Population Dynamics of the Olive Fly, Bactrocera oleae (Diptera: Tephritidae), Are Influenced by Different Climates, Seasons, and Pest Management

Abstract

Pest management practices interact with many species and have an impact on the ecology and the economy of the area. In this paper, we examine the population dynamics of the olive fly, Bactrocera oleae (Rossi), Diptera: Tephritidae, on Samos Island, Greece, observing the spatial and temporal changes of the pest along an altitude associated with area-wide pest management. More specifically, we analyze data from an extended McPhail trap network and focus on experimental sites, where we monitor the pest population in relation to sprays, temperature, and relative humidity inside the tree canopy during the season for a three-year period. Our findings indicate that fly populations are influenced mostly by climate and altitude over longer periods in the season and from bait sprays for shorter periods of time, which appeared to be less effective in autumn, probably due to population movements and overlapping generations. Apart from the factors that were taken into account, such as the weather conditions and pest management regimes that were proven important, more factors will have to be considered for infestation level, such as fruit availability, inhibition factors (natural enemies, symbiotic agents, food supplies), and cultivation practices. Site microclimate conditions and the landscape can be used to explain changes at the plot level.

1. Introduction

Bactrocera oleae (Rossi), Diptera: Tephritidae, known as the olive fruit fly (OFF), is perhaps one of the most studied pests due to its economic impact [1]. There are, however, still considerable knowledge gaps regarding its biology, like the diapause season and reproductive quiescence [1,2]. It is widely known that, OFFs reportedly overwinter as pupae on the ground or inside infested fruits, but in mild weather conditions, adults can overwinter in a facultative reproductive dormancy and remain active all year round in the canopy [3,4,5,6]. Adults can feed on many organic sources, like plant nectar, pollen, and insect honeydew [7]. Each fertilized female lays about 12 eggs a day and about 200–250 eggs in a lifetime [8].

Larvae feed inside fruits, causing premature fruit drop [3,9], and/or, if fruits are harvested, lower the quality of the pressed olive oil due to increased acidity [1,8,10]. Populations and the number of generations per year depend on many different factors, including microclimate (temperature and humidity), fruit availability, and quality [11,12,13,14]. Spring reproductive dormancy drives the synchronization of the first (‘base generation’) and the following generations [2]. Laboratory studies at constant temperatures show that temperatures of 35 °C and above are lethal to pupae [15], and older studies determined that the lower temperature threshold for larval development ranges from 10 to 12.5 °C and the upper temperature from 30 to 32 °C (summarized by Tsitsipis 1980 and Fletcher 1987) [16,17]. In the field, larval development occurs at 12–35 °C [18]. Higher temperature is a major mortality factor for early stages of OFFs, especially for eggs and young larvae.

Also, pupal mortality in the soil during winter plays a pivotal role in population dynamics [19]. To minimize production loss of quantity and quality, OFFs are usually managed with chemical attractants and/or pheromone traps, bait sprays, and cover sprays with insecticides as well as sanitation management, such as the removal of fallen fruits. Long-term use of certain insecticides causes gradual loss of their effectiveness in OFF population control. Examples in Greece include resistance to pyrethroids [20], spinosad [21], and organophosphates [22,23]. Results after a 9-year survey from various regions of Greece, including our study area, suggest increasing resistance for all above substances [24], with a few exceptions. A few developments, including Geographical Information System (G.I.S.) applications and more environmentally friendly spraying substances, but also the extensive abandonment of plantations and the decrease in funds for the practiced control program, brought forward the need for integrated pest management that considers landscape and pest ecology.

