DNA barcode for the identification of the sand fly Lutzomyia longipalpis plant feeding preferences in a tropical urban environment

Little is known about the feeding behavior of hematophagous insects that require plant sugar to complete their life cycles. We studied plant feeding of Lutzomyia longipalpis sand flies, known vectors of Leishmania infantum/chagasi parasites, in a Brazilian city endemic with visceral leishmaniasis. The DNA barcode technique was applied to identify plant food source of wild-caught L. longipalpis using specific primers for a locus from the chloroplast genome, ribulose diphosphate carboxylase. DNA from all trees or shrubs within a 100-meter radius from the trap were collected to build a barcode reference library. While plants from the Anacardiaceae and Meliaceae families were the most abundant at the sampling site (25.4% and 12.7% of the local plant population, respectively), DNA from these plant families was found in few flies; in contrast, despite its low abundance (2.9%), DNA from the Fabaceae family was detected in 94.7% of the sand flies. The proportion of sand flies testing positive for DNA from a given plant family was not significantly associated with abundance, distance from the trap, or average crown expansion of plants from that family. The data suggest that there may indeed be a feeding preference of L. longipalpis for plants in the Fabaceae family.

Sand flies require sugar as their energy source [10][11][12] . However, there is little information about specific sources of these sugars and about how sand flies are attracted to plants, especially under natural conditions 7,13 .
To extend current knowledge about the ecology of insects and plants, more information about insect-plant interactions is necessary 14 . Direct analysis of the plant content in an insect's digestive tract is a precise approach to understanding feeding behavior 15 .
The use of the DNA barcode tool, a technique that is able to identify the food content in invertebrates, using short sequences obtained in conserved regions of the chloroplast, has been widely reported in the literature [16][17][18][19] . Specifically, the ribulose diphosphate carboxylase (rbcL) gene has been successfully used to identify food source 14,20,21 . This same locus was used by Junnila et al. 22 to identify plant-derived food content in wild-caught Phlebotomus papatasi sand flies.
The DNA barcode approach can be used to identify relationships between presence of plant types and vectorial capacity or simply the likelihood of vector habitats in a particular area. One application of this technique, therefore, would be informing vector control measures, such as urban landscaping techniques, for leishmaniasis and other vectorborne diseases. Here, we describe a study designed to identify plant-based feeding preferences of L. longipalpis through the use of DNA barcode technology in a Brazilian city where visceral leishmaniasis is endemic.

Results
We used rbcL PCR to detect and identify plant DNA in the guts of 100 L. longipalpis captured during 5 days in the tropical urban environment of Teresina, Piauí State. Fifty-seven percent of the flies were found positive for plant DNA. In the vicinity of the trapping location, we identified 22 species of plants belonging to 14 families (Supplementary Table S1). DNA from the ubiquitous plant family Fabaceae was most prevalent and was identified in 94.7% (54/57) of sand flies in which plant DNA was detected. On average, each sand fly was positive for the DNA of 2.1 ± 1.4 plant families. To consider whether plant availability or proximity may have contributed to observed feeding trends among sand flies, relationships between plant characteristics and frequency of plant DNA detection in sand fly guts were assessed. The proportion of sand flies testing positive for DNA of a given plant family was not significantly correlated with local abundance of plants from that family, average distance of plants from the trap, or average crown expansion of plants in the family (Fig. 1). A statistically significant, negative correlation was found between the average distance between plants from a family and the trap and their average crown expansion (Pearson r = −0.77; p = 0.002) ( Fig. 1 and Supplementary Table S1).
On each of the five trap days, DNA from four to eight of the plant families was detected in the flies captured and analyzed (Table 1). Although there was little variation in measurements of abiotic factors (i.e., temperature, air velocity, relative humidity) across the sampling period, humidity was highest on the two days with lower temperatures (Day 1 and Day 4). DNA from 7.5 plant families on average was detected in sand flies captured on these two cooler, more humid days, while DNA from 5 plant families on average was detected in sand flies captured on the warmer, less humid days.

