Close Encounters Computational Models

Computer simulation highlights the importance of pollinator diversity

Sharing pollinators can cause problems when pollen competes for transport

How many pollinators does an ecosystem need? Does it matter if a few specialist pollinators are lost if there are enough generalists to keep browsing from flower to flower? A simulation published by Alan Dorin and colleagues in Theoretical Ecology shows that when pollen competes for pollinators, plants can have a lot to lose.

It might seem obvious that plants want specialist pollinators. Flowers produce pollen, but if it doesn’t arrive at a suitable partner it can’t fertilise a flower of another species. “Considering the plant that provided the pollen, this is potentially a wasted opportunity for reproduction since its pollen is seemingly lost for no reproductive gain,” write Doran and colleagues. “Our simulations reveal, however, how lost pollen may actually provide a competitive advantage by inhibiting the reproduction of a plant’s competitors, even if the competitor occupies a refuge zone immune from physical invasion by the pollen donor at that time.”

In the real world this kind of experiment would be insanely complex to conduct. However, using agent-based simulation it becomes possible to run the same experiment multiple times to account for random effects. Doran and colleagues set up a landscape in their simulation 200 × 200 squares big with two plants in competition with each other. At each side of the landscape were 40 × 200 strips, these were refugia that the opposing plant could not enter. The central zone of 120 × 200 squares was a region where either plant could take control. It was into this landscape that the virtual pollinators were released.

A virtual pollinator above a stripey field of flowers.
Image: Canva

The pollinators were quite simple. “Each pollinator tracks its current position and heading as it moves. It carries pollen collected from flower visits and main- tains a memory of its five most recently visited flowers. It will not revisit any flower on this list, in keeping with empirical data on bee short-term memory, scent marking, and foraging behaviour…,” write Doran and colleagues. The pollinators have one of two strategies to gather pollen. Either Forage Nearest flower or Forage Any flower.”

The plants were stocked with pollen grains. Each time a pollinator visited they for a fixed number of anther grains, and deposits some grains to the stigma of the flower. Importantly, the model keeps track of whether the plant receives conspecific or heterospecific grains. Only conspecific grains can pollinate the plant. This is crucial because once the model has run through the pollination phase, it then runs through a reproduction phase.

The viable plants produce seeds which scatter slightly. Then the plants are wiped from the simulation. Next the seeds become plants. Any plant in the opponent’s refugium gets deleted as ‘unviable’. A new population of pollinators is generated and assigned to random locations and the cycle begins again. After a number of loops you can see what happens over time.

The next element of the simulation is to change the conditions. Doran and colleagues did this by changing how pollen competes with each other, or by removing the refugia. The team considered three ways pollen could compete. In the first scenario, it didn’t. In the second and third scenarios ‘clogging’ happened.

Clogging is what happens when pollen from another plant gets in the way of pollen that could fertilise an ovule. So in the simulation if pollen from the wrong plant arrived first then clogging could happen. In one scenario clogging was one-way, so X could clog Y but not the other way round. In the other scenario both plants’ pollen could clog, so both plants could lose reproductive opportunities.

When the team ran the simulations, the results were striking.

Without clogging both plants usually co-existed. Most of the time both plants were around after 1000 generations. But not always. 14% of the time one of the plants was unlucky and eradicated. However, even in these results eradication took at least 400 generations to happen. When clogging was permitted, it could happen a lot faster.

When the simulation was set up for pollinators to forage on any flower, and one-way clogging, one species was always eliminated. On average it would take fewer than 17 generations, and the loser would be evicted from its refugium, even though the other plant could not enter. In the case of two-way clogging it took fewer than 20 generations, on average for one plant to disappear.

“Our results support the hypothesis that heterospecific pollen deposition has the potential to act as an “infertility bomb”. Under some circumstances, this may drive the exclusion of a competitor from a region that both plant species could otherwise potentially co-inhabit.” write Doran and colleagues. “Our simulation results suggest the extent that pollen misdelivery might need to be factored into our understanding of plant competition, even though the periods available for typical ecological studies might be too short to capture the entire sequence of events directly.”

Doran and colleagues note there are all sorts of evolutionary pressures on flowers, so there’s not a drive purely for pollen competition. However they argue that when there’s one-way clogging, pollen competition could be significant. Their results add some chronological depth to a proposal by Alexander Suárez‐Mariño and colleagues in an article in the American Journal of Botany from last year. They argued the success of some invasive plants could be due to their pollen blocking native plants. Their study examined the tolerance of Bidens pilosa, black-jack, to heterospecific pollen compared to native plants.

Together the two studies indicate that losing a specific pollinator could be terminal for a plant population, even if in the short term it can share a pollinator with its neighbours.

You can read the article for free via ReadCube at https://rdcu.be/cbmOE and if you’d like to run the simulation yourself, the source code is Open Access via GitHub at https://github.com/tim-taylor/evobee.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.