Theory, meta-analyses and stylised facts in ecology

What is a theory? Is ecology theory-poor and if yes, why? What are the paths to theory development in ecology? Meta-analyses? Data syntheses? Big data? Stylised facts? These are the questions we set out to discuss during  this week’s lab meeting. We extended an invitation to EdGE (the EdEN discussion Group for Ecology) to get more diverse perspectives, and shared our thoughts on these topics, largely inspired by Dynamic Ecology’s posts about stylised facts in ecology and why meta-analyses in ecology often don’t lead to theoretical insight. We also added in Marquet et al.’s 2014 paper “On Theory in Ecology” into our discussion, bringing forward many thoughts on the different types of theory in ecology, and whether theory in ecology is possible to begin with.

We defined theory as a hierarchical framework of postulates, based on a number of assumptions, and leading to a set of predictions. As we set out to do our research, we can use theory as the base on which you build your hypotheses – and if you find enough support for your hypotheses, in time they might grow into a theory, thus prompting more hypotheses – a self-propelling cycle of gathering empirical evidence and developing theory. But is the cycle broken, with empirical evidence (or its synthesis) becoming an endpoint that prompts little theoretical insight?


We had a mix of undergraduates, PhD students and PIs in the room, and it was interesting to see how our thoughts varied based on our career stage. We started off with a quick quiz on 1) whether we had heard of the theories covered in the paper before, and 2) whether we had thought deeply about them. Here are the results!

How do we find out about theories in ecology to begin with? It was interesting to note that at least in the ecology curriculum here at the University of Edinburgh, most theories are taught pretty late (3rd and 4th year), and many don’t make it into the curriculum to begin with. How do we decide which theories are worth teaching about? Linking back to Marquet et al. 2014, should we be focusing on teaching the most efficient theories? Should we teach ecological theories in year 1?

From our experience, a lot of ecologists don’t like to think about theory too much – after all, ecology is so complex, are generalisations even possible? Some might say yes! We did, however, wonder what is the role of theory in ecology, if it seldom holds true across organisms, ecosystems, biomes. But then again, theories don’t need to be always right to be useful. Neutral theory, for example, can be thought of as a strawman idea that has spurred many interesting discussions (and research) on how reality differs from the simple pattern described by the theory.

We thought that while theories can be useful, a really strong emphasis on theory can bring you astray – stuck in mathematical equations and too far from the real world. According to the undergraduate participants in our discussion, theories are great for conceptualising ecological processes and thinking about how patterns can be generalised across time and space. We then discussed the difference between meta-analyses and data syntheses, with our group being predominantly being in preference of data syntheses – perhaps they are one of the paths towards the development of more ecological theory. Has that happened in the past? Yes! We used species-area curve relationships  that led to the development of the Island Biogeography theory as an example.

So why isn’t there more ecological theory? We thought of a simple answer – ecologists like to hang out outside. We briefly imagined what first year ecology students would say if when they showed up for the Field Ecology course, where you get to run up and down the Pentland Hills and collect data, we tell them that instead, we will be staying inside, thinking, doing lots of maths, and learning how to develop theory. Most of us went into ecology because we love the natural world and want to 1) learn more about it, and 2) experience it relatively often.  Fieldwork is the highlight of ecology for many of us (though for some of us it’s a tie between fieldwork and coding!), and that, together with all the noise in our data and the many complexities of our field, makes us less likely to engage deeply with theoretical work. Finally, most of us are not exposed to much math, especially at the start of our careers, which again makes it hard to think about how we can turn empirical evidence into theory.

Nevertheless, we are jealous of evolutionary biology, where theories abound! We talked about why that is, reaching the conclusion that theory prompts more theory – because in evolutionary biology there is one major unifying theory, other theories can quickly follow from that – a self-propagating cycle.

Are ecologists too critical? For every theory that tries to make its way, there most probably be someone who says that doesn’t apply to their study organism/system. We thought that we shouldn’t expect theories to always be true, instead we should use them as a stepping stone to build our future work.

