Applied Ecology and Management Around Waterfowl Management in Nebraska
Ornithologists have been banding birds for over a century, but these data have been used less often than they could be to answer questions about population dynamics. In this seminar, I attempt to bridge the gap between data-rich species like mallards (Anas platyrhynchos), with over 1.2 million band recoveries, to data-poor species like wood warblers (Parulidae), with just over 6,000 recoveries from 53 combined species. The most common application of band-recovery data is estimation of annual survival. By using species as random effects, I show how we can also estimate juvenile and adult survival for data-poor assemblages such as wood warblers. Moving on to fecundity, I demonstrate how we can estimate fecundity at annual and regional scales using age ratios of birds captured for banding, and I apply these estimators to prairie dabbling ducks (Anatidae) and dark-eyed juncos (Junco hyemalis). Finally, I demonstrate how banding data can be combined with harvest data to estimate population size using Lincoln estimators, and apply the method to data-rich and data-poor examples.
icon search Searchable Transcript
Toggle between list and paragraph view.
[00:07:48.090]Thank you very much.
[00:07:53.280]So for the last decade. I've been doing a lot of work with banding data and and I want to talk about what we can do with banning data, some of the things that I think everyone realizes, we can do with banning data, but also some of the additional things that we can do with banding data.
[00:08:13.950]So the obvious thing we do with banning data is we estimate survival rates.
[00:08:19.860]I want to talk about doing additional things estimate fecundity estimating abundance and we do all those things. Basically, we can we can ask me and everything we need to build full population models using banding data.
[00:08:36.450]Start out with survival. You know why why focus on survival. Why care about survival well because survivalist, the most important vital rate affecting
[00:08:46.980]Population size next year. If you've seen projection models lesson projection models and you look at the know you look at the vital rates here inside the projection matrix.
[00:08:57.570]Even color coded it to make it easy for you to survival rates are essence and their most dominant thing they're more SS in there than anything else.
[00:09:06.840]They also have the most leverage the greatest sensitivity or elasticity and particularly this one down here in the lower right corner really drives. What happens to most populations, unless you work on Super for con Fisher bunny rabbits are micro this or something like that.
[00:09:27.510]This is the most important vital rate for whatever you probably work on
[00:09:34.320]How do we ask them in survival. There are lots of different approaches. One approach is population reconstruction. If we can age animals in life or in depth, we can figure out how many of them there are each age class, we can reconstruct populations we can estimate survival that way.
[00:09:53.220]We can estimate survival using telemetry. This is a picture from a PhD student of mine Courtney Amundson
[00:10:00.210]Studied survival mallard duck lengths from hatch to floods, unfortunately, we found out when we put five gram radios on on 35 gram ducklings they didn't survive from NASA to the web. And so, you know, telemetry gives you a really great information on survival unless it, it causes mortality.
[00:10:21.000]Market capture is a kind of way of estimating survival where we put colored markers on animals we observe them or we capture them live identify them live
[00:10:34.740]But the kind of data. I want to talk about our debt recovery data. We put tags or bands on animals, and then we we encounter them typically when they're dead. Often when they're harvested. And what do we know, we know that we they died here they died, then we know when they were marked
[00:10:55.890]If you're familiar at all with live encounter data.
[00:11:00.450]You probably familiar with the idea of capture histories, we can build a capture history for a particular piping clover. This one was first marked in 2005 as a juvenile
[00:11:12.810]Three years later it recruited as a breeder and it was cited alive and no way. And so we have these capture histories here for five different clovers and and once they recruit sometimes they're seen almost every single year.
[00:11:28.290]With debt recovery data.
[00:11:31.050]We have one observation or two. We have an observation, when we put the mark out and we maybe get one more observation, when you die and we can summarize the data from dead recovery data completely in what's called an MRI.
[00:11:46.740]And if you're familiar with program mark if you've done Mark we capture analysis. These triangular shaped
[00:11:54.120]Make a Caesar arrays, you'd be very familiar with them. The idea behind them is each row represents a release cohort so top row animals mallards released in year one.
[00:12:07.230]columns represent what year they were covered dead so released in year one recovered dead in your on cover dead near to
[00:12:16.380]Diagnose represent age or time since marketing. So there's a lot of information contained really compact format.
[00:12:26.190]And because you can only appear once here there's absolutely no loss of information with debt recovery data to summarize in this sort of way.
[00:12:35.250]You can build these with live encounter data but you lose all that information about this particular clover that we've seen every single year versus another clover that seemed to take sabbaticals every other year, because things like that.
[00:12:51.390]So I'm partial to debt recovery data and other important part of them is how many birds were banded each year and release. That's part of the data to
[00:13:01.230]So I first started working with debt recovery models and probably the way most people have worked with dad recovering models in the wildlife field by working with big data sets on harvested species and my introduction was with lesser Scott
[00:13:17.640]Scott populations were in decline. They were well below their long term mean no one knew why. But when no one knows why in the wildlife for waterfowl world. They always playing over hunting.