Discussion
This study verified the feasibility of detecting plant DNA in the digestive tracts of L. longipalpis using a barcode approach. The successful application of this approach is consistent with previous studies investigating feeding behavior of P. papatasi 22 . Furthermore, detection of plant DNA in sand fly guts is biologically plausible as sand flies feed directly from the plant's tissue 23,24 to obtain sugars that will be used as energy sources 25 . However, despite the success of the approach presented here in distinguishing L. longipalpis feeding sources, plant DNA was not detected in nearly half of the caught sand flies. This may be explained by factors such as DNA degradation by enzymes in the digestive tract-a process that occurs with blood DNA ingested by females during blood feeding 26,27 -or due to sugar acquisition from honeydew excreted by aphids and coccidians 28,29 .
At the sand fly trapping site, plant species belonging to 14 families were collected and identified. The Fabaceae family was the most frequently detected food source of L. longipalpis (Table 2) despite the low abundance (2.9% of local plants) and relatively high average distance from the trap of plants from this family. Additionally, DNA from plant families, specifically Anacardiacea and Meliaceae, that were most represented in the area (i.e., plants from these families had the highest abundance) was found in fewer insects than DNA from the Fabaceae family (Table 2 and Supplementary Table S1). Of further note, some plant families (i.e., Rutacea and Annonacea) with abundance higher than that of Fabaceae were found in none of the sand flies (Table 2). This attests to the fact that sand flies have a feeding preference for certain plants or plant families.
The plant families Fabaceae, Anacardiaceae, Meliciaeceae, Rutaceae and Annonacea have pantropical distributions. The Fabaceae family, commonly known as legumes, is considered to be the third largest angiosperm family 30 . The species belonging to this family have carbohydrate heterogeneity, varying from simple sugar to complex heteropolysaccharides 31 . The Anacardiaceae family is known for having many fruit trees of high economic value, due to their wood and their production of substances used in industry or medicine 32 . The Meliaceae family includes trees producing meliacin, known for its insecticidal properties 33 . Plants from the Rutaceae family have a large variety of secondary metabolites, as alkaloids, coumarins, flavonoids, limonoids, and volatile oils 34 . Species belonging to the Annonaceae family produce bio compounds with medicinal, allelopathic or pesticide properties 35 .
The attraction of sand flies to certain plants, as well as to certain honey odors, has been reported in different studies [36][37][38][39] . According to Muller and Schlein 12 sand flies seeking sugar sources are primarily guided by attractive factors. These factors can be the carbohydrate composition of certain plants or plant families, high CO 2 emission which can be detected by the insect 40 , or even the release of a phytochemical affecting the olfactory system 41,42 . Muller et al. 13 who evaluated the P. papatasi attraction index for different plant species, found that phytochemicals serve as potential attractants for insects. Magalhaes-Junior et al. 43 additionally verified that volatile plant compounds, 1-octen-3-ol, 1-nonanol and 1-heptanol, can act as attractive factors for insects from the L. longipalpis species. It is worth noting that the compounds 1-octen-3-ol 44 and 1-nonanol 45 have been identified in different species belonging to the Fabaceae family.
Through an assessment of the relationship between abiotic factors and the number of families detected in sand flies per collection day, it was evidenced that more plant families were used as a feeding sources on those days with lower temperatures and higher relative humidity. However, no reports were found in the literature about how these factors might influence the dynamics of sugar acquisition in these insects. Currently, it is known that abiotic factors such as temperature and relative humidity influence the population dynamics in the different sand fly species [46][47][48] including L. longipalpis 49 .
Accordingly, our study suggested there is indeed a feeding preference by L. longipalpis for the Fabaceae family, represented in this study by specimens of the species Albizia niopoides, Anadenanthera macrocarpa, Cenostigma macrophyllum and Tamarindus indica. At the same time, the data suggests that L. longipalpis had less preference for plants belonging to the families Anacardiacea, Meliaceae, Rutacea or Annonacea.
Upon further verification of L. longipalpis plant feeding preferences, it will be necessary to determine whether sand fly abundance varies with the presence or absence of plants from certain families. If so, the combined findings may have implications for controlling urban epidemics of vector borne disease. In particular, removal of preferred sugar sources for sand flies could reduce their prevalence and thus reduce density-dependent transmission of the pathogens they harbor.
It is also recommended that new studies relating the feeding interaction between sand flies and plants focus not only in the identification of plant species or plant families serving as a carbohydrate source, but also in the identification of the chemical components that lead to the insect preference for certain food source. Additionally, it is necessary to understand the role of each abiotic factor in this feeding process, and this way, sand fly population control measures may be applied.

