Biology subjects

Ogutu, J. O.

Publications and source records attributed to Ogutu, J. O..

8 recordsLinked to original sources

bioRxiv2026

Robust Random Forests for Genomic Prediction: Challenges and Remedies

Data contamination--from recording errors to extreme outliers--can compromise statistical models by biasing predictions, inflating prediction errors, and, in severe cases, destabilizing performance in high-dimensional settings. Although contamination can affect responses and covariates, we focus on response contamination and evaluate Random Forests through simulation. Using a synthetic animal-breeding dataset, we assess robust Random Forests across several contamination scenarios and validate them on plant and animal datasets. We thereby clarify the consequences of contamination for prediction, develop a robust Random Forest framework, and evaluate its performance. We examine preprocessing or data-transformation strategies, algorithmic modifications, and hybrid approaches for robustifying Random Forests. Across these approaches, data transformation emerges as the most effective strategy, delivering the strongest performance under contamination. This strategy is simple, general, and transferable to other Machine Learning methods, offering a remedy for robust genomic prediction. In real breeding data, robust Random Forests are useful when substantial contamination, phenotypic corruption, misrecording, or train-deployment mismatch is plausible and the goal is to recover a latent signal for genomic prediction and selection; in that setting, ranking-based and weighting-based robust Random Forests offer complementary remedies, the latter through a median-centred formulation that gives shrinkage a coherent interpretation across response distributions. Robustification is not universally necessary, but it becomes important when contamination distorts the link between observed responses and the predictive target; standard Random Forests remain the default for clean data, whereas robust Random Forests should be fitted alongside them whenever contamination is plausible, with the final choice guided by data, trait, and breeding objective. Author summaryMachine learning (ML) methods are widely used for prediction with high-dimensional, complex data, and supervised approaches such as Random Forests (RF) have proved effective for genomic prediction (GP) and selection. Yet their performance can be severely compromised by data contamination if the algorithms rely on classical data-driven procedures that are sensitive to atypical observations. Robustifying ML methods is therefore important both for improving predictive performance under contamination and for guiding their practical use in high-dimensional prediction problems. To address this need, we develop robust preprocessing, algorithm-level, and hybrid strategies for improving RF performance with contaminated data. Using simulated animal data, we show that ranking- and weighting-based robust RF provide the strongest overall compromise for genomic prediction and selection under contamination. Validation on several plant and animal breeding datasets further shows that the benefits of robustification are not universal, but depend on the dataset, trait, and breeding objective. Although motivated by RF, the framework we propose is general, practical, and readily transferable to other ML methods. It also offers a basis for deciding when robustness should complement standard RF rather than replace it outright.

bioinformatics↗

bioRxiv2024

Extensive population structure highlights an apparent paradox of stasis in the impala (Aepyceros melampus)

Impalas are unusual among bovids because they have remained morphologically similar over millions of years - a phenomenon referred to as evolutionary stasis. Here, we sequenced 119 whole genomes from the two extant subspecies of impala, the common (Aepyceros melampus melampus) and black-faced (A. m. petersi) impala. We investigated the evolutionary forces working within the species to explore how they might be associated with its evolutionary stasis as a taxon. Despite being one of the most abundant bovid species, we found low genetic diversity overall, and a phylogeographic signal of spatial expansion from southern to eastern Africa. Contrary to expectations under a scenario of evolutionary stasis, we found pronounced genetic structure between and within the two subspecies with indications of ancient, but not recent, gene flow. Black-faced impala and eastern African common impala populations had more runs of homozygosity than common impala in southern Africa, and, using a proxy for genetic load, we found that natural selection is working less efficiently in these populations compared to the southern African populations. Together with the fossil record, our results are consistent with a fixed-optimum model of evolutionary stasis, in which impalas in the southern African core of the range are able to stay near their evolutionary fitness optimum as a generalist ecotone species, whereas eastern African impalas may struggle to do so due to the effects of genetic drift and reduced adaptation to the local habitat, leading to recurrent local extinction in eastern Africa and re-colonization from the South.

evolutionary biology↗

bioRxiv2022

Identifying the potential for sustainable human wildlife coexistence by integrating willingness to coexist with habitat suitability models

Persistence of large mammals in the Anthropocene depends on human willingness to coexist with them, but this is rarely incorporated into habitat suitability or conservation priority assessments. We propose a framework that integrates human willingness-to-coexist with habitat suitability assessments to identify areas of high potential for sustainable coexistence. We demonstrate its applicability for elephants and rhinos in the socio-ecological system of Maasai Mara, Kenya, by integrating spatial distributions of peoples willingness-to-coexist based on Bayesian hierarchical models using 556 household interviews, with socio-ecological habitat suitability mapping validated with long-term elephant observations from aerial surveys. Willingness-to-coexist was higher if people had little personal experience with a species, and strongly reduced by experiencing a species as a threat to humans. The sustainable coexistence potential framework highlights areas of low socio-ecological suitability, and areas that require more effort to increase positive stakeholder engagement to achieve long-term persistence of large herbivores in human-dominated landscapes.

