Discussion
Here, we experimentally test the interactive effects of multiple
disturbances on resident diversity, by exposing a bacterial community to
different regimes of pulse mortality and invader disturbances. A
challenge in understanding the effects of multiple disturbances is the
potential for different interactive mechanisms to operate at different
levels of disturbance (Foster et al. 2016). We found a
significant interaction between the disturbance types, such that at low
pulse disturbance frequencies invasion had a positive effect on total
resident diversity, whereas at high pulse disturbance invasion had a
negative effect. Consequently, our results demonstrate that, depending
on the disturbance regime, multiple disturbances can act both
antagonistically (have reduced combined effects on diversity) or
synergistically (have greater combined effects on diversity).
It is commonly hypothesised that co-occurring disturbances have
synergistic effects on diversity because communities that are disrupted
by one disturbance may be less stable, and therefore less resilient to a
second disturbance (Christensen et al. 2006; Crain et al.2008; Burton et al. 2020). However, evidence for such synergism
between disturbances is mixed, with previous work frequently finding no
or even antagonistic interactions between multiple disturbances (Darling
& Côté 2008). Here we show a possible reason for such mixed empirical
results: the direction of the interaction can depend on the regimes of
the disturbances involved, and synergistic interactions may only be
apparent in highly disturbed systems.
The dependence of the direction of the interactive effect on the
disturbance regime strongly suggests that the negative effects of
invasion on resident diversity might be missed at some disturbance
frequencies. This may result in an invader erroneously being categorised
as beneficial based on its positive effects at other disturbance
frequencies. That the effect of the invader is highly contingent on the
pulse disturbance regime may be particularly important for the
passenger, driver and back seat driver hypothesis, which categorises an
invader based on its effect on resident diversity (Didham et al.2005; MacDougall & Turkington 2005; Bauer 2012; Wilson & Pinno 2013;
Fenesi et al. 2015). A ‘driver’ invader changes ecosystem
properties and causes a decline in resident diversity independently from
other factors of change going on in that environment (MacDougall &
Turkington 2005; Wilson & Pinno 2013). In contrast, a ‘passenger’
invader takes advantage of available resources created by other causes
of change, such as disturbance, but does not cause biodiversity or
functionality loss themselves (MacDougall & Turkington 2005; Bauer
2012). Lastly, ‘back seat drivers’ act synergistically with other
factors of change – they require ecosystem change to establish, like a
passenger, but once established will themselves cause change, like a
driver (Berman et al. 2013; Fenesi et al. 2015). In our
experiment, the invaders acted as a back seat driver: they benefitted
from higher disturbance frequency (as this lowered resident densities
and offered more periods of reduced priority and dominance effects, and
hence reduced invasion resistance), and subsequently affected the
resident community. However, at low disturbance this may be missed, and
the invader erroneously categorised as a passenger due to having no
negative effect on the resident community. Similar phenomena may be
observed with other pulse disturbance types. This shift in apparent
invader effect at different levels of disturbance highlights the need to
consider the characteristics of the disturbance regime when drawing
inference, and raises concerns that invaders previously considered
harmless (passengers) may shift to being detrimental for resident
diversity (back seat drivers) as other disturbances increase.
Our finding of the pulse-invader interaction shifting from antagonistic
at low levels of pulse disturbance to synergistic at high levels
highlights the importance of looking at multiple disturbances over
different regimes. A recurrent theme in understanding the effects of
disturbances on diversity is that different disturbance aspects can
interact in a complex way to shape communities, and that our
interpretations of the disturbance-diversity relationships depend
crucially on the extent and resolution of the observations. Even under a
single disturbance type, a gradient of aspects of disturbances (e.g.,
frequency, severity) interact to produce various disturbance-diversity
relationships. This may help to resolve previously reported conflicting
patterns on disturbance-diversity relationships (Miller et al.2011; Hall et al. 2012). Similarly, previous analyses of the
interactive effects of multiple disturbances (Darling & Côté 2008) have
shown that the empirical patterns generated by interactions are mixed.
Here, we use multiple disturbance regimes to show that, as in previous
disturbance-diversity studies, a gradient of disturbance is crucial to
fully understanding the effect of simultaneously co-occurring
disturbances.
