The Scientific Method, Research Tools, and Techniques. Objectives In this lab you will learn:

• How to utilize the scientific method • How to make observations and collect data • How to test an hypothesis using simple statistics

Welcome to Life Sciences 1! In this class, we will explore various aspects of biology focusing particularly on evolution, ecology, and biodiversity. The lab is designed to illustrate some of the important concepts from lecture. It is also your chance to explore some aspects of biology in further depth through in‐class exercises and observation of living and preserved organisms. Many of you will see organisms you have never seen before. Hopefully, all of you will learn much more about biology. Above all, be enthusiastic and have fun!

Scientific Method Modern science began during the Age of Enlightenment, which swept through Europe in the late 16th and early 17th centuries. Although science itself was performed in all parts of the world and in various manners long before then, the philosophical advancements that were brought about in the Age of Enlightenment codified a logically sound methodology that is the basis for all modern scientific knowledge. The methodology, naturally, is referred to as the scientific method. The scientific method is a process that starts with careful observation of the natural world. From these observations, a hypothesis is formulated to explain the observation. The hypothesis is used to make predictions, and these are tested through experimentation. The experimental results are then interpreted to either support or reject the hypothesis. This process repeats, until the hypothesis is either highly supported or rejected. The scientific method is based on deductive reasoning championed by Francis Bacon (1561‐1626), a philosopher and scientist from the Age of Enlightenment. Deduction is a philosophical process in which you can conclude that something is true because it is a logical extension of other things you know to be true. For instance, because you know that there are two sides to a coin, you can deduce that the odds of attaining a head is 50%. Induction, on the other hand, is a form of reasoning in which multiple, consistent observations yield a conclusion. In our coin example, if you were to flip a coin 10 times and attained 10 heads, you might inductively conclude that a flipped coin always lands on heads. A conclusion based on inductive reasoning, however, is easily negated: a single tail in our example would disprove our conclusion. Often induction is used to develop

hypotheses, because hypothesis development is based on observations. These hypotheses are in turn tested through experimentation and rejected or supported through deductive reasoning.

More specifically, the scientific method employs what is known as a hypothetico‐deductive (or hypothesis‐prediction) approach. In this logical method, a specific hypothesis is formulated to explain a particular phenomenon. The hypothesis is designed to make predictions about the phenomenon you observed. This hypothesis is tested with an experiment designed to permit deduction about that phenomenon. The experimental results will either support or refute your hypothesis, allowing you to draw conclusions about the truth or accuracy of your hypothesis. This process can be ongoing ad infinitum, allowing you to learn a great deal about a single phenomenon. The formulation of a good hypothesis is an important part of the scientific method. A hypothesis must give a testable explanation of the observations you have made of a particular phenomenon. For example, perhaps you have noticed

that your dog is fond of chasing some of your neighborhood cats. You may make a hypothesis to test, for example, whether your dog chases neighborhood cats more than out‐of‐town cats, black cats more than tabby cats, more cats in the morning than in the afternoon, etc. All of these can be tested experimentally and will provide information about your dog and its relationship to cats. However, it is important to realize that there are hypotheses that cannot be tested, that are outside the scope of scientific inquiry. It would be impossible to test, for example, whether your dog chases cats because it hates them or if it chases cats because it likes them. It is much easier to reject a hypothesis than it is to accept one. Recall our discussion on deductive vs. inductive reasoning. A single observation that disproves a hypothesis nullifies the validity of that hypothesis, whereas multiple observations that support it may not be sufficient to demonstrate that it is true. You might hypothesize that all crows are black, because every crow you have ever seen has been all black. But take a journey to Europe and you will observe a crow with a white stripe around its neck and shoulders. Your idea that all crows are black can be nullified with a single observation of black‐and‐white crow. This is why it is common in the scientific method to try to deductively refute competing hypotheses rather than gathering inductive support for your hypothesis. The competing hypothesis states that the factor you are testing experimentally has no effect. For example, if you hypothesize that birds prefer red berries, your competing hypothesis should be that berry color has no effect. This competing hypothesis is called the null hypothesis, and very often experiments are designed to reject the null. By rejecting the null hypothesis, you are in fact supporting your own hypothesis. Say you were to hypothesize that drug abuse during pregnancy causes low fetal birth weight. It would be easier to disprove the null hypothesis: that drug abuse during pregnancy has no effect on fetal birth weight. You could then experimentally sample birth weights of drug abusers and non‐drug abusers. If, based on your experimental data, you were able to reject the null hypothesis that drug abuse has no effect on birth weight, then your alternative hypothesis is supported. You were able to deductively reject the null hypothesis (no effect). An important thing to remember about science is that it is an ongoing enterprise, where hypotheses are constantly tested and retested. The field of scientific knowledge therefore is always changing as new ideas are supported and old ideas are rejected. A hypothesis that withstands a high degree of scrutiny through many experiments can be called a theory –a hypothesis that has withstood a great deal of scrutiny and has great predictive power – but even theories can be overthrown with a single observation. In 1919, for example, Einstein’s theory of relativity disproved the physics of Isaac Newton that had robustly explained so many natural phenomena regarding the movement of heavenly objects for over 200 years. Einstein’s