ecology↗

bioRxiv2022

Genomic prediction using machine learning: A comparison of the performance of regularized regression, ensemble, instance-based and deep learning methods on synthetic and empirical data

The accurate prediction of genomic breeding values is central to genomic selection in both plant and animal breeding studies. Genomic prediction involves the use of thousands of molecular markers spanning the entire genome and therefore requires methods able to efficiently handle high dimensional data. Not surprisingly, machine learning methods are becoming widely advocated for and used in genomic prediction studies. These methods encompass different groups of supervised and unsupervised learning methods. Although several studies have compared the predictive performances of individual methods, studies comparing the predictive performance of different groups of methods are rare. However, such studies are crucial for identifying (i) groups of methods with superior genomic predictive performance and assessing (ii) the merits and demerits of such groups of methods relative to each other and to the established classical methods. Here, we comparatively evaluate the genomic predictive performance and computational cost of several groups of supervised machine learning methods, specifically, regularized regression methods, deep, ensemble and instance-based learning algorithms, using one simulated animal breeding dataset and three empirical maize breeding datasets obtained from a commercial breeding program. Our results show that the relative predictive performance and computational expense of the groups of machine learning methods depend upon both the data and target traits and that for classical regularized methods, increasing model complexity can incur huge computational costs but does not necessarily always improve predictive accuracy. Thus, despite their greater complexity and computational burden, neither the adaptive nor the group regularized methods clearly improved upon the results of their simple regularized counterparts. This rules out selection of one procedure among machine learning methods for routine use in genomic prediction. The results also show that, because of their competitive predictive performance, computational efficiency, simplicity and therefore relatively few tuning parameters, the classical linear mixed model and regularized regression methods are likely to remain strong contenders for genomic prediction. The dependence of predictive performance and computational burden on target datasets and traits call for increasing investments in enhancing the computational efficiency of machine learning algorithms and computing resources. Author summaryMachine learning methods are well suited for efficiently handling high dimensional data. Particularly, supervised machine learning methods have been successfully used in genomic prediction or genome-enabled selection. However, their comparative predictive accuracy is still poorly understood, yet this is a critical issue in plant and animal breeding studies given that increasing methodological complexity can substantially increase computational complexity or cost. Here, we show that predictive performance is both data and target trait dependent thus ruling out selection of one method for routine use in genomic prediction. We also show that for this reason, relatively low computational complexity and competitive predictive performance, the classical linear mixed model approach and regularized regression methods remain strong contenders for genomic prediction.

genomics↗

bioRxiv2019

Spatial and temporal variation in small mammal abundance and diversity under protection, pastoralism and agriculture in the Serengeti Ecosystem, Tanzania

Land use is an important factor influencing animal abundance, species richness and diversity in both protected and human-dominated landscapes. Increase in human population and activities intensify changes in habitat structure and hence abundance, species richness and diversity. We investigated the influences of land use and seasonality on small mammal abundance, species richness and diversity in 10 habitat types distributed over protected, agricultural and pastoral landscapes in the Serengeti ecosystem in Tanzania. We used live traps (n = 141) and capture-recapture methods in each of 10 fixed plots distributed across three landscapes for a total of 28,200 trap nights of effort. Trapping was carried out in the wet and dry seasons for two consecutive years (April 2017 to October 2018). Small mammal abundance was higher in the pastoral than in the protected and in the agricultural landscape. Abundance was higher in the dry than the wet season across all the three landscapes. Species richness and diversity were higher in the protected, middling in the agricultural and lowest in the pastoral landscape. The high abundance in the pastoral landscape was due to the numerical dominance of two species, namely A. niloticus in the shrubland and M. natalensis in the cropland habitat, resulting in low species richness and diversity. Abundance was more evenly distributed across all habitats in the protected area due to less disturbance. The low abundance in the agricultural landscape, likely reflects disturbance from cultivation. High species richness and diversity in the protected area indicate high habitat heterogeneity while high species diversity in the agricultural landscape was likely due to high food availability during and soon after harvests. These findings emphasize the importance of protection in maintaining habitat heterogeneity for wildlife. They also reaffirm the need for buffer zones around protected areas to cushion them from intensifying human activities.

ecology↗

bioRxiv2019

Robust estimation of heritability and predictive accuracy in plant breeding: evaluation using simulation and empirical data

