Does Hartigiola annulipes (Diptera: Cecidomyiidae) distribute its galls randomly?

It is expected that environmental conditions impact the distribution of galls on host plants. Moreover, insects may induce their galls randomly or choose certain parts of a host to induce such growths. This study aimed to determine whether or not the gall midge, Hartigiola annulipes (Diptera: Cecidomyiidae), which induces galls on leaves of the European beech (Fagus sylvatica), prefers leaves facing a particular cardinal direction. In addition, we wanted to ascertain whether the galls are evenly spread across three leaf zones: proximal, median and distal, distinguished by dividing leaf area along the midrib. The results show that H. annulipes chose leaves facing various directions in different studied locations, and leaf choice is not restricted to the specifi c leaf area, a parameter that accurately refl ects the light conditions of leaf growth. Moreover, the medial leaf zone was preferred, while the distal zone was avoided. The choice of the leaf zone modifi ed the distance between the mid-rib and a gall. Gall distribution in the crown of trees is probably random, while at the leaf level, it is determined by leaf morphology.


INTRODUCTION
The search for factors infl uencing the distribution of insect galls within the crowns of trees and on plant organs has led to several hypotheses suggesting that herbivores choose plant modules to feed on depending on the vigour or size of these modules.One of the most popular is the Plant Vigor Hypothesis (PVH) (Price, 1991), which claims that herbivores choose the most vigorously growing modules to feed on.It appears that the distribution of the eastern spruce adelgid (Adelges abietis Linnaeus) refl ects PVH, but small galls develop on lateral shoots of Norway spruce (Picea abies (L.) H. Karst.) and large galls on main shoots (Pilichowski et al., 2014).The main shoots of spruce are usually larger than lateral shoots, but there are fewer of them, thereby limiting the available space for galls.The aphid Kaltenbachiella japonica Matsumura galls the most vigorously growing region of a leaf (Aoyama et al., 2012).Gall-inducers may depend on high vigour and reduce it at the same time by deforming or hindering the growth and development of infested modules (Mendel et al., 2004;Cuevas-Reyes et al., 2008;Urban, 2010).
In addition to different hypotheses (such as PVH, optimal module size hypothesis -OMSH), it is suggested that different abiotic factors may affect gall development, survival and oviposition behaviour, due to their direct effect on a gall-inducer or plant host.Among these factors, insolation, levels, the locations chosen for this study were different types of forests.Leaves collected were free of signs of herbivory and pathogens as far as it was possible to determine with the naked eye.
This study is based on two sets of data, the fi rst was used to determine preferences for galling leaves facing particular cardinal directions and for an initial study of the distribution of galls on these leaves.The second was used to model the relationship between the factors affecting the distribution of galls on leaves.The permission to collect leaves in nature reserves was provided by the Regional Directorate for Environmental Protection in Gorzów Wielkopolski, Poland (document ID: WPN-1-6205.28.2015.AI).To better interpret the results, each location was assigned a label (two beech forests: BF-01 and BF-02 -both are nature reserves, two Scots Pine forests: PF-01 and PF-02 -both are forestry managed forests, one park: P-01 and one urban forest: UF-01) (Table 1).

Distribution: cardinal directions and the preliminary study of the distribution of galls on leaves
To determine whether there is a preference for galling leaves facing a particular cardinal direction, three thousand leaves were randomly collected in October 2015 from three locations in western Poland (Table 1).
At each location (BF-01, BF-02 and PF-01), one thousand leaves (with and without galls) were gathered from fi ve randomly chosen beech trees (several years old) growing in the understory: 200 leaves per tree and 50 per cardinal direction (N, S, E, W).The maximum height at which the leaves were collected was 2 m.Galls of H. annulipes were counted in three zones on the leaf lamina: proximal, medial and distal.Each zone was determined by dividing the leaf perpendicularly to the midrib.The proximal zone was defi ned as the zone at the base of the leaf (Appendix 1).
To determine whether there were differences in the quality of leaves facing each of the four cardinal directions, their specifi c leaf areas (SLA [cm 2 /g]) were determined as follows: 100 leaves were collected from fi ve understory beech trees during late summer, including 20 per tree and 5 per cardinal direction (N, S, E, W) in each stand.The leaves were scanned using an Epson Perfection V700 scanner (Seiko Epson Corporation).The images were processed in ImageJ software version 1.48 by Wayne Rasband, National Institutes of Health, USA, to determine leaf areas.Then, the leaves were oven-dried (60°C, two days) and weighed using an OHAUS Corp. Adventurer Pro scale Model AV2102CM with d = 0.01 grams (precision of scale division).thasarathy & Anil, 2016).This may vary, however, with successional stage, and the outcome might be that food availability, not quality, determines herbivory (Poorter et al., 2004).Generally, SLA (and other morphological leaf traits) values are associated with herbivory and fungal damage, which effect the growth rates of trees (Schuldt et al., 2017).Thus, it is worth taking SLA into consideration when studying the distribution of galls on leaves.
In this study, we aimed to evaluate the preference of H. annulipes (Cecidomyiidae) to gall leaves facing a particular cardinal direction and determine what factors (location, leaf zone, leaf blade area, number of galls) infl uence the distribution of galls on leaves.Based on the results of previous studies we hypothesized that (1) light conditions for leaf growth (expressed as SLA) will affect the distribution of galls in the crowns of trees, with a preference for galling shaded leaves (high SLA), (2) H. annulipes avoids southfacing leaves, those that receive the most sunlight, and (3) H. annulipes prefers to gall the proximal zone of a leaf.