Population dynamics of most arthropod species are influenced at the landscape scale [25] by environmental (e.g., relief, microclimate, interactions between organisms, such as parasitoids) and anthropogenic factors (e.g., rural, urban, and industrial activities) [26,27]. Olive plantations are cultivated along a landscape that exhibits different microclimates, especially when these landscapes are continuous [6,13,28,29]. Along these gradients, surrounding land cover can also profoundly affect fly population levels by providing food and/or habitat for parasites/predators [30]. A relationship between population density change and elevation over time has also been found [31], suggesting that adult flies move actively through the landscape. In northern Greece [13], hot spots of OFF populations were observed in the summer months at mid-to-high altitudes (up to 700 m) where conditions were cooler, while the corresponding lowland populations were very small and the opposite phenomenon in the autumn months, when temperatures were closer to the OFF optimum were measured in the lowlands. In a similar study in Israel, apart from environmental factors (mainly temperature and availability of fruit), endogenous factors, such as reproductive quiescence [1,5,27], were found to be important in annual changes of population dynamics. Even in nearby plots (less than one km apart, with same cultivars, tree ages, planting distances, etc.), the different ground morphology yielded different levels of population and damage [27].

In this paper, we investigate spatial and temporal OFF population dynamics on Samos Island, Greece. A spatial Krigging model approach [32] was used on the island to determine spatial and temporal similarities and differences in OFF populations for three seasons. Here, we build on that model and consider applications of crop protection management (bait spraying) rarely studied in the literature. We separate after each spraying event the area into sprayed and unsprayed and compare the effects of spraying for three 10 day periods. This is an exploratory approach, with the use of data for three seasons (2017–2019) for the whole island, which integrates local meteorological (from our own network of temperature and relative humidity data loggers), insect, and spatial data to provide insights on the landscape ecology of OFFs and recommendations for area-wide plant protection management programs.

Our overall goal is to examine how climate, altitude, and pest management influence OFF population dynamics in the landscape. More specifically, the research objectives are:

(1)

To monitor the changes in the OFF populations over space in the period June to October for three seasons, both within each season and among the seasons;

(2)

To analyze the changes in OFF populations in relation to previous populations for all years and other climatic variables using a linear mixed-effects model and considering spraying events.

2. Materials and Methods

2.1. The Case Study Area: Samos Island

The study area is Samos Island, which is located in the North Aegean Region between circa 37.6 and 37.8 north latitudes and 26.6 and 27.1 east longitudes and covers an area of 477.395 km². The island has a varying topography with two mountain ranges, one in the center north (“Ampelos” or “Karvounis” at 1153 m) and the dry, rugged Kerkis mountain (1434 m) on the west coast (Figure 1). The climate is predominantly Mediterranean, with mild, wet winters and warm, dry summers and prolonged sunshine. Mean annual temperatures range from 10.0 °C in February to 28.5 °C in July. The coldest month is usually February (6.5 °C to 13.2 °C) and the warmest is July (22.2 °C to 32.5 °C). Depending on the local topography, rainfall is within the range of 700 to 900 mm per annum, with somewhat higher averages in the northern part. Summers are usually dry and hot. Almost 60% of the total rainfall occurs in winter, whereas rain is almost absent during the driest months of July and August. The prevailing wind direction in Samos is northerly (almost 60% of the total winds throughout the year). The rugged and diverse terrain and the steep slopes generate microclimates.

Land use is roughly 25% agricultural land (53% of which are olives and 10% vineyards); 54% shrublands, maquis, and phrygana; and 21% forests (pine and oak). Olive groves cover almost 10,000 ha with ca. 1.3 million olive trees, 55.6% of which are in lowlands (0–200 m), 27.4% in semi-mountainous areas (200–400 m), and 17% in mountainous areas (>400 m), and olive crops are not irrigated. An unknown proportion of fields are neglected (few cultivation practices, except for periodically collecting olives; [33] reports the same for neglected plantations on Lesvos Island) or abandoned, especially in remote and less accessible areas as a direct result of the depopulation of the island in the last 10 years (−36% from 1951 to 1981, +4% since), which was more pronounced in rural and mountainous settlements, and the declining olive oil prices. The most extended cultivar is non-irrigated “Throumpolia”, followed by “Koroneiki”, “Manaki”, “Kalamon”, etc. Throumpolia is classified as an intermediate-sized fruit cultivar (weight 2.7–4.2 gr) for mixed use (olive oil/table olives).

