Next time you eat chocolate, munch on a fresh apple or tuck into a plate of beans, remember that the production of these foods have relied on pollinators. While large proportions of the human population are under lockdown, these essential workers of the agricultural sector continue to make sure that a multitude of crops are still being pollinated. It is well established that ~76% of the crops that we consume globally depend to some extent on animal pollination, whether that be by bees, hoverflies, wasps or even bats. The incredible work of pollinators is not only limited to agriculture; the successful reproduction of wild plant species can also depend upon plant-pollinator interactions too.
Unfortunately, the pressures upon the pollinator community are no secret. In the UK, it has been shown that there are pollinator food shortages at certain times of year (such as early spring and autumn). These are also key times of year for insects, as it is critical to build up energy resources before and after hibernation. Pollinator food shortages are particularly a problem in intensively farmed agricultural landscapes, where changing land-use has resulted in a loss of nectar and pollen-rich flower supplies. Spatial distribution of food is also a key consideration. Some small-bodied pollinators only travel short distances when foraging and for these species, navigating a farmed landscape of many large mono-cropped fields with few flowering resources could be likened to crossing a desert.
Taking into account the spatial and temporal distribution of food (and nesting resources) can therefore be useful when farmers decide upon management actions for enhancing or creating on-farm pollinator habitat, and this is where my PhD research comes in.
Broadly speaking, I am looking at how we can better manage farmed landscapes for pollinators. More specifically, I am exploring whether remotely-sensed aerial imagery can be used to map nectar-rich flowers in both wildflower margins and hedgerows. A classification system on GIS can be used to assign the images to a plant species, which is what can be seen in the images below. The imagery used here is multispectral, fine resolution (3 and 7cm pixel sizes), with data gathered across the visible and near-infrared wavelength ranges.
Demonstration of a classification trained for hawthorn flower recognition. Left: Multispectral 3cm resolution image, with sections of hawthorn and cow parsley outlined. Right: Same image classified with categories defined as ‘hawthorn’ and ‘other’.
So far, I have been focusing upon a few key nectar-rich species such as common knapweed and hawthorn. We also need to investigate whether the approach can be extended to other species, particularly those of similar colours in the visible wavelength range. For example, will an oxeye daisy be separable within imagery from yarrow? We also need to establish a way of translating the number of pixels that have been classified as a particular species, into an estimate of the number of flowers that this represents on the ground.
Common carder bumblebee (Bombus pascuorum), Large patch of yarrow (Achillea millefolium) , Oxeye daisies (Leucanthemum vulgare)
These are all questions that I looked to address in the 2020 field season. After an initial panic as to whether fieldwork would go ahead due to the pandemic, I was pleased to have made it out to the field in July.
For the most part, social distancing wasn’t much of an issue as I was predominantly working alone. Pandemic-proof logistics were mostly associated with changing accommodation and collecting the drone imagery. I had already mapped out sections of flower species of interest using markers that would be visible from the sky, and the drone was going to take the images I required for analysis. Helping with this was a novel experience. The farm was located near to an airfield so I was put in charge of listening out on the radio for incoming light aircraft that might be in the path of the drone. Andy Bodycombe from HexCam had the trickier job of navigating it!
The data collected this season is currently being used to classify images into each of the flower categories of interest before calculating the accuracy of each classification. Eventually, the mapped flower resource will be translated into an estimate of the nectar-provision across the study site.
Andy from HexCam flying the drone
Setting out large whiteboards as markers
Whilst the last year has presented challenges, the pandemic has provided us with a golden opportunity to slow down. If you have a garden, front porch or window box, now is a great time to start growing some nectar-rich flowers species too – it all makes a difference! The key is continuity of resources throughout the whole foraging season so here are some suggestions of nectar-rich species that are great for pollinators throughout the year.
Early spring
Early flowering shrub species such as blackthorn and goat willow
Dandelion
Red campion
Cowslip
Late spring
Hawthorn
Hogweed
Early summer
Knapweed
Oxeye daisies
Bramble
Clovers
Late summer/autumn
Wild carrot
Yarrow
Ivy
Acknowledgements
Thanks go firstly to UKRI NERC and Hutchinsons for funding this PhD and to the EnvEast DTP. Many thanks go to my UEA supervisors Dr Lynn Dicks and Professor Andrew Lovett and my Hutchinsons supervisor, Stuart Hill. Thanks go to Spectrum Aviation and HexCam for acquiring the remotely sensed imagery and to everyone else who has helped me so far.
