WFSC 408 Ex 3

systematic placement on a grid intercept sampling method

grid that has been created both horizontally and vertically and having a vegetation sample at each grid intercept

resource units

habitat units, points in the habitat or food items

random coordinate placement sampling method

having a grid but not taking samples at each intersect but random

Total counts on sample plots: Strip count equation transect count equation (workout problem)

(Better definition above) - Density is the ratio of the sum of animals counted to the sum of area surveyed - Density is multiplied by the size of the study area to obtain populations size: N=DA - combined into a single equation- the simple strip abundance estimator Transect Variables: N= Population abundance D= density of animals in strips A= area of inference (Study area) a= area of each strip (Lx2w) x= number of animals seen on each transect w= preset .5-strip width (sample are on one side of transect line) L= length of transect ns= number of strips surveyed - why do you multiply LWNs by 2 • the survey area is on both sides of the line

Terrestrial vegetation assesment

- fairly straightforward - techniques vary with vegetation type • canopy coverage º Grass and forbs with tape measures º trees require angle gauges, or spherical densiometers

Total counts on sample plots: Point count equation workout problem

(better definition above) Variables N= population abundance A= area of study area X= number of birds seen n= total points sampled π= pi (ratio of the circumference of a circle to its diameter) r= present radial distance

Census

- A total count of an animal population (entire population) - usually not possible except in specific circumstances (rarely done) • rare or endangered species • small well defined areas such as islands • often can't be made with certainty and there is way to determine bias nor assess the precision - the data obtained are not a sample, but rather from the entire population - in most cases data resulting from these methods should be viewed skeptically

Incomplete sampling counts: modern distance sampling

- Based on the assumption that detection probabilities decrease with distance - distance data from the sample itself is used to estimate the specific shape of the detection function • estimated for a particular target species and set of conditions - more likely to detect something that is closer to you rather than something that is farther away - same types of survey methods used (ie. animals sighted along a transect) - analysis is more complicated - distance data can also be binned into categories (distance) • ex) 1-4.99 meter, 5-9.99m,10-14.99m Assumptions: 1) points or transects are located randomly with respect to the distribution of animals 2) all objects at the center of the point or transect are detected with certainty 3) objects are detected at their initial location 4) distances are measured accurately (ungrouped data), or objects are counted in the proper distance category (binned data) 5) objects are detected independently

Consider prior to marking

- Can natural markings be used instead - Do animals need to be individually identifiable or can they be marked as a group - Can animals be marked without capture - How visible do the marks need to be • Might affect an animal's behavior and survival • Separation of effects from capture/handling from those caused by the marking method have not been evaluated in most cases - Will the marking method cause pain and/or decreased survival - Will the mark affect the animal's health, reproduction, movement patterns, and/or behavior? - Will the proposed mark last for the duration of the study - Will the proposed marking method interfere with other studies - Will the marks attract public attention - Do your permits allow the proposed mark

Indicies

- Density index (density indicies) can be defined as any measure that correlates with density • any type of measurement that is correlated with the actual density of the animals in a given area - most indicies collect frequency (number of individual animals or animal sign) information along transects, at quadrates, or points. - inexpensive and simple to perform (no special equipment) but can be labor intensive - differ from population estimation methods because: • only relative abundance or relative density can be derived • little or no attempt to correct for incomplete detection or variable detection probability (not considering what animals we are missing or varying detection probabilities) - used to compare animal numbers between treatment and control areas or the same area over time • assumes only change is relative abundance • detection probability should be similar (detection probability not taken into account at all) • standardize season, time of day, weather conditions, habitat and observer experience - we are not going to talk about specific indicies but here are some examples - often species/habitat specific - pellet count • fecal droppings > Need to consider: 1) can you be positive that the species of interest actually created those pellets, 2) Need to consider the habitat (bear ground will be easier to see than dense vegetation) - track count • movement tracks (foot) > Need to consider: 1) if you are just sampling on trails where traveling is easy then the tracks can be misleading 2) they also are creatures of habit that continuously move back and forth on the same track even though its just one animal going multiple times. - nest count • Need to consider: 1) can be easy to count during breeding season , but could be hard outside outside of breeding season, 2) can be hard to determine which nests are created by what species

Reconnaissance survey

- ID what to sample - determine what environmental factors will influence how and when to sample - overview of vegetation structure - on the ground or with aerial photogrophy

Precision

- Is a measure of the variation in estimates obtained from repeated samples • Possible to be precise but not accurate > when you weigh a bird you use a bag and a spring scale the weight of the bag should be zeroed out thus to not interfere with the actual weight of the bird. If you forgot to zero out the bags however you would have the weight of the bag included in the measurements • however since there was only 1 bag being used all at the same weight the data would still be precise however not accurate • if there were more than one bag being used the results wouldn't be either precise or accurate

What is a bow net?*

- It is used to catch birds and is a large folding hoop kept under tension by springs. It is normally baited, with a rigid outer edge that closes really fast. It can easily hurt the animal if they are too close to the edge. To counter they use bigger nets (diameter) however they have to use more tension (stronger springs) that could hurt you.

Frequency of occurrence

- Observed number of an attribute relative to total possible number of that attribute - because you are not taking into account the total number of animals seen just whether they were seen or not you are not going to be able to get the true abundance but you can get the relative abundance • individuals was observed on 18 of 20 spotlight counts • Ex) Stomach contents in 100 coyotes. Collected Watermelon seeds in 75 coyotes we could then say the frequency of occurrence of watermelon seeds in our sample area is 75%. - since it is a simple yes or no question the data it is easy to collect and analyze and with high levels of accuracy

population estimate

- a numerical approximation of total population • we don't know the real answer º "best guess" based on our data, assumptions, and estimation technique - ideally the estimate is replicated in a short time frame to understand how precise it is • if each estimate is similar, can be useful as an indicator of population trend even if it is not accurate • however, if conditions change this my not be true • seasons effect this

Population index

- a statistical that is assumed to be related to population size • most indices collect frequency (number of individual animals or animal sign) information along transects, at quadrates, or points

