Niche differentiation in birds of prey

Niche differentiation in birds of prey

We are searching data for your request:

Forums and discussions:
Manuals and reference books:
Data from registers:
Wait the end of the search in all databases.
Upon completion, a link will appear to access the found materials.

I'm not much of an ornithologist but I know enough to distinguish most Central European birds of prey. To me it is amazing that there are so many species that seem to occupy the same niche. Especially the two native kite species (Milvus milvus and Milvus migrans) and the common buzzard (Buteo buteo) are -- to my impression -- quite similar in terms of size, preferred prey, habitat etc. According to the competetive exclusion principle, these species would not coexist if they would exploit the exact same set of resources. So my question is:

What are the major differences between the mentioned species that allow coexistence? Are there more such examples from other geographical regions?

What Makes Owls So Different From Other Bird Species?

Why are owls so different from other bird species? originally appeared on Quora: the place to gain and share knowledge, empowering people to learn from others and better understand the world.

Answer by Stefan Pociask, wildlife researcher, on Quora:

It’s sunset, and Harold the Red Tailed Hawk has had a long day watching over his field on the edge of the forest. Not a particularly successful day, but he did catch two mice in the grass. Time to call it a day. As he approaches the time-clock to check out of the day-shift and go home, he sees Oscar the Great Horned Owl approaching.

“How’s it going, Harold? Good day?”

“Eh. so-so. There’s a rabbit in the NW corner that got away from me.”

“Amateur. I’ll take him out tonight.”

“I’m sure you will Oscar. Good night, bud. Give em hell.”

They both punch the time clock. The hawk goes to sleep, the owl goes to work.

The territory of both of these predatory birds is the same field. They share it… one during the day-shift (the diurnal hawk), and one during the night-shift (the nocturnal owl). This scenario is another beautiful example of evolution at work. In particular, it’s call niche differentiation. The world is dark half of the time. Nature found a way to make full use of all twenty-four hours by assigning different animals to different times of day. We have the diurnal day hunters, the nocturnal night hunters, and just to make sure to cover the gaps… the crepuscular dawn or dusk hunters. Each has physically adapted to be at his best during certain hours.

The owl has physically changed from his long-ago ancestors, in many ways, to fill their special niche of darkness. We’ll start with the eyes. A hawk is nearly as blind as you and I during the night. An owl’s eyes are much, much bigger than any hawk’s eyes, in order to gather more light from the moon and stars. The eyes of a three pound Great Horned Owl are the same size as the eyes of a three hundred pound man! The eyes take up nearly all of the room inside an owl’s skull. An owl’s eyes aren’t even sphere shaped. They are elongated, shaped like a bulging barrel with rounded ends. This gives them a maximum distance between the front and back of the eyeball. Now this causes one particular problem. Since they aren’t shaped like a ball, they can’t shift from side to side, to look left or right. An owl’s eyeballs are fixed in place. This is why he must rotate his entire head to look left or right, and it’s why the neck bones allow it to swivel two-hundred and seventy degrees around. Most diurnal birds can’t go further than one-hundred and eighty degrees. Plus, owl eyes are on the front of the face, like a human’s not on the sides of the head, like other birds of prey. This makes for excellent binocular vision.

Owl hearing is also adapted to work at night. One ear is higher than the other so that they can triangulate to the exact spot of any noise in the dark. I’ve explained this here: Stefan Pociask's answer to Are all life forms relatively symmetrical?

Also, unlike diurnal birds, their face is shaped like an acoustic dish or bowl that funnels all sounds right to their ears.

Feathers! Yes, they are special for night hunting as well. Owls have very particular feathers with fluffy edges that make less sound when flapping through the air. You can test this by holding all your fingers tightly together, as if saluting, and waving your hand very fast past your ear very close to your ear, like swatting a fly. You will hear a whoosh sound. Now do the same with your fingers spread far apart. The sound will be much quieter (how many of you just smacked your ear by accident? Yeah, I know… it stings!) This is the way that owl feathers allow them to fly silently in the dark.

Speaking of feathers… when our friend, Oscar the Owl goes to sleep during the day, he is highly vulnerable to getting attacked and eaten in his sleep, by any number of daytime carnivores. He’s pretty defenseless while just sitting there in broad daylight, fast asleep. So, unlike diurnal birds, owls have evolved an incredible ability to hide, with camouflage. Check out this guy: [1]

It’s not difficult to imagine how this particular trait evolved. Obviously, the owls who were not born with excellent camo, all got eaten up in their sleep! It only left those who could “hide in plain sight” the best, and the camo kept getting better and better. Now they are invisible and can rest easy during the day.

Next… feet. Here is a hawk’s foot: [3]

The difference is obvious: three i n front and one in the rear for a hawk vs. two in front and two in the rear or three in front and one in the rear for the owl. An owl has the option to switch it up! This gives the owl an advantage of a better grip, when conditions are dark. Also, it gives an advantage to an owl’s favorite killing method… crushing. Owls mostly land on top of prey and stay on the ground while their dinner dies in their grip. Hawks will more frequently swoop down and lift off with their prey, not necessarily crushing it to death right away. Owls normally remain silent during a kill, since they are on the ground and vulnerable. So they need a fast way to kill. The two-by-two talon arrangement helps with this. They can safely make kills at night this way, then swallow their prey whole, if they can. It’s all about silence. A hawk will fly away with the food, so they aren’t that worried about a super quick kill. An owl’s crushing force is multiple times that of a hawk, of equal size. A Great Horned Owl could puncture your skull with all four talons. A Red Tailed Hawk can’t do that.

The bottom line is that owls are different because they live or die based on three things: darkness, silence and hiding (hmmm… ninjas?). Everything about them has evolved toward these three concepts, that other birds don’t worry about.

Owls are extremely specialized creatures. They work second shift. No other bird can do what they do, as well as they do it. These are the things that make an owl so different. There are other nocturnal birds, of course. But none are as perfectly adapted to be such deadly efficient killers. They are pretty cool dudes! As far as the original question comparing them to other bird species… well, they are as different as night and day!

[3] H. Alleyne Nicholson, An Introductory Text-book of Zoology(Edinburgh:William Blackwood and Sons, 1871) 152 Falcon Foot | ClipArt ETC

[4] The Alaska Owlmanac: A guide to the identification, habits, and habitat of ten owl species found in Alaska. Alaska Department of Fish and Game

This question originally appeared on Quora - the place to gain and share knowledge, empowering people to learn from others and better understand the world. You can follow Quora on Twitter, Facebook, and Google+. More questions:

You might also Like

Niches are a very important part of any ecosystem. It is not true that there can only be one specific animal in one specific niche – however, having more than one often creates competition which can lead to major ecosystem problems

I guess what I mean to say is that while it is possible for two different species to fill the same niche temporarily, it is not advantageous to either species.

That is part of the reason that many of us feel that we should conserve wildlife habitats as much as possible – that way animals who no longer have homes in their own niches are not moving in on other’s niches, creating big problems.

I think a simple way to put what a niche is this it is a specific place that one particular organism or species takes up in their world. In their own little ecosystem they have a very specialized job which without them would not be fulfilled.

Haven’t you ever heard about a person ‘finding a niche in life?’ Well, this has much the same meaning, just in a different context.

For instance, in my area of the world we have a lot of deer. They are everywhere! This has a lot to do with the fact that my area provides them a great ecosystem to live in. Lots of woods, water, and food!

Because we have plenty of deer, we also have plenty of deer hunters who use the animals for meat.

Beyond that, we also have buzzards that are there to help clean up the mess. Each one of those animals listed –people, deer, buzzard – have their own niche in the world that I live in.