[00:13:31.590]Exactly. Actually, no. Scott in that ducks. I see redheads and ring. Next, and looks like blue winged Teal. Maybe, but I'm not withstanding we blame hunters typically
[00:13:45.600]And so 2016 I teamed up with a bunch of waterfowl ecologist tantalize continental US Scott data going back to 1952 in 60 years of data from throughout North America.
[00:14:00.360]We found that actually harvest rates for stop. We're going down through time. Historically, back in the 50s and 60s and 70s had been quite high in the 90s, Peter down to nothing.
[00:14:14.640]Increased a little bit in most recent decades, but on average harvest rates were low. By contrast, survival rates fluctuated year that there is no pronounced trend for survival to decline through time. Here's adult males.
[00:14:31.560]Here's adult females more variable and lower but no overall trend.
[00:14:37.800]Here's juvenile males.
[00:14:40.980]And here's juvenile females.
[00:14:43.860]And when we put those two pieces together and looked at survival rates in relation to harvest rates NASA does higher harvest great cause got survival to decline.
[00:14:54.630]We found no evidence for it whatsoever. So here, applied survival rates on the Y axis recovery rates return index of harvest rates on the X axis, no relationship.
[00:15:12.360]Can still maintain their positions in the presence of data in the absence of data. We've been arguing about additive versus compensatory mortality for decades. I take two or three year
[00:15:23.100]Sabbatical from working on waterfall and work on other things, and I come back, and people are still arguing about this question.
[00:15:29.520]I don't want to talk about this question today, I want to take banning data further and trying to ask, what else can we do with it. Can we ask other interesting questions that any population, the college's should be interested in
[00:15:44.010]So going back to my roadmap, the past, I'd always thought of a band recovery or debt recovery data is being limited to game species to harvested species.
[00:15:54.570]Is that the case, can we use these data to estimate survival in other birds demanding millions of birds 30 million birds in the last 60 years. Surely we can do something with that.
[00:16:10.530]And like log log regressions. And here I've got a log log progression of number of band recoveries against number banded each each datum here is this one species of North American bird.
[00:16:24.630]And so what the equation here tells you is that
[00:16:31.140]For every thousand bands you put out you get six recoveries on average.
[00:16:39.480]Don't think you're going to write a Masters by putting 1000 bands out on an average bird.
[00:16:46.680]But this is a log log scale. It's based 10 some species or 10 100 fold above that regression line I'm speech there 10 100 fold, below it.
[00:16:57.990]Any guesses what those two are
[00:17:06.930]But number of recoveries to really high.
[00:17:11.099]Robbins backwards would be way over on the right, but not in terms of number recoveries.
[00:17:17.190]He's Canada piece is one of them.
[00:17:20.550]The most boring duck in the world.
[00:17:23.940]Now it's or the other one.
[00:17:27.780]I don't have all these points memorized. I know this is the yellow build moon. And I know that that's a yellow bellied flycatcher so there are some things have been banned in in force and there's really no data on will really do anything.
[00:17:44.460]So let me go back to live encounter data. These are what the cell probabilities look like if you have to dig in and say, okay, what are these data mean and you'll notice in the likelihoods
[00:17:57.690]There are no essence there whatsoever. We don't use asked to describe survival in live in countering data, we call it a parent survive, give it the Greek letter feet.
[00:18:11.790]For debt recovery data, we actually code survival with an S and we call it to survival.
[00:18:18.870]You don't have to be a population in colleges to know which would you prefer to survival or appearance survival that I could ask the bartender tonight or tomorrow morning. If you have a choice between two survival or apparent survival, which would you choose.
[00:18:35.700]I would choose true survival.
[00:18:39.660]And the reason for that distinction. The two are linked but appearance survival means that you're still alive and you come back to the studies say where you originally marked
[00:18:53.730]True survival just means you're still alive somewhere out there in the world. So the, the difference between the two is emigration out of a study area and demographically that potentially can be real important. And isn't something to just ignore
[00:19:15.000]If we look at patterns of permanent emigration and birds, we find that, you know, virtually all species young birds are more likely to emigrate or disperse an adult birds.
[00:19:30.690]Also got that backwards but females are more likely to disperse and young birds are more likely to disperse
[00:19:38.670]So we know we're going to have more problems estimating to survival of females or have more travel estimate to survival of juveniles. If we use live encounter parents survival Tana
[00:19:54.390]So if you look at some case studies and people trying to population models. You can see where this important limitation came into play.
[00:20:05.520]So people working on oven birds wouldn't like to estimate adult female survival but because adult females didn't come back to their study areas. Instead, they built their population model using adult male survival.
[00:20:21.330]If students. My population ecology class remember nothing else they remember that I always describe male birds and bags with wings.
[00:20:30.330]Their superfluous except for about, I don't know, two seconds during the breeding season. And if you really want to understand how a population changes, you need to focus on females not build a population on bags with wings.
[00:20:47.700]Do for juveniles because juveniles don't come back. People routinely build population models for birds, where they have assumed that you know survival is one half of adult survival and you can find dozens cases of this literature.