Incomplete sampling counts: Double observer

- Rather than using 2 different sampling methods we use one method and have multiple observers Independent observers - two observers collect observations along the same transect independently - location of each animal spotted is recorded Categories: 1) seen only by Observer A 2) seen only by Observer B 3) seen by both Observer A and Observer B Assumptions: - Observations must be independent or it will cause issues Problems: • violated if the activity of one observer alerts the other observer to an animals presence • different observers should be used to ensure independence - category assignment must be accurate • conducting surveys simultaneously decreases the chance that categories are misaligned • easier with stationary targets (nest, burrows) - targets must have equal detectability (but observers do not need to be equally good) • not always good • better than dependent observer method because each spotter has their own detection probability Dependent observers - two observers work in tandem - the primary observer detects animals and tells the secondary observer - the secondary observer then records any additional sightings independently - Categories: 1) seen by both primary and secondary observers 2) seen only by secondary observer Problems: - if detection probabilities are variable population estimates are biased - assumes observers have equal detection probabilities - probably not as useful as independent double observer method • practical or logistical reasons preclude independent observer method • independent observer method is more precise, simpler to understand, and allows the 2 observers to have different sighting probabilities

Sampling schemes to avoid non-probabilistic sampling

- Systematic sampling - Simple random sampling -

Mode

- The value that occurs most frequently in a dataset, arrange numbers form least to greatest and then determine which number is repeated most

Incomplete sample counts: Double sampling

- Using 2 different methods within the same study area 1) select set of random points 2) Then randomly divide them in half 3) Have 1st half do method 1 and then have the second half do both method 1 and method 2 • Method 1: fast and inaccurate • Method 2: slow and accurate - we do this because labor is often the most expensive so the longer we take the more expensive it becomes • if we do fast inaccurate methods we are saving money • we then use the slower but more accurate count (nearest neighbor) to get a more accurate idea of density but on a smaller number of points. - Now that we know both densities for the 2 groups we can get an idea of how inaccurate the group who just used method 1 (fast & inaccurate method) was by comparing it and correcting it to the data that used method 1 and method 2. Assumptions: - method 2, the intensive slow method, is a real representation of the density of individuals in the study area - both counts are performed at the same time or fairly close

population estimator

- a mathematical formula used to compute a population estimate calculated from data collected from a sampled animal population • many different ways to estimate population

measuring food availablity

- abundance and distribution of food influences habitat selection • fundamental component of behavioral ecology • energy needs, reproduction, and survival - abundance can be estimated using vegetation analysis techniques or assessing the abundance of potential prey - availability suggests food is accessible and usable • varies with weather predators or competitors

Remote vegetation surveys

- aerial or satellite based technologies can measure regional vegetation patterns at the largest scale of sampling • variety of sensors have potential applications

what makes a good food habitat study (4 characteristics to think about when designing a study): Data throughout the year

- animals do not eat the same foods year round • all foods are not always available • can be present but not available (after heavy snow or heavy rain or predators animals might not have access to specific food - animals do not eat the same foods year round • nutrition needs change throughout the year > birds preparing for migration > rearing young/lactating or pregnant individuals require more or different types of nutrients

Plotless methods (still total count methods)*

- area counts commonly used for both plant and animal sampling • What problem do plotless methods avoid compared to the use of traditional plots?:* Removes boundary effect, which is when we must determine whether or not to include each target observed on a plot boundary in the sample which can be time consuming. - Benefits: • removed the boundary effects • popular because they are fast, easy to perform • remove decision making and don't require much training • target must remain in place or can be measured before it moves (has to be done with a high level of accuracy not estimating) > useful for animal indices like nest, burrows, pellets, or other, animal, activities that remain stationary Assumptions: 1) rely on randomly sampled points • may not be practical (difficult to access in water, etc) • random points to select random individuals is biased towards isolated individuals 2) Distribution of the target species: •good for random or uniformly distributed targets • problems if target is clumped Examples of plotless methods: 1) Point-quarter method 2) Nearest neighbor 3) Nearest Individual

Incomplete sample counts (not total count methods)

- assuming you have an incomplete sample no matter how small of an area you survey and instead you seek to find how many animals you are failing to detect - estimates detection probability directly or collect data to indirectly model detection probability 1) Double sampling 2) Double observer 3) Modern distance sampling

Estimating area

- attempts to determine the sample area in which no animals were missed • distance at which the number of animals counted beyond that distance is equal to the number of animals missed within that distance - estimate the are (size of the plot) surveys Counts on sample plots (estimating area) - methods should allow all sightings to be used - labor is expensive and throwing out sightings outside the sample frame created more cost without an increase in precision - attempt to determine the sample area congruent to the area over which counts were obtained - early methods used sighting distances to estimate sample area • every time we sight an animal we also have to estimate how far that animal is (different methods measure this at different ways) - most not particularly accurate - these are traditionally distance sampling, better methods (modern distance sampling) exist, but some of these are still commonly used - methods that accurately estimate sample area surveyed to obtain counts without preset 0.5-strip widths - precision was proportional to the square root of the number of animals seen

Total mapping of bird territories

- breeding birds are trapped and color banded - surveyed in the sam manner as territorial mapping • thought to be accurate way to established population density of breeding birds (more accurate than spot mapping or territorial mapping)

Variance of a sample (workout problem)

- calculate sample mean - subtract mean from each sample value - square these values - add these values and then divide by one less than the number of samples - used to describe the spread between the numbers in a data set. The lower this number is the closer the individual measurements are to the sample mean • Can calculate 2 ways: > population data: you have measured every individual that you were interested in and then you divide it by the total number of samples • example: you took the weight of every black footed ferret in captivity) > sample data: we only have data from a subset of the population, we then subtract 1 from the number of samples and then divide by that (white-footed sparrow) - this means that the variance of a population will be smaller than that of the variance of a sample if the values of each sample and the number of each samples are the same in both examples

Vegetation sampling: Field data collection

- can be electronically entered on site - can be identified to specific transects, quadrats or points using GPS locator UTM coordinates

Vegetation

- can refer to a single plant/species on a specific site or a community in the landscape • natural or introduced, live or dead

