[CT Birds] Another good read: Is eBird this bad with hummingbirds?
ghanisek at rep-am.com
Fri Mar 23 16:25:18 EDT 2012
Marshall is on a bit of a roll here after his treatise on commonly misidentified species, but I've always found the "hummingbird map" puzzling to say the least. This is really long, but interesting.
----- Original Message -----
From: Marshall Iliff
To: BlkVulture at aol.com
Cc: clw37 at cornell.edu ; bls42 at cornell.edu ; MH1920 at aol.com ; bill_hubick at yahoo.com ; lrbevier at colby.edu ; ghanisek at rep-am.com ; birdfreak007 at yahoo.com ; smirick at comcast.net ; kmcfarland at vtecostudies.org ; ehynes at maineaudubon.org ; Shaibal.Mitra at csi.cuny.edu ; af27 at cornell.edu ; sam.galick at gmail.com ; coturnicops at gmail.com ; tbj4 at cornell.edu ; rjostrowski at gmail.com ; naswick at gmail.com ; hcybelle at aol.com ; biodiva at myfairpoint.net
Sent: Friday, March 23, 2012 1:21 AM
Subject: Re: Is eBird this bad with hummingbirds?
Thanks for your thoughts on this topic. Following on the heels of a cautionary email about spring misidentifications from two days ago, I collated my thoughts and those of this group and posted the below long rant to our local listserv. March Madness for me I guess...but I do worry about "early spring fever" sweeping through the birding world with time-honored skepticism falling by the wayside. Not from all of you of course...
I'll be interested in hearing more on hummingbird arrivals (or non-arrivals) in your regions. I firmly believe that something is amiss with this map, but there are some mid- to late March dates from Virginia and Maryland in the historical literature, and maybe there is a real early vanguard of hummers that the average citizen is more likely to detect. If so, that would be highly interesting I think. But I'm betting not.
You can't believe everything you read on the internet, or so some have said. http://www.hummingbirds.net/map.html
Apologies for posting a second cautionary and skeptical note within as many days. Again, I don't wish to come off as cranky, and instead just have an interest in making the most of birder sightings. I truly believe that the observations reported to eBird, here on Massbird, and through other channels are useful bits of scientific data. However, this is dependent upon them being data points that we can have confidence in.
This post refers to the hummingbird map recently posted by Rick Bowes (thanks Rick for alerting us to this!). Rick wrote.
"For those interested, there's a map that plots the northward movement of hummingbirds located at http://www.hummingbirds.net/map.html. Looks like they should be showing up any day now!"
The map, as of tonight, shows some 24 dots north and east of of Virginia-West Virginia. It reports 2012 Connecticut dates as early as 14 March and Massachusetts dates as early as 15 March. This map to me, is highly questionable and the arrival dates reported for Ruby-throated Hummingbird really challenge my notion of what is possible. Below are a few thoughts about this, and I invite feedback. But I'll start with what we know. The below is a data rich discussion, but I think serves to raise some serious questions about where and when we can expect Ruby-throated Hummingbirds to arrive in the central/northern US and Canada.
2) Why is the eBird map for Ruby-throated Hummingbird map so incredibly different from the above map? http://ebird.org/ebird/map/rthhum?&gp=true&bmo=3&emo=3&yr=on&byr=2012&eyr=2012. Most of those records are from the wintering grounds (including coastal North Carolina) and only the Arkansas, Texas Hill Country, and a few other northern outliers are likely to be the vanguard of arrivals. Could the http://www.hummingbirds.net/map.html, with records from northern New York, Maryland, Massachusetts, Michigan, Connecticut, Wisconsin, the Great Lakes etc. be correct?
eBird is now a source of up to 3 million records (from over 200,000 checklists) each month. Although I really wish that more people in more places were reporting their birds to it (contact me if you want to get started!), I believe it now has enough information to provide an incredibly accurate and detailed look at the advancing front of bird movement in the USA. Check out these maps, for example, of the next species that will hit Massachusetts.
Blue-gray Gnatcatcher: http://ebird.org/ebird/map/buggna?&gp=true&bmo=3&emo=3&yr=on&byr=2012&eyr=2012 -- already early to our south, they may arrive next week (and will be ridiculously early if they do)
Louisiana Waterthrush: http://ebird.org/ebird/map/louwat?&gp=true&bmo=3&emo=3&yr=on&byr=2012&eyr=2012. Likewise, ridiculously early for Maryland and New Jersey. They should reach us within the next 1-10 days, so check territories near you.
