Search for: All records

Over 140 million households cultivate rice worldwide. However, rice production contributes to exceedance of planetary boundaries, such as freshwater use and biogeochemical flows. In addition, rice-farming families may face increased risk of contracting schistosomiasis, a parasitic disease transmitted by freshwater snails that contributes to reinforcing cycles of poverty and disease. Here, using data from 405 households in rural northern Senegal, we show that children in rice-farming households had higher Schistosoma mansoni prevalence and Schistosoma haematobium intensities than non-farming peers. To address this environmental health challenge, we integrated native Nile tilapia (Oreochromis niloticus) and African bonytongue (Heterotis niloticus) into rice fields. The fish thrived, suppressed insects and snail pests, improved soil nutrients and boosted rice yields by >25% with a net benefit of US$1,805–3,415 ha−1 yr−1 (a benefit-to-cost ratio of 7.42). Hence, low-input rice–fish co-culturing offers a sustainable planetary health solution that simultaneously improves agricultural outcomes and rural livelihoods while reducing disease transmission risks. 

EXECUTIVE SUMMARY Training workshops for university students are important because they improve their knowledge and skills in the subject matter. Based on this premise, the study examined the before and after effects of leadership training workshops on the knowledge and skills of university students. The workshops covered fundamentals of leadership, leadership for advancing organizations, strategic planning, grant writing, and 501(c)(3) organizations. The objectives were to describe demographic characteristics, describe and assess other characteristics, conduct pre- and post-tests, and assess if there were any impacts because of the workshops. The data for the study were obtained from a group of university students on the campus of Tuskegee University in a series of training workshops focusing on the fundamentals of leadership, leadership for advancing organizations, strategic planning, grant writing, and 501(c)(3) organizations, in the first semester of the 2022/2023 academic year. Furthermore, the data were collected using pre- and post-tests and were analyzed using descriptive statistics and paired t-tests. The analyses were done using SPSS 12.0© (MapInfo Corporation, Troy, NY). The results revealed that there were more males than female participants; more Blacks than other races, and more sophomores than other classifications. Also, a majority had parents with a four-year college education than otherwise, more participants held a high school leadership position, and more participants held a college leadership position than otherwise. Moreover, the results of the paired t-tests for the various workshops showed statistically significant differences between the pre-and post-tests. The results indicate that the workshops impacted the knowledge and skills of the students, at least in the short term. The results of the study imply that when university students are provided with the appropriate training in a field, they are likely to obtain the requisite knowledge and skills. It is recommended that future leadership training be conducted to enhance the knowledge and skills of other students of identical classifications. Keywords: Before and After Effects, Knowledge and Skills, Leadership Training, Leadership Workshops, University Students 

Metalloenzyme superfamilies are typically defined by their protein scaffolds and active sites. Owing to the high tunability of protein structures, members of a single superfamily can catalyze diverse reactions with the same metallocofactor. Some superfamilies, such as Amidohydrolase-related dinuclear oxygenases (ARO), display further versatility by utilizing multiple metallocofactors. We have shown that certain AROs catalyze monooxygenation reactions with diiron, dimanganese and/or mixed manganese-iron cofactors, but the molecular factors governing the selection of a particular cofactor remain unknown and the extent of this superfamily in biology is unclear. Here, we report bioinformatic analyses that expand the ARO superfamily to approximately 17,000 unique UniProt sequences, far exceeding the number of previously characterized enzymes. Through the integration of structural, spectroscopic and thermodynamic analyses of representative proteins with a bioinformatic pipeline that identifies key secondary- and tertiary-sphere residues, we can predict in silico the metal preference for the majority of reported ARO sequences. These annotations were validated via the characterization of multiple new AROs, including ones implicated in key oxidative steps of natural product biosyntheses. This study establishes the key structure-function relationships governing metal preferences in AROs and highlights their vastly underappreciated role in myriad biological processes. 