[00:21:05.310]And I'll confess that the math is very, very easy. But where did this assumption come from.
[00:21:12.090]I think it actually traces itself back to Bob reckless, but that's another story.
[00:21:19.740]I want to suggest that that the distinction really shouldn't be between live encounters and dead recoveries. It should be about where you encounter birds.
[00:21:31.110]So if you're doing a live encounter study like these red knots and your places where you recite them are all along the Atlantic seaboard any place they might disperse and migrate through, you probably measuring true survival.
[00:21:48.330]If he ring shags on the Isle of May and you only recovered edge shags on the Isle of May, you have 10 recoveries. But you're not measuring true survival. You mentioned parents survival. So it's were not how
[00:22:03.870]Which leads me to this dichotomy for for treating your data, not just is it live or dead.
[00:22:12.240]But, but, whereas the individually encountered encountered at the site where you first marked it now you spanning site. But if it's mammals just marketing site.
[00:22:21.420]Or somewhere else in the world are you alive when encountered or dead one encountered and the likelihoods vary depending on that and and we've got these F fidelity terms involved, you're still at the side of market.
[00:22:39.090]So to do something completely different. I wanted to estimate survival in warblers would work. There's 45 species Canada in the US.
[00:22:50.880]Almost 4 million banded
[00:22:54.300]In the last 60 years, five years.
[00:22:59.160]Lot own banded with real spotty data. Sometimes they were aged. Sometimes they weren't. Sometimes they were that banding. Many times they weren't
[00:23:09.000]And from all those bands things. What do we have
[00:23:13.380]We have less than 3000 total recoveries.
[00:23:19.650]And that's fewer recoveries. Then from a Mallard study, a famous Mallard studying that Mark Mallard for eight years at one wildlife refuge in Colorado. So there's not much to work with here. But that's what I'm interested in asking what what can we learn from species without much data.
[00:23:41.280]So the approach. I'm going to use this. I'm going to, we're going to pay attention to how you encountered live or dead also. Where you are encountered were marked or somewhere else.
[00:23:53.790]I'm going to look for age effects and I'm going to look for effects and the parameters. I'm looking at her s for survival F for fidelity P for live encounter are for dead encounter.
[00:24:11.520]I'm going to ignore annual variation, because there just isn't enough data.
[00:24:17.040]And the approach. I'm going to use. So I'm going to combine the data from all 45 species.
[00:24:23.190]Are going to estimate a mean across all species, but I'm also going to know about species to differ from each other using random effects.
[00:24:38.550]And these are the hyper parameters or species means I get back for encounter rates.
[00:24:47.280]In market capture literature we often refer to encounter rates as nuisance parameters.
[00:24:53.490]When the default, you know, and cheesy plot is for the y axis to get, spit out and scientific notation. These are more than nuisance parameters and these are, that's awful.
[00:25:06.360]A meaning counter rate of of two individuals for every 10,000 bands deployed.
[00:25:14.040]Think about that.
[00:25:17.250]If the issue is not enough data, you're not going to get out of that hole by studying them for 10 more years or 50 more years, or for working twice as hard or four times as hard. This is what we have to work with.
[00:25:34.500]Who we actually do a decent job of estimating parameters fidelity parameters.
[00:25:40.890]data suggests that adult females and adult males are very high fidelity to marketing sites less. So for juveniles as we'd expect
[00:25:51.930]These are hyper parameters and apply to all species. So sort of average values. Let's break it out and look at individual species. So I've picked nine, you know, your four letter bird codes. You could pick out what some of these are
[00:26:08.790]And I found this interesting and I just picked these nine kind of nilly.
[00:26:14.190]And you know what I noticed is that there are some species that have really high site fidelity and both males and females and others that have really low side fidelity and I thought about the species and i i mean we came up with a biological explanation for this and
[00:26:33.720]Here it is.
[00:26:36.450]Something to look at common yellow throats and more detail. I'm going to look at Cape name warblers in more detail in the next two slides.
[00:26:45.330]So common yellow throats. Where do they breed everywhere kind of mid latitudes and their breeding distribution overlaps that beautifully with long term banding stations.
[00:26:57.810]So we're banding these birds, where they breed and it's no surprise to me that adults have very high fidelity to marketing sites and juveniles less so. And I interpret that as a biological phenomenon of the needle dispersal.
[00:27:16.140]About Cape May warblers
[00:27:18.900]They breed north of most major Manning stations. Some most of them are being banded in passage on the way to the breeding ground for the way back to the wintering grounds but not on their breeding grounds.
[00:27:33.240]If we look at fidelity parameters for them.
[00:27:38.070]Fidelity reflects there's no attachment to the sites where they're abandoned because they're migratory sites, not very strong fidelity, at least.
[00:27:46.530]And so here I don't interpret very much biological terms of this I interpret fidelity almost as a nuisance parameter. And it's a new sense because we're not banning them where we need to be a major monitoring stations are are in monitoring areas are in migratory areas.