2.2. The Olive Fly Management Program on Samos

Twelve plots were selected to represent the island’s olive groves according to altitude, percentage of olive cover in one km² zone around the trap, and ensuring that both northern and southern exposures were selected. The selected plots were two above 400 m, four in the 200–400 m zone, and six in the lowland area (<200 m). Two temperature and humidity loggers were set inside the tree canopy in each location [34,35].

Hourly minimum, mean, and maximum temperature and relative humidity measurements were taken from these loggers for the three seasons of the study (June to October). Daily maximum and minimum temperature and and relative humidity data were obtained from the Hellenic National Meteorological Service (HNMS, hereafter) reference station, which is stationed at the airport (latitude: 37.69, longitude: 26.91, altitude: 6.0 m a.s.l.). The long-term climatic series were utilized to define monthly means, identify extreme values, and present the average intra-monthly daily variations of maximum and minimum temperature and relative humidity. Long-term climatic averages for every day were calculated for the normal thirty-year period 1961–1990 for each month (June to October).

In Greece, as well as in our case study area, OFFs are controlled with administrative area pest management programs, recognizing the necessity of a area-wide approach. Insect populations are monitored and insecticide bait sprays are applied by tractors and pedestrian personnel, the frequency and intensity of which are determined by population levels, olive growth stage, and weather conditions. Within the framework of this program, a network of McPhail traps was installed. These traps are commonly used to monitor the OFF population, typically filled with 2% ammonium sulfate solution or protein food lures. On Samos, there is a dense network of around 400 (1 trap/2000 trees) McPhail traps (Figure 1), re-installed at the start of each season (end of May). Traps were checked, cleaned, and refilled every five days. Pest numbers in traps were regularly recorded in a database [36]. The first spray every year is applied at the pit hardening stage (usually starting in the second half of June) because it is when olives start to be susceptible to OFF attack, and then sprays are re-applied only in hot spots suggested by monitoring. The overall efficiency of these spraying programs has been questioned: (a) many fields, especially in remote and inaccessible areas, are abandoned and not sprayed, acting as sources for OFF populations [37]; (b) organic fields scattered throughout the plantations’ mosaic are not sprayed; and (c) as the available funds of the pest management program decrease, spraying has to be more selective in areas with more intense pest pressure. Nevertheless, and despite these shortcomings, the analysis can shed some light on the effectiveness and the impact of these events on OFF populations.

Transformation of the OFF data according to spaying events was performed on a 10-day basis, which was deemed more appropriate to estimate spraying event impacts for two reasons: the first is that the actual spraying event typically lasts more than three days. The five-day measurement may be wholly during the event itself. The second is that it agrees with the spraying directions for all substances used. Sprayed areas were calculated based on the routes of the tractors. Traps that fell within a 100 m radius from the line were considered “sprayed”, and the rest of the traps were “unsprayed”. Therefore, the 399 traps were divided into two subsets every ten days: spayed and unsprayed areas.

The data from the McPhail trap network and spraying data (date, area, substance) in the OFF-management program for three seasons (2017–2018–2019) were used to monitor OFF populations’ temporal and spatial dynamics. The trap network used to collect our data consisted of 399 McPhail traps (Figure 1) set on olive trees on June 1 and checked every five days until October 31. We used a mobile phone application [36] to automatically record and control the accuracy and reliability of measurements, which involved digitizing the trap location and the development of a geodatabase accessible via mobile internet where results were stored. The data for the insect populations were automatically stored in a geodatabase by the “trap-setters”. Agronomists performed occasional blind controls. Spraying data were provided by the rural development authority using an Android application to record the routes of the tractors used for spraying [36]. All experimental plot data (trap results and meteorological data) are derived from trees over one hundred years old and of the Throumpolia variety to exclude a potential selection bias of the OFFs [38,39].

2.3. Statistical Analysis

The measured variables are listed in

Table 1. List of variables used in the analysis.