A niche is specific thing that a specific organism needs for survival. Not two different organisms can share the same niche or competition will arise. In a nutshell, it's something an organism needs for survival. -Drew anon122075 October 26, 2010

A niche is a cycle an organism goes through every day to survive. xoxo anon121456 October 24, 2010

whats the niche of an onmivore? anon116453 October 6, 2010

We learned about a niche in class. It is anything that an organism uses or needs to be able to survive. Trust me I'm in honors biology. anon87039 May 27, 2010

i still don't really understand what a niche is. could someone explain it better? anon24652 January 15, 2009

On the relationship between niche and distribution

Institute of Ecology, University of Georgia, Athens, Georgia, 30602, U.S.A.

Institute of Ecology, University of Georgia, Athens, Georgia, 30602, U.S.A.


Applications of Hutchinson’s n-dimensional niche concept are often focused on the role of interspecific competition in shaping species distribution patterns. In this paper, I discuss a variety of factors, in addition to competition, that influence the observed relationship between species distribution and the availability of suitable habitat. In particular, I show that Hutchinson’s niche concept can be modified to incorporate the influences of niche width, habitat availability and dispersal, as well as interspecific competition per se. I introduce a simulation model called NICHE that embodies many of Hutchinson’s original niche concepts and use this model to predict patterns of species distribution. The model may help to clarify how dispersal, niche size and competition interact, and under what conditions species might be common in unsuitable habitat or absent from suitable habitat. A brief review of the pertinent literature suggests that species are often absent from suitable habitat and present in unsuitable habitat, in ways predicted by theory. However, most tests of niche theory are hampered by inadequate consideration of what does and does not constitute suitable habitat. More conclusive evidence for these predictions will require rigorous determination of habitat suitability under field conditions. I suggest that to do this, ecologists must measure habitat specific demography and quantify how demographic parameters vary in response to temporal and spatial variation in measurable niche dimensions.


2.1 Study area

Fieldwork took place during the austral spring (i.e., breeding season, November–December, 25–45 days/year) in 2014, 2016, and 2017 and during fall (i.e., nonbreeding season, May, 20 days) 2017 in Franklin Bay at the southwestern coast of Isla de los Estados (Staten Island), Argentina (54°85′30S, 64°83′90W). The island is 540 km 2 and is separated from the Tierra del Fuego Main Island by the 24 km wide Le Maire Strait. The climate is cold, humid, and oceanic, with winds mainly from the SW and a variable rainfall regime, ranging from 700 to 2,900 mm/year depending on the site (Morello et al., 2012 ). Mixed forests of Evergreen beech (Nothofagus betuloides) and Winter's bark (Drymis winteri) cover most of the island, but along the fjords and coasts, a grassland vegetation typical of Subantarctic islands is present. There are over 100 species of birds and several marine mammals. The island is an important site for South American Sea lions (Otaria flavescens) and Fur seals (Arctocephalus australis), both of which are recovering from past exploitation (Milano et al., 2020a , 2020b ). There are no native terrestrial mammals apart from the Chuanisín mouse (Abrothrix [Angelomys] xanthorhina), but rats (Rattus sp.), feral goats (Capra hircus), and red deer (Cervus elaphus) are commonly observed introduced species (Massoia & Chebez, 1993 ).

Within its tussock (Poa flabellata) grasslands, our 4 km 2 study site at Franklin Bay holds a large southern rockhopper penguin colony with 127,000 breeding pairs, plus 4,600 breeding pairs of imperial shag (Leucocarbo atriceps) and 1,600 Magellanic penguin (Spheniscus magellanicus) breeding pairs (Raya Rey et al., 2014 ). Grazing pressure by introduced herbivores (goats since 1856 and red deer since 1974) has apparently restricted nest site availability for caracaras, and some rockhopper penguin subcolonies have no associated caracara nests (Balza et al., 2017 ). Caracaras are the most abundant scavenger on the island (Frere et al., 1999 ), being over six times more abundant than the second most abundant species (i.e., southern crested caracara, Caracara plancus, UB unpublished).

2.2 Pellet analyses

On Isla de los Estados, caracaras are the most important predator of rockhopper penguin chicks (Liljesthröm et al., 2008 ), but no information on other prey items was available. To establish the prey items potentially included in the stable isotope mixing model analysis (see below), we analyzed pellets from nests and identified their remains. This technique is biased over prey that leave hard remains (e.g., hairs, feathers, exoskeletons), and, as in our case, are generally encountered in the vicinity of nest sites (Marti et al., 2007 Redpath et al., 2001 ). Each year, we searched for active caracara nest sites by walking systematically through the study area and observing territorial behavior of breeding caracaras (for details see Balza et al., 2017 ). The number of accessible, active nests found in each year was 11–13 and represented

70% of the observed breeding population. At first observation, the caracaras were in either the late incubation or early chick rearing stage. Pellets were dried and analyzed following Marti et al. ( 2007 ) and Rexer-Huber and Bildstein ( 2013 ). Hairs in pellets were identified following Chehébar and Martín ( 1989 ) complemented with a reference collection for deer and goats from the study area. Feathers, eggs, and bones were identified with a reference collection of adults, chicks, and eggs from the breeding species listed above as well as geese (Chloephaga picta) and gull species that also breed on Isla de los Estados.

2.3 Blood and feather collection

1 ml) were collected from the brachial vein of

20-day-old chicks (43 individuals from 17 nest sites 1–3 chick/nest*year) captured manually, and from juveniles, immatures, and adults during the breeding (n = 8) and nonbreeding (n = 8) seasons captured with walk-in and noose traps, and later stored in 70% ethanol (Hobson et al., 1997 ). Age of individuals was determinate by plumage cues (Strange, 1996 ). We used the mean value of each nest for those with more than one chick, obtaining 8–10 independent samples/year. Also, as in some cases we collected samples from the same nests in multiple years, when we estimate overall isotopic niche parameters for chicks, we use the mean isotopic values for each one of the 17 nest sites. All 59 captured birds were banded with plastic rings (Ecotone, Poland), and no individual was sampled twice during the study period. To obtain floater and breeding adult samples, we collected molted wing feathers and classified them in relation to their distance from the nests. When collected from nest sites, we assumed it was molted by a breeding adult (n = 13, one feather/nest) and when collected >300 m apart from any active nest, by a floater (n = 63). Caracaras nest in a nearly colonial arrangement with very small breeding territories (Strange, 1996 ). The >300 m threshold was assumed not likely to represent breeding adult samples because observed foraging of the breeding adults was mainly associated with the nearest penguin patch (i.e., median < 50 m and in all cases < 200 m) and floaters are two- to fivefold more abundant than breeding adults (UB unpublished). Therefore, we assume a distance of >300 m from any known nest site is an area unlikely to be used by a breeding adult. Feathers were identified as belonging to adult birds (i.e., >5 years old) following Strange ( 1996 ). Floater abundance was 92 (95% CI 62–139) individuals in 2018, and since we obtained 11–29 samples/year, we assume no double sampling in this part of the population either. Molting of feathers in the study area was only observed during the breeding season for both floaters and breeding adults, and thus, we assumed that feathers are synthetized during the period of rockhopper penguin presence. Samples used are summarized in Table 1.

