i have used the script from the biomass paper to process the environmental data for 2018
it is just saved in the main directory:
write.table(outputCast,file="environData_2018_DK.txt",sep="\t")

last time, we had a cleaned-data folder and put it in there, shall we do that again?

I will work on the 2019 data now but having some weird issues reading it in. Will keep trying.

Cluster analysis - look for natural land use compositions and relate them to biomass

We have a column with land use and one with sub land use type which originally was derived from the route. I have previously made a clunky attempt of standardizing the different variations (from Danish): https://github.com/CecSve/InsectMobile_Diversity_and_Biomass/blob/main/reports/02_Sampling_data.R - could we use this in any way for heterogeneity (count amount of land use + sub land use for each route)? I think it would probably make more sense to take this from the environmental variables we have with proportions. But just throwing the idea out there - otherwise I think we should disregard the rough land use categories all together and don't spend time standardizing them.

• double-check that metadata for imaged bulk samples are included
• check that 2017 samples are included (will be excluded) and how many samples they account for
• #18
• check the routes from AU match the routes we have in CPH

Lmer regression - land cover and heterogeneity prediction - test for biomass, richness and evenness

Need to think a bit more about what is our measure of heterogeneity.
As its simplest, it could be simply the Shannon diversity of the land cover proportions (but whether that is just based on the 5 dominant classes, or we use a finer habitat categorisation, is still to be decided)

https://www.pnas.org/doi/10.1073/pnas.2002554117

@PtrsBrt here's a list of to-dos for you:

• make sure all lab data is digitized
• prepare a 02 script for German sampling data and align to the Danish output
• prepare a 03 script for German lab data and align to the Danish output
• in script 03, make a ID column that incorporates size fraction
• in script 03, rename PCRIDs to S18_** and S19_**
• retrieve all environmental data from Germany

all analysis could involve:

• biomass
• richness
• rarefied richness
• eveness
• diversity

Previous analysis was just of 2018, but now we have 2 years. I propose we analyse both together:

• We can do this by:
• same model as before except with year as an additional fixed effect. We should not complicate it by any year x land cover interaction (i.e., we dont want to test whether land cover effects differ between the years). Instead, we should just focus on mean effects of the land cover across the years.
• We still need the transect as a random effect. It doesn't matter if some were only sampled in one year.

Same as original analysis
Lmer regression - continuous land cover (original analysis) - test for biomass, richness, evenness, diversity (already written, do we do rarefaction? iNext? incidence-based rarefaction)

General check-up when we merge everything (should perhaps be calculated by year):

• how many collected samples from DK?
• how many collected samples from DE?
• How many of the collected samples were processed in the lab (which were, which weren't)? DK
• How many of the collected samples were processed in the lab (which were, which weren't)? DE
• How many of the processed samples were sequenced DK?
• How many of the processed samples were sequenced DE?
• How many samples made it through the bioinformatic processing of the sequenced samples DK?
• How many samples made it through the bioinformatic processing of the sequenced samples DE?

All this should be documented in the supplementary file. In the main manuscript, we should write something in the lines of:

Of the 771 samples collected during June 2018 in Denmark, 561 and in June 2019 678 samples out of 798 collected insect samples were retained for analysis after quality checks in the laboratory and the bioinformatics pipeline (Fig XX (map of sampled areas in DK with transparent dots coloured by year collected), SXX).

Document the output for each country in a comment below.

Bottom section of script 01 checks the names of the species.

Mostly accepted (95%), we have 240 synonyms and 4 doubtful status

The function at the bottom gets the accepted name of these species according to GBIF.

Just guide me on what I can do next.
e.g., check genus too?
replace the species names with their accepted names in taxonomy_cleaned_sub ?

Relationships among the metrics - richness vs biomass

Should we carry out analysis for Class = Insecta only or should we include Arachnids and some other arthropod classes as well, or just stick to flying invertebrates (which may be difficult to pin point)? We should at least remove the weird ones:

• Phylum = Cnidaria (e.g. jelly fish)
• Phylum = Nemertea (worms, unsure whether some species could be associated with insects?)
• Class = Pycnogonida (sea spiders)
• Class = Polychaeta (we have one marine species in the taxonomy list)
• Class = Turbellaria (again a marine species)
• Kingdom = Protista
• Phylum = Onychophora (velvet worm, would be super cool if we found one, but they all live in the southern hemisphere and are extinct in the north)
• Phylum = Rotifera
• Phylum = Chordata (we have wild boar, cows, human and wolf/dog)
• Order = Isopoda
• Order = Decapoda
• Class = Malacostraca (would take care of Decapoda and Isopoda)

NMDS - Community composition dissimilarity (mapping the clusters onto land cover)

this route has an incorrect eventDate

 PIDRouteID SampleID  PID DOFAtlasQuadrantID    subLandUseType  eventDate StartTime  EndTime

1099 P346.2 P346.2A P346 5684 bog, meadow, lake 23-06-2010 14:29:00 14:56:00
Wind Temperature Notes PilotNotes RouteID_JB utm_x utm_y decimalLongitude
1099 gentle breeze 3.4-5.5 25+ 20 641080.4 6173608 11.24422
decimalLatitude eventTime Time_band Route_length Distance_driven
1099 55.68773 2019-06-23 14:29:00/2019-06-23 14:56:00 midday 5000 10000
Date yDay Time_driven Velocity Year
1099 2010-06-23 174 27 370.3704 2010

Looking at eventTime, it looks like eventDate year 2010 should be changed to 2019

I leave that for you to do!

Once done, please send me the updated sampling_data_cleaned.txt file

I checked this at the start of 04

Land cover:

• we have all data for 2019
• we have almost all data for 2018 (based on the covariate-data folder files of the biomass paper).
  -Exception: route "P115.2" - but maybe there is a reason

Traffic light counts:
We only have one file for this. I guess for 2018.
Not sure how to check for 2019??

     PIDRouteID SampleID  PID DOFAtlasQuadrantID    subLandUseType  eventDate StartTime  EndTime
1099     P346.2  P346.2A P346               5684 bog, meadow, lake 23-06-2010  14:29:00 14:56:00
                      Wind Temperature Notes PilotNotes RouteID_JB    utm_x   utm_y decimalLongitude
1099 gentle breeze 3.4-5.5         25+  <NA>       <NA>         20 641080.4 6173608         11.24422
     decimalLatitude                               eventTime Time_band Route_length Distance_driven
1099        55.68773 2019-06-23 14:29:00/2019-06-23 14:56:00    midday         5000           10000
           Date yDay Time_driven Velocity Year
1099 2010-06-23  174          27 370.3704 2010

from eventTime, looks like eventDate should be 2019 not 2010

update it within sampling_data_cleaned

Used for the carnet + metabarcoding manuscript - check and reuse code for this calculation.