Coming back to stylised facts, which may or may not lead to theory, we went around the room and each thought of a stylised fact from our field:

  • Plant growth is more temperature sensitive in wetter vs. drier sites (Soil moisture hypothesis, Myers-Smith et al. 2015Ackerman et al. 2016)
  • Biotic interactions shift from negative to positive with increasing environmental severity (Stress gradient hypothesis, Bertness and Callaway 1994)
  • Negative frequency dependence driven by higher trophic levels can maintain diversity (Paine 1966 and the Janzen–Connell hypothesis)
  • Phenology responses to global change drivers are stronger at lower throphic levels than higher (Thackeray et al. 2016).
  • Decomposition has a saturating relationship with temperature (e.g., Sierra et al. 2015).
  • Bigger and older trees are more prone to damage, increasing fungal infection rates (Basham 1958).
  • Plants with bigger foliar volume have more biomass (Greaves et al. 2015Cunliffe et al. 2016).
  • Remotely-sensed plant attributes can’t be accurately estimated at scales finer than the individual level (Cunliffe et al. in prep).
  • Big trees suffer more than little trees in rainforests experiencing drought (Rowland et al. 2015).
  • As the trait diversity of plant communities increases, so does the resource usage efficiency (Lasky et al. 2014).
  • Things that grow fast rot fast (Cornelissen et al. 2007).
  • The effect of agri-environment schemes is moderated by landscape complexity (e.g. Concepción et al. 2008).
  • The effectiveness of conservation interventions is proportionate to the ecological contrast they create (or their additionality) (e.g. Maron et al. 2013).

We finished our discussion where we started – at the definition of what stylised facts are, and whether there is one universal definition – thus showing that ecologists do care about generalisation!

The role of β-diversity in conservation

What indicators should we use in conservation? Why do different biodiversity indicators seem to disagree? What is the role of beta-diversity in conservation? This week we extended our usual TeamShrub lab meeting to hold a discussion on two recent biodiversity papers, as part of the EdEN (Edinburgh Ecology Network) EdGE (EdEN Discussion Group for Ecology) meetings. We talked about what are the best indicators to assess biodiversity change, whether there is a place for β-diversity metrics in guiding conservation actions, and why do different indicators of biodiversity change seem to disagree with one another.

We all had an interesting and jolly discussion, inspired by the following papers:

Socolar, Jacob B., et al. “How should beta-diversity inform biodiversity conservation?.” Trends in ecology & evolution 31.1 (2016): 67-80.

Hill, S. L.L., Harfoot, M., Purvis, A., Purves, D. W., Collen, B., Newbold, T., Burgess, N. D. and Mace, G. M. (2016), Reconciling Biodiversity Indicators to Guide Understanding and Action. CONSERVATION LETTERS, 9: 405–412. doi:10.1111/conl.12291


As we work in the Arctic, we appreciated how the papers recognised the fact that regions which are not particularly rich in biodiversity still deserve to be on the conservation radar.

We started off by identifying what β-diversity is and how we measure it – we discussed temporal β-diversity (how has species composition changed through time) and spatial β-diversity (commonly known as just beta-diversity, how do communities differ across space – i.e. measures of similarities, etc.) and what are the implications of using β-diversity metrics in conservation. We can mostly agree that one of the goals of conservation is to maximise biodiversity, but what diversity? Alpha, beta, gamma?

Unlike α-diversity (diversity at the local scale) and γ-diversity (diversity at the global scale), β-diversity does not refer to a spatial extent, but to the comparison between communities, and as such is is often used as an indicator of biotic homogenisation.

Calculating β-diversity allows us to understand biodiversity loss from a different perspective – we can look beyond species richness increasing or decreasing, and think about whether communities are becoming more similar, and what the implications of that might be for ecosystem functionality and the provision of ecosystem services. Nevertheless, β-diversity has to be used carefully – if two communities are both changing, β-diversity might stay the same (i.e. they might still have the same amount of species in common), but their current species composition might have changed. We also discussed how increasing the spatial extent of agri-environment management (or other conservation measures) might not always have the desired outcomes – such actions might decrease β-diversity by favouring the same set of species over large spatial extents. Communities can shift in many ways, which don’t necessarily fit in the biodiversity loss toolbox we most often use.

