Earliest evidence of birds visiting flowers

Angiosperms, or flowering plants, first appear in the fossil record about 160 million years ago. A new paper in Biology Lettersby Gerald Mayr and Volker Wilde (reference below and—I think—a free download) shows that by about 50 million years ago, birds had already evolved to take advantage of this new food source.

Mayr and Wild report a new bird fossil from the famous Messel formation of Germany. The specimen, Pumiliornis tessellatus, is remarkably well preserved as a complete skeleton and is dated at roughly 47 million years, in the middle Eocene.  It is not a member of any of the three modern groups of birds that independently evolved the ability to eat nectar and pollen: hummingbirds, lorikeets + hanging parrots, and some groups of the Passeriformes (“perching birds,” whose nectar-and-pollen eaters include sunbirds, honeycreepers, etc.).

The remarkable thing about this specimen, as shown in the photo below, is that there is a clump of pollen grains near the femur—right where the stomach would be in a living bird.  Although there are also a few insect parts (perhaps accidentally ingested along with the pollen), the number of grains, their clumping, and their position suggests that this bird was in fact eating pollen. Notice the wonderful feather impressions in the fossil below:


Here are some scanning electron microscope (SEM) photos of the fossilized pollen:

Screen shot 2014-05-30 at 6.36.48 AM

The other clue that this bird didn’t accidentally eat pollen, and was adapted to a flower-feeding lifestyle, comes from its appearance. It has a long beak and enlarged nasal openings characteristic of modern birds that sip nectar, and it has “zygodactyl” feet, meaning that the fourth toe could be turned backwards—a trait of perching birds that climb branches and flower stems. The #1 toe is the one you should look at in the photo below:

Screen shot 2014-05-30 at 6.40.42 AM

Now it’s not clear if pollen was the primary object of this bird’s diet, was ingested accidentally while drinking nectar, or if the bird ate both pollen and nectar.  What is pretty clear is that by the middle Eocene, when this bird lived, birds had already evolved to use as food flowering plants that had been around for over 100 million years.

Although the authors were unable to identify the plant that produced this pollen, they suggest that it was already itself evolutionarily adapted to pollination by animals rather than wind:

Although pollen size does not allow discrimination of insect and bird pollination, the large size of the grains and the fact that some are still clumping (figure 2c) indicate direct ingestion from a plant adapted for animal rather than wind pollination.

This pushes back the earliest known bird/nectar/pollen interaction by 17 million years, as heretofore the earliest such specimen dated at about 30 mya.


Gerald Mayr and Volker Wilde. 2014. Eocene fossil is earliest evidence of flower-visiting by birds. Biol Lett 2014 10: 20140223


  1. Maurizio Casiraghi
    Posted May 30, 2014 at 4:56 am | Permalink

    Johnny, giusto per il tuo lato “zoology love”. Guarda questo: interessante. E’ la più antica evidenza di un uccello che si mangia del polline

  2. darrelle
    Posted May 30, 2014 at 5:05 am | Permalink

    That is a magnificent fossil.

    • John Scanlon, FCD
      Posted May 30, 2014 at 8:20 am | Permalink

      One of many thousands from the site!
      This exceptional preservation is because the animals at Messel were killed either by noxious gas (CO2 released from the bottom of the crater lake) or toxic water (seasonal cyanobacteria blooms), and buried in ultra-fine sediment where no scavengers could live. The deposit was never buried terribly deep so the shale is quite soft and moist, and crumbles to dirt if it dries out; to prevent this, fossils are prepared by the ‘transfer’ method. After splitting the shale through the middle of a specimen in the field, each side (part and counterpart) is embedded in a slab of epoxy resin (in earlier times they used wax), and the soft shale then removed (sometimes not very delicately, e.g. scrubbing with a wire brush in the case of common fish fossils). Feathers, hair etc. show up quite well as dark stains against the pale yellowish resin.
      For most kinds of fossil, this epoxy method works beautifully. I found it unsuitable for snakes (where it’s really useful to be able study disarticulated vertebrae in the round; wax would have been preferable), but with modern HRXCT methods that should no longer be a problem.

      • Gregory Kusnick
        Posted May 30, 2014 at 4:12 pm | Permalink

        This does raise an interesting question, though. If collection technology is still evolving, then might science be better served in the long run by leaving some fossils in the ground to await better collection methods?

        I’m imagining some sort of portable CT or 3D ultrasound scanner capable of seeing right inside rock at high resolution. Would future paleontologists equipped with such devices thank us for breaking open and grinding down valuable specimens that they might have extracted more information from?

