Arcadia University

Sexual selection theory traditionally emphasizes female mate choice, yet an expanding body of research highlights the prevalence and evolutionary importance of male mate selectivity. In species where ejaculate production is energetically costly, males may exhibit cryptic mate choice by modulating pericopulatory or postcopulatory investment in response to female quality. Here, we tested for cryptic male mate choice in the long-bodied cellar spider Pholcus phalangioides, a species in which males and females mate multiply and males repeatedly induct their pedipalps with sperm throughout their adulthoods. We mated virgin males with either large or small virgin females, then remated the same males with medium-sized virgin females after a recovery period. Offspring production from medium females served as a correlate for ejaculate investment. As predicted, males that first mated with large females had reduced reproductive success in subsequent matings than males whose initial mates were small, indicating a trade-off in ejaculate allocation. However, offspring survival did not differ between groups, suggesting that sperm quantity rather than quality was affected. Although alternative explanations for our results cannot be dismissed and should be considered for future studies, our results are consistent with the hypothesis that P. phalangioides males strategically allocate reproductive resources in response to female phenotype and face sperm limitation under repeated mating. Our findings provide evidence of cryptic male choice in spiders and underscore the fitness consequences of strategic ejaculate allocation in systems lacking paternal investment.

Estimating task progress requires long-horizon and dynamic reasoning, going beyond static visual perception. Although Vision-Language Models (VLMs) excel at describing what is visible in a single observation, it remains unclear whether they can infer how far a task has progressed from partial information. To study this question, we introduce Progress-Bench, a benchmark with over 3K instances for evaluating progress reasoning from a single observation. We further examine a human-inspired two-stage paradigm that combines episodic retrieval with mental simulation. We instantiate this paradigm through both training-free prompting and a training-based approach using the automatically curated ProgressLM-45K dataset. Experiments on 14 VLMs show that most models struggle with reliable progress estimation, and that training-free reasoning provides only limited and model-dependent benefits. In contrast, the training-based ProgressLM-3B achieves consistent improvements in accuracy, robustness to viewpoint variation, and handling of unanswerable cases, despite its small scale. Additional analyses reveal common failure patterns in existing VLMs and clarify when and why progress reasoning succeeds or fails.

The Microhaplotype Working Group (MWG) has made significant progress in defining key criteria for identifying microhaplotype (MH) loci that will enhance forensic genetics. The growing number of publications on MHs reflects increasing interest in these markers and highlights the need for greater standardization in the field. This paper summarizes the group's achievements since its formation following forensic community discussions at the 29th ISFG Congress in 2022. In this paper, we focus on locus nomenclature, allele definitions for MHs, and the challenges of marker identification and characterization. With the progress achieved on the core published MH locus database, MicroHapDB (https://github.com/bioforensics/MicroHapDB), it is now possible to use unique names for all currently published MHs. The MWG has reached a consensus on the critical marker characteristics for selecting MH loci for forensic use, with the effective number of alleles (Ae) metric as the primary selection parameter. Criteria for excluding otherwise suitable candidate MHs include genome location, length limitations, close physical linkage with forensic STRs, and sequence complexity considerations. Specifically, MHs in Long and Short Interspersed Nuclear Elements (LINEs and SINEs) and Long Terminal Repeats (LTRs) should be avoided, along with loci longer than ∼250 base pairs (bp). Finally, loci containing Indels or low complexity sequence patterns near their allele-defining SNPs should be excluded. MHs less than roughly 100 bp are potentially useful for typing degraded samples while those up to 250 bp are broadly useful and generally more informative if they include a higher average number of informative SNPs. The potential development of specialized MH panels for applications such as mixture deconvolution or biogeographic ancestry estimation is discussed. Considering these parameters, the MWG has identified 1148 loci theoretically suitable for forensic applications. The assembly of a final panel will maximize Ae while ensuring marker independence and loci with flanking sequence suitable for robust primer designs. This work serves as a foundation for the future selection of a core set of MH loci for forensic applications and development of recommendations for their implementation in forensic casework.

Causal inference from electronic health records (EHR) is fundamentally limited by unmeasured confounding: critical clinical states such as frailty, goals of care, and mental status are documented in free-text notes but absent from structured data. Large language models can extract these latent confounders as interpretable, structured covariates, yet how to effectively integrate them into causal estimation pipelines has not been systematically studied. Using the MIMIC-IV database with 21,859 sepsis patients, we compare seven covariate-integration strategies for estimating the effect of early vasopressor initiation on 28-day mortality, spanning tabular-only baselines, traditional NLP representations, and three LLM-augmented approaches. A central finding is that not all integration strategies are equally effective: directly augmenting the propensity score model with LLM covariates achieves the best performance, while dual-caliper matching on text-derived categorical distances restricts the donor pool and degrades estimation. In semi-synthetic experiments with known ground-truth effects, LLM-augmented propensity scores reduce estimation bias from 0.0143 to 0.0003 relative to tabular-only methods, and this advantage persists under substantial simulated extraction error. On real data, incorporating LLM-extracted covariates reduces the estimated treatment effect from 0.055 to 0.027, directionally consistent with the CLOVERS randomized trial, and a doubly robust estimator yielding 0.019 confirms the robustness of this finding. Our results offer practical guidance on when and how text-derived covariates improve causal estimation in critical care.