A study of diel and seasonal patterns of loss of commercial lychee fruits to vertebrate frugivores: implications for mitigating a human-wildlife conflict

Site and species description

Mauritius (20°20′S; 57°34′E) is a volcanic island 7.8 M years old in the Indian Ocean about 900 km east of Madagascar (Fig. 1). It covers 1,865 km², spanning about 60 × 40 km, has a maximum elevation of 828 m and is located within one of the world’s biodiversity hotspots (Myers et al., 2000). The mean annual temperature is 22 °C, and the mean annual rainfall varies from 800–4,000 mm (Staub, Stevens & Waylen, 2014). Habitat destruction due mainly to agriculture and urban development decreased its native forest cover to 4.4% (Hammond et al., 2015). Sugarcane plantations have occupied more than half of Mauritius (Nigel, Rughooputh & Boojhawon, 2015). However, sugarcane plantations are being converted to other crops including fruit (Ministry of Agro-Industry and Food Security of Mauritius, 2015). Lychee is one of the three main fruits produced on the island (Statistics Mauritius, 2017), and it has economic importance to both the local market (around 7,000 tonnes) and export industry (around 1,000 tonnes) (Fresh Plaza, 2016). However, data on its market value remain unavailable.

The Mauritian flying fox is a medium-sized fruit bat weighing 380–540 g and the last Pteropus species surviving on Mauritius following the extinction of two other species (Cheke & Hume, 2008). Its natural diet consists of leaves, nectar, flowers and mainly a wide variety of native fruits, making this species a keystone seed disseminator (Florens et al., 2017) and a potential pollinator (Nyhagen et al., 2005). Following the extinction of the two other Pteropus species, giant tortoises (Cylindraspis spp.), the dodo (Raphus cucullatus) and other large frugivores, P. niger has become the largest native seed disseminator of the Mascarenes (Hansen & Galetti, 2009; Heinen et al., 2023). It is known to be able to cross the whole island in a single night for foraging (Oleksy et al., 2019). The island has other native vertebrate frugivores, such as birds, comprising six passerine species and one parrot (Heinen et al., 2023). Apart from the Mauritius grey white-eye (Zosterops mauritianus), all the other native birds’ distribution is restricted to native forests (Jones, 1987; Safford, 1997). The Mauritius black bulbul (Hypsipetes olivaceus) is the main frugivore among native passerines but exotic birds (common myna and red-whiskered bulbul) competitively excludes it from non-native habitats (Safford, 1996). These exotic birds can move between native and secondary vegetation since they are highly mobile (Dulloo, Kell & Jones, 2002). Other alien frugivores include rats (R. rattus and R. norvegicus) and long-tailed macaques, but the red-whiskered bulbul is considered one of the most invasive species on Mauritius (Cheke, 1987).

Sampling and statistical analyses

With permission from the Ministry of Agro-Industry and Food Security, we sampled two lychee orchards, one at Calebasses (north, nine hectares) and one at Beaux Songes (west, seven hectares) and two backyard gardens in the north and one in the central upland region (Fig. 1). Twelve non-netted unharvested fruiting trees of ∼eight m height and 9–11 m canopy diameter were randomly selected in each orchard and a total of six trees of similar height and canopy diameter were selected in backyards. Random selection was done by assigning numbers to all non-netted fruiting trees in orchards and numbers were drawn using a random number generator. Backyard tree selection was dependent on the owner’s permission to enter the property at night and whether the tree would be left non-netted. Lychee season in Mauritius starts in September (fruit set), and the ripening stage spans November to end of December. The selected sites were visited regularly to monitor fruit growth and quadrats were set up before damage started, which was approximately six weeks after fruit set (following lychee ripeness stages (Wei et al., 2013; Chang et al., 2015)). Permanent quadrats of one m² were placed at the four cardinal points, midway between the trunk and the canopy edge of each assessed tree, while slightly adjusting the placement of quadrats if the midpoint was located under foliage only.