Trophic category Tissue sampled Group n Aim
Selective Blood Juveniles, immatures, and adults in breeding season 8 Isotopic niche width estimation and mixing model analysis
Opportunistic Juveniles, immatures, and adults in nonbreeding season 8
Selective Chicks (nests) 8−10/year
Selective Molted wing feathers Breeding adults 13
Opportunistic Floaters 11−29/year

2.4 Prey sample collection

To describe the potential prey resources for caracaras for building mixing models, we collected tissue samples of representative prey based on prey remains observed in pellets, published literature, and field observations (Catry et al., 2008 Rexer-Huber & Bildstein, 2013 ). From 2017 to 2018, we collected samples from marine and terrestrial prey on Isla de los Estados (Table 2). Mussels were collected manually from the intertidal during low tide. Birds and invasive mammal samples were collected from fresh dead animals in our systematic surveys along the shores and at seabird colonies. Recently abandoned eggs were collected manually. Rodents were collected using Sherman-like traps, and insects were collected using pitfall traps. Sea lion feces observed to be eaten by caracaras were collected in the nonbreeding season Observatorio Island, 40 km to the NE of our study area. All other samples were collected in Franklin Bay during the breeding season.

Trophic web Species Tissue n δ 15 N δ 13 C
Marine Mussel (Mytilus edulis) Muscle 5 10.8 ± 0.6 −14.7 ± 0.3
Rockhopper penguin (egg) Egg membrane 3 9.6 ± 0.4 −21.4 ± 0.5
Rockhopper penguin (chick) Muscle 11 9.7 ± 2.5 −21.8 ± 1.7
Rockhopper penguin (adult) Muscle 2 8.0 −22.9
Imperial shag (egg) Egg membrane 3 15.7 ± 0.7 −15.1 ± 0.4
Imperial shag (chick) Muscle 6 14.8 ± 1.0 −16.8 ± 0.4
Sea lion Feces 3 15.4 ± 0.8 −18.8 ± 0.8
Terrestrial Red deer* Muscle 3 4.6 ± 4.4 −24.9 ± 0.1
Goat* Muscle 1 3.3 −24.7
Goose (Chloephaga picta) Muscle 2 12.2 −30.4
Rat (Rattus rattus)* Muscle 3 9.5 ± 4.7 −19.8 ± 4.2
Chuanisín mouse Muscle 3 4.7 ± 2.3 −23.9 ± 2.2
Beetle (Ceroglossus suturalis) Muscle 6 5.7 ± 5.2 −27.9 ± 0.8

2.5 Stable isotope analysis

Stable isotope (SI) analysis provides useful insights on species trophic ecology that avoids many of the biases of traditional diet study methods (Bearhop et al., 2004 ). Isotopic ratios for those elements that are incorporated through diet can be interpreted as a reflection of consumer's food webs pathways (Chisholm & Nelson, 1982 Hobson & Clark, 1992a ) and trophic level (Minagawa & Wada, 1984 ). However, consumer SI values also reflect spatial and temporal variation in food sources' SI values and thus are not necessarily equivalent to niche variation (Matthews & Mazumder, 2004 ). Also, the SI values of consumer tissues are context dependent, and quantifying baseline information is important when applying this technique in new study sites/species (Phillips et al., 2014 ).

SI analysis allows testing hypothesis of OFT for two reasons: First, it provides quantitative, individual-level and temporally integrated data. Therefore, diet variation considers temporal consistency and is not a snapshot of the diet of individuals (Novak & Tinker, 2015 ). Second, intraspecific variation in resource use is reflected by shifts in consumer tissue isotope ratios in a predictable way (Hammerschlag-Peyer et al., 2011 ). Additionally, SI mixing models can be useful to detect the importance of prey such as carrion that are not well represented in classic techniques.

where X is 13 C or 15 N and R is the corresponding ratio 13 C/ 12 C or 15 N/ 14 N. The Rstandard values were based on Vienna Pee Dee Belemnite (VPDB) for δ 13 C and atmospheric N2 (AIR) for δ 15 N values.

Since physiological, tissue-dependent traits are known to be relevant for isotopic ratios (Hobson & Clark, 1992b ), we needed to assume the differential factors acting for different tissues. As a consistent linear relation between blood and feather samples in bird chicks with marine diets exists, we normalized blood stable isotope values to reflect feather stable isotope values (Cherel et al., 2014 ), to further compare blood from chicks with feathers of breeding adults and floaters.

2.6 Isotopic niche metrics and statistics

We compared isotopic niche width and overlap among groups to test hypotheses regarding trophic expansion/reduction dynamics and changes in resource use between seasons (Hammerschlag-Peyer et al., 2011 ). We first quantified isotopic niche width, which is a common metric used to quantify variability in trophic diversity and resource use (Bearhop et al., 2004 Newsome et al., 2007 ). We calculated Bayesian standard ellipse areas (SEAB) using the SIBER package in R software (Jackson et al., 2011 R Core Team, 2018 ). SEAB are iteration-produced, posterior probabilities of the 2-dimensional isospace of the groups that allow comparison between unbalanced sample sizes (Jackson et al., 2011 ). For each model, we ran 10,000 Markov chain Monte Carlo iterations, discarding the first 1,000 of the analysis with default priors. For posterior comparisons, we tested the probability of one group's SEAB being bigger than the other group by comparing the proportion of posterior ellipses (PP) that differed between groups. We considered PP ≥ 0.95 to reflect relevant differences in SEAB. Interannual variation was studied for chicks and floaters only, because the breeding adult sample size was too low, and thus, only a pooled analysis was used for them.

To compare overlap in resource use among groups and between seasons, we estimated the probability of individuals in one group to be contained in the ellipses of another using the nicheROVER package (Swanson et al., 2015 ). Overlap values range from 0 (i.e., no overlap) to 1 (i.e., complete overlap). To test the occurrence of nested patterns, we looked for differences in the 95% credible intervals (CI) of the estimations among reciprocals. For example, for supporting the hypothesis of group A being a subset of group B, we looked for asymmetry in overlap, meaning that individuals in group A are more likely to be encompassed in the ellipses of group B than vice versa.

2.7 Mixing model analysis

We built Bayesian stable isotope mixing models using the MixSIAR package (Stock et al., 2018 ). We separated our caracara samples into groups according to age, season, and breeding status. Stable isotope mixing models can be sensitive to the trophic discrimination factors (TDF) used (Bond & Diamond, 2011 ), and having the consumer data included in convex hulls is a necessary though insufficient condition for mixing models to work properly (Phillips et al., 2014 ). For this, we used the method described by Smith et al. ( 2013 ) to simulate stable isotope mixing polygons and to select TDF sources that would allow for a suitable mixing model. Depending on the tissue and age class considered, we contrasted up to four TDF sources: from a related species (peregrine falcon, Hobson & Clark, 1992b ), from a scavenger bird of prey (California condor, Kurle et al., 2013 ), from a subpolar raptor (snowy owl, Therrien et al., 2011 ), and from a meta-analysis-derived TDF using the SIDER package (Healy et al., 2017 ) (Table S1). We ran the mixing models with all suitable TDF to explore possible effects on the election of TDF on final output (Figure S2, Table S4).

Also critical to the performance of mixing models is the election of priors. Informative priors are recommended, when information is available, to constrain the output of indeterminate models (Phillips et al., 2014 ). They can accurately describe the diet input in some cases (Chiaradia et al., 2014 ), but they can also produce poor model performance when pellet/scat analysis are used, because they tend to overestimate the importance of prey with indigestible parts (Swan et al., 2020 ). In our case, we first used pellet analysis to constrain the selection of potential prey for breeding adults and their chicks, assuming that potentially important prey types should occur at least once in this analysis (Table 3). Then, we used informative priors based on abundance of prey types for breeding season models only, which were available in published works for seabirds (Raya Rey et al., 2014 ) and our own estimations for geese (UB unpublished). Rockhopper penguins are 27- and 85-fold more abundant than shags and geese, respectively, in our study area. We set our informative priors to reflect a 5% minimum importance of all prey types other than rockhopper penguins, therefore setting a precautionary underweighting of rockhopper's signature in the starting point of the models. For the nonbreeding season model, we used uninformative priors because of the lack of more detailed information. Deer and goat were combined in one signature as we had only one sample of the latter. The potential prey used in each model are detailed in Table 3. Following Phillips et al. ( 2014 ), we combined sources a posteriori into “terrestrial” and “marine” to distinguish between these two trophic pathways. Because marine and terrestrial prey were not evenly sampled (e.g., for the nonbreeding, season four terrestrial sources and two marine sources were used), even the “uninformative” prior models were informative of marine and terrestrial input (see Phillips et al., 2014 ). Depending on the model, initial terrestrial input varied from 9% to 67% (Table S4). For chick's models, we used nest id as random effects. We ran all our models with 300,000 Monte Carlo iterations. We checked whether the models converged with two different diagnosis statistics (Stock et al., 2018 ), and we informed all plausible models and ranked them using deviance information criteria (DIC, Ando, 2010 ) (Table S4).

