Native and invasive ants affect floral visits of pollinating honey bees in pumpkin flowers (Cucurbita maxima)

Global pollinator decline is a major concern. Several factors—climate change, land-use change, the reduction of flowers, pesticide use, and invasive species—have been suggested as the reasons. Despite being a potential reason, the effect of ants on flowers received less attention. The consequences of ants being attracted to nectar sources in plants vary depending upon factors like the nectar source's position, ants' identity, and other mutualists interacting with the plants. We studied the interaction between flower-visiting ants and pollinators in Cucurbita maxima and compared the competition exerted by native and invasive ants on its pollinators to examine the hypothesis that the invasive ants exacerbate more interference competition to pollinators than the native ants. We assessed the pollinator's choice, visitation rate, and time spent/visit on the flowers. Regardless of species and nativity, ants negatively influenced all the pollinator visitation traits, such as visitation rate and duration spent on flowers. The invasive ants exerted a higher interference competition on the pollinators than the native ants did. Despite performing pollination in flowers with generalist pollination syndrome, ants can threaten plant-pollinator mutualism in specialist plants like monoecious plants. A better understanding of factors influencing pollination will help in implementing better management practices.


Scientific Reports
| (2021) 11:4781 | https://doi.org/10.1038/s41598-021-83902-w www.nature.com/scientificreports/ primarily engaged in foraging floral nectar, a few also have attempted to prey upon pollinators 18,20,[27][28][29] . Although some studies suggest that the invasive ants protect some plants against herbivores, it is likely that their aggressiveness wards off some useful mutualists, too 29,35 . Although a few studies examined the effect of invasive ants on pollinators, a comparative study between native and invasive ants has not been done yet 12 . In this study, we have attempted to compare the effects of ants on pollinators and compare the competition exerted by non-native and native ants on the pollinators of Cucurbita maxima-a monoecious globally-important, pollination-deficient crop plant.
To understand the effects of ants on pollinators, we took three attributes of pollinators and observed how these were affected by flower-visiting ants (hereafter, floral ants). They were the pollinators' choice to visit or not to visit the flowers, the frequency of pollinators' visits (visitation rate), and the time the pollinators spent inside the flowers. We expected the visitation rate and visit duration of pollinators to be negatively affected by ants' presence. Since invasive ants are known for displacing biodiversity and interrupting species interactions, we hypothesize that the invasive ants exert more pressure on plant-pollinator interaction than the native ants do. We considered the ant's ability to occupy the maximum number of flowers and maximize the number of individuals occupying a single flower as the parameters for assessing ants' exploitation competition on pollinators. Our expectation for this was that the flowers occupied by invasive ants in the population are higher than the flowers occupied by native ants. Since the plant in our study is monoecious, we also asked whether the ant-pollinator interaction was affected by flower sex.

Material and methods
Study site. Our study was conducted at Kasaragod, the northernmost district of Kerala state in the peninsular India. The district is located at 12.5° N 75.0° E with an average elevation of 19 m above sea level. The highest and lowest temperature is 37 °C and 17 °C respectively. The district enjoys a tropical and sub-tropical climate, with winter (January and February), hot summer (March-May), and two rotations of Monsoon, the Southwest Monsoon (June-September) and Northeast Monsoon (October to November), giving an average annual rainfall of approx. 2000 mm. The district exhibits a topographic division into three, the lowland bordering the sea, the mainlands, and the forest highlands on the eastern side. The coastal strips are sandy, the lowland has red ferruginous loam with a mixture of clay and sandy soil, and the highland region is laterite.
The study took place in seven independent farmer-managed vegetable gardens in Kasaragod. These sites stood 3-10 km apart. The farmers of these sites cultivate rice paddy as their main crop, sowing in June to August and harvesting in November to January months of the year. From November to April, they cultivate vegetables belong to Amaranthaceae, Cucurbitaceae, Malvaceae, Solanaceae, and Fabaceae families as smallholdings (approximately one-acre land).
Study system. We took pumpkin (Cucurbita maxima) as our study system. It is a monoecious plant with separate staminate (male) and pistillate (female) flowers. Both flowers are borne on different nodes with a disproportionately high number of staminate flowers per field 37 . The staminate and pistillate flowers are large, showy yellow with abundant floral nectar-the pistillate flowers producing a slightly higher quantity of quality nectar 38 . We harvested nectar from pre-caged staminate (N = 14; 2 per site) and pistillate flowers (N = 14; 2 per site) soon after anthesis using Drummond micro-capillary tubes (10 μL) and estimated the sugar concentration using a hand-held refractometer.
