Data from: Environmental DNA metabarcoding reliably recovers arthropod interactions which are frequently observed by video recordings of flowers
Data files
May 06, 2024 version files 186.60 MB
-
2023-01-26-HSP_fastq.zip
-
Annotated_family_interactiontype.xlsx
-
Arthropod_list.xlsx
-
Arthropods_video_camera.xlsx
-
Meta.xlsx
-
OTU_tables.xlsx
-
README.md
-
Taxonomic_annotation.xlsx
Abstract
Environmental DNA (eDNA) metabarcoding promises to be a cost- and time-efficient monitoring tool to detect interactions of arthropods with plants. However, observation-based verification of the eDNA derived data is still required to confirm whether the arthropods indeed previously interacted with the plant. Here we conducted a comparative analysis of the performance of eDNA metabarcoding and video camera observations to detect arthropod communities associated with sunflowers (Helianthus annuus, L.). We compared the taxonomic composition, interaction type, and diversity by testing for an effect of arthropod interaction time and occupancy on successful taxon recovery by eDNA. We also tested if pre-washing of the flowers successfully removed eDNA deposition from before the video camera recording, thus enabling a reset of the community for standardized monitoring. We find that eDNA and video camera observations recovered distinct communities, with about a quarter of the arthropod families overlapping. However, the overlapping taxa comprised ~90% of the interactions observed by the video camera. Interestingly, eDNA metabarcoding recovered more unique families than the video cameras, but approx. two-third of those unique observations were rare species. The eDNA-derived families were biased towards plant sap-suckers, showing that such species may deposit more eDNA than e.g. transient pollinators. We also find that pre-washing of the flower heads did not suffice to remove all eDNA traces, suggesting that eDNA on plants may be more temporally stable than previously thought. Our work highlights the great potential of eDNA as a tool to detect plant-arthropod interactions, particularly for specialized and frequently interacting taxa.
README: Data from: Environmental DNA metabarcoding reliably recovers arthropod interactions which are frequently observed by video recordings of flowers
We compared flower-derived eDNA with video camera-based observations of arthropods on sunflowers. The eDNA was recovered using a washing protocol. We also pre-washed the sunflowers in some samples and compared them to non-pre-washed samples to remove undesired eDNA contamination from before camera observation time.
Data collection
For more details please see DOI: 10.1002/edn3.550
The sampled sunflowers were situated in a commercial field facing southwest in front of steep vineyards in the outskirts of Trier (Rhineland-Palatinate, Germany), a typical agricultural habitat of the region. To capture plant–arthropod interactions, five video cameras were positioned in front of the flower heads and recorded between 9 a.m. and 5 p.m. for a maximum duration of 5 hours on August 25, September 1, and September 2, 2022. The duration of some recordings was shortened from the initially planned 5 h due to video camera batteries draining before the intended duration was completed.
To test whether eDNA on the sunflower heads could be effectively removed with water before recording, 21 flower heads were prewashed using a hand-held pressure sprayer. We sprayed a jet of deionized water on the flower heads for 1 min by applying the stream evenly across the entire surface and letting the water drop of the flower head. As a control, ten samples were not treated with deionized water before the video camera recording. After recording, the flower heads were cut off with sterile stainless-steel scissors, leaving one centimeter of the peduncle intact. They were then placed in a plastic bag on dry ice in the field, transferred to a laboratory freezer, and stored at −28°C. To test for possible contamination of the sampling process and the materials, one plastic bag was filled with 100 mL of deionized water from the pressure sprayer in the field.
We distinguished different qualitative interaction types of arthropods with plants by identifying the most likely interaction type for different arthropod families based on their taxonomy. The groups we assigned were honeydew feeders, predators (for example, spiders), plant gallers, plant sap-suckers, plant miners, and plant chewers. We further differentiated between the pairs of pollinating nectarivores and palynivores, parasitoids and parasites, as well as coprophagous and saprophagous arthropods. For the analysis of the ecological composition, the values of each of these pairs was combined into the following categories: “palynivore/nectarivore,” “parasitoid/ parasite,” and “sopro-/ coprophag.” The combination of these groups was introduced to facilitate annotation at the family level (for example, nectarivores, and palynivores are not always distinguishable at the family level) and to tidy up the analysis by aggregating ecologically similar groups. The pairs were subsequently used as one interaction type. If more than one interaction type could be defined for a family, the values of the family would be treated as equal parts of a whole. For instance, when a single family was annotated to two different interaction types, the value of the family was multiplied by 0.5 and separately assigned to each interaction type. Hence, if one arthropod would appear in ten out of twenty samples, each of the two interaction types would be considered five out of twenty times.
The interactions derived from the video camera recordings were quantified using three approaches, resulting in the values used for the analysis. First, we counted a single interaction whenever one arthropod entered and thereafter left the video camera frame (hereafter called interaction count). Second, we measured each arthropod's interaction time with the flower head as the number of seconds passed during one interaction count (hereafter called interaction time). Furthermore, the occupancy of arthropods was used to describe the relative number of samples where a positive detection of an arthropod could be found. For instance, if an arthropod is present in five out of ten samples, the occupancy would be 50%.
