A Manual for Undergraduate Research in Field Biology
Part II. Design— how to collect your data

A Manual for Undergraduate Research in Field Biology
Part II. Design— how to collect your data

Introduction

Part II of this research manual is designed to help students of biology, ecology, and environmental science design a field study to answer a scientific question or to test a specific hypothesis or set of hypotheses. Because it's web-based, it can be updated, changed, modified and improved over time. I've tried to include a variety of sampling strategies that can be used to address different types of organisms, habitats and ecosystems. That said, I'd love to add more examples—send any suggestions to:

singler -at- franklinpierce -dot- edu
(funky format to prevent emails from robots)


The Variability of Field Sites—

Though I truly hope and expect the suggestions here will be valuable as you plan your fieldwork, it's simply not possible to use a "cookbook" approach when designing a field study. Every situation in the natural world is different. For example, no two field sites are shaped exactly the same—they can be roughly square, circular, rectangular, or wrap around the edge of a pond—and those are just some of the simplest examples! This guide should give you some basic tools, approaches and general guidelines to get you up and running. In the end, there's no substitute for thinking through and understanding the question or hypothesis motivating your study, the issues you face at your field sites, and most importantly, getting advice from your professor or advisor. If you design your field study well, your project can be fun, rewarding and relatively stress-free. Good luck with your pursuits!



Your Question or Hypothesis

So you want to design a field study . . . or maybe it’s a requirement for a class or to graduate . . .

You'll only be able to design a successful field study if you have a clearly defined scientific motivation, either a question or a hypothesis about the natural world. Of course, you may also have a few related questions or hypotheses, but, in every case they must include at least one dependent and independent variable and be specific and focused enough to guide your study. In fact, it's simply not possible to design a good study without a clear question or hypothesis. If you don't yet have one, see Part I of this guide.



Four Steps for Designing a Field Study

Once you're ready with your question or hypothesis and you've clearly identified your dependent and independent variables, you can use these four basic steps to design your study:

  1. Determine the size and number of field sites you will use.

  2. Identify the approach you will use for sampling your sites.

  3. Figure out exactly what data you will collect and observations you will make.

  4. Make sure your design clearly addresses your question or hypothesis and printout your data sheets.

Each of these steps can require a fair amount of work, but if you plan carefully, you’re likely to get meaningful information that will help answer your question or test your hypothesis.



Step 1: Determine the size and number of field sites you will use.

What is it about the natural world that you're studying? For example, if your focus is soil chemistry or leaf litter invertebrates (such as spiders), your field sites can be much smaller than if you're studying larger or more mobile organisms or ecosystem processes.

Size of Field Sites—

The size of a "field site" in a field study can vary dramatically. On one end of the spectrum, a field site could be as small as 15 m x 15 m (or perhaps even smaller); on the other extreme, a field site could be as large as many hectares (1 hectare, abbreviated ha, is 100 m x 100 m). Also, in many cases you may not know the precise dimensions of your field site, but you definitely need to be able to visually recognize the boundaries and know the approximate dimensions of your sites.

Some examples of field sites can be seen here.

So how do you know the appropriate size of a field site for your study? The guidelines below should help get you started, but nothing can replace getting feedback from your professor or advisor.

Here are some general examples of study topics and very rough suggestions for sizes of field sites that may be appropriate; some of these examples might apply directly to your study.

  • Soil chemistry, soil or leaf litter invertebrates, insects: perhaps as small as 15 m x 15 m or as large as one hectare.
  • Relatively small organisms like rodents or herbaceous plants: at least 30 m x 30 m; more likely one to several hectares.
  • Large or mobile organisms like trees or birds: two to several hectares.
  • Highly mobile organisms like deer, moose, bear: ten or more hectares.

*IMPORTANT NOTE: The list above refers to the suggested size for field sites, the area within which you'll take your samples. The size for sampling plots (if you choose to use them) will be much smaller and is addressed below.

Number of Field Sites—

The number of field sites included in studies also varies dramatically, especially in undergraduate research. In almost all cases, studies that are published in the research literature are based on more than one field site. For your research, it might be acceptable to focus on only one site, but you should consult with your professor or advisor early in the process of designing your study if you're considering including only one site.

If you're planning to compare two habitat or ecosystem types (as specified in your question or hypothesis), it's best to include a minimum of two field sites to represent each type (more would certainly be better!).

