skip to main content

Title: Reproductive biology of harvestmen (Arachnida: Opiliones): a review of a rapidly evolving research field

Harvestmen are a major arachnid order that has experienced a dramatic increase in biological knowledge in the 21st century. The publication of the book Harvestmen: The Biology of Opiliones in 2007 stimulated the development of many behavioral studies. Although the book is relatively recent, our understanding of the reproductive biology of harvestmen is already outdated due to the fast accumulation of new data. Our goal is to provide an updated review of the subject to serve as a benchmark for the following years. In the pre-copulatory phase, we explore the evolution of facultative parthenogenesis, the factors that may affect the types of mating system, and the role of nuptial gifts in courtship. Regarding the copulatory phase, harvestmen are unique arachnids because they have aflagellate spermatozoa and a penis with complex morphology. We discuss the implications of these two features for sperm competition and cryptic female choice. In the post-copulatory phase, we connect oviposition site selection and climate conditions to the widespread occurrence of resource defense polygyny, alternative reproductive tactics, and sexual dimorphism in several clades of tropical harvestmen. Finally, we present the different forms of parental care in the order, and discuss the benefits and costs of this behavior, more » which can be performed either by females or males. Throughout the review, we indicate gaps in our knowledge and subjects that deserve further studies. Hopefully, the information synthesized here will stimulate researchers worldwide to embrace harvestmen as a study system and to improve our effort to unravel the mysteries of their reproductive biology.

« less
; ;
Publication Date:
Journal Name:
Current Zoology
Oxford University Press
Sponsoring Org:
National Science Foundation
More Like this
  1. Synopsis

    The lower reproductive tract of female mammals has several competing functions including mating, tract health maintenance, and parturition. Diverse vaginal anatomy suggests interactions between natural and sexual selection, yet despite its importance, female copulatory morphology remains under-studied. We undertook a comparative study across the species-rich mammalian order Chiroptera (bats) with a focus on the suborder Yangochiroptera (Vespertilioniformes) to examine how female vaginal features may have coevolved with male penis morphology to minimize mechanical damage to their tissues during copulation. The penis morphology is diverse, presenting great potential for post-copulatory sexual selection and coevolution with the female morphology, but vaginas have not been carefully examined. Here we test the hypotheses that vaginal thickness and collagen density have coevolved with features of the male penis, including the presence of spines and a baculum. We present histological data from females of 24 species from 7 families of bats, and corresponding data on male penis anatomy. We also examine the role of phylogenetic history in the morphological patterns we observe. We found evidence that female vaginal thickness has coevolved with the presence of penile spines, but not with baculum presence or width. Collagen density did not appear to covary with male penile features.more »Our findings highlight the importance of considering interactions between the sexes in influencing functional reproductive structures and examine how these structures have been under selection in bats.

    « less
  2. Abstract

    Quantification of genital variation in males and females can inform our understanding of likely copulatory interactions and evolution of genital diversity. However, no studies have quantified genital shape variation within a single snake species or examined the shape and size of both the vaginal pouch and hemipenes. Here, we examine the shape and size of the genitalia of female and male diamondback water snakes, Nerodia rhombifer, using a three-dimensional automated landmark geometric morphometric approach on models of the lumen of the vaginal pouch and inflated hemipenes, applying these techniques for the first time to the genital shape of vertebrates. Vaginal pouch shape is significantly associated with body size and reproductive status. As females grow larger and become reproductive, the vaginal pouch enlarges, widens and becomes more bifurcated. In reproductive males, the shape of the hemipenes is also significantly associated with body size. As males grow larger, the hemipenes enlarge and widen; their bifurcation becomes more defined and the spines at the base become more prominent. Vaginal pouch and hemipenial centroid size are isometric with respect to body length. The centroid sizes of the hemipenes and vaginal pouch are not significantly different from one another, hence the genitalia match inmore »size. Reproductive females and males covary in the degree of bifurcation and size of their genitalia. We demonstrate the utility of three-dimensional analysis in studies of the shape of soft tissues and advocate its use in future studies of genitalia.