[00:28:06.240]This is a survival hyper parameters and
[00:28:10.980]It's not a very great precision for juvenile survival. That's a common problem.
[00:28:17.370]But I will point out about seasonal survival. If it's not one half of adult survival.
[00:28:23.550]Not even close.
[00:28:25.800]Nice thing about Beijing analyses is it's easy to summarize are called posterior distributions.
[00:28:32.340]This is a posterior distribution of the ratio of juvenile to adult survival. Here's the 50% line and everybody in their brother and sister of us for population modeling.
[00:28:43.560]If you're going to pull a number out of you bound to use for population modeling. I'd say use two thirds, at least, instead of one half. But why not use the data.
[00:28:57.780]And for the nine species here. Here the survival rates.
[00:29:02.910]And what I haven't done up till now on to share with you the sample sizes that went into this analysis. But usually when I do an analysis. I like to think in terms of
[00:29:13.470]Let's use a really good strong example a week example and sort of middle example. I don't have that my continuum is weak data or maybe extremely weak data no data at all.
[00:29:27.930]And I want to point that out, no data at all Colima warbler I wouldn't know one if it flew in here and landed on the bench, but probably someone would
[00:29:38.400]Why would you include a species with no recovery data.
[00:29:43.680]And the answers to prove a point that even when you know nothing.
[00:29:48.630]You should know something, you know, it's a warbler, you know, it should share survival rates of its conjures
[00:29:58.350]Amidst the distribution of its current owners. You do know something. I mean, that's not a very precise estimate of adult male survival and that's pretty horrid estimate of juvenile male survival but uh you know a little bit about the expected distribution even, even with no data at all.
[00:30:21.840]So let me move on to estimating candidates from banding data. So, something we typically think about
[00:30:32.550]So these are the Emirates. This is that famous data set of San Luis Valley mallards you ever want to build a better
[00:30:42.000]Band encounter dead recovery model. This is the famous data set, you have to work with. Do you want to build a better quarterback receiver model, you have to apply it to the European differ data.
[00:30:53.280]Well, this is the test data set for for dead recoveries. And there are a lot of recoveries there. But what I'm drawn to is, there's really a lot of information down here. The release calm.
[00:31:06.330]And for something like warblers, that's where all the data is somewhere up in the recovery array. There might be a one someplace. He looked for it and so can we get information from all this data that we collect a time of banding.
[00:31:22.800]We determine if you're a juvenile we determine if you're an adult, can we estimate. So I am using African this time to meet for quantity instead of fidelity. But can we estimate fecundity as a number of juveniles banded over the number of adults beyond it.
[00:31:40.560]An index of candidate.
[00:31:45.000]And I'd worked with data sets with mallards and black ducks, where if I looked at H ratio from harvest data as we traditionally collect it.
[00:31:55.380]seemed really similar to age ratios at time of banding. You know, caught some the same temporal dynamics.
[00:32:03.150]In NH ratios and seems a lot easier to use the data you have in hand from banding birds and the way for hunter recoveries to come in and they also need band recovery data just for vulnerability, why not just use the banding data.
[00:32:19.350]So again, this really savvy PhD student to just matriculated last year. Hannah Speight
[00:32:27.660]Took this I gave it to us an idea was kind of my baby. I was afraid to give it away, and she took it. She ran with it. She did more wonderful things with it than I could ever do. I'm really happy to work with her we published it this year. Journal of Applied ecology.
[00:32:46.020]We're also getting in the habit of whenever we publish something to publish all our data sets and all our code and make that accessible as well. So that's out there and what she did is she spatially analyzed 15 years of banning data.
[00:33:02.880]All across the prairie provinces and the US prairies Prairie Pothole Region basically
[00:33:09.330]Looked at all. The duck banning data all the duck survey data and and all the breathing bird survey data collected at all and
[00:33:20.400]Did a lot and I just want to highlight a couple things that came out of that analysis is that across for the five species we documented.
[00:33:31.440]Long term to clients and fecundity a prairie nesting dabbling ducks. The only exception here blooming to have really responded increase the kind of the starting about 1995 that's when the Dakotas got wet. That's when CRP came on the landscape. It's when you expect a big bounce in productivity.
[00:33:53.490]Color coding also indicates that habitat conditions have got a lot worse in Canada and blue. And they have in the US eagle.
[00:34:07.380]I want to keep working with this. And so, so I wonder, could we correct those eight races at banding for vulnerability, just like we do in harvest data so we get right ratios at time of banding.
[00:34:22.020]Potentially juveniles, they're easier to capture than adults. And if that's the case, then he would be inflated maybe adults are easier to capture and then Nate ratios would would be underrepresented.
[00:34:35.820]If somehow we could estimate capture probabilities
[00:34:40.110]We could control for this. And so what we need to know is what's the relative vulnerability of of juveniles and adults to been catching a trap for I missed that.
[00:34:52.739]This would be really straightforward to do. Everyone had bands birds. I'm going to advance ducks gets thousands of traps and you know what they do they pick him up there looking said, we've been you already do that.