Trophic category Model (group, tissue) Marine signature components Terrestrial signature components Justification
Selective Chicks, Blood Breeding adults, wing feathers Rockhopper penguin and imperial shag (eggs, chicks and adults) Insects and geese No evidence of terrestrial prey other than insects and geese by pellet analysis (Table 3). Although caracaras can also predate chicks of the other breeding seabird in the area, the Magellanic penguin (K. Harrington com. pers.), we have no records of such behavior in our site (see Results, pellet analysis) and this species is the less abundant seabird breeding in the study area. Therefore, we assume its importance to be no significant
Opportunistic Floaters, wing feathers Insects, geese, deer, goat, and rodents Uncertainty about prey taken all observed and potentially important prey sources included in terrestrial items
Selective Breeding season, blood
Opportunistic Nonbreeding season, blood Sea lion feces and mussels Deer, goat, insects, and rodents No seabirds and geese available during nonbreeding season. Association with pinniped feces are “probably the most important source of food in the feeding cycle of Phalcoboenus in the winter” (Strange, 1996 ), and caracaras feeding on sea lion excreta were observed in nearby Observatorio Island in the nonbreeding season, where the sea lion samples were collected. Bivalves observed as prey in other populations (Catry et al., 2008 Rexer-Huber & Bildstein, 2013 ) and in our case, although available all-year round, are considered potentially important only in the absence of seabird colonies


The guidelines of the EU Habitats and Bird Directives make provisions to ensure the protection of wildlife against WT structures and recommend wind projects to be preceded by impact assessment studies and succeeded with post-construction (baseline) collision monitoring programs to determine impacts on wildlife at the project sites 37 . We used long-term collision detections from wind farms in the state of Brandenburg for the assessment of the worst hit groups of birds at WTs – Buntings, Crows, Larks, Pigeons, and Raptors. The main intent behind our examination was to assess to which particular land-use types and at what distances to these land-use types do WTs promote or reduce the collision risk. Distances are often required when policymakers ask for information ensuring safe deployment of WTs. Therefore, the results can be helpful in showing the increase and decrease of the collision risk at distances in the immediate vicinity or distant away from specific land-use types, thereby facilitating proposing safer placements of WTs in the landscape. Therefore, we analyzed the carcass detections in relation to the local landscape, specifically against the distances between and within multiple land-use types to the WT sites, to ascertain special combinations of distances leading to a higher risk of bird collisions.

The marginality coefficients for each group depict strong relationships between the turbines where carcasses have been detected and the following key land-use types: fields and other arable lands, forests and forestry areas, green and open areas outside human settlements and grassland and forb areas. With increasing or decreasing absolute values, signifying proximity with respect to the sign (inside−, outside+ announcing the direction). It is noteworthy that the proximity of the detections (group-wise) to particular land-use types on which our collision sensitive niche analyses (group-wise) are based, are alike.

The marginality factor of the data from Raptors further suggested higher importance of distances between turbines and green, open areas in and around human settlements as well as distances of turbines to forests. These findings are reflected in their observed carcass detections at turbines closer but outside the borders of forests and forestry areas up to distances of 2000 m (Annex: Figure A3) and is in line with expectations based on Raptor proximities to forests and forestry areas that provide them with suitable nesting and breeding places 38,39 . Our results are also in accordance with the minimum distances of wind turbines to breeding sites of Raptors as recommended by the Working Group of German State Bird Conservancies, based on species-specific telemetry studies, collision data, functional-spatial analyses, long-term observations and expert assessments, taking into account the risk of collision, avoidance and barrier effects caused by wind turbines 40 .

Raptors are also highly abundant in the fringe zones of infrastructures 41 , primarily due to adequate hunting options 42 , especially of many human-commensal small mammals 43,44,45 and the availability of roadkill carrions 46 , with observed carcasses at turbines situated from their borders between 400–2400 m distances (Annex: Figure A3). They are also observed using features of the urban landscape, such as trees adjacent to open covers, fences and buildings, as shelter from wind, pollution, domestic predators, and concealment in ambush attacks on their prey and for purposes of perching, utilizing new and artificial nesting substrates 47,48,49,50 . Pigeons likewise, another abundant bird species in built-up environments, have also adapted their nesting requirements and foraging habits to be conducive with the urban lifestyles 51 and particularly, green and open areas and urban parks surrounding heavily urbanized areas, settlements and infrastructures have higher densities of these species, as they take advantage of food discarded by humans favoring a more stable presence 52 , explaining the increase in Pigeon carcass detections at turbines closer to their borders, with detection primarily observed between 1000 m and up to 1700 m (Annex: Figure A3).

The marginality and specialization factorial axes of all the bird-groups also indicate strong relationships with distances to arable lands, highlighting their impact in limiting their collision sensitive niches. In case of Raptors, their associations with certain elements of the agricultural landscapes, especially arable lands and open fields, is primarily because of hunting facilitated by mowing or use of low-stature crops 53 , exposing preys to aerial predators 54 . Moreover, the fallow land at the mast foot provide suitable small-mammal habitat in the agricultural landscape, irrespective of low- or high-stature crops 55 . Placement of WTs generally has to follow many criteria the site under consideration should have a strong potential for wind and should neither be near to settlements nor to areas of important habitats for birds or protected species that could be harmed 56 . With the reluctance of local people to install WTs near their homes, project developers often attempt to install wind energy facilities on agricultural land, particularly on arable land dominated by open fields 57 . These areas are also characterized by large plots of grassland or large fields of crops. Therefore, we can find almost all of the already constructed WTs inside of fields or open grasslands. This spatial preference also adds on to the ecological affinities certain bird-groups, particularly Larks show towards open landscapes. They avoid tall, dense vegetation cover 58 , and nest and forage in open agricultural fields, that influences most of their habitat preferences and reproductive success 59,60 , which in turn increases their risk of colliding with the turbine structures closer to the borders of fields, grasslands and open areas. With carcasses detected near to wind turbines situated between −400 m up to 100 m distances from the borders of fields and majorly detected between 300 m and 700 m distances from the borders of grasslands and open areas (Annex: Figure A3).

ENFA results also show that Raptors have the lowest global specialization value in comparison to the other bird-groups and also a comparatively larger niche breadth as per Hulbert’s niche breadth analysis. The ENFA analyses and the LDA analyses also denote that the coverage of the collision space by Raptors is larger compared to that of the other bird-groups, explaining their asymmetrical niche overlap with the other bird groups. Raptors have a greater home range 61,62 as compared to many other birds of smaller size, and venture across distances to utilize perch and prey availability 49 . This indicates that the greater Raptor overlap is either an effect of the comparably larger parameter space covered by the Raptors or a better coverage of the detections in the study area because of their larger sample size, i.e. the exceptionally high number of Raptor carcasses detected at WTs in comparison to other smaller birds. This is primarily due to higher searcher efficiencies in combination with longer carcass persistence times for Raptors 13 .

