Migratory flight behaviour of the pollen beetle Meligethes aeneus †

Abstract BACKGROUND The field ecology of the pollen beetle Meligethes aeneus and its damaging effects on oilseed rape crops are well understood. However, the flight behaviour of M. aeneus, in particular the drivers for migratory movements across the landscape, is not well studied. We combined three established methodologies – suction traps, vertical‐looking radar and high‐altitude aerial netting – to demonstrate that M. aeneus flies at a range of altitudes at different points during its active season. RESULTS By linking evidence of high‐altitude mass migration with immigration of pollen beetles into oilseed rape fields, we were able to ‘ground‐truth’ the results to characterise the seasonal movements of this pest across the landscape. CONCLUSION We demonstrate that this novel combination of methodologies can advance our understanding of the population movements of pollen beetles and could provide an opportunity to develop predictive models to estimate the severity and timing of pest outbreaks. © 2017 The Authors. Pest Management Science published by John Wiley & Sons Ltd on behalf of Society of Chemical Industry.


INTRODUCTION
The pollen beetle Meligethes aeneus (Fabricius) (synonym Brassicogethes aeneus Fabricius) (Coleoptera: Nitidulidae) is a small, black, univoltine beetle with a dependence on brassicaceous plant species for oviposition and larval development. 1 -3 The introduction of large areas of oilseed rape (Brassica napus L.) (Brassicaceae) into the European agricultural landscape in the 1970s was rapidly exploited by M. aeneus, and consequential changes in its reproductive success and intraspecific competitive ability 4 have resulted in its present-day status as an important economic pest. 5,6 The migration behaviour of M. aeneus is not well studied to date; however, the life cycle and many population parameters are well known. 5,7 M. aeneus beetles are not ground active, so it is assumed that they rely on flight for all dispersal movements across the landscape when in search of host plants for feeding, mating and oviposition. Temperature is a limiting factor on insect flight activity, and recent laboratory studies have shown that their propensity to fly follows a sigmoid temperature-response curve between 6 and 23 ∘ C. 8 Field studies have recorded solitary flights starting from 10.2 ∘ C 9 and gregarious flights 10,11 from 12.3 ∘ C upwards. 12,13 The main migratory movements of M. aeneus across the landscape occur at recognised stages within their life cycle: 1,5,7 1. On emergence from overwintering sites, adult beetles fly to flowering wild plants to feed on pollen. 2. Adult beetles then migrate to oilseed rape crops to feed and reproduce; firstly to winter-sown crops in early spring (a high intensity of flights at this time was recorded by Sedivy and Vasak 14 ), and then to spring-sown crops in late spring. 3. New-generation adults emerge mid-summer and fly from winter oilseed rape crops that have finished flowering to locate new host plants for feeding; these may include spring brassica crops, but once flowering of these also subsides, the beetles move to forage on pollen from wild plants. 4. As pollen availability and day temperatures fall, the beetles move to overwintering sites.
The rise in temperature in early spring and other meteorological factors 8,15 trigger migration flights from overwintering sites and also regulate the development of the oilseed rape crop. As a consequence, the timing of the main migration into winter oilseed rape fields usually coincides with the damage-susceptible early bud stage. Climate-change-related variations in air temperature and precipitation are likely to affect crop invasion by pollen beetles, and multimodel predictions by Junk et al. 16 suggest that the onset of crop invasion will be 10 days and 23 days earlier in the near future (2021-2050) and in the far future (2069-2098) respectively. However, the implications for the status of M. aeneus as a pest are unclear, as the effect of climate change on the overwintering survival and behaviour of the insect are as yet unknown.
The migratory movements of individuals described above have mostly been inferred from counts of adults on plants throughout the year at different sites, and not from direct recording or tracking of individual flights. Owing to their small size, it has proved difficult to track individuals; however, some studies have been able to estimate the distances flown. Individuals have been recorded travelling 13.5 km using radioactive tracers, 17 but it is likely that they can travel considerably further.
In addition to the longer migratory flights, M. aeneus also perform shorter-distance, lower-altitude flights to locate their host plants. The 'flight boundary layer' was described by Taylor 18 as a hypothetical layer of air near the ground within which insects are able to control their movements relative to the ground because their flight speed exceeds wind speed. Taimr et al. 17 showed that M. aeneus beetles were able to locate and travel to fields of oilseed rape regardless of wind direction, indicating that they used self-powered, directed flights within their flight boundary layer, and were recorded at a distance of 300 m from the release point within 2 hours of release. Further evidence from detailed field experiments has confirmed that adult M. aeneus use upwind anemotaxis to locate oilseed rape plants. 19 -21 To date, little research has focused on the flight behaviour of M. aeneus above the flight boundary layer to uncover population-scale patterns of longer-distance, higher-altitude dispersal movements. Therefore, this study investigates the potential for combining data records of pollen beetles from three established recording techniques -suction traps, vertical-looking radar (VLR) and high-altitude aerial netting -and combines these data with phenological monitoring of beetles in oilseed rape crops to gain an understanding of their migration movements.
Suction traps collect flying insects by constantly sampling from the air. 18,22 The Rothamsted Insect Survey operate suction traps that sample at 12.2 and 1.5 m heights, providing a constant, easily comparable measure of insect density at the two heights. High-altitude insect flights have been studied using scanning radars since 1968. 23 However, the recent development of automated, vertical-pointing systems has enabled radar to be used for routine, long-term monitoring of aerial migration. 24,25 Vertical-looking radar (VLR) records information about overflying insects that is related to their speed, direction of movement, orientation, size and shape. Sampling of the aerial fauna at similar heights is required to calibrate the radar records for individual species, and this has been achieved using a balloon-supported net, which samples the aerial fauna between 180 and 200 m above ground level. 26,27 Applying this combination of methodologies to advance the understanding of the migratory ecology of M. aeneus is novel, and for the first time we aim to 'ground-truth' these approaches by linking evidence of high-altitude mass migration with immigration of pollen beetles to fields containing oilseed rape. This study was conducted over two years to determine 'proof of concept' of this approach for developing an understanding of population dynamics at the landscape scale and the potential for advancing forecasting approaches.

