Study of the overwintering ecology of the hazelnut pest, Palomena prasina (Hemiptera: Pentatomidae) in a perspective of Integrated Pest Management

Palomena prasina, the green shield bug (GSB), is widely distributed in the Eurosiberian region. In the Southwest of France it is considered a serious pest of hazelnuts, its feeding punctures lead to blank hazelnuts and kernel necrosis, causing heavy losses in commercial orchards. To date, no Integrated Pest Management strategy is available to control P. prasina.

Control strategies often focus on the pests’ spring-summer ecology, when they are in the field or in the vicinity of crops. However, the abundance of pest populations in crops is also related to their autumn-winter ecology.  The present work focussed on the autumn-winter ecology of P. prasina to identify new opportunities for this pest suppression. We investigate (i) where P. prasina overwinters, (ii) if it aggregates in its overwintering sites, and (iii) if it mates while overwintering. Samples were collected over a two-year period in different ecosystems (forests, hedges, orchards), in human-made structures, and habitats (litter, bushes/trees, dead trees). The reproductive status of GBS individuals was monitored in winter, and in spring when they emerged from overwintering sites.

Our results show that 97% P. prasina adults overwinter  in the leaf litter of orchards and natural ecosystems and that 70% overwinter individually. The abundance of GSB in those sites is negatively correlated with litter temperature and positively correlated with humidity levels. Furthermore, adults only mate after leaving their overwintering site. Finally, there was an important number of overwintering adults hosting endoparasitoids (32%).

The fact that GSB overwinters alone in the leaf litter means controlling its populations by destroying the overwintering sites is not a solution. All the same, our results do open new perspectives for the control of P. prasina. First, the emergence traps, in particular the cone traps, proved efficient for collecting emerging adults and could be used for monitoring. Moreover, our observations point out the existence of long-range mating signals that could be exploited for trapping. Last but not least, the important number of overwintering parasitised adults is a promising biocontrol avenue.

Methods
Material and methods
1. Characterisation of overwintering sites

A field study was conducted in a zone with a radius of 20 km around the city of Cancon, France (44°32'09.7"N, 0°37'30.1"E, 160 m a.s.l.) from December to February (10 weeks), both in 2021 and 2022.

In 2021, three different types of ecosystems were studied: hazelnut orchards, linear hedges and woods. The hazelnut orchards were over 20 years old, at least 1 ha in size, and most of them were planted with the Pauetet cultivar, which is sensitive to P. prasina. They were all commercial orchards with no cultural practices or pesticide spraying during the study period (the last management practices before that period correspond to pesticide spraying in July and harvest in August-September). The linear hedges were lines of trees and bushes that separated two fields, and were more than 116 m in length. The woods were at least 0.68 ha in size. The hedges and the woods were mainly composed of local deciduous trees, different species of oak, chestnut, hornbeam, checker tree, and some shrubs including dogwood and hawthorn; evergreen bushes included wild blackberry, butcher’s broom, and wild madder ( Bonneaud, 2014; Badeau et al., 2017). In 2022, apple orchards were added to the above-mentioned ecosystems. The apple orchards were at least 3.36 ha in size. All the orchards were managed conventionally and most were surrounded by crops, mainly hazelnut, plum, and apple orchards. Ten sites representing each ecosystem, located at least 8 km apart, were selected for the study. Each week, one site of each ecosystem type was prospected (i.e. each site was prospected only once per year).

Three main types of habitat were sampled in each ecosystem: leaf litter, the bark of trees and the foliage of evergreen bushes. Five samples were collected at 10-m intervals along a 50 m transect in the hedges and at the edges of orchards and woods starting at 0 m and continuing to 40 m. A further 5 samples were collected at the centre of the orchards and woods, at a distance of between 10 and 50 m from the edge.

A total of 600 leaf litter samples were collected, of 1 m2 each (250 and 350 samples were collected in 2021 and 2022 respectively). For each sample, four abiotic parameters were recorded at the surface of the litter using a TR-74Ui Series data logger (TandD corporation, Japan): temperature, relative humidity, illuminance and UV. The depth of the leaf litter was also measured. After the measurements were complete, the leaf litter covering each 1 m2 area was collected separately, placed in a black plastic bag and transported to the laboratory. The contents of each bag were placed under a photo-eclector trap (Soil photo-eclector Ø1m2, ecoTech Umwelt-Meßsysteme GmbH, Germany) for 24 h at 25 °C under constant light (Philips Master TDL 18W/840 Cool white), after which the presence of insects was checked in each trap. At the end of the 24-h period, the photo-eclector traps were removed, and leaf litter was carefully inspected for any remaining insects.

We then defined an area with a radius of 2 metres around each leaf litter sampling area. In this area, the bark of one dead or standing tree was inspected visually for 2 minutes, and one evergreen bush was prospected for 10 seconds using a beating tray.

Finally, during the same monitoring period, a survey was conducted of potential overwintering sites in human-made structures (buildings and houses) based on a call for witnesses broadcast by radio, on Internet, and using flyers. Witnesses were asked to take photos of stink bugs present in their houses and buildings and to send them to the laboratory by email together with information on the general geographical location, plus the exact place and date of collection. Thirty-seven volunteers in the Cancon region responded to the call.

All the individuals belonging to the Pentatomoidea collected in the above-mentioned samples were identified to species level.

