Open Textbooks

2018

Quantitative Problem Solving in Natural Resources

Peter L. Moore
Iowa State University, [email protected]

Follow this and additional works at: https://lib.dr.iastate.edu/opentextbooks

Part of the Environmental Sciences Commons, and the Mathematics Commons

Recommended Citation
Moore, Peter L., "Quantitative Problem Solving in Natural Resources" (2018). Open Textbooks. 1.
https://lib.dr.iastate.edu/opentextbooks/1

This Book is brought to you for free and open access by Iowa State University Digital Repository. It has been accepted for inclusion in Open Textbooks
by an authorized administrator of Iowa State University Digital Repository. For more information, please contact [email protected].
P. L. Moore

Quantitative
Problem Solving in
Natural Resources
O C TO B E R 5, 2018

Iowa State University Digital Press

A G R E AT D I S C O V E R Y S O LV E S A G R E AT P R O B L E M B U T T H E R E I S A G R A I N O F

D I S C O V E R Y I N T H E S O L U T I O N O F A N Y P R O B L E M . Y O U R P R O B L E M M AY B E

M O D E S T ; B U T I F I T C H A L L E N G E S Y O U R C U R I O S I T Y A N D B R I N G S I N T O P L AY

Y O U R I N V E N T I V E F A C U LT I E S , A N D I F Y O U S O LV E I T B Y Y O U R O W N M E A N S ,

Y O U M AY E X P E R I E N C E T H E T E N S I O N A N D E N J O Y T H E T R I U M P H O F D I S C O V -

E R Y. S U C H E X P E R I E N C E S AT A S U S C E P T I B L E A G E M AY C R E AT E A TA S T E

F O R M E N TA L W O R K A N D L E A V E T H E I R I M P R I N T O N M I N D A N D C H A R A C T E R

FOR A LIFETIME.

G E O R G E P Ó LYA , H O W T O S O LV E I T
Copyright © 2018 P. L. Moore

This work is licensed under a Creative

Commons “Attribution-ShareAlike 4.0
International” license.

published by iowa state university digital press

typeset with the tufte-latex class: tufte-latex.github.io/tufte-latex/

Contents

1 Introduction 11

Part I PROBLEM SOLVING

2 Problem Solving as a Process 19

3 Some teaser problems 29

Part II NUMERICAL REASONING

4 Quantities in the Real World 37

5 Working with Numbers 47

6 Reasoning with Data 59

7 Interlude: Collecting and managing data 73

Part III SPATIAL REASONING

8 Geometry and Geography 79

9 Triangles 93

Part IV ALGEBRAIC REASONING

10 Generalizing Relationships 107

11 Relationships Between Variables 117

Part V MODELING

12 Modeling 125

13 Models of growth and decay 133

Index 141
Preface

This book is a collection of resources for students of natural

resource science and management1 . Many of you will be required 1
I interpret this broadly, so that we
to complete some mathematics and/or statistics courses during may include study of fisheries, wildlife,
forestry, and water resources, among
your undergraduate studies, and some of you will likely dread the other things
prospect.
However, if you’re reading this, you have probably been presented
with an alternative means of satisfying at least some of your quan-
titative skills requirements. This is the origin of the course NREM
240 at Iowa State University, for which this book was originally pre-
pared. NREM 240 is a class about solving quantitative problems that
non-mathematicians interested in biology and environmental sciences
may find compelling.
I use the word problem here in the same way that math education
scholar Alan Schoenfeld does — to describe an intellectual challenge
that is quantitative in nature2 . A problem is distinct from an exer- 2
Schoenfeld, A. H., 1985, Mathematical
cise in subtle, but crucial ways. An exercise is a prompt for which Problem Solving, Academic Press.

a student must select one or more of a small number of recently-

demonstrated procedures or algorithms to reveal a clear and known3 3
at least to the instructor and the owner
result. Contrast that with the intellectual impasse of Schoenfeld. In of a solutions manual

this spirit, a problem is a deeper challenge that often contains a mix-

ture of complexity, uncertainty, and ambiguity and requires some
technical skill or knowledge. A problem also often requires creativ-
ity — or as Pólya says in the quote4 , “inventive faculties” – and the 4
Excerpted from the preface of Pólya,
willingness to explore, to try and fail, to persist, and to learn from G., 1945, How to Solve It, Princeton
University Press.
mistakes. This kind of task can be frustrating and uncomfortable for
those of us not accustomed to it, particularly if we have no interest
or investment in the problem being posed. But life rarely provides us
with exercises. If we are confronted with exercises, we become bored
and uninspired by the idle redundancy5 . But a completely differ- 5
Mathematician Paul Lockhart has
ent sense of achievement and satisfaction is realized when we solve written an engaging, though scathing
critique of current school math curricu-
problems because, by their very nature, they lead us to new ways of lum in a book called A Mathematician’s
thinking and understanding, if we are willing. Lament.
NREM 240 at ISU was initially conceived as a bridge to link the
8 iowa state university

quantitative skills developed elsewhere with some common ap-

plications in the natural sciences. Experience has shown that, once
students know which techniques to apply to which given quantities,
the computational task is rarely challenging. A greater challenge is
the selection of appropriate techniques and assembly of relevant in-
put quantities when neither are given. Since both of these processes
are hallmarks of authentic problem solving, NREM 240 evolved to
embrace the problem-solving process as the central objective, calling
upon quantitative concepts and techniques as needed to address par-
ticular problems. Therefore, applied problems and the strategies used
to address them have become the focus of the course, and this text
has evolved to support that focus.
The philosophy of this text has been strongly influenced by John
Harte’s fantastic books, Consider a Spherical Cow and Consider a Cylin-
6
Harte, J., 1988, Consider a Spherical drical Cow6 . Both books pose environmentally-themed problems
Cow: A Course in Environmental Problem that are compelling and maddeningly open-ended. But Harte boldly
Solving, Sausalito, CA, University
Science Books; Harte, J., 2001, Consider demonstrates how idealizations, approximations, and analogies can
a Cylindrical Cow: More Adventures in be leveraged to make sense of complex problems. On a tip from a
Environmental Problem Solving, Sausalito,
CA, University Science Books. friend in grad school, I picked up a copy of each of these books and
was amazed. I had always been an average or slightly-below-average
student of math all the way through grade school and college, and
had never enjoyed it. Harte’s problems hooked me because they
offered a means to address problems I found compelling. I saw, per-
haps for the first time, a way that math could help me gain insights
into things I already wanted more insights into. I experienced an
unfamiliar willingness to labor over unit conversion details, to really
wrestle with what it meant to integrate a function, and to chase wild
and risky ideas to see where they led. That feeling has stuck with me
over the years, and I have sought to facilitate some semblance of that
experience in the students I now teach.
The methods or strategies that are highlighted here are drawn
in large part from ’s problem-solving framework and derivatives
thereof. Excellent tutorials on problem-solving include Thinking
7
Mason, J., L. Burton, and K. Stacey, Mathematically by Mason, Burton and Stacey7 , Crossing the River
2010, Thinking Mathematically, 2nd ed., with Dogs by Johnson, Herr, and Kysh8 , and Ants, Bikes, and Clocks
Pearson Education Ltd.
8
Johnson, K., T. Herr, and J. Kysh, 2012, by Briggs9 . Notwithstanding the odd titles of some of those books,
Crossing the River with Dogs: Problem all provide interesting perspectives and tips for problem-solving.
Solving for College Students, 2nd ed.,
Wiley.
Even so, many of the problems in these texts are of the sort that you
9
Briggs, W.L., 2005, Ants, Bikes and might find on standardized tests: “Janet leaves on a train heading
Clocks: Problem Solving for Undergradu- east at 42 miles per hour, while Mark...”. If there is truth to the idea
ates, Society for Industrial and Applied
Mathematics.
that we embrace challenges when they address topics that we are
already interested in, these approaches may still fail to engage stu-
dents. With this in mind, the examples and exercises in this text are
aimed at engaging the natural resource student in thinking about for-
 quantitative problem solving in natural resources 9

