Watch videos with subtitles in your language, upload your videos, create your own subtitles! Click here to learn more on "how to Dotsub"

BITC / Biodiversity Diagnoses - Inventory Completeness 1

0 (0 Likes / 0 Dislikes)
OK, So a very key concept in dealing with a regional diagnosis of biodiversity is how complete those data are across that region and essentially what we want is a map that says here my inventory is very complete and very reliable, but here, eh, not so much OK? And that's gonna form the basis for a lot of the downstream inferences both on the biological side and on the sampling side So, this is a pretty key set of concepts. You know, in theory you would think it's easy You find somebody with expertise in a taxon and you put that person in that place and you wait a little while until the expert says "I'm done, that's everything". OK? And it's not that easy. I'm just gonna click down As you all know many sites of not been surveyed in detail, many taxa are hard to survey. OK? I suspect that Dorothy can go into a site and find 5 species and not find 50 other species that come out only during the rains Or only after the rains, or only in the peak of the dry season. So, there are hard taxa to detect and then a lot of cases, you go in and your first task is to describe the new species. So, Moses can tell us about that with his micro-plants where even he has one name after him. And then even after the expert or somebody who wants to be an expert says: "OK, I think I'm done", you need to think about that more carefully and make sure that you believe the expert. So, that's where all of these other considerations come in, the spatial distribution of sampling, errors in taxanomy, etc. So, we really do need to be quantitative about this, so let's look at it a bit. This is from the Mexican Atlas, this that phenomenon that several of you have noticed that sampling follows roads, and that is true for very obvious reasons. I'd been involved in surveys in this gap here is called Chimalapa region and weeks of work just to get into the sites and between political problems, and logistic problems, and you know no trail going above 600 meters. Those of you who have done field work, know what I am talking about. OK? Those gaps, are gaps for a reason. So, those are spatial gaps and then we have temporal gaps, we already talked about that one with the Mexican Bird data, and then we just have these random gaps So, this is 3 different measures of degree of inventory, and you can see in all cases, these are frequency diagrams, so this is the number of specimens for example And the highest category by far is zero. OK? Which is to say for Mexican birds, obviously depends on the size of the block, but most of the blocks have not been surveyed. OK? And that is common. So, very generally, biodiversity tends to be too big to do what I would call inventory. Think about if you were the owner of a car dealership, OK? And at the end of the year, you need to understand what your stock of cars is. You're gonna go in and see I have 5 of this model, 3 of this model, 4 of this model, and one of this model. You're not gonna go in and sample. I have around, you know 4-6 of this model, and somewhere between 0-10 of this model So, a car dealership is easy because nobody has more than a few dozens. But when we are talking of millions, you really can't do that excessive inventory. I'll come back to this concept at the end But it really takes us to a world of sampling, rather than inventory, OK? So, there, in sampling we're talking about estimating a quantity. And that quantity should be local single site diversity, OK? So, Arthur's gonna talk with this a quite a bit about what diversity means, but not today. But my point is simply and I'm not gonna in a great detail to leave this to him is that ideally we want a picture of the diversity at individual sites. And we will start aggregating and remember our conversations about you know a tenth of a degree, a half a degree, a degree, two degrees That's aggregating and so you start contaminating local site diversity with among site diversity. No more said about that, but that's what we would love to have, and it's very rare that we can't have it. So, really briefly let's talk a little bit about sampling. OK, when we talk about sampling, there are going to be some assumptions. And almost always we violate those assumptions, but we should at least be aware of them. We are talking about Homogeneity or at least randomness of sampling across space, we're talking about homogeneity of sampling through time, and again both of those almost always we violate those assumptions So we have to be thinking about how much those violations of assumptions mean. And this is perhaps the most important one, we have to be thinking about how comparable individual samples are. So, this for example, Rodrigue, this is why we were talking about, should we combine GBIF data with your survey data? They are really different. And so, one unit of sampling in you know, found data coming from herbaria around the world is very different from one unit of your sampling, OK? So, that comparability is really key. We're talking about samples accumulating knowledge in a consistent, comparable way so that we can develop quantitative estimate from the samples. Now, really this slide which I thought was really creative, but others have said it's rather silly, this slide summarizes everything. Imagine we go to two different places and we start sampling. We are just detecting species. Which inventory is done at the end of these detections? [woman]: The one on the right. The one the right has one detection of each of nine different organisms! The one on the left, has five and four detections of only two. But if I'm just going out and grabing animals I think I've come to an end point here. I grabbed one and it was a rabbit, I grabbed another and it was a kitten, I grabbed the third and instead of being a dog, it was a rabbit. So, I grabbed lots of samples And I really only found two species! There over here, every time I grabbed a sample, it's a different species. So, if I graph the accumulation of species in this inventory, every time I take a new sample, I get more species. Whereas here, after a certain point, every time I took a sample, I detect another of the same species. So, over here, I'm done! That's a completed inventory. Over here, definitely not! OK? So we can summarize this information ... Got it. OK, thanks. Do you need this? So we can summarize our sampling information as matrix of species by samples, OK? And so, this matrix which is just gonna have kind of ones and zeros, OK? That matrix is what would base a lot of these inferences on. We can talk about three levels of detail, a full matrix is going to have all of the information in this matrix, including abundances. Now, for some of you abundances make sense, like when an Entomologist do trapping. You put out of trap, that's a very nice comparable unit of sampling, you know one night of trapping, and you might get, you know 40 of this species, 20 of this species, zero, 2, 5 and 10. OK? Or mammal people with traps. A simpler version is a presence-absence matrix, where the content of the matrix is binary. And then, what you'll see, perhaps most commonly, is just the accumulation of species records. So that would be in the first sample, maybe, it's the first day, I got one species. In the second day, I added one species to that. In the third day, I added three species to that. Clearly, more information is better, OK? So, up here is better than down here. Many people believe that a lot of abundance data are falsely precise, and that is definitely the case with observational data, and a lot of the point counts and things like that, that bird people do are all false precision, and so I personally don't put my stocks in those data So, the detail possible is often very much determined by what is possible, not by your preference. And then, there is some kind of basics of this that communities that have a lot of more or less equally abundance species would be the easiest ones to characterize. So, communities with high evenness if you remember back to your basic ecology classes. When you have done sampling, obviously you gonna be able to do a better job But when you have communities that have very rare species, so very big disparities in abundance, with some species that are hard to detect or very uncommon. Those are gonna be the hardest. That's all obvious. OK? So, we can look at some examples of species accumulation curves. Here is marine benthic communities, and you can see some of these communities, They start sampling and it levels off, and maybe these last 10 sampling periods, they didn't find any new species. And then you can see other inventories that are still going up even after lots of samples.

Video Details

Duration: 14 minutes and 58 seconds
Language: English
License: Dotsub - Standard License
Genre: None
Views: 5
Posted by: townpeterson on Jul 26, 2016

This talk was presented in the course on National Biodiversity Diagnoses, an advanced course focused on developing summaries of state of knowledge of particular taxa for countries and regions. The workshop was held in Entebbe, Uganda, during 12-17 January 2015. Workshop organized by the Biodiversity Informatics Training Curriculum, with funding from the JRS Biodiversity Foundation.

Caption and Translate

    Sign In/Register for Dotsub above to caption this video.