    « less
  3. Abstract

    In live-bearing animal lineages, the evolution of the placenta is predicted to create an arena for genomic conflict during pregnancy, drive patterns of male sexual selection, and increase the rate of speciation. Here we test these predictions of the viviparity driven conflict hypothesis (VDCH) in live-bearing poecilid fishes, a group showing multiple independent origins of placentation and extreme variation in male sexually selected traits. As predicted, male sexually selected traits are only gained in lineages that lack placentas; while there is little or no influence of male traits on the evolution of placentas. Both results are consistent with the mode of female provisioning governing the evolution of male attributes. Moreover, it is the presence of male sexually selected traits (pre-copulatory), rather than placentation (post-copulatory), that are associated with higher rates of speciation. These results highlight a causal interaction between female reproductive mode, male sexual selection and the rate of speciation, suggesting a role for conflict in shaping diverse aspects of organismal biology.

  4. Obeid, I. (Ed.)
    The Neural Engineering Data Consortium (NEDC) is developing the Temple University Digital Pathology Corpus (TUDP), an open source database of high-resolution images from scanned pathology samples [1], as part of its National Science Foundation-funded Major Research Instrumentation grant titled “MRI: High Performance Digital Pathology Using Big Data and Machine Learning” [2]. The long-term goal of this project is to release one million images. We have currently scanned over 100,000 images and are in the process of annotating breast tissue data for our first official corpus release, v1.0.0. This release contains 3,505 annotated images of breast tissue including 74 patients with cancerous diagnoses (out of a total of 296 patients). In this poster, we will present an analysis of this corpus and discuss the challenges we have faced in efficiently producing high quality annotations of breast tissue. It is well known that state of the art algorithms in machine learning require vast amounts of data. Fields such as speech recognition [3], image recognition [4] and text processing [5] are able to deliver impressive performance with complex deep learning models because they have developed large corpora to support training of extremely high-dimensional models (e.g., billions of parameters). Other fields that do notmore »have access to such data resources must rely on techniques in which existing models can be adapted to new datasets [6]. A preliminary version of this breast corpus release was tested in a pilot study using a baseline machine learning system, ResNet18 [7], that leverages several open-source Python tools. The pilot corpus was divided into three sets: train, development, and evaluation. Portions of these slides were manually annotated [1] using the nine labels in Table 1 [8] to identify five to ten examples of pathological features on each slide. Not every pathological feature is annotated, meaning excluded areas can include focuses particular to these labels that are not used for training. A summary of the number of patches within each label is given in Table 2. To maintain a balanced training set, 1,000 patches of each label were used to train the machine learning model. Throughout all sets, only annotated patches were involved in model development. The performance of this model in identifying all the patches in the evaluation set can be seen in the confusion matrix of classification accuracy in Table 3. The highest performing labels were background, 97% correct identification, and artifact, 76% correct identification. A correlation exists between labels with more than 6,000 development patches and accurate performance on the evaluation set. Additionally, these results indicated a need to further refine the annotation of invasive ductal carcinoma (“indc”), inflammation (“infl”), nonneoplastic features (“nneo”), normal (“norm”) and suspicious (“susp”). This pilot experiment motivated changes to the corpus that will be discussed in detail in this poster presentation. To increase the accuracy of the machine learning model, we modified how we addressed underperforming labels. One common source of error arose with how non-background labels were converted into patches. Large areas of background within other labels were isolated within a patch resulting in connective tissue misrepresenting a