MATERIALS AND METHODS
Beech gall midges, Hartigiola annulipes Hartig (Cecidomyiidae), produce one generation each year and spend most of their life within single-chambered, hairy galls that develop on leaves of European beech (Fagus sylvatica L.) (Rohfritsch, 1971).The galls detach from the leaves during late summer and early autumn and overwinter in the leaf litter.Imagines emerge from galls in spring, and after mating, females lay eggs on the bottom side of leaves at the beginning of May (Rohfritsch, 1971).In this study, the number of galls counted on leaves included galls that failed to develop and survive, galls that matured and detach from the leaves, and galls still attached to leaves.The purpose was to study the preferences of H. annulipes for inducing galls on different parts of a leaf.Whether the galls successfully matured was not measured in this study.Due to the possible effect of microclimatic conditions on the distribution of galls at tree crown and leaf

Distribution of galls on leaves -a model
To determine the frequency of occurrence of the galls on different parts of a leaf, 1092 leaves of the common beech, each bearing at least one gall of H. annulipes, were collected at four locations (BF-01, P-01, UF-01, PF-02) in Poland from trees growing in the understory (up to 3 m tall) between September and October 2015 (Table 1).Locations of the material collected differed in terms of type of forest, which is expected to be an important factor affecting the results.The number of leaves collected at each location was different due to differences in the availability of beech trees, abundance of galled leaves and forest management.In the park in Nowy Kisielin (P-01), there was only a single European beech tree growing in the understory suffi cient for material collection, while fi ve trees were chosen for collection at the other three locations.In Zielona Góra (PF-02), the infestation level was very high, and thus, the number of leaves collected there was the highest.
The area [mm 2 ] and length [mm] along the midrib of each leaf was measured using ImageJ software.ImageJ was also used to measure the perpendicular distance of H. annulipes galls from the midrib [mm].The total number of H. annulipes galls was recorded.

Statistical analyses
The χ 2 test was used to check differences between the presence and absence of galls in each leaf zone, as well as the preference for particular cardinal directions (data set 1).An ANOVA was used to estimate the effect of individual beech trees, cardinal direction and interaction on SLA.One gall was randomly chosen out of 1092 galls for each leaf analyzed.We used log-transformed values of the distance between the midrib and galls to normalize the distribution of the data.Then, a linear mixed model (LMM) was used to determine the effect of the factors (leaf zone, leaf blade area, number of galls, location and number of beech trees) on the distance between the midrib and galls (dependent variable) (data set 2).The numbers of beech trees and locations were treated as random factors and beech trees were nested in the locations (the script is shown in Appendix 2).
Statistical analyses were performed using JMP PRO 13 software (SAS Institute Inc., Cary, NC, USA).

RESULTS
Hartigiola annulipes galls were very common as they were found on more than 35% of beech leaves.A total of 4512 H. annulipes galls were recorded on 3000 leaves.The abundance of galls differed at the different locations.Most galls (3375) were found at PF-01; and 655 and 482 were recorded in the nature reserves BF-02 and BF-01, respectively.Of the collected leaves, 1949 were not galled and 346 had only one gall.The highest number of galls per leaf was 76 (in PF-01).The numbers of galls and galled leaves are shown in Appendix 3.
The infl uences of location, cardinal direction and interaction factors on the SLA were not signifi cant (Table 2).
The distance between the mid-rib and galls was signifi cantly associated with two factors: leaf blade area and leaf zone (Table 3).The distance between the mid-rib and galls increases with leaf area, and the distance between the mid-rib and galls in the distal zone was shorter than in the proximal zone (p < 0.0001).The analyses were done for the proximal zone.The model equation obtained was Y = 0.3273 -0.2334DZ + 0.0002A, where (Y) was the distance between the mid-rib and galls, (DZ) was the distal leaf zone and (A) was the leaf area.

