Data from: The island syndrome hypothesis is only partially validated in two rodent species in an inland-island system

According to the island syndrome and island rule hypotheses, island
isolation and reduced area lead to phenotypic shifts in ecologically
relevant traits in insular populations compared to mainland ones. These
hypotheses have been built up with oceanic islands in mind or islands
where isolation is high and colonization rate relatively limited. This set
of hypotheses, however, may not be applicable to other inland-island
systems or recently fragmented landscapes. We investigated how island life
leads to phenotypic changes in two rodent metapopulations: deer mice and
red-backed voles in a fragmented system on a river in Canada. From 2013 to
2016, we sampled 454 deer mice and 665 red-backed voles spread across 10
islands and six mainland sites on river shores. We compared body mass,
tail and hindfoot length, exploration, and reaction towards predators,
between individuals from island and mainland sites, and assessed the role
of connectivity, isolation and area of islands. Insular individuals from
both species were less aggressive towards potential predators and insular
mice were more thorough explorers compared to mainland ones. Male mice
were heavier and juvenile mice had longer tails on islands. However,
contrary to expectations, we found negative relationships between
aggressiveness and juvenile exploration score with patch connectivity for
mice. Island connectivity significantly affected tail length of mice
through a concave relationship. Finally, vole aggressiveness and
exploration of male mice were positively correlated to island isolation.
Our study supports only partially the predictions of the island syndrome
hypothesis. It provides empirical evidence that inland-island life can
modify traits in deer mice, but less so in red-backed voles in a
non-oceanic system. It also raises questions about the direct causal
factors of these changes in a naturally fragmented landscape, and urges
for more predictive models about phenotypic/genetic divergence among
populations in natural and artificial fragmented landscapes.