EXPERIMENTAL METHODS
The aerial netting data were collected in July of 1999 and 2000. All other data (suction traps, VLR and counts on crop plants in the field) were collected from the start of March to the end of August in both 2001 and 2002 and were summed into weekly totals for comparative analyses.

Suction traps
The aerial density of M. aeneus was measured daily at two heights -12.2 and 1.5 m -using the Rothamsted Insect Survey suction trap 28,29 at Rothamsted (Harpenden, UK). Daily counts of pollen beetles were converted to aerial density in 10 3 m 3 air, assuming a 12 h (daytime) period of flight activity, and summed for weekly values.

Vertical-looking radar (VLR)
The VLR was operated continuously and was also sited at Rothamsted Research; the equipment and operating procedures are detailed by Chapman et al. 25,30 Briefly, the VLR emits a narrow, vertical beam, and overflying insects modulate the radar signal in a way that is related to their speed and direction of movement, their orientation, size and shape. It is capable of detecting individual insects flying within 15 height bands between 150 and 1200 m above the radar, depending on their weight. As M. aeneus beetles only weigh between 1 and 2 mg, the radar can only detect this size of insect at the lowest sampling band of 150-195 m above ground level. 25 Any signal captured within that range was recorded for a 5 min period every 15 min, 24 h a day. The aerial density of overflying insects was found by calculating the volume of air sampled by the VLR for every target, and densities were expressed in terms of the mean number of insects per 10 7 m 3 as calculated for each 5 min period. The VLR is unable to identify the species of insect detected using this method alone, so the continual recordings are restricted to specific times during the day and to specific insect weights, to reduce the 'noise' from other flying insects. Therefore, the radar data were filtered to daytime records (07:00-19:00 h GMT) of targets between 1 and 2 mg over the sampling period for both years. The aerial densities were then summed to provide the daily density of correct-sized targets per 10 7 m 3 and then corrected to density in 10 3 m 3 air to be comparable with the suction trap density data. The aerial densities will be overestimates of true pollen beetle numbers owing to the fact that the VLR will also record other species weighing 1-2 mg, but the data are indicative of periods when lots of pollen beetles are flying.

Collection of migrating pollen beetles using aerial netting
Clarification of the insect species flying at approximately 200 m above the ground was achieved by aerial sampling using a net suspended from a tethered, helium-filled blimp. 26

Meteorological data
Meteorological data were collected from a meteorological station sited within 20 m of the Rothamsted suction traps. The data were collected every 15 min, 24 h a day, and were combined into daily or weekly means. The variables used were: mean temperature ( ∘ C), mean solar radiation (W m −2 ), mean wind speed (m s −1 ), total rainfall (mm) and mean relative humidity (%).