These maps above match the chatter on the listservs about what people are seeing; eBird and the listservs provide quasi-independent verification of what birders are seeing.
2) Yes, this spring is remarkable, and *population-level* arrival in New England of things like American Woodcock, Killdeer, Eastern Phoebe, Tree Swallow, and Pine Warbler are all far ahead of schedule (averaging 10-14 days or more in most of the cases I have looked at). However remarkable though this is, these are all birds that would be expected in Massachusetts within two weeks of their actual arrival. There is a temptation to say -- hey, it's an early spring, "anything is possible." But that simply is not true. I am willing to stick my neck out and say that, in this decade anyway, a March Eastern Wood-Pewee, Blackpoll Warbler, Yellow-billed Cuckoo, or Alder Flycatcher is simply NOT possible. These are species that do not even arrive on the Gulf Coast until mid to late April, and their arrival in Massachusetts follows 2-3 weeks behind that (only three of the four have occurred in April, ever, and those only barely). There is concern among scientists that these birds may *never* "learn" to arrive earlier, which could have drastic consequences for those species. My point is that anything is NOT possible, and that despite the remarkable weather, bird migration is also governed to some degree by an internal clock and by the simple challenge of geography. All four of these birds I mentioned winter in South America, and it takes a long time and a lot of preparation (i.e., fattening up) for these birds to make the jump.
Granted, Ruby-throated Hummingbirds, which winter in Mexico and Central America, and increasingly in the Southeastern US from North Carolina to Florida and west to central east Texas, are a different case. Their normal arrival dates are earlier (about 25 April-5 May in New England). But regardless that is about one month from now! Could hummingbirds be one month early? Could any bird?
3) A high percentage of birders use eBird to track ALL their bird sightings -- not just noteworthy early dates and rarities, but every single bird they see, common or rare. This allows us to know not just who is seeing Ruby-throated Hummingbirds, but who is not. A different view of the eBird map for March 2012 shows this map http://ebird.org/ebird/map/rthhum?&gp=false&bmo=3&emo=3&yr=on&byr=2012&eyr=2012, with a bunch of gray cells -- those indicate areas where birders are reporting everything they see, but no hummingbirds. So eBird is collecting a lot of negative data (i.e., a lot of folks that would report a hummingbird if they saw one, are not seeing any). How much data was that based on? The eBird line graphs tell the story -- 23378 checklists from 15-21 March and 27198 checklists from 8-14 March. Looking just at Maryland, New Jersey, New York, Connecticut and Massachusetts, this graph (click "show sample size") http://tinyurl.com/6slfnhs shows how many checklists there were -- 4378 complete checklists for that region last week. Yet none of them reported Ruby-throated Hummingbird.
4) eBird has a rigorous data quality system, with over 370 volunteer reviewers managing several thousand regional checklists that define bird occurrence in each of the twelve months of the year. Any submitted record that falls outside of expectations is flagged for review. I double-checked to make sure there were no Ruby-throateds lost in the eBird review purgatory. There were two, form Indiana and Wisconsin, and one of those lists has many other suspect species on it (Wood Thrush!). In other words, it isn't like the eBird review process is keeping hundreds of hummingbirds hidden.
The New England result of that data quality process? eBird has this graph of Ruby-throated Hummingbird migration, averaged across all years for these states: New Hampshire, Massachusetts, Rhode Island, Connecticut, and New York: http://tinyurl.com/75gt2js. The actual numbers? In early April, 1/12355 has reported Ruby-throated; in the second week of April, 0/12826; in the third week, 15/12778; and in the third week, 308/17522 (giving a whopping 1.27% chance of seeing a hummingbird in any of those states in the last week of April…although this chance increases later in April and is probably higher in lowland areas vs. mountainous ones).
Furthermore, the New England-New York arrival takes a full month to occur, with obvious arrivals continuing through the week of 15-22 May and beyond. So we are at least two months from the end of hummingbird migration into New England.
5) What data quality process does that hummingbird map have? I honestly don't know. It looks like one can sign in, report a hummingbird, and it will show up. The presence of one not valid record provides a sort of confirmation for subsequent ones. Some say a dataset is only as good as its weakest record, and in this case, I worry that there is a sort of "confirmation bias" going on, where one questionable record provides implied support for additional questionable ones. Would a typo in zip code get caught? Would a maliciously false entry get caught?
