Fallen fruit: A backup resource during winter shaping fruit fly communities

Fallen fruits provide important feeding and breeding substrates for insects such as Drosophilidae and can be a potential trophic reservoir when usual host fruits become scarce. Recently, two invasive fruit fly species, Drosophila suzukii and Chymomyza amoena, have become established in Europe and are expected to alter existing Drosophilidae communities. In this study, carried out between September 2021 and April 2022 in northern France, we aimed to disentangle the relative roles of microclimatic, landscape and local factors driving the diversity of the Drosophilidae community in decaying fruit across seasons. Minimum site temperature during the week preceding sampling and the proportion of rotten fruit tissue had the strongest positive influence on Drosophilidae abundance and species richness. Drosophilidae abundance also increased with urbanisation (portion of building cover) around the sampled trees. Decaying apples were important breeding sites for C. amoena across seasons, but provided a suboptimal substrate for D. suzukii, which was only present in late summer. This study sheds light on the important role of unharvested fallen crop fruit in maintaining the diversity of an insect family that is generally overlooked in field studies. It also emphasises the importance of considering multiple scales and factors when studying the interactions between invasive species, native species and their shared trophic resources. Finally, our data highlight the importance of the Drosophilidae community in recycling agricultural waste.


INTRODUCTION
In agricultural systems, herbivorous insects attract attention because of their pest status and their damaging impacts on crop production and storage (Manosathiyadevan et al., 2017).Most research studying fruit pests focuses on the effects of insect attacks before or during fruit ripening, when the economic consequences are greatest (Alford, 2007;Herrera, 1982).However, what happens after the harvest period is often neglected including, for instance, the role of insects in organic matter recycling in crop fields.Fallen fruit can play a significant role as feeding and breeding substrates for arthropod communities (Alamiri, 2000;Mészárosné P oss et al., 2022;Rohlfs & Hoffmeister, 2004).They could also represent a potential reservoir for insect pests when cultivated fruits become scarce, providing a seasonal refuge and potentially initiating future pest outbreaks (Liquido, 1991).Nevertheless, the temporal dynamics of arthropod communities across the 'lifespan' of a decaying fruit have largely been overlooked.
Drosophilidae species are of particular interest due to their significant role in the recycling process of decaying fruit.The success of fruit infestation by Drosophilidae species is influenced by numerous environmental factors.At the regional scale, climatic and microclimatic conditions influence their presence.In particular, temperature and precipitation patterns affect thermal limits and desiccation tolerance (Hoffman, 2010;Kellerman et al., 2009;Ulmer et al., 2022).At a finer scale, landscape composition, that is, the cover of natural and anthropic habitats surrounding a site, is a key driver of their diversity and abundance (Delbac et al., 2020;Furtado & Martins, 2018;Poppe et al., 2016;Wen et al., 2023).At local microclimate scale, during cold seasons or cold parts of the day, decomposing fruit can provide niches that are warmer than the surrounding environment, as microbial communities produce heat during organic decomposition (Ryckeboer et al., 2003).All these factors can independently influence the infestation rates of different Drosophilidae species.Finally, the structure of Drosophilidae communities is likely to be highly dynamic given that fruit can display profound gradual changes over the period of decomposition, affecting the quality and quantity of substrate they can provide for insects.
Drosophilidae community structure is also strongly influenced by biotic interactions.Individual species may limit or facilitate infestation by other species, depending on their ability or preference to oviposit and develop in a fruit at a given stage of decomposition and on their competitive ability (Davis et al., 1998;Hoffman, 2010).The number and identity of competitors can influence the range and abundance of species (Davis et al., 1998).Furthermore, invasive species can disrupt the native community assemblage and cause cascading effects across trophic networks, sometimes also facilitating native species infestations (Poyet et al., 2014;Rodriguez, 2006).Worldwide, communities of frugivorous species have been impacted by invasive exotic species.
For instance, in Europe, the invasive members of the Drosophilidae family, Drosophila suzukii Matsumara and Chymomyza amoena Loew, are highly polyphagous and could profoundly alter the structure of Drosophilidae communities developing in fresh and decaying fruit by using the same resources as native species (Band et al., 2005;Pajač Živkovi c et al., 2017;Poyet et al., 2014Poyet et al., , 2015)).