Statistical analyses 2.7.1 Identification of meteorological factors influencing flight at different altitudes
Daily meteorological variables and beetle densities at the three altitudes were compared by calculating Spearman's rank correlation coefficients. The significance levels for rejecting the null hypothesis (of no association between the two variables) based on Student's t-approximation were calculated (see supporting information 1), and the false discovery rate 32,33 for these comparisons was also calculated to quantify the expected proportion of type I errors.

2.7.2
Correlation between the phenology of the oilseed rape crop and field counts of M. aeneus The mean number of pollen beetles per plant and the proportion of the crop in flower (GS 60-65) were calculated weekly for all the fields in both years and compared for both winter and spring oilseed rape by calculating Spearman's rank correlation coefficients. The significance of the correlation was determined using a t-test.

Correlation of population flight patterns with immigration and emigration from field crops
Weekly data were divided into periods of immigration and emigration. Immigration was defined as those weeks leading up to and including the maximum number of beetles per plant. The mean field count of M. aeneus and the proportion of plants in flower (GS 60-65) for each week during immigration were then compared with the previous week's insect densities at 1.5, 12 and 150-195 m, and correlation matrices were produced using Spearman's rank correlation coefficients. Emigration was defined as the period following the maximum number of beetles per plant in the field, and the data for these weeks were compared with the same week's insect densities, again using Spearman's rank correlation coefficients. These comparisons were chosen to model the movement of insects between flight and the crop.

Diurnal activity
The total catch of M. aeneus from the 12.2 m suction traps is shown in four discrete time periods (Table 1), demonstrating that flight in this species occurs during the daytime.  (Fig. 1a) and 2002 (Fig. 1b). Around 3 times more beetles were found in 2002 than in 2001, although both years show similar patterns phenologically. The highest density of beetles occured at 1.5 m, whereas the 12 m and 150-195 m densities were approximately one order of magnitude smaller. Their use of flight at different altitudes varied through the season. Early in the season (March-early April; weeks 1 to 6 in Fig. 1) there was a predominance of flight at 12 m, followed by a period (May to June; weeks 9 to 15 in Fig. 1) where flights occurred at all altitudes.

Altitudinal flight profile of M. aeneus
wileyonlinelibrary.com/journal/ps The highest density of beetles flying at any height occurred in July (weeks 18 to 21 in Fig. 1), with by far the most at the lowest altitude, 1.5 m. In August (weeks 22 to 25 in Fig. 1) there was a shift towards high-altitude flight (at 150-195 m) and the lower-altitude flights tailed off rapidly.

Identification of meteorological factors influencing flight at different altitudes
Insect density at all altitudes was positively correlated with temperature and radiation while negatively correlated with rainfall and wind speed (which was also found by Skellern et al. 21 ) (see supporting information 1).  (Fig. 2). Spring oilseed rape also showed the same trend, but there were low numbers of replicates; in 2001 the correlation was not statistically significant (r s = 0.533, df = 8, P = 0.115), and in 2002 there were too few replicates to conduct the Spearman's rank correlation test.

Correlation of population flight patterns with immigration and emigration from crops
Correlation matrices were formed for the period of immigration (S2) and emigration (S3) into and out of winter and spring oilseed rape fields. By focusing on the weeks around the initial colonisation of winter and spring crops in each year (Fig. 1), it is clear that daily patterns of flight are different for the two colonisation periods. Winter oilseed rape colonisation is characterised by low insect wileyonlinelibrary.com/journal/ps Emigration was less well correlated with insect aerial density. As the number of beetles in winter oilseed rape crops declines (seen by the negative correlations with week), there is a correlation with insect density at 150-195 m (r s = 0.475, df = 38, P = 0.002), but not with densities at any of the other altitudes, while emigration from spring oilseed rape is strongly correlated with insect density at 1.5 m (r s = 0.741, df = 9, P = 0.009).