6) Delving into the literature for this region, here is what I find.
- Cape May, New Jersey (Sibley 1997): extreme early date = 29 March, with bulk arrival not until late April (based on his histogram)
- New York (Levine 1998): "extreme early date = 14 April (coastal and inland)"
- Massachusetts (Veit and Petersen 1993): extreme dates of 18 Mar 1973 in Sudbury (fide Baird) and 26 Mar 1969 in Chatham (Fuller, fide Baird). [Bird Observer database may have additional info]
- New Hampshire (fide Pam Hunt and New Hampshire Bird Records): record early date 8 April, and 14 total April records, with just 5-6 before 26 April
7) I can't help but wonder if the early dates for Cape May, Massachusetts, and eBird (1st week of April) were carefully vetted. All are far enough outside of the main arrival period as to raise questions. They may be correct. But maybe not. No system is infallible, and all of these raise my eyebrows a bit given how far outside of the norm they are.
8) Prior to posting this, I checked with bird records managers in Maine, New Hampshire, Massachusetts, Connecticut, New York, New Jersey, Maryland, and North Carolina. Without exception, they responded with skepticism about the posted arrival dates and had never heard of verified (i.e., documented with photo) or even credible reports of hummingbirds as early as mid-March in their regions. Never, and that's including this year. Early spring or not, all were skeptical that the data could be believed.
9) Can hummingbirds be misidentified? One would think that even a non-birder could recognize a hummingbird, and any hummingbird in the East must be a Ruby-throated, right? Well...not exactly. There are actually a large number of cases of people (nonbirders, usually!) reporting hummingbirds at odd times of year that have turned out to be day-flying months (especially "hummingbird moths", not expected this early either), or large beetles or even bumblebees. Honestly, this might account for some or all of these hummingbird map sightings in March--I have noticed lots of bumblebees recently in my birding (but no hummingbirds).
Also, it must be said that Rufous Hummingbird successfully overwintered in New York City, as well as at numerous points south of there. Every year there are more and they push the boundaries more. What happens when they migrate. Do they all go straight west to Oregon? The "internal clock" for Rufous is to be on the move in the west from late February to late April, with peaks in March in much of California. Should we not take some care to eliminate western hummingbirds before we expand the early arrival date of Ruby-throated Hummingbird by 3-4 weeks?
10) Counterpoints -- There are many: 1) hummingbirds really are easy to identify and almost all nonbirders are qualified to recognize one; bugs and rare western species aside, some significant proportion of the reported Ruby-throated Hummingbirds MUST be correct, right? 2) Birders are but a tiny TINY fraction of the total population. If the hummingbird map reaches 5x, 10x, 100x, or 10,000x more people than eBird or Massbird or the entire, integrated bird reporting network, then maybe they are getting an accurate picture of the extreme ends of the bell curve of hummingbird migration. (Of course, without displaying who is NOT seeing hummingbirds, we have no sense of the sample size to compare to eBird.) 3) hummingbirds are no more likely to turn up on Plum Island as in someone's yard, so the types of spots that birders go looking for birds may not be the best spots to find hummingbirds, which frankly. are probably gardens with flowers and areas where hummingbirds know they can expect a feeder.
Given some of the counterpoints I mention above, and some of the historical early dates reported for Massachusetts, it gives me pause. I know eBird is still a drop in the bucket for what data on bird occurrence *could* be collected if everyone with an interest in birds participated. But eBird and litservs reach people with a real interest in birds--real students of bird identification as it were. Every one of us has some notion of taking care to get our identifications, right? Although not a fair blanket statement (blanket statements are rarely fair!), I am quite certain that your average non-birder would not take as much care with their average hummingbird identification. And I also find myself wishing that the hummingbird map provided photos of some of these extreme early arrivals -- someone must have gotten one, with the current abundance of easy digital photography, right?
I am not currently prepared to say that this map is categorically bogus, but I do hope that the Massbird community will think about hummingbird arrival dates, provide some feedback, and maybe get a photo tomorrow of a clear Ruby-throated to prove all my skeptical thoughts wrong! If so, it would redefine some of my notions about the extremes of the bell curve for migrants, since most species seem to have very short early "tails" in the spring. Eastern Phoebe for example has arrived statewide within 7-10 days of the very first report (see http://tinyurl.com/7oudvhg). If we are seeing the early tail of hummingbirds, then will the rest of the population arrive by 2 April? I am sure it won't (although if this keeps up, I would bet on a record-early hummingbird year).