Fruit ripening begins in the north region, and proceed southwards. Fruit damage started on 12 November 2022 in the north and 30 November 2022 in the west and central uplands and damage assessment lasted until no fruits remained in orchards or until fruits were harvested in backyards (28 November and 27 December 2022, respectively). We collected all the fruits falling into quadrats, identified the agents of frugivore damage (flying fox, bird, rat and macaque) using bite marks (Senar et al., 2016; Oleksy et al., 2021) (Fig. 2) and separated the amount of pulp eaten per fruit into five categories (0%, 25%, 50%, 75% and 100%) (Fig. S1). The corresponding ripeness stages of damaged fruits that still had their pericarp were also classified into percentages (characterised by change in pericarp colour from green; 0% ripe to red; 100% ripe) (Fig. S1). Sampling was done every six hours (06:00–12:00, 12:00–18:00, 18:00–00:00, 00:00–06:00) for 24-hour cycles. We recorded complete cycles three to five consecutive days per week for a total of 13, 11 and 15 days at Beaux Songes, Calebasses and backyards respectively throughout the fruiting season. The different bird species observed foraging on lychee during sampling were also noted. We estimated visitation frequencies of flying fox, bird, rat and macaque based on the presence of damage per tree, time and day. Data collected for backyard trees were focused on temporal quantification of flying fox damage only. One AudioMoth (recording 55 s every five minutes from 18:00–06:00) was used for two consecutive weeks in each orchard to record any deliberate disturbance made to deter flying foxes. We randomly selected 42 ripe fruits from a backyard tree and measured their weight (average 21.3 ± 2.3 g) to estimate fruit loss in kg.

All statistical analyses were done in R (version 4.3.1) (R Core Team, 2023). Analyses were done for flying foxes and birds only as we had too few observations for other frugivores. We tested the hypothesized effects of time period (06:00–12:00 (morning), 12:00–18:00 (afternoon), 18:00–00:00 (evening) and 00:00–06:00 (night)) and successive sampled days on the amount of fruit eaten by flying foxes and birds per tree in every site (Beaux Songes and Calebasses for flying foxes and birds, and backyards for flying foxes) using generalized linear models (GLMMs) with negative binomial error distributions. We first fitted global models (using package glmmTMB (Magnusson et al., 2024)) with two-way interactions between day and site and between time period and site, to account for within-site variation in the relationships between number of fruits eaten, time period and day. For flying foxes, we only included two time periods (18:00–00:00 and 00:00–06:00), as they almost exclusively fed during these periods. For birds, we included all time slots, as they usually started feeding before 06:00 and finished feeding after 18:00. We included tree as random effect to account for correlation of repeated measurements per tree across time (Harrison, 2014). We evaluated model fit of all global models using residual diagnostic plots from package DHARMa (Hartig, 2022).

We used the global model for inference, as this also provides a balanced representation of statistically non-significant results. We first tested the global model against a null model with a likelihood ratio test (Forstmeier & Schielzeth, 2011). Next, we reported the model estimates and evaluated statistical significance of observed estimates using 95% confidence intervals (CIs) (Nakagawa & Cuthill, 2007). We considered evidence for an effect as weak, moderate and strong when the 90, 95 and 99% CIs did not overlap zero, respectively (Muff et al., 2022). For the two-way interaction between site and day in each GLMM, we calculated regression coefficients, standard errors (SEs), and CIs for the day slope for site, correcting for multiple comparisons using the Tukey method (package emmeans (Lenth et al., 2017)). For the two-way interaction between site and time period in each GLMM, we carried out post hoc contrast tests for the pairwise comparisons between time periods in every site (using package emmeans).

We also tested our hypothesised effects of stage of fruit ripeness on the proportion of pulp eaten by flying fox, parakeet or other birds in each orchard using a GLMM with ordered β distribution (and using individual fruits as sampling units). The ordered β distribution is similar to the zero–one-augmented β distribution (recommended for continuous proportions) (Douma & Weedon, 2019) but produces more accurate estimates and needs less processing time (Kubinec, 2023). In this GLMM, we included a three-way interaction between site, flying fox, parakeet or other birds and fruit ripeness stage and between site, flying fox, parakeet or other birds and day as fixed effects. We also included tree as a random effect to account for pseudoreplication. We used the same validation procedures as described for the previous GLMMs and used the global model for inference. We calculated regression coefficients, standard errors (SEs) and CIs for the slope of fruit ripeness stage for every species in every site and carried out post hoc contrast tests for the pairwise comparisons between predicted means in proportion of pulp eaten by different species for different stages of fruit ripeness in every site.

