Dismantling the Mantel tests

By Gilles Guillot, François Rousset. 2011

508 citations recorded by the source corpus.

The simple and partial Mantel tests are routinely used in many areas of evolutionary biology to assess the significance of the association between two or more matrices of distances relative to the same pairs of individuals or demes. Partial Mantel tests rather than simple Mantel tests are widely used to assess the relationship between two variables displaying some form of structure. We show that contrary to a widely shared belief, partial Mantel tests are not valid in this case, and their bias remains close to that of the simple Mantel test. We confirm that strong biases are expected under a sampling design and spatial correlation parameter drawn from an actual study. The Mantel tests should not be used in case autocorrelation is suspected in both variables compared under the null hypothesis. We outline alternative strategies. The R code used for our computer simulations is distributed as supporting material.

Subfields: Ecology, Evolution, Behavior and Systematics, Genetics, Nature and Landscape Conservation

View Paper: https://www.semanticscholar.org/paper/9b949e5b8be2a6344d66c3424a9dc96490183e91

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Ecology, Evolution, Behavior and Systematics

In Genetics

In Nature and Landscape Conservation

Citations in the corpus

Isolation by environment (2014, background): “However, when the data are spatially autocorrelated, the partial Mantel has a pathological type I error rate (Guillot & Rousset 2013), as the significance procedure implicitly rejects the spatial structure of the data.”
Testing environmental and genetic effects in the presence of spatial autocorrelation (2014, methodology): “The epidemiological study of Diggle et al. (2007) provides another realistic design, already used by Guillot and Rousset (2013) to illustrate the performance of partial Mantel tests.”
Robust Identification of Local Adaptation from Allele Frequencies (2012, background): “However, the partial mantel approach is known to perform very poorly when both genotypes and environmental variables are spatially autocorrelated (see Guillot and Rousset, 2013, for discussion).”
A practical guide to environmental association analysis in landscape genomics (2015, methodology): “Guillot & Rousset (2013) showed that, if there is spatial autocorrelation in the two matrices, Mantel tests result in P-values that are not well calibrated, because the permutation procedure fails to produce a valid null hypothesis.”
Landscape Genomics: Understanding Relationships Between Environmental Heterogeneity and Genomic Characteristics of Populations (2017, methodology): “Mantel tests have been severely criticized for various reasons, and alternative approaches should be used whenever possible (Balkenhol et al. 2009; Legendre and Fortin 2010; Guillot and Rousset 2013; Legendre et al. 2015).”
Testing the Mantel statistic with a spatially‐constrained permutation procedure (2019, methodology): “To sum up, as reported by Guillot and Rousset (2013), we confirmed that partial Mantel tests failed to adequately correct for the effect of spatial autocorrelation observed in Mantel tests.”
Should the Mantel test be used in spatial analysis? (2015, methodology): “…data that were spatially correlated, for example Manly (1986), Oden & Sokal (1992), Legendre, Borcard & Peres-Neto (2005), Legendre & Fortin (2010), Guillot & Rousset (2013) and section ‘Simulations involving spatially autocorrelated data: comparison of Mantel test and dbMEM analysis’ of this…”