News and Views

Invasion ecology: Unpredictable arms race in a jam jar

Microcosm experiments show that post-invasion evolution of residents and invaders means invasive species effects are even harder to predict than previously thought.

We live in an invaded world. Non-native species have been introduced to every continent and countless islands. Predicting the fate of introductions is difficult, as many introduced species integrate into resident communities through a wide range of ecological interactions with residents including predation, competition and facilitation. In this issue of Nature Ecology & Evolution, Faillace and Morin perform an elegant ‘jam jar’ or microcosm experiment, demonstrating that post-invasion evolution of both invaders and residents may further contribute to this unpredictability1.

Invaders may evolve in response to the new environmental conditions they encounter in the non-native range2 and also in response to differences in species interactions3 in their new communities. However, resident communities are not all helpless victims of invaders as resident species can present ecological obstacles to invasion in the short-term4 and present evolutionary responses in the longer term. Faillace and Morin show that the evolutionary arms race between invader and invaded can have different, and unpredictable, consequences: for some communities the ecological impacts of invasion become less apparent over time but for others the impacts become more severe. Invasion microcosm studies like this enable comparisons that are just not possible in real world invasions and shed light on a wider range of potential outcomes from species introductions. We rarely get the chance to study why introductions fail and therefore have only a biased sample of successful invasions from which to generalize.

In their experiment, Faillace and Morin assembled aquatic bacteria, ciliate protists and rotifers into two different communities with one species from each community designated as the invader of the other. In each case the invaders were two functionally similar ciliate protists with the ability both to predate bacteria and to photosynthesize. The first phase of the experiment was to establish different interaction histories of the two communities over 200–400 generations and the second stage assessed the post-invasion abundance of residents and invaders exposed to different treatments. Invader species and resident communities were classed as either experienced (+I, +R) if they had a recent evolutionary scale interaction or naive (–I, –R) if they had no recent interaction history. All combinations of experienced and naive residents and invaders were used as treatments. Uninvaded controls of both communities were maintained throughout the experiment to enable comparisons with invasion scenarios.

Evolutionary experience might be expected to benefit both the invader and the residents, increasing the performance of the invader through time and increasing the resistance of the residents. Evolutionary experience for the invader should be advantageous, enabling invader populations to achieve higher abundances in naive resident communities than in experienced communities. By the same token, naive residents should perform more poorly against experienced invaders than against naive invaders. Evolutionary experience could also benefit the residents, with naive invaders doing relatively poorly in a community of experienced residents, and experienced residents outperforming naive residents in any invasion scenario.

In the experiment, experience benefited one of the invading protists, with highest abundance in the experienced invader/experienced resident scenario (+I/+R) analogous to a long-term invasion where both residents and invaders have had the opportunity to co-evolve. In fact this invader performed even better in the new community after a 200–400 generation interaction time compared with its performance in the original source community. In contrast the community to which the other invading protist was introduced fought back. Evolutionary experience of the recipient community with the invader increased the residents’ resistance to subsequent invasion. The responses of resident species in each assemblage were idiosyncratic with some species declining after prolonged interaction with the invader and some increasing. These idiosyncratic responses led to divergence in the community composition of each treatment. Interestingly the performance of the rotifers in each community was not affected by the invasion treatments.

Faillace and Morin's results show that long-term invader and resident performance may not be predictable, based on either performance in the source environment or the effects of invasions into naive recipient communities. This is not good news for invasion ecologists, as source environment performance5 or effects of recent invasion are often used to assess the risk of new invasions6. In these experiments naive invaders introduced to experienced resident communities performed worse than in their respective source communities. Invader performance subsequently improved after long-term experience within the community. We may therefore underestimate the performance of invaders if we use initial performance as a guide. Co-evolution within recipient communities may turn ‘sleepers’ (naturalized but not yet invasive species)7 into invaders. In Faillace and Morin's experiment, two resident species in the experienced resident/experienced invader treatment (+I/+R) declined drastically in abundance, an effect that was not detectable in the naive invasion scenario where neither invader nor resident community had recent prior experience of each other (–I/–R).

The elegance of microcosm invasion experiments comes at the cost of real world complexity. As Faillace and Morin have shown, the effects of invasion may be highly dependent on species identity, with sampling biases in introduction pathways contributing to the suite of traits of introduced species and effects of invasions in plant communities8. A fundamental disjoint between the jam jar and real world is that, in contrast to the microcosm where performance was assessed using abundance only, our identification of an ‘invasive’ species as distinct from an introduced species is not consistent. A recent study of plant invasions9, demonstrated that four demographic dimensions of invasion (local abundance, geographic range, environmental range and spread rate) are largely decoupled. Traits linked to one dimension (for example, local abundance) may have different effects on another dimension (for example, spread rate), making the discovery of a general trait syndrome for invaders, without distinguishing between different routes to invasiveness, difficult. More progress may be made in predicting the kinds of species likely to be problematic through their demography and impact by linking species traits to the different underlying ecological and evolutionary processes in play.

Faillace and Morin's work escalates the prediction problem further by asking how we might predict which invasion impacts will be mitigated through time by co-evolutionary processes and which impacts may persist or even grow. It still remains to be seen how our management efforts impose further selection pressures on residents and invaders10.


  1. 1.

    & Nat. Ecol. Evol. 1, 0013 (2016).

  2. 2.

    & Global Ecol. Biogeogr. 23, 1356–1365 (2014).

  3. 3.

    & New Phytol. 198, 916–924 (2013).

  4. 4.

    Science 288, 852–854 (2000).

  5. 5.

    et al. Ecol. Lett. 14, 274–281 (2011).

  6. 6.

    J. Appl. Ecol. 49, 10–19 (2012).

  7. 7.

    Euphytica 148, 111–120 (2006).

  8. 8.

    & J. Ecol. 104, 4–17 (2016).

  9. 9.

    et al. J. Ecol. 104, 1745–1758 (2016).

  10. 10.

    , , , & J. Ecol. 104, 182–192 (2016).

Download references

Author information


  1. Yvonne M. Buckley is in the School of Natural Sciences, Discipline of Zoology, Trinity College Dublin, Dublin 2, Ireland.

    • Yvonne M. Buckley


  1. Search for Yvonne M. Buckley in:

Corresponding author

Correspondence to Yvonne M. Buckley.