non-background label. In response, the annotation overlay margins were revised to exclude benign connective tissue in non-background labels. Corresponding patient reports and supporting immunohistochemical stains further guided annotation reviews. The microscopic diagnoses given by the primary pathologist in these reports detail the pathological findings within each tissue site, but not within each specific slide. The microscopic diagnoses informed revisions specifically targeting annotated regions classified as cancerous, ensuring that the labels “indc” and “dcis” were used only in situations where a micropathologist diagnosed it as such. Further differentiation of cancerous and precancerous labels, as well as the location of their focus on a slide, could be accomplished with supplemental immunohistochemically (IHC) stained slides. When distinguishing whether a focus is a nonneoplastic feature versus a cancerous growth, pathologists employ antigen targeting stains to the tissue in question to confirm the diagnosis. For example, a nonneoplastic feature of usual ductal hyperplasia will display diffuse staining for cytokeratin 5 (CK5) and no diffuse staining for estrogen receptor (ER), while a cancerous growth of ductal carcinoma in situ will have negative or focally positive staining for CK5 and diffuse staining for ER [9]. Many tissue samples contain cancerous and non-cancerous features with morphological overlaps that cause variability between annotators. The informative fields IHC slides provide could play an integral role in machine model pathology diagnostics. Following the revisions made on all the annotations, a second experiment was run using ResNet18. Compared to the pilot study, an increase of model prediction accuracy was seen for the labels indc, infl, nneo, norm, and null. This increase is correlated with an increase in annotated area and annotation accuracy. Model performance in identifying the suspicious label decreased by 25% due to the decrease of 57% in the total annotated area described by this label. A summary of the model performance is given in Table 4, which shows the new prediction accuracy and the absolute change in error rate compared to Table 3. The breast tissue subset we are developing includes 3,505 annotated breast pathology slides from 296 patients. The average size of a scanned SVS file is 363 MB. The annotations are stored in an XML format. A CSV version of the annotation file is also available which provides a flat, or simple, annotation that is easy for machine learning researchers to access and interface to their systems. Each patient is identified by an anonymized medical reference number. Within each patient’s directory, one or more sessions are identified, also anonymized to the first of the month in which the sample was taken. These sessions are broken into groupings of tissue taken on that date (in this case, breast tissue). A deidentified patient report stored as a flat text file is also available. Within these slides there are a total of 16,971 total annotated regions with an average of 4.84 annotations per slide. Among those annotations, 8,035 are non-cancerous (normal, background, null, and artifact,) 6,222 are carcinogenic signs (inflammation, nonneoplastic and suspicious,) and 2,714 are cancerous labels (ductal carcinoma in situ and invasive ductal carcinoma in situ.) The individual patients are split up into three sets: train, development, and evaluation. Of the 74 cancerous patients, 20 were allotted for both the development and evaluation sets, while the remain 34 were allotted for train. The remaining 222 patients were split up to preserve the overall distribution of labels within the corpus. This was done in hope of creating control sets for comparable studies. Overall, the development and evaluation sets each have 80 patients, while the training set has 136 patients. In a related component of this project, slides from the Fox Chase Cancer Center (FCCC) Biosample Repository ( -facility) are being digitized in addition to slides provided by Temple University Hospital. This data includes 18 different types of tissue including approximately 38.5% urinary tissue and 16.5% gynecological tissue. These slides and the metadata provided with them are already anonymized and include diagnoses in a spreadsheet with sample and patient ID. We plan to release over 13,000 unannotated slides from the FCCC Corpus simultaneously with v1.0.0 of TUDP. Details of this release will also be discussed in this poster. Few