The stone pine, Pinus pinea L., a new highly rewarding host for the invasive Leptoglossus occidentalis

The invasive seed bug Leptoglossus occidentalis , a species native to Western North America, is of major concern for the producers of stone pine seeds in the Mediterranean countries. The large size of these edible seeds and their nutritive content may represent a pull factor for the seed bug. Cone and seed traits of three main Mediterranean pine species: P. pinea , P. pinaster , and P. halepensis, were evaluated. Preference trials with cone-bearing branches, individual cones and seeds were conducted to test host preference among the three host species. Considering the kernel size, stone pine seeds provide 4 to 13 times more reward than P. pinaster and P. halepensis seeds, respectively, but also needed a greater effort to be reached as measured by coat thickness. Still, the benefit/cost ratio was higher on P. pinea . Individual seeds and cones of P. pinea were 2 to 3 times more consumed than those of the two other pine species. However, branch preference trials did not reveal any difference in bug visits. Moreover, adults manifested strong group behaviour on branches, frequently dissoci-ating into two persisting groups. The implications of these results for P. pinea producing areas are discussed. .


Introduction
) and between cones of P. pinea of different physiological status (Farinha et al. 2018b). This host selection capability strongly indicates that the bug can probably discriminate between different host conifers.
Evaluating host preference of this seed feeder under natural conditions is not a straightforward task due to the difficulty in detecting the bug and the absence of visible damage on cone surface. The only study on L. occidentalis feeding preference was carried on under laboratory conditions using individual mature seeds, extracted from cones (Lesieur et al. 2014). Having offered such seeds to adults in choice tests, Lesieur et al. (2014) did not find any preference among several European conifers (P. sylvestris, P. nigra, Abies nordmanniana subsp. bormuelleriana, Larix decidua and Picea abies) compared to the Nearctic Douglas-fir, Pseudotsuga menziesii var. glauca. However, although quite extensive this study did not include P. pinea seeds and to our knowledge, no other studies tested bug preferences for this host species.
Seeds of P. pinea are large-sized and highly nutritive which can be a pull factor to a seed-eater like L. occidentalis. On the other hand, larger seeds also mean a thicker seed coat which may represent an obstacle for the piercing mouthparts of the bug. We hypothesized that in preference trials using seeds, the bigger individual seeds of P. pinea might visually lead to a host preference towards a higher reward whereas the harder seed coat may constitute an additional cost. Seed volume is a proxy to the seed reward and thickness may represent a proxy to the effort.
Selective behaviour in the field is known to operate at sequential levels. First, individuals select a tree, then a cone and lastly a seed to feed upon (Blatt 1997). So, we presumed that host preference may differ regarding the plant component tested; either using seeds, cones, branches or trees. Furthermore, it would be relevant to compare results and discuss the pros and cons when using different methodologies.
In this study, our objectives were to evaluate the host preferences of L. occidentalis for branches, cones, and seeds of the three main pine species in the Mediterranean Basin, P. pinaster, P. halepensis and P. pinea. For this purpose, we compared cone volume, seed weight and volume and seed thickness of the three species, and then tested bug preferences in choice tests.