Arctic rivers transport water, sediment and carbon, playing a central role in coastal stability and biogeochemical cycling. Although freshwater discharge to the Arctic Ocean has increased in recent decades, limited observations have hindered system-wide assessment of long-term, reach-level sediment dynamics. Here we develop a pan-Arctic-specific, satellite- and machine learning-based framework to reconstruct four decades of suspended sediment concentration dynamics for 4,331 river reaches. Our analysis reveals a significant increase in suspended sediment concentration in 40% of river reaches (858 out of 2,158) draining continuous permafrost zone, primarily driven by increasing discharge, intensified thermokarst disturbances and fires. The pan-Arctic land–ocean sediment flux averages 315 ± 33 Mt yr−1, with 198 ± 35 Mt yr−1 (63%) from the six major rivers (Yenisey, Lena, Ob’, Kolyma, Yukon and Mackenzie) and 117 ± 13 Mt yr−1 (37%) from 263 previously overlooked small- and medium-sized coastal rivers. The total land–ocean sediment flux has increased by ~15%, from ~299 ± 28 Mt yr−1 in the 1980s to 344 ± 29 Mt yr−1 in the 2010s. These results provide a baseline for pan-Arctic river sediment dynamics and underscore the essential yet underappreciated contribution of small- and medium-sized coastal rivers to Arctic landscape and carbon cycle changes. 

Production LLM serving must simultaneously deliver high throughput, low latency, and sufficient context capacity under non-stationary traffic and mixed request requirements. Data parallelism (DP) maximizes throughput by running independent replicas, while tensor parallelism (TP) reduces per-request latency and pools memory for long-context inference. However, existing serving stacks typically commit to a static parallelism configuration at deployment; adapting to bursts, priorities, or long-context requests is often disruptive and slow. We present Flying Serving, a vLLM-based system that enables online DP-TP switching without restarting engine workers. Flying Serving makes reconfiguration practical by virtualizing the state that would otherwise force data movement: (i) a zero-copy ModelWeights Manager that exposes TP shard views on demand, (ii) a KV Cache Adaptor that preserves request KV state across DP/TP layouts, (iii) an eagerly initialized Communicator Pool to amortize collective setup, and (iv) a deadlock-free scheduler that coordinates safe transitions under execution skew. Across three popular LLMs and realistic serving scenarios, Flying Serving improves performance by up to 4.79× under high load and 3.47× under low load while supporting latency- and memory-driven requests. 

Large language model (LLM) coding agents are increasingly applied to code translation and optimization, yet their effectiveness in performance-critical high-performance computing (HPC) settings remains poorly characterized. This paper evaluates LLM-based coding workflows on SZ-family errorbounded lossy compression kernels, which combine numerical constraints with memory-intensive and control-flow-heavy implementations. We study two representative CUDA workloads (SZp and SZx) and target two heterogeneous execution platforms: NVIDIA GPUs and Cerebras wafer-scale accelerators. Focusing on single-agent iterative generation, we analyze not only final throughput but also agent runtime behavior, including iteration patterns, sensitivity to prompt specification, and characteristic failure modes. Our results reveal a pronounced cross-architecture divergence. On GPUs, stronger models can achieve substantially higher throughput but exhibit increased sensitivity to prompt precision and optimization guidance, whereas on Cerebras the dominant challenge lies in producing runnable programs under a PE-centric spatial execution model. We further observe that LLM agents are more effective on modular kernels (SZx) than on tightly coupled bit-level pipelines (SZp), where structural dependencies hinder optimization progress. These findings suggest that evaluating LLM coding agents for HPC requires accounting for both performance outcomes and architecture-specific robustness, and that success on thread-based platforms does not directly transfer to spatial accelerators. 

Abstract Modification of river corridors, particularly deforestation and the removal of large wood, has greatly altered the abundance and influence of large wood in most rivers in the temperate latitudes. The conceptual framework of large wood process domains can assist in both directing research and facilitating large wood‐related management and restoration in rivers. Large wood process domains are spatially or temporally distinct portions of a river network or region with distinct processes of wood recruitment, transport, and storage. Previous research has shown wood to be unevenly distributed across space and time. We use a data set of logjam distribution density (# of channel‐spanning logjams/100 m length of channel) in 304 spatially distinct reaches of mountain streams in the Colorado Front Range, and up to 11 years of repeat measurements at some reaches, to (a) statistically evaluate whether a priori designated process domains for logjam distribution density are distinctly different and (b) evaluate the sensitivity of process domain delineations to spatial and temporal sample size. Our results indicate that the major spatial controls on logjam process domains for logjam distribution density in the Southern Rockies are drainage area, reach morphology, and wildfire disturbance history. Greater logjam distribution densities were present in wide reaches and undisturbed catchments. Using subsets of the data set composed of under 100 reaches created similar results. The relationship between geomorphic and hydrologic characteristics and their ability to describe logjam distribution density was minimally affected when using fewer than 10 years of data.