Lidar

- characterize 3-dimensional canopies at fine scale resolution • can tell you how much canopy there is in an area, how many layers of canopy there are in an area

T-test (be familiar but don't have to calculate)

- compares two means and determines if they are different from each other - tells you how significant the differences are • could the difference be due to chance or not (if its not then its significant)

Complete sample counts

- completely count a sample by counting every individual within a small area and then using that information to determine how many individuals occur in a larger area - goal is to actually count every individual in a given area by either reducing the study area to a few survey areas to assure complete detection of every individual and then extrapolate, or don't use set plot size at all (not a census which attempts to count every individual within a study area) - goal to either standardize or estimate the parameters (like distance) necessary to obtain accurate estimates • no evaluation of detection probability

Study site selection

- critical for vegetation study - directly related to the objectives • many factors influence selection study sites (soil types, management history, human disturbance, topography (flooding) - select sites so intersite variation is natural • not affected by some factor not accounted for in the objectives or study design • mapping all vegetation types, their location, and size may be required

what makes a good food habitat study (4 characteristics to think about when designing a study)

- data throughout the year - data from all age/sex classes - must be associated with availability data - must answer why animals are eating what they are eating

Chi square test: goodness of fit test

- determines if a sample data matches the population as a whole • can use it for non-normal data - need expected and observed numbers for a given trait - expected values calculated using proportions • used to determine if animals are using their resources in proportion to their availability or if they are selecting for or against a certain resource

Median

- determining the middle value in a data set • calculate: put all numbers in order from least to greatest and then cross off one number at a time taking turns from the right to the left until you meet the number in the middle (if there are two numbers that make up the middle you should take the mean of them by adding them together and dividing by two.

Vegetation sampling methodologies

- develop objectives - identify vegetation aspects to sample • vertical and/or horizontal spatial distribution • surrounding environment (landscape structure) • temporal variation in structure •species composition • overall stand structure • biomass

what makes a good food habitat study (4 characteristics to think about when designing a study): Data from all age/sex classes

- different nutritional needs • diet of many animals is substantially different between the juvenile and adult classes - young northern bobwhites eat many insects but adults are primarily vegetation) - may be prevented for accessing better resources by older/more dominant individuals

Ground based vegetation surveys

- digital cameras - passive sensors

Quantification of Plant use

- estimate carrying capacity • stem-count method (used and unused stems) • browse use estimated by dimension analysis > predict pre-browsing lengths or weights of twigs from diameter-weight or diameter- length relationships

Standard error (workout problem)

- estimate of how much your sample mean deviates from the true mean - divide the standard deviation by the square root of n - larger sample size the smaller your sample of error (need to do variance of sample first (look above), then find the standard deviation, then divide that by the square root of the total number of samples)

Hayne method

- estimates population density of flushing birds - assumed a fixed flushing radius for each bird species and habitat - An observer walks on a transect and counts the number of birds flushed - assumes the sine of the angle for each observation came from random distribution ranging from 0-1, with an average of 32.7 degrees. • later determined to be around 40 degrees

Use of vegetation measurements

- evaluate vegetation response to management - estimate carrying capacity - characterize cover and habitat components for species - long-term monitoring of plant vigor or vegetation type condition

Techniques for sampling fruits

- few habitat analyses include an inventory of fruit production • important when target species utilize fruit • traditionally just number and size of fruiting plants - generally are referred to as mast • soft mast > fleshy exteriors ex) berries, drupes, and pomes • hard mast >dry or hard exteriors ex) achenes, nuts, samaras, cones, pods, seeds, and capsules Large or heavy fruit of trees: - if interested in landscape level mast production need large numbers of points - sometimes best to only sample under the canopies of mast-producing trees - mast production can be estimated by counts of mast in ground plots, counts of mast on trees, or use of seed traps Small or light fruits of trees: - seed traps - direct count (binoculars) • if evenly distributed can use extrapolation - categorizing relative abundance Fruits of shrubs and bushes: - counted or harvested directly from the shrub fruits of herbaceous vegetation: - samples may be take from the ground, from traps, or directly from the plant

Sampling schemes to avoid non-probabilistic sampling: Stratified random sampling

- first place experimental units into categories and then randomly select from each category - implicit differences (before treatment) that must be accounted for in the analysis • experimental units are categorized and sample units selected randomly from within categories

Techniques for Sampling vegetation

- frequency of occurrence - density - cover - biomass or standing crop - visual obstruction - vegetation height

Vegetation sampling: Preparations and getting started

- good leadership is essential - initial planning and preparation • list of supplies and equipment - develop data forms - small- scale preliminary field trst - training of the field crew

Techniques for Sampling vegetation: Vegetation height

- height of herbage • easiest vegetation attributed to measure in grasslands • can be estimated with high precision in many grasslands - effective plant height • usually is the maximum height or leafy cover for grasses and forbs • can be shorter if plants naturally droop or bend to a certain height (more effective in shorter plants)

stratified random sampling method

- if the area you are sampling is not uniform you might need to stratify your sample - pick the number of points you are going to do and equally divide them based on the amount of landcover that they have in your sample - then appropriately distribute based on size (left is correct right is wrong)

Central Limit theorem

- if you take an infinite number of random samples from a population and calculate a sample mean for each, the distribution of means will be approximately normally distributed - even if data is non-normal, the distribution of the sample means has an approximate normal distribution if sample size is large enough (usually at least 20) and all samples have the same size - useful because statistical analyses of normal distributions are easier

Techniques for Sampling vegetation: Biomass (standing crop)

- included both live and dead vegetation - total biomass and biomass of edible components • estimated directly by clipping and weighing (dry it first so you don't get the weight of water) • estimated indirectly (less accurate) by measuring the size of the leave or the stem and estimating

Avoid convenience sampling

- individual animals should be equally likely to be included - NOT • animals brought in by the public • animals that are easy to get to (close to roads/trails)

Spot mapping or territorial mapping

- locations of individual birds seen or heard are plotted during repeated visits by observers • clusters of locations of birds are assumed to be individual territories. • Total number of individuals equals the number of clusters, plus the sum of any fractional parts of clusters on the boundaries multiplied by the mean number of birds per cluster (usually 2) - probably not a complete count, as results can vary among observers and map analysts - at beset, spot mapping yields an index • can double count individuals (no marking) ex) • 4 individuals per territory and 3.5 territories on the total study area. We would then calculate the total population of birds on our study area by multiplying 3.5 x 4 to get 14. • However if our mean number of birds in a territory was 2 we would have multiplied it by 3.5 and have gotten 8 birds in our study area. • The larger number of territories there are and the more far off the mean is from the actual number of individuals there are in the territory the larger the error can be in a given sample.

