Pollination Services of Mango Flower Pollinators.

Measuring wild pollinator services in agricultural production is very important in the context of sustainable management. In this study, we estimated the contribution of native pollinators to mango fruit set production of two mango cultivars Mangifera indica (L). cv. 'Sala' and 'Chok Anan'. Visitation rates of pollinators on mango flowers and number of pollen grains adhering to their bodies determined pollinator efficiency for reproductive success of the crop. Chok Anan failed to produce any fruit set in the absence of pollinators. In natural condition, we found that Sala produced 4.8% fruit set per hermaphrodite flower while Chok Anan produced 3.1% per flower. Hand pollination tremendously increased fruit set of naturally pollinated flower for Sala (>100%), but only 33% for Chok Anan. Pollinator contribution to mango fruit set was estimated at 53% of total fruit set production. Our results highlighted the importance of insect pollinations in mango production. Large size flies Eristalinus spp. and Chrysomya spp. were found to be effective pollen carriers and visited more mango flowers compared with other flower visitors.

Pollinator species differ in morphological characteristics and behaviors that determine their ability to pollinate (Horsburgh et al. 2011). Various insect pollinators frequent plants differently and thus vary in visitation rates, removal, and deposition of pollen within spatial as well as temporal scales (Sahli and Conner 2007). Paralleled with emerging need for multi-year and multi-site comparisons of risk assessment in conservation and sustainable agriculture, measuring pollinator performance has become increasingly important (Ne’eman et al. 2010, Benjamin and Winfree 2014). According to Rader et al. (2009), the most effective pollinator species are those occurring in high abundance, actively moving from flower to flower (has a high visitation rate) and transferring many pollen grains on the stigmas.

Pollinator efficiency or effectiveness measures the amount of pollen deposited by an insect onto a stigma. This technique can be extremely difficult for many small flowers held on a panicle because of uncertainty in locating the pollinated flowers. Moreover, pollen deposition depends totally on insect’s flower-visiting activity that is strictly species specific (Howlett et al. 2011) in many flowering plants, therefore the technique becomes very time consuming. Alternatively, counting pollen grains on flower-visiting insects is a quicker and an easier method because the pollinators can be directly captured from the flowers (Howlett et al. 2011). Although not the best technique to measure pollinator efficiency, the amount of pollens carried by each insect also reflects its effectiveness based on the foraging behavior, pollinating ability, and ecology of the pollinator (Borkent and Schlinger 2008, O’Neill and O’Neill 2010).

Mango (Mangifera indica) flowers are pollinated by various insects such as wasps, ants, flies, butterflies, beetles, and bees as well as by wind (Bally 2006, Aliakbarpour and Che Salmah 2010). The important role of insect pollinators in mango production has been recognized in many mango producing countries in the world. Sung et al. (2006) reported that honey bees (Apis spp.) and flies (Musca domestica and Chrysomya megacephala) were responsible for successful pollination hence good mango production in Taiwan. In India, Sharma et al reared Lucilia sp. (Calliphoridae) and Sarcophaga sp. (Sarcophagidae) flies in their attempt to increase the abundance of pollinators in a mango orchard. These flies were supplemented with fish or mutton pieces in mesh bags hung on the lower branches of mango trees. Meanwhile, Apis cerana was kept for pollination of mango flowers in addition to producing honey in a small scale industry in Thailand (Wongsiri and Chen 1995).

Although mango panicles has a lot of hermaphrodite flowers (Bally et al. 2009), Usman et al. (2001) found that cross pollination had contributed to large increase in mango fruit set. Hermaphrodite flowers are self-pollinated but incompatibility of some pollen and stigmas cause failure in mango fruit set (Singh et al. 1962, Mukherjee et al. 1968, Sharma and Singh 1970, Dag et al. 2006, Gehrke-Vélez et al. 2012). In Chok Anan cultivar for example, Ding and Khairul Bariah (2013) detected some degree of self-incompatibility and hence cross pollination is necessary for a successful fruit set.

Up to the present time, modern agriculture shows heavy reliance on domesticated pollinators for crop production (Benjamin and Winfree 2014). For that matter, more focus is given to managements of pollinators particularly honey bee (Apis mellifera) populations within agricultural environments (Abrol 2012). Unfortunately, mango flowers do not appear to be overly attractive to honey bee (Popenoe 1917, Usman et al. 2001) due to little amount of its honey production. In an attempt to improve mango production in Malaysia, this study was aimed to assess the contribution of wild pollinators to successful fertilization of mango flowers by quantifying the number of fruit set produced. The second objective estimated the efficiencies of various pollinator species through their visitation rate on the mango flowers (Sahli and Conner 2007). We presumed that the visitation rates of mango pollinators and their pollen loads were species specific and differed between male and female pollinators.