[00:35:06.810]So we don't have the data to do this with and I want to change that. And if nothing else comes out of this seminar, the importance of of same season. Same station live encounter data, we can do a lot more with it and people realize
[00:35:26.340]And if we have that vulnerability, then we can estimate true age ratios and get a vulnerability adjusted estimate for candidate.
[00:35:35.580]At the same time, we use the banding data estimates, you know, adult survival. Now we filled in that matrix. This room banding data where estimate everything we need project this population.
[00:35:48.180]So on the field test this on a population that I knew from auxiliary data was declining. So I picked a 2324 year timeframe for northern pin tales.
[00:36:02.400]With carefully selected them points when the survey data suggested that this population was declining by five and a half percent per year and a half. Could I make cover that using only banding data.
[00:36:19.230]Working in when bugs I estimated survival rates with a lot of precision. That's all I want to emphasize here recovery rates from hunters are estimated with a lot of precision.
[00:36:31.080]He's lining counter rates are asked me that really poor decision. And that's because in the entire bird banning lab data bank on 10 tails there only 90 recorded live encounters.
[00:36:45.930]And I know that there were thousands and thousands of birds were looked at as it always been you yesterday.
[00:36:51.540]So the data exists that don't exist that could exist as opportunity to collect that data. We don't get very good precision on vulnerability or fecundity, but it's because we don't have good estimates of live encounter rates.
[00:37:09.330]But we nail lambda, just from banning data we recover lambda point estimate at least that bang on to what the survey data suggests yeah I'll admit that conference intervals wide enough to drive a truck through, but, um, but I think it has some promise.
[00:37:27.780]Let's try something more difficult about women. Let me show you the annual parameter estimates juvenile survival doesn't fare in much adults survival doesn't vary at all for females.
[00:37:38.970]All the annual variation is in the quantity and that's what's driving in your population growth. So 10 tales declined over this 24 year time period because fecundity was declining turn 13 years
[00:37:59.430]And for and giggles I find it to jump goes
[00:38:04.350]Which is a species that breeds way up north isn't well covered by Breeding Bird survey data at all, but they're banded lot of them are abandoned and passage during migration banding.
[00:38:19.500]I said, let's try this on telcos with Dan and a lot of junk goes, you know, a third of a million.
[00:38:26.400]But less than 200 dead encounters. And here I've only got 60 live encounters to work with. But I'm I can estimate juvenile and adult survival. I can estimate quantity and I can estimate a lambda that
[00:38:42.870]Not very precise, but it's bang on with what the map stations get for dunk codes and it bang on with what reading bird survey gets in terms of point asked him, So I think, I think this approach can have promise for other other birds and if it can work for exact goes
[00:39:00.870]Probably can't work for warblers because there's not enough recovery data, but we've got a few hundred recoveries. And especially, we have live encounter data, we could do this.
[00:39:13.290]Okay, last piece of the puzzle.
[00:39:16.470]Estimating abundance from banding data and my subtitle is a new old method for estimating abundance. What do we mean by that.
[00:39:25.770]I was excited to be a part of this paper with colleagues back in 2014 where we used Lincoln estimates banding and harvest data estimate population sizes mallards
[00:39:37.980]But fact that we call them Lincoln estimates suggest that someone else came up with this idea that was Frederick Lincoln back in 1930
[00:39:47.580]We outline this method and nobody really used it for 80 years sprinkling here in there. And this is the Lincoln, for which the Lincoln Peterson estimator his name. So he is known for that. But the idea of combining harvest data and recovery data estimate population size.
[00:40:06.420]Didn't see a lot of us.
[00:40:09.330]So the logic of it is really straightforward.
[00:40:13.620]It's why Petersen independently discovered it 50 years earlier, or something like that and why gone discovered 100 years before that, but isn't remembered for it. It's just some pretty simple division.
[00:40:30.270]If we can count the number of birds that we band, we should be able to do that. They come in strength of 100 that makes it easier
[00:40:39.570]We can count the number of recoveries we get back then we get a recovery rate and I've expressed this as an inverse
[00:40:51.300]Then if we can estimate TOTAL HARVEST we've got TOTAL HARVEST divided by per capita harvest that's population size.
[00:41:00.750]There's one nice little correction. We'd like to put in, if we can, we'd like to recognize that not all bands get reported to sororities
[00:41:09.570]And so this green band circled in red. That's a reward band. If you ever find one of those check it out carefully and it might say reward $100 or $20 a free report it. You can attack US government for reporting the band and we put those three pieces together.
[00:41:29.940]We can estimate population size and I put tax over everything, because we're estimating all those things. So there's some uncertainty involved, but most of the uncertainty is involved in how many recoveries. We get down here in the denominator.
[00:41:48.390]Now he's, he's a lot now and lots of different species of waterfall used to my Woodcock now to and the precision of this estimator depends on how many recoveries. You can get
[00:41:58.650]So if you only got one band recovery, you know, don't try to do much with that CV of plus minus 100% of your estimate but 10 recoveries. That's not bad 100 recoveries.