est measurements, fisheries management, habitat conservation, and

the like.
You will find in the pages that follow an introduction to problem
solving as a process. You’ll also find a review of some frequently-
used mathmatical concepts and procedures. For some of the more
powerful concepts and techniques, exercises are provided to assist
with reviewing (or exploring for the first time) by doing. The booklet
is not intended to be a guide to be followed through a series of skills,
but rather a resource to support the problem-solving process and
help lower the conceptual and computational barriers along the way.
Schoenfeld argues that what constitutes a problem to you depends
strongly upon your experience and formal training. This relativ-
ity makes it challenging to keep a diverse group of students on the
same page, and able to succeed at similar rates. With this in mind,
in preparing the materials for this book and the course it supports,
I have presumed 1) that you are sincerely interested in the natural
sciences; and 2) that you have had a typical sequence of high school
math courses, including a few years of algebra, some geometry, and
perhaps trigonometry and statistics. It may also be true that you have
found it challenging or even unpleasant when asked to recruit your
quantitative skills to interpret information or address a question in
the coursework in your discipline. If any of this describes you, you’re
in the right place. Let’s go!

Note to instructors

I have written this book with my students and their needs in mind,
but hope that others may find it useful. It may be worth taking a
moment to clarify how I use this book. While I expect my students
to read this text and work on the exercises in it, I build the course
around a group of “focus problems” not contained here, but with an
open-ended form and practical flavor like the sample problems in
Chapter 3. As we work on the first focus problem, I guide students
through the problem-solving process described in Chapter 2, making
frequent reference to examples contained in the book. Therefore, to
appreciate the processes involved, I think it is wise to have students
read the first three chapters early. The remainder of the book is or-
ganized more by problem type, and can be assigned or referred to as
appropriate to support the focus problems, rather than proceeding
linearly through each chapter.
I have deliberately avoided discussing particular software tools or
internet resources, partly in the interest of ensuring that this text does
not rapidly become obsolete, but also to allow for flexibility. In online
video tutorials and in-class exercises, I ask my students to work with
10 iowa state university

data and create graphs using spreadsheets, but those seeking richer
computing environments can just as easily use this text alongside
instruction in R, Matlab, Mathematica, and others.
1
Introduction

1.1 Some Philosophical Notes

In a conventional math course, you might be confronted with a ques-

tion like the following:

Find the roots of x in the expression:

4x2 − 13x + 6 = 0.

You may be instructed or implicitly expected to apply an algorithm

to this problem and provide the two possible roots. Most likely this
would be an opportunity to use the trusty quadratic formula:
√
−b ± b2 − 4ac
x=
2a
If you weren’t already turned off, you might plug the values 4, -
13, and 6 in for a, b, and c in the quadratic formula, perform some
arithmetic and find the roots to be 0.56 and 2.69. Alternatively, if
you’re like me, you’d let a computer program like Geogebra1 apply 1
Geogebra is a free, multi-platform
this algorithm, since that is what computers are for. software package that combines a
CAS (computer algebra system) with
In any case, with these two roots in hand, you have an answer, a dynamic, interactive geometry and
isn’t that satisfying!? OK, maybe it is for some of you, but this has graphing interface.
never been satisfying for me. No meaning was ever assigned to the
variables or constants, nor was it claimed that the result had any con-
text or significance. I don’t really know what to do with the answer,
now that I have it. I’ve learned very little, except perhaps that I can
enter the proper numbers into an algorithm. That is not to say that
learning the algorithm is without value – indeed it is very valuable.
But for most of us, the algorithm itself is not an end in itself, it is
a means to an end. It is a useful tool that allows us a shortcut to a
result when an equation presents itself in a quadratic form.
Though the problems in a conventional math class may look ar-
bitrary, they are often designed to be “well-behaved”. You wouldn’t
12 iowa state university

often see equations exactly like the example above, because the roots
turn out to be icky decimal numbers rather than nice, clean integers.
Furthermore, things get complex (literally!) if the numerical coeffi-
cients on the left-hand side of the equation are such that the term un-
der the square-root in the quadratic formula turn out to be negative.
Such a poorly-behaved case belongs to a completely different subject
in the mathematics curriculum (complex analysis), and so cannot
be imposed upon an unsuspecting algebra student. However, in the
“real world”, there is no more reason to suspect a real result than a
complex one, in those rare practical instances when one needs to find
the roots of a second-order polynomial. Thus, in this approach we
learn a very strict set of rules applicable only to an idealized subset
of problems that may or may not have any significance outside of
abstract trivia.
The approach we use in this course is to encounter math and
statistics in the process of finding solutions to real—or at least plausible—
problems in the natural sciences. Sometimes these real problems are
messier than those out of a textbook. Often they will be open-ended
and will require multiple steps and a variety of techniques. We’ll
need to decide for ourselves what tools and techniques to use, ac-
cording to the needs of the problem. Being creative in math classes
isn’t what we’ve been trained to do, and at times it may be uncom-
fortable. That’s OK, we’ll take our time. But no matter the problem,
we’ll always have a reason to be doing math or statistics – every quan-
tity in an equation will have real meaning or role, and we can apply
our non-quantitative knowledge and experience with these entities to
help us solve our problems.

Let’s have a look at the kind of problem that a natural resource

manager or ecologist might care more about, a problem that we’ll
return to periodically in this course. Consider the Iowa DNR’s es-
timates of the statewide pheasant population shown in the graph
below. Now consider the question that is important to many hunters
and game managers around the state: what should we expect pheas-
ant populations to look like in 5 years? The black dots in the graph
above are annual results from roadside pheasants surveys, the dots
2
note that connecting dots like this can are connected in chronological order with black lines2 , and a blue
imply continuity between data points, line traces the long-term trend. In the roadside survey, DNR biolo-
which may or may not be what we wish
to indicate. gists travel 30-mile segments of rural roads on dewy late-summer
mornings, count the pheasants observed in each stretch, and compile
3
for more information on the pheasant the data across the whole state3 . The changes from year to year in
population trends and the roadside this measure of pheasant population are similar to the changes in
survey method, consult the IDNR small
game website hunter harvest and are thought to be a good indication of the pheas-
ant population as a whole.
 quantitative problem solving in natural resources 13

Figure 1.1: Record of pheasant counts

per 30 miles from the Iowa August
Roadside Pheasant Count. Data from
the Iowa DNR.

