Plant-pollinator mutualisms are key to sustaining ecosystem function and biodiversity. The study of plant-pollinator networks has conventionally focused on diurnal interactions, while flower-settling moths are among the most diverse yet least understood pollinator groups. Our main objective was to bring to focus the structure and function of a flower-settling moth network based on a previous pollination study in the Florida Sandhill. Specifically, we considered key taxonomic, life history, and functional traits of moths as potential drivers of network structure. We integrated two network types: 1. Flower-visitor, based on direct field observation of moths visiting flowers for nectar; and 2. Pollen transport, inferred from the identification of pollen found on trapped moths. Flower-settling moth networks were diverse and significantly structured (modular). Moth phenology and caterpillar host-plant interactions associated with module formation. Additionally, functionally similar moths were linked more often to particular modules. Notably, the average proboscis length and wingspan of moths varied significantly among modules, thus revealing potential functional niches. Accordingly, we propose three potential functional groups of flower-settling moths: micro, small, and macro - as supported by modes in the distribution of proboscis length and wingspan, and as reflected in modularity analysis. In addition, we recommend the use of wingspan as a potential proxy for the moth functional group. We conclude that flower-settling moth assemblages are more functionally diverse than previously understood and offer a glimmer of hope in the darkness for pollinator conservation.

Dataset for moth-flower interactions was provided from a previous study published by the lead author on the diversity and flower-nectar hosts of settling moths within a sandhill ecosystem. We integrated two network types, based on two methods applied by the latter study to document flower-settling moth interactions: 1. Flower-visitor network - based on individual field observations of moths taking nectar on flowers; and 2. Pollen- transport network - based on the identification of pollen found on individual moths sampled from traps. 

For each interaction, the dataset includes Moth and Flower identification (Family, genus, species), the network type, specimen voucher number, date of interaction, and Functional traits, including flower tube-length and moth wingspan, proboscis length and body size were measured and included for each interaction in the dataset.

Input for both the flower-visitor and pollen-transport networks was structured the same for easy integration. Input included a list of all individual interactions, recorded as the species of moth in the first column, with the flower species in the second column. Networks excluded flower species linked by only a single interaction (Observed or inferred from pollen). Most pollen samples were easily identified to the Family level; however, network analysis only included interactions with pollen identified at the species level. Meanwhile, for a description of moth functional traits, we included all measurements from moths with pollen. 

Measurement of functional traits 

Moth proboscis length was measured using fresh moth specimens (after sampling pollen), secured with a pin in modeling clay, and placed in view under a light microscope. A micro-pin attached to the end of a wooden probe was then used to extend the proboscis for measurement under an ocular scale to the nearest 0.01 mm. After moths were pinned with wings spread, wingspan was measured as the length between the apex of one forewing to the apex of the facing wing, using a caliper or ruler (for larger moths) to the nearest 0.01 mm. Body size guild in moths was indicated by modes in the frequency distribution of wingspan and proboscis length of individual moths. Tube length was measured for five individual flowers per flower species using a ruler to the nearest 0.01 mm.

Field Methods & Pollen Microscopy

Field methods and pollen microscopy were originally conducted by the lead Author. Moth-flower interactions were recorded weekly, during the peak moth flight period (first 2-3 hours after sunset; or ≈17.00 to 24.00 hr, depending on the day length), from 15 July 2010 to 17 November 2010, totaling 9 nights of sampling and observation sessions. Beginning just after sunset, moths were sampled from a light trap, while observation sessions were conducted within a 25 m radius of the light trap. Sampling and observation sessions continued until moth activity on flowers and at traps stopped, typically 2-3 hours after sunset.

1.     Field Observations Direct field observation sessions of moths taking nectar on flowers were conducted using a 1-watt LED headlamp set on red illumination (white light distracts the moths). Focusing on one flowering plant at a time, flowers were scanned, and flower-settling moth interactions were recorded (15-20 minutes). This continued until all flowering plants within the plot were scanned (≈ 1-2 hours). For each moth observed on a flower, nectar feeding behavior was recorded for 10-20 seconds, and included photographic or video documentation of moths taking nectar on flowers before the moth was sampled. To prevent damage to the flower and pollen contamination, moths were collected off flowers by holding a clean vial directly above the moth, allowing the moth to fly up into the vial.

2.     Trapping moths. Moths were sampled manually from a light trap composed of one 40-watt ultraviolet light and one 70-watt mercury vapor lamp suspended on each side of a vertical white sheet (approximately 2.0 × 2.4 m). All moths were collected individually in sanitized vials or envelopes to prevent contamination (moth scales and pollen). Samples were then placed in a small cooler while in the field to sedate the moths and stored in a deep ultracold (-80˚C) freezer until ready for processing and identification.

Pollen sampling and preservation. All moths sampled at light traps were scanned for pollen and subsequently included for pollen microscopy. Pollen was stained and preserved in glycerin gel on a microscope slide with a cover slip. Pollen samples were visualized and measured using a reflective microscope 40 × (objective) and 1.25 × (optivar), and imaging software was used to capture an image of pollen. Pollen characteristics such as shape, number, and length (μm) of spines, diameter (μm), and number of furrows or pores were considered for identification.

Pollen Reference Library A pollen reference library was prepared from all flowering plants recorded in bloom during the course of the study and used as a reference for pollen identification. All plants were photographed, collected, pressed, and identified, and pollen was sampled from flowers. Photographs were taken of pollen and used as reference images for the identification of pollen on moths. When pollen could not be matched with the reference collection, local pollen keys and manuals were also used to identify pollen (Kapp et al. 2000, Jelks 2001). Pollen samples from moths and the pollen reference library were processed and deposited at the Palynology Pollen Library, Florida Museum of Natural History (Gainesville). Plant vouchers were deposited with the moth reference collection at the McGuire Center for Lepidoptera & Biodiversity.