[00:42:12.930]Equals the precision of the Eastern waterfall survey done by US Fish and Wildlife. There is an enormous expense.
[00:42:21.570]Thank you could get 100 black duck and Mallard recoveries for less than that thousand recoveries equals the precision of the entire continent waterfall survey, we're not going to hit that for most PCs, but we get it from mallards and Canada geese certainly
[00:42:37.380]And so I've used this for black ducks, where we have survey data we can combine it with harvest data. And what I've been able to do with black ducks is historical survey data go back to 1993
[00:42:52.080]Harness data go back to late 1960s.
[00:42:57.210]Using both data streams reverse time models to take advantage of stronger data. More recently, I can ask them a black duck population abundance back before we had aerial surveys to do that.
[00:43:16.290]There's some critics of Lincoln estimators as we've applied in waterfowl. The main reason here I'm showing you the population estimate the official population estimate from the US Fish and Wildlife Service.
[00:43:29.580]Here's what reality discuss in and colleagues, including me get with Lincoln estimators and what do you do when you're continental population sizes of 95% confidence incredible intervals don't overlap each other at all.
[00:43:45.240]One of us is wrong, probably were both wrong. There are lots of potential reasons for that. But, but I don't put a lot of confidence anymore and population estimates enough they got nice 95% confidence intervals around them.
[00:44:01.230]Pay attention to the assumptions and realize we might we might be, we might be a long way away from where we think we are
[00:44:10.020]So once more to apply this to something completely different. Let's use dark. It just goes, we don't have a dark I Junko hunting season yet.
[00:44:18.060]But we band them a two times of the year, you can treat spring banding as if it were a harvest season so so I've taken data by decade.
[00:44:29.400]This red line is partners in flight puts up population estimates for every bird in North America. Every man bird don't know where these numbers come from, but I've used them Bob and I use them when we published about
[00:44:44.070]Window and tower mortality and birds because we needed something
[00:44:49.290]But when I asked me population size of market capture for exact goes yeah made sense. Back in the 80s, but but if you're rich uncle's now maybe
[00:45:03.450]Last bit putting all these pieces together in the last few years have become a real fan of integrated population models.
[00:45:13.020]Because I don't trust abundance data all by itself. We have abundance data and we have survival data and we have fecundity data, why not try to combine them all into into a model and put all the pieces together.
[00:45:29.520]So the example I want to share some work I've been doing with thumb Dave Coons at Colorado State because shop at Swiss on a theological Institute and
[00:45:41.520]Based on what I've just described in the preceding elements we can use Lincoln estimate estimate abundance of juvenile and adult black ducks in the fall.
[00:45:51.750]We also do in the spring we can estimate survival rates been estimate fecundity rates we kind of have the workings here have an entire population model.
[00:46:03.450]One of the nice things about black ducks is that they've been banded historically at two times of the year.
[00:46:10.050]Right before hunting season. And right after hunting season.
[00:46:15.720]And most of the recoveries come during the hunting season.
[00:46:20.070]So given this banding a two times of the year, we can actually estimate survival for the time in between.
[00:46:27.300]So survival from here to here. That's sort of breeding and migratory Survival. Survival from August, September, all the way around to next January, February, includes winter natural mortality and also includes the hunting season.
[00:46:47.520]So just using Lincoln estimates of population size at two times of the year for different age cohorts. We get a lot of information by that. In fact, I could build a
[00:47:00.870]Integrated population model just using these data, I can illustrate sort of what's what we can estimate from this if I know adult population size.
[00:47:15.630]The spring and I know adult population size here and fall. The difference between those and summer mortality of adults.
[00:47:24.690]Well, then all the kids come in. That's a measure of Kennedy.
[00:47:30.150]January one comes along all the black duck panders stop differentiating between hats here. And after hats here birds because they don't do that anymore. They call them all. After hats years
[00:47:40.950]And I lose the ability to differentiate age. So I just call them adults but I get winter mortality of the combined ages, so that that's the vital rates is from abundance
[00:47:52.590]But I can also ask me the abundance rates from the banding data. So here's the idea with band and strength.
[00:48:00.810]We differentiate males and females.
[00:48:04.500]And so we've got population size in spring of adult males and adult females.
[00:48:11.370]We banned in the fall and we differentiate by agent.
[00:48:16.020]And so I've got population size by age and
[00:48:22.200]during the breeding season, what's happening. We've got breeding mortality. We've also got fecundity changing that present matrix projection model, the changes that so I get fecundity and survival rates by season.
[00:48:39.030]And in the winter. No fecundity both got winter mortality, some of it today content and I protect that forward with survival rates. So I can go all the way around the annual cycle determining abundance determine the key vital rates that cause abundance to change.
[00:48:58.800]And here I'm plotting survival during the winter period, which includes hunting up against the harvest rate on the x axis and this diagonal line.
[00:49:10.500]represents what additive mortality would look like if hunting was additive and I don't see any evidence of additives it in adults. But the caution, I'd say is that historically we pushed right up against the very limit of where there's potential for compensation.