BackgroundGenomic prediction (GP) is used in animal and plant breeding to help identify the best genotypes for selection. One of the most important measures of the effectiveness and reliability of GP in plant breeding is predictive accuracy. An accurate estimate of this measure is thus central to GP. Moreover, regression models are the models of choice for analyzing field trial data in plant breeding. However, models that use the classical likelihood typically perform poorly, often resulting in biased parameter estimates, when their underlying assumptions are violated. This typically happens when data are contaminated with outliers. These biases often translate into inaccurate estimates of heritability and predictive accuracy, compromising the performance of GP. Since phenotypic data are susceptible to contamination, improving the methods for estimating heritability and predictive accuracy can enhance the performance of GP. Robust statistical methods provide an intuitively appealing and a theoretically well justified framework for overcoming some of the drawbacks of classical regression, most notably the departure from the normality assumption. We compare the performance of robust and classical approaches to two recently published methods for estimating heritability and predictive accuracy of GP using simulation of several plausible scenarios of random and block data contamination with outliers and commercial maize and rye breeding datasets.\n\nResultsThe robust approach generally performed as good as or better than the classical approach in phenotypic data analysis and in estimating the predictive accuracy of heritability and genomic prediction under both the random and block contamination scenarios. Notably, it consistently outperformed the classical approach under the random contamination scenario. Analyses of the empirical maize and rye datasets further reinforce the stability and reliability of the robust approach in the presence of outliers or missing data.\n\nConclusionsThe proposed robust approach enhances the predictive accuracy of heritability and genomic prediction while alleviating the need for performing outlier detection for a broad range of simulation scenarios and empirical breeding datasets. Accordingly, plant breeders should seriously consider regularly using the robust alongside the classical approach and increasing the number of replicates to three or more, to further enhance the accuracy of the robust approach.

genomics↗

bioRxiv2019

Wildebeest migration in East Africa: Status, threats and conservation measures

Migration of ungulates is under pressure worldwide from range contraction, habitat loss and degradation, anthropogenic barriers and poaching. Here, we synthesize and compare the extent of historical migrations of the white-bearded wildebeest (Connochaetes taurinus) to their contemporary status, in five premier East African ecosystems, namely the Serengeti-Mara, Masai Mara, Athi-Kaputiei, Amboseli and Tarangire-Manyara. The current status, threats to migration, migratory ranges and routes for wildebeest were characterized using colonial-era maps, literature reviews, GIS and aerial survey databases, GPS collared animals and interviews with long-term researchers. Interference with wildebeest migratory routes and dispersal ranges has stopped or severely threatens continuation of the historical migration patterns in all but the Serengeti-Mara ecosystem where the threat level is relatively lower. Wildebeest migration has collapsed in Athi-Kaputiei ecosystem and is facing enormous pressures from land subdivision, settlements and fences in Amboseli and Mara ecosystems and from cultivation in Tarangire-Manyara ecosystem. Land use change, primarily expansion in agriculture, roads, settlements and fencing, increasingly restrict migratory wildebeest from accessing traditional grazing resources in unprotected lands. Privatization of land tenure in group ranches in Kenya and settlement policy (villagization) in Tanzania have accelerated land subdivision, fencing and growth in permanent settlements, leading to loss of key wildebeest habitats including their migratory routes and wet season calving and feeding grounds. These processes, coupled with increasing human population pressures and climatic variability, are exerting tremendous pressures on wildebeest migrations. Urgent conservation interventions are necessary to conserve and protect the critical wildebeest habitats and migration routes in East Africa.

ecology↗

bioRxiv2019

Long-Term historical and Projected Herbivore Population Dynamics in Ngorongoro Crater, Tanzania

The Ngorongoro Crater is an intact caldera with an area of approximately 310 km2. Long term records on herbivore populations, vegetation and rainfall made it possible to analyze historic and project future herbivore population dynamics. In 1974 there was a perturbation in that resident Maasai and their livestock were removed from the Crater. Vegetation structure changed in 1967 from predominately short grassland to mid and tall grasses dominating in 1995. Even with a change in grassland structure, total herbivore biomass remained relatively stable from 1963 to 2012, implying that the crater has a stable multi-herbivore community. However, in 1974, Maasai pastoralists were removed from the Ngorongoro Crater and there were significant changes in population trends for some herbivore species. Buffalo, elephant and ostrich numbers increased significantly during 1974-2012. The zebra population was stable from 1963 to 2012 whereas numbers of other eight species declined substantially between 1974 and 2012 relative to their peak numbers during 1974-1976. Numbers of Grants and Thomsons gazelles, eland, kongoni, waterbuck (wet season only) declined significantly in the Crater in both seasons after 1974. Wildebeest numbers decreased in the Crater between 1974 and 2012 but this decrease was not statistically significant. In addition, some herbivore species were consistently more abundant inside the Crater during the wet than the dry season. This pattern was most evident for the large herbivore species requiring bulk forage, comprising buffalo, eland, and elephant. Analyses of rainfall indicated that there was a persistent annual cycle of 4.83 years. Herbivore population size was correlated with rainfall in both the wet and dry seasons. The relationships established between the time series of historic animal counts in the wet and dry seasons and lagged wet and dry season rainfall series were used to forecast the likely future trajectories of the wet and dry season population size for each species under three alternative climate change scenarios.

ecology↗