Population estimation units

- population estimates include some sort of distance/area unit - make sure all measurements are in the same unit • convert all meters, kilometers etc > 1 Km = 1000 m > .1 Km = 100 m > .01 Km = 10 m - note if total study area is already give (2 Km^2) or if it needs to be calculated (2 Km x 2 Km = 4 Km^2)

Treatment types

- manipulative (applied by the experimenter) • the researcher is directly changing to observe the impacts of these actions (ie. prescribed fires, presence or absence of predator control, restoration of native plants) - changes in space or time • calculating biomass in an ecosystem over a multi-year period, changes in an animals body condition over time, size variation across a species range, plant species composition over grassland habitats - organismal (natural categories like age/sex) • example Identifying nutritional needs of different age classes, or determining if males and females have different range sizes

How to determine "importance" (what food item is the most important to each species)

- many different ways to determine which food items are most important • be clear about what methods are used - 4 ways to determine a foods importance 1) number 2) weight 3) volume 4) frequency of occurrence (rank from largest to smallest in all categories for each food type, then add all the rankings for each individual food (do not average just add) then pick the one with the lowest value)

Standard deviation (workout problem)

- measure how spread out the data is - square root of the variance (look at the sample variance above and then square root the final answer)

Techniques for Sampling vegetation: visual obstruction

- measurement of horizontal cover • used to assess wildlife habitat suitability, habitat preference, and impacts of land use practices on wildlife habitats - cover boards: • used to index or quantify cover on provide visual record to changes in coverage when photographed from the same reference point (cover board with checkered squares) - robel range pole: • use to obtain visual obstruction estimates from a specific height and distance - staff-ball method • fast and accurate the point where the ball meets the staph that supporting it is either obscured by vegetation or its not.

Why sample vegetation?

- mixtures of plant species provide wildlife with the 3 essential habitat elements • food • cover • water - vegetation quantity and quality within habitats are influence wildlife research and management - inferences about the total plant population within a given vegetation type

Aquatic vegetation assesment

- more complicated submergent, and emergent plant species - often conducted while wading, from a boat, or from aerial photography - floating equipment a good idea

Overcoming problems with population estimates

- multiple samples - random samples

Traditional ways to ID food

- observing animals - visual examinations of stomach/crop contents or fecal samples - the farther down the digestive tract the food gets the harder it becomes to identify • even in the crop digestion has started • sometimes possible to guess food items based on color or other characteristics • histological examination

Marked-resight (mark recapture studies)

- percentage of marked animals in the population will affect the accuracy of the estimates • ideally, at least 25% of the population should be marked Marked Sample - marked animals within the population estimate detection probabilities - marked animals "released" into the population, an available for detection during the survey - marked and unmarked animals are counted and the probability of detection of the marked animals can be estimated - equal detection of marked and unmarked targets - actual number of marked animals is known - telemetry can determine size of marked • can't be used to find animals during the survey - individually identifiable marks not required if number of marked animals is known - any marked animals not "known" to be present prior to the survey are treated as unmarked Only 1 Assumption: - the proportion of marked to unmarked individuals is the same in both the sample and study population • many not be true though - can be violated by: • have to be sure we are counting every individual that was marked as marked and not counting unmarked individuals as marked • if animal loses mark overtime • the actual number of marked animals is known (will be smaller at the end of the study due to death or markings going away) Marked-resight variables: N=total population size in the surveyed area n1= number of marked animals present in the area at the time of the survey n2= number of animals (both marked and unmarked) seen during the survey m= number of marked animals seen during the survey ß= the probability of detection • Lincoln-petersen estimator is biased at small sample sizes; chapman estimator is less biased - Chapman estimator • more accurate when working with smaller sample sizesx

Exploitative/removal methods

- popular because it relies on others such as hunters, to collect data - removal doesn't necessarily mean removed (catch and release, photographed, etc) Exploitive Assumptions: - population is closed - all removals are known - estimates are imprecise and work well only when a large portion of the population is "removed" Removal methods: - as more animals are removed, fewer remain to be "caught" - catch per/day will decline - developing a linear regression Catch per unit effort: - Animals are captured and removed from the population • removal could be part of normal activities such as hunting • each day fewer animals are available and so catch per/day will decline • animals don't have to be actually removed, could be marked - eventually the expected catch will become zero • total number of animals removed equals the initial population size • in practice we never get to zero, but use a linear regression to estimate the total population - not likely to be accurate unless a large proportion of the population is removed Assumptions (catch per unit effort): - sampling units taken at random - population is actually closed - all individual animals have an equal probability of being caught - unit effort is constant - all the removals are known Change in ratio: - used to compare change in ratios between two groups (doesn't have to have = chance of being captured and marked) - can be used on any 2 classes of animals as long as harvest varies between them • 1 sex or age class favored over the other • 1 species favored over the other • 1 species hunted; other not Assumptions (change in ratio): 1) proportion of the classes will change after the hunt due to selective harvest of one class over the other 2) observed proportions of the 2 classes are unbiased Change in ratio variables: N1= Pre-hunt populaiton T= Total kill (all animals harvested regardless of class) F= number of unfavored in survey before hunt p1= proportion of unfavored in survey before hunt p2= proportion of unfavored ini survey post hunt N2= Post-hunt population • To determine number of animals in population post hunt N2 subtract T from N1

Assumptions of spot mapping and total mapping

- populations are constant • birds should be the same in May and July - birds remain within territories during sampling period - songs/calls are frequently enough to be observed on multiple visits - estimated proportions of territories along boundaries are accurate - estimated mean number of birds per cluster is accurate - observers are skilled and consistent