[00:49:29.070]And you can't keep extending this line horizontal pass pass this line for young birds.
[00:49:40.170]There is a negative correlation here.
[00:49:44.130]But they're also well below the line indicating that there's a lot of natural as well as hunting mortality. So my conclusion.
[00:49:50.640]Hunting mortality doesn't seem to be affecting adults with the caveat that we've taken it right up to the very limit of compensation, we can add 50% to it or anything like that.
[00:50:05.160]Winner survive on males declines with increased harvest. So there's a significant effect of harvest on males, but both males and females experienced a lot of natural mortality during the winter. In addition to hunting mortality.
[00:50:21.420]Thinking about black ducks during two seasons.
[00:50:26.160]I think we tend to think about populations and we want to identify what's the carrying capacity for black ducks.
[00:50:33.180]I think that's a really silly question for something as migratory we could ask it, but I think we should ask, what's the carrying capacity for the breeding grounds and what's the carrying capacity on the wintering grounds.
[00:50:48.000]Here my estimates of population size during the spring during the fall. One based only on Lincoln estimates one based on the IP.
[00:50:58.740]And if I asked you to identify what is the carrying capacity of this or if you asked me to identify what's the carrying capacity of this
[00:51:06.060]I could I could construct the basie and state space model that would estimate our max you carrying capacity and what draw a line through this. I can also hand you a blue and an orange crayon and do just as well as I would do the checks and you might do something like that.
[00:51:27.060]And this is my key point.
[00:51:30.300]Here, I think those lines are absolutely wrong. They should be switched
[00:51:38.580]In spring birds come back to the breeding grounds. They lay eggs. A 35 gram black duck can come out of the egg on day one and find enough food over the next 60 days to turn itself into a duck. Does that sound.
[00:51:54.690]Like a population that carrying capacity or population, way below carrying capacity.
[00:52:01.200]Conversely, in the winter.
[00:52:04.620]Lot more birds fly down to the wintering areas, they're ever going to come back. That sounds like a population way above carrying capacity.
[00:52:13.800]Building lines need to be switched
[00:52:20.520]And I can model that show that makes more sense in terms of the data as well.
[00:52:31.260]Where we find out in terms of population size having influence on survival rates is that
[00:52:40.620]Fall population size.
[00:52:43.710]Doesn't really influence adult survival. So adult survivors constant regardless of population size but juveniles do less well when the populations high in the fall. And that's exactly what you'd expect juveniles or
[00:52:57.510]Less position to compete with adults and a resource limiting environment we get, especially for females, where they expect to see it the most have a strong negative correlation with survival.
[00:53:11.400]During during the winter.
[00:53:18.300]With respect to spring.
[00:53:21.090]The key driver of population growth rate and black ducks when looking at here in the y axis is lambda to year over year population growth, the only important driver was adult female survival during summer.
[00:53:37.050]So that's not that's not hunting. It's not harvest. I don't know what it is. I mean, most, most adult female mortality during the breeding season.
[00:53:46.320]Black ducks nesting the plants there a wetland bird that necessarily a plan, they're vulnerable to predation my
[00:53:53.670]You know my gut suspicion would be that it's prediction during the nesting season but but it could be other things. But during this time stress. This was the most important thing affecting survival rates and black.
[00:54:11.730]I'm not going to run through that I've run through that already. I'll leave it up there. Just as a reminder of what I've touched upon, but but what I hope I left you with is that
[00:54:21.750]Banning data can be used for a lot more and just survival and we can do unique things with it by borrowing ideas from other fields other disciplines.
[00:54:36.030]And doesn't just go for for banning data and goes for anything that we think about how we might use data limited
[00:54:44.640]Population data that we're never going to acquire enough of, like, we do a lot more of it with it and we imagine and for for species that we care about.
[00:54:55.920]That are less abundant declining that don't have long term monitoring programs. We need to get more efficient about what we do for the hundreds of species of passwords that that we care about that.
[00:55:08.670]We're never going to get enough data to estimate survival all by themselves. But I think there's some tricks we can play to get closer to that. So I'm going to stop it there and we have time for questions.
[00:55:42.780]society puts out some kind of forecast for American birds and mostly blue are those estimates based
[00:55:58.020]That's a good question. I mean,
[00:56:01.260]Every year the Fish and Wildlife Service puts out a duck report that's based on much better data and I don't pay a lot of attention to that. I think
[00:56:10.680]You know, over a time span of five years, you need trends there there are real, but on a one year by one year basis know
[00:56:19.470]The best way we have of gauging North American bird populations. Love it or hate it is the Breeding Bird survey. And if you aggregate all the species with enough data and look at him.
[00:56:33.030]You know, an average things aren't looking too bad. And there are species are doing extremely well their species that seem to be doing extremely poorly.
[00:56:43.170]But if you think about all those 400 species as an idea of the news as random effects.
[00:56:50.190]Some of them just too bad luck or going to suggest increases that aren't real. Some of them through bad luck or kind of success populations are declining. The data don't really support the uncertainty about the parameters report.
[00:57:04.590]I think we spent a lot of time.