Materials and Methods
Field collection of phlebotomine sand flies. Insect and plant collection were conducted in a region of the city of Teresina (05°07′700″S/42°46′426″W), capital of Piauí State, in Brazil's Northeast. The sand flies were captured using a CDC light trap, deployed in an animal shelter outside of a household. The trap remained at the site for five consecutive nights working 12 hours per night (6 P.M. to 6 A.M.) in January, 2015. The captured sand flies were taken to the laboratory for identification and processing. Only L. longipalpis sand flies were separated randomly from the sample of sand flies collected every trapping night. In order to avoid the contamination of the samples during the DNA extraction process, every plant fragment was removed from the sand fly bodies. Previously collected insects were submerged in a solution of 0.5% hypochlorite with 0.01 ml/ml Triton X-100 detergent, agitated gently for 1 min with forceps, and then rinsed in double distilled water (ddH20) for 1 min as proposed by Matheson et al. 15 . After this procedure, the insects were stored at −80 °C for further DNA extraction. All the techniques used for plant identification had been previously tested using DNA directly from plants (Lima et al., submitted). For every capture day, values of temperature, relative humidity and wind speed were obtained in collaboration with the National Meteorology Institute (INMET).
Field collection of plants. The botanical material (leaves and flowers) was collected under the criteria that plants should be, at the most, 100 meters apart from the sand fly trap, which is within the interval of dispersal of the sand fly Lu. longipalpis 50 . In total, 22 plant species were collected and subsequently identified at the Graziela Barroso Herbarium at the Federal University of Piauí. For genetic analysis, species were labeled and stored at −80 °C until DNA extraction. Information on species included in the present study is provided in Supplementary  Table S1.
Distances of each plant specimen from the CDC traps were recorded using a Garmin Dakota 10 GPS device (Garmin, USA) and the average expansion of the crown was measured according to methods proposed by Blozan 51 . DNA extraction, PCR and sequencing. During the DNA extraction process, sand flies were put in 1.5 mL test tubes and 300 μl of CTAB lysis buffer (50 mL 2x CTAB + 50 mL ddH 2 0 + 200 mL mercaptoethanol (0.2%)) was added. The samples were macerated with an Argos Pellet Mixer (Argos, USA), and 10 μl K proteinase (Invitrogen, USA) was added to each tube (concentration 10 mg/mL). Immediately after, the samples were incubated for two hours in a 60 °C hot bath. Extraction was then conducted using the PureLink Genomic DNA Mini Kit (Invitrogen, USA) according to the manufacturer's instructions. The plant genomic DNA extraction was carried out using the kit BIOPUR Extração Mini Spin Planta (Biometrix, Brazil). The DNA was extracted and analyzed in 1% agarose gel under UV light and subsequently was quantified by spectrophotometer NanoDrop ™ 2000 (ThermoScientific, USA) and the fluorometer Qubit ® 2.0 (Invitrogen, USA).
For each separated DNA sample, a segment of chloroplast gene rbcL was amplified by PCR using the PCR Master Mix (Promega, USA) according to the manufacturer's instructions for a 50 μL final volume, containing 1 μL of each primer (rbcLaF and rbcLaR 52 ), and 5 μL DNA template. The reactions took place in a T100 (Bio-Rad, USA) thermal cycler. The thermal cycle conditions have been described by Bafeel et al. 52 . PCR products were visualized in 1% agarose gel under UV light and purified using ExoSap-IT (Affymetrix, USA) for the sand fly samples and PureLink PCR Purification Kit (Invitrogen, USA) for the plants. Further, this material was quantified by using fluorometer Qubit ® 2.0 (Invitrogen, USA) and purity verified in spectrophotometer NanoDrop ™ 2000 (ThermoScientific, USA). Sequencing was carried out in an ABI 3500 automatic sequencer (Applied Biosystems, USA). Only sequences with a PHRED score above 30 were used in the analysis. Contig assembly was carried out using CodonCode Aligner (CodonCode Corporation). Local alignments were done using BLAST 53 . Assembly of DNA barcode library. DNA sequences obtained from each primer were deposited in Genbank database, linked to the National Center for Biotechnology Information (NCBI) (Supplementary Table S1). The deposited sequences were set up for starting a reference library of plant species from the city of Teresina, and from other tropical cities.
Identification of the plant DNA in sand flies. Each sequence of plant DNA found in the insects was compared to a local plant library using the BLASTN. Only matches with highest score and, at least, 90% identity were considered for the sand flies' plant feeding preference 54 .
Statistical Methods. Descriptive statistics were summarized for individual plant characteristics and abiotic factors (i.e., temperature, relative humidity and wind speed) across the five trap days. Correlations between the average distance of the plant family from the trap, the average crown expansion of plants from a given family, the proportion of plants in the area from a given family, and the proportion of sand flies testing positive for DNA from a given plant family were investigated for the entire sample of sand flies.
All statistical analyses were performed using R version 3.2.1 (R Foundation for Statistical Computing, Vienna, Austria).