On a short timescale, changes in the nutritional composition of fruit during the decomposition process can affect fly foraging decisions and adult oviposition choices (Silva-Soares et al., 2017).Fruit with characteristically slower and longer decomposition allows the study of the temporal dynamics and environmental drivers of the Drosophilidae community depending on them.Prolonged decomposition could include successive decomposition stages characterised by different chemical compositions and textures, sometimes spanning different seasons and potentially offering successional niches to different Drosophilidae species.Apples (Malus domestica Borkh) provide an excellent candidate fruit for such studies.Apples are a major crop fruit in Europe and worldwide (FAOSTAT, 2023;FranceAgriMer, 2021).However, a considerable proportion of many apple crops is damaged and commercially lost (Jeannequin et al., 2015).The large quantity of apples remaining on the ground through damage or as windfalls at the end of summer/beginning of autumn can provide an important potential trophic resource for both vertebrates (Bouvier et al., 2020;Castañeda et al., 2022;Myczko et al., 2013) and a range of invertebrates (Morimoto & Pietras, 2020;Papadopoulos et al., 2001), including exotic species (Bal et al., 2017).The apple decomposition process can extend throughout the winter, providing both food for many organisms when other sources are scarce (Folwarczny et al., 2022;Myczko et al., 2013;Stockton et al., 2019) and a microclimatic environment protecting from environmental extremes (Jiménez-Padilla et al., 2020).Rotting apples contribute to maintaining the diversity of decomposer insects in the ecosystem, among which frugivorous Drosophilidae species (MacMillan et al., 2016;Morimoto & Pietras, 2020) are of particular interest because of (i) their significant role in the decomposition of decaying fruit and (ii) the recent establishment in Europe of two major invasive Drosophilidae that can utilise these decaying fruits.
Originating from Asia, the invasive D. suzukii fly spread to Europe in the late 2000s (Asplen et al., 2015).Because of their serrated ovipositor, females can attack healthy, ripening, fruit making this species a major threat to fruit crops including berries and stone fruit (Tait et al., 2021;Walsh et al., 2011).However, interactions between D. suzukii and apples have been poorly documented, due to the low (if not imperceptible) level of damage it causes, leading to very little commercial impact.However, decaying apples are likely to be attractive to adult flies, as D. suzukii is attracted to apple juice (Feng et al., 2018), products of apple degradation (Feng et al., 2018) and apple cider vinegar (Lasa et al., 2020).Few studies have reported infestations of apples by D. suzukii in orchards (Asplen et al., 2015;Bal et al., 2017;Briem et al., 2015;Walsh et al., 2011;Weydert & Mandrin, 2013) and these infestations probably primarily occurred on already damaged fruit with easily accessible flesh for oviposition (Cai et al., 2019;Lee et al., 2011).
The Nearctic invasive species C. amoena is known to oviposit on already damaged or parasitised fleshy fruit and nuts (Band et al., 2005), but its contribution to fruit decomposition has not been studied.It is not considered a true pest (Band & Band, 1980;Pajač Živkovi c et al., 2017) at present and has received much less research attention than D. suzukii.Native to North America, it has become widely established in Europe since its first detection in 1975 in Czechia (Band et al., 2005;de Jong & van Zuijlen, 2003;Pajač Živkovi c et al., 2017).
The species' spread across Europe is likely to have been facilitated by association with regional transport of fruits, probably apples, across the region (Band et al., 2006;Burla & Bächli, 1992;Withers & Allemand, 1998).Questions about its potential impact on the native Drosophilidae community have been raised previously (Withers & Allemand, 1998), but no further investigations have been made.The trophic resources used by C. amoena are varied, spanning multiple wild and domestic fruits including apples (Band, 1991;Band et al., 1998Band et al., , 2005Band et al., , 2006;;Burla & Bächli, 1992).Oviposition mostly takes place in fruit scars, codling moth (Cydia pomonella Linnaeus) tunnels or frass (Band, 1988;Band et al., 2005) and is usually aggregated (Band, 1989).
C. amoena takes advantage of previous physical damage to fruit, including from previous pest infestations (Band et al., 1998(Band et al., , 2005)).Both C. amoena and D. suzukii have been reported co-occurring in apple orchards in Italy (Amiresmaeili et al., 2019).
This study aimed to disentangle the relative roles of landscape and local factors in controlling the diversity and composition of Drosophilidae communities using decaying apples as breeding sites, including these two invasive species.We hypothesised that (i) the highly polyphagous D. suzukii would preferentially utilise apples in the early stages of decomposition at the end of summer, as it is a pioneer species preferentially infesting ripening fruit (Poyet et al., 2014), and (ii) C. amoena would use this fruit resource in autumn, as this fly is commonly found infesting apples that have been first attacked by codling moth C. pomonella (Band et al., 2005).We also hypothesised that (iii) both native and exotic species would be positively associated with the proximity of natural habitats (particularly woodland), which generally provide a refuge (and thus reservoir) for Drosophila species (Basden, 1954), (iv) they would be affected by low temperature and particularly frost days and (v) they would be influenced by apple tree traits, such as the presence of large canopy trees providing shelter or high apple production for feeding and oviposition.It is worth mentioning that we focused on apples for the abovementioned reasons but other fruits could play this backup resource role for Drosophilidae (e.g., pears, grapes, figs and persimmons).

Study area
The study was conducted between September 2021 and April 2022 in the region of Amiens (49 53 0 40 00 N, 2 18 0 07 00 E) in northern France.The region's climate is oceanic with a mean annual temperature of 10.7 C and average rainfall of 691.9 mm (data from meteorological station Dury-les-Amiens, StatIC network).The landscape of this region is largely characterised by agricultural production and is a mosaic of open fields cultivated for cereals, rapeseed and sugar beet, interspersed by apple orchards, grasslands and woodland patches.The apple orchards are managed using various strategies including conventional and organic fruit production.However, apple trees are also commonly present in domestic gardens, green spaces in towns and villages, along roadsides and in pastures and woodlands.In the study area, only one study assessed the Drosophilidae diversity, recording 25 species (Ulmer, 2022).