Materials and Methods

Study Area

The assessment of pollinator services was conducted at Universiti Teknologi MARA (UiTM) Perlis, Malaysia; fruit orchard located in Arau district at coordinates of 6°26′59″ N and 100°16′47″ E. Two 78 m2 plots of 40 trees, 10-year-old Chok Anan-MA224 and Sala-MA164 cultivars were selected for this study during the flowering season of 2014 from 28 January to 7 March. Annual temperature in the district ranges between 24 and 36°C with humidity of 52.6–95.6%. This area receives mean daily precipitation of 16 mm and wind speed at 0.3 ms−1, respectively.

All trees were fertilized regularly and a flower inducer (FlowStar 250, active ingredient is Paclobutrazol 25%, Brightonmax Sdn. Bhd., Malaysia) was applied 2–3 months prior to the flowering season. Weeds were manually cut monthly. Out of 40 trees in each plot, 20 relatively uniform sized, flowering Chok Anan, and Sala trees were randomly selected for the assessment.

Assessment of Pollinator Performance

Pollinator performance was assessed following the methods of Freihat et al. (2008) and Gemmill-Herren and Ochieng (2008). Three flower panicles from each Chok Anan and Sala trees were chosen from 20 trees of each cultivar and a total of 120 panicles were used in this study.

Pollination Treatments

Pollinator performance was assessed by comparing the number of mango fruit set produced after the flowers were exposed to three pollination treatments arranged in a completely randomized design. Each of the three flower panicles from a mango tree used in this study was exposed to one of the three pollination treatments. The first panicle was allowed to be naturally pollinated by available wild pollinators in the orchard throughout the flowering period. In addition to natural pollination, the second panicle was artificially provided (hand transferred) with additional pollens from different flowers within the same tree (Azhar and Ithnin 2009, Bally et al. 2009). Due to asynchronous blooming of flowers, provision of pollens was repeated at 3-day interval until all flowers completely bloomed. The third panicle served as a control (pollinator’s exclusion). This flower inflorescence was inserted into a net bag of 0.25 by 0.25 mm mesh size a week prior to flower bloom to prevent from insect pollination. The bag was removed when all flower petals dried and fell off, and the ovaries became considerably swollen. The number of mango fruit set in each panicle was recorded 30–35 days after the commencement of pollination in each treatment.

Estimation of Hermaphrodite Flowers on a Flower Panicle

Ten fully developed Sala (20–25 cm long) and Chok Anan (25–35 cm long) panicles were cut off and the numbers of hermaphrodite and male flowers on each panicle were counted. The mean number of hermaphrodite flowers on each inflorescence was used to calculate the percentage of fruit set per hermaphrodite flower of respective cultivars at 30–35 days after blooming.

Assessing Pollinator’s Contribution to Plant Reproductive Success

Pollinator Visitation Rate

To assess the attractiveness of mango flowers to pollinators, field observations were conducted weekly for 5 weeks during pollinators’ foraging hours from 1000 to 1100 hours. The time was selected based on our preliminary observation during flowering season of 2013. At this hour, the foraging activity fell in the middle of anther dehiscence period (from 0900 to 1200 hours) for mango grown in Malaysia (Azhar and Ithnin 2009). Camponotus and Iridomyrmex ants (Family Formicidae) were included for this study because of their high abundance on flower panicles. In addition, three fly genera, Stomorhina, Chrysomya, and Sarcophaga (Diptera) moved actively from panicle to panicle and we assumed that they contributed to cross pollination of flowers. All flower visitors used in this study have never been reported as pests of mango flowers and they visit the flowers mainly to obtain nectar and pollen (Waterhouse 1993, Pena et al. 1998). Fifteen individuals from each genus were observed for their handling time (period of insect spend on flowers to collect nectar or pollen) on flower panicles in a 2-min observation period.

In this study, visitation frequency represents the number of single-visit flower for each insect species during a 2-min observation period which took place within 1000 h 1100 hours. The number of visits also indicates individual pollinator effectiveness (Gemmill-Herren and Ochieng 2008, Jacobs et al. 2010). The duration of floral manipulation and handling time of each visitor was measured by a digital stopwatch (±0.1 s) (Casio Stopwatch HS-3, Casio Electronic Manufacturing Co., Ltd., Japan). For each visit, the category (crawling, flying) of insect was recorded and the behavior of the insect was observed.

