[MUSIC] Hi guys. Welcome to the 12th lecture on my course, biological diversity, theories, measure and data sampling techniques. Today, I will continue to explain to you how to analyze and measure biological diversity. So, let's continue discussing how to sample biological diversity. Of course, it is impossible to collect all species in a community. So our aim should be to collect the greatest number possible of individuals, sensor, or species. A general tape is that it's better to collect many small samples than few big samples. Another important thing is to ensure to get right and permission to sample. For instance, you have to ask national park or protected areas authorities to be allowed to enter the park and to collect your data for scientific purposes. You need to know statistic. That's a very important point, and otherwise, if you don't know, okay it's better to know something. It's better to learn how to carry statistic, but at least it's important to know statistician. So, in order to sample biological diversity, we need to select the target community. And for doing this, we need to chose the sampling frame. The sampling frame is just the range of the sampling that can be a political or geological or a administrative boundary. And we always need to specify in our papers, in our research report the sampling frame. When sampling biological diversity, it's very important to mind the accuracy. Accuracy is the function of the bias and the sampling error. Also, of course, we need to mind the bias. We have two different types of bias. Measurement bias and sampling bias. To reduce the bias we need to increase the number of samples and to vary the sampling technique, for instance, just increasing the number of leeches. So, we need to mind the sampling error. Sampling error can be detected unlike bias, measured and quantified and reported into the papers as a standard deviation, error standard, etc. Sampling error is kind of equivalent to genetic drift in evolution, I've explained to you, when you're starting to sample populations, then you draw, and then you stop and draw and stop again, and you have the ratio that is changing from the beginning. Then we need to select the sampling unit. A sampling unit is a square plot, a circle plot, a plot of grid or something else. Choosing between design is a key question in biological diversity sampling. So, how to choose between the sampling areas? It depends on the experimental question. For instance, if we are working in gradients, probably the best is to use trisects. Or in this case, ensure the compatibility of different sample techniques. If you are walking in relative uniform sampling area, we probably need to random our sampling so if done frequently enough. We need to get equal representation of our areas that are included. We may use green, that are very useful when need to uniform samples in the area. Green can be a good choice when we need to use a regular intervals along two dimensional design. Transect are very useful, instead, when you have to sample with reference to a straight line. Random sampling can be used to side point quadrants, quadrants, or other sampling methods. There are other types of stratified sampling, for instance, as a very important condition that every element of the population must be presented in only one stratified layer. We have different citing techniques. In short, a citing technique is just any viable form of collecting samples. So, we need to be cited at the level appropriate to the question. For example, We can use point quarter in the proximity to a central point within a cross. We need to use product if just sampling within a small area. We can use pitfall traps. We can use beetle sheets, we can use meat snack things, seining, etc. One of the most used sampling technique in plant survey are plots. Plots can be of different types, can be nested, can be winded course plots. It depends on the sampling that we are going to organize. In animal survey for instance it's very important to understand how to work distant sampling technique. In this case, we just note the distance from our object from the angle we spotted, and the angle from our line, that is a for instance, and the animal we spotted. That's very useful, because we can construct a kind of histogram in this way, and use this histogram to estimate the number of animals that are in the area. Another key aspect of biological diversity sampling is replication. Why we need to replicate? We need to replicate because we need to control for random and stochastic error. For instance, I'm testing the independent factor that may otherwise determine the outcome of the experiment. We need to increase the precision of our test. We need to increase the generalizability of the test. So, if you test across many sites you can safely generalize too many others. We have different definition of replicates. For instance, to consider a sample a true replicate we need to maximize these in our experimental design. The greatest number possible, of course, but given logistical limitation. So if you are very professional in what we are doing, we need to use a power analysis. Otherwise, the sub sample can be absurd to replicate. It means that this is only true if the sub samples are incorrectly treated as true replicates for statistical analysis. So, sub samples are useful to increase the accuracy of the data estimate for the replicate. A special type of statistics analysis are therefore possible. So cellular replication is due to incorrect replication, for instance when we are replicating samples and not treatments, or when we replicate without independence. So the problem, the main problem the problem is that this violates the key assumption of statistical analysis. So, for instance, the independence replicates, we need to increase the precision of studies if independent. We need to approximate through If vital if they are independent. We need to account for normal random error. We need to allow to set our key factor such us alpha factor and keep it constant. And all of these are violated if our samples are replicated I'll provide you an example. For instance, we have two treatments, the question is what is the effect of Treatment A and Treatment B? If we do cellular replication, we treat each stock inside the square, that's our sample of the same color as a replicate. If we do the right replication, we include only a single star of each color or their average. It's very interesting to notice that almost 48% according to cellular replication. And 71% of studies that are using ANOVA, that's a common statistical test and design errors. So this is important because particularly [INAUDIBLE] in studies with logistic problems for instance for [INAUDIBLE] animals for transportation or foreign as a limitation. But most of these studies are already in print. Some examples of cellular replication, when we have many samples from a single side, these are actually sub samples. [INAUDIBLE] that only a single for each treatment condition. These are actually replicas, but we cannot use statistics on a sample size of one, or single samples from a single side, but replicating time would be true samples if the experimental question is time dependant. Vice versa, it just cellular the replication. So how to control [INAUDIBLE] cellular replication. We need to know our question. Question that time whether the design includes cellular replication. If we are working on taxonomy levels different from we are working on ecological level. So, we clearly need to define our independent and dependant variables, what constitutes the [INAUDIBLE] for instance if we plant [INAUDIBLE] in the [INAUDIBLE] population. So, we identify what is the true unit of repetition, is very important for to [INAUDIBLE] of them are the level of replication that are needed for ecological studies. So, we need to randomize our sampling design. This kind of randomization allows us to decrease sampling error and to increase the accuracy of estimation. There is a kind of stoppage rule on how many samples we should take for a good study. Someone suggested that we need to stop at the reference point, that we can get from literature, for instance. Some others suggested that we need to stop at the sample gathered from a statistical distribution of a ranked abundance plot. Another option is to select the minimal sampling area from the species area curve. Or to estimate the samples number from the Chow, Alpha Fisher, or other indices. So I hope that information that I provided you on how to analyze and measure biodiversity will help you to fulfill the tests that are included in this course.