The rebound in the roadside counts from 2013 to 2014 was char-
acterized in the DNR report as a 151% increase, and the change from
2014 to 2015 was described as a 37% rise. The last few numerical
value pairs plotted in the graph are also shown in Table 1.1 to the
right. year count
This example may seem straight forward at first glance, and in 2012 7.8
2013 6.5
some ways it is. The trend appears to indicate an overall decline in 2014 16.3
pheasant numbers across the state, and perhaps we should be pre- 2015 23.2
pared to take a more hands-on approach to managing pheasant num- 2016 20.4
2017 14.4
bers if we wish to sustain a viable game resource in the coming years. 2018 20.6
On the other hand, the year-to-year changes seem to be erratic, ris-
ing and falling in a way that seems to lack a pattern. Addressing the Table 1.1: The most recent five years of
data from the roadside pheasant count.
guiding question with any confidence, however, could be a bit chal-
lenging. If, for example, we were looking at the dataset at the end
of 2011 following 6 straight years of steady decline, would we have
been able to anticipate a rebound in 2014 or 2015? Probably not with-
out a robust and reliable model4 of the factors that cause population 4
a model in this context means an ap-
proximate mathematical representation
changes and how those factors could change in subsequent years.
of the real system from which predic-
These are advanced topics, but ones that wildlife managers have to tions about the behavior of the real
incorporate into their management strategies in some situations. system may be made and tested.

On an even deeper level, the roadside pheasant count itself is

a strange quantity that doesn’t exactly represent what we wish to
know (i.e., the pheasant population). Instead, it is an easy-to-estimate
approximation of the real population. As such, it is a sample from
14 iowa state university

the larger population, at least one step removed from the quantity
we seek. How does such a sample relate to the larger quantity we are
after? That’s a pretty simple question in theory (i.e., in stats class),
but when we account for the sampling methodology, timing, and
observer variability, and we consider that pheasant visibility may not
always be directly linked to population, it isn’t so straight-forward
after all.
The observations we’ve made from this dataset are just some of
the many complexities that we might uncover as we endeavor to
solve problems in pheasant population or habitat management in
Iowa. This example hints at the concepts of time series analysis,
forecasting, and measurement uncertainty, as well as functional rela-
tionships between multiple variables and between samples and pop-
ulations. Each of these concepts represents a quantitative tool that
can be applied toward the larger problem-solving task. We’ll visit
most of these concepts and many more en route to addressing prac-
tical problems and methods. But as we will see in the next chapter,
the quantitative procedures that we employ in the problem-solving
process are just a part of the arsenal necessary to solve practical prob-
lems. Furthermore, it should go without saying that some of the
quantitative tools we do have at our disposal are not appropriate for
some problems, and a key job of the problem-solver is to determine
which tools those are. We’ll delve deeper into this issue a few pages
ahead.

1.2 An ancient puzzler

“In a lake the bud of a water-lily was observed, one span above the
water, and when moved by the gentle breeze, it sunk in the water at
two cubits’ distance. Required the depth of the water.”
-Henry Wadsworth Longfellow

The lines above are from a poem in which a Mr. Churchill play-
fully challenges his wife with mathematical puzzlers. The problem
he describes actually dates back many centuries, to a 12th-century In-
dian mathematician named Bhascaracharya who posed the problem
5
One translation that can be readily in verse in his book Lilavati5 . As we look forward to learning the pro-
found online is an 1817 translation by cess of problem-solving, let’s imagine a dialogue between a student
H.T. Colebrooke: Algebra, with Arith-
metic and Mensuration, Brahmegupta (S) confronted with this problem and a patient instructor (I). This
and Bhascara, London, John Murray. isn’t quite the sort of problem we’re interested in really diving into,
but the dialogue serves to illustrate a few points that we’ll address in
the next chapter.
 quantitative problem solving in natural resources 15

1.2.1 A problem-solving dialogue

S: Hmm, so lemme see if I understand this. Wait, is a span an actual

distance? And a cubit?
I: Yes, they are old-fashioned and inexact measures of length. The
cubit at least is about the length of a forearm, and people often take
that to be about 10.5 inches. A span is from the tip of the thumb to
the tip of your pinky if you stretch your hands out as far as possible,
and that’s usually interpreted as 9 inches.
S: OK, so a span is 9 inches, so the water lily was 9 inches above the
water at first. Then when the wind blows, the bud moves two cubits,
so about 21 inches to one side until it is submerged. So we want to
know how deep the water is. Umm...
I: Good start.
S: Well how am I supposed to... Hmmm... Alright, so the would the
depth just be 30 inches, I mean that’s 9 plus 21? No, that doesn’t
make sense. Shoot... I’m not sure where to start?
I: Do you have a picture in your head of the situation you’re thinking
about?
S: Yeah, sort of.
I: Do you think you could draw it?
S: Um, I guess so... Yeah... here’s the lily bud 9 inches above the
water when it’s standing straight up... Now when the wind blows it
stretches to the side and then the bud is over here, 21 inches away.
Wait, is that 21 inches to the right, or 21 inches from here to here
[pointing from the upright bud to the bud at the waterline]?
I: I think you can assume 21 inches to the right, but that’s a really
good question, and an important one to get sorted out before you get
too far along. Did drawing the picture bring that question to mind?
S: Yeah, definitely. OK, so when the wind blows the lily over there,
the stem is slanted like this... and ... so we want the depth of the
water, which is this distance [drawing a vertical line from the lake bottom
to the waterline]... so this makes a triangle, is that what I’m supposed
to do?
I: There are no “supposed to’s” in this class.
S: Oh, so how will I know when I’m doing it right?
I: There’s more than one way to solve the problem, so there isn’t just
one “right way”. If you’ve interpreted the problem correctly, chosen a
solution procedure that doesn’t violate the fundamental principles of
16 iowa state university

mathematics, and all your arithmetic is sound, you’ll get the correct
answer and you’ll know it.
S: Oh. But... how do I know if I’ve got the right answer? You’re not
going to tell me?
I: Welcome to the real world.

Exercises

1. Make a list of the variables or factors that one might need to take
into account when projecting future pheasant populations from
historic data such as the Iowa Roadside Pheasant Survey.

2. How do game wildlife managers benefit from making population

estimates like the roadside survey?

