Fossil fungal ascospores assignable to the genus Potamomyces have been recovered from Early Pliocene deposits at the Gray Fossil Site, a Fossil-Lagersta & uml;tte in Tennessee, USA. Characterized by four prominent equatorial verrucae, these ascospores closely resemble the fossil-species Potamomyces invaginatus, P. batii, and P. pontidiensis, as well as the extant species P. armatisporus. Modern species of Potamomyces are primarily saprotrophic fungi that occur on decaying wood in freshwater and brackish environments, as well as in moist terrestrial habitats. During the Pliocene, Potamomyces formed part of the late Neogene wetland ecosystem associated with sinkhole ponds at the Gray Fossil Site, which developed under a warm temperate to subtropical humid climate. The remains described here represent the northernmost known occurrence of Potamomyces in North America, based on both fossil and extant records. These findings further suggest that Potamomyces appears to have considerable potential as a non-pollen palynomorph proxy for reconstructing past environments and climates.
The interactions between vegetation and hydro-climatic factors play a critical role in key geophysical processes, including carbon and hydrological cycles. However, the interaction between vegetation and hydro-climatic factors in China remains unclear. Here, nonlinear Granger causality tests were employed to analyze the bidirectional relationships between vegetation and key hydro-climatic variables from 2002 to 2021. The results indicate that temperature was the dominant Granger-causal factor influencing vegetation growth (43.49 % of grid cells), followed by terrestrial water storage (16.49 %), soil moisture (11.44 %), precipitation (10.92 %), and solar radiation (3.11 %). In the reverse direction, vegetation exerted the strongest feedback on terrestrial water storage (31.18 %) and precipitation (28.77 %), with weaker effects on soil moisture (20.68 %), solar radiation (9.67 %), and temperature (3.79 %). Bidirectional Granger causality was observed in 33.57 %-46.77 % of the assessed areas, with grasslands showing the highest proportion of significant causal relationships. These findings improve our understanding between vegetation and hydro-climatic factors across China, providing valuable insights that can guide the refinement and optimization of these processes in ecological and hydrological models.
Agents that answer questions from compressed or retrieved memory must recognize when the evidence a query needs is no longer in memory. Benchmarks for this task usually create insufficient-evidence examples by deleting supporting passages. We show that this construction leaks the label through memory size: on MuSiQue, a classifier that only counts paragraphs reaches an area under the ROC curve (AUROC) of 0.979 for detecting unsafe memory, higher than the lexical estimator we initially evaluated. We propose a size-matched construction that provably removes this shortcut, and use it to study MemSafe, an estimator that cross-encodes the query with each memory unit and aggregates the units with a set transformer. Across three multi-hop question answering datasets and five seeds, MemSafe reaches 0.968 and 0.983 AUROC on MuSiQue and HotpotQA, 0.26 to 0.39 above a lexical baseline, while the third dataset, 2WikiMultiHopQA, is saturated. A frozen pretrained cross-encoder with a logistic head already closes 41% of the MuSiQue gap between the lexical baseline and MemSafe. At the same time, MemSafe degrades more than a weak baseline on the unanswerable questions released with MuSiQue, reaches only 0.639 AUROC on SQuAD 2.0, and needs several thousand clinical training examples before it outperforms a feature-based estimator. Used as a gate for a 7B reader, it reduces the error rate on answered questions from 0.850 to 0.631 at 5% coverage, outperforming both reader confidence and, on average, the ground-truth integrity label, although a 7B LLM judge is the better gate at 10% coverage. These results indicate that the way insufficient evidence is constructed matters as much as the estimator that detects it.
We report a metal-free, light-driven platform that expands amine chemical space using amino acid derivatives and simple alkanes as carbon-based synthons. The transformation demonstrates an underexplored sequential decarboxylative-dehydrogenative process, employing in situ generated aryl radicals to activate a broad range of C(sp3)-H bonds. Representative amino acids, including glycine, alanine, valine, and leucine derivatives undergo efficient decarboxylative coupling with activated and unactivated C(sp3)-H bonds affording structurally diverse amine products. This strategy underscores the untapped potential of simple next-generation bioderived and hydrocarbon building blocks for light-enabled C(sp3)-C(sp3) cross-couplings as a complementary strategy to reductive amination.
Caregivers often turn to online communities for informational and emotional support. In these spaces, peer supporters frequently draw on personal narratives to respond to emotionally complex caregiving situations. As LLMs are increasingly designed as peer-like sources of support, they introduce a critical tension: AI can provide immediate, private, and nonjudgmental support, but it cannot authentically possess the lived experiences that make human peer support meaningful. Yet, when prompted to sound peer-like, LLMs may generate language that implies lived experience. This creates a synthetic lived experience paradox: the same experiential language that may make AI support feel warm, relatable, and peer-like can also falsely position the system as someone with lived experience. We examine this paradox in the context of family caregivers of people living with Alzheimer's Disease and Related Dementias (ADRD). Drawing on caregiver support exchanges from online communities and prompted peer-like responses from three LLMs – LLaMA, GPT-4o-mini, and MedGemma – we analyze how human peers use personal narratives and how AI incorporates similar narrative forms. Psycholinguistic analysis shows that peer responses used significantly more first-person and past-focused language than peer-like AI responses. Qualitatively, we identify seven types of personal narratives in human peer support and show that AI often captures their emotional work, but can fabricate experiential grounding. These findings reveal a narrative authenticity gap: peer-like AI can generate synthetic lived experience without the real experience that makes peer support meaningful. We argue that caregiver-support AI systems need mechanisms to distinguish supportive peer-like framing from fabricated lived experience, ensuring that models can offer warmth and validation without falsely positioning themselves as experiential peers.