For those of you with statistical training, you should realize that comparing one site representing Habitat A with one site representing Habitat B is essentially including a sample size of 1 for each study group. Ideally (from a statistical perspective), you would include a bare minimum of ten for each group, but this often isn't practical for short-term undergraduate field studies. As with most of the recommendations in this guideline, you should consult with your professor and match your study to course or project expectations.

(back to list of four steps)



Step 2. Identify the approach you will use for sampling your sites.

In almost all cases, it will be impossible to measure every individual organism or sample an abiotic factor at every single location within your field site, so, you will need to take many subsamples within your site. How do you design your sampling strategy? What approach will you use to organize your samples? Where exactly within your site will you take them? How many samples will you take?

This section discusses three topics to help you determine how best to sample your site: basic tools for sampling; determining the exact location for sampling; figuring out the number of samples to take.

Two additional sections, one on examples of how to determine plot placement and the other with diagrams showing examples of field sampling designs should also help you come up with a plan that's most appropriate for your study.

A. Tools

Transects—

  • Transects are lines through your field site (often marked with meter tapes) that help you organize your samples. Examples can be seen here.
  • The length and number of transects you use will depend on the details of your study and the particular features of your site (size, shape, etc.). For example, it might be best to use two long transects to sample a narrow band of habitat along the edge of a wetland; several shorter transects may work better to sample a meadow that's roughly circular in shape.
  • It's always best to use a minimum of two transects per site so that you have some replication built into your field methods—this will help with statistical analysis.

Sampling Plots—

  • Sampling plots provide a designated area for taking measurements. Examples can be seen here.
  • The appropriate size of plots will depend on what you're studying. They can be as small as 10cm x 10cm (or even smaller), or as large as 20m x 20m (or even larger).
  • It's always best to use multiple sampling plots per field site (usually a bare minimum of ten).

Plotless Sampling—

  • For some field studies, plotless sampling is preferable to or more efficient than laying out sampling plots.

  • The Point Method: Under this method, sampling points can be located randomly throughout a field site (see section below on random sampling), or they can be located at regular or random intervals along transects. This method works well for abiotic factors that can be measured at specific points (such as soil moisture, soil pH, forest canopy openness, etc.) or for biotic samples of small organisms (such as soil microinvertebrates or bacteria).

  • The Transect-Intercept Method: Under this method, all environmental features of interest crossed by the transect are recorded. This method can work well for sampling environmental features such as coarse woody debris (branches and logs on the forest floor), potential cover type for salamanders (such as rocks and logs), or forest microtopography (treefall pits and mounds). For more, see figure and figure legend here.

  • The Point-Quarter Method: Though this method was originally developed for sampling forest trees, under certain circumstances it works well for sampling evidence left by animals, such as rodent burrows or animal sign. For full details, see figures and figure legends here.

    A simpler version of the point-quarter method, sometimes called the nearest-feature method, can be used to sample the nearest feature of interest (such as a certain size or species of tree, type of mushroom, animal sign, etc.) to each of your sampling points. Details can be found here.



B. Location, Location, Location!

Unbiased Representative Sample—

  • As previously mentioned, it's almost always impossible to take measurements on every organism you're interested in (like the tail length of all Red-backed Salamanders on the Franklin Pierce Campus). As a result, you'll need to take a subsample of all individuals. In order to get meaningful information on the population (or community, ecosystem, etc.) as a whole, your subsample should be an unbiased representative sample of the entire population (or community, etc.) you are studying.
  • It's NOT OK to sample the easiest organisms to measure, the most interesting section of your site, the easiest area to walk, etc. – this will give a biased representation of what you're trying to measure.
  • There are several approaches for getting an unbiased representative sample, including random sampling, uniform sampling, and haphazard sampling. Though the least prone to bias is probably random sampling, the other methods do have some advantages and may be appropriate.
  • Another issue to keep in mind is repeat sampling—measuring the same individuals or sampling areas more than once. Repeat sampling can cause bias because the individual that was repeatedly measured becomes over-represented in your sample. Unless you're purposely re-sampling individuals as part of your study (such as measuring growth or movement over time), it's best to avoid repeat sampling. In some cases, individuals or locations can be temporarily marked (such as marking tree bark with chalk) in order to prevent repeat sampling.
  • Regardless of what method you choose to achieve your unbiased representative sample, make sure the sample you take is guided by your question or hypothesis and awareness of your dependent and independent variables! For example, if you're interested in the effects of canopy openness on the abundance of shrubs in the understory of a forest, it's important that your sample includes plots or points that vary in canopy openness.