Pollen Carrying Capacity by Pollinators

A pollinator species that landed on a flower panicle was captured carefully into a glass vial (3 by 2 cm—height by diameter) containing a piece of tissue paper soaked in 98% diethyl ether, a rapid killing agent (Jacobs et al. 2010, Howlett et al. 2011). The pollinator captured had to be killed immediately to prevent pollen loss from the insect body and to prevent the insect from grooming the pollens off its body. This method of study offers two major assurances: (1) that it confirms the insects collected are floral visitors; and (2) that it prevents pollen contamination among pollinators. However, due to sampling difficulty and pollinator availability, the number of replicates in this study varied among species.

Pollen Atlas

A temporary pollen atlas was prepared to accurately identify the mango pollen and to differentiate it from weed pollen that the pollinators might carry. Mango pollen were collected from fully bloomed flowers of several panicles and soaked in 3 ml of 75% ethanol. Simultaneously, the pollen of 16 common weed flowers growing within the orchard were collected and preserved in the same manner. All pollen were observed on slides under a compound microscope Olympus CX41 (Olympus UK Ltd., England) with 40× magnification. Pictures of various pollen were captured using a digital camera (Xcam-α) equipped with an acquisition software DigiAcquis version 2.0 (2006) by Matrix Optics (M) Sdn. Bhd, Malaysia.

Estimation of Pollen Grain

The number of pollen grains adhering to a pollinator’s body was estimated following the methods of Freitas et al. (2002) and Godini (1981). Each captured pollinator was soaked and washed thoroughly with 2 ml 75% ethanol in a vial. Dislodged grains were counted using a Neubauer-improved haemocytometer under a compound microscope Olympus CX41 (Olympus UK Ltd., England). The haemocytometer is a depression microscope slide with two counting chambers, each with 9-mm2 surface area. When covered with a cover slip, it gives the chamber depth of 0.1 mm. The chamber is subdivided into nine small chambers of 0.1 ml and 18 small chambers can be examined at a time (two counting chambers × nine small chambers). Pollen sample in the solution was mixed thoroughly and using a micropipette, a drop of the suspension was extracted and placed into the haemocytometer chambers. The 2 ml pollen suspension was completely withdrawn in five extractions. Each of the small chamber in the haemocytometer retained 1/20,000 pollen suspension from the original amount of 2 ml in each vial giving a total proportion of 90/20,000 ml (18 small chambers × five drops) suspension observed for each sample. All pollen grains inside the small chambers were counted. In this study, the amounts of pollens on Camponotus, Iridomyrmex, Stomorhina, Chrysomya, Sarcophaga, Eristalinus (Syrphidae: Diptera), and Specodes (Halictidae: Hymenoptera) were examined. Unequal numbers of replicates were collected for each insect genus depending on its availability in the field. Although the two later genera were not commonly found on mango panicles, few of them were collected during the sampling periods. Therefore, their pollen carrying capacities were compared with five other genera. Body sizes, head width, head length (O’Neill and O’Neill 2010), and body length (head to tip of abdomen) of all pollinators were measured under an Olympus SZ61 (Olympus China Co. Ltd., China) microscope attached with an Olympus UC30 (Olympus Europa Holding GMBH, Germany) digital camera equipped with an analySIS LS Starter (Olympus Soft Imaging Solutions GMBH, Germany) program. The sex of each insect was recorded. Then all insect specimens were individually preserved in universal bottles filled with 75% ethanol.

Statistical Analyses

The entire pollinator performances were assessed using the modular approach derived by Ne’eman et al. (2010). Open pollination efficiency (O) serves as an estimate of the performance of the whole pollinator assemblages or the degree of pollination limitation. O = o/s where o is the proportion of fruit set in naturally pollinated flowers out of the maximal fruit set production (s) achieved by supplementary cross pollination (hand pollination) of naturally pollinated flowers.

Autonomous selfing efficiency (A) estimates the self-pollination of flowers in the absence of pollinators. A is the proportion of the number of fruit set produced in the absence of pollinators (a) out of the maximal potential fruit set production by supplementary cross pollination (hand pollination) of naturally pollinated flowers (s), therefore, A = a/s. Variations in fruit set produced among treatments were evaluated using the Kruskal–Wallis test at P = 0.05, followed by mean comparisons using Mann–Whitney U-test.