3. See if you can arrive at a solution for the Lilavati water lily prob-
lem. While you’re working on this, occasionally step back and
think about what sorts of things you are doing to make progress.
Are you trying to understand what the problem is about? Trying
to appreciate the geometry? Trying to recall algorithms that relate
different quantities in a geometric shape? If you feel stuck at any
point, what do you think is preventing you from making progress?
Record and annotate all of your thoughts and solution attempts,
and try not to erase or scribble anything out.
 Part I

PROBLEM SOLVING
2
Problem Solving as a Process

2.1 What is a problem?

Let’s establish this right away: a problem is an intellectual challenge.

Solving a problem is then a process of undertaking and overcoming
the challenge.
As we have indicated elsewhere, authentic problems are often
poorly-structured and vague. They are not carefully-crafted to yield
whole-number answers with a few minutes of symbolic manipula-
tion, like those you typically encounter in school. Authentic problems Many mathematics teachers and schol-
could take hours, days, or even longer to solve, and you may not al- ars distinguish between problems and
exercises. An exercise prompts you to
ways know with confidence that you have succeeded because the an- practice a method that you recently
swers aren’t in the back of the book. Problems don’t require only the learned or for which you recently
studied examples. Exercises usually
application of a recently-learned method or algorithm; indeed, you have answers that can be compared
may not know at the outset which methods are appropriate for solv- with an answer key. A problem is
ing the problem. You may not be given all the information needed more challenging because there may be
fewer cues to the appropriate solution
to achieve a complete solution, or information you are given may be methods and there may be no explicit
uncertain or incomplete. In short, authentic problems are hard, and relationship between these methods and
the instruction received.
that can be frustrating.
But wait! Don’t close your book (or laptop) and walk away just
yet! Having just explained the difficulties, consider the flip side of the
problem-solving coin: problems that are authentic are also inherently
interesting, particularly when they address contemporary issues or
puzzles in your chosen area of study. Solutions and solution methods
for such problems are therefore not just an academic dead-end, but
can lend themselves to practical applications in the real world. The
rewards of achieving a clever and well-justified solution to a practical
and interesting problem should outweigh by far the moments of
uncertainty, frustration, or disappointment encountered along the
way.
Most of the problems discussed in this book are designed to mimic
authentic problems, and in many cases are drawn from or inspired by
encounters with local researchers and practitioners. As we endeavor
20 iowa state university

to address these problems, we’ll often find it helpful to explore sim-

pler problems and exercises to help with sense-making. Thus, our
time will be spent moving back and forth from focus problems to
auxiliary problems and exercises. Read on through the end of this
chapter to understand how this approach can lead to more successful
problem-solving.
Problem solving cannot be reduced to a simple recipe or fast and
easy method, but in the past 70 years, much has been learned about
how successful construction of solutions differs from unsuccessful
attempts. A key component is the use of heuristics, or habits of mind
that are useful in solving problems. The modern idea of heuristics
has its origin in the work of Hungarian mathematician George Pólya
in the mid 20th century. Heuristics help guide us in decisions about
how to approach a problem. With the help of heuristics and the ben-
efit of experience, we may develop problem-solving strategies that
A strategy is a definite sequence of lead to successful solutions. We’ll begin our study of problem-solving
steps or operations that leads to a with a brief look at Pólya’s method and some of his heuristics and
solution.
then consider how they might apply to problem-solving in the natu-
ral science and natural resource management contexts.

2.2 Pólya’s method and beyond

A credentialed mathematician and academic, Pólya was no stranger

to the struggles of solving difficult problems. But he was also a
teacher and concerned himself with the development of problem-
solving skill and intuition in students. He studied his own problem-
solving process and that of his professional colleagues and distilled
his observations into four essential principles. These principles are
general–that is, their use needn’t be limited to mathematical problem-
solving. The method can be summarized as follows:
 quantitative problem solving in natural resources 21

Pólya’s Method, Condensed and Slightly Revised

1. Understand the problem. What is the unknown or target

quantity? Is there enough information to find a solution?
How is the information that is available relevant to the un-
known?

2. Plan a solution strategy. How can you proceed from the

information available to the unknown? What steps are neces-
sary, and how will the given information be used?

3. Execute the solution plan. If the solution plan is chosen well,

the implementation of the plan should yield the sought-after
result. If unsurmountable difficulty arises, an alternative plan
may need to be formulated.

4. Check the result. Does the result satisfy the conditions stated
in the problem? Is it consistent with expectations or within
reasonable bounds? Can you arrive at the same result using a
different approach?

If it helps to have a mnemonic to

These principles may seem obvious, but when the time comes to remember these, how about UPEC for
Understand, Plan, Execute, and Check.
actually solve a problem it is easy to overlook one or more of them
or to lose track of what we are after. Employing this method as a
general framework for approaching problem-solving will yield more
consistent success and more reliable results.
Consider the most common words out of a typical college stu-
dent’s mouth when confronted with a novel problem: “I don’t know
where to start”1 . Perhaps the student really means “you haven’t yet 1
If I had a nickel for everytime I’ve
told me exactly what to do to get the answer”. But if the instructor heard that...

were to point the student toward a solution method every time she
was confronted with a challenging problem, she would learn only
two things: 1) how to implement algorithms and compute numerical
results as instructed; and 2) to relinquish all control of choosing how
to approach and solve a problem to somebody else. Sadly, this is of-
ten the best outcome of the standard school mathematics curriculum.
The worst outcome is that students dismiss math as boring, difficult,
or irrelevant. In some cases, a diligent student develops some facil-
ity2 with basic manipulations of mathematical symbols and numbers, 2
Though the memory of how to use
but little or no ability to create the frame of mind and methodological mathematical algorithms certainly
degrades with time unless used or
structure needed to begin and confidently proceed trying solutions. reviewed frequently
For the challenge of getting started, Pólya’s framework offers Un-
derstand. What is the problem really asking, and what exactly do
you want to end up with as a result? Take, for instance, the pheas-
ant count problem that we introduced in the first chapter: what is
22 iowa state university

the unknown in that problem? Essentially, the question we posed

there was how many pheasants we should expect to be living in Iowa
in the next few years. In some ways, this is a specific reinterpreta-
tion of the problem statement, but the simple act of making that re-
interpretation not only helps us understand what we are looking for
concretely, but our formal statement of it might clarify to colleagues
or readers what our solution is driving at.
Since Pólya’s framework was developed from the perspective of a
mathematician, some of the questions and suggestions pertain mostly
to abstract problems. The framework doesn’t take advantage of the
fact that most of our problems are situated in real-world contexts
and involve quantities whose properties can be used as an asset in
identifying, constructing, and evaluating solutions. If you’re not
exactly clear on what I mean by that, go ahead and peak at the next
chapter where we discuss the definition and properties of quantities.
On the next page, I have expanded and elaborated upon Pólya’s
framework and adapted some of the details for problem-solving in
natural sciences.

Pólya’s framework, and our elaboration of it for problems in the

natural sciences, may help us be better organized, but when it comes
to applying the quantitative skills we spent years developing in math
and statistics classes, we still have little guidance. In Pólya’s How to
Solve It, this is the point where the idea and utility of heuristics was
introduced. In the next section, we will introduce and review some
generic heuristics that can aid with understanding the problem and
inspiring a solution plan.