Methods
Three separated choice experiments using different substrates, cone-bearing branches, fresh last year cones and mature seeds, were conducted to evaluate the preference of L. occidentalis adults among Pinus pinea, P. pinaster and P. halepensis. All the experiments were carried out under laboratory conditions. Only adults of L. occidentalis were used since nymphs are apterous and thus not capable of actively choosing the tree or the host where they will feed in natural conditions. All individuals came from a permanent laboratory colony with adults collected in Santarém region, Portugal during the summer of 2015. The colony was supplemented each summer with more adults from the same region to avoid consanguinity thus consisting of individuals with mixed ages.
The colony was reared at Centro de Estudos Florestais, Lisbon, Portugal in a climatic room under the following controlled conditions: 21 °C with 60% RH and 14:10 light/ dark cycle. Branches and cones from P. pinea were used as food source. Trials began by removing experimental adults from the permanent colony and marking them with an individually coloured and numbered label painted in the thorax. All marked adults were put in a cage with cone-bearing branches and seeds of the three hosts during one week. Individuals were then subjected to a 24-hour starvation period after which the trials began. Adults used to replace dead ones were removed from the permanent colony, marked but were immediately placed in the cages or test boxes.
Cone-bearing branches used in trials were collected in different locations for each of the pine species: stone pine branches were collected in Monsanto, Lisboa (38°43.09'N; 09°12.41'W) in a natural pure stand of adult trees; maritime pine branches were collected in Sobreda, Almada (38°38.06'N; 09°12.66'W) in an urban patchy stand; finally, branches of aleppo pine were collected in the university campus, Lisbon (38°72.80'N; 09°12.66'W). Cones / seeds used in preference trials were from branches / cones from the same locations as above with the exception of stone pine seeds which came from a pure, grafted stand in Santarém region, Portugal (39°6.50'N; 08°21.91'W) and maritime pine seeds in the two-choice trail which came from Setúbal region, Portugal (38°34.82'N; 09°11.09'W).

Assessing the differences in size of cones and seeds among the three Mediterranean pine species
The volume of a sample of the cones used in the experiments was measured by displacement of water in a graduated cylinder (n=6 for P. pinea and P. pinaster and n=12 for P. halepensis).
Respecting mature seeds, all that were used in preference trials were weighted at the beginning and at the end of the experiments. At the end of the trials, all seeds from the three pine species were opened, and the volume of the kernels showing no feeding damage were measured by displacement of water in a graduated cylinder with a sensitivity of 0.25 ml. Due to the very small size of P. pinaster and P. halepensis, these seeds were measured in batches of 20 seeds and then the individual volume was extrapolated. The thickness of the seed coat for each host species was measured on the images collected by the Scanning Electron Microscope (SEM) using Image J software. The coats of three seeds per host were photographed in SEM with 20 measures taken in each photo.

Branch preference trials
Choice experiments were conducted in large cages (100×50×40) cm made up of a wooden frame and mesh walls. Preference among the three host species, P. pinea, P. pinaster and P. halepensis was tested in pairs by offering two potted branches of differ-ent plant species per cage to ten adults. All branches used in the trials bear cones at all development stages (1 st and 2 nd year for P. pinaster and P. halepensis and 1 st , 2 nd and 3 rd year for P. pinea). The number of last-year cones (2 nd year for P. pinaster and P. halepensis and 3 rd year for P. pinea) in the tested branches was the same for the pair P. pinaster -P. pinea (ranging from 1 to 2 cones each), but not for the pairs P. pinaster -P. halepensis and P. pinea -P. halepensis in which the number of P. halepensis cones varied between 2 and 5 due to their smaller size. Young conelets (1 st cones for P. pinaster and P. halepensis and 1 st and 2 nd cones for P. pinea) varied in number in all three species between 1 and 4. The experiment was replicated three times, on 21-22 April, 28-29 April and 5-6 May 2016. Branches for each experiment were collected at the end of the afternoon of the day before the start of the experiment and kept in the refrigerator until then. Ten adult bugs were assigned to each of the three big cages. Cage number 1 had three males and 7 females and cages number 2 and 3 had four males and 6 females. All adults were individually marked in the thorax with a colour and number. The group of insects of each cage remained constant in all three trials varying only the host pairs to be tested. Between trials all individuals were kept together in a single big cage in the laboratory under controlled conditions and with branches from all the three hosts. For each trial, the ten adults were introduced into the cages by placing them one by one, within a two minutes interval, at the centre of the cage, between the two potted branches. Individual bug's behaviour and localization was recorded after that, at 1-h intervals from 8 a.m. until 6 p.m. for two days.

