UC Merced Undergraduate Research Journal

Antibiotic resistance is especially rampant among Gram-negative ESKAPE pathogens due to their structurally protective outer membrane, adaptive metabolic functions, and selective porins that make them more resistant to antibiotics than Gram-positive bacteria, which lack an outer membrane and instead possess a thick, exposed peptidoglycan layer that is more easily targeted by many antibiotics. This study investigated bacterial colonies in soil from an urban garden to identify bacteria capable of producing antibiotics effective against Gram-negative bacteria. One isolate, designated Isolate #9, demonstrated a clear zone of inhibition against Escherichia coli, a safe relative of K. pneumoniae, a Gram-negative bacterium. Morphological, biochemical, and metabolic characterization revealed that Isolate #9 is a Gram-negative bacillus with catalase activity, gelatin hydrolysis, glucose fermentation, nitrate reduction, and optimal growth at 30 °C, while lacking phospholipase, amylase, oxidase, and pigment production. Antibiotic susceptibility testing showed that the isolate is sensitive to trimethoprim and rifampin but resistant to tetracycline, penicillin, and gramicidin. Trimethoprim inhibits DNA synthesis by blocking dihydrofolate reductase, while rifampin targets RNA synthesis by binding to RNA polymerase. Although 16S rRNA sequencing produced inconclusive BLAST results, the biochemical profile suggests potential affiliation with the genus Proteus. These findings suggest that soils in high-traffic areas may harbor previously uncharacterized Gram-negative bacteria capable of producing antimicrobial properties that can combat clinically significant antibiotic-resistant Gram-negative pathogens.

The surge of generative AI in education calls for new learning approaches resistant to cognitive offloading. Our work explores a solution inspired by the Latin proverb, docendo discimus: “by teaching, we learn.” We developed and piloted Teach2Learn, a web-based educational platform where CS1 students demonstrate their knowledge by teaching an LLM-simulated learner. This paper presents the pedagogical framework, design, and findings of two pilot tests (N = 87). We conducted pilots with two distinct student cohorts: an interdisciplinary summer cohort of non engineering majors (N = 32) with limited prior programming experience and a more homogeneous fall cohort of engineering majors (N = 55). Our preliminary results indicate a positive impact on student self-efficacy. In both pilots, the majority of the students reported an improved under standing of concepts after the activity and felt that the simulated student challenged their thinking. However, the impact on conceptual understanding, measured by exam-style questions, was mixed. The non-engineering cohort showed improved scores, while the engineering cohort’s performance decreased on several questions. The mixed results underscore the need for further iteration.

Fine particulate matter (PM2.5) exhibits pronounced spatial variability even across the scale of a university campus and readily infiltrates the lungs and bloodstream, exacerbating respiratory and cardiovascular health burdens. Recognizing this heterogeneity is essential for accurately characterizing exposure, particularly for the student population that may be unaware of its adverse impacts. The San Joaquin Valley, home to UC Merced, remains out of compliance with federal PM2.5 and ozone standards and has received an F grade from the American Lung Association. However, campus air quality is monitored by only one fixed PurpleAir sensor at the Science and Engineering 2 Building (SE2), which may not capture local PM2.5 variability. To address this gap, we undertook a mobile monitoring campaign during the 2025 spring term using a handheld EXTECH VPC300. Hundreds of geo-referenced readings were gathered along walkways, parking areas, construction zones, and other busy sites, then compared with daily values from the fixed PurpleAir sensor. The handheld and PurpleAir measurements exhibited similar day-to-day and weekly trends, indicating that the fixed PurpleAir unit effectively captures overall temporal variations in PM2.5. Midweek elevations compared to weekends likely reflect increased campus activity, such as traffic, construction, and maintenance. However, the fixed PurpleAir sensor cannot capture elevated PM2.5 near parking areas and construction zones detected by the handheld device. These findings suggest that while a single monitor can track general trends, a distributed network of low-cost sensors is needed to capture local exposure differences and guide targeted mitigation for the UC Merced community.

Flooding and irrigation uncertainty in the Upper Merced River watershed present serious challenges for water managers and farmers. This project investigates how snowmelt, precipitation, and dam operations interact to influence river overflow and water availability ,especially near Yosemite National Park,. By comparing a dry water year (2022) with a wet year (2023), the project combines remote sensing data, streamflow records, and dam release patterns to model potential flood risks and seasonal irrigation supply. High-resolution snow data from the Airborne Snow Observatory (ASO) [1], which uses LiDAR to measure snow water equivalent (SWE), revealed significant snowpack differences between years. ENVI software was used to visualize snowmelt rates using band math and custom color lenses, while precipitation records from the National Oceanic and Atmospheric Administration (NOAA) [3] and river flow data from the California Data Exchange Center (CDEC) [2] helped map hydrological trends. Dam operation reports from the Merced Irrigation District (MID) [4] were manually compiled into an operational timeline. Results showed that although 2023 had greater SWE, the melt was slower and better regulated by MID dams, reducing immediate flood risk. In contrast, 2022’s lower snowpack melted rapidly, overwhelming limited flow controls. These findings support the adoption of more adaptive irrigation planning and early-warning systems tied to snowmelt dynamics. While the model remains simplified, it demonstrates how open-source data and remote sensing tools can enhance regional water management, especially under intensifying climate variability.