That the invader shifts to having severe negative consequences for
diversity at high mortality pulse disturbance, and even qualitatively
changes the pulse disturbance-diversity relationship, raises applied
concerns as the number, frequency and intensity of disturbances is
increasing with climate change (Essl et al. 2020). In our
experiment, one mechanism underlying the synergistic effect of high
disturbance in the invaded treatments is that the invader could have
been interacting similarly with each member of the resident community
(i.e. , reducing each resident species’ population equally). We
observed, however, that this is not the case; the invader only
negatively affects some species. This highlights that conservation
efforts need to consider a species’ vulnerability to invasion under
different disturbances and disturbance regimes, as well as that of the
entire community. For example, if just pulse mortality is considered in
our system, A. agilis would require greater resources to conserve
than P. corrugata as it is more vulnerable under a changing
disturbance regime. However, if we then additionally consider
vulnerability to exclusion by an invader in a situation with high pulse
mortality, P. corrugata would be regarded as more at risk thanA. agilis and therefore would receive more resources. One reason
for this interaction being particularly present in P. corrugatais that it is closely related to our invader, P. aeruginosa , and
so is more likely to compete for the same niche. We find that at low
pulse frequencies, P. corrugata benefits from a priority effect
and excludes P. aeruginosa . High frequencies, however, can reduce
this priority effect and consequently, P. aeruginosa (a better
competitor) can outcompete P. corrugata . This suggests that the
finding that invaders that are more related to residents have less of an
impact (Ricciardi & Atkinson 2004; Cox & Lima 2006; Saul & Jeschke
2015) may be in part due to priority effects, and may therefore not hold
when disturbance regimes change. These findings further suggest that
increasing disturbance may cause previously latent invaders to start
competitively driving diversity loss. Moreover, this in itself may act
as a disturbance, creating a positive feedback loop that facilitates
further invasion (Wonham et al. 2005).
The differences in vulnerability to invasions among the resident species
and the resulting changes in diversity imply that community robustness
against invasion may significantly change over different disturbance
regimes. This could be because frequent disturbances open up resources
and niches, and therefore reduce the competitive advantage of some
resident species over the invader (i.e., eliminate any priority or
dominance effects they have). It is also important to note that in a
stably coexisting community, such as the one studied here, an invader
may indirectly reduce a resident species’ fitness by outcompeting a
different species on which the resident relies.
Importantly, our findings can be applied to other natural microbial
communities. For example, frequent disturbances by antibiotics have been
shown to not only facilitate the invasion of clinically relevant
opportunistic pathogens, such as Clostridium difficile , into a
host microbiome, but also to have catastrophic effects for the diversity
of the resident community (Shah et al. 2021). The antibiotic and
pathogen may then interact synergistically and feedback to one another
to reduce microbiome diversity, and consequently may further increase
the risk of dysbiosis and infection (Blaser 2016; Ribeiro et al.2020), including from pathobionts (Ribeiro et al. 2020). Our
results suggest that this interaction may reduce diversity principally
by eliminating specific species – this could have particularly severe
consequences in the microbiome, where species can play specific roles in
host health and defence (Lange et al. 2016).
In conclusion, we provide experimental evidence that the interactive
effect of two disturbances on resident diversity is dependent on their
regime. Specifically, we find higher frequencies of pulse disturbance to
increase the success of an invader, and to lead to multiple extinctions
and a collapse in diversity, whereas at low pulse disturbance the
invader had a positive effect on resident diversity. These findings help
to create a predictive framework to understand how multiple disturbances
interact to affect diversity – a pressing issue under climate change.
Moreover, they demonstrate the importance of recording or deliberately
using different disturbance regimes when testing for the interactive
effects of multiple disturbances on diversity. Furthermore, we show that
invaders can aggravate the negative effect of high pulse frequency on
diversity. This raises serious concerns for global biodiversity, as the
global trend towards increasing disturbance frequency may amplify
negative effects of invasion worldwide, and the increasing number of
invasions may amplify the negative effect of frequent pulse disturbances
worldwide.