Cone preference trials
Three separate laboratory trials, with 3 to 5 days length each, were conducted from April to September 2017, using a video camera BRINNO TLC200 Pro. In each trial, two video cameras each videotaping two plastic boxes (23×20×19) cm simultaneously, were set. Each box contained a small branch of P. pinea, a petri dish with wet cotton to keep moisture and small aluminium cases filled with sand to place the cone. In this way, the insect was not allowed to hide underneath the cones. Similarly as in the branch preference trials, the bugs were individually marked with a coloured label. One cone of P. pinea, one of P. pinaster, two of P. halepensis and three adults of L. occidentalis were then added to each box. All cones were from last year of development which corresponds to the 3 rd year in P. pinea and 2 nd year in P. pinaster and P. halepensis. Experiments were carried on in a room under control conditions (26 °C, 60 % RH, 16:8 L:D) from 20-24 April, 2-6 May and 12-14 September of 2017. The videotaping was done using the time lapse function with one picture taken every two minutes, and played back at a speed of one frame per second. Videos were analysed with the program VLC media player 2.2.6 Umbrella for windows (https://www.videolan.org/vlc/index. html). Both the number of times each bug started feeding on a cone (frequencies), and the duration of the feeding was recorded. Feeding was assumed to have occurred whenever stylet insertion was observed.

Testing bug preference for seeds among the three pine species
Two laboratory trials were carried out to evaluate bug preferences for individual seeds of the Mediterranean pines. A two-choice test compared P. pinea and P. pinaster whereas a three-choice test included the three species. The first trial lasted three weeks whilst the second one lasted four weeks. The experiments were carried out using small plastic boxes (20×15×10) cm with a perforated lid for gas exchange in a climatic chamber under controlled conditions (21 °C, 60 % RH, 16:8 L:D).
The two-choice trials were carried out at INRA Orléans, France, and the threechoice one at Centro de Estudos Florestais Lisbon, Portugal.
In the two-choice experiments boxes containing both P. pinea and P. pinaster seeds (n seeds =5 and n seeds =12, respectively) were set (n box =2). No-choice, control experiments were conducted using boxes with only P. pinea seeds (n seeds =10 per box; n box =6) and boxes only with P. pinaster seeds (n seeds =24 per box; n box =2). The experiments were carried out in February and March of 2015 with four adults per box.
The three-choice experiment was performed during January and February of 2017. Twelve boxes, each with ten seeds per pine species, acted as replicates. Each box had three adults.
All boxes included for water supply and as a substrate for resting and laying eggs, a twig of P. nigra in the two-choice and of P. pinea in the three-choice trial. Previous trials using boxes have shown that insect mortality increases greatly when there is no fresh twig inside (personal observation). The use of different pine species was dependent on conifer availability near the laboratory where the experiments took place. In addition to the twig, a petri dish with wet cotton to keep moisture and foam to support the seeds were also added to each box. Control boxes with no bugs were present in both trials. The sex of the adult specimens was not considered since previous studies found no significant differences in the consumption of conifer seeds between sexes (Bates et al. 2000, 2002, Lesieur et al. 2014. Bug mortality was checked every working day, and dead individuals were replaced immediately, using the stock available from the permanent colony. All seeds from the two-choice trial were radiographed before the trial using the HP Faxitron-43855 X-raying apparatus and X-ray sensitive films (Kodak 'Industrex M'), following the procedures described in Roques and Skrzypczyńska (2003) but optimized for the seeds of the pine species tested. Seeds from the three-choice trial were X-rayed at the University of Lisbon, Faculty of Veterinary using the Philips Practix 300 machine and the constants 45Kv / 25mAs -1 . Only seeds showing no damage on the X-rays were used in the trials. Each seed was followed individually. Kernel consumption was estimated by subtracting the final seed weight from the initial. During the trial, seeds were radiographed every week for four weeks to track the damage by L. occidentalis. In the end, all seeds were opened manually, and kernels were separated into damaged and not damaged. Damaged kernels were photographed using a camera Canon 1100 D and their seed coat was carefully analysed under a stereomicroscope and a Scanning Electron Microscope (SEM) TM3030Plus Tabletop microscope Hitachi.