        • John Scanlon, FCD
          Posted May 30, 2014 at 11:08 pm | Permalink

          Vertebrate palaeontologists have spent time daydreaming about portable scanners for decades, and there are some real examples – the in-ground seismic scan in the first Jurassic Park movie wasn’t pure science fiction. CT-scanning specimens that have been excavated and brought to the lab but not prepared (e.g. still inside plaster field jackets) is easier than doing it in the field, and becoming quite common.
          Most fossil digs are just scratching the surface of large deposits (dependent on natural erosion to expose sparse remains) so there’s plenty left for the future, but of course there are many cases where the deposit is limited (cave and fissure fills, deposits in small ancient pools and channels, burrows, nest sites) and the practice is usually to leave at least half in the ground (or unprepared if it must be removed) for future studies. Leaving it in the ground can be a dead loss in the case of rapidly eroding sites, or salvage digs where the deposit’s going to be dug up for a highway or building foundation.
          The Messel oil-shale was mined for petroleum during WWII, but it was low-grade and inefficient; then commercial fossil collectors had several decades of open slather to play with the site, till the early 90’s when there were plans to use the pit for landfill, stopped by the World Heritage listing. Collecting is now carefully regulated and small-scale (the Senckenberg Institut and Hessisches Landesmuseum are the main groups involved, though there are other important collections), but most of the research being published is based on more detailed study of specimens collected decades ago that have been sitting in museum drawers or private collections.

      • Diane G.
        Posted May 30, 2014 at 11:19 am | Permalink

        Really interesting!

      • darrelle
        Posted May 30, 2014 at 9:35 am | Permalink

        Thank you!

    • gravelinspector-Aidan
      Posted May 30, 2014 at 7:31 am | Permalink

      Look at the scale bars. You could fit this bird onto the flexor muscle at the base of your thumb.

  3. Posted May 30, 2014 at 5:07 am | Permalink

    Just a note of appreciation of the science posts in the mix on WEIT. Beautiful fossil.

  4. Dominic
    Posted May 30, 2014 at 5:19 am | Permalink

    My question when I saw this was how much pollen would a bird need to eat to make a living in that ‘niche’ or would it form just a part of the diet, or only at certain times?
    Was this in a tropical climate where there might be flowering plants all year?

    • Torbjörn Larsson, OM
      Posted May 30, 2014 at 8:03 am | Permalink

      I’m fairly certain there must have been a transitional period where the pollen hadn’t yet evolved large size and clumping and hence become easier to eat.

      Maybe the presence of insects is also a clue on earlier or still essential supplementary feed? E.g. some flowering plants initially coevolved with insects for pollination after wind et cetera mechanical pollination evolved, did they not? So plant and animal could both have relied on insects initially, then starting to cut out the middle man.

      • Jim Sweeney
        Posted May 30, 2014 at 11:15 am | Permalink

        Both hummingbirds and orioles eat insects as a source of protein. Birds do not live by nectar alone!

      • Posted May 30, 2014 at 10:10 am | Permalink

        I was thinking the same thing. Insects were early pollinators, and among the first such pollinators were the beetles. Their was a very well established coevolution between beetles and plants by the Cretaceous. Beetle pollinated flowers provide a lot of pollen for beetles to wallow in. The pollen would be a very attractive source of nutrients for other animals.

    • gravelinspector-Aidan
      Posted May 30, 2014 at 7:28 am | Permalink

      Was this in a tropical climate where there might be flowering plants all year?

      The paper is free to download, but doesn’t particularly address the environment of the Messel deposit. Since it’s a very well-studied deposit, they obviously don’t feel the need to expend column inches on that topic. If you remember some hype a couple of years ago about a novel Eocene primate fossil, Something somethingicus Darwinii, nick-named “Ida” ; that’s from the same deposit.
      Bristol University has a page on Messel’s ecology, including the snippet that “Konservat-Lagerstatte means literally ‘conservation warehouse’” (having tried to find the stationery cupboard in a Norwegian office one day … ‘Lager’ doesn’t just mean a cold, top-brewed amber nectar to me any more), and the more climatological statement that

      Plants – Lake Messel was situated in a dense rainforest, shown by the presence of preserved plant material, including palm leaves, fruits, wood, pollen and some water plants.