[00:57:07.710]Chasing ambulances isn't the right term for it but focusing on things that seem to be declining, but maybe the day they're deficient indicate that they're really declined.
[00:57:24.480]Just a source of nature. Somebody increases or decreases.
[00:57:29.940]I think if we took anthropogenic effects out that on average they all be stable and not population growth would be zero. I mean, that'd be the expectation and less
[00:57:42.870]And some would go down because of habitat and yes when the next glacier comes in, which seems like it won't be for a while that might change everything and drive it down. But on average, I would expect.
[00:57:57.750]Most things to be holding steady and
[00:58:02.220]You know, probably not going to follow a normal distribution. But if it did two and a half percent would be increasing above expectations and two and a half percent though.
[00:58:13.050]There are a lot of good examples of species. They're increasing exponentially. And there are some good examples of species i think they're disappearing their species like black hole warblers would seem to be going down. Yes. Can certainly based on PBS. It's big enough to include zero. Yes.
[00:58:57.300]Yeah, that's a good question. And it really depends. It's study area specific. So some of the big study area and a thorough study area their scale of what's local could be really big and someone they're really highly focused any area.
[00:59:14.700]Chickens in my backyard is going to be pretty small. What I used for this analysis and it was the course of the data from the bird banning lab come at
[00:59:24.150]Is is a 10 minute block which is pretty large. And that's because historical banning data weren't recorded with GPS accuracy. They recorded to the nearest 10 minutes of a degree block which is
[00:59:39.240]Pretty big chunk.
[00:59:41.940]So that's what I mean by fidelity is, you know, pretty hard to him.
[01:00:05.160]For the most part, I helped with the with the workload example I showed where I knew that the the species breathing and the boreal forest were
[01:00:15.060]Had really low fidelity because of where they're being handed I I quoted that as a simple deep breathing the boreal forest or not.
[01:00:23.760]That Kofi had had tremendous explanatory power in terms of the fidelity parameter and it's able to include that in the model, if you had reason to suspect that if you're able to quantify that you could include it.
[01:00:38.190]I also tried that with warbler is differentiating between long distance migrants and short distance migrants on survival and didn't find any difference.
[01:00:48.000]So, but in the case study like that if there's a cove area that you think might influence juvenile survival or adult survival and you can measure it over all the species to be really straightforward to include an analysis.
[01:01:32.250]And most everything I touched on here. I actually selected the entire North American range and speeches and one one nice thing about that. If you think about how populations change birth immigration death emigration, select the whole range and you can not worry about it anymore. But, um,
[01:01:52.200]Do they work on smaller ranges.
[01:01:55.440]Within those 45 species of warbler the very best data existed for you could probably guess curtains warbler
[01:02:05.490]Fredonia endangered species that has lots of banning effort to encounter data. So, so the species that were outliers. There were were curtains warblers
[01:02:17.430]Range wide but restricted range wide and for solitary warblers will nest nest boxes. They've been well studied in some smaller study areas. And so those were the two where the vital rates were actually the best
[01:02:32.130]And and it was because of localized studies. I'm a fan of local intensive studies and I don't do them anymore.
[01:02:41.400]But, but I'm still a fan of doing now. I'm trying to work more big picture, because I think that's getting neglected by people and the data exists to try and do the big picture questions.
[01:03:14.190]Sorry, I missed
[01:03:31.020]So, so sort of the key thing to be able to do that is, is to isolate that season.
[01:03:38.760]That you care about the most with an encounter occasion on each side of it or marketing or encounter occasion on each side of it.
[01:03:47.460]By and large the bulk of bird marking and encounter data occur in the breeding grounds were most ornithologist work. And so we have early breeding season, maybe we have late breeding season.
[01:04:00.990]And the other bit you know we can close the breeding season. But actually, we've also included the other but includes to migration periods and a winner and period and a lot can happen.
[01:04:12.600]Over a lot of geography, she really wanted to do that focus on wintering areas, you need to counter birds as they first arrived on winning areas, maybe mark them for the first time encounter them again before they leave.
[01:04:29.880]Probably best done with telemetry but marketing can work for it for some species that have really restricted ranges. They're confined to
[01:04:41.040]coastal regions read knots are a good example. I think clovers are a good example where you get detections on the wintering grounds.
[01:04:49.470]And the breeding grounds and the beginning in the end of both of those. I think it's feasible to do that. And I've actually started to work with Great Lakes piping clovers to divide the season up into four and partition out 3D wintering
[01:05:04.290]fall migration spring migration and using live encounter data and the most critical periods seems to be falling migration.
[01:05:15.300]But both migratory periods.
[01:05:24.840]You know, yep. And, and there are banding networks that have started big banning networks have started
[01:05:33.300]In Latin America, some of them are focused year round. And so they're worried about their resonant birds, but they also pick up winter migrants, beginning in the end if they're doing monthly banding.
[01:05:43.470]There is the potential to to estimate survival during that period, you have to be in the right place at the beginning in the end of the wintering season.
Log in to post comments