In the lead-up to the 2024 presidential election, Donald Trump worked a 30-minute shift in a McDonald's kitchen in Buck County, Pennsylvania. This seemingly mundane publicity stunt at a McDonald's franchise reveals a complex narrative of class dynamics, political performance, and the ongoing struggle to connect with America's working class. This performative labor, which I define for the purposes of this essay as any activity which generates the appearance of busyness and production rather than true labor, becomes a microcosm of broader social tensions. This thus exposes the intricate ways political candidates negotiate their relationship with working-class identity and experience. This analysis seeks to unpack these social tensions between the American proletariat and U.S political entities by examining Trump's McDonald's shift alongside both Harris’ and Trump’s socioeconomic and political backgrounds using Marx and Engels’ ideas of worker alienation of labor and class consciousness.

2025AbstractAs automated driving technologies such as Adaptive Cruise Control (ACC) become commonin the automotive industry, the risk of chain-like crashes and degraded traffic flow increases,especially if string stability and vehicle safety is not ensured. Demonstrating and improving string stability is essential to advancing the design of ACC systems for smoother, safer, and more energy-efficient vehicle platoons. We study how parameter excitability in the regressor matrix influences accuracy and adaptability in ACC systems. When excitation is low, it reduces sensitivity and weakens parameter estimation, making the system less responsive to dynamic conditions. As a result, prediction reliability is compromised, and designing controllers that maintain string stability in actual traffic becomes difficult. We model the ACC system using an ordinary differential equation in which the acceleration of the ego vehicle depends on the spacing, relative velocity, and a constant time progress parameter. Online parameter estimation is performed using a Recursive Least Squares algorithm to capture dynamic changes. To evaluate the role of excitability in the matrix, we analyze the regressor matrix at each update step, quantifying excitability through condition numbers, and convergence of parameters. We

introduce diverse driving scenarios that simulate lead and ego vehicle interactions in real-world settings. These driving scenarios are simulated with a lead and ego vehicle velocity modeled in various scenarios: random walk in equilibrium, random walk in non equilibrium, induced curve, and aggressive lead vehicle. Our findings demonstrate that situations with little to no excitation, like random walk equilibrium, had difficulty achieving precise convergence because of a rank deficiency in the regressor matrix. Higher excitation scenarios, such as induced curves, aggressive lead drivers, and random walk (non-equilibrium), on the other hand, showed better convergence and reduced estimation error. In highway scenarios with extended constant speeds, limited excitation was also noted, which resulted in degraded trajectory prediction, parameter drift, and ill-conditioned regressors. In contrast, mixed-driving conditions with periodic excitation showed improved performance, maintaining estimator stability over long periods of time. Overall, these findings show that sustained excitation reflecting realistic traffic variability is necessary for both strong ACC performance and precise online parameter estimation in these driving scenarios.

Cancer is frequently regarded as a modern illness; however, its origins are intertwined with the evolutionary background of multicellular organisms. The persistence of cancer reflects a fundamental evolutionary challenge: how organisms maintain cooperation among billions of cells while preventing the rise of selfish, malignant ones. In this review paper, we will examine how natural selection influences the development of cancer suppression mechanisms within individual organisms and across various species. We will concentrate on two key areas: first, the Darwinian characteristics of cancer as a clonal evolutionary process driven by mutation, competition, and selection within the body; and second, the evolution of the immune and genetic defense system that inhibits tumor formation to a limit. The development of somatic defenses that have delayed cancer’s progression in long-lived species usually entails trade-offs such as aging, diminished regenerative abilities, and reduced reproductive capacity. We will also mention comparative studies that show that species such as naked mole rats, elephants, and whales have independently evolved distinct anti-cancer mechanisms, illustrating the varied approaches to a shared issue. Considering cancer from an evolutionary perspective can broaden understandings of its original traces and endurance. It also provides valuable insight into how organisms balance longevity and cellular integrity. Ultimately, integrating evolutionary theory with oncology may guide the development of more adaptive therapies and preventive approaches inspired by nature’s own long-term solutions.