DISCUSSION
Our results show that H. annulipes chooses leaves facing different cardinal directions.Skrzypczyńska & Lorenc (2005) and Wrzesińska (2012) claim that H. annulipes galls are more common on shaded leaves.The amount of light available for plants in a forest depends on the type and age of the forest (Peters, 1997;Guo et al., 2017) and differs in forests composed of conifers and broad-leaved species (which may shed their leaves seasonally) due to variations in the densities of their canopies, species of trees present, their growth rates and requirements.Older trees, which are usually taller than those growing in the understory, as well as fast growing individuals, compete more effi ciently for light than smaller or slow-growing trees (Peters, 1997;Kobayashi et al., 2000;Schulte et al., 2013).These facts may be an important issue affecting the results of this study since the trees studied were growing in the understory of different types of forest.However, since gall midges are considered to be poor fl yers (Urban, 2000), studies of gall distribution in mature tree crowns may be pointless.In contrast, Kampichler & Teschner (2002) do not agree that gall midges are poor fl yers.The overstory tree species composition in forests may infl uence the amount of light reaching the understory when the leaves are mature (Sercu et al., 2017); moreover, natural and managed forests with different species compositions may infl uence understory plants differently (Thomsen et al., 2005), and various types of managed forests infl uence the biodiversity (Gao et al., 2014).These fi ndings indicate it is likely that the stands studied differed in terms of environmental conditions (e.g., light transmittance), which may have infl uenced the preferences for leaves facing in particular directions.
There were no differences in SLA, which was assumed to be an indirect measure of light conditions.The expectation was that SLA would be important in determining preferences for leaves facing different directions since there are reports of relationships between SLA and herbivory (Poorter et al., 2004;Parthasarathy & Anil, 2016).The results presented do not support the fi rst hypothesis that SLA depends on the direction that leaves face.At BF-01, the galls were mainly on south-facing leaves and north-facing leaves were avoided.At the BF-02 reserve, where there is a natural beech forest by a lake, and at PF-01, where we collected leaves from understory trees growing in a managed conifer forest, the fi rst leaves to be galled were eastfacing (which are the fi rst leaves to be exposed to sun in the morning).The comparison of the frequency of H. annulipes galls at BF-02 and PF-01 indicates that even though insects prefer east-facing leaves, environmental conditions modify preferences towards other cardinal directions.By environmental conditions, we specifi cally mean microclimatic differences (abiotic factor) between beech and pine forests and the availability of trees for infestation (biotic factor).At BF-02, where leaves were collected from beeches growing on the western side of a lake, the east-facing leaves may receive more solar radiation due to openings in the canopy and distribution of the trees.However, at PF-01, a managed pine forest with a few beech trees, similar preferences were recorded for west-and east-facing leaves.It seems that in this case, H. annulipes is more opportunistic and uses the resources randomly.It is likely that females choose oviposition sites based on the abundance of food rather than its quality (vigour).Statistical analyses indicated no infl uence of the factors studied on SLA.Even though the trees chosen for SLA analyses were solitary and in the understory (not in contact with other trees) in order to avoid the effect of shading, it was not possible to determine the effect of cardinal direction on leaf quality (expressed as SLA).These results led to the assumption that the light conditions were similar in each of the four main cardinal directions.We could not fi nd any evidence in the literature that females of H. annulipes were attracted to light, which could account for their oviposition behaviour.Do females become active in the morning after sunrise and lay their eggs on young, east-facing leaves?We speculate that they do, and later during the day, they may randomly lay their eggs on leaves, due to an increase in the activity of the insects in response to rising temperatures during the day.Skrzypczyńska & Lorenc (2005) and Wrzesińska (2012), who claim that H. annulipes galls were more common on shaded leaves, did not support this thesis with a statistical analyses, and the aim of their studies was not to determine the preferences of H. annulipes.Yukawa et al. (1988) show that Mycophila sp.(Cecidomyiidae) exhibits phototaxis.Similar conclusions come from the study of Dasineura oxycoccana Johnson (Cecidomyiidae) by Roubos & Liburd (2010) as it has the same preference for light and a southerly direction, whereas females of Stenodiplosis sorghicola Coquillett (Cecidomyiidae) are colour-and odour-attracted (Sharma & Franzmann, 2001).If females of H. annulipes show the same pattern of behaviour, they may also react to the different light wavelengths refl ected by or passing through young leaves during the day according to the position of the sun.The optical properties of leaves and bark of common beech change during the plant's lifetime (Pilarski et al., 2008).Leite et al. (2011) report that an unidentifi ed galling species belonging to the Eurytomidae (Hymenoptera) prefers the middle zones of Caryocar brasiliense Camb.north-facing leafl ets.The positioning of galls may be strongly infl uenced by inter-specifi c competition within a tree crown (Leite et al., 2011), with fungal endophytes being a threat to the survival of gall insects (Wilson & Carroll, 1997) and leaf-miners (Wilson & Faeth, 2001), or as McGeoch & Price (2004) show, positioning may be weakly affected by inter-specifi c competition.For H. annulipes, such a competitor may be Mikiola fagi Hartig (Diptera: Cecidomyiidae), which also galls the leaves of F. sylvatica.Nonetheless, M. fagi galls were absent at PF-01, while they occur at BF-01 and BF-02, but H. annulipes galls were by far the most abundant (personal observation).Thus, a comparison of distributions of the galls of both species occurring in the areas studied is diffi cult.This raises a need for further studies.Aphidoletes aphidimyza Rondani (Cecidomyiidae), a predatory gall midge, does not avoid ovipositing at sites already occupied by conspecifi c females and their eggs (Kohandani et al., 2017).Females of A. aphidimyza tend to clutch their eggs at a single oviposition site (Kohandani et al., 2017).Perhaps females react to a site that is acceptable for other females as an optimal site.However, intraspecifi c competition cannot be avoided, as the food resources for larvae are limited.This would account for the occurrence of 76 H. annulipes galls on a single leaf (Appendix 3) at PF-01, where beech trees are rare.