The least observed niche overlaps based on turbine sites where collisions were detected show that the rather restrictive collision niche of Buntings has an insignificant overlap with the collision niches of other bird-groups, especially Crows. Crows being generalist omnivores 63 and Buntings being shrub-land specialists 64,65 , mostly show niche differentiations on grounds of their specific preferences towards proximity to green and open areas in and around settlements and proximity to shrub-lands respectively. This is in accordance with our pairwise discriminant analysis, showing turbines with Bunting and Crow detections having fundamental niche separations related to the distances to the edges of shrub-lands (favoring Bunting detections) and green areas around human settlements (favoring Crow detections). These results are also consistent with ENFA, where Buntings show higher global specialization values as compared to other groups.

Overlaps of the respective collision niches of the bird-groups indicate similar sensitivities of birds to the multiple land-use combinations, whereas niche differentiations indicate the reverse. Niche overlap is often used to indicate potential for competition between species 33,35,66 . However, in this study, with respect to renewable energy infrastructure the overlaps between species provides insights into their similar or disparate sensitivities to distances from different land-use types that allow directing safer turbine positioning for protecting multiple bird-groups at once as well as for targeting specific groups with limited overlaps with other groups.

DNA analyses as a tool for the protection of species

The example of Manx Shearwater (Puffinus puffinus) seabirds living in Mediterranean regions shows the practical use of these new findings in the systematics of birds. DNA analyses identified the two subspecies P. p. velkouan and P. p. mauretanikus as two distinct species, which led the protective status of the birds to be raised from Birdlife International’s SPEC category 4 to SPEC category 2.

Manx Shearwater (Puffinus puffinus) (Photo:


This research was supported by the Keren Kayemeth LeIsrael (KKL-JNF), the Ministry of Science and Technology (Israel), Smaller-Winnikow Foundation, Israel Nature and Parks Authority (INPA), Hoopoe Foundation (SPNI), Rieger Foundation (USA), Raptor Research Foundation (RFF), the International Center for the Study of Bird Migration at Latrun (Israel) and Kfar-Etzion field school (Israel). GB and KMM were supported in part by NASA grant #NNX11AP61G, US National Science Foundation grant #IOS-1145952 and the Jacob and Lena Joels Memorial Foundation Visiting Professor Award. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation. We thank many people for their useful advises and helpful analysis, including O. Spiegel, R. Harel, Y. Orchan, R. Nathan, Y. Bartan, and S. Rotich (Movement Ecology Lab-HUJI) R. Rabinovich, and R. Biton (HUJI) Y. Nissan, R. Talmor, A. Sherman (KKL) F. Kuemmeth and W. Heidrich (E-Obs GmbH) We are grateful to D. Brand, Y. Rozental, H. Mazar, D. Alon, N. Leader, Y. Malichi, O. Hatzofe, for their generous logistical support and research permits. We thank our accompanying committee: U. Motro, A. Lotem, and G. Katzir for their useful guiding. We are grateful to many people who helped during fieldwork, including U. Kaizer, Y. Ben-Ari, O. Sulimani, D. Gad, Y. Miller, G. Perlman, Y. Perlman, Y. Kiat, J. Meyrav, D. Oz, A. Saieg, Y. Motro, D. Oberman, I. Eitan, I. Ben-Dov, Y. Milo, G. Milo, R. Izraeli, Y. Friedemann, B. Porat, D. Gavish, G. Kali, B. Rinat, A. Naor, Y. Sela, E. Haklai, A. Ben-Gigi, N. Laufer, Y. kaufman, E. kaufman, Y. Siman-Tov, U. Levin, D. Ashkenazi, O. Ben-Shabat, T. Levi, D. Barel, A. Goldfarb, I. Shifman, Y. Shamir, H. Ben-Yaakov, B. Trooper, and R. Milgalter. This paper is dedicated to the memory of R. Trabelsy.


The role of light in structuring the ecological niche remains a frontier in understanding how vertebrate communities assemble and respond to global change. For birds, eyes represent the primary external anatomical structure specifically evolved to interpret light, yet eye morphology remains understudied compared to movement and dietary traits. Here, I use Stanley Ritland's unpublished measurements of transverse eye diameter from preserved specimens to explore the ecological and phylogenetic drivers of eye morphology for a third of terrestrial avian diversity (N = 2777 species). Species with larger eyes specialized in darker understory and forested habitats, foraging manoeuvres and prey items requiring long-distance optical resolution and were more likely to occur in tropical latitudes. When compared to dietary and movement traits, eye size was a top predictor for habitat, foraging manoeuvre, diet and latitude, adding 8–28% more explanatory power. Eye size was phylogenetically conserved (λ = 0.90), with phylogeny explaining 61% of eye size variation. I suggest that light has contributed to the evolution and assembly of global vertebrate communities and that eye size provides a useful predictor to assess community response to global change.

1. Introduction

Light is a pervasive component of ecosystems, producing the tapestry of optical sensory environments through which organisms navigate their daily lives. Despite exhaustive research on how light structures plant communities [1], the role of light in defining the ecological niche for vertebrates remains relatively enigmatic [2–4]. This is surprising given that vertebrate biodiversity persists across vast climatic and habitat gradients that vary widely in ambient light conditions [5,6]. Identifying functional traits linking light to the ecological niche will help more precisely quantify local community assembly, regional patterns in niche packing and species turnover, and the drivers of biodiversity loss in the face of global change [7–9].

Birds rely heavily on vision to detect food and predators, and the eye is the primary external anatomical trait specifically adapted to sense light [10]. For species operating at the extremes of optimal resolution, extraordinary adaptions in retinal anatomy allow individuals to hunt nocturnally (e.g. owls), target fast-moving and distant vertebrate prey (e.g. raptors), and forage in aquatic environments [11–14]. The vast majority of terrestrial avian biodiversity, however, navigates dramatically shifting mosaics of colour and luminosity that interact with structurally complex habitats [15]. Even within relatively dark forests, irradiance can vary across orders of magnitude, depending upon location within canopy strata, the position of gaps and abundance of sun flecks [5]. Despite comprehensive knowledge of ecomorphological relationships for traits related to movement and diet [16], little research has explored the one functional dimension specialized for light: the eye.

While the avian eye possesses micro-anatomical features exquisitely adapted for interpreting variation in colour and brightness, eye size is a useful trait for full community studies, because it directly relates to focal length and image resolution and can be quantified efficiently in the field or museum [3,17–19]. Specifically, larger eyes have more retinal ganglia cells, collect more light and are thought to expand the perceptual range by improving visual acuity and sensitivity to contrast [20–22]. Evolving a larger eye imposes significant trade-offs, however. Increased metabolic investment is required to maintain anatomical structures and process neurological information [23,24]. Birds have relatively large eyes and brains compared to other vertebrate taxa, posing spatial constraints within the cranium and aerodynamic consequences for maintaining flight performance [4,20,25–29]. While the pupil and optical adnexa deter optical damage in bright conditions, large eyes are thought to be more susceptible to overexposure or ‘disability glare’ that can impact the detection of food or predators [30,31]. Given such evolutionary trade-offs, residual variation in eye size after correcting for body mass allometry should represent an adaptive trait related to the optical environment [32].