Statistical analysis
The size of the mature seeds, the thickness of the seed coat and the volume of the cones all had a normal distribution. The analysis was made using a one-way ANOVA, with three levels corresponding to the three-host species (P. pinea, P. pinaster and P. halepensis). Post-hoc pairwise comparisons were made using the Student-Newman-Keuls (SNK) test. In both the cone and branch preference trials we used Generalized Linear Models (GLM) with repeated measures (each bug was an individual with repeated observations). To compare frequencies among cones and branches, we used GLM with a Poisson distribution, log link function, and Wald Chi-Square test. In the branch preference trials, we performed the analysis for the overall data for each pine species combination, pulling the three cages, as well as per cage. Finally, we used GLM with normal distribution and log link function to analyse differences in the feeding duration time in the cone trials. Again, each bug was considered an individual with repeated measures. For the seed preference trials, to compare frequencies among seeds, we used GLM with a Poisson distribution, log link function, and Wald Chi-Square test. In the two-choice seed trial, we compared: i) the mean number of seeds consumed between choice and non-choice tests for each pine species; ii) the mean number of seeds consumed between pine species on both choice and iii) on non-choice tests. In the three-choice trial, we analysed the differences in the mean number of seeds consumed between the three pines species with boxes considered as repeated measures. We further compared the seed weight consumed and the percentage of kernel consumed per host and box in both seed trials using GLM with normal distribution and log link function. Boxes with no consumption were removed from the analysis.
All statistical analyses were performed using SPSS, version 24.0 (IBM Corp., Armonk, New York) with a statistical significance level of 0.05.

Size of cones and seeds and coatw thickness
The volume of last-year cones differed significantly among pine species (F=92.38; df=2; p<0.001), with the volume of P. pinea cones being two and three times larger than those of P. pinaster and P. halepensis, respectively. The weight of mature coated seeds also differed significantly among species (F=10387.92; df=2; p<0.001) as well as the kernel volume (F= 1526.33; df=2; p<0.001), and coat thickness (F= 4681.251; df=2; p<0.001: Table 1). Seed kernel was 4 times larger in P. pinea than in P. pinaster and 13 times larger than in P. halepensis. Seed coat thickness was more than 3 times greater in P. pinea than in P. pinaster and 12 times greater than in P. halepensis. The ratio kernel volume/coat thickness (KV:CT) that could be used as a proxy of benefit/cost for the bug showed a higher value in P. pinea (Table 1).

Branch preference
No host preference was detected in either of the three host pine choice combinations on the frequencies of visits per bug (P. halepensis x P. pinea: Wald Chi 2 =2.485, p=0.115; P. pinaster x P. pinea: Wald Chi 2 =0.005, p=0.943; P. halepensis x P. pinaster: Wald Chi 2 =0.008, p=0.927). Overall, 70% of the individuals remained on the same host species during the trial period (48h) with the few changes happening on the first day. When each cage was consider separately, a significant preference was observed for one branch or the other, whereas preferred host species may differ from one trial to the other for the same host species combination (Figure 1).
As a general trend, we observed that the ten individuals from each cage dissociated into one or two fix groups in the three trials (Figure 1). The record of the specific place in the plant where the adults were revealed that in more than 80% of the observations the insects were resting between the needles or on the last year cones, regardless the host.

Cone preference
We found no differences between the three trials and so results were analysed together. The adults were observed visiting and feeding more frequently on P. pinea cones in comparison with cones from the other two species (visiting: Wald Chi 2 =17.42; p<0.001; feeding: Wald Chi 2 =15.31; p<0.001). Per feeding meal, the adults also spent more time feeding on P. pinea cones in comparison with other cones (Wald Chi 2 =12.05; p=0.002) ( Table 2).