A large ovary at the flower base and three stigma lobes inside the flower distinguishes the pistillate flower from the staminate flower. The staminate flower has its anthers united into a single long filament of about 10 cm. The pollen grains are large and sticky; thus, pollination is dependent on the animal pollinator. Flower longevity is one day. The flower opened between 0630 and 0700 h and remained open for about 6 h. Native honey bee, Apis cerana, is the primary pollinator of C. maxima in our study area 37 . Farmers raised plants from the seeds collected from their previous years, with "Ambili" being the predominant variety in field populations. Four to six seeds were sowed in a pit of about 60 cm diameter and 30-45 cm depth, mixing it with neem cake and cow dung. The field is irrigated daily morning using the traditional pot method or water hose 38 . To observe pollinator and ant activity, we spent 15 min per flower during the early hours of the anthesis and pollinators' peak activity (0700-0800 h). We observed the flowers in which ants were present and the flowers in which ants were absent in the fields. During this observation window, we assessed the visitation rate, the duration of the visit, and pollinators' hovering activity. We observed the visitation rate as the number of times the pollinator entered the flower. We recorded the time of entry and exit of a pollinator, the duration the pollinator spent inside the flower in a single visit. This time was recorded using a stopwatch. In cases, when a large number of ants colonised the flower, the pollinators did not enter the reproductive part of the flower; the pollinator rather hovered around the flower or landed on the tip of petals for very less time (< 3 s). Such an activity of the pollinators was recorded as 'hovering' and not as a visit. In flowers occupied by ants, we collected the ants in 5 mL vials with ethanol and identified them later from the lab. Based on the identification and the invasive ant list 39  Data analysis. First, we examined the preference of pollinators and ants to flower sexes for foraging resources. The visitation rate of pollinators on ant-unoccupied staminate and pistillate flowers was compared to understand pollinators' general preference for flowers for foraging floral resources using a Generalized Linear Mixed Effect Model (GLMM). In the model, pollinators' visitation rate was fitted as the dependent variable, flower sex was fitted as the fixed variable, flower ID nested in plant ID and sites was fitted as a random factor, and negative binomial distribution as an error type. The overall ants' abundance on staminate and pistillate flowers was compared using another GLMM. We repeated this for native ants and invasive ants, separately. In the models, respective ants' abundance was fitted as response variable, flower type as fixed effect, flower ID nested in plant ID and sites as random factors and negative binomial distribution as the error type. We took three main attributes of pollinators-flower choice (on a binary scale; ant-occupied flowers or antunoccupied flowers), visitation rate and visit duration-to study the effects ants have on flowers. The pollinators' choice of visit or not visit the ant-occupied and ant-unoccupied flowers was tested using a generalized linear mixed model with binomial distribution as the error type, visit-no visit data of pollinators on flowers (binary response) as the dependent variable, flower type as the fixed factor, flower ID nested in plant ID and sites as random factors and logit as a link function. We studied the effect of flower type (ant-unoccupied and ant-occupied) on visitation rate and the time the pollinators spent in flowers using two generalized linear models with negative binomial distribution as an error type, flower type and flower sex (staminate and pistillate) as the fixed effects and visitation rate and time spent by the bees as the response variables. We studied the interactive effects of ant type (invasive/native), number of ants and flower sex on a) visitation rate and b) the time the pollinators spent on ant-occupied flowers using two GLMMs. In the models, the ant type, number of ants on flowers, and flower sex were the fixed factors, visitation rate and the time the pollinators spent on flowers were the response variables, flower ID nested in plants and sites were the random factors and negative binomial distribution as the error type. We also examined the effect of ant species on visitation rate and time the pollinators spent on flowers by constructing two other GLMMs. These overall models' significance was tested using Wald's Chi-square test available in the R-package car. In the results, mean ± standard deviation and estimate ± standard error are given throughout. All analyses were carried out using R version 3.5.2 40 .

