Changes in the status of Bees in the Washington DC region (Part 2 of 3)

[Sam Droege]

See the previous email for background information

Results

I will take two approaches to evaluating bee change using these data.  First will be a simple inspection of the species lists between the two time periods (what species appear and disappear and stories about why that might be).  Second will be a standardized evaluation of changes in proportions of these species captures using just the netting data.

All data are not equal.  Five obviously spurious names were dropped from the GBIF dataset (A very small number in my opinion, obviously data providers are doing a pretty good job at this level). An additional 53 species were eliminated because either recent taxonomic changes likely would not be reflected in the old GBIF data or within portions of BIML data or I thought that certain groups of Nomada and Sphecodes were difficult to impossible to safely identify. That left 319 species and species groups with information to play with. 

Comparisons of Species Lists

Our analysis areas for this study includes the jurisdictions of DC and 2 adjacent MD counties.  GBIF detected 202 species, BIML when looking only at netting results found 213 (interesting how close these two numbers are), slightly more bee species were present in the BIML trapping data (225), and when you combine all sources of bee data (this includes iNaturalist and the literature) for the DC region the list jumps to 319 species. 

As a historical contrast, Ted Mitchell in the early 60's heroically published 2 taxonomic volumes on the bees of the eastern United States and also tabulated species by state (including DC).  He recorded only 79 species for DC and 115 for Maryland. Note that Ted's results for Maryland include data from the entire state not just the DC area.  In an alternative universe using Ted's data we could conclude that bees are doing roughly two time better now than in the past, but we know that is a misleading story and we will investigate to find out why. 

How do past and recent netting results compare?

Again, in the following exercise we are just using the species lists and not including any information about the number of specimens captured.  Of particular interest are the number of species ONLY detected during each time period and not present in the other time period (see the associated Excel file for the full species lists).  Inspecting the list of bee species found while netting during the historic data we find that of those bee species 58 species were only detected in the early years and not in the later years.  All else being equal (when is that ever true with bee data?) if bees were declining like mad in the region, the recent BIML data netting data should have a shorter list of species seen in recent times but not in historic times than visa-versa. This is not the case. BIML netting data detected more missing species (70 for BIML vs 58 for GBIF). This differential is highlighted by the fact that there were far fewer sampling days during the recent time period and that the recent time period only covered approximately 25 years while the historic period covered over 100 years.  Only 144 species were seen in both time periods.

Very interesting.  One could interpret this as indicating that bees are doing better now in the DC region or maybe that we are just bad at measuring bee species' change.  Things certainly seem imperfect, yet there are things to learn here by going deeper about the status of bees in the DC area.

Rare things are rare things.  Look at the list of bees only found during the early period when comparing netting data between the two time periods.  With a couple of exceptions these are all species that are rare.  Eucera hamata is an exception and does appear to be relatively common in the area, but for some reason only seems to show up in bowl type traps not in netting.   

A few species do seem to be now truly absent from the area.  Those include bumble bees associated with the Bombus population collapse of the  90s (B. fraternus, B. affinis, B. terricola). Bombus ternarius was also only present in the early time period but not found later.  This bumble bee species is not associated with declines during but it is a northern species that has not been recorded in recent times in the state (who wouldn't collect a bumble bee with a bright red butt?).  The one record was collected by L.O. Jackson in 1914 along the Anacostia (Entomology Collections Search).  Megachile latimanus also needs to be on the list as this obvious species has retracted its range in Maryland into the mountains, but was regularly found in the region in the past.  More interesting patterns will be revealed when we look at changes in the number of collection dates.

Several factors associated with past records need be kept in mind. While rare species would have been preferentially kept in collections in the past it does not mean that they all have shown up in current databases that house historic records.  For example, the NMNH collections have a great deal of regional data that will not appear in any database because the data have simply not been entered and thus some of the species "missing" during the early time period's data will disappear once that effort is complete.  Similarly, there are specimens languishing in the unidentified pro tem sections of museums awaiting identification or were misidentified that will also show up. This may be some of the reasons that despite having 20% more collecting days involved in the historic data they caught 5% fewer overall species than in the recent period (202 species versus 213).