Bug preference between seeds of P. pinea and P. pinaster
For four weeks, the four individuals always ate two, out of ten, seeds of P. pinea per box, either if it was offered alone (non-choice tests) or mixed with P. pinaster seeds (choice tests) (Table 3). Conversely, when adults fed on P. pinaster alone, the number of seeds consumed was on average 5.5 ± 1.7, which was higher than the P. pinaster seeds consumed in choice tests, 0.5 ± 0.5 (Wald Chi 2 =5.271, df=1, p=0.022). Considering non-choice tests Table1. Cone and seed average measures (± SE) of the three host species. Different letters within a column indicate significant differences between the values per host species after ANOVA tests (p-value=0.05) followed by SNK test.

Host species Visiting frequency Feeding frequency Time spent per feeding meal (minutes) P. pinea
13.4 ± 4.2 a 5.5 ± 1.2 a 131.9 ± 25.9 a P. pinaster 5.1 ± 1.2 b 2.0 ± 0.6 b 53.2 ± 12.0 b P. halepensis 3.7 ± 1.5 b 1.1 ± 0.5 b 62.5 ± 13.3 b Figure 1. Branch preference trials. Bugs distribution per host in each cage and for each host pair comparison in the preference trials using potted branches. Numbers with asterisk on the x axis correspond to male bugs. The absence of bars means that individuals were never observed on the branches during the trial, but remained on the floor or on the walls of the cage.

Testing bug preference among seeds of P. pinea, P. pinaster, and P. halepensis
Seed consumption was observed on 8 out of the 12 boxes ( Figure 2). On these boxes overall, there were no significant differences in the number of seeds consumed between species (Wald Chi 2 =0.011, df=2, p=0.995) ( Table 3). The total number of seeds consumed was 15, 16 and 15, respectively for P. pinea, P. pinaster, and P. halepensis. However, when considered the weight of the seed kernel eaten by the adults, we found significant differences among pine seeds (Wald Chi 2 =117.632, df=2, p<0.001). On average the bugs ate more P. pinea seed kernel than the two other species (p<0.001) which, in turn, had no difference between them (p=0.726). When analysing the consumption in terms of the percentage of consumed kernel per host in each box and trial differences were obtained in both two and three-choice trial (Wald Chi 2 =1002.485 df=1, p<0.001, Wald Chi 2 =15.625, df=2, p<0.001, respectively). The adults consumed a higher percentage of P. pinea kernels that the other two hosts (p<0.001 for both comparisons) (Table 3).

