Can developments in sensor technology improve our understanding of honey bee foraging behaviour and contribute to the development of new models for almond pollination economics.

The amount of bee foraging activity during almond bloom is of considerable interest to growers. Currently, bee foraging activity is inferred using Bee Flight Hours, which are calculated purely based on meteorological data. A bee flight hour is defined as one hour where: there is no rain; the temperature is above 55℉ (12℃); and wind speeds are less than 15 mph (24 km/h). This provides a basic estimate of the amount of good bee foraging weather throughout the bloom period. The hypothesis is that more bee flight hours means more opportunities for pollination, making flight hours a useful predictor of nut set, and ultimately yield. While this concept seems logical, in reality, the situation is much more complicated. There are various in-colony and out-colony factors that have an impact on foraging activity levels[1] [2] [3]. In-colony factors include the colony size and demography, queen and brood status. Out-of-colony factors include weather conditions and forage availability. The dynamics of forage, weather conditions and the changes in colony foraging behaviour are the key determinants of the number of visits received by blossoms.

However breakthrough developments in sensor technology mean it is now possible to directly monitor bee foraging traffic. A unique optoelectronic bee counter sensor that is being developed  and tested through the iPollinate project provides accurate real-time counts of the number of bees leaving and returning to hive. As will as providing a much deeper understanding of honey bee foraging behaviour, it will also contribute to the development of more accurate models and metrics for assessing foraging performance and pollination efficacy. With this aim in mind, the iPollinate project deployed bee counters on 16 hives used to pollinate almonds at two different orchards during the 2021 pollination season. In addition to the bee counter sensor, the hives were also fitted with an In Hive Sensor (Colony acoustics, brood temperature and humidity), a hive scale (nectar flow), and a weather station (local microclimate). This article presents some examples of the insights generated in the trial, focussing on the impact of weather conditions and forage availability (stage of bloom) on bee behaviour, and pollination progress.

Daily foraging pattern and real flight hours

Figure 1 below shows the foraging rate (bees per minute leaving the hive) from one of the colonies throughout a day during the full bloom period in late February 2021.

The precision with which the instrument measures bee flight uncovers a number of patterns of daily bee flight behavior: the exact time of the start and end of foraging, (and hence the real flight hours), and the total number of foraging trips made by the colony during the day.  Foraging activity started at 9.00am, and increased rapidly to reach a peak of  150 bees per minute leaving the hive by 11.00. The foraging rate drops off rapidly at 17:00, and the workforce is back in the hive by 17:45.  Therefore the actual ‘flying time’ on this day was 8 hours 45 Minutes. What is also interesting is that the bees were still foraging quite late in the afternoon. However, perhaps even more valuable is that the total number of foraging trips made by this colony during that day was 50,529.

Figure 2 compares the actual flying hours as measured by bee counters for all the hives throughout the study period with the Bee Flight Hours calculation based on ambient weather conditions for the orchard locations.

This shows that the Bee Flight Hour calculation underestimated the actual flying time of the colonies at both orchards. Over the time period of the study this totalled 38 hours of flying time. Bees will sometimes fly in weather conditions outside the thresholds used for the Bee Flight Hours calculation, particularly when high-value pollen is available. This is consistent with previous studies, which showed foraging activity commencing on some mornings when temperatures were only 48- 50F [4].

Another reason for the difference is that the microclimate in an orchard can vary from the prevailing weather conditions of the wider area.  Even the positioning of a hive in the shade or full sun, or on a NW slope instead of a SE slope can impact colony activity. Bees continuously monitor their external environment to make decisions about whether to embark on foraging journeys. The reaction time between changes in microclimate and changes in colony activity is less than 1 minute. This is exemplified in Figure 3, which shows flying activity of a hive on 10 March, when there was cold, wet weather.  This shows how the bees dodged the rain showers throughout the day; indeed we can see that bees started to return to the hive before it started raining.

Not all bee flight hours are the same

While Bee Flight Hours is a useful measure of the available foraging time, we know that  actual foraging rates vary significantly over time in response to a range of intra- and extra-colony stimuli. Therefore an hour defined as Bee Flight Hour can differ drastically in the number of actual foraging flights, and thus visits to flowers, for any given period.