Variation in residual eye size has been linked to multiple ecological factors for small subsets of the terrestrial avifauna. Eye size is strongly correlated with the actual light intensity micro-environments used by free-ranging birds [33] as well as several behavioural traits related to ambient light levels, including initiation of singing at dawn, arrival at feeding stations in the early morning and prevalence of nocturnal foraging [12,34–36]. Because visual acuity tends to increase with eye size, large-eyed species appear more responsive to experimental predators [37] and more likely to engage in aerial foraging manoeuvres [33,38]. The relationship between eye size and habitat disturbance is less clear, with some evidence that species with larger eyes disappear from brightly lit agricultural landscapes and forest edges [33,39]. Avian reproductive phenology is also potentially affected by the interaction of artificial light and eye morphology [40,41]. To date studies have been restricted to single communities or comparative efforts with less than 200 species, and the lack of a large-scale comparative analysis across the avian phylogeny has hampered efforts to make generalizations about the contribution of eye morphology to global patterns in community assembly.

Here, I use a dataset of eye measurements collected from preserved specimens by Stanley Ritland (SR) [32] to explore the ecological and phylogenetic correlates of eye morphology for a third of terrestrial avian diversity (N = 2777 species). While Ritland provided an initial draft of ecological correlates at the family level in his dissertation, I expand upon his work by (1) examining the ecological correlates of eye size at the species level within a modern phylogenetic framework, (2) including macro-ecological variables reflecting contemporary knowledge of species range sizes, latitude and life history, (3) comparing the explanatory power of eye size with a suite of morphological traits related to movement and diet, and (4) partitioning the phylogenetic and ecological contributions to eye size variation.

Regarding ecological correlates, I first hypothesized that residual eye size is correlated with habitats that constrain the quantity of light, with the prediction that species with larger eyes specialize in forests and forage in the understory where increased visual acuity is required to overcome dark foraging environments. I further predicted that increased eye size would be associated with ranges centred in the tropics, a region housing some of the darkest forests on Earth [42]. Second, I hypothesized that eye size varies by foraging behaviour and diet due to the role light plays in mediating food identification and capture, with the prediction that species employing long-distance (hyperopic) foraging manoeuvres and pursing arthropod or vertebrate prey would have larger eyes associated with heightened visual acuity. Third, I predicted that species with larger eyes would exhibit non-migratory tendencies and smaller range sizes due to less varied optical environments and the aerodynamic advantages of minimizing eye mass during long-distance flight.

2. Databases

(a) Eye morphology

I extracted measurements of transverse and axial diameter (AD) from SR's unpublished dissertation [32]. All measurements were collected by SR from specimens of whole eyes preserved in formaldehyde and/or alcohol using 0.05 mm Vernier callipers. I focus on transverse diameter (TD) because (1) SR noted that his measurements of TD were more reliable than those for AD and focused on them for his dissertation, (2) TD and AD were highly correlated (r = 0.99 electronic supplementary material, figure S1), and (3) my analysis did not include birds of prey for which AD is considered a proxy for visual acuity at high speeds [14]. To ensure that there was no systematic difference between AD and TD, I repeated all analyses for both metrics and report the AD results in the electronic supplementary material. SR included the pre-preservation mass for most specimens measured, and I supplemented missing species with values from the Elton Traits database [43]. Sample sizes were less than or equal to five for 99% of species.

(b) Species

I included only extant, terrestrial, diurnal and non-raptorial species. This excluded seabirds, shorebirds, wading birds, birds of prey and nocturnal/crepuscular species whose eye morphologies are adapted to aquatic foraging, extreme long-distance prey resolution and nocturnal vision, respectively [12–14]. This resulted in 2777 species from 139 families (electronic supplementary material, table S1).

(c) Habitat

I defined habitat specialization using classifications published by Bird Life International [44]. For each species I extracted all habitats classified as ‘Major’, which is defined by BLI as ‘required for survival’. For the 541 species lacking a ‘Major’ classification, I extracted habitats from the next hierarchical level of ‘Suitable’. If a species was classified as having only one ‘Major’ habitat I considered it specialized to that specific habitat. Species with more than one ‘Major’ habitat were classified as habitat generalists.

(d) Foraging behaviour

I used a recent ecomorphological analysis of the avian tree of life to define the predominate foraging manoeuvre employed by each species [16]. Specifically, I used the ‘Foraging Niche’ classification to categorize species as using predominately myopic (near-sighted) versus hyperopic (far-sighted) manoeuvres (electronic supplementary material, table S2). Because the study was interested in classifying species into specialized manoeuvre types based on a combination of dietary guild and fine-scale foraging behaviour, it did not provide classifications for omnivores or species that exhibited a mix of manoeuvre strategies that did not easily conform to their fine-scale classification system. Because I was interested in a more generalized binary classification unrelated to diet or behavioural specialization, I used Birds of the World to classify those remaining species as either myopic or hyperopic (N = 704 species) [45]. I classified myopic manoeuvres as those that target food in the immediate visual plane (i.e. glean, pick, probe and hammer) and hyperopic manoeuvres as those that target food at a distance (i.e. sally, hawk, screen and pounce) [46]. For foraging stratum, I calculated the per cent use of the middle and upper canopy strata by combining the middle canopy, upper canopy and aerial foraging strata published in the Elton Traits database [43], creating a continuous scale ranging from 0 (ground) to 1 (canopy).

(e) Diet

I extracted diet percentiles published in the Elton Traits database [43] and converted them to four principal component axes explaining 99% of variation in dietary preference (electronic supplementary material, table S3).

(f) Macro-ecological variables

I extracted migratory tendency, range size and mean range latitude from a previously published global analysis of dispersal ability [47].

(g) Morphological traits

To examine the relative importance of eye size compared to other morphological traits, I used recently published ecomorphological databases [16,47]. Specifically, I used nine principal components decomposing variation in morphospace for a suite of traits related to diet (bill length, width and depth) and movement (wing, tail, tarsus and mass), as well as the hand-wing index, an assumed correlate of dispersal ability.

(h) Correlation of variables

All ecological variables were largely uncorrelated, and eye size was largely uncorrelated with the 10 previously published morphological axes (electronic supplementary material, table S4 and figure S2).

(i) Phylogeny

I pruned 100 randomly selected trees downloaded from a previously published avian phylogeny [48] (Hackett backbone). Trees were based on molecular data for 2311 species (82%), and I ran all analyses across the 100 trees to account for topological uncertainty among the species with no molecular data.

3. Analysis

All analyses were conducted in R 4.0 using the ‘phylolm’, ‘MASS’, ‘emmeans’, ‘visreg’, ‘nlme’, ‘ape’ and ‘rr2’ packages [49–52].

(a) Eye residuals

To correct for body size allometry, I extracted residuals from phylogenetic regressions of log(TD) and log(AD) on log(mass) (electronic supplementary material, figure S3A). I compared models incorporating evolutionary models for Brownian motion (BM), Pagel's λ (PA) and an OU model of stabilizing selection, looping across 100 trees. The PA models best fit the data as ranked by AIC for TD and AD (electronic supplementary material, figure S3B), and I extracted residuals from the PA models with the median slope coefficients among the 100 trees.

(b) Part 1: ecological correlates of eye size

I used phylogenetic multiple linear regression to examine the ecological correlates of residual eye size with all phylogenetic models repeated across 100 trees. I scaled all continuous variables to a mean of 0 and standard deviation of 1 to allow comparison among effect sizes. The following steps were completed for both TD and AD. First, I converted the following variables to binary states: habitat (forest versus non-forest), foraging manoeuvre (myopic versus hyperopic) and migratory tendency (migratory versus non-migratory). Second, I constructed a global additive model containing all variables, with interactions between each binary variable and all continuous variables. Third, I fit the global model to phylogenetic regressions for BM, PA and OU models of evolution the PA models consistently fit best for both TD and AD (electronic supplementary material, figure S4), so I used PA models for all subsequent analyses. Fourth, I used forward and backwards AIC model selection to eliminate all variables and binary × continuous variable interactions with insufficient explanatory power. The retained variables and interactions were largely the same for TD and AD (electronic supplementary material, table S5). Fifth, I constructed phylogenetic regression models using the retained variables and binary × continuous variable interactions from the stepwise procedure to construct the final ecological regression model. Finally, I used partial residuals and the estimated marginal means and pairwise contrasts of model coefficients corrected for multiple comparisons (Tukey adjustment) to visualize and summarize the coefficients of each individual variable and interaction while controlling for all others retained in the model. Statistics exhibited little variation among trees, and I used partial residuals and statistics extracted from the trees with median coefficient estimates for subsequent inference (electronic supplementary material, figures S5 and S6). For continuous variables and interactions, inference was made on mean coefficient estimates and whether 95% confidence intervals overlapped zero.