Differences in visitation frequency, amount of pollens carried as well as size difference among flower visitors were analyzed using the non-parametric Kruskal–Wallis test at P = 0.05. Mean comparison of various fly sizes was carried out using Mann–Whitey U-test with Bonferroni correction at P < 0.0083. The Mann–Whitney U-test (at P = 0.05) was also used to examine variation in pollen carrying capacity by male and female pollinators.

The amount of pollen load on each insect (A) was calculated as A = n × B where n is the average number of pollen grains counted inside a small chamber and B is a fraction of suspension retained by each small chamber in relation to the original amount of suspension in the vial. Finally, the Spearman’s correlation analysis was used to investigate the influences of various body size parameters of flower visitors (body length, head length, and head width) on the amount of pollens deposited on their bodies at P = 0.05.

Results

Role of Pollinators in Mango Fruit Set

Observation in the field indicated that mango flowers bloomed all day but peaked between 0900 and 1100 hours. Table 1 shows the abundance of flowers in each Sala and Chok Anan flower panicle. Chok Anan produced 57% more flowers than Sala corresponded with its longer panicle (25–35 cm). High percentage of hermaphrodite flowers was also recorded in Chok Anan (15.7%) compared with Sala (9.24%). The ratio of hermaphrodite to male flowers was lower in Chok Anan (1:5) and slightly high in Sala (1:10).

Table 1.

Number of flowers (mean ± SE) per panicle of Sala and Chok Anan cultivars

Cultivars	Total flowers	Hermaphrodites	Males
Sala	732.7 ± 51.23	67.7 ± 8.03	665.0 ± 49.13
Chok Anan	1,153.2 ± 144.65	181.2 ± 34.32	972.0 ± 119.39

In the first pollination treatment, naturally pollinated flowers of Sala mango produced approximately three fruits in a panicle. More successful pollination was observed in Chok Anan flowers and the number of fruits per panicle doubled that of Sala (six fruits). About 95.2 and 96.9% of hermaphrodite flowers of Sala and Chok Anan, respectively, dropped off without fruit set. In the second treatment, pollen supplement for naturally pollinated flowers had significantly improved the number of fruit set for Chok Anan (Fig. 1) but was not statistically significant (P > 0.05) for Sala compared with other treatments. Chok Anan failed to have any fruit set in pollinator exclusion treatment emphasizing the importance of insect pollination for this cultivar.

Fig. 1.

Pollination treatment effects on fruit set of two mango cultivars (Sala and Chok Anan).

Selfing efficiency was significantly higher in Sala (47.2%) than in Chok Anan (11.9%) (Mann–Whitney U = 89.00, z = −3.608, P = 0.02). In contrast, pollinator efficiency was higher in Chok Anan (71.2%) but lack of statistical significance (P > 0.05) between cultivars. High rate of flower drop that resulted in very low number of fruit set in Chok Anan had further confirmed that this cultivar strongly depended on insect pollinators for maximum fruit production. In general, total pollinator performances for both mango cultivars as indicated by fruit set production in naturally pollinated flowers, contributed about 53% of total estimated fruit set production (achievable through supplementary cross pollination).

Visitation Frequency

The visitation frequencies of five pollinator genera in two mango cultivars are listed in Table 2. In general, Chrysomya visited more flowers compared with other insects. Sarcophaga on the other hand, visited the least number of flowers (approximately six flowers) within a 2-min observation period. Most of the insects seemed to spend more time on Chok Anan flower compared with Sala flower as shown by low visitation frequency of all insects on Chok Anan panicle. Meanwhile, both ant genera, Camponotus and Iridonyrmex, visited approximately nine flowers within a 2-min period. However, visitation frequency was not significantly different among all insects observed in both cultivars (Kruskall–Wallis, P > 0.05).

Table 2.

Visitation frequencies (mean ± SE) per 2 min observation of common flower visitors in Sala and Chok Anan flower panicles

Insects	Cultivars		Total
	Sala	Chok Anan
Iridomyrmex	12.71 ± 5.190	5.14 ± 0.829	8.93 ± 2.735
Camponotus	10.62 ± 1.917	7.69 ± 1.157	9.15 ± 1.135
Stomorhina	8.93 ± 2.295	7.64 ± 1.574	8.36 ± 1.439
Chrysomya	17.00 ± 4.824	9.50 ± 3.069	14.00 ± 3.242
Sarcophaga	8.50 ± 1.893	4.38 ± 0.944	6.14 ± 1.089

Pollinator Behavior

We observed that flies and ants displayed different visitation behavior on mango flowers. Sarcophaga demonstrated consistent grooming behavior after each visit and this activity was commonly performed between small branches of flower panicle. Most of the flies usually landed on the flowers with their abdomens directly in contact with the stigmas or anthers and the legs grabbed the oblong petals. Ants Camponotus and Iridomyrmex were continuously moving on sepals and petals of each flower and avoiding the middle part of the flower that had a 5-lobed nectary. This nectary lobe is a secretory structure that releases nectar of high sugar content at the base of stamen, anther, style, and stigma.

Deposition of Pollen Grains as Estimator of Pollinator Effectiveness

A simple pollen atlas was produced to compare the general shape and size of mango pollen (Fig. 2) with pollens of various weed species (Fig. 3) commonly found around the orchard. In moist condition, medium size (∼30 μm) mango pollen was spherical in shape. Its outer membrane (exine) became swollen particularly at areas having a few small furrows or pores. Weed pollens such as Asystasia intrusa and Passiflora foetida were very large (>40 μm) or very small (<20 μm) (e.g. Gomphrena serrata, Cleome rutidosperma, and Mimosa pudica) compared with the mango pollen. Some weed pollens such as Paspalum conjugatum and Cyperus kylingia had smooth surface but most of the Compositae weeds had echinate exine (e.g. Tridax procumbens, Emilia sonchifolia, Synedrella nudiflora, and Vernonia cinerea).

Fig. 2.

Mango pollen (Magnification 40×).

Fig. 3.

A pollen atlas of various weed species in the mango orchard. a, Ageratum conyzoides; b, A. intrusa; c, Borreria laevicaulis; d, Cy. kylingia; e, Echinochloa colona; f, E. sonchifolia; g, Euphorbia hirta; h, G. serrata; i, M. pudica; j, P. foetida; k, T. procumbens; l, Vernonia cinerea; m, Choromolaena odorata; n, Cl. rutidosperma; o, Pas. conjugatum; p, S. nudiflora (magnification 40×).

Variations in pollen carrying capacity were observed among different pollinator genera (Table 3) and were statistically different at P = 0.00, H(6) = 111.082. Comparing the dipteran pollinators, Eristalinus had the highest number of pollen on its body followed by Stomorhina and Chrysomya. However, Eristalinus was not a common visitor on mango panicles in this orchard and only three individuals were collected. The result also showed that a fast flying fly, Sarcophaga, only carried a small amount of pollen. The carpenter ant, Camponotus, carried very few pollens on its body compared with the dipterans while Iridomyrmex ant, did not carry any pollen.

Table 3.

Estimated pollen abundance (mean ± SE) on ♀ = female, ♂ = male insects visiting mango flowers

Genera	Sex	Mean ± SE	Number of insects	P
Stomorhina	♀	4888.89 ± 1639.091	9	0.472
	♂	1414.14 ± 421.257	11	0.474
	Total	2977.78 ± 787.874	20	0.487
Eristalinus	♀	14,222.22 ± 3661.987	2	0.437
	♂	12,000.00	1	–
	Total	13,481.48 ± 2514.781	3	0.450
Chrysomya	♀	2370.37 ± 701.612	3	0.450
	♂	2555.56 ± 810.506	4	0.457
	Total	2476.19 ± 544.875	7	0.468
Sarcophaga	♀	1,629.63 ± 458.337	3	0.450
	Total	1,629.63 ± 458.337	3	0.450
Camponotus	♀	622.22 ± 176.086	20	0.481
	Total	622.22 ± 176.086	20	0.481
Iridomyrmex	♀	–	10	–
	Total	–	10	–
Specodes*	♀	14,222.22	1	–
	Total	14,222.22	1	–

*Contributed by single individual. This genus was used as a reference for a wild bee.

Only Stomorhina females had significantly more pollen on their bodies compared with the males (U = 935.00, z = −2.394, P = 0.017) (Table 3). In general, less pollen were carried by most female anthophiles (2314.81 ± 417.391) compared with the males (2361.11 ± 469.983) although there was no significance difference in pollen deposition by the two sexes (U = 8597.00, z = −1.681, P = 0.093).

Estimated number of pollen on insects visiting Chok Anan flowers (2614.38 ± 421.724) was higher than those visiting Sala (2000.00 ± 528.786) but the difference was not statistically significant (U = 11,520.50, z = −1.788, P = 0.074). However, Stomorhina and Chrysomya carried more pollen from Sala panicles. There was no difference in the amount of pollen on Camponotus from both Sala and Chok Anan panicles (Table 4).

Table 4.

Estimated pollen abundance (mean ± SE) on insects visiting Chok Anan and Sala flowers

Genera	Mean ± SE		P
	Chok Anan	Sala
Stomorhina	2017.09 ± 445.839	4761.90 ± 2079.732	0.951
Eristalinus	16,000.00 ± 3,471.222	8444.44	0.254
Chrysomya	2111.11 ± 829.523	2962.96 ± 638.438	0.122
Sarcophaga	1777.78 ± 831.479	1555.56 ± 578.537	0.859
Camponotus	622.22 ± 201.181	622.22 ± 291.234	0.599
Iridomyrmex	–	–	–
Specodes*	14,222.22	–	–

*Contributed by single individual. This specimen only used as a reference for a wild bee.

Influence of Body Size of Pollinators on Pollen Carrying Capacity

Variations in body size parameters; head width, head length, and body length were significantly different among seven genera of insects at P = 0.00 (Hbody length = 170.669, Hhead length = 128.104 and Hhead width = 167.499) (Table 5). Among them, Iridomyrmex was the smallest, followed by Stomorhina, and Specodes and the others were relatively large. Significant variations in some body parts were also observed between the sexes (P < 0.05). The males of Stomorhina, Eristalinus, and Chrysomya had bigger heads, Uhead width = 1172.50, z = −6.936, and Uhead length = 2418.50, z = −3.112 than the females but no sex difference in body length was detected.

Table 5.

Measurements of body sizes (mean ± SE); body length, head width, head length of ♀ = female, ♂ = male flower visitors

Genera	Sexes	Number of insects	Body length (mm)	Head width (mm)	Head length (mm)
Stomorhina	♀	9	5.03 ± 0.089	0.73 ± 0.052	1.91 ± 0.044
	♂	11	5.07 ± 0.104	1.11 ± 0.049	2.15 ± 0.028
	Total	20	5.52 ± 0.074	0.94 ± 0.043	2.04 ± 0.029
Eristalinus	♀	2	8.61 ± 0.357	1.87 ± 0.320	3.23 ± 0.199
	♂	1	9.49	2.58	3.68
	Total	3	8.91 ± 0.273	2.11 ± 0.238	3.38 ± 0.148
Chrysomya	♀	3	8.29 ± 0.173	0.92 ± 0.078	3.47 ± 0.089
	♂	4	8.08 ± 0.027	1.20 ± 0.147	3.47 ± 0.089
	Total	7	8.17 ± 0.077	1.08 ± 0.094	3.35 ± 0.068
Sarcophaga	♀	3	9.75 ± 0.008	1.35 ± 0.133	3.58 ± 0.015
Camponotus	♀	20	7.70 ± 0.431	1.96 ± 0.033	1.29 ± 0.036
Iridomyrmex	♀	10	3.89 ± 0.055	1.28 ± 0.012	0.89 ± 0.005
Specodes*	♀	1	6.15	1.40	1.91

*Contributed by single individual. This specimen only used as a reference for a wild bee.

In general, more pollen was found on bodies of large pollinators compared with small pollinators (Tables 4 and 5). Long bodied (ρ = 0.305) and big headed (ρ = 0.615) pollinators carried more pollen (P = 0.01) but pollinators with longer heads had less pollens on them (ρ = −0.235, P = 0.01). Among the flies, the body sizes varied from relatively small to large (Fig. 4). Stomorhina was the smallest fly and Chrysomya was the next smaller fly. Except for its head length, the size of Chrysomya was comparable to Eristalinus while Sarcophaga was the biggest fly in the orchard. However, small size Stomorhina carried more pollens than slightly bigger Chrysomya and the largest Sarcophaga but much lesser than relatively big size Eristalinus.

Fig. 4.

Mean values of body sizes (body length, head length, and head width) of four genera of Diptera; Stomorhina, Eristalinus, Chrysomya, and Sarcophaga. Small letters indicated significant differences of mean at P < 0.0083.

Discussion

A mango tree produces masses of flowers as a survival mechanism to ensure successful production of seeds for perpetuation of the species. Hermaphrodite and male mango flowers are borne within a single inflorescence and their ratio is mainly cultivar related (Usman et al. 2001). Environmental and cultural factors may contribute to variations in flower components to a certain extent (Bally et al. 2009). In various cultivars, flowers reach full bloom in 25–30 days after initiation. Differences in flower or fruit abortion may lead to large variation in yields among mango cultivars.