2.3 A Few Versatile Heuristics

The heuristics we consider here are just a snapshot of the generic

methods we might employ in many problems, and some of these are
further elaborated in subsequent chapters. These should be some
of your most frequent go-to tools for the initial understanding and
planning phases, and can be interpreted differently according to
the constraints or conditions of each problem. In selecting optimal
strategies, the field of options may be narrowed by examining the
nature of the unknown. Think about the unknown and what it rep-
resents, not only for the focus problem but for any sub-problems
or for any goals identified as necessary to solving the focus prob-
Heuristic: Narrow the options lem. Could the problem be stated in the form “how much...?”, “how
Use the known properties of the many...?”, or “is...or not?”. If so, the problem might require arithmetic
unknown quantity to identify strategies
appropriate to the problem. and/or algebraic reasoning. If data is available or supplied and the
problem can be stated in the form “what relationship...?” or “how
 quantitative problem solving in natural resources 23

Solving Ill-Structured Problems

1 Understand the problem

Do you understand the problem as it is stated?
Can you restate the problem in your own words?
What is the unknown or desired quantity or output (be specific)? Is it a number? A function? A
procedure?
Can you make a drawing or diagram to illustrate how the unknown relates to any known quanti-
ties or to the broader problem-space?
Do you already know approximately what the value should be? Can you guess a ballpark or range
of reasonable values?
How accurate does your solution need to be? What are the consequences of errors?
What information do you already have?
Is the information that you already have sufficient to solve the problem?
If appropriate, can you write the problem as an algebraic equation with suitable notation?

2 Plan a solution. Consider multiple approaches if possible

Have you successfully solved a problem like this before?
Has somebody else documented a solution method to this or a similar problem?
If the problem can be written explicitly as a mathematical statement, do you recognize an algo-
rithm or heuristic that can yield the desired unknown?
If a solution method is apparent, can you assemble all the needed quantities?
If the problem is not immediately solvable, or a solution method not yet apparent...
Can you break the problem into smaller sub-problems that may be easier to solve?
Can you approximate uncertain or unknown quantities?
Could you solve a related auxiliary problem to gain insight?
If data are given, what exploratory analyses could be done to spark ideas?

3 Execute the plan

At each step, check to see if the incremental result matches expectations.
Double-check all formulae and algebraic manipulations.
If appropriate, is unit/dimensional homogeneity satisfied?
If you encounter difficulties, revisit the plan and alternatives.

4 Check the solution

Is the result reasonable?
Is it consistent with ballpark estimates or benchmarks?
If appropriate, can you substitute the result into the original problem and satisfy the assumptions
and conditions?
Double-check algebraic manipulations.
Double-check numerical compuations.
Could you document your full solution with a concise but complete summary of steps, justifica-
tion of assumptions and methods?
Could a colleague reproduce your approach and find the same solution?
24 iowa state university

does...change as you vary...?”, graphical and statistical reasoning are

probably appropriate. If the problem can be stated in the form “how
big...?”, “what distance...?” or “where...?”, then geometric or spatial
3
Each of these types of reasoning and reasoning could be necessary3 . There are certainly problems that
the context-specific strategies that won’t be easily stated in any of these terms, and all options should
are particularly helpful for them are
reviewed in separate chapters in this then remain open. Nevertheless, recognizing common properties in
book. the nature of problems can sometimes narrow down your choices of
strategies and make promising solution methods or approaches more
apparent.
• Break the problem into sub-problems. Complex problems often
require multiple steps that can be divided into discrete sub-goals.
For example, determining the value of an unknown quantity that
is required to solve the focus problem can be considered distinct
from solving the focus problem itself. Therefore mapping out the
solution in terms of incremental sub-goals can clarify the pathway
to a solution.

• Express conditions algebraically. Assign symbols to relevant

quantities and express the relationship, if known, as an equation
relating the quantities. If relationships are not known beforehand,
use dimensional analysis to suggest them.

• Guess the correct answer or solution. Usually a guess or ball-

park estimate is not sufficient if the issue is truly a problem, but
estimates can still be used to help you recognize if you are on the
right track in later computations. If you know approximately what
the solution is or what range it should lie within, use this value
to check your work. If the solution is not known beforehand but
algebraic or practical constraints are available on related quanti-
ties, use those quantities to get a ballpark estimate. At this stage,
back-of-the-envelope calculations in scientific notation can make
quick work of it.

• Try a few values. Where some values are unknown but are not
the desired quantity, try to solve the problem with a few supposed
numerical values. Sometimes the outcome of algebraic relation-
ships is relatively insensitive to the precise value of the quantities
included in the relationship. When the relationship is strongly sen-
sitive to unknown or poorly-constrained quantities, identify those
quantities as important intermediate goals or sub-problems.

• Draw a picture or diagram. When the problem is inherently spa-

tial, such as in the case of habitat or landscape ecology problems,
make a map or schematic drawing of the geometric or spatial rela-
tionships between known and unknown quantities. To the extent
that it is possible, scale distances or spatial dimensions accurately.
 quantitative problem solving in natural resources 25

• List all possible cases. If you’re confronted with a logic or simple

probabilistic puzzle, make a list or matrix containing possible
permutations or combinations.

• Work backwards. Where the desired ending condition is known

or can be approximated but the steps to reach it are not known,
use the end result to help “back out” the steps.

• Visualize the data. If data or input values are given and a rela-
tionship or summary-statistic is desired, make a graph or diagram
from the data. In some cases, visual representation of the data can
suggest or substantiate approximate values for the unknowns or
can illustrate the form of functional relationships.

• Solve a simpler problem. Sometimes a condition or relationship

is too complex to solve easily in its full form. In these cases, it
may be possible and helpful to simplify the condition to make the
problem tractable. You can do this by assuming that an unknown
value is known (as in Try a few values), by eliminating one or more
terms in sums and differences, or by using a simpler statement of
the original condition.

The heuristics above are by no means a recipe for success in every

situation, but they should be available to you in your repertoire of
things to consider. In the chapters that follow, we will elaborate on
some of these strategies and add more context-specific tools that can
be applied in practical problems.

2.4 Stepping back

Before we set you loose on solving problems, it is important to ad-

dress the mind-set of problem solving. If you have ever uttered the
words “I suck at math” or something similar, this section is partic-
ularly for you. But as far as I’m concerned, realizing how the mind
constructs knowledge and understanding in a problem-solving task
is an empowering notion. Alan Schoenfeld, mathematician and math-
education specialist, has identified four aspects of the mental process
of problem-solving that are essential: Resources, Heuristics, Control,
and Belief. Each is necessary and problem-solving cannot or will not
proceed without them. First, let’s look at what each means, and then
we’ll consider how they contribute to our problem-solving success.
26 iowa state university

• Resources
These are the things you know or understand about the prob-
lem domain, constraints on quantities and their representation
in the problem domain, and the skills you possess in perform-
ing algorithmic procedures.