digitally annotated databases of pathology samples like TUDP exist due to the extensive data collection and processing required. The breast corpus subset should be released by November 2021. By December 2021 we should also release the unannotated FCCC data. We are currently annotating urinary tract data as well. We expect to release about 5,600 processed TUH slides in this subset. We have an additional 53,000 unprocessed TUH slides digitized. Corpora of this size will stimulate the development of a new generation of deep learning technology. In clinical settings where resources are limited, an assistive diagnoses model could support pathologists’ workload and even help prioritize suspected cancerous cases. ACKNOWLEDGMENTS This material is supported by the National Science Foundation under grants nos. CNS-1726188 and 1925494. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation. REFERENCES [1] N. Shawki et al., “The Temple University Digital Pathology Corpus,” in Signal Processing in Medicine and Biology: Emerging Trends in Research and Applications, 1st ed., I. Obeid, I. Selesnick, and J. Picone, Eds. New York City, New York, USA: Springer, 2020, pp. 67 104. [2] J. Picone, T. Farkas, I. Obeid, and Y. Persidsky, “MRI: High Performance Digital Pathology Using Big Data and Machine Learning.” Major Research Instrumentation (MRI), Division of Computer and Network Systems, Award No. 1726188, January 1, 2018 – December 31, 2021. https://www. [3] A. Gulati et al., “Conformer: Convolution-augmented Transformer for Speech Recognition,” in Proceedings of the Annual Conference of the International Speech Communication Association (INTERSPEECH), 2020, pp. 5036-5040. [4] C.-J. Wu et al., “Machine Learning at Facebook: Understanding Inference at the Edge,” in Proceedings of the IEEE International Symposium on High Performance Computer Architecture (HPCA), 2019, pp. 331–344. [5] I. Caswell and B. Liang, “Recent Advances in Google Translate,” Google AI Blog: The latest from Google Research, 2020. [Online]. Available: [Accessed: 01-Aug-2021]. [6] V. Khalkhali, N. Shawki, V. Shah, M. Golmohammadi, I. Obeid, and J. Picone, “Low Latency Real-Time Seizure Detection Using Transfer Deep Learning,” in Proceedings of the IEEE Signal Processing in Medicine and Biology Symposium (SPMB), 2021, pp. 1 7. https://www.isip. [7] J. Picone, T. Farkas, I. Obeid, and Y. Persidsky, “MRI: High Performance Digital Pathology Using Big Data and Machine Learning,” Philadelphia, Pennsylvania, USA, 2020. [8] I. Hunt, S. Husain, J. Simons, I. Obeid, and J. Picone, “Recent Advances in the Temple University Digital Pathology Corpus,” in Proceedings of the IEEE Signal Processing in Medicine and Biology Symposium (SPMB), 2019, pp. 1–4. [9] A. P. Martinez, C. Cohen, K. Z. Hanley, and X. (Bill) Li, “Estrogen Receptor and Cytokeratin 5 Are Reliable Markers to Separate Usual Ductal Hyperplasia From Atypical Ductal Hyperplasia and Low-Grade Ductal Carcinoma In Situ,” Arch. Pathol. Lab. Med., vol. 140, no. 7, pp. 686–689, Apr. 2016.« less
  5. Abstract

    In her influential book “Developmental Plasticity and Evolution,” Mary Jane West-Eberhard introduced the concept of cross-sexual transfer, where traits expressed in one sex in an ancestral species become expressed in the other sex. Despite its potential ubiquity, we find that cross-sexual transfer has been under-studied and under-cited in the literature, with only a few experimental papers that have invoked the concept. Here, we aim to reintroduce cross-sexual transfer as a powerful framework for explaining sex variation and highlight its relevance in current studies on the evolution of sexual heteromorphism (different means or modes in trait values between the sexes). We discuss several exemplary studies of cross-sexual transfer that have been published in the past two decades, further building on West-Eberhard’s extensive review. We emphasize two scenarios as potential avenues of study, within-sex polymorphic and sex-role reversed species, and discuss the evolutionary and adaptive implications. Lastly, we propose future questions to expand our understanding of cross-sexual transfer, from nonhormonal mechanisms to the identification of broad taxonomic patterns. As evolutionary biologists increasingly recognize the nonbinary and often continuous nature of sexual heteromorphism, the cross-sexual framework has important utility for generating novel insights and perspectives on the evolution of sexual phenotypes acrossmore »diverse taxa.

    « less