Agents of fruit loss

Fruit loss by frugivores in orchards was caused by flying foxes (native) and birds, rats and macaques (all alien). We also accounted for fruit loss by other factors which included mainly fungal diseases, fruit cracking and natural fruit fall. The number of fallen fruits per square meter per day varied from 0–67 at Beaux Songes and 0–102 at Calebasses. At Calebasses and Beaux Songes respectively, fruit loss by flying foxes was 62.6–81.2% (n = 2,401) and 61.3–94.4% (n = 3,392); introduced birds was 9.8–30.4% (n = 711) and 2.4–26.7% (n = 289); and other factors was 1.2–12.2% (n = 191) and 1.3–12.0% (n = 183). Rat and macaque damage occurred only at Beaux Songes where it represented less than 1% (n = 5) damage. Only parakeet damage was distinguished from other birds. However, a total of four avian frugivores were observed feeding on lychees and they were all introduced species namely ring-necked parakeets, red-whiskered bulbuls, common mynas and village weavers (Ploceus cucullatus). The highest visitation frequency was by flying foxes (n = 446) and birds (n = 306) and the lowest was for rats (n = 3) and macaques (n = 1). Hence, flying foxes and birds were classified as regular visitors while rats and macaques were considered occasional visitors.

Temporal pattern of flying fox foraging behaviour

Of the fruits eaten by flying foxes, an average of 59% were eaten in the early night (54% Calebasses, 59% Beaux Songes, 68% backyards). The number of damaged fruits per tree (four m²) at Calebasses, Beaux Songes and backyards respectively varied from 30–280, 26–425 and 8–521 before midnight and 14–222, 20–226 and 7–196 after midnight. Although the estimated difference between early and late night was on average one (Calebasses), two (backyards) and three (Beaux Songes) fruits per tree per night, the GLMM analysis showed that the number of fruits consumed was significantly less in the late night than in the early night in Beaux Songes and backyards (Fig. 3, Table 1). At Calebasses, flying fox damage started with 44 fruits, peaked at 413 fruits (5th sampled day) after four days and dropped to 143 fruits on the last (11th) sampled day. The total number of fruits eaten at Beaux Songes started with 409 fruits on the first day, peaked at 573 fruits (4th sampled day) after four days and ended with 123 fruits on the 13th sampled day. They ate on average ∼1 fruit less per tree with increasing sampling days in both orchards but not in the backyards (Table 1, Fig. 4). The low R²_marginal indicated that our predictors only explained 11% of variation in the amount of fruit eaten by flying foxes, meaning other factors unaccounted for probably considerably influenced the flying foxes’ feeding behaviour.

Temporal pattern of bird foraging behaviour

Fruits eaten by birds were on average 49% (n = 141) in the morning, 13% (n = 38) in the afternoon, 2% (n = 6) in the evening and 36% (n = 102) at night in Beaux Songes and 70% (n = 500), 28% (n = 199), <1% (n = 2) and 1% (n = 10) respectively at Calebasses. We found strong evidence that birds ate most fruits at night (00:00–06:00) and in the morning (06:00–12:00) in Beaux Songes and in the morning and afternoon (12:00–18:00) in Calebasses (Table 1, Fig. 5). The R²_marginal indicated that our predictors explained a considerable amount of variation (65%) in the amount of fruit eaten by birds. The total number of fruits eaten at Beaux Songes started with 17 fruits on the first day, peaked at 39 fruits (9th sampled day) after 14 days and ended with 23 fruits on the 13th sampled day. At Calebasses, bird damage started with 11 fruits, increased to 90 fruits (5th sampled day) after four days and dropped to 66 fruits on the last (11th) sampled day. In contrast to flying foxes, we found weak to moderate evidence that birds ate slightly larger number of fruits toward the end of the study in both orchards (Table 1, Fig. 4). However, the relatively large SEs indicate that more observations were needed to obtain more precise estimates.