Random Sampling—

In random sampling, you use a random number table (or some other way of generating random numbers) to select randomly chosen points within your field site. These points might mark the beginning of transects, locations for placing sampling plots, locations for using plotless sampling techniques, etc.

There are two main ways to conduct random sampling. The first is to convert your field site into a coordinate system. For example, if your site is 100 m x 100 m, you can think of one boundary as the x-axis and the other as the y-axis. You can then choose pairs of random numbers between 0 and 100 to select random points within your coordinate system. See example diagram here.

An alternative that can be slightly easier to implement in the field (less measuring with meter tapes required) is to measure random distances from points along randomly generated compass bearings. See example diagram here.

Here you will find random number tables that will help with these techniques.

Uniform Sampling—

In some cases, it might be appropriate to sample your site using some type of uniform method, like placing transects at evenly spaced distances within your site and locating sampling plots at evenly spaced intervals. The advantage to this method is that it can be less time consuming than locating random points. Here's the possible drawback: You need to be sure that your sampling interval (like placing a sampling plot every 5 m) doesn't coincide with a natural pattern within your field site. For example, if you sample every 5 m and your field site happens to be particularly wet every 5 m due to unique microtopography at your site, you will get a biased sample. Every situation is different, so my best advice is scope out your field site, think through this potential drawback and whether it might apply to your study, and (once again), consult with your professor!

Haphazard Sampling—

Haphazard sampling is a variation on random sampling—by using this approach you're trying to get an unbiased representative sample by choosing points "blindly" without actually using random number tables and locating truly random points. For example, you could toss a sampling flag behind you over your shoulder (without looking) to find a point where you sample. Next time, you choose a different direction and toss a different distance. This approach will only work if you don't pay close attention to where you're tossing the flag—if you pay attention to where you're throwing, you may be biasing your sample. The advantage is that it can be less time consuming than a truly random approach. The disadvantage—it may be more prone to bias. I recommend clearing this approach with your professor before proceeding.

What it all means for plot placement—

So, once you determine whether you're going to use random, uniform, or haphazard sampling, how do you then figure out exactly where to put your plots? Well, the details vary according to your study, of course, but see these Sampling Design Diagrams for some examples.



C. Number of Samples

The number of samples that you collect at each field site will depend on many factors, including how long it takes to collect a sample, what your course or project requirements are, and most importantly, your motivating question or hypothesis. Make sure you're getting enough samples for various values of your dependent and independent variables—if unsure, see your professor or advisor.

If you're comparing two or more groups (such as different habitat types) represented by different field sites, you should collect a bare minimum of ten samples per field site—twenty (or perhaps much more) would be better.

From the perspective of statistical analysis of field data, more samples are almost always better! There's a point beyond which that's no longer true statistically, but that point is seldom reached in undergraduate field studies. Again, consult with course or project guidelines and your professor.

An additional topic to consider regarding number of samples is . . . sampling effort . . .

Sampling Effort—

Sampling effort refers to how much area you include or how much time or how many transects, etc. you use to sample your field site. If you're comparing different field sites, you should use equal sampling effort at each site.



D. So where exactly do you place your plots (or sample points)?

Once you've decided what size your field sites should be, whether to use transects, whether to use plots or plotless sampling (or a combination of the two), whether to use random sampling or another method, and how many plots or sampling points you need, how do you put it all together and decide exactly where to take your samples in your sites? There's no one quick answer to this question, but once again it's worth stressing that you keep your question or hypothesis in mind and that you make sure you get enough samples for various values of your dependent and independent variables (if unsure, see your professor or advisor).

The written examples below should help give you a sense for how you can come up with a specific plan. The bullet lists under each example provide one possible way to include the desired number of samples and sampling approach; there are other possible designs that will meet the same goals.