Using residuals in subsequent linear models can create spurious results if collinearity exists among them [53]. Hence, I duplicated the initial global model using the log of absolute eye size (TD or AD) as the dependent variables and the log of mass as an explanatory variable. Results were extremely similar. I used residuals for the final analysis because they provided an intuitive index of relative trait size.

(c) Part 2: morphological predictors of ecology

I examined the relative explanatory power of residual eye size (TD and AD) in predicting each of the previously described ecological variables compared to the 10 previously published morphological axes related to movement and diet using phylogenetic regression. For binary dependent variables, I used phylogenetic logistic regression, which accounts for the evolution of binary traits [50]. Each model was additive and consisted of one dependent ecological variable and the 10 independent morphological traits. All traits were scaled to mean of 0 and standard deviation of 1. I ranked the predictive power of each trait by comparing the absolute value of median z-scores across all trees. To determine the degree to which eye size improved model fit, I calculated the coefficient of determination (pseudo R 2 ) for the 10 ecological variables used in this analysis by sequentially adding (i) phylogeny, (ii) the 10 previously published morphological trait axes and (iii) residual eye size. R 2 values were extracted using the R function ‘R2.lik’ [52]. The R 2 values did not vary substantially among trees, and I used the median values for inference (electronic supplementary material, figure S7).

(d) Part 3: phylogenetic correlates of eye size and variation partitioning

I partitioned variation among the phylogenetic and ecological correlates of eye size (TD and AD) using phylogenetic and non-phylogenetic regression. For phylogenetic models, I first calculated the cumulative R 2 for a model incorporating a phylogenetic correlation matrix based on PA and containing no variables versus a non-phylogenetic null model. I then sequentially added variables and interactions retained from the stepwise model selection results in the following order: habitat, foraging and diet (foraging manoeuvre, stratum and dietary axes), and macro-ecology (latitude, range size and migratory tendency). Values for R 2 did not vary substantially among trees, and I used the median values for inference (electronic supplementary material, figure S8). To better understand how each of the 10 ecological variables contributed towards the role of phylogeny in explaining residual eye size variation, I calculated PA for each variable independently using the functions ‘phylolm’ (package ‘phylolm’) for continuous traits and ‘fitDiscrete’ (package ‘geiger’) for the three binary traits.

4. Results

(a) Part 1: ecological correlates of eye size

As predicted species specializing in forests and using hyperopic foraging manoeuvres had significantly larger eyes (forest versus non-forest: β = 0.036, p < 0.001 hyperopic versus myopic: β = 0.099, p < 0.001) (figure 1 electronic supplementary material, table S6). Eye size decreased with increasing foraging strata for resident forest-dwelling species using myopic foraging manoeuvres. Migratory non-forest species with larger ranges had smaller eyes, and eye size increased with proximity to the tropics (figure 2). Results for AD were similar, except that AD did not vary with foraging stratum for any category (electronic supplementary material, figures S12 and S13).

Figure 1. Partial residual plots for the relationship between residual eye size and habitat, foraging manoeuvre and migratory tendency for 2777 species of terrestrial birds. Boxplots represent the median, 25% and 75% quartiles, interquartile range and outliers. Pairwise contrasts corrected for multiple comparisons are significant where p ≤ 0.05 significantly different categories are denoted by unique lower case letters.

Figure 2. Estimated marginal means (+/− 95% CI) of model coefficients for (a) foraging stratum, (b) latitude and (c) range size in relation to residual eye size for 2777 species of terrestrial birds.

Correlations with diet were partially mediated by habitat, foraging behaviour and migratory tendency. As predicted invertivores had larger eyes compared to herbivores. The pattern was significant only for myopic foragers based on TD and for both myopic and hyperopic foragers based on AD (figure 3 electronic supplementary material, figure S14). Eye sizes of myopic invertivores were larger for forest specialists compared to non-forest species, suggesting the existence of visual constraints within darker environments. Large eyes were also associated with frugivory compared to granivory for myopic species. Finally, small and large eyes were associated with increasing amounts of nectivory and carnivory, respectively. The results for AD were similar (electronic supplementary material, figure S14).

Figure 3. Estimated marginal means (+/− 95% CI) of model coefficients for diet in relation to residual eye size for 2777 species of terrestrial birds.

(b) Part 2: morphological predictors of ecology

Eye size had the largest z-score when predicting specialization in forests and the second largest z-score when predicting foraging manoeuvre, consumption of fruit versus seeds and latitude. It was the fourth ranked predictor for invertebrate versus plant consumption (figure 4). Eye size increased the total coefficient of determination represented by ecological factors for these variables by 8–28% (electronic supplementary material, table S7). Results were extremely similar for AD (electronic supplementary material, table S7 and figure S15).

Figure 4. z-scores of residual eye size and 10 previously published morphological axes as predictors for 10 ecological variables from phylogenetic regression for 2777 species of terrestrial birds. Boxplots represent variation in z-scores across 100 trees. Scores are significant (p > 0.05) outside the dashed lines. Eye size is highlighted with its predictive rank relative to the other morphological axes.

(c) Part 3: phylogenetic correlates of eye size and variation partitioning

Residual eye size was highly conserved across the avian phylogeny (λ = 0.90 figure 5), producing families with strikingly different eye morphologies (electronic supplementary material, figure S9). There was strong phylogenetic signal among residual eye size and ecological variables for the final ecological model (PA = 0.87), and phylogeny explained the majority of variation in eye size (R 2 = 0.61) (figure 6). Ecological variables collectively explained considerably less variation in eye size for the phylogenetic model (R 2 = 0.09) compared to the non-phylogenetic model (R 2 = 0.41). This was largely driven by foraging and dietary traits, which also exhibited the highest levels of phylogenetic signal among the 10 ecological variables used in this analysis (electronic supplementary material, figure S10). Results were extremely similar for AD (electronic supplementary material, figures S11 and S16).

Figure 5. Phylogenetic distribution of residual eye size (TD) for a third of terrestrial avian diversity (N = 2777 species). Select families are noted. Photo attributions: A. Morffew, A. Radhac, B. Matsubara, B. McCauley, D. Coetzee, D. Church, D. Greenberg, D. Sutherland, F. Veronesi, F. Franklin, G. Smith, J. West, J. Thompson, J. Harrison, J. Boone, Kuribo, L. Boyle, L. Docker, M. Thompson, Mike's Birds, N. Borrow, P. Gaines, P. Kavanagh, T. Benson, T. Wilberding, S. Price, USFWS. All photos are held under CC-BY licenses.

Figure 6. Variance partitioning for phylogenetic and non-phylogenetic models of residual eye size for 2777 species of terrestrial birds.