We tested two sets of primers for arthropod DNA amplification (PCR amplification), both targeting the mitochondrial COI gene: The first primer combination was ZBJ-ArtF1c/ ZBJ-ArtR2c and will hereafter be referred to as “ZBJ.” The second primer combination was NoPlantF_270/mlCOIintR_ W and will hereafter be referred to as “NoPlant.” The water from the field, which was subsequently used as a control during filtering and extraction (hereafter called material control), and the PCR controls were processed and sequenced along with the experimental samples.
Briefly, sequences were merged using PEAR (Zhang et al., 2014) and then quality filtered with the FastX Toolkit (Gordon & Hannon, 2010). The merged sequences were then dereplicated, cleaned from Chimera, and clustered into OTUs with USEARCH (Edgar, 2010). We generated zero-radius OTUs (e.g. haplotypes, hereafter called zOTU). The taxonomy was assigned using BLASTn, with a maximum of 10 target sequences, and against the NCBI GenBank reference database. The dataset was further filtered by using a minimum similarity of 98% for annotation to species level, 97% to genus level, and 95% for family level if the taxonomic annotation was used for the analysis. As the control samples mostly contained under three reads per zOTU, we set all reads in the dataset under three to zero. Two zOTUs of the ZBJ dataset that had three and six reads in the controls but were the most abundant zOTUs of all samples were not considered for the filtering cutoff. To retain only arthropods, all non-arthropod zOTUs were removed before the analysis. To quantify the interactions derived from eDNA metabarcoding, we used the relative read abundance within a sample by deviding the read number of each zOTU in the sample by the total read number of the sample. The occupancy of arthropods and their interaction types were assigned, as described before, for the video camera recordings.
Files
1. Arthropods_video_camera: This file contains the families and orders identifiied from the camera recordings. We counted each arthropod that entered and left the frame (interaction_count) and noted the passed time during that interaction count for each arthropod interaction (interaction_time). If the interaction time was very short, we noted 0 interaction time but an interaction count of 1. Since the original values were notated in minutes, we had to transform them to seconds, resulting in decimal numbers.
- sample_ID: ID of the sunflower
- PCR_ID_NoPlant: sample_ID related ID of the NoPlant primer used for eDNA metabarcoding
- PCR_ID_ZBJ: sample_ID related ID of the ZBJ primer used for eDNA metabarcoding
- Order: Order of the arthropod that was identified via video cameras
- Family: Family of the arthropod that was identified via video cameras
- interaction_count: The interaction count is always one, since all occurences are listed in the file
- interaction_second: Time in seconds that passed during one interaction count
2. Annotated_family_interactiontype: Interaction types annotated to the families observed with the "NoPlant" and "ZBJ" primer used for eDNA metabarcoding and the video camera recordings.
- Family: Arthropod family that was identified either by video camera or eDNA metabarcoding
- Family_interaction_type: Ecological guild that the observed family belongs to. At the family level, often more than one interaction type occure.
3. Arthropod_list: List containing all species derived from NoPlant and ZBJ primer (cutoff of species with similarity under 98% to the reference from NCBI blast) and all families (cutoff at 95%) found in the video camera recordings.
- Species_98_percent_similarity_to_zOTU: Species detected via eDNA metabarcoding
- Family_95_percent_similarity_to_zOTU: Family detected via eDNA metabarcoding
- Order: Order the species and family belong to.
- NoPlant_occurence: Does the family occure in the NoPlant dataset?
- ZBJ_occurence: Does the family occure in the ZBJ dataset?
- Camera_occurence: Does the family occure in the video camera dataset?
4. Meta: Meta table
- sample_ID: The ID repeats three times, for each of the two PCR primers and the video camera samples
- PCR_ID: Unique ID that describes the samples. Note that the PCR_ID containing "_V" belongs to the camera samples and is naturally not processed using a PCR.
- primer: Primer used for PCR during eDNA metabarcoding processing. Video camera samples are represented with a NA.
- sampling_date: Field sampling date of both eDNA metabarcoding and video camera recordings.
- pre_wash_off: Yes = The samples were pre-washed before video camera observation, No = No wash off before video camera observations. We intended to wash off undisered eDNA contamination. All of those samples were then washed in the lab to recover eDNA from the surface of the pre-washed and non-pre-washed sunflowers that were exposed to arthropods interactions after pre-washing for up to 5 hours.
- obseravtion_time: Duration of video camera observation. After the observation time, the samples were cut off for eDNA recovery in the lab using a washing protocol.
- comment: Shows material control (field control of water, and all material used in field, filtering and in DNA extraction) and PCR controls that were not used as samples, but to control the eDNA metabarcoding results.
5. OTU_tables: 2 sheets including all zOTUs (column OTU_ID) of all samples (header) from the NoPlant and ZBJ libraries.
6. Taxonomic_annotation: 2 sheets including taxonomic annotation of the zOTUs shown in the OTU tables.
- hit_percentage: Similarity to the NCBI reference sequence
- base_length: length of the recovered fragment
Missing data code : NA
The zip file 2023-01-26_HSP_fastq contains all fastq files used in the study.
Methods
We compared flower-derived eDNA with video camera based observations of arthropods on sunflowers. The eDNA was recovered using a washing protocol. We also pre-washed the sunflowers in some samples and compared them to non-pre-washed samples to remove undisired eDNA contamination from before camera observation time. For further methods please see the paper: DOI: 10.1002/edn3.550