Changes in recorder techniques - can we detect differences in the dataset?

I have long felt that the shift away from records based largely upon specimens to records based on non-lethal methods such as photography was likely to be influencing the outputs of analyses to investigate trends in species' abundance. Whilst revising the text for the Provisional Atlas, I became acutely aware that some of the trends did not seem to fit my perceptions from field work and from monitoring the UK Hoverflies Facebook page. I therefore suggested to Stuart that it would be helpful to run two separate analyses; one for all data and the other for a subset of the data that excluded known photographic records. The results are really very interesting and can be summed up as a table (Figure 1).

Figure 1. Permutations of possible trends in the HRS dataset and in a subset that excludes photographic records. The final column highlights whether the permutations were found in analysis.

I think the overall results need to be published in the peer-reviewed press because they show how research teams must treat trends with caution. I expect that all of the graphs will become available at some point once we have decided how we might use them in the provisional atlas but probably not readily apparent in any printed version because of the cost of printing in colour. We can of course do so as a pdf without any problem. Here are a couple of examples:

Figure 2. Trends for Cheilosia proxima, All records in blue and with photographic records excluded in red.

Figure 3. Trend for Cheilosia impressa. All records in blue and with photographic records excluded in red.

Figure 4. Trend for Epistrophe diaphana All records in blue and with photographic records excluded in red.