On Thu, Mar 22, 2012 at 11:52 PM, <BlkVulture at aol.com> wrote:
Like others, I am glad Marshall has brought this up. I've never been sure how to handle the "hummingbird people" with eBird, nor when I was a regional editor for North American Birds. Marshall, did you look at how the Purple Martin data in eBird compare with the martin people? I vaguely remember years ago that there was a martin website similar to this hummingbird site. I think there are some similarities between the two groups.
I have a couple similar assumptions as Marshall, and I also have some skepticism that probably steps beyond what it typically regarded as healthy.
Certainly eBird is more heavily vetted, but neither I nor any other Virginia reviewer have seen a report for RTHU yet this year.
While hard to misidentify, as Marshall notes, it isn't impossible. I suspect that there is also motivation on the part of people that live for hummingbirds. Many of them are quite different from other birders in their zeal, as we've certainly all noticed. I do wonder if in their eagerness to see one, that they aren't particularly careful. Being such a skeptic, I wonder if their eagerness to be the first in their area, or even state, to see one, they just enter a bogus report. I think there is some of that at play.
I've always thought that the coverage that birders offer in terms of likelihood of detection of a rarity is pretty thin in general. I do wonder if the notion that gardeners and random homeowners might be better suited to detect a hummingbird is accurate. How many birders get their feeders out weeks before typical arrival dates? Are we perhaps a bit lackadaisical about it if we know they shouldn't be here until some point in April? I suspect there is some of that. I certainly don't have feeders out yet. I bet far more "hummer people" have them out than birders. All that said, as Marshall said, it seems that a birder would eventually get one of these early birds, if there is widespread early arrival. But are we talking about that? Or are we talking about very few early arrivals? As of about midnight Thursday night, there are only 24 dots on that map north of Virginia and east of West Virginia. Are they all bogus?
One thought that I am curious about is the over-wintering birds along the coast of the Carolinas and such. Is it not safe to speculate that those birds might wander north as things start to bloom, and insects start to fly, and temperatures sail into realms more associated with May, especially those overnight lows? Obviously they cannot forecast the weather, but I do wonder if they can follow leaf-out on trees as an indicator that it is safe to creep north. Hell, maybe their trigger to move isn't solely based on photo-period, and overnight temperature plays a role in it. This has been an exceptional month for warmth, and southerly flow to the wind.
No real answers here, but I do think the map is part bumblebee map, partly fueled by zeal, and part hummingbird map. I also think that for some species that have appeal to non-birders, especially hummingbirds and martins, eBird might be a bit cocky in its presumptions of its ability to track arrivals.
Todd Michael Day
blkvulture at aol.com
In a message dated 3/22/2012 7:25:28 P.M. Eastern Daylight Time, miliff at aol.com writes:
East Coast eBird Guys and Gals,
This is a serious question actually. When do hummingbirds arrive in your state?
eBird Map for March 2012 shows migrants barely north of the Gulf Coast. This matches my impression of reality.
The much publicized hummingbird map is creating buzz in New England that hummingbirds are in CT (and even se. MA now). Could these dates be right? Or is this thing total bogus?
A few assumptions of mine:
1) eBird data are heavily vetted; hummingbird map data are not
2) Hummingbirds are hard to misidentify (in the East)
3) People DO misidentify hummingbirds, and can call large moths AND bumblebees hummingbirds
4) Hummingbirds might be better detected by gardeners and random homeowners than by birders
5) Even if #4 is true, some lucky birder would probably see a few hummingbirds if there was a widespread early arrival
6) there are just two hummingbirds awaiting validation in eBird, from MI http://ebird.org/ebird/view/checklist?subID=S10195928 and IN, http://ebird.org/ebird/view/checklist?subID=S10236040. The latter has Wood Thrush also, which means that the person is clearly not a skilled birder!
So what is going on here. If we have no eBird reports in VA or MD, can we really believe CT, MA, and northern NY ones on 3/14, 3/15, and 3/22 respectively? Central WI on 3/20?
Is this thing just a bumblebee map, labeled as a hummingbird map?
Do we know enough to provide a skeptical view here? Or is eBird too cocky to think that we know al about all birds, including hummingbirds that even my 80-year old non-birding garbage man could identify?
Marshall J. Iliff
miliff AT aol.com
West Roxbury, MA
eBird/AKN Project Leader
Cornell Lab of Ornithology
More information about the CTBirds