Apple collection and fruit and tree trait measurements
Decaying apples were obtained from 19 sites randomly selected within a landscape window of 30 Â 45 km (see Table S1 for site characteristics and apple varieties).To capture the temporal dynamics of communities (apples mainly fall in September and October, and some apples still remain on the ground in early Spring), eight sampling sessions took place between September 2021 and April 2022 (21-22 September 2021, 19-20 October 2021, 16-17 November 2021, 14-15 December 2021, 18-19 January 2022, 16 February 2022, 15 March 2022and 12 April 2022).Each month, at each site, six decaying apples were randomly collected from the ground below the canopy of a single apple tree.After the last session in April, no decaying apples remained on the ground at any of the sites.A total of 633 apple were collected over the study period.
To evaluate the fruit resources available for the Drosophilidae, the following quantitative and qualitative traits were measured for each apple collected: height (from the flower scar to the fruit peduncle), diameter, mass, volume (mean radius 3 Â π Â 4/3), the proportion of healthy skin of the fruit (based on colouration), the proportion of rotting surface (based on visual aspect), proportion of fruit surface with no skin (removed by physical wounds or animal foraging), presence of mould (i.e., Penicillium and Monilinia fungal sporophores following the identification keys of FREDON, 2002), and the number of holes associated with galleries drilled by pests (including those from the codling moth C. pomonella).Since fruit sugar levels are known to be associated with egg laying choices and/or larval development of some Drosophila species (Lee et al., 2008), Brix values (% sugar content) were measured both in the healthy and rotting parts of each collected fruit using a refractometer (Bellingham+Stanley sugar 0%-50%, serial number 019568 for low volume).The pH was also measured in both parts of the apples using pH indicator paper (Special indicator paper, REF 902 05, pH 0.5-5.5 (MACHEREY-NAGEL) and pH indicator paper, pH 5.5-9.0 (Supelco).In all cases, pH and sugar contents were used as proxies for the general decaying stage of the apple, including both internal and external decays.
The following vegetative and reproductive traits were recorded for each apple tree under which fruit was collected in order to estimate their influence on community dynamics: tree height, minimum and maximum canopy diameter, canopy area (π Â max canopy radius Â min canopy radius) and trunk circumference (at 1.30 m height).The numbers of fruit hanging on the tree and the ground surface under the tree canopy in a 5 m radius around the trunk were recorded.The relative proportions of healthy vs. damaged fruit on the ground were measured.For each apple tree, the mean volume of an individual fruit (mean radius 3 Â π Â 4/3) and the total volume of fruit on the tree were calculated.The apple variety was also noted (Table 1).

Emergence and identification of Drosophilidae flies
Immediately after collection, apples were individually placed on wet cotton wool in cylindrical plastic transparent containers (diameter = 118 mm, height = 135 mm, volume = 1476 cm 3 ), covered with a nylon mesh, and maintained in a temperature-controlled room at 20 C under a 16:8 L:D regime until all insects had emerged.Adult insects emerging were collected daily and stored in 70% ethanol until identification.All collected fruits were maintained in the emergence room for 100 days to check for any late emergence of imagos.
Emerged insects were subsequently sorted to select taxa representing Drosophilidae, which were identified to species level using a Leica M205C stereomicroscope (equipped with a Leica MC170 HD camera and Leica Application Suite software) following the keys provided by Bächli et al. (2004) and a collection of standard specimens available in the laboratory.