Fruit ripeness and foraging behaviour of flying foxes and birds

Flying foxes ate between 20–28% more lychee pulp per fruit than parakeets at Beaux Songes and we found strong evidence of this trend at 0%, 25% and 50% ripened fruit (Table 2). In Beaux Songes and Calebasses, flying foxes ate 7% and 6% less fruit pulp than birds at the 25% ripened stage respectively, and it differed significantly at both sites. From unripe to 50% ripe fruits, parakeets ate 38–26% less fruit pulp than birds at Beaux Songes (strong evidence) and 7% less fruit pulp at 25% ripe fruit at Calebasses (weak evidence). Furthermore, flying foxes ate 39% and 42% more lychee pulp per fruit with an increase in fruit ripeness from unripe (0%) to fully ripe (100%) at Beaux Songes (strong evidence) and Calebasses (strong evidence) respectively (Table 2, Fig. 6). Parakeets ate 7% more fruit pulp with an increase in fruit ripeness at Beaux Songes and we found moderate evidence for this trend (Table 2, Fig. 6). With every increase in sampling day, flying foxes ate 1% more pulp per fruit at Beaux Songes (strong evidence) (Table 2). However, the small number of observations for parakeets and other birds for some levels of fruit ripeness resulted in large SEs, meaning additional observations were required to estimate the trends for parakeets and other birds more accurately. The low R²_marginal indicated that other factors that we did not account for may explain a larger amount of variation in the proportion of fruit pulp eaten by different species.

Deliberate disturbances in orchards

The AudioMoth recorded deliberate disturbances at both orchards. The number of events per six hours varied between 12-27 in the evening (18:00–00:00) and 1-30 at night (00:00–06:00) at Beaux Songes and 0-18 and 1-11 respectively at Calebasses. The duration of the sampled disturbances accounted for 3.0% (SD 1.4%) of the evening time period and 2.7% (SD 2.3%) of the night time period at Beaux Songes and 1.6% (SD 1.1%) and 1.0% (SD 1.0%) respectively at Calebasses. Five different types of disturbances were recorded and they were either noise-based deterrence (firecracker/gunshot, sound of stick hitting empty barrel, whistling and shouting) or smell-based (burning of leaves/branches/tyres) which could also be detected with the AudioMoth by the sound of crackling and sputtering during burning. However, since we were restricted to using a single AudioMoth throughout the fruiting season, we did not have enough data for more detailed analysis about the influence of such disturbances on foraging flying foxes.

Impact of flying foxes and other frugivores on fruit loss

During our study, the percentage of damaged lychee attributed to flying foxes in orchards was 61–94% per tree. Given that we sampled an area of four square meter per tree, we estimated fruit loss by flying foxes between 19–272 kg of lychee per tree in orchards (based on an average tree diameter of 10 m (canopy area: 78.54 m²) and a fruit weight of 21.3 g). Oleksy et al. (2021) assessed fruit damage in 2015 at the same orchards as in our study to show the efficacy of netting trees and they recorded flying fox damage of 9% and 53% for non-netted lychee trees at Calebasses and Beaux Songes respectively. Both orchards covered between seven and nine hectares but Calebasses was divided into several smaller plots of approximately 0.4 hectares each. As a result, Calebasses had multiple owners, whereas Beaux Songes had only one. Hence, the reported differences between the two sites appear to have been influenced by the intensity of active deterrence, as the presence of multiple owners at Calebasses likely led to more pronounced deterrent measures. Furthermore, since 2009, the Mauritian Government has been providing subsidies to purchase nets for orchards and backyard gardens (Government of Mauritius, 2010). The higher (63% and 32% more at Calebasses and Beaux Songes respectively) flying fox damage recorded in our study suggested that the considerable increase in the use of nets in orchards during the last seven years may be displacing the damage by frugivores away from netted trees to concentrate them onto the fewer trees that remained unprotected. This would highlight the importance of protecting trees with methods proven as effective.