95% confidence interval (workout problem)

- probability that a given estimate will fall within +- 2 standard errors of the mean - with 95% confidence the population mean is between x and y How to calculate: - calculate the mean - find the standard error (look above, have to do variance, then standard deviation, then you calculate for standard error) - once you have your standard error multiple it by 2 and then add and subtract the that number from the mean to get your lower and upper limit

Techniques for Sampling vegetation: Frequency of occurrence sampling

- proportion of samples in which a species occurs • low frequency, plants are clumped • high frequency, plants are uniformly distributed - when comparing between time or communities size and shape of the sampling unit is important • smaller the plant the smaller the sampling unit º herbaceous vegetation (1-2 m^2) º trees (100m^2)

Techniques for Sampling vegetation: Point-centered quadrat

- randomly or systematically select points - measure distance to the nearest plant within each quadrant around the point

Replication

- repeated sampling from each experimental unit • multiple sampling plots within each unit (sample multiple places within a a single burn site in order to get a more accurate result regarding the different plants that are within a single plot not just one specific area. Some areas within the plot may have burned hotter in certain places or some areas in a plot may have more fire resistant vegetation so it is important to take multiple samples)

Avoid selective sampling

- selecting areas where animals are more abundant • example: 95% grassland and 5% tree cover. Determining fox and squirrel population abundance. If you look only at animal counts from the tree covered area (5%) you will have many squirrels however if you were to translate that data to the 95% grassland area the data would be misleading because you would be assuming the same squirrel density for grassland as it is in the trees. - selecting animals you think are "typical" • example: deer point counts, excluding animals that you think are abnormally small or consume odd dietary needs

Experimental unit

- smallest entity to which a treatment can be randomly assigned (can be an individual animal or an area like a field) • homogenous • representative of the population or treatment to which inferences is to be applied - ex) can be assigning anti-parasitic drugs to certain individuals or determining which 10 acre plots would need herbicides to remove an invasive plant

Method considerations

- some animals can be problematic to sample • clumps (not randomly distributed- in heard, packs, or clustered around resources) • sample distribution correlates with animal distribution - most survey methods do not observe all individuals within the population - detection probability may vary

Problems with spot mapping and total mapping

- species must vocalize often (or perch conspicuously) • if they don't you will be missing many individuals - only estimates the population of relatively conspicuous birds holding territories • does not take account floaters or other non territorial individuals

point-quarter method (workout problem)

- square each value of x and then add them together

Techniques for Sampling vegetation: Density

- total number of objects (individuals, plants, seeds) per unit area) - measured using quadrats or plotless methods • absolute density determined using quadrats º square, rectangular, or round - relative density determined if points or line transects are used

when an animal is unable to eat its desired food or obtain specific nutrients they may be

- underweight - undersized - not produce as much young - be more at risk for predation - more susceptible to disease

Ways to account for detection probability* (1 way to limit its impact on population estimation)

- standardize methods • recognize sources of variation in detection probability and control those that can be controlled > methods, effort, observer experience, weather, time - use of covariates in analyses of survey statistics • develop models to estimate change in population size as a function of the relevant variables (observer, weather, etc) • covariates cannot be associated with both detection probability and true abundance > ie vegetation type can influence both detection probability and actual abundance - detection probability is not constant over space or time and not all variables can be measured modeled, or perceived • best estimated directly from your exact sample and situation - implement methods that permit direct estimation of detection probability • not commonly done (hard to do)

Total counts on sample plots

- strip counts - point counts

Normal distribution

- symmetric bell shaped curve, distribution in which there is a singular hump and within the hump the mean median and mode are all occurring because they are equal (50% of the values are less than the mean and 50% of the values are greater than the mean) - continuous for all values of X between negative infinity and infinity so that each conceivable interval of real numbers has a probability other than zero - many things actually are normally distributed or very close to it - easy to work with mathematically • methods developed using normal theory work quite well even when the distribution is not normal

Analysis of variance (ANOVA)

- test for differences between multiple groups that could be interacting - can look at interactions between categories - more conservative (less chance of Type 1 errors) • single test that is not likely to have false positives

Mean

- the average value in a data set • calculate: adding all numbers within the data set and divide that sum by the total number of observations

Range

- the difference between the lowest and the highest values • calculate: put numbers in order from least to greatest and then subtract the smallest number from the biggest

Sample unit

- the entity form which measurement are obtained • could be animal, plot, etc • can often not be equal to experimental units

Sample size (n)

- the number of samples required to obtain the desired precision is an estimate, or the desired power in a hypothesis test - Having the correct sample size is critical • if too high resources are wasted • if too low, the information obtained may be incapable of producing useful results, leading to incorrect conclusions

Tree structural characteristics

- the size of characteristics of individual trees affect forest structure - three common, interrelated measures of tree size are height, crown volume, and trunk diameter - tree height may be measured using a standard clinometer, a laser rangefinders or hypsometers (tall trees less accurate) - crown volume calculate by measuring minimum and maximum canopy height and horizontal canopy diameter - trunk diameter and cross sectional area can be directly measured with diameter tape or with calipers • by convention the measurement (DBH) is made 1.5 m above ground level (rest height or breast height) - tree age determined using growth rings • trunk cross-sections • cores obtained with a wood increment borer

Survey extent

- the spatial and temporal extent over the area over which inference is to be made - can be defined geographically or temporally

Total counts on sample plots: Fixed area method

- total counts on limited sample areas may be possible • must be able to counts every individual (depends on species) - the length/width being surveys is defined prior to the start of the survey and held constant - the mean density from sample plots is used to extrapolate across the entire study area - sampling shape (quadrats, strips, plots, etc) and size, dependent on circumstance and target species - both have predetermined survey area 1) Strip counts - one of the post commonly used methods - the counting unit is a strip or transect (a long thin plot) with a fixed area - Assumptions • transects are randomly selected • transect lines is then transferred and all animals within a predefined distance (on either side) are counted > animals observed outside this distance are not counted, and it is assumed that all animals within the strip are counted with certainty 2) Point counts - observer detects animals for a specified time within a preset distance from the point • before counting, a set amount of time may be allowed for animals to settle down • considered to be an index of relative density - Assumptions • all animals are detected within sample radius > often false unless the area is small or animals conspicuous • points are random