Environmental variables
Local, landscape and meteorological variables were measured at each sampling site or obtained from online databases to examine their relationships with the Drosophilidae community emerging from the apples.The full list of variables, their units and their codes are given in Table S1.
For each sampled tree, the altitude and the slope of the site were recorded.Within a 5 m-radius plot centred on the tree trunk, the cover and height of the tree layer, shrub layer and herbaceous layer were estimated, as well as soil litter thickness.Leaf litter, dead wood debris and the presence of a composting agent, such as apple pomace waste from cider factories, were also noted.The proportions of local habitats surrounding the apple tree (wood, hedgerow, orchard, grassland, crop, garden, building, road, river and pond) within 10 and 50 m radii were also noted.The management of the site herbaceous vegetation (mown vs. grazed grasslands) and apple trees (yearly pruned vs. not or rarely pruned, pesticide treated vs. non-treated) were noted.
T A B L E 1 Site, apple tree and apple characteristics.Note: Site: site code, Coordinate: geographic coordinates of the apple tree (coordinate system: WGS 84), Altitude: mean altitude of the apple tree, Variety: variety of the apple tree, sum collected apples: total of apples collected, mean sugar content: mean sugar content in healthy part of apples sampled ± standard error, mean pH: mean pH in healthy part of apples sampled ± standard error.
The landscape composition around each sampled apple tree was characterised.A geographic database was created using a Geographic Information System (GIS; ArcGIS Pro v.2.5, ESRI).The sampled apple trees were positioned in the GIS and buffers of 10, 20, 50, 100, 250, 500, 750 and 1000 m radii around each tree were created for subsequent analyses of landscape composition.Landscape elements (woodland, grassland, crop, building, road, water) were extracted from the OSO database (Centre d'Expertise Scientifique OSO, 2021) and updated using aerial photographs and field observations (in buffers <100 m).The BD-Carto ® database from French National Geography Institute was used to refine mapping and georeferencing.For each sampled tree, the cumulative area of the different habitats composing the landscape was then computed in each concentric buffer.
Macroclimatic conditions were characterised for each sampling site using regional measurements.The daily meteorological data were retrieved from the three nearest meteorological stations to each site (https://www.historique-meteo.net/france/),representing large-scale climatic conditions.Daily minimum and maximum temperatures, rainfall and snowfall were calculated for all sites using inverse-distance weighting interpolation (Willmott et al., 1985) from the data from the three nearest weather stations (detailed in Tables S3 and S4).Daily temperatures were extracted for each day from January 1 2021 to the date of sampling, allowing us to calculate the mean daily temperature range (daily maximum-daily minimum) as well as the number of frost days since the beginning of the year.Accumulated degree-days (growth degree-days) were calculated using a lower threshold of 0 C between 1 January 2021 and the date of sampling (Baskerville & Emin, 1969).The same calculation was also performed from 1 September 2021 to the date of sampling to better match the ripening and collection period of the apples.The threshold value of 0 C is a standard threshold commonly used in insect and plant studies (McNeil et al., 2020;White et al., 2012), although it may not represent the precise threshold for all Drosophilidae.This threshold is particularly appropriate for the study of temporal synchrony between flies and plant resources (Iler et al., 2013).It was also selected as several Drosophila species are known to be active at very low temperatures (<5 C) during winter days (January-February in Amiens and Bordeaux, Ulmer et al., 2022).Cumulative precipitation between 1 January 2021 and the date of sampling was also calculated from daily precipitation values.
Microclimatic temperatures were recorded at each sampling site, noting the temperature of the ground surface on which the apples were laying and the air temperature within the apple tree canopies.
Hobo loggers (TIDBIT data logger V2 TEMP TBI-001, ONSET Company, Bourne MA, USA) were used, recording every 60 min.Each sampling site was equipped with two loggers (soil surface and within tree canopy).The first was placed on the ground surface among decomposing apples ('bottom hobo').The second was suspended 1.5 m above the ground in the tree canopy, under the shade of a branch to avoid direct exposure to solar radiation ('top hobo') and oriented northward.The minimum, mean and maximum temperatures for ground and air were extracted every day to compute the mean daily minimum, mean and maximum temperatures, and used to assess the buffering ability of the soil relative to the local air temperature.Accumulated degree-hours (growth degree hours) were calculated using a lower threshold of 0 C (Baskerville & Emin, 1969) during the week preceding the day of each sampling.

Statistical analyses
General Linear Mixed Models (GLMMs) were used to test the influence of environmental factors (apple traits, local conditions, landscape composition and meteorological variables; see list in Table S1) on the temporal dynamics of species richness and abundance of the Drosophilidae community emerging from the decomposing apples (n = 633 fruits).Part of the analysis specifically focused on the number of D. suzukii and C. amoena individuals emerging per fruit.'Site' was included as a random effect in the GLMMs to account for the non-independence of apples sampled from the same site.Response variables were checked for normality and transformed (log 10 + 1) when necessary before analyses (Quinn & Keough, 2002).In the models, only the environmental variables that were significantly correlated with the response variables, and with R > 0.2 (Pearson correlation), were considered to avoid overfitting and excessive complexity.
After backward selection of explanatory variables, the Akaike information criterion (AIC) was used to select the most parsimonious model, that is, the most significant model with the lowest AIC (Harrison et al., 2018).Model residuals were checked for homoscedasticity using biplots of model predictions.Kruskal-Wallis tests were used to test whether the numbers of species and individuals of Drosophilidae emerging from apples varied between months.Post hoc pairwise comparisons between months were performed using Mann-Whitney U tests (α = 0.05).
The influence of environmental conditions on Drosophilidae community composition was examined using a redundancy analysis (RDA; Jongman et al., 1995;McCune & Grace, 2002) of the Drosophilidae species matrix constrained by environmental variables.Before performing the RDA, a detrended correspondence analysis (DCA) (Hill & Gauch, 1980) was run on the species matrix to assess (based on the gradient lengths depicted by DCA axes) whether a linear or unimodal model was appropriate in the subsequent multivariate analyses.DCA results on the species matrix showed short gradients (<3 S.D.), supporting the use of an ordination technique based on a linear model, such as RDA (Jongman et al., 1995).The Drosophilidae species matrix was built using the mean abundance of Drosophilidae species calculated from the six apples collected per site for each month (91 samples Â 7 Drosophilidae species).The Drosophilidae species matrix was standardised with a general relativisation by sites (McCune & Grace, 2002) and log 10 + 1-transformed before analyses to decrease the influence of extreme variability in the numbers of individuals (Baar & ter Braak, 1996).Drosophilidae species present in <0.8% of the sampled apples (i.e., in <6 apples among the total of 633 apples collected) were removed from the matrix before analysis to avoid outlier effects.To avoid multicollinearity among the numerous environmental variables measured at the different spatial scales of the study (see list in Table S1), only the variables that correlated with DCA axes and were ecologically non-redundant or statistically not inter-correlated were retained, generating a final subset of 91 samples Â 18 environmental variables.