Nor Hazlina et al. (2008) reported that the ratio of perfect flower to the male mango flower was as high as 1–75. However, in this study a much lower ratio of 1:5 was recorded for Chok Anan and 1:10 for Sala. In spite of this, low fruit set and high dropping off of perfect flowers in both cultivars hampered high mango production in this orchard.

Based on the number of perfect flowers, each mango panicle can bear up to 500 fruits. Although extremely low percentage (0.1–0.25%) of hermaphrodite flowers developed into fruits in some self-pollinated mango cultivars (Chapman 1964, Jabatan Pertanian Malaysia 2009), more flowers were fertilized in this study and higher percentages of fruit set by naturally pollinated Chok Anan (3.1%) and Sala (4.8%) flowers were recorded. Other flowers were unfertilized, shed off, or failed to set fruits. Progressive thinning of fruits which occurs during their development to maturity (Bally et al. 2009) further reduces high production potential of Malaysian mango (Azhar and Ithnin 2009).

Chok Anan flowers failed to produce fruit set in the covered treatment and hence were strongly pollinator dependent. In the presence of pollinators, Chok Anan is capable to bear fruits in the rainy season which is unusual for other mango clones (Nor Hazlina et al. 2008). The result of this study also showed that Chok Anan has a high pollinator efficiency of 0.711. Therefore, increase in pollinator abundance would certainly amplify its productivity.

Hand pollination has effectively fertilized a considerable amount of flowers. Azhar and Ithnin (2009) successfully produced 26.7% fruit set from 1000 Chok Anan flowers that they hand pollinated. In this study, more than 100% increase in fruit set was recorded for Sala compared with naturally pollinated flowers. In Chok Anan, fruit set production increased about 33%. Eardley et al. (2006) and Ya et al. (2004) also reported large increase in yields of hand pollinated oil palm and pear respectively.

Although a lot of hermaphrodite flowers were hand pollinated in the second treatment, the resultant fruit set were far smaller than the number of pollinated flowers. Unsuccessful delivery of male gamete to the ovule might be the case in Chok Anan because Ding and Khairul Bariah (2013) have discovered that the style of its hermaphrodite flower was longer than the filament of the stamen. Physiological incompatibility has been the main problem for hand pollinated flowers. Mango flowers are usually self-compatible but varying degree of self incompatibility may result from environmental changes. Previously Mukherjee et al. (1968) and Singh et al. (1962) also discovered self-incompatibility in Indian mango. Other mango cultivars like Dashehari, Langra, and Chausa are also found to be self-incompatible (Sharma and Singh 1970).

Apart from poor fruit set, hand pollinating mango flowers is a difficult task because of very small flower size. The flower stigma is easily damaged during the pollen transfer despite exercising extra care. In addition, heavy rain washes off pollen and thus prevents fruit set for most crops (Nakasone and Paull 1998). Coincidently, our study orchard is located in the Northern Peninsular Malaysia which receives very little rain between December and March (Mohtar et al. 2014).

The results of our study emphasized the need for supplementary pollination for a better mango fruit production Malaysian orchard. Our findings also highlighted the importance of various wild pollinators that contributed ∼53% of estimated maximum fruit production. Therefore, there was an urgent need for pollination service to ascertain better fruit production particularly for Chok Anan and Sala mangoes in Malaysia (Usman et al. 2001, Sung et al. 2006). Furthermore, several reports have confirmed increased fruit/seed set and better quality fruits produced by crops that receive multiple pollinator visits (Free 1993, FAO 2008).

Usually insect pollinators have different foraging behavior that may influence pollinator efficiency (Ne’eman et al. 2010). Among the flower visitors observed in this study, Chrysomya displayed higher visitation frequencies and visited more flowers within the allotted time. High visitation frequency may increase the chances of pollen delivery thus increases the chance that a flower matures into a fruit (Mitchell and Waser 1992). In contrary, Mayfield et al. (2001) suggested that pollinator with lesser frequency of visit deposited more pollen on the stigmas. Likewise insects with slow foraging rates (therefore low visitation frequency) can ensure more pollen transfer than those with high foraging rates (Ivey et al. 2003). Longer foraging times may translate into greater pollen removal and more pollen deposition on stigmas (Horsburgh et al. 2011). Two fly genera, Stomorhina and Sarcophaga, certainly had low visitation frequencies on mango flowers thus they were valuable pollinators. However, Sarcophaga flies spent more time grooming their bodies on flower branches instead of spending more time on flowers.