5. Discussion

As predicted, eye size was strongly correlated with ecological variables related to light, such that species specializing in darker habitats and foraging manoeuvres requiring long-distance prey resolution had larger eyes. Across the globe terrestrial birds with the smallest eyes were those living in non-forested habitats and using myopic foraging manoeuvres, while species with the largest eyes specialized in forests and used hyperopic manoeuvres. Species at tropical latitudes had larger eyes, suggesting that light may contribute towards increased sensitivity to anthropogenic habitat disturbance in the tropics [54]. Despite many significant correlations with ecological variables, residual eye size was highly conserved across the avian phylogeny, and phylogeny explained the majority of eye size variation. Strong correlations between eye size and foraging manoeuvre and diet were reduced in phylogenetic models, likely due to correlation between the evolution and ecology of those traits.

(a) Eye morphology

The avian eye is unique among anatomical structures in being the one external trait specifically adapted to interpret light and is attuned to two interacting properties: intensity (brightness or luminosity) and wavelength (colour) [10]. Light intensity is mediated by the density and orientation of retinal cell ganglia (RCG), whereas colour is interpreted by specialized visual pigments and oil droplets found within RCG [3,13]. Although orientation of RCG and topography of the fovea contribute towards articulating precise aspects of light interpretation [18], larger eyes accommodate greater numbers of photoreceptors and improve visual acuity [55]. My results demonstrated that residual eye size correlates with multiple ecological factors related to light and can be used as a functional trait related to the optical environment.

(b) Foraging and diet

The majority of terrestrial avian diversity encompasses a complex array of dietary and foraging strategies, all of which require mediating rapidly changing light micro-environments when identifying and capturing food [15,56,57]. Past studies of specific avian communities or small subsets of the avian phylogeny have proved inconclusive regarding the relationships between eye size and foraging behaviour [33,37,38]. Using a much larger sample of the avian phylogeny, I demonstrated that species employing myopic (near-sighted) foraging manoeuvres and species foraging in higher canopy strata that are presumably exposed to more light had smaller eyes compared to those capturing distant, mobile prey, especially at lower, darker habitat strata. The fact that hyperopic species did not vary in eye size across foraging strata suggests that the adaptive benefits associated with a longer focal length both compensate for a darker understory and outweigh potential overexposure to increased luminosity higher in the canopy. Interestingly, the association between foraging stratum and eye size was fairly weak, with a significant relationship documented only for resident forest myopic foragers and effect sizes being generally small. Birds are known to forage across widely different canopy strata [58], meaning that the highly variable light environments experienced by most species may provide weaker selective pressure on eye size than other traits, such as diet or manoeuvre.

I documented several previously unrecognized relationships between diet and eye morphology. Diet predicted eye size mainly for myopic species, which had larger eyes when consuming more fruit and invertebrate prey. However, eye size based on AD also increased with arthropod diets for hyperopic foragers, suggesting an adaptive advantage provided by increased focal length when pursuing distant prey [38]. Nectivory was strongly associated with smaller eyes, implying that the probing manoeuvres associated with extracting nectar require less visual acuity. Instead, colour recognition may be a more important component of the visual system for nectarivores when identifying flowers [59,60]. Vertebrate capture was strongly associated with large eyes, likely because capture requires long-distance detection as employed by raptorial species [14]. In sum, small eyes appeared related mainly to nectivory and granivory and large eyes with insectivorous, frugivorous and carnivorous diets.

(c) Habitat and macro-ecology

The dramatic variation in optical environments found across a species' range requires visual adaptations that optimize survival [61,62]. Species adapted to forest interiors have been hypothesized to have larger eyes (i.e. ‘dim forest hypothesis’) [32], yet studies remain inconclusive on the associations between eye size and habitat [33,39,63]. This study confirms that forest specialists have larger eyes, supporting the idea that species adapted to extremely dark environments may respond negatively to abrupt spectral changes. Interestingly, eye size was inversely correlated to range size for migratory and non-forest taxa. Species employing these life-history strategies likely encounter the widest variation in light conditions across the terrestrial avifauna, suggesting that smaller eyes are more adaptive to increased habitat niche breadth [64,65]. Hence, the relative lack of light may constrain minimum eye size for species occupying the dimmest conditions, whereas abrupt transitions to bright light may pose a constraint in maximum eye size for species occupying heterogeneous habitats.

(d) Ecomorphological comparisons

Avian ecomorphological relationships have long been established for traits related to movement (wing length and shape, tail and tarsus), diet (bill length, depth and width) and size (mass) [16,66]. Given the links between morphological form and function, morphology is often used to explain rates of species diversification and trait evolution [67–69], population-level genetic differentiation [70], niche expansion and packing [8] and functional collapse in disturbed ecosystems [71]. Here, I demonstrate that eye size provides strong relative predictive power for explaining variation in traits related to habitat, trophic niche and life history beyond previously published ecomorphological trait axes. In particular, eye size was the strongest single predictive trait in determining whether a species specialized in forest and was a top predictive trait for foraging manoeuvre, diet and latitude. Despite recent exhaustive research describing the ecomorphological relationships for the complete avian tree of life, 22% of variation in the avian foraging niche remains unexplained [16], and eye morphology may contribute towards resolving such residual variation.

(e) Role of phylogeny

Phylogeny played a predominate role in explaining eye size variation, and residual eye size was highly conserved throughout the avian phylogeny. Despite many significant relationships between ecological variables and eye size, ecological factors explained far less overall variation after controlling for phylogeny. This mirrors evidence that trait evolution has a predictive accuracy of 65% in explaining ecomorphological relationships across the avian tree of life, with contemporary adaptations explaining a further 20% [16]. Results from this study are strikingly similar, attributing a comparable proportion of variance in eye size to ancestral relationships (61%) and recent adaptation (9%). Collectively, high phylogenetic signal in both residual eye size and correlated dietary and foraging traits suggests that light has contributed to the evolution of avian lineages, producing families with markedly different eye morphologies and providing tantalizing evidence of correlated evolution between eye morphology and traits related to the foraging niche.

(f) Conservation implications

Morphology is often interpreted as a set of functional traits that contribute to the disassembly of avian diversity in anthropogenic landscapes [9,72]. Strong predictive power and correlations with forest specialization and latitude suggest that the relatively dark light micro-environments found within forests, especially in the tropics, contribute towards species-specific sensitivity to habitat disturbance [73]. Given that altered light intensity regimes have been linked to avian community disassembly [33,39], light may act as an environmental filter in anthropogenic landscapes, and morphological adaptations to light should be considered when assessing interspecific sensitivity to habitat fragmentation and land use conversion.

Data accessibility

All datasets and scripts are available from the Dryad Digital Repository: [74].

Author contributions

IA: conceptualization, data curation, formal analysis, funding acquisition, investigation, methodology, project administration, software, validation, visualization, writing-original draft and writing-review and editing.

Competing interests

I declare I have no competing interests.


The Katherine Ordway Chair in Ecosystem Conservation at the Florida Museum of Natural History provided funding for data entry.

What are some examples of a niche?

One example is a primate called the aye-aye. Found only on the island of Madagascar, it's niche is actually very similar to a woodpecker's because this primate uses it's long middle finger to tap on trees and listen for larvae underneath. It then uses that middle finger to dig out the larvae and consumes them. It is active at night and is arboreal. The fossa, birds of prey, and snakes may eat aye-ayes.

The niche of a polar bear is very specialized. These arctic dwellers have no natural predators. Polar bears are carnivores and catch seals. The majority of seals caught are not on water nor land but at the interface of the two (or water and ice). Polar bears spend a majority of their time in the water. Given that polar bears obviously cannot swim forever, thus they require ice to periodically rest from time spent traveling and hunting. Polar bear cubs are raised in snow dens, thus these animals also require snow in order to raise their young.

Watch the video: Η διατροφή του διαβητικού. Η δική μου εμπερία. (May 2022).