Coexisting or not coexisiting? That's the....

Submitted by editor on 2 September 2016.Get the paper!


 A lot of species coexist. A lot don’t. Distinguishing between the two groups is a core component of community ecology. The inherent complexity of natural systems, and the quantity of data needed to fill in theoretical models, makes species coexistence particularly tricky to predict.

When we refer to the coexistence of species, we’re really talking about ‘stable coexistence’. Stable coexistence has several components. First, the species need to occur in the same places: co-occurring, as it were. But that’s not enough. The species need to be actively interacting with each other as well. Generally this means competition for resources, though any type of interaction is fine. We can check for this by seeing if the species co-occur at a scale where they’re likely to interact. So far, so good; these are very easy data to collect.

The final component of ‘stable coexistence’ is that both species exhibit no long-term population trends. This is the tricky bit. Populations always wobble a bit over time as conditions vary. So how long would we need to observe them for? More than one year? Definitely. Longer than a PhD or grant cycle? Almost certainly (for anything but very short lived organisms, like microbes). Even for short-lived plants, we’d need data from many generations to be sure (e.g., Figure 1). For late-successional rainforest trees, we could easily be staring down the barrel of several millennia of data collection! Ideally, as ecologists, we would like to make predictions about species and communities, so it would be unfortunate if decades of research were required in every system before we could start making predictions.

Thankfully, there are some alternatives. The theory behind species coexistence includes some fantastic models; models we can use in lieu of long-term data. For the most common framework (Chesson (2000) etc.), we need to quantify niche and fitness differences between the species, and test whether species can increase from rarity to stabilise their population. How do we do that? It’s a good question. Thankfully HilleRisLambers and her colleagues (2012) offer some suggestions.

Our first option is to conduct experiments with the species, manipulating limiting resources and seeing how the fitness of species changes. With this approach we could quantify niche and fitness differences, feed them into the theoretical model and get a prediction about stable coexistence. That prediction, however, would only be valid for the particular context (such as specific environmental conditions) in which we test the model. This is a problem: we know that sometimes species coexist because resources vary across the landscape (‘the storage effect’), but an experiment that includes all important spatial and temporal variation would be unfeasible.

Given this difficulty, another option is to look at the distributions of functional traits in communities and compare them to some ‘null’ or random expectations. This would solve the above problem, but raises issues of its own. First, we’re not entirely sure which functional traits represent species’ differences that permit stable coexistence; it could be very different in different environments (Mayfield and Levine 2010). Second, we’re only able to detect a signal using this method where the species’ functional traits differ. Sometimes, as in our recent study, species are functionally very similar and they still appear to coexist stably.

Are there any other options? There are many options in the recent literature. In the paper we recently published in Oikos, we take a multifaceted approach where we build a logical case for coexistence by assessing multiple lines of evidence. Essentially, we put on our ‘deerstalkers’ (a la Sherlock Holmes) and try to collect extensive circumstantial evidence. This is not a superior approach to the direct methods we discussed above; in fact, we’d argue it’s worse. It doesn’t allow us to make predictions using the robust coexistence models; it doesn’t let us make predictions full stop. It is, however, more feasible considering the type of ecological data that are commonly available and obtainable over normal granting and PhD cycles.

In our study, we look at the abiotic conditions each species is found in, we use a modest experiment to test for inherent fitness and niche differences, and look for differences in seed pollination and pollinator assemblages. Though we can’t conclusively state that our study species coexist, we feel we provide sufficient circumstantial evidence to satisfy reasonable doubt. That said, we leave it to you to act as the jury. Read our paper and return the most reasonable conviction.


References cited above:

Chesson, P. 2000. Mechanisms of maintenance of species diversity. - Annu. Rev. Ecol. Syst. 31: 343-366.

HilleRisLambers, J. et al. 2012. Rethinking Community Assembly through the Lens of Coexistence Theory. - Annual Review of Ecology, Evolution, and Systematics 43: 227-248.

Mayfield, M. M. and Levine, J. M. 2010. Opposing effects of competitive exclusion on the phylogenetic structure of communities. - Ecol. Lett. 13: 1085-1093.

Figure 1: Population size of two short-lived plant species (red and black) over time (years). Alternating grey and white bars show 5 years: a not-uncommon but considerable study period. Could we confidently predict the red species’ slow decline over this time period?


The beautiful winter annual wildflower communities of south-west Western Australia. This system is highly diverse, with on average 34 (± 1.7) species in a 30cm x 30cm quadrat. Photo credit: John Dwyer

Our study species co-occurring: Trachymene ornata (with white seeds) and Trachymene cyanopetala (purplish seeds). Photo credit: John Dwyer

The authors through Tim Staples

Insights into Oikos papers