Food resources and natural enemies are the most important biotic factors infl uencing the success and survival of herbivores (Whitham, 1978;Kaplan et al., 2014) along with host plant phenology (Urban, 2000;Aoyama et al., 2012).In addition, ontogeny and microclimatic conditions modify the degree of herbivory (Stiegel & Mantilla-Contreras, 2018).Since galling animals are specialists, it is expected that they are favoured or limited by these two factors.Hence, galls should occur in a place where it can be well-supplied by the host plant and where the risk of predation or parasitism is low.Furthermore, the cardinal direction in which a leaf faces and on which galls are located could determine the likelihood of it being attacked by herbivores, predators (Evans & Gregoire, 2006) and/or parasitoids (Zargaran et al., 2011).In contrast, some studies indicate no preference in the occurrence of galls facing any of the four cardinal directions, e.g., Forda riccobonii Stefani (Hemiptera: Pemphigidae) forming galls on the Pistacia atlantic Desf.(Martinez, 2009).The random distribution of galls according to cardinal directions minimizes parasitoid attacks at some sites since there are studies indicating that light conditions may modify the behaviour of parasitoid wasps (Luo & Chen 2016;Cochard et al., 2017;Kelch et al., 2017).Consequently, such a distribution of galls might be considered as a trade-off between avoiding parasitoids and occupying the most benefi cial galling sites.
Moreover, there are reports that herbivores prefer particular parts of plants.Various chemical compounds protect plants against herbivores and pathogens.For example, phenolic compounds, the quantities and concentrations of which differ in different plant species and their parts (Abrahamson et al., 2003;Oliveira Ribeiro et al., 2014).Furthermore, they change seasonally (Bernal et al., 2013;Oliveira Ribeiro et al., 2014) and with age (Bhakta & Ganjewala, 2009) and may be infl uenced by environmental factors (Estiarte et al., 1994;Loponen et al., 2001;Borges et al., 2013).Petrakis et al. (2011) report phenols in the leaves and bark of F. sylvatica in Greece and that infested beech leaves have higher concentrations of total phenolic compounds and that it increases with the number of insects attacking beech trees.The concentration of phenolics in host organs can vary, which affects the distribution of galls (Zucker, 1982;Bergvinson, 1997;Giertych & Karolewski, 2000;Gutterman & Chauser-Volfson, 2000;Ferreira et al., 2014).Thus, we expected that H. annulipes would avoid galling the distal zone and prefer the proximal third of leaves.Indeed, the results show that the frequency of galls occurring in the distal zone (Fig. 2) is the lowest when compared to the median and proximal zones.However, they also show that H. annulipes slightly prefers the median zone to the proximal zone of leaves, but only in natural beech forests.In the managed Scots pine forest, with a few beech trees in the understory (PF-01), the frequency of galling of the proximal and median zone is much lower.This may be due to the rarity of the host plant, which results in a high number of galls per leaf, the highest recorded number of H. annulipes galls on one leaf in PF-01 reached 76 (Appendix 3), while in the BF-01 and BF-02 reserves, it reached 20 and 12, respectively (Appendix 3).
The position of a gall on a leaf may also be important if the distance from the midrib is taken into consideration.The LMM model incorporates only two signifi cant factors (zone of occurrence and leaf area) that infl uence the distance between a gall and the midrib.Relative to the proximal zone, the distance of a gall in the distal zone is signifi cantly shorter, and galls that developed on the largest leaves tend to be further from the mid-rib.These results correspond with the obvious presumptions that as the distal zone narrows towards the tip, galls have to develop closer to the mid-rib, which also becomes thinner than in the proximal zone.There is still too little knowledge about the distributional patterns of galls on leaves.Giertych et al. (2013) report that leaf size and distance from the petiole to the gall (C.quercusfolii) are positively correlated, but they did not discuss the results for the distance of the gall to the midrib as they were not signifi cant.Leite et al. (2011) show that the central parts of C. brasiliense leafl ets are more frequently galled than the borders and midrib.We suppose that large leaves allow the establishment of galls further from the mid-rib than small leaves, because a larger area and thicker lateral veins provide galls and larvae with enough resources to ensure their growth and development.Furthermore, galls may not be affected by the high concentrations of harmful secondary metabolites fl owing in the mid-rib.In this way, H. annulipes can optimize the location of its galls.Stiegel & Mantilla-Contreras (2018) studied herbivory rates of common beech at three different heights.The authors assume that it is impossible to establish a herbivory-pattern for a single species.Instead, they propose to study such patterns at the guild level.
It must be mentioned that the material used in this study was collected from understory trees, and thus, the results may be limited to explaining the distributional pattern of galls of H. annulipes on young trees.If this gall-midge is sensitive to concentrations of various chemical compounds, especially secondary metabolites, then it may respond with different distributional patterns at different heights and in different parts of the canopy.Dudt & Shure (1994) report that the shade-tolerant plant, Cornus fl orida, L. contains more phenolics than shade-intolerant Liriodendron tulipifera L. F. sylvatica is also considered to be a shade-tolerant species, and Stiegel et al. (2017) suggest that the high carbon content of F. sylvatica leaves in upper canopies refl ect high concentrations of phenols.Numerous studies report a connection between the chemical quality of leaves and light availability (Crone & Jones, 1999;Roberts & Paul, 2006;Oliveira Ribeiro et al., 2014), with shaded leaves having high concentrations of phenols (Dudt & Shure, 1994); however, the latter is not a general rule, as proven by Karolewski et al. (2013) and suggested by Stiegel et al. (2017) in the case of F. sylvatica.