Discussion
The impact of an invasive species must be assessed at different levels from the individual to the ecosystem processes level (Parker et al. 1999). The seed feeder, L. occidentalis is classified as having a negative impact to the native biodiversity at the individual level (herbivory, predation, competition, disease transmission) and, in addition, a negative economic impact (Rabitsch 2008). Indeed, the presence of the bug in the Mediterranean Basin is a severe threat to the Mediterranean pine nut production as very relevant non wood forest product (Roversi 2009, 2017 by causing direct damage to cones and seeds and as putative vector of the fungus Diploidia pinea (Luchi et al. 2012, Tamburini et al. 2012). Here we tested the preference of the bug on the three main lowland conifer species in the Mediterranean Basin region, P. pinea, P. pinaster and P. halepensis at three levels of selection: branch, cone, and mature seed. These three pine species significantly differ in the shape of the tree silhouette and also in the cone size, seed size, and seed coat thickness. Cues behind tree selection by cone feeders may be related to these morphology traits (Turgeon et al. 1994), but also to chemical factors which are indicators of the tree nutritional quality or of its level of chemical defence (Schultz 1988). In fact, there are two predominant hypotheses for the process of host selection by insects: first, that an increased abundance of insects is explained by increased plant vigor (Plant Vigor Hypotheses, Price 1991), or secondly, that it is explained by increased plant stress (Plant Stress Hypotheses, White 1969). Studies on how herbivores select the host plant reported responses according to the theory of plant vigor (Moran and Whitham 1988, Waring and Price 1988, Kimberling et al. 1990, Mopper and Whitham 1992, Hull-Sanders and Eubanks 2005, Mitchell 2006), plant stress (Bjorkman et al. 1991, Rappaport and Wood 1994, Cobb et al. 1997, Virtanen and Neuvonen 1999 or even both simultaneously (Fernandes 1992, Pérez-Contreras et al. 2008) depending on the insect species, its trophic sub-guild (e.g. herbivores that feed on growing plant parts seems more likely to attack vigorous plants (Price 1991, White, 1993 or even on the type of experiment (Waring and Cobb (1992), in a review found that on observational studies, White's theory was prevalent whilst on experimental studies was the vigor hypothesis).
Under natural conditions, L. occidentalis has been shown to select a host in a multilevel process. At first, the bug selects a tree, then a cone and finally a seed where to feed upon (Blatt 1997). Regarding the selection of the tree, the bug tends to select vigorous trees, i.e. denser canopies and longer needles, (Farinha et al. 2018b) bearing moderate crops because it is usually a sign of having larger cones and therefore larger seeds (Blatt 1997, Blatt and Borden 1999, Richardson et al. 2017. With all that has been said in mind, we hypothesized that branch selection by this bug, a polyphagous insect that feeds on growing plant parts, should rely mostly on vigour (e.g., increased resources, higher food quality, and lack of induced defensive compounds) and cone size and not so much on plant species and their chemical profiles.
In our experimental trials using potted branches with cones, the seed bug showed no clear preference between the three host species but rather a preference for one of the two branches on each trial. The plant vigour hypothesis (Price 1991) for the selection of the host may explain this result. The vigour of the chosen branch (e.g. nutritional state, morphology) and the size of the cones on the branch may have been more critical for a polyphagous insect like L. occidentalis than the species of the host plant. Although, we choose branches with similar size, with no signs of diseases and always bearing cones of all ages, yet differences on the number of cones and on its physiological status (e.g. nutritional quality, allelochemicals) might have caused differences between host species and between trials. Furthermore, since branches came from different locations and trees, there could be both a tree and site effects.
Preference studies on a related species, L. phyllopus (L.), which is also polyphagous, have evidenced that nutritional and host quality issues (wild vs cultivated plants) are more determinant in host plant selection rather than plant species (Mitchell 2006). A study of L. occidentalis impact in a P. pinea plantation, also revealed higher bug damage on irrigated and fertilized trees than on control ones, showing a bug preference for high vigour trees (denser canopies, greater needles) (Farinha et al. 2018b).
The branch preference trial also revealed an overall trend for L. occidentalis to form two groups of individuals per cage, one in each plant. In general, the composition of the two groups remained similar in each cage in all three trials. During the time between trials, all insects (n=30, 10 from each cage) were placed together in a single cage but when they were replaced one by one again in the test cage they regrouped in the same way as in the very first trial. Furthermore, once the individuals had chosen one of the plants, they remained there, in 70% of the cases, throughout the experiment. Group dynamics and not an individual host selection is, thus, a more suitable explanation for the results obtained. The gregarious behaviour of this insect has already been proposed by other authors (Koerber 1963, Mitchell 2006. Preference for a host species was further tested exposing cones to adults in trials using video cameras. The use of video recording can be very informative on the insect preference because it allows capturing the behaviour of the insect continuously. Moreover, the use of cameras with time lapse mode made it possible to process all data since it condenses several