The abundance of P. prasina in the 600 leaf litter samples was analysed. We fitted a Generalised Linear Mixed Model (GLMM) with a Poisson error distribution and the following fixed effects: type of ecosystem (categorial factors: hedge, woods, apple and hazelnut orchards), zone (categorial factors: interface and centre; interface corresponds to samples collected in the hedges and edges of woods and orchards), abiotic parameters of the leaf litter (continuous factors: litter depth, temperature, humidity, light and UV), study year (categorial factors: 2021 and 2022). To consider the spatial autocorrelation, we added the sampling site as a random effect.

The goodness of fit was evaluated through the randomised quantile residuals calculated using the DHARMa package (version 0.4.6; (Hartig, 2022). The GLMM was performed with the HLfit function of the spaMM package (version 3.13.0; Rousset and Ferdy, 2014) using R software version: 4.1.2.; (Development Core Team, 2021). P-values were calculated using the type II log likelihood ratio test (hereafter L.R.). We started with a model including all the double interactions between the zone and the other seven independent variables and removed them when they were not significant. This first model is called “Full model”.

The same statistical procedure was applied to two partial models focused on P. prasina abundance in the 250 samples collected in the centre of woods and orchards (called the “Centre model”) and in the 350 samples collected at the edge of woods and orchards and in the hedges (called the “Interface model”). The effect “zone” was removed from each of those GLMM and the remaining fixed and random effects were tested. Using the two final partial models and the "predict" function in R, we calculated the predicted number of P. prasina individuals. We computed predictions and their 95% confidence intervals according to each type of ecosystem and for the year in which the largest number of individuals was recorded, except the apple orchard ecosystem which was only sampled in 2022. Furthermore, the quantitative predictive variables with no significant effect were set to their average value and the random effect was not included in the calculation.

Next, a qualitative variable with two categories corresponding to high and low values of the parameter was created for the quantitative abiotic parameters which showed a significant effect in the partial models. The threshold for defining these categories was determined by selecting a value that minimised the P-value of the effect associated with this new qualitative variable in the partial model. In simple terms, this threshold represents a value above which the number of P. prasina individuals in our leaf litter samples changed significantly.

2. Reproductive status

The reproductive status of P. prasina was assessed for the overwintering adults collected in the leaf litter, as described in the previous section, plus adults emerging from overwintering sites at the end of the hibernation period, captured as described below. The study was performed in 2021-2022.

From the end of February to the end of April 2022, adults of P. prasina newly emerged from overwintering sites were collected in two ecosystems: a  2.3 ha hazelnut orchard surrounded by lakes and other hazelnut orchards, located near the village of Moulinet (44°31'23.9"N, 0°36'00.5"E, 117 m a.s.l.), and a 2.25 ha wood surrounded by buildings, crop fields, and hazelnut orchards, located close to the city of Cancon (44°32’34.6”N, 0°35’34.2”E, 97 m a.s.l.). As no information was available concerning the most efficient traps to catch P. prasina emerging from overwintering sites, we used two different kinds of traps. The first was a homemade cone trap (Raney, 1969) adapted from a pop-up mosquito transparent net tent measuring 1.8 m x 2 m x 1.5 m (LxWxH, ®GLKEBY). The adaptation consisted of cutting off the floor of each tent and placing a plastic cone collector, with a1.5 cm Ø entrance at the top of the tent. The tent was kept in place by a 1.50 m wooden pole inserted into the ground at the centre of the tent. The second trap was a homemade trap based on the Circle trap design (Mulder, Reid, Stafne & Grantham, 2012), modified and adapted for P. prasina by enlarging the opening at the top of the trap top from Ø 0.80 cm to 1.50 cm, and by reducing the width of the trap from 81 cm to 66 cm so it fitted the circumference of the tree trunks better.

Ten cone traps and 20 Circle traps were installed at each site. The cone traps were spaced about 20 m apart and the Circle traps were fixed to trees growing about 10 m apart. The presence of P. prasina in the traps was checked three times every day: in the morning (at 9 am, at noon, and at 5 pm), to avoid leaving emerging individuals together which could bias the evaluation of mating status (see below).

We checked the sexual maturity of both male and female P. prasina collected weekly in the overwintering sites between December and February, in 2020-21 and 2021-22, and in the cone and Circle traps between the end of February and the end of April 2022. All parasitised individuals were excluded from the analyses, as they could interfere with reproductive status (De Salles, 1992). First, we observed the colour of the individuals, which is considered as a proxy for reproductive status in the close species Nezara viridula (L.) and Plautia stali Scott (Musolin & Numata, 2003; Kotaki & Yagi, 1989): russet or reddish brown during the winter reproductive diapause and green during the period of active reproduction. The insects were then killed by freezing at -20 °C for 2 min, dissected and the sexual organs extracted in distilled water under a stereo microscope (Nikon, SMZ1270). Ovarian development was assessed using a scale starting at 0 (previtellogenic females) and going up to 4 (post-reproductive females) (Kiritani, 1963; Nielsen et al., 2017). The mated status of females was evaluated by the presence of spermatozoids in the spermatheca (Golec and Hu, 2015; Hamidi et al., 2021) observed under the microscope. In males, sexual maturity was assessed through the presence of living (i.e. moving) spermatozoids in the testes.