• Heuristics
The decision-making tools used to make sense of challenging
problems that allow you to make progress or develop insight.
Most of the chapters of this booklet are devoted to developing
strategies useful in natural science or resource management
domains.

• Control
Control is the management and self-awareness of the
problem-solving process. It includes planning, execution
and evaluation decisions and and the selection of resources
and heuristics for the problem.

• Belief
Belief includes the set of notions one has about the problem
domain as well as one’s own abilities or challenges in ap-
plying math and statistics to the problem domain; this also
includes preconceptions and (mis)understandings that could
lead to the use of (in)correct resources and heuristics in a
given problem.

Your mathematics and statistics education up to this point has al-

most certainly stressed resources. Thus, the quantitative resources
you bring to a problem consist of all the algorithms and methods you
know how to use and your understanding of what they do or mean.
Unless you’ve followed a curriculum through school and college that
has deliberately made use of these resources, you’ve probably forgot-
ten many of them, but re-learning them may not be as challenging as
learning them naively. Control and belief are gained from experience,
and can build from any foundation of resources and heuristics. The
heuristics and strategies themselves can be a bit of a problem though.
You may have been instructed some in the development solution
strategies in school, but there’s an important distinction between
learning what to use and learning when to use it. Some people may
argue that it is not a mathematician’s prerogative to instruct students
in their service courses in more than resources, since heuristics vary
from one discipline to another and control and belief grow with
 quantitative problem solving in natural resources 27

experience. This argument is fair, but as non-mathematicians we’re

left with training in how to implement algorithms, but little idea
about how to use those algorithms unless presented with problems
that aren’t really problems but exercises.
As a result, when the going gets tough,...we get stuck. That’s
where this course comes in (hopefully to the rescue??). This is your
opportunity to work with resources you already have at your dis-
posal, perhaps learn a few more, and to be introduced to strategies
for using them in problems that you might encounter in other nat-
ural resource courses, in internships, or in your career. By working
through these problems in a systematic manner, you’ll learn how to
control your problem-solving process while building your experience
base. I sincerely hope that your belief system evolves in such a way
that you become confident that you can solve quantitative problems too!

Exercises

1. Think of a challenging problem that you needed help to solve in

one of your high school or college courses. It could be a mathe-
matical problem, but doesn’t have to be. Describe the problem and
reflect on what assistance you needed to arrive at a solution. Were
you unable to get started? Did you need help with recognizing
and implementing the appropriate algorithms? What was the na-
ture of the assistance that helped you solve the problem? Was it
satisfying to arrive at the correct solution?

2. Now reflect on a challenging problem that you were able to solve

correctly without assistance. Why were you successful? Were you
able to overcome any difficulties or hurdles along the way? Was it
more or less satisfying to solve this problem on your own than to
solve a problem with help? Why?

3. What resources do you think a person needs to be able to make

good predictions of pheasant population over the coming 5 years?
3
Some teaser problems

To illustrate some of the ways that we can employ a deliberate prob-

lem solving process, apply general heuristics, and look toward de-
velopment of solution strategies, I offer some teaser problems. We’ll
revisit these problems throughout the text where they serve as fodder
for more targeted discussions of solution strategies. These problems
vary in complexity and are in most cases fairly open-ended. As we’ve
already discussed, these are features of real problems and though we
may feel a bit of anxiety about that, we should also realize that this
presents opportunities for creativity and ingenuity.

3.1 Waterfowl easements

To have the greatest benefit to migratory waterfowl, is it better

secure easements containing many small wetlands or fewer, larger
wetlands?
One of the greatest threats to waterfowl populations is loss of
habitat. Organizations like Ducks Unlimited and The Nature Conser-
vancy have advocated for preservation and restoration of wetlands
in key migration corridors and breeding areas. Among the strategies
that organizations like this have used is acquisition of conservation
easements, wherein a private land owner agrees to limit develop-
Figure 3.1: Two adjacent sections, each
ment on land containing quality habitat in exchange for benefits like approximately 1 square mile, with
management assistance and tax deductions. In practice, these orga- different number and size distribution
nizations cannot accept all easement donations but need to prioritize of wetlands.

those that will have the greatest long-term benefits to waterfowl.

Thus the question: if all else is equal, is it better to prioritize a parcel
with many small wetlands, or a parcel of the same size with fewer,
large wetlands (Figure 3.1)?
30 iowa state university

3.2 Herbicide purchase

How much herbicide is needed to kill and suppress regrowth of

invasive woody shrubs in a wooded urban park?
A common management problem in urban woodlands is the con-
trol of invasive or unwanted understory vegetation. In the midwest-
ern USA, bush honeysuckles (genus Lonicera) and European buck-
thorn (genus Rhamnus) are particularly challenging to manage, and
the most effective means of control is usually with herbicide. The
simplest approach for many of these woody invasives is to cut all the
stems and paint stumps with a general herbicide like glyphosate. So
what must we know and determine to decide upon the quantity of
glyphosate required?

3.3 Deer-automobile collisions

How likely are collisions between automobiles and whitetail deer

in your county, and how might they be influenced by changes in
deer population management?
If you live in an urban environment without open spaces and
wildlife corridors, the answer to this may be easy: zero likelihood.
But in many areas both periurban and rural, deer are a familiar sight.
In these settings, deer-automobile collisions may unfortunately be all
too common. We might infer that the frequency of collisions would
be greater where both deer and drivers are more numerous, but can
we quantify this?

3.4 Prairie dog plague

Is it possible to prevent the spread of Sylvatic plague through

prairie dog colonies?
Yersinia pestis is a flea-borne bacterium that causes Sylvatic plague,
an often fatal disease that primarily affects rodents. The black-tailed
prairie dog (Cynomys ludivicianus), already threatened by land-use
change and hunting/poisoning within it’s native range in the North
American Great Plains, periodically experiences plague epidemics
that can decimate colonies. In addition to harming prairie dogs di-
rectly, however, plague can be transmitted both directly and through
fleas from prairie dogs to their predators, including the endangered
black-footed ferret. In isolated populations of prairie dogs, moni-
toring and active management can potentially be used to reduce the
spread of plague, but designing such a strategy requires that we first
understand the dynamics of disease transmission within colonies.
 quantitative problem solving in natural resources 31

3.5 Forest fire losses

How much fuel management is optimal to minimize risk of fire

damage in a forest managed for timber production?
In many western forests managed for timber production, historical
fire exclusion has led to a buildup of fuel wood, leading to height-
ened risk of destructive fire. When fires do occur in these settings,
high fuel density can lead to destructive crown fires that greatly di-
minish the value of the timber and can damage or destroy surround- cost
ing infrastructure. To combat this, forest managers can undertake C + NVC
fuel reduction treatments such as thinning and prescribed burning, C
and emergency managers can aggressively fight fires when they do
start. However, such suppression and pre-suppression efforts can
be costly and can have diminishing returns in terms of cost and re-
sources (Figure 3.2). Is there an optimal amount of fuel reduction and NVC
suppression that provides some protection from risk while containing suppression

costs?
Figure 3.2: Schematic illustration of
one conceptual model for the optimal
management of fire fuels. Total cost
3.6 Maximized effluent C of management increases with
suppression and pre-suppression effort.
How much of phosphorus (P) can be discharged from point sources With little suppression effort however,
the risk of high losses or net change
into an urban stream without exceeding total maximum daily loads in value, NVC, is high, declining
(TMDLs)?1 with suppression effort. This view of
The U.S. Clean Water Act of 1972 establishes criteria for identify- management incentives indicates that
overall cost is C + NVC.
ing polluted water bodies and designing plans to reduce pollutant 1
Inspired by Litwack et al., 2006, Journal
loads and improve water quality. One of the measures available to of Environmental Engineering, 132(4):
water resource managers is the establishment of Total Maximum 538-546.