Variable	Name	Definition	Data Availability
variable 1	Spatial location and altitude of the traps	Geographical coordinates of trap locations	once
variable 2	Hours with maximum temperature exceeding 32 °C	Hours per 10 days and month when maximum temperature exceeds 32 °C calculated by hourly measurements of temperature	10 days/monthly
variable 3	Hours with maximum temperature exceeding 35 °C	Hours per 10 days and month when maximum temperature exceeds 32 °C calculated by hourly measurements of temperature	10 days/monthly
variable 4	Hours with Relative Humidity below 60%	Hours per 10 days and month when RH is below 60% calculated by hourly measurements of temperature	10 days/monthly
variable 5	OFF population	Olive female fly populations as measured in each trap per 5 days, recalculated for 10 days	10 days/monthly, seasonal
variable 6	OFF population	Olive fly populations as measured in each trap per 5 days, recalculated for 10 days	10 days/monthly, seasonal
variable 7	% fertile infestations on the olive fruit/month	Percentage of fertile infestations of total infestations, measured in sampled olives per month for August, September, and October	monthly
variable 8	Spraying incident	If a spray is performed, the area is considered as sprayed for the particular 10 day period	random

.

Climatic analysis, where maximum and minimum temperature and relative humidity records from the data loggers were transformed into (1) the number of hours within the period used where the temperature exceeds 35 °C represents the absolute maximum temperature for the insect’s movement; (2) the number of hours within the period used where the temperature exceeds 32 °C that hinders its movement and activity; and (3) the number of hours within the period where the relative humidity is below 60% as unfavorable conditions.

We performed linear regressions every ten days with (i) the population in the next 10-day period (T1) and (ii) populations for the next three 10-day periods (T1, T2, T3) for the 399-trap network. These regressions are performed for each calculation’s spayed and unsprayed areas in T0. We also calculated % population differences for these periods (T1–T0%, T2–T1%, T3–T2%).

More details can be found in the Supplementary Materials. The efficacy of the treatments was verified by monthly olive fruit sampling and infestation control by the local Office for Rural Economy and Veterinary of Samos: Olives are collected randomly from a set number of trees for 22 local communities of the island, the number of trees is determined by the type of terrain (flat, intermediate, mountainous), and the total number of trees in each community is also determined (twenty trees are sampled and marked per 10,000 trees in total). Ten olives per tree are picked in the last 10 days of August, September, and October from different heights in each tree and are dissected to determine the ratio of infested to uninfested olives. The infested olives are further classified by the age of larvae and/or exit points on the fruit.

2.4. Analysis

We performed linear regressions for every 10 days with (i) the population in the next 10-day period (T1) and (ii) populations for the next three 10-day periods (T1, T2, T3) for the 399-trap network. These regressions are performed for the spayed and unsprayed areas in T0 of each calculation. We also calculated % population differences for these periods (T1–T0%, T2–T1%, T3–T2%).

OFF population fluctuations over longer time periods in the season per altitude zone were also performed with descriptive statistics and ANOVA tests per altitude zone for each month.

OFF numbers per trap numbers were log-transformed (log(x + 1)) to achieve a normal distribution and the data were checked for homoscedasticity. We then built a linear mixed-effects model for non-negative count data in R [40] using the function lme in library nlme to test the effects of various explanatory variables on fly numbers in the traps on the basis of maximum likelihood. All fixed effect explanatory variables (altitude (continuous), sprayed (yes/no), olive infestation levels (continuous), fly numbers after 10 days (continuous), fly numbers after 20 days (continuous), and fly numbers after 30 days (continuous) were added to the full model as main factors. In addition, the variables fly numbers after 10 days, fly numbers after 20 days, and fly numbers after 30 days were also added as two- and three-way interactions. We used the following random effect structure in the model: Region/Trap ID/Year/Sampling date to account for the temporal and spatial non-independence of the fly counts and hence to avoid pseudoreplication. All correlations (Pearson’s r) between the main factors were well below 0.7. Consequently, we used the function step (AIC) in the library MASS to simplify the model. During this process, the explanatory variables olive infestation, the three-way interaction fly numbers after 10 days–fly numbers after 20 days–fly numbers after 30 days, and the interaction fly numbers after 20 days–fly numbers after 30 days were eliminated from the full model. We then performed an ANOVA test of the simplified model using the function anova.

