Post provided by Matthias Becher, Grace Twiston-Davies & Juliet Osborne

The BEEHAVE Team Osborne Becher and Twiston-Davies. Credit: Pete Kennedy.

Everyone, well, almost everyone, loves honey – that sweet, liquid gold laboriously collected by busy bees from countless little flowers. But of course, much more important than honey or wax or even cosmetic royal jelly products are the pollination services that bees provide to wildflowers and crops. In this blog post, authors Matthias Becher, Grace Twiston-Davies & Juliet Osborne discuss their latest paper published in Methods in Ecology & Evolution, “BEE-STEWARD: a research and decision support software for effective land management to promote bumblebee populations”.

Pollination is the effective transmission of pollen grains from one flower of a species to another flower of the same species, which enables fertilisation and finally seed production. Seed production allows the plants to reproduce – or humans to harvest them as energy-rich oilseeds or deliciously plump raspberries. No, humans would not die in a few years without bees – but our diets would be frustratingly dull.

While on the one hand, the demand for pollinating highly-profitable crops is on the rise, bee numbers, on the other hand, are on a dramatic decline. The loss of biodiversity caused by urbanisation and the intensification of modern agriculture leaves little nectar and pollen resources for pollinators to feed on. In addition to this nesting habitats are removed, excessive use of pesticides, and newly introduced parasites transmitting diseases take a further toll on pollinators.

Bee Bias and Population Declines

When in the mid-noughties ‘CCD’ hit the headlines, this mysterious Colony Collapse Disorder made honeybee workers disappear from the hive, leaving only the queen and young nurse bees behind to die, the public became more and more aware of the problems bees are facing – and increasingly concerned.

Red-tailed bumblebee (Bombus lapidarius). Credit: Matthias Becher.

While the focus was mainly on honeybees, we are now in a situation where wild bees also get the attention they deserve.  Especially bumblebees, those fluffy, colourful cuties, gain the publics sympathies. And thanks to their obvious stripe patterns, they are easy to identify, making them ideal subjects to citizen science projects like the UK’s Bee Walk survey scheme, organised by the Bumblebee Conservation Trust. With their long tongues, willingness to forage at low temperatures, and capability to collect pollen by ‘buzz pollination’, bumblebees can be highly efficient pollinators. But like so many other insect species, their numbers have been declining since the mid-20th century. Therefore, it is essential to understand which stressors have the most devastating impact on their population sizes and which screws we might turn to improve their situation.

Although empirical work is indispensable for this purpose and more data is desperately needed, particularly on the spatial and temporal distribution of resources, large-scale field studies are highly time and money consuming. As an alternative, in silico experiments, i.e. simulations of complex systems using computer models, can provide valuable services and allow us to test a multitude of scenarios once they have been developed. For example, implementing the findings of a laboratory study to assess how individual effects may or may not scale up to affect the whole population.


Over the past ten years, we have developed a family of sophisticated bee models which try to capture the complexity of a social bee colony and its interaction with a heterogeneous and ever-changing environment. In 2014, we published BEEHAVE, an agent-based model of a single honeybee colony which can take forage resources, weather, varroa mites and viruses, and pesticide effects into account. We wanted this model to be freely available and easily accessible by everyone including researchers, laypersons, beekeepers, and NGOs, and hence decided to implement it in NetLogo. NetLogo is a free and user-friendly programming language has been specifically developed with educational purposes in mind. Consequently, this model has since been applied by many research groups to study the impact of seasonal forage gaps, predation by Asian hornets, pesticide-induced bee mortalities and other factors. BEEHAVE was positively evaluated by an expert panel of the European Food Safety Authority (EFSA) and subsequently is used by EFSA for simulating the background variability of honey bee colony sizes on a European scale.

Foxglove (Digitalis purpurea). Credit: Matthias Becher.

Encouraged by the success of BEEHAVE we then developed Bumble-BEEHAVE a bumblebee counterpart of the original honeybee model, but addressing a whole population of bumblebee colonies and simulating their fate over many years. However, unlike the one Western honeybee species Apis mellifera, there are 24 bumblebee species in the UK alone and ca. 250 worldwide. Also, bumblebees show far less flower constancy than honeybees and due to their smaller colonies, smaller food patches may also be of more importance. This made it necessary to take flower resources in greater detail into account, as was the case with BEEHAVE. We updated our landscape module BEESCOUT to feed food patches composed of several flower species into the model. This created quite realistic patterns of the temporal and spatial nectar and pollen availability, which was derived from a habitat map. Still, the number of habitats we could define was limited.

While this approach was working and valid, it felt somewhat cumbersome to have to deal with two programs – BEESCOUT and Bumble-BEEHAVE – in parallel. Making small changes to the landscape could become a tedious exercise, the restriction to not more than nine habitat types often was a limitation and to add a new bumblebee species, we had to adjust the source code. In short, Bumble-BEEHAVE, together with BEESCOUT was a great model for researchers, but it was not the easily accessible, user-friendly tool that anyone could use which we wanted to see.


This, finally, led to the development of our new BEE-STEWARD tool, which is now published in Methods in Ecology and Evolution online. In order to fulfil the needs of our stakeholders, we closely worked with farmers, landowners and land advisors to make sure BEE-STEWARD benefits bees and business on the ground and to optimise the features of this tool while keeping the interface clean and tidy. With BEE-STEWARD we combined the landscape defining features of BEESCOUT with the bumblebee population model Bumble-BEEHAVE. We removed some of the previous limitations: new bumblebee species can now be easily defined in the input files, and the number of habitat types is no longer limited. Changing a landscape, e.g., adding margins around a field or creating a new patch of wildflowers, can now be done with a few mouse clicks. A simple report to compare two scenarios (e.g. with or without bee-friendly measures) can quickly be created, while more sophisticated ways of model analysis can still be applied.

We have tested the tool out with stakeholders and researchers through a range of projects to pilot and refine BEE-STEWRD so then it can be used for research, policy and practice. And, of course, BEE-STEWARD is freely available here and we would like to encourage everyone interested to give it a go! The model comes with detailed documentation and a manual. But if you are still stuck, drop us a line – we are happy to help!

To read the full Methods in Ecology and Evolution article, click on the following link: “BEE-STEWARD: A research and decision-support software for effective land management to promote bumblebee populations.”

To access the BEE-STEWARD tool, visit the BEEHAVE website here.