days of filming in movies of few minutes. So, by tracking the feeding behaviour, a clear bug preference could be observed for visiting and feeding on P. pinea cones. Furthermore, the individuals spent twice more time feeding, by each feeding meal, on this host species than on the other two species. A higher reward per seed would probably keep insects feeding longer times. It has also to be noted that the seed coats were not yet totally hardened inside cones because we used last year cones collected in the spring. Under natural conditions, the larger cones of P. pinea could be more attractive since visual stimuli are important to this bug in the process of host selection (Blatt andBorden 1999, Richardson et al. 2017). The higher reward value may further benefit the performance of L. occidentalis, when feeding on P. pinea seeds, namely through a higher survival rate and faster development, as indicated by Ponce et al. (2017). Bernays and Minkenberg (1997) in an experimental study with seven polyphagous insects (four Lepidoptera and two Hemiptera) came to the conclusion that it is the greater resource availability rather than the nutritional enhancement or differences in allelochemicals among host species that resulted in a higher performance (survivorship, gain in mass and fecundity). We assume that in our cone preference experiments, the larger cones of stone pine represent a greater resource availability comparing to the other hosts.
In respect to seed trials, no clear preference between mature seeds of P. pinea, P. pinaster and P. halepensis was observed if we compare the number of seeds consumed. However, P. pinea kernels are much larger than the others. Furthermore, it has been shown that the same insect feeds several times on the same seed and that different insects may also feed on that same seed by sharing the feeding hole (from video recording observations, data not showed) (Farinha et al. 2018a). Therefore, the mass or the percentage of kernel consumed is a better indicator of the real consumption by the individuals. When considering kernel consumption, P. pinea emerges as the most consumed host species in all trials. When expressed in percentage of consumed kernel per host species and per box, L. occidentalis consumed about 97% and 92% of P. pinea kernels in two-choice and three-choice tests, respectively.
Other preference study at the seed level showed that L. occidentalis appeared capable of differentiating a viable seed from one infested by chalcid, Megastigmus spermotrophus (Hymenoptera: Torymidae), discarding the latter (Blatt 1997). However, since all the seeds used in our trial were first radiographed and only the healthy ones were used, we hypothesize that feeding onset in a given seed might be random or visually determinate. After first opening a hole on the seed, feeding activity would be concentrated on that seed, which in the case of P. pinea seeds would satiate longer, further requiring less number of consumed seeds per bug. Also, bigger seeds may result in higher nutrients income in shorter periods which optimize the feeding. This result becomes evident when we compare P. pinea and P. pinaster kernel consumption in non-choice tests (over 5 times more mass consumed in boxes with P. pinea seeds) ( Table 2). In another preference trial using mature seeds, Lesieur et al. (2014) found no difference between host species. Still, in that study, the size of the seeds did not differ so much among tested host species.
Despite the larger size of P. pinea seeds, which constitute a more significant reward, the seed coat implies a higher cost, being three and twelve times thicker than that of P. pinaster and P. halepensis, respectively. Even so, a benefit/cost analysis pointed to P. pinea seeds as being more advantageous. Feeding behaviour videos show that drilling a hole in P. pinea seeds can take more than 8 hours to complete (unpublished data) but then the benefit is high and, most importantly, it is shared by the remaining insects of the box as other bugs use the same hole to feed. We should note that, although easy to replicate under laboratory conditions, host selection trials resourcing to mature seeds have a limited ecological significance because the seeds, enclosed within the cones, are not subjected to selection in natural conditions.

Conclusions
In spring or early summer, depending on the climatic conditions, this insect becomes active and begins to search for a site with coniferous where to feed and reproduce. No data can be found on the bug behaviour when leaving its winter shelter except for an inconclusive study by Richardson (2013) in Lodgepole pine seed orchards for two consecutive years. Does it return to the same place as the year before or disperse elsewhere? How does it select the site to colonize? Is there any host preference at this moment?
Among the three main pine species in the Mediterranean Basin, P. pinea, P. pinaster and P. halepensis, our results support evidence that the cones and seeds of P. pinea are highly rewarding for L. occidentalis. From a nutritional point of view, we may then expect that P. pinea trees and plantations may favour L. occidentalis population growth. Whereas in seeds and cones there was always a preference trend for P. pinea, no clear preference for host species was detected when we used potted branches.
It must be highlighted that the quality of the host plant rather than the species, and the aggregation behaviour of this bug are important factors to take into account when designing the methodology of future host preference studies. Furthermore, larger scale trials are required. The population dynamics of this insect must be a priority research topic. No management plan will succeed without understanding which factors influence the distribution and abundance of this pest, including the availability of, and its performance on, different hosts.