To estimate the importance and effect of spraying, we numbered all counts of each year in 10 day periods from one to thirteen to cover the 150 days of the season (130 days + 10 before the first count and 10 days after the last) and then divided the total number of traps into those that were in areas that were sprayed (and labeled “sprayed”) and the rest (labeled “unsprayed”).

3. Results

3.1. Measured Weather Parameters

The weather conditions on Samos Island according to the measurements of our sensors appear rather similar concerning T_mean (

Table 2. Mean temperature (T_mean), mean relative humidity% (RH%), and average hours with temperatures over 32 °C per month and year on Samos.

	Average Hours with Temp > 32 °C			Tmean °C			RHmean%
Month	2017	2018	2019	2017	2018	2019	2017	2018	2019
6	24.9	7.7	30.1	25.8	24.9	27.2	52.5	61.4	55.8
7	52.3	34.5	30.4	28.2	27.7	27.0	46.5	56.2	50
8	36.2	33.2	47.3	27.7	27.7	28.3	52.4	58.3	49.9
9	13.2	11.9	4.7	24.3	24.7	23.5	51.8	58.1	60.2
10	0.7	0	0.2	18.6	19.1	20.7	62.1	72.7	70.9

): 2019 was slightly hotter in most months of the season than the other two years except June, but there is no overall pattern detected, and most early differences were ironed out later in the season. Relative humidity% values indicate again that 2018 was cooler on average and with higher humidity in the start of the season than both 2017 and 2019, and this trend continued until late in the season. Lower average humidity was recorded in 2017 for almost all months of the season (Table 2). This picture of relative uniformity is not confirmed though when the average hours with temperatures over 32 °C are examined: 2018 is characterized by significantly fewer hours of higher temperatures than the 2017 and 2019 in June, a fact that could potentially lead to more favorable conditions for early development of OFF populations (RH% is also high, which also is expected to favor early growth), while in hotter 2019, lower OFF numbers are expected early in the season. July was significantly hotter in 2017 (20 more hours of high temperatures on average than 2019, Table 2), but then the trend reversed, and August 2019 was hotter than 2017 and 2018. September was again cooler in 2019, which seems to be a year of temperature extremes. These lower temperatures late in the season are expected to favor OFF populations significantly. What becomes evident is that average temperatures and relative humidity figures appear to be unsuitable to describe some trends that could be of importance for OFFs.

3.2. OFF Populations: Temporal and Spatial Differences

OFF populations measured in our trap network were not statistically significantly different in the three reference years. (Table S1). The season in 2019 starts with significantly lower OFF adult counts in the traps, compared to 2017 and 2018 (Figure 2), even though maximum values do not differ so much due to the presence of so-called “hot spots”. This pattern continues into July, with the counts of 2019 being as low as 25% of those in other years but with a very high maximum value. In August, OFF counts in the traps in 2019 catch up with those of two previous years, while the highest maximum values of arrests are also recorded in 2019 (one count is more than 1350 OFFs). The lowest counts for August are recorded in 2017. This high growth rate continues in September 2019 with very high average trap arrests, as well as the absolute maximum value of arrests. Counts in October 2019 drop again below the average values of the two previous years. These developments seem to follow the weather conditions of the month in question and those of the previous month.

A closer look at the monthly differences with elevation classes reveals similar patterns in population changes over the season in different altitudes despite the significant differences in each year and especially 2019 in our case study. In the first month of each season, OFF counts in the lowland traps and 200–400 m altitudes are typically larger, but in hotter July, counts at higher altitudes increase (

Table 3. Average counts of OFFs per month and year for different altitude zones (asterisks indicate statistically significant differences—between values with the same indication e.g., *1, *2 etc.—of the average values with ANOVA tests and p < 0.05 for each month per year and altitude zone).