Ants, particularly Camponotus and Iridomyrmex, were also active flower visitors in this orchard despite no record of ant pollinating crops in Malaysia (Bakhtiar and Maryati 2009). From observation on their behavior, ants contributed very little to pollination of mango flowers. They moved actively on mango panicles to search for honey dew secreted by homopteran pests instead of flower nectar. In this study, Camponotus and Iridomyrmex ants had relatively high visitation frequency. Gomez et al. (1996) suggested that the role of ants as pollinators depends heavily on their high relative abundance. Ant pollination becomes evident when they outnumber other floral visitors. Unfortunately, ants are poor pollinators and can also disrupt pollination by deterring other flower visitors, or by stealing nectar (Corlett 2004, Ballantyne and Willmer 2012).

Pollen carrying capacity varies among pollinator species. In this study, a small parasitic sweet bee, Specodes (Halictidae) had the highest pollen load on its body. Bees are known to be effective pollinators compared with other insect pollinators. They spent relatively long time on the flowers and their hairy bodies make excellent contact with the female flower parts (Mayfield et al. 2001). Comparing with dipteran pollinators, bees, and many other hymenopterans carried a significantly higher proportion of pollen on their bodies (Howlett et al. 2011, Rader et al, Abrol 2012).

Among the dipteran pollinators, Eristalinus carried the highest amount of pollen on its body. Another fly, a small size Stomorhina, also carried relatively large amount of pollen and was frequently observed on mango flowers. Stomorhina is an active pollinator and besides mango it has been reported to pollinate other plants including Tectona grandis in Thailand and tuckeroo (Cupaniopsis anacardioides: Sapindaceae) in Australia (Hawkeswood 1983, Tangmitcharoen and Owens 1997). Chrysomya (fly) also displayed high capability to carry mango pollen on its body suggesting its potential as a good mango pollinator. One of its species, C. megacephala carried more avocado pollen than other pollinators in Mexico (Perez-Balam et al. 2012). We found that the sex of the pollinator was not a contributing factor to pollen carrying capacity. Although Stomorhina and Eristalinus females had slightly more pollen on their bodies, the sex difference was not statistically significant. Borkent and Schlinger (2008) also reported no variation in pollen carrying capacity between male and female Eulonchus tristis (Diptera: Acroderidae).

Our study confirmed a considerable involvement of Camponotus ant in mango pollination. This ant also pollinates Mediterranean shrubs Retama sphaerocarpa, Frankenia thymifolia (Gomez et al. 1996), Jatropha curcas (Luo et al. 2012), and plants in the family Euphorbiaceae (Reddi and Reddi 1984, Schurch et al. 2000). Unlike Camponotus ant, we found that Iridomyrmex ant did not carry any mango pollen on its body. A hard-bodied Iridomyrmex may have difficulty to come in contact with flower anthers thus reduced its chances to contact with the pollens (Armstrong 1979). The lack of bristle on its smooth integument makes it unsuitable to carry mango pollen (Corlett 2004). However, Iridomyrmex anceps was reported by Luo et al. (2012) to be a pollinator of J. curcas in China.

Influenced of Body Size of Pollinator on Pollen Carrying Capacities

According to Pearce et al. (2012) and Luo et al, large pollen load was common in large, long, and more robust insect bodies. O’Neill and O’Neill (2010) also reported that the body size of female leaf cutting bee, Megachile rotundata (Hymenoptera: Megachilidae), was positively correlated with the pollen loads it carries. Paralleled to the previous findings, the result of this study showed that a pollinator with big head (head width) and long body carried more pollen than a small-headed and short-bodied pollinator. A relatively big Eristalinus (Syrphidae: Diptera) collected the highest number of pollen and a large Camponotus ant also carried a lot of mango pollen.

The behavior of the pollinators may also influence their pollen carrying capacity. We observed that Sarcophaga flies actively groomed its body after each flower visit. According to Castellanos et al. (2003), intensive grooming is a way to regulate the amount of pollens from a donor flower to be deposited onto a receiver flower. The Specodes bees typically feed on pollen and nectar thus collect high amount of pollen on their bodies (Abrol 2012). In addition to feeding behavior, pollination efficiency was also influenced by anatomical features of the pollinators (Larson et al. 2001).

As a conclusion, this study highlighted the importance of pollinators in Malaysian mango industry. Self-pollinated and naturally pollinated flowers resulted in very low fruit yield. Some of the mango cultivars such as Chok Anan were completely dependent on insect pollination. Enhancing population of wild pollinators such as Eristalinus, Chrysomya, Stomorhina, Sarcophaga, and Camponotus especially those with big size and hairy body may result in improve pollination service in this mango orchard.