Importance of accounting for context: a re-analysis as proof of
concept
To illustrate the importance of considering context when testing the
ERH, we re-analysed a recent synthesis (Xu et al. 2021). Xuet al. used the dataset of Turcotte et al. (2014), which
reports annual herbivory damage for a range of plant species. Xuet al. categorised each plant species in the dataset as ‘exotic’
or ‘native’ and compared their damage rates, accounting for plant growth
form and latitude of each observation. We repeated their analysis, but
also included information about enemy type, which was available in
Turcotte et al. (2014) (context = type of enemy). See
Supplementary Analysis for details.
We found that damage rates were generally higher on native than exotic
species (F1,1660=9.12, p=0.003, Fig. 6a; N=137
[exotics], N=1527 [natives]), consistent with the conclusion of
Xu et al. (2021). However, when we included the type of enemy as
a random variable, we found no difference in damage between native and
exotic species (χ21=0.13, p=0.719,
Fig. 6b). We suggest that these distinct findings can be largely
attributed to vertebrate impacts. Effects of vertebrates were only
recorded on natives (Fig. 6b; N=0 [exotics], N=76 [natives]),
and as vertebrates cause higher mean damage than other types of enemy in
the database, average damage on natives was disproportionately higher
than damage on exotics. In this case study, unless type of enemy is
explicitly accounted for, it can erroneously appear that overall damage
is lower on exotics, as found in Xu et al. (2021)’s study. In
fact, if only damage from insects is examined (N=84 [exotics],
N=1341 [natives]), damage appears higher on exotics than natives
(Fig. 6b), contradicting the ERH and the conclusion of Xu et al.(2021). This brief example demonstrates how evidence for the ERH is
contingent on context, and failure to account for context affects
conclusions. It also highlights that examining context can reveal data
gaps (here, vertebrate herbivory on exotic plants) that should be
prioritised in future work.