Altitude Zone	Year/Month	June	July	August	September	October
0–200 m	2017	23.7	38.8 *2	25.3 *3	53.7 *4	82.2 *7
	2018	25.5	49.0	49.6	69.4 *5	100.6 *6
	2019	6.0 *1	13.5	53.4	83.6	59.7
200–400 m	2017	18.6	51.5	66.7 *3	70.6 *4	39.9 *7
	2018	26.0	50.6	52.8	50.1 *5	51.7 *6
	2019	4.0	13.6	68.2	102.0	68.6 *8
>400 m	2017	27.8	74.4 *2	33.8 *3	36.8 *4	28.4 *7
	2018	21.6	42.4	49.0	51.8	41.2 *6
	2019	1.1 *1	27.9	35.9	54.3	45.2 *8
Total	2017	22.9	43.2	33.8	56.0	71.0
	2018	25.4	49.0	50.2	64.7	87.8
	2019	5.4	14.3	55.2	85.5	60.6

). In August, most counts are reported in middle altitudes, and in the final two months, lower temperatures and higher relative humidity favor lowland populations.

For the importance and effect of spraying, the findings vary significantly (

Table 4. The 30-day differences per number of trap counts for traps in sprayed and unsprayed areas for three 10-day periods: 10 days after spraying (T1), immediately before spraying (T0)%, 20 days after spraying T2–T1 (%), and 30 days after spraying T3–T2 (%).

10-Day Periods for 3 Years	T1−T0 (%)		T2–T1 (%)		T3–T2 (%)
	Unsprayed	Sprayed	Unsprayed	Sprayed	Unsprayed	Sprayed
1st	69.8		−26.0		4.1
2nd	−28.7	20.5	−1.4	65.1	−7.4	11.4
3rd	5.7	2.6	−27.4	19.7	−14.7	29.3
4th	0.5	−34.8	17.1	−34.5	36.6	90.1
5th	−8.4	98.7	40.1	39.1	14.1	40.6
6th	42.5	9.5	22.4	32.1	14.0	−24.4
7th	52.8	−17.3	12.5	8.5	−10.5	18.5
8th	21.9	2.3	11.4	−16.7	−0.4	−4.2
9th	−3.7	18.0	−3.4	17.1	−15.2	−28.1
10th	2.1	−9.1	−28.1	6.6	46.6	−0.5
11th	−3.4	−41.3	22.6	37.8	−16.4	3.2
12th	15.7	51.0	−9.0	−15.6
13th	−11.8	−10.2

), but overall and especially in the period of the first 10 days after the event, counts in areas that were sprayed do not seem to correspond to decreases in the OFF populations. In some cases, the effect of the spraying seems to be present in the second 10-day period. But, in many cases, the third 10-day period marks a rapid recovery of the OFF population. The timing within the season also seems to be important, as spraying events in the hotter months and until roughly the 6th period (end of July) appear more effective than in the last part of the season in the autumn (10th–13th period). Finally, infestation results from olive fruit sampling indicate that infestation rates rise as the fruits mature and grow in September and October each season, and climatic conditions also become more favorable for OFF populations.

The findings of the linear mixed-effects models (

Table 5. R-core Data Analysis Results.

	Value	Std. Error	DF	t-Value	p-Value
Intercept	0.5037640	0.11655140	718	4.322248	<0.001
Altitude	0.0032617	0.00037427	346	8.714936	<0.001
Spray	0.3988375	0.06502091	718	6.133989	<0.001
1st 10-days	0.0282857	0.00176229	718	16.050507	<0.001
2nd 10-days	0.0131370	0.00153348	718	8.566800	<0.001
3rd 10-days	0.0176605	0.00211540	718	8.348552	<0.001
1st: 2nd 10-days	−0.0000568	0.00000792	718	−7.174855	<0.001
1st: 3rd 10-days	−0.0002365	0.00003094	718	−7.644837	<0.001

) that combined many of the aforementioned explanatory variables for OFF numbers in the traps of our network did not show some significant results when all explanatory variables were included (altitude, sprayed, olive infestation levels, fly numbers after 10 days, fly numbers after 20 days, fly numbers after 30 days, and their interactions) (Table S2). The removal of the 20- and 30-day fly numbers and infestation levels improved the reliability of the results.

4. Discussion

4.1. Relevance of the Approach

In this paper, we investigate OFF population dynamics in relation to applications of crop protection management (bait spraying), with an approach that separates areas after each spraying event into sprayed and unsprayed and compares the effects of spraying for three ten-day periods. This approach is descriptive and exploratory, considering only macroscopic or landscape level effects of management practices. It does not take into account tree-level and field-level factors (see e.g., [13,30,31]). Nevertheless, its landscape scale is the actual scale at which OFF management is practiced, and we therefore argue that it should be practiced at that scale. The use of data for three seasons (2017–2019) for the whole island ensures that meteorological differences are considered, along with spatial data. We discuss some of the most important findings and lessons from and for better management.

4.2. Temporal Patterns of OFF Populations

The findings of the counts of OFF populations show that while weather conditions have a strong impact, conventional indicators such as the average daily temperature, or even the maximum daily temperature, are not suitable to describe or explain differences in OFFs over the different seasons or even different altitudes. Cumulative indicators though seem to correspond better to these differences in OFF populations. The average hours with temperatures over 32 °C are the indicator that best describes both higher OFF counts in cooler early 2018 and lower early 2019 counts during the hot start of the season. The values of RH% have to be also taken into account as they may favor or hinder insect movement and growth [41,42]. Later in the seasons, the hours over 32 °C again seem to describe the rapid growth of OFF counts in September of 2019. But, overall for the three seasons, despite the seasonal and the spatial differences of OFF counts, the overall populations counted in our trap network are not significantly different. Even in seasons such as 2019, where OFF numbers initially lagged, the populations quickly reached average levels. This seems a fundamental result and has to be taken into account in any efforts for collective management. At the same time, counts are not necessarily infestations. The data collected by the local office on fruit infestations do not indicate significant differences within each year. Infestation rates rise as the fruits mature and grow in September and October and climatic conditions favor OFF population growth. It has to be stressed though that many other factors affect infestation rates, such as unsuitable fruit (shrunken) due to low precipitation, higher predation rates, and others [43]. Differences are observed though, during and between the seasons, from north to south, and at high to low altitudes. This could be attributed to humid and cool northerly winds that prevail in the Aegean in the summer months that lower temperatures. OFF populations in the south increase later in the season due to hotter conditions in the southern lowlands and flatter terrain.

These findings also suggest that adults actively move in the landscape, as has been documented elsewhere [13,27,31,35]. Changes in favorable conditions that the climatic analysis has revealed may also affect reproductive success and population growth rates [41,44,45]. In fact, both factors could explain the differences observed here, and other authors have suggested that these factors are related: movement towards areas with more favorable conditions affects population dynamics [6,13,27,46]. The inclusion of more seasons would help in making sense of the climatic changes in the landscape.

4.3. Spraying

The analysis for spraying seems to reveal that the results are at best mixed in terms of the effectiveness of the management approaches used. This can be explained by a combination of different factors. The first, and most important one, is that they represent local events, typically one or two communities and their fields. To be ordered, OFF populations in the area have to be high and/or rising. Therefore, they are focused on “hot-spots” of OFF populations and a proper comparison has to take into account not just comparing these areas with other areas where OFF populations were lower or decreasing but with the same area populations if the spraying did not take place [27].

Another factor is the type of substances used: a typical example is the second period of the season, where counts in columns T2–T1 (%) and T3–T2 (%) are increasing and then decreasing due to the use of pyrethroid insecticides (deltamethrin) in the spray. These have a rapid effect but reduced effectiveness over time, and therefore OFF populations recover fast. Kampouraki et al. corroborate this [24], but Varikou et al. (mostly in 2018, but also in their 2017 and 2015 publications) question such claims and state that all attractants used are not very efficient, especially late in the season [47,48,49].