CONCLUSIONS
This study shows that Hartigiola annulipes galls the leaves of common beech randomly at the crown level and leaves facing different cardinal directions.
H. annulipes exhibits some preference for inducing galls on east-facing leaves and tends to avoid south-facing leaves.
H. annulipes prefers to induce galls in the medial zone of a leaf and avoids the distal zone.
SLA values did not explain the distribution of galls on leaves.
The choice of leaf zone can infl uence the distance between a gall and the mid-rib.

Fig. 1 .
Fig. 1.Radar charts based on χ 2 tests, presenting frequency of occurrence of H. annulipes galls on leaves facing each of the four cardinal directions (N, E, W, S) at each study site.BF-01 ("Bukowa Góra" reserve) and BF-02 ("Nad Jeziorem Trześniowskim" reserve) are natural beech forests.PF-01 (Stary Kisielin -Zielona Góra) -a managed Scots pine forest with beech trees occurring in the understory.Probabilities shown in each chart represent p values obtained from the χ 2 tests.

Table 1 .
Description of study sites included in the two data sets, with the numbers of trees chosen for collecting leaves and the numbers of leaves collected at each location.

Table 2 .
ANOVA results of relationships between specifi c leaf area (SLA) of beech leaves and the following three factors: individual beech tree, cardinal direction and interaction.Abbreviations: DF -degrees of freedom, DFE -degrees of freedom for error, F -result of the F-test, p -probability.

Table 3 .
Linear mixed model relating distance between the mid-rib and galls on different parts of a leaf (zone), leaf area and number of H. annulipes galls.Model includes locations and trees nested in location as random effects.SE -standard error.