Can we use beta-diversity to link local scale observations to global scale inferences on biodiversity trends?

We thought that this is theoretically a great idea, but logistically, there are difficulties in going from the local scale observations to inferences on γ-diversity – gaps in the data, understudied regions, etc. We also pondered the dangers of promoting rare species at the expense of common species, and also what about disturbance-tolerating species? It is easy to say that e.g. Plot1 has lost/gained one species, but hard to have confidence in how the world has changed over time. Perhaps it is β-diversity that will help us link our local-scale observations to inferences on the global scale.

By Gergana


Our BES Annual Meeting highlights

Christmas arrived early for Jakob and I – we attended the British Ecological Society Annual Meeting in Liverpool at the start of December and got to enjoy the jolly and festive atmosphere of hundreds of ecologists keen to share their research. Having just wrapped up Coding Club for 2016, it seemed very appropriate to start the BES conference with their Best Practices for Code Archiving workshop. I was very keen to learn more about code archiving, but I also wanted to see how other people teach coding and organise workshops. I will be leading a GitHub and version control workshop for Coding Club soon, and I’m looking forward to sharing the knowledge and skills I gained at BES with the Coding Club members.

czacjgzwiaalapbThe Coding workshop communicated a great message – writing reproducible code and archiving it is not hard. I wouldn’t say it’s easy, but it’s certainly not hard, either, and it’s something we all could (should) do. I particularly liked this graph, as it conveys what we’ve been trying to tell Coding Club members (and ourselves) – little investments in learning good coding practices can deliver big benefits. It’s even better if there is a community around you which is also keen to learn. When I was wrapping up my dissertation, I was feeling a bit intimidated by Markdown and decided that I couldn’t learn it in a day (since the dissertation was due the following day). This year, I went to a Markdown workshop led by PhD students from the Australian National University, and after and hour, I could make a beautiful report of my code and results. It was great to see students pick up Markdown quickly at the Coding Club workshop, too – if you are keen to learn it as well, check out our online tutorial here. Similarly, syncing RStudio with GitHub doesn’t take long, and is a great way to keep track of your code and its many versions (you can check out Coding Club’s Github tutorial here).

There were many great talks in the following days, and I was particularly impressed by the PhD talks. It was so interesting to learn about Francesca Mancini‘s interdisciplinary conservation research on how social media images can be used to infer eco-tourism hotspots – and good to know that my photos on Flickr could have been used for science! Sarah Scriven gave a very well presented talk on butterfly movement through oil palm plantations (and in the spirit of #BEScode, the data and code are publicly available). Great to highlight not just connectivity between fragmented landscapes, but also functional connectivity – even if butterflies can move through agricultural land, their larval host plants might not occur there, thus preventing breeding.

czjqnpewiaa8hnmThe conference was great at prompting me to think about topics that hadn’t crossed my mind before – for example, whether mountain bikes are good at seed dispersal, and what implications that might have for plant colonisation and perhaps even the spread of invasive species. Prof Anne Chao delivered a fantastic plenary talk on biodiversity estimators and rarefaction, from which we all learned a lot, and I’ve put down exploring the R package, iNEXT, as a reward for after I’ve finished the first draft of my manuscript. You can also check out their Shiny app online, where you can upload your own data and get plots and values for the different biodiversity estimators. I also got to meet Prof Chao at the Meet the Speaker session after the plenary talk, which only inspired me further for biodiversity analyses.

Chatting with keen ecologists over lunch was super fun (and the food was tasty, too!), the Science Comedy Slam was hilarious, and in general we loved the dynamic of the whole conference – we came home inspired and thankful that we get to be a part of the ecology research community.

By Gergana and Jakob