Morphology, Life Cycle, Environmental Factors and Fitness – a Machine Learning Analysis in Kissing Bugs (Hemiptera, Reduviidae, Triatominae)

Abstract

Populations are permanently evolving and their evolution will influence their survival and reproduction, which will then alter demographic parameters. Several phenotypic, life history and environmental variables are known to be related to fitness measures. The goal of this article was to look into the possible types of those relationships in insects of the subfamily Triatominae, vectors of Trypanosoma cruzi, the causative agent of Chagas disease. After an exhaustive literature review of 7,207 records of publications referring exclusively to all possible features of the triatomines, using 15 keywords those records were reduced to 2,968 publications, that were analyzed individually; after deleting those publications that did not have the data in quantitative form as needed for the objective of this article, I found that 171 papers were adequate for the present analysis. From them I compiled a dataset of 11 variables and 90 cases from 36 triatomine species. Those variables included four environmental, two life cycle, and four morphological variables, and one demographic parameter: a fitness measure (the population intrinsic rate of natural increase, r0), used as dependent variable. However, the relationship between T. cruzi and its vector host was not included in this analysis despite triatomine-T. cruzi interactions constitute an important factor in the evolution of triatomine’s life history. I resorted to the Random Forest method as a machine learning approach for the analysis of this dataset, and found that –in addition to the triatomine species themselves– only the two life cycle variables (mean development time from egg to adult, and mean fecundity, expressed as the average number of female eggs laid per female per day) were statistically significant in determining fitness (r0). The machine learning approach used in the analysis provided a similar but deeper insight into these relationships than classical regression. Except for an analysis on senescence, this is the first study in triatomines addressing these questions. These results will be useful for other theoretical optimization approaches (frequency-dependence, density-dependence, evolutionary game theory, and adaptive dynamics), thus contributing to the theoretical framework for interpreting the succession of stages in insect adaptations, a framework yet to be constructed.

Precision Prediction for Dengue Fever in Singapore: A Machine Learning Approach Incorporating Meteorological Data

“…These models can capture complex non-linear relationships and interactions among various factors that influence vector-borne disease transmission dynamics. Recent studies have exemplified the effectiveness of machine learning in studying vector-borne diseases like malaria [ 31 ], Zika virus infections [ 32 ], and Chagas disease [ 33 ]. Nonetheless, challenges remain in data quality, interpretability, and the need for continuous model validation.…”

Challenges and opportunities in applying AI to evolutionary morphology

“…The end goal of many studies using this approach is to assign an ecomorphological characterisation to phenotypic traits and to parse their ecological signal (Barr, 2018). AI has been implemented in this field through the use of algorithms that infer present and past ecomorphologies by reducing the dimensionality of ecomorphological data through ML pipelines such as Random Forest analyses (Mahendiran et al, 2022;Rabinovich, 2021;Sosiak and Barden, 2021;Spradley et al, 2019). Similarly, ML procedures have been used to discriminate and sort phenotypes (especially morphology) based on their belonging to specific ecomorphs or ecological guilds (MacLeod et al, 2022).…”

Nutrigonometry I: using right-angle triangles to quantify nutritional trade-offs in multidimensional performance landscapes

“…We then made statistical inferences on peak area from 95% confidence intervals using the 'ci' function of the 'Rmisc' package (Hope et al 2013) whereby we resampled with replacement the selected random points obtained from steps 5 and 6 above. To test the performance of nutrigonometry in estimating nutritional trade-offs, we used the most commonly used models to test relationships between traits in behavioral ecology (e.g., general linear model), machine learning models used in regression models in ecology and evolution [e.g., Random Forest, (Rabinovich 2021)], as well as models that have been specifically used to analyse multidimensional performance landscapes in GF studies (e.g., SVM, GAMs) (Ponton et al 2015;Morimoto and Lihoreau 2019). In particular, we tested the performance of Bayesian linear regression (Bayes), general linear regression (LM), k-nearest neighbors (KNN), Gradient boost (GBoost), random forest (RF), support vector machine (SVM) with radial basis function as well as generalized additive models (GAMs) with both smooth term or tensor product term.…”

“…In the last decades, however, a method known as Geometric Framework of Nutrition (GF) has emerged as a powerful unifying framework capable of disentangling the multidimensional effects of nutrients (both ratios and concentrations) on life-history traits and fitness (Simpson and Raubenheimer 1993a). The GF has been applied to a diverse range of nutritional studies across species such as flies (Lee et al 2008;Reddiex et al 2013;Jensen et al 2015;Ponton et al 2015;Morimoto and Wigby 2016;Kutz et al 2019) ( Barragan-Fonseca et al 2018, 2021, crickets (Ng et al 2018) (Rapkin et al 2018) (Maklakov et al 2008), cockroaches (Bunning et al 2015), domestic cats and dogs (Hewson-Hughes et al 2011, and mice (Solon-Biet et al 2014;, being paramount for advancing our understanding of complex physiological and behavioural processes across ecological environments and even human health (Simpson et al 2017). As a result, developing a simple, intuitive, and accurate quantitative method for quantifying nutritional trade-offs has become a key issue for comparative nutrition, which will allow new avenues of research for insights into the evolution of physiological and behavioural modulation of nutritional responses (Morimoto and Lihoreau 2020).…”