DISCUSSION AND CONCLUSIONS
This study has characterised, for the first time, the altitudinal flight profile of M. aeneus throughout their active season. It has provided evidence that this species uses high altitude flights (up to ca 200 m) at specific points during the year, as well as showing multiple periods of low-altitude flights. It has also been confirmed that there is a strong bias for flight during the day, which could be due in part to their reliance on visual cues during flight 34 and/or the need to reach the minimum temperature threshold required for flight. 8 It may also be influenced by the fact that they use plant volatiles for host location and plants release more of these volatiles during the day. These results concur with the findings of Lewis and Taylor, 35 who found that the peak flight time in this species was 12.44 GMT.
Emergence of overwintered adults (March-early April) is followed by a peak in higher-altitude flights (at 12 m), which may suggest that medium-range dispersal movements are occurring from the overwintering site to food sources. Mating occurs on brassicaceous plants from mid-May onwards, 36 so dispersal from overwintering sites also ensures mate location and increases population heterogeneity. 37 The emergence seems to be spread over several weeks that are characterised by warmer weather and low rainfall. These variables have previously been linked with pollen beetle immigration into oilseed crops. 15,16,21 However, we found that after emergence these meteorological factors have less impact on the timings of flights throughout the rest of the season. Flight in M. aeneus was most common at low altitude (1.5 m suction traps), indicating high levels of population redistribution at the local scale. Such dispersal enables the population to relocate across the landscape to find their ephemeral, but highly concentrated host plants. 37 During the winter oilseed rape crop colonisation period, there is still a considerable level of flight activity, which could be due to continuing emergence from overwintering and/or population redistribution to localised areas of resource availability. Following this period, there is an interesting lull in flight activity that coincides with a drop in the number of beetles in the crops. This occurs in the time between winter oilseed rape flowering and spring oilseed rape flowering, during which time there is no detectable movement of beetles to other food sources. This might indicate that a large proportion of the overwintered adults die at the end of the winter oilseed rape flowering and it is mainly the new-generation adults that colonise the spring crops. However, it is thought that the new generation does not reproduce in their first year, 36 yet the spring oilseed rape crops can sustain high levels of oviposition damage; 38 therefore, some reproductive females must move from the winter to the spring crops.
Immigration of M. aeneus into spring oilseed rape is very rapid, and the population reaches its peak in mid-July, with the arrival of new-generation adults. The use of low-altitude flights at this stage is likely to be the most efficient way of quickly locating the nearest resource that is still in abundance at this time. Spring oilseed rape emigration at the end of the active season may be followed by migratory flights at high altitude (150-195 m), as there was evidence of this from the 2001 data (but not 2002, resulting in a non-significant correlation overall). Utilising the strategy of wide-range migration prior to overwintering would increase genetic variation in the population in the following year and may increase the geographical range over which the population will emerge. Further investigation into emigration patterns over several years would provide better understanding of this part of the life cycle.
In both years the highest number of beetles making flights at low altitude and subsequently colonising the nearby crops was in early July owing to the population comprising both oldand new-generation beetles. Therefore, timing the sowing of the spring crop to ensure that the vulnerable bud stage does not coincide with this could provide an efficient means to reduce pest damage at this crucial point in crop development.
The combination of sampling techniques used in this study presents a novel approach for understanding population movements of M. aeneus at a landscape scale and provides the opportunity to develop predictive models to estimate the size of the following year's population (by comparing the size of the new generation with the overwintered generation the following year). Such ecological understanding would enhance decision support systems such as proPlant expert (http://www.proplantexpert.com) (in the United Kingdom http://www.bayercropscience.co.uk/ wileyonlinelibrary.com/journal/ps pollenbeetlepredictor/) that help growers predict the risk period for crop immigration by pollen beetles. proPlant is a web-based decision support system that has a phenological model of population dynamics of M. aeneus that is related to local weather forecasts to predict the start and end of crop immigration as well as major risks of migration events in-between, 15,39 enabling more focused monitoring. 40 The work presented here demonstrates that there is also potential to refine such decision support systems by integrating crop phenology and pest population dynamics. These refined models could provide estimates of pest abundance in addition to forecasts of risk of immigration. With just two years' worth of data, this study highlights the potential of these methodologies for collating a dataset that could be used to refine such models and the value in investing future research effort in this area.
The description of the flight patterns of M. aeneus will also be of great importance in implementing integrated pest management control methods such as the push-pull strategy most effectively. 41,42 The push-pull strategy is likely to be most effective at controlling pests during immigration to oilseed rape crops. 38,43 -45 Important considerations established from this study include the fact that the immigration pattern is generally linked to the crop phenology, and the arrivals have not always moved from nearby flowering plants, but have potentially flown from long distances at a range of altitudes to reach the crop.
The synergistic output of the novel combination of techniques employed here provides a rationale for long-term monitoring of the population movements of this pest, and puts value on the modelling of several years' worth of meteorological and biological data potentially to yield specific predictors of immigration to oilseed rape crops.