More understanding of status can be garnered by comparing the list of historically detected species now appearing to be missing if we simply expand our investigation to additional sources of recent data on bee species besides BIML netting and see if any of these historically "missing" species might have been found using these trapping techniques (they also greatly increase the cumulative amount of sampling for bees during this time period so it is both technique as well as the additional days that result in finding more species).  The result of expanding the search for recent bee records from just BIML netting data to BIML trapping data and iNaturalist indeed does detect an additional 21 species of bees as recently present that were previously considered historically missing in the GBIF dataset when the comparison was restricted to netting data.  This drops the "missing" list of species to 36.

Let's stop and reflect a bit on these results because there are lots of numbers with qualifications being thrown about and story telling is sometimes more effective at grilling the meat. In the olden days a bunch of guys would go out to their favorite collecting spots in the DC area and hunt bees for their collections. Each time their goal was to find new and rare species and to enjoy a day in the countryside. In the past 25 years a different bunch of guys did the same thing, going to different places, but the goal was the same: find rare and cool bees in places that might have new bee species. Comparing their results, each group found bees that the other group did not find with the recent guys finding more species than the historic guys using similar techniques and strategies.  Other people (many not "guys") were also looking for bees recently in the DC area but instead of netting were using traps and taking pictures.  They found a bunch more (21) of the species the historic guys had found but the recent netting guys did not.  When you mush all the data together it turns out the recent people appear to have done a way better job at finding bee species than the historic guys.  So good at finding bees were the recent folks that they found 98 species of bees that the historic folks did not find while the historic folks only found 37 species that the recent folks did not find.  How cool is that even though the DC area has clearly gone to biodiversity hell during the past 125 years (i.e., more people, more buildings, more pavement, more deer, more invasives und so weiter) we can find more bee species than in the region than in the past.

So, is everything good?

Let's look at those 37 species a bit more (you can see the whole list in the Excel file) that have not shown up recently and see if there are any patterns that might influence why they have not been seen recently and then we will ask a question about how complete those historical data really are.  It is important to note that even if the historic data are pretty poor, any species they found but recent efforts did not needs to be investigated as a "missing" species (could this be made into a true-crime podcast?  "Who killed Andrena braccata?", but I digress). 

Forty-one percent of the total missing species (15 species) are pollen specialists (Andrena accepta, Andrena andrenoides, Andrena braccata, Andrena phaceliae, Colletes simulans, Colletes willistoni, Epimelissodes compta, Macropis patellata, Megachile integra, Perdita halictoides, Perdita novaeangliae, Protandrena abdominalis, Protandrena pauper, Protandrena illinoiensis, and Xenoglossa strenua) and by their nature both have lower natural populations and if their plant species is absent or no one investigates those plant species you would not find them. 

Twenty-four percent (9 species) are nest parasites and also have naturally lower population numbers (Coelioxys galactiae, Epeoloides pilosulus, Epeolus autumnalis, Epeolus banksi, Epeolus scutellaris, Holcopasites illinoiensis, Nomada integerrima, Triepeolus atripes, and Triepeolus pectoralis). 

Eight percent (3 species) are Bombus species that we know have crashed from introduced pathogens (Bombus affinis, Bombus fraternus, Bombus terricola).

Fourteen percent (5 species) can be tricky to identify and could possibly be incorrect (Andrena ceanothi, Andrena integra, Andrena miranda, Colletes eulophi, Osmia proxima)

Thirty-two percent (12 species) are at the edges of their range and their absence could indicate changes in overall range limits.  The letter after their name indicates which direction the bulk of their population lies (Andrena accepta (W), Andrena phaceliae (N), Anthophora ursina (W), Bombus fraternus (S), Bombus terricola (N), Coelioxys galactiae (S), Andrena integra (N), Colletes eulophi (W), Megachile centuncularis (N), Megachile latimanus (N), Protandrena abdominalis (W), Xenoglossa strenua (S)).

Sixteen percent (6 species) are rare throughout their range.  Rare is defines as having fewer than 100 records in GBIF (Coelioxys galactiae, Epeolus banksi, Nomada integerrima, Perdita novaeangliae, Protandrena pauper, Triepeolus atripes).