Daily Loads (TMDLs), which typically limit the permissible concen-

tration of a pollutant in an impaired water body. If the goal in this
case is to limit the amount of P delivered to downstream water bod-
ies, we need to determine how much P constitutes the maximum
allowable concentration.

3.7 Brook trout recruitment

Given data from four consecutive electrofishing passes in an iso-

lated stream reach, what is the stream’s age-0 brook trout popula-
tion?
One method of estimating fish populations in streams is to iso-
late a stream reach with barriers upstream and downstream of and
performing an electrofishing transit of the reach, collecting and mea-
suring each fish retrieved. When extra detail is needed or fish are
elusive, multiple passes through the reach may be required, and re-
trieved fish are removed to a live well or adjacent reach. Through this
32 iowa state university

method, the change in catch between consecutive passes can be used

to estimate the actual population within the reach. To isolate the pop-
ulation of age-0 (young of the year) fish, age must be assessed from
2
A nonlethal age estimate can be each fish either directly2 or indirectly from the length distributions in
obtained by collecting and observing the catch in each pass.
fish scales under a microscope

The problems above deal with a variety of practical issues that

could be addressed by a natural resource professional. None of them
are expressed explicitly as math problems, but quantitative meth-
ods could be instrumental in solving each of them. Indeed, this is
characteristic of many real science and management problems. Be-
fore they become quantitative problems, they must be interpreted
and carefully re-framed (Figure 3.3). This is not only an important
part of the process, but in many cases is the most challenging and
pivotal step. In practice, this must be part of the process of under-
standing the problem and perhaps even planning a solution strategy.
In most cases, not enough information is given up front and the ex-
pectations for the kind of solution desired are vague. In this sense,
they are ill-structured problems, requiring some digestion and careful
re-phrasing before they can be understood as solvable quantitative
problems.
Gathering the information deemed necessary to arrive at the solu-
tion is, of course, an essential step in the process. When possible, we
should strive to include this step as part of the coursework. However
the design of experiments and logistics of making novel field mea-
Figure 3.3: A pathway between prob- surements can and should be addressed in their own courses, and we
lems and solutions in practical natural
resource management, including vari- cannot hope to do justice to those concerns here. As a consequence,
ous quantitative approaches. where the teaser problems are addressed here and in the chapters
that follow, we’ll either work with hypothetical data or engage real
data from the literature or from government documents and web
resources.
Importantly, these problems also vary in the degree to which we
could ever hope to know that we have the “right” answer. For ex-
ample, It may indeed be possible to know exactly the maximum P
load delivered in Problem 3.6 or the herbicide volume needed in
Problem 3.2, provided that we had perfect information on stream
discharge and stem basal area, respectively. However, for most of
the other problems, confidence that we have a good solution must
come from confidence that we have made well-reasoned interpreta-
tions and assumptions and used technically-correct manipulations
and analysis. Thus, part of the burden of solving such a problem is
articulating how the problem is interpreted and how chosen solution
strategy follows from that interpretation.
Each of these problems provides hints or implicit cues to what
 quantitative problem solving in natural resources 33

kinds of strategies are appropriate to assemble a solution. For exam-

ple, there are clear spatial elements to Problem 3.1 and Problem 3.2,
but there may also be subtler spatial components in others. By spa-
tial, I mean that we need to incorporate information about how big or
how long something is into our solution. Addressing these elements
of problems requires some tools from geometry and trigonometry, as
well as perhaps the language and conventions of geography or car-
tography. Therefore, this book contains several chapters exploring the
aspects of spatial reasoning that arise frequently in natural resources. Spatial reasoning is the use of spatial
Not surprisingly, several of these problems point to issues of how information or relationships, like
lengths, areas, volumes, or directions, in
many, particularly Problem 3.3 and Problem 3.7. In these and other the solution of problems.
problems, we may wish to characterize individuals or populations
in terms of a representative value or distribution of values, or we
may need to compare values or proportions. These tasks engage our
experience with arithmetic, descriptive statistics, and probability, but
in many cases also require that we are careful with units and that
we can work efficiently with the extremely large or small numbers
sometimes encountered in the sciences. For these issues, we have
several chapters devoted to numerical reasoning. Numerical reasoning, as used in this
Aspects of most of our problems can be expressed in terms of the book, is the manipulation, characteriza-
tion, comparison, and interpretation of
relationship between multiple variables, or relationships between numerical values (such as data) in the
cause and effect. For example, in Problem 3.5 we may need a way service of problem solving.
to represent the relationship between how much effort is made in
fire-suppression activities and their cost. Since we don’t necessarily
have an idea ahead of time about how much effort is appropriate,
an algebraic expression relating the variable C (cost of suppression
efforts) to the amount of suppression effort itself (call it S) could
stand in and allow us to employ algebraic reasoning as a means to a Algebraic reasoning is the use of
solution. generalized variables and formal
relationships between them, rather than
Finally, several of these problems imply that we seek an assess- numbers, as a means of constraining
ment or prediction of future, hypothetical, or unobservable quantities solutions.
or events. For example, Problem 3.4 seems to ask whether something
is even possible, even though we don’t know what that something is.
In these cases, finding a useful solution may require modeling. As in- Modeling in this book refers to the con-
struction of a simplified representation
dicated in Chapter 1, a model is simplified representation of real sys-
of real or hypothesized systems, often
tems and is constructed for the purpose of exploring cause and effect described with one or more equations
or functional relationships between variables as system conditions are relating variables to one another and to
system properties, and used to explore
changed. To model the spread of plague among prairie dogs, then, complex or unobservable phenomena or
we may need a way to characterize the transfer of fleas between an relationships.
infected individual and an uninfected, but susceptible individual.
This type of reasoning will often require algebraic constructs and
simplifying assumptions, and may therefore require competency in
algebraic reasoning. Even so, however, some insights can be gained
from model construction even if the detailed relationships between
34 iowa state university

variables aren’t formally articulated with equations.

The Parts and Chapters that follow are elaborations and demon-
strations of problem-solving strategies employing each of these types
of reasoning. Not all will be applicable to every problem, and there
isn’t necessarily a sequence of interdependence. Therefore, I hope
that rather than reading the text in order, that you will consult it
as the need for ideas and insight arises in problems you are pre-
sented with. I don’t provide complete solutions to any of the teaser
problems above, but do use parts of them to illustrate concepts and
strategies as they arise.