hypothesis was that space and time were curved, whereas in Newtonian physics (the kind you may have learned in high school), space was flat and time linear. In 1919, an expedition was conducted by the British Royal Society to Brazil and the island of Principe off the western coast of Africa to observe a total solar eclipse. Einstein’s hypothesis predicted that in those two places, astronomers would be able to observe stars that were located directly behind the sun because the light from those stars would curve around the mass of the sun. The Royal Society astronomers undertook the expedition and were able to observe light from the stars behind the sun. That single observation was sufficient to disprove Newton’s theory of physics. Such is the nature of science: nothing can ever be proven. Even the most robust of theories must be revised in light of new evidence. However, this is also the strength of science. It is not dogmatic in its beliefs; it constantly evolves so that our level of knowledge only improves with each experiment that is performed. Next time you are walking to school, sitting outside studying, or walking to class take a look around and try to watch some of the animals you see on campus (trust me there are more than just hungry squirrels). Formulate a hypothesis regarding something you have seen. State your null hypothesis, and try to design a simple experiment to address this hypothesis. While you don’t have to turn this in, you may be asked to present your hypothesis in lab this week, so make sure to take some time to look around and think about what you see.

Observing and Quantifying Animal Behavior

Making an interesting Observation

Most investigations of scientific questions begin with an interesting observation. People have noticed that some species of ground squirrels exhibit a very peculiar behavior when they encounter anything that is saturated in snake scent. For example, California ground squirrels and rock squirrels will chew on shed snake skins and apply it to their fur by licking their bodies. This behavior has been called “snake scent application” (SSA), and is also found in other rodents (e.g., chipmunks and mice). Have a quick look at this behavior now. In this lab exercise you will make observations, record data, quantify observed behavioral patterns, and learn a simple statistical test to evaluate your hypothesis. You will watch videos of a behavior exhibited by two species of ground squirrels to determine if the SSA behavior to various regions of the body is random.

Asking a Question and Developing a Testable Hypothesis Intriguing observations about behaviors often lead to questions about how or why they occur. For instance, we may wonder how rodents apply snake scent – do they apply the scent to the different parts of their bodies in a particular order, or is

the sequence random? Do they apply scent to one region of their body more than others, or do they cover all regions of the body equally? We also may wonder about the function of applying the scent. Rattlesnakes have been a major predator of ground squirrels for millions of years and this coexistence has led to the evolution of several unique anti-predator strategies in ground squirrels. For instance, in some species of ground squirrels adults are resistant to rattlesnake venom and will actively harass and attack these predators.

Group Exercise

A) Rodents are a major prey source for snakes – why do you think rodents apply their predator’s scent? Tentative explanations to your question are called hypotheses. Within your group, brainstorm as many hypotheses as possible to answer the question about function above. List your hypotheses. Are all of them testable? Give at least one example of a non‐testable hypothesis.

B) Choose one testable hypothesis and suggest specific ways to test the hypothesis. What predictions would you make to support or refute the hypothesis?

In this lab we will focus on the SSA behavior. We will address the question: is SSA applied randomly over the squirrels’ bodies? We will be making observations of two different Spermophilus species (S. beecheyi and S. variegatus). Write your hypothesis and null hypothesis below. Why might you expect SSA to be non-random? While you are thinking about this consider some things about the ecology of squirrels: where do they live, where are they likely to be found in their environment (i.e. on the ground, rocks, in trees), could there be a benefit of applying snake scent to some regions versus others on the body? You will be collecting data on more than one species, do you expect different patterns between species? What other things did you consider when trying to address this question?

Testing Your Hypothesis In this section, you will test the hypothesis that SSA behavior is random with respect to areas of the body. That is, whether the squirrels simply apply the snake scent or whether they apply the scent to specific areas of their bodies. You will watch videos of SSA behavior from two species of ground squirrel, California ground squirrels (S. beecheyi) and rock squirrels (S. variegatus). To obtain the videos, a researcher staked out a shed skin of a sympatric rattlesnake species (Crotalus atrox for rock squirrels, Crotalus oreganos for California ground squirrels) and surrounded it with a minimal bait trail (sunflower seeds) to attract the squirrels to the skin. The squirrels were marked with numbers (fur dye) to identify individuals. Each video represents a different individual. The study sites were in Winters, CA (California ground squirrels) and Caballo, NM (rock squirrels) in county or state parks where squirrels were used to cars. The squirrels were videotaped from a car, ~20‐30 m away.