Variations in quantity and quality of the apple resource
Across all the sampling sites, apple resources declined progressively over the study period (Figure 1).Initially, 7969 apples were present in September 2021, of which 7040 were hanging on the trees and 929 were laying on the ground.The number of apples progressively declined until April, when no apple remained on the trees and only 65 on the ground.
The quality of the apples also changed across the study period, with a gradual decrease in sugar content and increase in pH, especially in the rotten parts of the fruit (Figure 2).Overall mean sugar content decreased from 12.28% ± 0.28% to 4.5% ± 1.03% in the rotten parts of the fruit between September and April, with the lowest concentration measured in February (2.84% ± 0.5%).In the healthy parts of the fruit, sugar content decreased from 16.41% ± 0.85% to 10.9% ± 2.12% between September and March (apples collected in April were entirely rotten).The mean pH of the rotten parts increased from 2.86 ± 0.07 to 3.36 ± 0.09 between September and April, with the maximum value observed in February (4.66 ± 0.1).In the healthy parts, pH remained relatively stable from September to February (between 3.35 and 3.60), but then increased to reach 4.55 ± 0.51 in March.
Microclimatic differences between soil and air over the study period Ground surface and canopy air temperature measurements confirmed that the soil acted as a thermal buffer, especially during the coldest months.From February to April, differences in mean monthly temperature between air and ground were observed (Figure S1A).Notably, even in months where the mean temperatures were similar, temperature variability was higher in the air than on the ground (Figure S1B), reaching both higher and lower temperatures daily than the soil with differences of up to $10 C. At the site scale, temperature variation was even more apparent when comparing the weekly variations between autumn, winter and spring (Figures S1C-E, respectively).In winter and spring, negative canopy air temperatures were experienced, whereas the ground surface was protected from frost by the soil's buffering effect.

Drosophilidae species abundance and richness
From the 633 apples collected on the ground, a total of 6283 individual flies representing 14 Drosophilidae species emerged (Table 2, detailed by month in Table S2).The Drosophilidae community was Drosophilidae species richness per apple was 0.86 ± 1.24 species (n = 633 apples) but differed significantly between months (X 2 = 92.08,p < 0.001; Figure 4a) with a maximum of 2.02 ± 0.17 in

Influence of environment on Drosophilidae community composition
Individual Drosophilidae species presence could be differentiated according to their ecological niche in terms of substrate acidity and sugar content (Figure 5).Two groups of species were apparent: one,

DISCUSSION
Our study showed that four main categories of environmental drivers systematically influenced the diversity and composition of the Drosophilidae community present in fallen apples (i) local and seasonal thermal variation, (ii) habitat composition and (iii) fruit quality and quantity.

Climatic variation and the seasonal dynamics of the Drosophilidae community
Local-scale variations in microclimate temperature appeared to influence Drosophilidae diversity and species abundance more than regional-scale variations derived from meteorological stations.Local and hourly microclimatic data are likely to provide a better indicator of the impact of temperature extremes on the diversity and abundance of Drosophilidae species, especially that of negative temperatures (freezing events).In temperate regions, temperature is known to be a major driver of insect population dynamics and geographic range (Sinclair et al., 2003) because most insects are chill-susceptible and die due to injuries caused by ice formation (Lee, 1991;Stephens et al., 2015).Soil appeared to buffer the ground surface temperatures experienced across seasons, especially during the coldest months.In autumn, canopy shading may explain the smaller difference between air and ground surface temperature.During winter and early spring, when the trees have lost their leaves, the soil becomes the last compartment of the ecosystem capable of buffering the colder open-air temperatures and protecting on or near the ground surface from freezing temperatures.
Finally, D. tristis was found only in apples collected during the early T A B L E 3 Effect of environmental variables on the species abundance and richness of the full Drosophilidae community, Chymomyza amoena abundance and Drosophila suzukii abundance, analysed by GLMMs.Note: BROWNROTTEN2: % of rotten part on apple surface, MASSAV2: apple mass, BUILD50: % of buildings in a radius of 50 m, MINTH7Jhb: soil minimum temperature in the preceding 7 days, VOL2: apple volume, MAXTH7Jhh: air maximum temperature in the preceding 7 days, MOW: mowing of herbaceous vegetation, CARPOPAPSE2: presence of codling moth borehole, CANOPY: radius of the apple tree canopy, MINTH7Jhh: air minimum temperature in the preceding 7 days, and PENICILI2: presence of Penicillium sporophore on apple.
months of autumn and in March, probably indicating freeze intolerance, although studies of this species' cold tolerance are lacking (Basden, 1954).
The cold-tolerance strategies of different species are likely to structure seasonal community composition, whereas partitioning may also take place when two species sharing the same fruit have different fitness values (e.g., developmental time, number of eggs laid and percentage of adult emergence) when developing at the same temperature (Grimaldi, 1985).For example, D. melanogaster lays more eggs than D. simulans at 15 C (McKenzie, 1978), and their development times vary according to the season, with D. melanogaster developing more rapidly in September than D. simulans (Behrman et al., 2015).
Laying more eggs and developing more rapidly would favour D. melanogaster over D. simulans, as it results in increased larval competitive ability (Grimaldi, 1985;Sevenster & Van Alphen, 1993).This could explain why, in our study, the abundance of D. melanogaster was greater than that of D. simulans in September but the reverse was apparent in October after the temperature dropped.Exploiting the resource first and having a short development time is another strategy that would benefit a species (Nunney, 1990).This is the strategy adopted by D. suzukii, which can utilise the fruit as it ripens before any decomposition takes place (Clemente et al., 2018;Walsh et al., 2011).
Over the thermally governed seasonal gradient extending from September to April, the decaying apples were exploited by a succession of Drosophilidae species, which supports the hypothesis of temporal partitioning of the use of this food resource.This strategy adopted by organisms using a common resource has been reported in studies of Drosophila species (Hodge et al., 1996;Matavelli et al., 2015) and in insects in general (Vindstad et al., 2020;Wettlaufer et al., 2021).The sequential use of resources allows coexistence through minimising competitive interactions, potentially avoiding exclusion.This temporal separation of the use of decaying apples is not only determined by temperature changes differentially affecting Drosophilidae species over time across seasons but is also associated with the gradual modification of fruit quality, which varies with its stage of decay, as shown in other guilds of insects (Koskinen et al., 2022).The hierarchical continuum model, primarily used in plant communities (Collins et al., 1993;Hanski, 1982), may be applicable to Drosophilidae species coexistence in decaying apples.Based on this concept, four strategies can be identified and applied to species, termed 'core', 'satellite', 'urban' and 'rural' species, based on the responses of their abundance (i.e., dominance) and distribution (i.e., frequency changes between months) to the seasonal progression (Figure 7).Only one species, D. subobscura, was typically dominant and was present across the full seasonal sampling period (see Figure 3 and Table S2) and, thus, can be considered a 'core' species (sensu Collins et al., 1993).This species plays a major role in apple decomposition and its role in structuring the community by transforming the substrate is fundamental.'Satellite' species are found at low density and restricted to a smaller element of the seasonal progression.This was the case for late-summer/early autumn species (D. suzukii, D. melanogaster, D. simulans, D. tristis), which are probably opportunistic species ovipositing and developing in fallen apples, using the freshest and sweetest wounded fruits (Cai et al., 2019;Lee et al., 2011).
D. immigrans conformed to the description of an 'urban' species in the hierarchical continuum model, with a very high abundance peak (particularly in September and October) restricted to the beginning of the seasonal progression.This strategy indicates that D. immigrans is a strong performer in early autumn but is also very dependent on the substrate quality (high sugar levels; Figure 5).The invasive C. amoena was the only 'rural' species in the current, present throughout most of the seasonal progression from September to March, but at low to moderate densities.Even though not dominant, its strategy of persisting through the seasons could contribute to its invasion success.Its persistence in decaying apples could indicate a high tolerance to acidic substrates but a preference for basic substrates (Figure 5).

Landscape and local factors driving community diversity and invasive populations
Insect communities are affected by both local conditions and landscape composition (Mitchell et al., 2014).At the landscape scale, variables associated with direct human activities and semi-natural habitats were significant drivers of the Drosophilidae community in our study.Abundance and species richness increased with the cover of buildings in a 50 m radius (even though this variable was not retained in the final GLMM of species richness).Anthropogenic stress factors, such as urbanisation, reduce the overall abundance and richness of insect communities (Vaz et al., 2023) and may lead to ecological homogenisation (Eggleton, 2020;Knop, 2016;McKinney, 2006) with specialist species being negatively impacted (Knop, 2016;Shuisong et al., 2013).However, some species benefit from urbanisation, including non-native or invasive species (Bertelsmeier, 2021;McKinney, 2006), generalist species (Shuisong et al., 2013) or human commensal species such as mosquitoes (Perrin et al., 2022).Some members of Drosophilidae could also benefit from urbanisation.
D. immigrans, D. melanogaster and D. suzukii were strongly correlated with the cover of buildings.The former two species are known to be associated with human activities (David, 1979;Kojima & Kimura, 2003), especially D. melanogaster, which is considered the most commensal Drosophila species associated with humans (Lachaise & Silvain, 2004).In winter, all three of these species may find overwintering refuges around human structures (Ulmer, 2022).
Other species in this study were affected by the presence of seminatural habitats.D. subobscura was found in apples collected under trees close to roads.Road edges impact community structure, yet can have positive effects on biodiversity (Rotholz & Mandelik, 2013).
C. amoena abundance was positively correlated to the presence of orchards with woodland within a 50 m radius, consistent with previous studies (Band et al., 2005).competition in larger fruits.In association with a higher detectability, grass mowing was associated with increased Drosophilidae abundance.Mowing could also negatively impact the presence of some predators (Dobbs & Potter, 2014;Horton et al., 2003), contributing further to increased Drosophilidae abundance.The quality of the resource, that is, the percentage of rotten surface, the presence of codling moth or Penicillium and the presence of damage of biotic origin, also influenced the Drosophilidae community.Apple skin is waxy and robust and most Drosophilidae ovipositors may not be able to penetrate it (Atallah et al., 2014).Damage of biotic origin provides a point of entry for oviposition.This is especially the case for C. amoena, which uses codling moth tunnels to oviposit in walnuts and apples (Band, 1988).Similarly, D. melanogaster needs naturally damaged grapes to oviposit but can also benefit from D. suzukii's oviposition sites (Rombaut et al., 2017).In contrast, the presence of Penicillium negatively affected D. suzukii abundance.Penicillium commune is found in D. suzukii artificial diet (Gao et al., 2017).It has been isolated from D. suzukii and had no effect on host survival (Bing et al., 2021).The odour profile of Penicillium spp.contains geosmin and octenol, molecules associated with other fruit-associated plant pathogens, which could act as an indicator of unsuitable substrate for oviposition for D. suzukii (Wallingford et al., 2016).Large trees were also negatively associated with D. suzukii presence, possibly as their canopy shading is unsuitable for this species, which may be more attracted by thermophilic and sites exposed to full sun for oviposition as autumn approaches.

Trophic niche partitioning and the role of Drosophilidae in fruit decomposition
Trophic niche partitioning occurs when different species share the same nutritional resource but use it differently, limiting competition (Pocheville, 2015;Roughgarden, 1976).In our study, up to eight species were able to coexist in one fallen apple.This suggests temporal niche partitioning.Temporal partitioning in relation to seasonal temperature variation was addressed above.Another driver of temporal partitioning may relate to the fruit decay process and physical and chemical changes over shorter timescales.When fruit is damaged, either by the fall from the tree or by insect oviposition, microbial communities, mostly bacteria and yeasts, immediately start to proliferate (Rombaut et al., 2017), inducing changes in biochemical composition (Awmack & Leather, 2002;Matavelli et al., 2015;Nunney, 1990).In particular, the protein:carbohydrate (P:C) ratio increases, allowing flies with different dietary requirements for protein to use the resource sequentially (Matavelli et al., 2015).For instance, the successive stages of decay in figs are exploited sequentially by different Drosophilidae species as the sugar and protein contents change (Lachaise et al., 1982;Matavelli et al., 2015).In early autumn, apples are sweeter and more acidic than in winter and spring.Drosophilidae species appear to be sensitive to sugar concentration and pH variation.
Sugar and pH are known to impact fecundity, survival and microbiota composition (Deshpande et al., 2015;Fellous & Xuéreb, 2017).For example, D. suzukii prefers low sugar content substrate in which to oviposit and shows lower survival and fecundity at higher sugar concentrations (Fellous & Xuéreb, 2017).In adult D. melanogaster, acidic food increases palatability, food intake and survival (Deshpande et al., 2015).As decay continues, the fruit texture changes, and firmness of the substrate can be of importance for oviposition (Kienzle & Rohlfs, 2021;Sato et al., 2021).For example, D. suzukii oviposits in firm substrates (Kienzle & Rohlfs, 2021;Sato et al., 2021), whereas D. melanogaster selects softer substrates (Sato et al., 2021).Decaying apples can be considered ephemeral resource patches, as their chemical and structural compositions vary in space and time, and therefore, the diversity of niches they provide can be large, thereby enriching local community composition and interspecific interactions at the individual scale (Butterworth et al., 2023).A study on the Drosophilidae succession across the decaying stages of the apple would nicely complement our results in the future.We discuss in the Supporting Information how spatial partitioning may also have occurred between and within apples.