Bird damage was relatively low (3–68 kg per tree) and bird sonic deterrence (clapping, shouting and music) was occasionally used. Birds also seemed less affected by sonic deterrence, possibly explaining why a much larger amount of variation in bird foraging intensity was explained by time slot and day. Fruit loss by rats was negligible as they tend to hoard food (Bindra, 1948) and because rodenticides were used after fruit set. Even though macaques do visit Beaux Songes orchard (G Bhanda, pers. obs., 2018), their damage remained negligible. Macaques typically do not swallow lychee-sized seeds (Tsuji & Su, 2018), but may store them in their cheek pouches or carry them away in their hands over distances >20 m (Corlett & Lucas, 1990), particularly when disturbed by humans during crop raiding (from up to 200 m; R Reinegger, pers. obs., 2024). Hence, alternative methods for assessing crop raiding by macaques using camera traps paired with direct observations could be explored. Fruit loss by other factors was low because practices like watering adequately and frequently during fruiting to decrease the occurrence of fruit cracking (Marboh et al., 2017) and preventively spraying fungicides, were common. Furthermore, the maintenance of windbreaks around orchards could have reduced natural fruit fall. Apart from flying fox damage, damage by birds, alien mammals and other factors aligned with Oleksy et al. (2021).

Diel patterns

We confirmed that flying foxes followed a pattern while foraging as the amount of fruit eaten varied during the night. However, much of the variation in flying fox feeding intensity remained unexplained by foraging time or study period. Differences between trees explained a greater proportion of variation in feeding intensity. Hence, factors such as proximity to other fruiting trees and crop size could have affected fruit choice and frugivore’s foraging intensity (Manasse & Howe, 1983; Ortiz-Pulido, Albores-Barajas & Díaz, 2007). Nevertheless, since GLMMs residual plots indicated homoskedasticity and no other deviations, the effects of our predictors were still relevant. Hengjan et al. (2018) found a correlation between number of fruits dropped at different hours and frequency of flying fox visits. Hence, the amount of fruit eaten by flying foxes was considered as an indicator for its density. The higher flying fox density recorded in early night at Beaux Songes orchard and undisturbed backyards suggested that they were targeting familiar foraging patches with energy and water-rich foods. Flying foxes in Mauritius demonstrated seasonal movement patterns depending on food availability (Oleksy et al., 2019) with more flying foxes foraging outside natural forest areas during commercial fruiting seasons (Seegobin, Oleksy & Florens, 2022). Studying the black flying fox (P. alecto), Markus & Hall (2004) showed its tendency to fly directly towards known foraging sites at nightfall. Another study on Egyptian fruit bats (Rousettus aegyptiacus) suggested that experienced bats time their visits with an understanding of tree phenology and showed that bats that have not eaten much and did not drink for 12 h would leave their colony early to target water-rich fruits compared to late-leaving bats seeking protein-rich fruits (Harten et al., 2024).

Non-netted orchard lychee trees sustained high fruit losses to flying foxes despite being subjected to some degree of sound and smoke deterrence. These deterrence practices are common in most Mauritian orchards. Orchard owners would either camp there during fruiting season or employ people to do so to actively deter flying foxes and thieves at night. Despite the higher flying fox damage recorded before compared to after midnight at Calebasses, the difference was not statistically significant. Since both orchards were subjected to active deterrence, it seemed that flying foxes at Calebasses could have been more affected as the difference in the duration of disturbance before and after midnight was slightly higher than at Beaux Songes. Disturbances may have varied in intensity between different parts of the orchards too, further explaining between-tree differences in flying fox foraging intensity.

Using active deterrents only during early nights would protect on average ≤1 kg more lychees per tree per night. In some cases, it could protect up to 42 kg more lychees per tree per night, but in other cases, it could result in a maximum loss of 20 kg. Additionally, animals tend to shift their activity with human disturbance (Gaynor et al., 2018; Lee et al., 2024). Consequently, deterring flying foxes only before midnight could lead to more flying foxes foraging in orchards after midnight. Although effective, as a result of the short term renting of orchard trees for exploitation that precludes cost-effective multi-year netting, fruit sellers find netting trees to be more costly, labor-intensive, time-consuming, than it ought to be, and point out that net supplies are often limited, especially during the fruiting season. For example, adopting active deterrence method can cost 2.5 times less than netting a lychee orchard of one hectare (∼180 trees) (orchard owner, pers. comm., 2024). Hence, for orchard owners relying only on active deterrence to protect trees from flying fox damage, it is recommended to protect the orchard during the whole night. While deliberate human disturbances involving the use of smoke or sound has been reported in Australia and India to deter frugivorous bats (Srinivasulu & Srinivasulu, 2001; Bicknell, 2002), their effectiveness in decreasing bat damage in orchards remained to be studied. No study demonstrating the effectiveness of these commonly used methods has been published in Mauritius.

