16.02.2023

New technology enabling the automated monitoring of moths has been put to rigorous testing in tropical conditions in Panama by an international team including researchers from UKCEH. Dr Tom August explains more...

The rustling sounds of howler monkeys passing through the canopy is abruptly interrupted by a wail of excitement as one of our entomologists spots a butterfly flitting through the dappled light, “WOW! Oh boy! Did you see that?!”

Our team was in Panama to test new technology for monitoring moths during the night, but in the day we made the most of our time in one of the most biodiverse regions of the world.

Back at the research station we set about assembling our moth monitoring systems. Batteries on charge, SD cards readied, and we all try to keep our expectations in check. After all, this is the first time anyone has tried any of these automated insect monitoring systems in the tropics where the heat and humidity is intense.

Our home for the week was Barro Colorado Island in the middle of the Panama Canal. The US-based Smithsonian Tropical Research Institute has been doing research here since the 1920s and it has everything needed for a successful trip. Accommodation, canteen, and a diverse and well-studied community of moth species.

The research team from Europe and North America have been motivated to explore new ways to monitor insect populations in the wake of shocking reports in recent years of large, long-term declines in insects as a result of human activity. Bringing together engineers, entomologists, computer vision experts, and data specialists, we were in Panama to work out if we can develop automated systems for monitoring insects. The systems use solar power to run for months at a time, and use the latest AI to analyse thousands of images collected each night.

 

Lights on the systems are crucial for attracting moths.

Image
Equipment for automated monitoring of moths in a forest location

Powering the units in closed canopy forest, where solar panels are ineffective, will require some innovation of low energy components and low intensity sampling designs.

Early nerves dissipate as the results of the first night come in. Aside from a few minor glitches the systems are working well, and the diversity of moths we are observing on the cameras is impressive. It is immediately clear that the lights on the systems are critical. Moths are attracted to a range of colours of light, but are particularly attracted to UV light and the budget lights simply don’t cut it.
 
Over the following four nights we learn many important lessons that will guide the development of these systems over the coming year. First, the activity and diversity is far above the norm seen in Europe and North America, and in combination with the heat and humidity this pushes the hardware to its limits. Second, powering the units in closed canopy forest, where solar panels are ineffective, will require some innovation of low energy components and low intensity sampling designs.

It’s also clear that moths are not our only visitors. During the nights we saw praying mantis, wasps, beetles and even a Coati (see picture). At the moment our AI algorithms ignore everything that is not a moth but, with all these other creatures making an appearance, we should perhaps be growing our ambitions.
 
The four of us from UKCEH are proud of the AMI-trap, one of the systems being tested. Described as “relentless” by one of our colleagues, it was developed over the course of a year by UKCEH engineers and scientists, building on early designs from Aarhus University. Having tested it underwater and in a 60C oven, perhaps we shouldn’t have been surprised it ran flawlessly in the tropical climate.

Image
A coati, a mammal native to parts of the Americas

A coati visiting the research station on Barro Colorado Island, Panama.

Image
Grasshopper attracted to the lighting

Moths weren't the only insects attracted to the monitoring systems. Other visitors included praying mantis, wasps, beetles and grasshoppers.

The AMI-trap collected tens of thousands of images of moths over the week. An upgraded light, specially designed to emit light most attractive to moths, performed well, and will now become a standard add-on. However, the development never ends and our wish list of new features has got a lot longer while being on the island.
 
After five days, we left feeling excited that the systems performed well and motivated to develop them further in order to realise their potential for tropical research.
 
As we disembark the boat from the island and load our backpacks into the awaiting minibus I overhear a teammate earnestly say, “I think this is the most exciting project I’ve ever been involved in”… I think I agree.  

Dr Tom August

Thanks to the Natural Environment Research Council (NERC) for funding the Easy RIDER project, allowing this work.

Image
Screen showing many different moths and suggested identification

The systems use the latest AI to analyse thousands of images collected each night.

Image
Dr David Roy, UKCEH, working on an AMI-trap

Dr David Roy from UKCEH working on an AMI-trap

Related links