Sampling schemes to avoid non-probabilistic sampling: Systematic sampling

- uniform coverage of area under investigation • takes out biases - randomly assign the first point and then 50m away from that point will be the next and continue from there

Bias

- used to describe how accurate our data is. - it is the the difference between the parameter (true value) and statistic (estimated value). - however, without knowledge of the true population size, bias unknown • the larger our sample size the less bias except in population estimates • repeatedly performing population estimates will not allow us to estimate bias however, this doesn't mean we shouldn't be taking multiple samples

Time-area squirrel survey workout problem

- used to estimate tree squirrel abundance (but can be used for any type of animal with a few modifications) - point-based example of using distances to sample area of the counts - select random points - start at sunrise and lasts for 30mins • bc count is so standardized it helps make it easier to use over different sites Differences - survey area is determined by average distance to each squirrel when first spotted - observer can't watch entire circle and the proportion observer is variable • use compass to estimate the portion of a circle viewed factored into the estimation equation

King method

- uses average radial distance to all observed animals to estimate of .5-strip width used in the calculations of animal abundance - formula very similar to Hahn formula since the method only differs in how the width of the strip is calculated (uses average radial distance rather than average visibility. Variables: N= population abundance A= Area of the study Xi= number of animals seen on each transect r(bar)= the .5-strip width determined by average sighting radius L= total length of all transects*

Kelker method

- uses perpendicular distances to generate a histogram - from the histogram subjectively determine the 0.5-strip width over which all animals were likely detected • the point is very subjective - often not very accurate: because its so subjective not very repeatable

Survey design

- usually not possible to survey entire area of interest • a sampling scheme is devised to select a portion of a study area to be sampled (experimental units) • minimize the effects of uncontrolled variation (differences that aren't caused by what we are studying but rather something that is just already present in the area (ie, the climate or soil type between the different study areas) • ex) 2/3 of refugee got treated for fire ants and the other 1/3 did not the issue was that the areas they treated were a different soil type than the areas they didn't treat. This made it hard to determine if the variation in insect composition between sites was due to the insecticide or the soil type (soil type could have impacted the type of plant communities grown and therefore the types of insects attracted such as the fire ants)

Study scale

- vegetation sampling generally conducted within vegetation patches or field-size units - landscape-level assessments usually conducted via satellite, aerial, or video photography coupled with GIS

Techniques for Sampling vegetation: Cover (canopy cover and basal cover)

- vertical projection of the crown or stem of a plant onto the ground surface - often expressed as a percent • total cover may exceed 100% in multilayered community - canopy cover serves as a criterion for relative dominant within a community - influences many things including light, precipitation, and soil temperature Estimating Percent Overstory cover: - spherical densitometer: used to estimate percent overstory cover in woodlands, mirror with a grid on it and then you count the number of grid cells that canopy cover in them to determine the percentage - moosehorn densitometer: used to estimate percent overstory cover in woodlands Basal Cover - measured at a height of about 2 cm on bunchhgrasses and tussocks - single-stemmed trees measured at 1.5 m above ground (DBH) - Ground surface on trees with multiple stems or on trees with buttressed trunks

Hahn method Workout problem

- very similar to the strip count only difference is the .5 strip width is estimated instead of predefined - transects are randomly places and animals counted along both sides of the transect - density estimates based upon all detected animals - uses maximum visibility measurements to estimate the area over which animals are counted - estimates of maximum visibility are taken repeatedly on both sides of the transect • maximum visibility is the maximum distance an observer could see a target animal perpendicular to the transect at each point - average visibility is set as the .5-strip width - often used for deer populations (because it is easy and cheap) - standardized for use by game agencies - deer are counted along a 2 mile strip by one person walking - visibility taken at 100 yard intervals • one person stands on the line while another walks perpendicularly away • distance at which the walker disappears from view is the visibility Varaibles: N=popualtion abundance A=area of study area Xi= number of animals seen on each transect v= the .5 strip width determined by average visibility measurements L= total length of all transects (only 1 transect) workout* (THE ACTUAL ANSWER FOR THIS ONE IS 916.66)

modern ways to ID food

- video cameras • collars that record what an animal eats • cameras on nest boxes to record food delivery - stable isotopes • good for broad understanding of diet • won't tell you specifically what an animal is eating (broad, meat, fish, or plant based diet) - DNA barcoding • sequence a small fragment for which there are lots of existing sequences IDed to species • can work on fecal samples or from internal samples • easier to do this in an area where all potential food items have previously been barcoded

Sampling plots

- visualizing how sampling plots, or plots on a transect, will appear in the field is difficult • detailed maps of study area are key - many layout designs are possible • depends on the objectives and requirement of statistical analysis

non-probalistic sampling

- want to avoid - results in biased estimates and are not representative of population or area as a whole - results in inaccurate results because you are often introducing biases into your sampling, and therefore the sample is not representative of the total population

Pseudoreplication

- want to avoid (one of the most common sampling design errors) - repeated sampling from the same sampling unit and treating them as different sampling units - why marking is critical

what makes a good food habitat study (4 characteristics to think about when designing a study): Why are animals eating what they are eating

- what does the food provide • fats • proteins • micronutrients

Sampling schemes to avoid non-probabilistic sampling: simple random sample

1 of the 2 ways to randomly sample within a geographic area: - each sample unit has an equal probability of being selected - works well when your sampling unit is pretty homogenous (if its not and you might have multiple ecosystems or some other characteristic that might impact the results of your study other than your treatment, a random sample like this one might result in some ecosystems being over sampled or under sampled and it might be better to do stratified)

target to nearest neighbor or nearest neighbor

1 of the 3 plotless methods: - at a point, lay out a circle of radius sufficient to enclose (on average) the 5 nearest targets - number these individuals and select n at random - from the randomly selected individuals, measure the distance to then nearest target species Variables: D= density n= number of samples Xj= distance from target j to nearest neighbor

Point-quarter method (workout problem)

