Materials and Methods

3.1 Study Area

Tannirkuli Village (14.45876 ºN and 74.42694 ºE), in the Hegde Panchayat of Kumta Taluk of Uttara Kannada in the South Indian west coast was chosen for the study. Tannirkuli is dominated by Halakkivokkal community farmers who are skilled growers of vegetables using traditional agricultural techniques. During the intense coastal monsoon, the farmers mainly grow cucurbits in a 10 ha well-drained ground, while rice is cultivated in the flooded fields. Groundnut, vegetables and cucurbits constitute the second crop in the post-monsoon fields when the hill slopes are too dry for cultivation (Fig. 1).

Figure 1 Map of study area: Tannirkuli Village, Uttara Kannada District, Karnataka State, India

3.2 Sampling Design

The preliminary studies were carried out from July to early August 2013. Observations on insect visitations on flowers were conducted from mid-August to the end of September. The crops studied were cucumber, bitter gourd, snake gourd, ridge gourd and Mangalore gourd (Fig. 2). The observations on pollination were started at 10% flowering and continued at weekly intervals until flowering nearly stopped (Belavadi & Ganeshaiah 2013). Numbers of visitor insects were counted species-wise. The frequency of visitation by a species was categorized as very frequent (>50% of the floral visits); frequent (>20% and <50%); and rare (<20%). Insect specimens were collected using a sweep-net and identified using taxonomic keys (Michener 2007; Belavadi & Ganeshaiah 2013). The specimens were examined using stereo microscope for the presence of pollen and segregated as pollinators and non-pollinators. The pollen grains were matched with the host pollen separately collected before ascertaining the visitor insect as a pollinator of a specified crop, in accordance with Belavadi and Ganeshaiah (2013). The insect specimens were maintained in the Kumta field station of Centre for Ecological Sciences, Indian Institute of Science.

Figure 2 Flower and fruit of cucurbits plants.

a & b = male flower of Cucumis sativus, c & d = female flower of Cucumis sativus; e & f = male flower of Momordica charantia, g & h = female flower of Momordica charantia; i = male flower of Luffa acutangula, j = female flower of Luffa acutangula; k = Delias eucharis on Trichosanthes anguina; l = fruit of Cucumis pubescens. m = fruit of Trichosanthes anguina; n = fruit of Luffa acutangula; o = fruit of Momordica charantia; and p = fruit of Cucumis sativus.

3.3 Foraging Dynamics

Insect foraging behavior was studied between 06:00 and 22:30 hours, at hourly intervals during species-specific blooming times. The observation protocols and frequency of visitors were measured using a hand tally counter and stopwatch following Free (1993). Sampling units of 1 m2 (approximately covering 20 - 30 flowers) were selected and the visitor insects were documented. Each observation period for any crop selected was for 5 minutes, at 3 replications per hour. The foraging patterns were analyzed to determine the dominant groups of visitor insects and peak foraging times.

3.4 Landscape Analysis

Landscape elements of 3 km radius around the study area were deciphered using remote sensing data from Google Earth and Landsat-8 imageries of 2014. The time series spatial data acquired from Landsat Series Multispectral sensor and thematic mapper sensors were downloaded from http://glcf.umiacs.umd.edu/data. Land Use (LU) analysis involved Preprocessing, Classification and Accuracy Assessment. Land use classification was done using supervised pattern classifier-Gaussian maximum likelihood algorithm through open source GIS i.e. GRASS-Geographic Resource Analysis Support System downloaded from http://ces.iisc.ernet.in/grass (Fig. 3). The classifications were based on derived signatures (training polygons). Classifier performance was assessed considering reference pixels using kappa (κ) statistics (Ramachandra et al. 2013).

Figure 3 Method followed for land use assessment