12 July 2019 in Evolution
21 April 2018 in Applied Artificial Intelligence
Statistical predictive models play an important role in learning analytics. In this work, we seek to harness the power of predictive modeling methodology for the development of an analytics framework in STEM student success efficacy studies. We develop novel predictive analytics tools to provide stakeholders automated and timely information to assess student performance toward a student success outcome, and to inform pedagogical decisions or intervention strategies. In particular, we take advantage of the random forest machine learning algorithm, proposing a number of innovations to identify key input thresholds, quantify the impact of inputs on student success, evaluate student success at benchmarks in a program of study, and obtain a student success score. The proposed machinery can also tailor information for advisers to identify the risk levels of individual students in efforts to enhance STEM persistence and STEM graduation success. We additionally present our predictive analytics pipeline, motivated by and illustrated in a particular STEM student success study at San Diego State University. We highlight the process of designing, implementing, validating, and deploying analytical tools or dashboards, and emphasize the advantage of leveraging the utilities of both statistical analyses and business intelligence tools in order to maximize functionality and computational capacity.
7 December 2022 in Evolutionary Applications
An essential goal in conservation biology is delineating population units that maximize the probability of species persisting into the future and adapting to future environmental change. However, future‐facing conservation concerns are often addressed using retrospective patterns that could be irrelevant. We recommend a novel landscape genomics framework for delineating future “Geminate Evolutionary Units” (GEUs) in a focal species: (1) identify loci under environmental selection, (2) model and map adaptive conservation units that may spawn future lineages, (3) forecast relative selection pressures on each future lineage, and (4) estimate their fitness and likelihood of persistence using geo‐genomic simulations. Using this process, we delineated conservation units for the Yosemite toad (Anaxyrus canorus), a U.S. federally threatened species that is highly vulnerable to climate change. We used a genome‐wide dataset, redundancy analysis, and Bayesian association methods to identify 24 candidate loci responding to climatic selection (R2ranging from 0.09 to 0.52), after controlling for demographic structure. Candidate loci included genes such as MAP3K5, involved in cellular response to environmental change. We then forecasted future genomic response to climate change using the multivariate machine learning algorithm Gradient Forests. Based on all available evidence, we found three GEUs in Yosemite National Park, reflecting contrasting adaptive optima: YF‐North (high winter snowpack with moderate summer rainfall), YF‐East (low to moderate snowpack with high summer rainfall), and YF‐Low‐Elevation (low snowpack and rainfall). Simulations under the RCP 8.5 climate change scenario suggest that the species will decline by 29% over 90 years, but the highly diverse YF‐East lineage will be least impacted for two reasons: (1) geographically it will be sheltered from the largest climatic selection pressures, and (2) its standing genetic diversity will promote a faster adaptive response. Our approach provides a comprehensive strategy for protecting imperiled non‐model species with genomic data alone and has wide applicability to other declining species.
1 January 2001 in Israel Journal of Zoology
8 September 2022 in Heredity
15 March 2024 in Molecular Ecology
Genomes are heterogeneous during the early stages of speciation, with small ‘islands’ of DNA appearing to reflect strong adaptive differences, surrounded by vast seas of relative homogeneity. As species diverge, secondary contact zones between them can act as an interface and selectively filter through advantageous alleles of hybrid origin. Such introgression is another important adaptive process, one that allows beneficial mosaics of recombinant DNA (‘rivers’) to flow from one species into another. Although genomic islands of divergence appear to be associated with reproductive isolation, and genomic rivers form by adaptive introgression, it is unknown whether islands and rivers tend to be the same or different loci. We examined three replicate secondary contact zones for the Yosemite toad (Anaxyrus canorus) using two genomic data sets and a morphometric data set to answer the questions: (1) How predictably different are islands and rivers, both in terms of genomic location and gene function? (2) Are the adaptive genetic trait loci underlying tadpole growth and development reliably islands, rivers or neither? We found that island and river loci have significant overlap within a contact zone, suggesting that some loci are first islands, and later are predictably converted into rivers. However, gene ontology enrichment analysis showed strong overlap in gene function unique to all island loci, suggesting predictability in overall gene pathways for islands. Genome‐wide association study outliers for tadpole development included LPIN3, a lipid metabolism gene potentially involved in climate change adaptation, that is island‐like for all three contact zones, but also appears to be introgressing (as a river) across one zone. Taken together, our results suggest that adaptive divergence and introgression may be more complementary forces than currently appreciated.