1 of the 3 plotless methods: - don't have to use it to only describe density • can be used to define community assembly or species composition - classic method for sampling vegetation - Surveyors would locate and describe the 4 trees nearest to each corner of a section of land • estimate forest species - selected points are located and the area around the point is divided into 4 quadrants • perpendicular to the transect if point-transect sampling • compass bearing for random points - the distance from the point to the nearest target within each quadrant is measured • 4 distances are obtained at each point Variables D= point-quarter estimate of density n= number of points sampled Xij= distances from point i to the nearest target in quadrant j SQUARE EACH INDIVIDUAL VALIUE AND THEN ADD THEM TOGETHER TO GET XIJ

Nearest individual (workout problem)

1 of the 3 plotless methods: - measure the distance from the point to the nearest target species

Census techniques

1) drive counts • beaters drive animals into an enclosure or past counters to count all individuals in the study area > best with large, easy to detect animals • drivers remain in sight of one another and traverse the area • preferably under high fence or surrounded by water • additional observers are placed along the boundaries to count animals • count animals that pass them on the right (eliminates double counting) • the total count is the sum of the number of animals moving out o the area or back through the line of drivers, minus any moving into the area ahead of the drivers • error as large as 20-30% > at low densities: underestimated the true population > at high densities: overestimated the true population > probably best viewed as an index of population size 2) aerial photography • low altitude photography can be used as a census technique • animal groups are photographed and later counted • difficult to ascertain if all individuals are visible > count not completely 3) territory mapping 4) total mapping of birds territories

2 ways tor randomly sample within a geographic area

1) simple random sampling 2)

What factors determine sample size

1. significance level • the odds that the observed result is due to chance > type 1 error: rejecting a true null hypothesis (false positive) >> probability of committing this error is controlled by the alpha level of the test (frequently= .05) • the value that reflects the probability of incorrectly rejecting the null hypothesis when it is actually true is acceptable • want to avoid type 1 error so will set alpha level closer to 0 > type 2 error: failing to reject a false null hypothesis (false negative) >> probability of committing this type of error is controlled by the beta level of the test (frequently = .05) > adjust alpha and beta according to experimental needs 2. power • the odds that you will observe a treatment effect when it occurs • increased power requires an increase in sample size 3. effect size • difference between treatments (eg in number of animals seen) relative to the noise in measurement • magnitude of difference between 2 sample means • smaller the effect, the more difficult it will be to find, therefore requiring larger sample size • two means are similar=noisy data = higher standard deviation=smaller effect size OR very different means=not as noisy data = smaller standard deviation= larger effect size 4. variation in the response variable • use to estimate variability in the parameter of interest (eg population mean) • typically obtained from either the literature or pilot study • measurement of how variable your data is > if your data is very tight and clustered around the mean = low standard deviation > if your data is spread out over larger distance = higher standard deviation

Population (sharp shinned hawk)

A group of animals of the same species occupying a given area (study area) at a given time - must be defined • have to define a study area • have to define a time (individuals who occupy an area in the winter may be different than the individuals in the summer) - sharp-shinned hawk, migratory species in some areas of its range and not in other. • in June to July they are found in the Eastern and Western coasts of America and much of Canada and Alaska and not in the great plans (middle of america) because this is during breeding season and they like to build nests in forested areas • in December to February however they are found all across the US and not primarily in Canada or Alaska at all - Sharp-shinned hawks in the Caribbean, not migratory but are not ALL across the island they are sedentary (stationary) problems/ issues - we don't always know what a species is - these sharp-shinned hawks are actually at least 5 species - if prior studies failed to define exactly what population they were studying we now won't know what populations those finding apply to (blue grouse are actually 2 separate species but have not been adequately defined in the past studies which is a problrm)

Advantages and disadvantages of using a cannon net over using a rocket net*

Advantages: - less expensive - charges can be shipped - don't start fires - no federal permit required for use Disadvantages: - can not be mounted above ground to accommodate large animals - cannons must be cleaned after firing

what makes a good food habitat study (4 characteristics to think about when designing a study): Must associate diet data with availability data

Animals eat to survive - preferred food may inaccessible or unavailable • defended by conspecifics • predators • snow/flooding • out of season Ex) we may prefer steak but are stuck with hotdogs - presence doesn't tell full story - are plants defenses impacting food availability • change with season, leaf position, and leaf age • some vegetation is only available to certain species > adaptations to overcome the plant defenses - more complex interactions • ants that live inside specialized structures on plants and protect them from herbivores

Problems with population estimates

Animals may not be evenly distributed: 1) may live in groups all/part of the year 2) territory size is variable 3) resources are not evenly distributed 4) targeting areas animals are known to frequent 5) detection probability is not the same across the study area/observer combination of factors that cause error - deer counts from roads for example • deer preferentially occur near roads during certain times of year (hunting season) • area deer are known to frequent (easy to travel along, food, hunters can't hunt too close to road) • easy to detect - all these characteristics can result in artificially high deer population estimates

examples comparing traditional and modern methods of diet determination

Caribou: Compared the results of 3 different methods of diet determination: - histology (traditional) • only 15% correlated to the other 2 types of methods - DNA barcoding • most accurate - Video • results highly correlated with DNA Bats: Compared the results of 2 different methods of diet determination: - visual examination of fecal material • different results than barcoding because most of the different parts insects were not as likely to preserve as other parts making it easily skewed towards insects that were primarily made up of hard parts rather than soft parts - DNA barcoding

Hahn method vs king method (differences)

Hahn method: - use an average visibility King method: - use radial difference between us and the target animal

Describe flight intercept and malaise traps and what is the main difference*

Malaise Traps are passive traps set along flyways. The placement of the traps determines what they will catch which is primarily Hymenoptera and Diptera species. The light bodied insects then move up on the vertical net toward a bottle filled with ethanol. Flight Intercept Targets insects malaise traps don't like Beetles and Orthoptera species. These insects fall into a trough filled with preservative rather than moving up the mesh net like in Malaise traps