Results
The pistillate flowers produced an average of 5.2 μL (± 0.09) nectar and the staminate flowers produced an average of 2.8 μL (± 0.08) nectar immediately following the anthesis, which continued and maintained through the entire anthesis period. In non-harvested flowers, the accumulated nectar volume can go up to about 40 μL in pistillate flowers and 28 μL in staminate flowers (Sinu, personal observation). The quality of nectar produced by staminate flowers averaged 29.4% (± 1.05) sugar concentration by weight and 31.1% (± 2%) in pistillate flowers.
The native honey bee, Apis cerana, was the primary pollinator of pumpkin flowers (92% of the total sample). In 8% of the flowers, another native bee species, Apis dorsata, also visited the flowers along with A. cerana. Since A. dorsata was a minority, we used pooled data of visitation characteristics of both the species in the analyses after realizing that the response of both the species of bees was the same towards the ants.
Nine species of ants visited the flowers of pumpkin as soon as the anthesis had occurred (Fig. 1). Four were non-native-Anoplolepis gracilipes (Smith, F.  2) on staminate flowers was lesser than on pistillate flowers, but the difference was marginally or not significant. The pollinators preferred to visit ant-unoccupied flowers (75%) over ant-occupied flowers (62%; 1.94 ± 0.6, z = 3.25, p = 0.001). The visitation rate of pollinators on ant-unoccupied flowers (4.14 ± 5.6) was significantly higher than on the ant-occupied flowers (2.4 ± 3.08; 0.50 ± 0.09, z = 5.35, p < 0.00005) (Fig. 2). On ant flowers, the visitation rate of pollinators to the native ants-(2.43 ± 2.29) and invasive ants-occupied flowers (2.38 ± 3.34) was same (0.41 ± 0.25, z = 1.62, p = 0.10), but decreased on native ants-occupied flowers when their numbers increased on the flowers (− 0.11 ± 0.04, z = − 2.41, p = 0.01). However, the visitation rate of pollinators to the flowers varied with the ant species occupied on the flowers (Wald's χ 2 = 18.75, df = 4, p = 0.0008) (Tables 1 and2). The pollinator visits also decreased with the number of ants occupied on flowers (− 0.04 ± 0.003, z = − 4.87, p < 0.00005). The response (visitation rate) of pollinators to the number of ants and ant type was not driven by flower sex (three-way interaction model: Wald's χ 2 = 0.93, df = 1, p = 0.33). It suggests that the pollinators, despite preferred pistillate flowers for foraging, absconded when the number of ants, though belonging to the native species group, was high on flowers ( Table 2).
There are a handful of studies exploring the impacts of invasive ants on pollinators, however, a comparison of invasive and native ants concerning their interaction with floral visitors are scarce 12 .
In this study, we attempted to explore the effects of ants on pollinators and compare these effects in cases of invasive and native ants. The study found that ants, regardless of their nativity, gave a competitive pressure for honeybees-the sole pollinators-in pumpkin flowers. The visitation rate and the duration they spent foraging floral resources were significantly low for ant-occupied flowers. Although the bees tend to spend more time on native-ant-occupied flowers, when the number of ants is high, avoid such flowers.
With its showy flowers with a reasonable amount of floral rewards in the form of both pollen and nectar, pumpkin is a noteworthy lure for both ants and bees. We found that ants occupied more than 75% of the flowers during the early hours of anthesis in different fields. The ratio of staminate flowers to pistillate flowers counted in our study agrees that the number of staminate flowers is mostly higher in a field 37,41,42 . The ants preferred pistillate flowers over staminate flowers for foraging nectar, although they were less in field populations. They even tended to be crowded on pistillate flowers than on staminate flowers. The relatively good amount of nectar in pistillate flowers might be explaining the differential level of attraction of ants to flower sexes.
Lach 32 defined interference and exploitation competition in the context of floral ants and pollinators. Exploitation competition is considered dependent on how much a competitor can access and utilize a resource. In our case, we thought the ant's ability to occupy the maximum number of flowers and crowding a single flower as the parameters for assessing the exploitation competition. Interference competition is the efficiency with which one competitor ward off the other.
Ants generally repel pollinators by their aggressive nature 43 . Sometimes, the pollinator might choose not to enter a flower due to ants 20 . The scent of ants negatively influences the choice of pollinator's entry into flower 19 . Our study found that the pollinator entry was significantly lower in the ant-occupied flowers. We also observed several hovering cases, wherein the bee hovered around the flowers or landed on petals but flew off. In such cases, the bees avoided contact with ants and did not touch the flower's reproductive structures, which may be a massive loss for the plant as pollination is likely to be directly affected. The pollination efficiency was also significantly lowered in ant-occupied flowers 20 . Although there were cases of predation attack by yellow crazy ants previously 20 , we found the native ant, Diacamma ceylonense was opportunistically preying upon honey bees in the present study (pers. observ.). The duration of pollinators' visit was also significantly higher in ant-unoccupied flowers over ant-occupied flowers. This indicates direct interference competition that ants offer to the pollinators.