Observations and Data Collection The behavior you will be scoring is what body regions ground squirrels apply snake scent to after chewing shed skins. You will be tallying snake scent application on the following body regions: Head Tail Front Legs Flank Hind legs Figure 1: Ground squirrel body sections http://pad1.whstatic.com/images/thumb/d/d3/Draw-a-Squirrel-Step-19.jpg/550px-Draw-a-Squirrel-Step-19.jpg

You will record the frequency of application behavior for both species. Make observations for four individuals of each species. You will find the video clips on the desktop of the computer (labeled SSA video clips). Choose four from each species (labeled Clip b# for S. beecheyi, and Clip v# for S. variegatus), watch each of these and record your data for each focal animal in the table above (or you can make one of your own). Record the total frequency, from all four individuals for each species, on the table below. At the end, you should have data collected for four different individuals of each species, eight individuals total. Species Ind Flank Head Front

Leg Hind Leg

Tail Row Total

S. beecheyi 1 2 3 4 Total S. variegatus 1 2 3 4 Total

Interpreting Your Results

The data collected from the videos allows us to statistically test whether SSA behavior is random or not. The statistical method employed is the chi-square test. The chi-square (X2) test compares the observed behaviors to what would be expected if the behaviors were randomly distributed. The basic Chi-square formula used to compute the statistic is:

where: o = observed frequency, and e = an expected frequency, asserted by the null hypothesis For example, let us say we were to observe 100 SSA behaviors, divided into 4 groups: the flank, the tail base, the middle tail, and the tail tip. If the behavior were truly random, we would expect those 100 behaviors to be distributed evenly among our 4 body parts, 25 each. This gives us our expected values for our chi‐square calculations. These expected values represent our null hypothesis; in this

case, our null hypothesis is that squirrel snake scent application behavior is randomly distributed over the body (hence 25 expected touches to each of the 4 body parts). If our statistical analysis allows us to reject this hypothesis, we may conclude that SSA behavior is non‐random. (NOTE: Your data is divided into 5 body regions, not 4. The expected number of touches to a particular body region therefore would be your total number of observations divided by 5.) In our 100 SSA observations, let us say we see squirrels lick their flanks 21 times, their front legs 22 times, their hind legs 29 times, and their tail 28 times. We would see that there is some variation between our observed and expected values. Although a chi‐square calculation is simple enough to do as a single‐line equation, some might find it helpful to organize your data into a table like the ones below. (You will need to do these calculations for both species). Flank Front Leg Hind Leg Tail Total Expected 25 25 25 25 100 Observed 21 22 29 28 100 To begin our chi-square calculations, we subtract the observed values from our expected values and then square them: Flank Front Leg Hind Leg Tail Total Expected 25 25 25 25 100 Observed 21 22 29 28 100 o-e -4 -3 4 3 (o-e)2 16 9 16 9 Next, we divide the squares of our expected-observed observations by the expected value: Flank Front Leg Hind Leg Tail Total Expected 25 25 25 25 100 Observed 21 22 29 28 100 o-e -4 -3 4 3 (o-e)2 16 9 16 9 (o-e)2/e 0.64 0.36 0.64 0.36 Finally these values are summed giving a chi-square value of 2. By itself, a Χ2 value of 2 is quite meaningless. We need to determine whether or not this indicates a significant deviation from our expected distribution. Therefore, once our Χ2 value is obtained, we need to compare it to a table of critical values (like that shown below) which will allow us to determine whether our observed data deviates significantly from our expectations due to random chance. The first step in this process is determining the degrees of freedom we have in our sample data. Degrees of freedom refer to how many independent pieces of data are

assembled in our final data set. In our case, there are 4 body parts in our data set. Only three of these are independent, because if we were given three pieces of the data table, we would be able to determine what the fourth was. As such, the fourth value is not independent and cannot be included in our degrees of freedom, leaving us with a df of 3. Generally speaking for simple data sets, the degree of freedom is one less than the number of observed classes of data. Table of Chi-squared Critical Values: Degrees of Freedom 0.10 0.05 0.025 0.01 0.001

1 2.706 3.841 5.024 6.635 10.828 2 4.605 5.991 7.378 9.210 13.816 3 6.251 7.815 9.348 11.345 16.266 4 7.779 9.488 11.143 13.277 18.467 6 9.236 11.070 12.833 15.086 20.515

In our table of critical values, we see that the critical Χ2 value to reject our null hypothesis is 7.815 at a confidence of 0.05 (an arbitrary standard cutoff, although others are sometimes used) with 3 degrees of freedom. This means that our Χ2

score would have to exceed 7.815 in order for us to reject our null hypothesis (that SSA body application is random). In our case, a chi‐square value of 2 does not allow us to reject this hypothesis. Our data therefore cannot be said to be significantly different from random. Discussion Given the data example worked in the lab manual, do you accept or reject your hypothesis for each of the species observed? Based on your raw data do you see any differences between species? What might explain these differences, if any, that you observed? How might you test hypotheses about different species?