Ex 1: Field sites roughly 100 m x 100 m (1 ha); forty plots per site; uniform sampling:

  • four 50 meter transects
  • transects located 20 m apart with at least 20 m between each transect and the edge of the field site
  • beginning of first transect is located 20 m in and 20 m up from corner of field site
  • ten plots per transect at 5 m intervals


Ex 2: Field sites roughly 50 m x 50 m; twenty plots per site; random sampling:

  • four 20 m transects
  • transects starting at random points within field sites chosen using either a coordinate system or the compass bearing technique
  • beginning of first transect is located at a random distance from corner of field site using either random coordinates or compass bearing technique
  • five plots per transect at 4 m intervals


Ex 3: Field sites several hectares (ha) in size; point-method with as many points along transects as possible:

  • figure out the maximum possible length for a transect (this may be 100 m as that's typically the longest meter tape available)
  • transects placed within sites to achieve an unbiased representative sample (random, uniform, or haphazard technique used for locating transects)
  • samples taken at random or uniform intervals along as many transects as time permits


Ex 4: Field sites several hectares (ha) in size; twenty points per site using the point-quarter method:

  • samples taken along two transects, ten points per transect
  • the interval between points must be great enough to avoid repeat sampling (see Figure 2.6b)
  • once interval is known, transects should be ten times that length
  • field sites could be divided in half with each transect running through the middle of each half




E. Examples of Sampling Designs

Here you will find a number of diagrams that show a variety of possible sampling designs. Feel free to base your design on one of them, but keep in mind you will probably need to sample at more than one field site, and, your design needs to address your question or hypothesis.

(back to list of four steps)




Step 3. Figure out exactly what data you will collect and observations you will make.

What exact data should you collect and observations should you make within each of your sampling plots or at each of your sampling points? Clearly the details will vary and be topic specific. However, in all cases be sure you keep your motivating question or hypothesis in mind! Make sure that you'll be measuring your dependent variable (or in some cases variables) and your independent variable(s). It may also make sense to record additional information that's not addressed specifically in your question or hypothesis. For example, if you hypothesize that salamander abundance will be greater in forests with greater soil moisture, clearly you need to measure the number of salamanders and soil moisture in all of your samples. In addition, you might want to measure other features of the environment (though not too many) that could influence salamander abundance, such as number of cover types available, or forest type. Often in science the original hypothesis is not supported, but additional information can lead to new questions or hypotheses that can be tested and ultimately increase our understanding of the natural world.

I recommend coming up with a detailed list of variables to measure and observations to make while using your question or hypothesis as a guide. Also, don’t try to do too much or you'll get overwhelmed and run out of time—this is a common mistake with beginning projects. Once you have your list, check it with your professor or advisor.

(back to list of four steps)



Step 4. Make sure your design addresses your question or hypothesis and printout your data sheets.

So you’ve figured out the size and number of field sites you should use, whether to use plots or sampling points, where you’re going to locate them, and exactly what data you’ll collect in each plot or at each point. Great—you’re design is nearly done! Now it’s time to ask yourself the following key questions:

  • Is your sampling design clearly guided by your question or hypothesis with appropriate emphasis on answering your question or testing your hypothesis and measuring your dependent and independent variable(s)?
  • Are your field methods designed to get an unbiased representative sample of what you're studying?
  • Have you included enough field sites and samples within each site to do statistical analysis (if required) or to meet the requirements of your project?

If the answer is yes to each of these questions, then you have just one last step before heading out to the field.

Field Data Sheets—

It will be extremely helpful if you make a detailed field data sheet in MS Excel (or comparable spreadsheet software) and take to the field with you the necessary number of printed copies (plus a few extra). A good field data sheet is almost like a set of directions for field sampling and can help make your fieldwork as stress-free and trouble-free as possible!

Below is a link to examples of field data sheets. You may be able to modify one to fit your purposes, or in some cases, one "as is" may work for you!

Link to Examples of Field Data Sheets

Once you print out your data sheets (and make sure you have the necessary field equipment, of course), you're ready to head to the field!


A final thought —
A successful study is not defined by whether you answer "yes" to your question or support your hypothesis; a successful study happens when meaningful data are collected and interpreted properly so that you can clearly address your question or hypothesis. If you support your hypothesis, that's fine, but it's not as though you've figured everything out — there's always more to learn! If you reject your hypothesis, that's ok too — you've learned something about the natural world and it's time to look for other explanations for what interests you. And, if your results are inconclusive, don't fret — it happens in science all the time. It's not easy to get clear answers, and, many scientists work for years before their studies produce clear results. If you design and implement your study well and reach appropriate conclusions, you've succeeded, and, you will most certainly learn more about the world around you!


Now, get off of the computer and head out into the field! Good luck!


(back to list of four steps)








This site created & maintained by
Rhine Singleton
Professor of Biology & Envi. Science
Franklin Pierce University
singler -at- franklinpierce -dot- edu


More about this site
©2008-2019, ecologyandevolution.org