One more issue of concern for spraying seems to be its effectiveness in areas where populations are consistently high. The findings seem to suggest that spraying is more effective in areas of relatively low populations, even if it was found that the OFF population is vulnerable to insecticides on Samos island [24].

Finally, the changes must be weighed against the eventual movements of OFF populations due to weather conditions and the need to find food. These movements are a well-documented fact [13,27,31,50,51,52,53]. On Samos, the presence of many organic fields (almost 20% of the total area that is not sprayed) and abandoned fields (around 15% of the total area) in the margins or within the landscape matrix of the olives allows the movement of flies from unsprayed fields to the ones that were sprayed.

4.4. OFF Population Management

One of the objectives of our approach is to provide insights for OFF management. Four issues stand out:

(a) Spraying seems to be more effective in areas of average or relatively low overall populations but less so in areas where populations are consistently high;

(b) Spraying late in the season seems to be less effective in the lowlands, where weather conditions support a rapid increase in the population;

(c) Even in seasons like 2019, where OFF counts were low at first and infestation rates were also low, in late September and especially October in all periods, OFF populations increase significantly. This highlights the need of integrated pest management and use of alternative pest control methods, like repellents or mechanisms to confuse host recognition [54], which is more necessary considering the late stage of the fruit maturation, the greater damage inflicted, and the requirement of no residues in olive oil.

(d) The management regime of the olive plantation and most importantly the presence of neglected or abandoned plantations needs to be recorded. Mapping of these sites will support the effectiveness of the pest management program. Such fields are mostly located in uplands and intermediate altitude zones and can serve as “reservoirs” for flies as they are unsprayed, and fruits are not harvested.

The findings also indicate some shortcomings of the current management system. Possible improvements include:

(i) Mapping of abandonment hotspots;

(ii) Zone sprays between treated and untreated olive plantations that reduce pest movements, which is an old practice (Greek Ministry of Agriculture, 1949), discontinued in recent decades due to: (a) the decrease in available funds for spraying and (b) the false feeling that collective management can be successful in local areas due to extremely active substances that were used in the past (all out of circulation now) and air applications;

(iii) Grazing in abandoned fields can reduce pest population by consumption of the infested fruits;

(iv) Ground cover of (organic) olive groves with mixtures of selected flowering plants can enhance the presence of natural enemies of pests (Braconidae, Miridae, Chrysopidae) by providing them habitat and food and could be part of a sustainable management system [55];

(v) Mass trapping of fruit flies, especially in organic olive orchards, could decrease the OFF population [56] and improve the efficacy of management programs;

(vi) Improvement of bait sprays and frequent treatment in hot spots could be applied depending on the active substance and lure duration [49,57], as reduced efficacy of the usually applied food attractants mixed with registered plant protection products has been reported [58], along with reduced attraction of hydrolyzed proteins [47], indicating the need for refreshing bait spots of the treated olive trees only with attractants reducing the insecticide doses per hectare by two- or three-fold [48].

5. Conclusions

Insect dynamics in the field are a complex topic due to a plethora of interrelated factors. In this paper, we investigated the landscape-level dynamics of OFFs to provide insights for more effective management at the area level. We advocate for landscape-level management to account for the landscape-level factors that need to be considered. Weather conditions and pest management regimes were proven important, but more factors will also have to be considered, such as host (fruit) availability; inhibition factors, like natural enemies, symbiotic agents, or other food supplies; and cultivation practices that were not measured in our approach.

The results nevertheless indicate that the landscape level can indeed be used to describe changes in detail that add depth to explanations at the individual-plot level. These descriptions seem to confirm movements in the population. The shortcomings of management were also identified in our approach, the most important of which refer to the presence of neglected or abandoned plantations interspersed in the olive mosaic that underline the need for more comprehensive area-level management and the incorporation of the landscape matrix in the estimations, which can reduce the need for spraying and overall costs.