Although bees did not distinguish ants on their nativity while venturing flowers occupied by ants, bothered by the ant species occupied on the flowers. Although our previous study showed that yellow crazy ant could be a major threat to the foraging bees, this study showed that Solenopsis geminata (invasive ant), Tetramorium mixtum and Camponotus variegatus (native ant) could be the worst competitors for foraging bees on flowers. Although other ant species occupied flowers, only five species frequently encountered on the flowers.
There are several cases of ant-pollinated plants. In the Western Ghats, ants were recently reported as a pollinator of Syzigium occidentale inflorescence 26 . In the Atlantic rainforest system, ants are a major pollinator of aggregated inflorescence 44 . Although ant-visited or pollinated flowers are not rare in tropics 18,25,45 , they are mostly bisexual and have either generalist pollination syndrome or have no bee pollinators on them. We found some amount of pollen sticking onto the ants during our observation. However, as other functional traits of pumpkin flowers-monoecy and highly skewed male-female ratio in populations-did not resonate with the characters of ant pollinated flowers 20,46 , ants might not play a major role in the pollination of pumpkin flowers. Additionally, exposing flowers to ant body secretions could deter pollen viability 17 , which, however, is not universal for all ant species 47 . While they feed nectar, ants sometimes damage the pistil base affecting the seed set 38 . Pumpkin is a major floral resource for specialist bees, including honey bees, bumblebees, and squash bees in different parts of the world, which perform cross-pollination in their monoecious flowers. Because they are monoecious, the visitors should travel from staminate to pistillate flowers for effecting pollination. It is doubtful that the ants switch visits between staminate and pistillate flowers in pumpkin fields as staminate flowers are abundant and pistillate www.nature.com/scientificreports/ flowers are not necessarily to be closer to staminate flowers. Although not investigated on all the nine species of ants in this study, Sinu et al. 20 found that none of the A. gracilipes-only visited flowers sets fruits. The efficiency with which ants used up resources determined the exploitation competition given by ants to the pollinators 32 . The main parameters of the ants that we used to measure this were the number of individuals that occupy a single flower and the number of flowers occupied by a single species of ant. Solenopsis geminata was found with the highest number of individuals per single flower. Although not particularly common for S. geminata, the mass recruitment is a key character of invasive ants 36 . Similarly, in general, invasive ants had a higher number of individuals recruited to a single flower than the native ants. The black crazy ant, P. longicornis occupied the highest number of flowers. This implies that they are highly efficient in locating resources. P. longicornis has a complicated modular recruitment system and polydymous nesting 36,48 . This minimizes the distance between the nest and food resource and increases their foraging efficiency. The other globally-important invasive species colonized in pumpkin flowers, A. gracilipes, is also known to make supercolonies and exploit resources quickly 49 . Thus, invasive ants exerted a higher exploitation competition on pollinators than native ants. Both the visitation frequency and the duration of visits decreased with the increasing number of ants. Therefore, even a native ant, when recruited in more numbers to the flowers, can also influence the pollinators' visitation attributes. The interference competition given by invasive and native ants seemed to be similar. Bees seemed not particularly showed any discretion between the scents of an invasive ant occupied and native ant occupied flower.
Our study further confirms the adverse effects the ants have on pollinators. However, such a generalization is challenging as the interactions between ants, plants, other herbivores, and pollinators are quite complicated 18,24 . For instance, Gonzálvez et al. 18 showed that the native ants on flowers defend inefficient pollinators, but unaffected the visits of efficient pollinators. Unlike many previous studies, we compared the competition given by invasive ants and native ants on the pollinators. The invasive ants do most of the exploitive competition as they have their supercolonies and huge population to their advantage. However, we found that any ant-invasive or native-can exhibit interference competition by being aggressive and preying upon the pollinators. But it is not possible to draw a fine border between these two competitions. Information on the effects of ants on pollinators will help manage pollinators in agricultural lands 12,50 .