Our study was done in summer with sunrise around 40 min before 06:00 and sunset 35 min after 18:00, explaining the record of bird damage at “night” (approximated to 18:00–06:00 in our study). Studies that modeled the optimal foraging behaviour of birds by including predation and starvation risks often predicted a bimodal feeding pattern with early morning and late evening peaks (Bednekoff & Houston, 1994; McNamara, Houston & Lima, 1994). Since our study was limited to diurnal data collected before and after noon, we could not investigate this bimodality. However, we found an early morning foraging peak followed by a decreasing rate of foraging throughout the day at both orchards. This peak could be explained by unpredictable food sources (influenced by weather changes and interruptions by competitors and predators) and a higher starvation risk in early morning pushing the birds to replenish energy reserves exhausted overnight (Bednekoff & Houston, 1994). When feeding is uninterrupted, birds are expected to decrease their foraging activity the rest of the day, feeding to maintain energy reserves and low predation risk (Bednekoff & Houston, 1994; McNamara, Houston & Lima, 1994). At Calebasses, foraging was significantly higher approximately one hour after sunrise. At Beaux Songes, however, foraging was significantly higher from sunrise possibly due to the residing large colony of village weaver nesting on ironwood trees (Casuarina equisetifolia) found inside the orchard. This is an additional and fourth alien bird species feeding on lychee in Mauritius compared to Oleksy et al. (2021).

Relationship between seasonality, fruit ripeness and foraging

Flying foxes and passerine birds at Beaux Songes ate more lychee pulp from ripeness stages 0% to 50% compared to parakeets. Birds at both orchards ate more pulp from 25% ripe lychees compared to flying foxes. Flying foxes ate more fruit pulp with increasing ripeness at both orchards and more fruit pulp towards the end of the study at Beaux Songes. This trend was not apparent at Calebasses as, for most of the assessed trees, all fruits were eaten before they ripened. While flying foxes are known to prefer ripe fruits (Luft, Curio & Tacud, 2003), we could not assess this preference in our study as we did not account for the availability of fruits of different ripeness stages on trees. However, consistent with previous findings, we could speculate that they favour ripe fruits (Krivek et al., 2020; Reinegger et al., 2021). Flying fox damage began in the seventh week after fruit set, when fruits were still unripe, with most damage occurring between the unripe and 25% ripened stages, during which they consumed little fruit pulp. This behaviour is common in foraging animals as they tend to become more choosy during high fruit availability when the probability of obtaining fruits of higher quality increases (Pyke, 1984; Janson, 1996; Whitehead, Quesada & Bowers, 2016). Hence, netting trees at latest in the sixth week after fruit set is advisable, a practice that is rare among most orchard owners who often end up netting their trees when fruits start ripening (G Bhanda, pers. obs., 2018–2022).

Ring-necked parakeets also ate larger proportions of pulp with increasing fruit ripeness in Beaux Songes, indicating that they could also target ripe lychee pulp. This suggested that they could be more frugivorous than granivorous (Shivambu, Shivambu & Downs, 2021). Consequently, like flying foxes, ring-necked parakeets were also more wasteful when feeding on unripe fruits, similar to findings by Sebastián-González et al. (2019), as they would only bite the fruit or partially consume it before discarding the rest. However, additional observations were needed to confirm how strong this trend was among parakeets in Mauritius because most fruits had been consumed when unripe leaving few ripe ones. For both orchards, frugivores ate less fruits towards the end of the fruiting period because fewer fruits remained on trees as they had already started depleting fruits from the beginning.