Population Estimates

Need to know 3 things 1) the size of our entire study area 2) Count a number of individuals on a small portion of that area then calculate animal density on the small survey area 3) Use that density of the small survey area to determine the total population of the total area 2 main ways to do population estimates: 1) Reduce study area down to where you can see all animals within your study area and call it a sampling area (not a census because that is counting every animal within the entire study area and in a population estimate you are doing a complete count over a small transect) 2) Don't use set plot size - often just proportions - two key things are collected • number of animals seen • distance from observer (or transect) - using number of animals seen and the distance from observer number of animals/area is calculated for the area surveyed - this is then used to extrapolate how many animals occur across the study area Example: if you see 3 birds on 1 half of the study area you will just multiple that by 2 and estimate that you have 6 birds in all of your cover area) - But is this really the case? • usually not: MORE BELOW

Parameter vs Statistic

Parameter: Is an attribute of the population (ie percent males in the population given you know the sex of each individual and how many individuals their actually are in your population) • real value, if you know the parameters of the population, you do not need statistics Statistic: Is an attribute from a sample taken from the population (ie percent male of a subset of the population) • Do not know the real value--- estimated value based on sampling, can be wrong • the smaller your statistic the more likely it is different than the parameter

study area

consider resource unit distribution, scale, availability, and budget constraints for collection of data

use

consumption of a resource

detection probability

The probability that an individual animal within a sampled population is detected while in a survey • the bird on the right is more likely to be detected because its perched up on a pole, in the open, and is most likely making mating calls to draw attention of the females - detection is variable • can change with time • individual animals º behavioral differences (incubation, nest provisioning) º pure chance • species • observer • habitat Examples - drab birds incubating will have a much lower detection probability than flashy birds feeding young or displaying - Black Throated Trogan might stick out like a sore thumb (or a female that is a bit harder to spot), or easy to overlook if they are facing away from you - Size will increase detection probability Detection varies by observer - some people are better than others at spotting/counting animals

Marked-resight: Chapman estimator (workout problem)

Variables: N=total population size in the surveyed area n1= number of marked animals present in the area at the time of the survey n2= number of animals (both marked and unmarked) seen during the survey m= number of marked animals seen during the survey Less biased

Marked-resight: Lincoln-Petersen estimator (workout problem)*

Variables: N=total population size in the surveyed area n1= number of marked animals present in the area at the time of the survey n2= number of animals (both marked and unmarked) seen during the survey m= number of marked animals seen during the survey More biased

open population

a sampled population this is not closed (has immigration, emigration and births and deaths)

Closed population

a sampled population where births, deaths, emigration, and immigration does not occur during the sampling period - easier to get an accurate estimate

preference

as selection independent of availability

Random sampling method

before going into the field you would select random places along the transect

cafeteria experiemnets

captive animals are presented a variety of foods and allowed to choose among them - may be completely guessing and not offer them anything that they actually will eat

Population trend

change in number of individuals over time (animal abundance over the course of 20 years) • can be misleading to people based on the scales we use in our graphs show people because people don't often look at scales. Thus we should use appropriate scales because it could look bad on the profession as a whole or lead to people not trusting us)

selection

components are exploited disproportionally to their availability (ex in a certain area milo makes up 20% of available food while corn makes up 80%, after a study you find that milo makes up 60% of the food consumed while corn makes up only 40% meaning that your species is selecting against corn)

Relative density

density or animals within 1 place and/or time period, relative to the density of animals in another place and/or time period

Vegetation type

differences in vegetation stands (e.g marsh vs. tall grass prairie)

Food habits

introduction - wildlife managers are interested in food selection • foods used vs food available - food availability and use impacts animal abundance and distribution in space/time - biologists must collect site and time specific information on patterns of food use • what they are eating and what is available - "any fool can do a food habits study" • but is it right? - early studies were based on samples collected at hunter check stations Early studies many biases • seasonality (only animals could be observed during a hunting season-short period of time and eating different things depending on season) • certain age classes targeted (many hunters primarily get adult males) • animals with certain behaviors more likely to be sampled - historic methods limited to visual observation Modern Studies overcome biases • not limited to hunter check stations (attempt to sample broadly, and sample across the seasons) • techniques are not limited to visual observation

Accuracy (accurate)

is a measure of bias error, or how close a statistic (ie a population estimate) taken from a sample is to the population parameter (ie actual abundance)

diet of a predator

need to know the species that are being eaten while also knowing how abundant each species is

Population density

number of individuals per unit area - large range animals will have less density per unit area and vice versa

Absolute abundance

number of individuals that is present in a given area

relative abundance

number of individuals within a population at 1 place and or time period, relative to the number of individuals in a different place and/or time period

point area count vs time-area squirrel survey

point area count - fixed radius time-area squirrel survey - calculate an average sighting distance - time and duration is standardized to 30 minutes (makes points more accurate)

What makes modern distance sampling different from traditional distance sampling?*

specific shape of the detection function is estimated for a particular target species and set of conditions using data from the sample itself

strip count transect count vs point count location count (differences)

strip/transect: - moving through the habitat and causing animals to move and flea (easier and more accurate) Point/location: - arrive at a point and don't move and flush animals, just count animals that are in their natural routine

equidistant sampling method

taking a new sample every certain number of meters (measrements)

Equidistant step-spacing sampling method

taking a new sample every certain number of steps not measuring exact distances

alternating sides sampling method

taking equally distance measurements that are alternating on each side of the transect

Science communication

taking science that is often very difficult and translating it in an easier way to the general public, kids, or other scientists in different fields. - examples of magazines: • national geographic • popular science • parks and wildlife • Audubon - BEATS: "beat reporting, also known as specialized reporting is a genera of journalism focused on a particular issue, organization, or institution over time" - how does this apply to wildlife "science needs to be communicated and needs to be communicated well" - the book; National Geographic Book of Mammals

Easy ways to describe data

test question:* What the method is describing, how to calculate it, or how to calculate the value from the given data set - mean - medium - mode - range

Mean estimates

the average of repeated sample populations estimates usually taken over a short time period

resource availability

the quantity accessible to an animal (a resource can be extremely abundant but might not be available to the animals)

abundance

the quantity for the resource in a study area (a resource can be extremely abundant but might not be available to the animals)

habitat

use of a vegetation type by an animal (e.g. deer habitat)