Conclusions and perspectives
Ecological processes during the post-harvest period in agricultural sys- As it is already recommended to reduce other apple fruit pathogens and parasites (brown rot fungi scab), actively removing these 'waste' fruits could, therefore, reduce the pest's abundance in the subsequent spring.However, the native Drosophilidae community, which plays a key role in the recycling process of organic matter, would clearly also be impacted if this management strategy was applied.Similarly, while C. amoena is can be considered as a pest species, by being present throughout the winter, it also contributes to reducing available resources in spring for D. suzukii (see Wilson et al., 2012).Finally, although this study focused on Drosophilidae, other species will certainly benefit from these fallen fruits, including parasitoids and other arthropods.Future studies should focus on the dynamics of these multitrophic relationships across winter time in relation with these ephemeral resources.
dominated by two species, Drosophila immigrans (41.03% of individuals) and Drosophila subobscura (39.77%).Five other species contributed >1% of total individuals: C. amoena (7.93%), Drosophila melanogaster (5.06%), D. simulans (2.72%), D. suzukii (1.39%) and D. tristis (1.39%).The remaining species (<1% of total) were Scaptomyza palida, Drosophila ambigua, D. bifaciata, D. cameraria, Drosophila obscura, D. kunzei, and Hirtodrosophila cameraria.Excepting D. suzukii and C. amoena, all species were native.The abundance patterns of Drosophilidae species were compared between months over the study period (Figure 3).The total number of individuals recorded per month rapidly and consistently declined, from 3635 emerging from apples collected in September to 84 from apples collected in April.D. subobscura was the only species that emerged from apples collected throughout the sampling campaign.C. amoena emerged from apples collected in all months except April.D. immigrans emerged from apples collected from September to December and D. suzukii only emerged from apples collected between September and November.
September and a minimum of 0.38 ± 0.09 in March.Up to eight different species were present in one apple in September.Drosophilidae abundance also decreased markedly over the study period and also showed high variability, averaging 9.93 ± 36.66 emerging individuals per apple overall, with a maximum in September of 33.66 ± 76.81 (Figure4b) and a minimum in March of 2.43 ± 9.28.Drosophilidae species richness and abundance were significantly influenced by numerous variables (GLMMs, Table3), increasing with apple size and the percentage of rotten surface (MASSAV2, BROWNROTTEN2) in F I G U R E 1 Variation in apple abundance between September 2021 and April 2022, in relation to their position (in the trees vs. on the ground) and the status of the latter (healthy vs. rotten).siteswhere the minimum ground temperature was mild (i.e., highest local daily minimum temperature on the ground during the week before sampling, MINTH7hb).The total abundance of Drosophilidae also increased with variables related to direct human influence, such as the area represented by buildings within a 50 m radius around the apple tree (BUILD50).The abundance of C. amoena was positively correlated with apple size and percentage of rotten surface, as well as the presence of codling moth holes (CARPOCAPSE2).The abundance of D. suzukii decreased with the presence of Penicillium (PENICILI2) and with increasing apple tree canopy size (CANOPY).Both C. amonena and D. suzukii abundances increased with higher local air temperature during the week before sampling (MINTH7hh and MAXTH7hh) and with mowing of the local vegetation (MOW).
including D. suzukii, D. melanogaster, D. simulans and D. immigrans, was F I G U R E 2 Variation of mean (±SE) sugar content (a) and apple pH (b) between September 2021 and April 2022, with healthy (plain line) and rotten parts (dotted line) of the apples presented separately.T A B L E 2 Distribution of the Drosophilidae species in the study.
associated with the sweetest and more acidic apples that were freshly fallen in early autumn; the second, comprising C. amoena, D. subobscura and D. tristis, was mostly present in rotten apples that had lower sugar content and increased pH.In the RDA coupling environmental factors and Drosophilidae species matrices (Figure6), the eigenvalues of the first two ordination axes explained 54.8% of the data variance.In the ordination diagrams, the plots formed groups associated with the successive seasons (Figure6a).Along axis 1 (27.8%)groupings discriminated samples from autumn (empty circles, positive part), associated with the warmest temperatures (MAXMAX7Js; Figure6b), and winter and early spring samples (grey and black squares, respectively, negative part).The negative part of axis 1 was associated with D. subobscura, damaged apples (BIOORIGIN2), apple trees close to roads (ROAD10_%) and apple trees surrounded by grasslands and shrubs (GRASS50_%, SHRUB50_ %).Axis 2 (26.9%) discriminated apple samples with C. amoena, collected under tall apple trees (HEIGHT), growing in orchards (ORC10_ %), and surrounded by woodland (WOOD50_%).The positive part of axis 2 was associated with autumnal species (D. immigrans, D. suzukii, D. melanogaster) present in the warmest periods (MAXMAX7Js), apples with the highest sugar content (SUGAR2) and mass (MASSAV2), degree of rot (BROWNROTTEN2), and collected from a human-influenced environment (BUILD50_%).In summary, the RDA showed that local and seasonal thermal variation, habitat composition and fruit quality and quantity were the major drivers of Drosophilidae species assemblages.F I G U R E 3 Distribution of the most abundant Drosophilidae species emerging from apples collected during the study, from September 2021 to April 2022.DROTRI: Drosophila tristis, DROSUZ: Drosophila suzukii, DROSIM: Drosophila simulans, DROMEL: Drosophila melanogaster, CHYAMO: Chymomyza amoena, DROSUB: Drosophila subobscura, DROIMM: Drosophila immigrans, OTHERS: species representing less than 1% of the emerged flies.F I G U R E 4 Mean Drosophilidae species richness (a) and abundance (b) per apple (±SE) in the 19 sites sampled from September 2021 to April 2022.
At the local scale, Drosophilidae richness and abundance were mostly influenced by the quantity and quality of the food resource, the management of the site's herbaceous vegetation and the tree canopy.The larger the apples, the greater the richness and abundance present.This could be explained by higher detectability of larger fruit, either visually or by olfaction, a selection process similar to that of other frugivorous insects(Sallabanks, 1993), or by a decreased larval F I G U R E 7 Distribution of the most abundant Drosophilidae species collected during the study, from September 2021 to April 2022.Adaptation from the hierarchical continuum concept described by Collins et al. (1993).DROTRI: Drosophila tristis, DROSUZ: Drosophila suzukii, DROSIM: Drosophila simulans, DROMEL: Drosophila melanogaster, CHYAMO: Chymomyza amoena, DROSUB: Drosophila subobscura, DROIMM: Drosophila immigrans.
tems are of particular interest but have received limited research attention.Unharvested fruits can remain on the ground for long periods of time, providing potential breeding sites, food resources and/or microclimatic refuges across seasons.In our study, decomposing apples hosted up to eight species of Drosophilidae, including two important invasive species, D. suzukii and C. amoena.The fly community structure was strongly influenced by local and seasonal thermal variation, habitat composition, and fruit quality and quantity.Decomposing fruits are a dynamic substrate allowing resource niche partitioning, both temporal and spatial, with competition and facilitation processes occurring.Our study raises questions relating to postharvest pest management strategies.Known serious pest species such as D. suzukii can use these post-harvest resources to increase overwinter survival and accelerate subsequent proliferation after the cold season.
Note: N ind: total number of individuals emerged from apples, Min: minimum number of individuals per apple, Max: maximum number of individuals per apple, Mean: mean number of individuals per apple, SE: standard error, %: percentage of individuals of the species among all individuals that emerged from apples, AFreq: frequency of apples infested by the species among all apples sampled.