      The Unesco record justifying it’s World Heritage status says “The location of the Eocene Lake Messel lay south of its present position. This accounts for the site appearing to have had a tropical to subtropical climate.
      The Senckenberg Museum (somewhere in Germany, probably near Messel) held a conference on Messel in 2011, and make 17MB of conference proceedings available from here. A whole day of the conference studied the deposit’s environment, with papers titled (for example, “Lake Messel, an extraordinary archive for the middle Eocene greenhouse

      In short – it was a warm to hot, damp climate. Seasonality is a more complex question – warm versus cool, or dry versus wet?.

      • John Scanlon, FCD
        Posted May 30, 2014 at 11:23 pm | Permalink

        The Senckenberg’s in Frankfurt. I was at a conference on Messel in 1991 at the Hessisches Landesmuseum in Darmstadt; Messel’s a little place in between those two cities.

        • gravelinspector-Aidan
          Posted May 31, 2014 at 2:30 am | Permalink

          Hmmm, SWMBO is planning on taking us to Bavaria this year – for a family thing – so a day hiding from culture in a good museum …
          I’ll file that datum for use one day, but this trip we’re not likely to get north of Munich.

        • John Scanlon, FCD
          Posted May 30, 2014 at 11:27 pm | Permalink

          I’d somehow missed seeing the 2011 conference abstracts. Right up my alley, there’s one describing a mammal from fossil snake vomit.

  5. Diana MacPherson
    Posted May 30, 2014 at 5:50 am | Permalink

    That is so cool and the pollen is so obvious when scanned like that (or at least comports with my brain’s concept of what it would look like).

    • gravelinspector-Aidan
      Posted May 30, 2014 at 7:29 am | Permalink

      That looks better preserved than most of the pollen that my palynological colleagues dig out of the rocks we drill.

      • Torbjörn Larsson, OM
        Posted May 30, 2014 at 8:13 am | Permalink

        Clearly there is a selection bias here. (No find, no article; et cetera.)

        But if it isn’t all there is to it, I wonder what mechanisms makes animals fossilize vs plant material? Also, the relative lack of oxygen vs the usual pollen fossilization may be part of it.

        • gravelinspector-Aidan
          Posted May 31, 2014 at 2:28 am | Permalink

          In regards of the Messel fossil, then clearly there is a selection bias operating. Less so in our industrial work. For substantial intervals of the well, every (that’s “every”, not “most”) sample from the wellbore, each one aggregating rocks taken from successive intervals of the well, gets passed to the “bug watchers” for study. Of the order of a thousand samples per well, and a thousand fossils per samples, for a million data points per well, each in a reasonably well constrained geological context.
          IDiots really don’t know how much work has gone into developing the science that they blithely dismiss. Not being famous enough to institute a GravelInspector Award along the lines of the Aldrin Award I propose above, I guess I’d have to plead to the judge that I was instituting the Stopes Award – named for the most famous of palynologists.

  6. John Harshman
    Posted May 30, 2014 at 6:36 am | Permalink

    Wait: how do zygodactyl feet indicate a pollen diet? Of all the extant nectar-feeders, only parrots are zygodactyl.

    • Diane G.
      Posted May 30, 2014 at 11:25 am | Permalink

      “Wait: how do zygodactyl feet indicate a pollen diet?”

      I wondered exactly that. Glad I’m in such good company. :)

    • Posted May 30, 2014 at 10:16 am | Permalink

      The modern birds that eat a lot of pollen are hummingbirds, honeycreepers, and sunbirds. This bird does not strike me as being like a hummmingbird, so perhaps it has a closer affinity to one of the other two kinds of birds. Both honeycreepers and sunbirds have a rotated toe for perching.

      • John Harshman
        Posted May 31, 2014 at 6:08 am | Permalink

        Almost all birds have a rotated toe (the first toe). It’s the also rotated fourth toe that makes a bird zygodactyl.

    • Posted May 30, 2014 at 9:20 am | Permalink

      To me it’s not at all clear from the photos that the feet are zygodactyl; a couple of diagrams (at least one made from a tracing of the photo) would help demonstrate this.
      As presented, I think the presumed zygodactyl habit and its implication are both a bit of a stretch. The presence of the pollen and the shape of the beak are more convincing.

      Woodpeckers are zygodactyl; while most aren’t nectar feeders, Gila woodpeckers do visit the flowers of tall cacti. Roadrunners (predators, nearly flightless but strong runners) are zygodactyl. I think this type of foot is more likely to be an adaptation for stability and/or agility rather than indicating a particular diet.