These categories cover all 37 of the species.

So, a reason that the list of "missing" species from historic GBIF for the region is relatively small is that the historic data found on GBIF is incomplete, meaning that there are specimens and species out there that have not made it into the database.  Why might that be?  Some specimens may have been collected but not identified, others not entered and some misidentified or destroyed before being entered.  One way to investigate that is to look for alternative sources of historic data other than GBIF and compare.

Enter Ted Mitchell and his wonderful revisions of eastern bees from the early 1960s.  You may recall that his list of bees for DC (79 species) was quite a bit lower than GBIF's list for DC (116 species).  It was not Mitchell's main intention to create species lists and his list are also likely to be incomplete because of that. But. His list of species does allow for one useful comparison with GBIF.  Recall that Mitchell's books were published in the early 60's. Thus the data for his species lists (and the associated specimens) would all fall in what I am calling the historic period in this essay.  Therefore, in a perfect world, all the species recorded by Mitchell for DC would have also been entered into and present in GBIF.  But.  That is not the case. There are 30 species listed as present in DC by Mitchell that do not have a record in GBIF (but should).  This represents a minimum 38% miss by GBIF. 

So now we have, unsurprisingly and as suspected, independent knowledge that the GBIF database is incomplete. There are specimens we know were historically found and written about by an expert but are not found in the GBIF database.  But. How many of those 30 species Mitchell detected in DC but were not found in GBIF have been now been found in the past 25 years or found in GBIF within the two Maryland Counties?  Only four. These would therefore also count as "missing" species:  Andrena ziziae/verna, Megachile relativa, Melissodes apicatus, and Protandrena virginicus . Andrena ziziae/A. verna and M. relativa are both at the edge of their ranges while P. virginicus is globally rare with only 16 records total in GBIF. Melissodes apicatus is an interesting story. 

The old record(s) for M. apicatus almost certainly would have come from within DC in the marshes that lined the freshwater tidal Anacostia river, which would have been easily accessible historically.  However, those marshes were in the poor part of town and as one does back in the day they got largely converted to dumps. The extent of the freshwater tidal plant pickerelweed (Pontedaria cordata, the pollen plant used by M. apicatus) consequently was greatly and obviously reduced from historic amounts.  That said, in Prince George's County pickerelweed beds are still extensive along the freshwater tidal portions of the Patuxent River and this bee has been found in those pickerelweed beds but only on the Anne Arundel county side of the river which is much more accessible to bee biologists than the P.G. County side. Thus M. apicatus readily falls into the loose category of "likely present but undetected" in the current study area.

OK, that seems like a reasonable exploration of historic bee species that are either are no longer in the area or were, in truth, still here but we did not detect them recently.  We have listed explanations (or are they Just So stories in the best Kiplingesque sort of way) for why these bee species are not on our current lists.  Keep in mind that our missing species all (with the exception of bumble bees) were never common and that is telling.  Rare bees are both hard to count/find and are rare for reasons that make them among the most likely things to become extinct (at least locally) or receding out of the area as their range changes. Perhaps there is some insights to be gained from an exploration of the bee species that have been detected recently but were not historically detected.   

In review, GBIF found 202 species, BIML when  looking only at netting results found 213, slightly more bee species were present in the BIML trapping data (225), and when you combine all sources of bee data (this includes iNaturalist and the literature) for the DC region the list jumps to 319 species.  Comparing recent to historic data we find 98 species on the recent list but not on the historic list and 37 on the historic list but not on the recent.  That is a big differential. 

Why might that be? 

Well we have already documented that the historical dataset is very incomplete.  How complete (in terms of grabbing any existing bee data from the region) is the recent list?  I would say pretty darn complete.  I don't know of any people/groups in the region that are collecting bees that we have not worked with directly, double checked their identifications, and extracted any useful data from their efforts.  There is iNaturalist data on bees for the region and we have also extracted those records.  That doesn't leave a lot of possibilities of other sources of data, though, I suppose, there could be some very secret deep state sampling of bees I am not aware of.

OK, let's explore these recent species that could be new to the area, keeping in mind that many probably were around historically, but we just can't prove that.  That said, some we know some species were almost certainly not here historically because they were brought in from another country, many rather recently.  There are eight species in that category.  Some of these invasives are now very common caught in the region (e.g., Osmia taurus, Megachile sculpturalis, Anthophora villosula).  Other native species are caught commonly now, but mysteriously did not show up on the historic list (e.g., Andrena illini, Halictus confusus (perhaps the most surprising one not on the historic list), Nomada imbricata, Megachile georgica).  Many species are seen regularly but in more modest numbers and some are quite rare (e.g., Andrena fulvipennis, Colletes brevicornis, Colletes productus, Hylaeus sparsus, Lasioglossum semicaeruleum, Panurginus atramontensis).

It is very likely that some of these recent records resulted from the use of newer trapping techniques deployed during the recent time period (re: bowl, pan, cup, malaise, canopy traps) and that increased the likelihood of some of these species being detected.  In particular, you can see that nicely illuminated in the genera commonly found in traps Lasioglossum (15 species added), Nomada (10 species), Andrena (18 species) and Osmia (6 species).  Indeed trapping data provided about twice the number of records of bee species not detected in the historic data (30) compared to the more traditional netting (17). Some of this differential has to do with effort, but it is difficult to compare the two technique categories except by number of days each were deployed. We can see that there were many more sampling dates for trapping (856) than netting (286) and likely that within an individual date more trapping locations than netting locations (on average) were sampled.  Thus trapping is a big factor here in terms of detecting recent species and has added many more days of sampling in the recent time period than the 582 netting dates in which there was  historic collecting activities.

In the next section we will look at a more quantitative exploration between historic and recent bee numbers in databases of the bee species that occurred in both time periods. But.  Before we do let's big picture up and look at the patterns of presence and absence of bees between historic and recent time periods. 

First we see that plenty of bee species have been seen in the Washington D.C region.  We have a firm list of 319 species and a few species groups.  Not included are another ambiguous bunch of species that we decided to not include. Thus we do have a fair number of species to play with. Inspecting the data within any combination of technique, time period, subregion or total there will be quite a few species of bees that have only been seen one time.  These singletons tell us that we are pretty bad as sampling.  If there are lots of species detected only once then there are lots (probably more) that are not detected at all.  Its just mathematically and biologically improbable that only single individuals of a bee species represent the only members of their species present in the region (not a sustainable reproductive strategy).  This also means, with almost equal certainty, that we don't really don't have a complete picture of what bees are in the region in any of these analysis categories. 

We have just demonstrated that records of our recent sampling efforts are much better than those we have for the past ... even after lumping over 100 years of historic data. When we turn these results and patterns over and over in the illuminating light of the conditions in which those data were collected we can rather unexpectedly conclude that we have not lost that many bee species in the region. We do have some pretty good evidence that a few species that regularly occurred in the region have now left  (e.g., Bombus affinis, B. fraternus, Megachile latimanus) and others have moved in...a natural state of affairs. 

The Apocalypse Will Not be Televised.

In sum, at the level of patterns of species being present or absent in the region, there is no evidence of anything that could be defined as even mildly apocalyptic.  And.  As mentioned earlier in this essay, at the very same time we are deeply into the apocalypse because, you see, unthinkingly, we have taken away from bees vast swathes of their landscape and converted it to houses, roads, parking lots, lawns, and croplands forcing our bees into a smaller and smaller number of locations along with the native plants they require.  We can still find those bees, but the apocalypse clock ticks down and we must move our thinking, actions, and hearts more and more toward conscious management of our residual bee reservoirs and into repatriating what we can of that which we have taken away.  In particular, we must take a pause and reflect on our cultural desire to turn everything into lawns.  Not useful if we care about the creatures of the planet or the planet itself.  Now is the time to turn lawns and the mowing of that which does not need to be mown back into regionally appropriate habitat.  Lots of opportunities are available but the average citizen needs to get on board and they are not.

Excuse my wandering.

How can one be orderly with this?

It's like counting leaves in a garden,

along with the sound notes of partridges,

and crows.

Sometimes organization

and computation become absurd.

• Rumi

(the next section will come in a couple of weeks as I am away at the moment and don't have time to complete part 3)