Figure 4 compares actual bee flying time and average daily bee trips per hive from full bloom to the end of pollination. This shows a steady decline in foraging trips over the period, from 35,000-40,000 a day in late February to less than 20,000 in the second week of March. The other noticeable feature of this graph is the drop in activity and flight hours on 8-10 March. Bee flight hours return to 8 hours/day on 11 March, but foraging trips are only half what they were in late February when flying time was also 8 hours/day.  Adding the orchard weather conditions and stage of bloom provides the vital context that explains these patterns of foraging behaviour (Figure 5).

The graph clearly shows that the overall decline in foraging trips is the colony response to the reduction in forage availability as the bloom progresses.  Against this overall trend, we can see the impact of cool, wet weather on 8th – 10th March, which significantly reduces foraging activity. The number of foraging trips made in 1 hour ranges from 4,900 on 28 February to 1,700 on 10 March.

These data illustrate the dynamic foraging activity of a single hive over time in response to changes in local environmental conditions. However it is also useful to compare data from hives at two separate orchards on the same day, to illustrate the impact of variations in local conditions.  Figure 6  compares bee foraging trips (from hives of equal strength) and Bee Flight Hours (calculated using ambient weather model) at the 2 study orchards on 27th February. This shows that the hives at Orchard 1 completed a total of 194,523 bee trips compared to  132,826 trips at Orchard 2, a 30% lower activity level. The Bee Flight Hours calculation for both orchards was 8 hours.

So why the big difference? The prevailing weather conditions for both orchards indicated no rainfall at either location, daytime temperatures well above minimal flying temperature, and average wind speeds below the threshold of 15mph. However our weather station at Orchard 2 recorded wind gusts of 15-25mph, enough to reduce foraging activity. At Orchard 1 the peak wind speed recorded at the hives was much lower, only 8.3 mph (Figure 7). The Actual Bee Flight Hours calculated from the bee counter traffic was 8:38 hours in Orchard 1 and 7 hours in Orchard 2.

The real time direct measurement of colony foraging activity using a bee counter will generate an incredibly rich source of data that could provide the basis for more accurate metrics and models of almond pollination.

New metrics for foraging activity and pollination input

Perhaps the most useful measures of foraging activity at the hive level are Foraging Rate and Foraging Force.  The Foraging Rate of a colony can be calculated by combining the number of bee trips (exits from the hive) with the number of flight hours.  This could be represented as Bees per Minute (BPM) or Bees Per Hour (BPH) leaving the hive. The latter is analogous to Miles per Hour (MPH) as the speed or rate of travel. Figure 8 shows average Bees Per Hour (BPH) for 4 hives in our experiment on 25th February, along with the average BPH for that pallet of hives.

The Foraging Force of a hive can be defined as the total number of foraging trips delivered by that hive to the crop. This could be calculated as a daily Bee Trips per Hive, the cumulative bee trips delivered during the bloom, and ultimately the Total Bee Trips for the whole pollination season. Figure 9 provides an illustration of the Foraging Force for one of the hives over the study period.  Further research will be conducted to assess the Foraging Rate and Foraging Force of hives of different strengths (bee frames) and under different environmental conditions.

Furthermore, the total number of bee trips of a group of hives can be apportioned over an orchard (usually 2 hives/acre in almonds) to provide a figure for the number of  Bee Trips per Acre. Dividing this figure by the number of trees per acre provides an estimate for the number of Bee Visits per Tree. Figure 10 shows average Bee Trips per Acre and Bee Visits per Tree for one day during full bloom at Orchard 1 in our study.   Figure 11 illustrates cumulative Bee Visits per Tree over the study period.

This exploratory trial has demonstrated how the application of advanced sensor technology and data analytics can not only provide deeper understanding of bee foraging behaviour in almonds, but also enable the development of new metrics and models for measuring pollination inputs and efficacy.

However further research is required to generate larger data sets and undertake more in depth analysis for robust model development. To this end iPollinate is carrying out further crop trials on larger numbers of hives in 2022 and 2023. In essence, the bees themselves are the ‘sensors’ of the environment, and harnessing the wisdom of the hive can help to improve our understanding and management of one of the most important processes for food production.


[1] Abou-Shaara, H.F; Veterinarni Medicina, 59, 2014 (1): 1–10; The foraging behaviour of honey bees, Apis mellifera: a review
[2]  Klein et al Honey bees increase their foraging performance and frequency of pollen trips through experience Scientific Reports volume 9, article 6778 (2019)
[3] Tenczar, P., et al Automated monitoring reveals extreme interindividual variation and plasticity in honeybee foraging activity levels. Anim. Behav. 95, 41–48
[4] Sandra Evans; Pollination Insights 10.12.20; Every Bee Counts.