      • John Harshman
        Posted May 30, 2014 at 10:01 am | Permalink

        I think the important point is that both zygodactyly and nectar-feeding are widely distributed within birds, and not only don’t they overlap much, they also don’t coincide in any single instance; that is, if you map both characters onto a tree, they don’t arise on the same branches, and zygodactyly never arises in a nectar-feeding group.

        But I’m willing to take Mayr’s word that the fossil is zygodactyl. There are features of the distal end of the tarsometatarsus that indicate zygodactyly even if the toes of the fossil don’t happen to be arranged that way in death.

    • Torbjörn Larsson, OM
      Posted May 30, 2014 at 8:16 am | Permalink

      Hummingbirds may be an unfair comparison vs perching. =D

      I think they meant an added constraint, not a decisive one.

      • John Harshman
        Posted May 30, 2014 at 9:08 am | Permalink

        I will also point out that most zygodactyl birds, including most parrots, are not nectar-feeders, and that nectar-feeding parrots are nested within non-nectar-feeding parrots, so zygodactyly could not have evolved for the purpose of nectar-feeding. There seems in fact to be no relationship between the characters.

  7. Ron DeBry
    Posted May 30, 2014 at 7:21 am | Permalink

    Cool fossil – but the tone of 1st paragraph of this write-up is odd. “by about 50 million years ago, birds had *already* evolved to take advantage of this *new* food source” – 110 million years old counts as ‘new’?

    • Curt Cameron
      Posted May 30, 2014 at 9:30 am | Permalink

      That was my thought too. It’s surprising to me that it it took more than a million years. But I’m not a biologist, what do I know?

      From a layman’s point of view, I think one of the biggest problems in our understanding paleontology is that we don’t really grasp the timeline. To most laymen, sabertooth tigers and mammoths existed a long time ago, and dinosaurs existed a long time ago, so they’re about the same, right? Our view of time into the past seems to be logarithmic – take one step back 100 years ago, the next step is 1000 years ago, the next step is 10,000 years ago, etc. Getting an understanding of the timeline is something that I’m trying hard to get right.

      • Posted May 30, 2014 at 1:11 pm | Permalink

        Here’s the Geological Society of America geologic time scale chart:

        My high school earth science teacher had students measure out the time scale on a roll of cash register paper. I still think it’s a good way to show the relative length of the Precambrian.

      • Posted May 30, 2014 at 10:22 am | Permalink

        Timelines can really throw you, because diagrams of the timelines are usually not represented in proportion.
        The first 90% of earth history is Precambrian. That is, all animal life is still in the water. Arthropods coming onto land to eat moss is ‘recent’ (!)

        • Diane G.
          Posted May 30, 2014 at 11:36 am | Permalink

          The Brights distribute a pretty cool, representative poster:


          • Curt Cameron
            Posted May 30, 2014 at 12:38 pm | Permalink

            That poster is cool in a way, but in another way it reinforces the problem – the time scale on it is not linear – 30 million years on the left side (the most recent 30 million) takes up the same amount of distance that a billion years takes in the middle, and several billion on the right side.

            When I was in high school, my biology teacher took the class on a long walk, probably a half-mile. He had marked distances along the walk, so that when we got to the farthest point from the classroom, that was the beginning of the Earth. We walked a while back towards the school, and that was the beginnings of life. It was a long walk until we got to the Cambrian. The last bit of the timeline was drawn on a lab table in the classroom, with all human history jammed into the final 1/8 inch. I think about that often.

            I’m not sure that a teacher could do that nowadays, taking the class off-campus. This was about 1976 in conservative west Texas, but I think I got a decent understanding of evolution from that class.

            • Diane G.
              Posted May 30, 2014 at 1:45 pm | Permalink

              I love examples like that, and the teachers that provide them. :)

  8. JBlilie
    Posted May 30, 2014 at 8:13 am | Permalink

    Really cool — and yes, so tiny!

    In a previous home, I had a huge maple tree in my back yard. Each spring when it bloomed, the local gray squirrels would congregate in the tree and stuff themselves on the flowers.

    I had never seen this before.

  9. Posted May 30, 2014 at 10:42 am | Permalink

    Fascinating post, and very well-written. I just wanted to offer a point of clarification: this is not evidence of a bird eating pollen 50 million years ago. It’s evidence that I, as part of my divine mission to create a fossil record as a test of humanity’s faith in God, accidentally smeared some pollen on one of my bird fossils. Sorry for the confusion.

  10. Filippo
    Posted May 30, 2014 at 1:39 pm | Permalink



Get every new post delivered to your Inbox.

Join 37,123 other followers

%d bloggers like this: