neural network models using convolutional neural networks (CNN) and neural attention model (NAM) Inter-rater agreement was computed using kappa statistics. Accuracy, receiver operated curves, and area under the curve were calculated and tabulated. P-values < 0.05 was significant and 95% confidence intervals were computed.\n\nResultsRadiologist agreement was 86-94% and Cohens kappa was 0.667-0.762 (substantial agreement). Accuracies of the CNN and NAM (range 0.90-0.94) were higher than SVM (range 0.88-0.92). NAM showed relatively equal accuracy with CNN for three classes, severity, mass effect, and hydrocephalus, higher accuracy for the acute bleed class, and lower accuracy for the acute stroke class. AUCs of all methods for all classes were above 0.92.\n\nConclusionsO_LINeural network models (CNN & NAM) generally had higher accuracies compared to the non-neural models (SVM) and have a range of accuracies that comparable to the inter-annotator agreement of three neuroradiologists.\nC_LIO_LIThe NAM method adds ability to hold the algorithm accountable for its classification via heat map generation, thereby adding an auditing feature to this neural network.\nC_LI\n\nAbbreviations

neuroscience↗