Exercises

1. Which of the teaser problems outlined above do you find most

interesting or compelling and why?

2. Using the table on solving ill-structured problems in Chapter 2 as

a guide, attempt to Understand a teaser problem of your choice
from the list above. Write your response to each of the questions
under the Understand prompt, or write “n/a” if not applicable.

3. Try to write one or part of one of the teaser problems as a more

conventional math problem, with only equations and no words.
What is challenging about doing this?

4. Write a problem of your own in a format similar to the teaser

problems above: a short title, a boldface question, and a short
paragraph elaborating on the context or significance of the ques-
tion. Choose a natural resource topic that interests you.
 Part II

NUMERICAL
REASONING
4
Quantities in the Real World

In this course, we seek to solve practical problems in natural re-

source management and ecology, but we focus on the use of quanti-
tative tools in service of this objective. Before diving too deeply into
problem-solving, we should ensure that we know what is meant by
quantities, quantitative tools, and quantitative reasoning. We will
also establish a few conventions for how quantities are represented
in science and how quantitative information can be most effectively
communicated.

4.1 Quantities in Natural Resources

If we can assess the presence or absence of something, count its num-

ber, measure some property that it has, or compare it to another
object, it can be quantified. That quantified thing is then represented
by a quantity that is itself now a property of the quantified thing.
If that sounds confusing, read on to some of the examples below. A
fully-defined quantity has five components:

Properties of quantities

• Name: what we call it.

• Procedural statement: how it is measured or computed.

• Number: numerical value(s) corresponding to magnitude or

multitude.

• Units: how it is scaled.

• Symbol: a character that stands for the quantity in equations.

Defining a quantity might seem somewhat pedantic, but it has

important implications for what we can and cannot do with it. This
contrasts fundamentally with the abstract variables we encountered
38 iowa state university

in high school math. In that setting, there is rarely any reason to

question whether it is OK and meaningful to add 3x and 8y, we just
do what we’re asked. But in the world of real quantities, if x stands
for “milligrams of sodium chloride” and y stands for the number
of eggs in a Northern Cardinal’s nest, it’s not so clear that we can
perform that addition. Even if we do, it is not so clear what the result
means.
In the introductory chapter, we pondered the Iowa DNR’s roadside
pheasant survey and what it means for pheasant populations across
the state. The pheasant count yields a single number each year, for
example 23.9 individuals per 30 miles in 2015. We discovered what
this quantity means and how it is measured in the Introduction, so
we already have most of the ingredients of a fully-defined quantity.
We only need a symbol. This is a pretty trivial step in simple prob-
lems, where the primary constraint is to make the symbol unambigu-
ous and suggestive of the quantity it represents. Perhaps we should
then choose P for our symbol. If we were to prepare a document de-
scribing the DNR roadside pheasant survey, once we establish each
1
Name, procedural statement, number, of the five properties1 of our quantity, we can thereafter use P with
units, and symbol confidence that the information conveyed by that symbol is clearly
established: within the context of our document, P would refer to
the series of annual estimates of pheasant density according to the
method established by the DNR. This formality thereby provides a
shorthand name and eliminates any ambiguity in discussion of quan-
tities. Note that this is different than P, which we used previously as
the chemical symbol for the element Phosphorus. If we happened to
be working on a problem involving both pheasants and Phosphorus,
we might select our symbol differently to avoid ambiguity.
As we indicated above, there are many types of quantities that we
encounter in science. We can treat presence/absence information as
2
Nominal-scaled quantities take the a quantity, measured with a nominal scale2 . Was there a black-capped
form of categories; a few pairs of chickadee on the bird feeder at 3:30 PM? Is there a beetle in the pit-
categories that fit this definition might
be present or absent; infected or not trap we set up in a field? An ordinal scaled quantitiy is one in which
infected. individual measurements or components are ranked or ordered. In
what order did different birds arrive at and depart from the bird
3
Ordinal-scaled quantities have values feeder3 ?
according to their rank among the In the above cases, distinctions between the scales are clear and ob-
population or data.
vious, but in others they can be more subtle. Consider the expression
of the quantity temperature. We have several scales we can choose
from for measuring and expressing temperature. Most important and
familiar are the Fahrenheit, Celsius (Centigrade), and Kelvin (abso-
lute) scales. The Celsius scale, for example, is defined according to
the freezing and boiling points of water, and the temperatures that
those phenomena corresponded to were defined to have values of
 quantitative problem solving in natural resources 39

0◦ C and 100◦ C, respectively. A degree of Celsius is therefore defined

on an interval scale4 , where a unit of temperature change is 1/100th 4
Interval-scaled quantities may take
of the difference between the boiling point and freezing point of wa- negative values.

ter. In contrast, Kelvin is a measure of the absolute thermal energy

contained in a substance, in reference to a theoretical state of zero
energy5 . Kelvins are a ratio scale6 , with a natural zero point and a 5
the Kelvin scale, named for Lord
unit magnitude that represents the value on a scale relative to that Kelvin (William Thomson), is consid-
ered an absolute temperature scale,
zero point. It can be confusing to distinguish between interval and measuring the absolute thermal energy
ratio scaled quantities, but the following example might help illumi- in a substance.
6
Ratio-scaled quantities cannot have
nate the difference. Imagine that you measured a lake’s surface water negative values; a non-positive value
temperature at one moment in mid March to be 1◦ C. Now suppose indicates an absence of the quantity.
you returned a week later to find that the temperature has increased
to 2◦ C. Wow, the temperature doubled, since 2 is twice as large as 1!
Well, no it didn’t. Because Celsius is an interval scale, chosen arbi-
trarily to have a value of zero at the freezing temperature of water,
we cannot say it doubled. The issue with this might be even more
clear if our first measurement had been -0.1◦ C (supercooled!) instead
of 1◦ C. Then if we apply the same logic as before to characterizing
the change to 2◦ C, we would have to say that the temperature is -20
times the initial value, which is ridiculous. It’s ridiculous because our
scale does not have a natural zero point, and instead describes inter-
val differences between a measured temperature and a standardized
reference point. In contrast, a temperature of 300 K is actually twice
as hot as a temperature of 150 K, since this is a ratio scale7 . 7
For additional topical perspective
It might seem a bit esoteric to define quantities and their unit on quantities and scaling in ecology,
consult a resource like Schneider, D.C.,
scales in so much detail, but hopefully you can now see how different 2009, Quantitative Ecology, 2nd ed.,
scales might entail different rules for what manipulations are or are London, UK, Academic Press, 415 p.

not meaningful.

4.2 What Quantitative Reasoning Is

Using, manipulating, and interpreting quantitative information are

the prerogative of professional scientists and natural resource man-
agers. The term quantitative reasoning is often used to describe the
processes of constructing and making sense of contextualized quan-
titative information8 . When we look for patterns in water quality 8
One researcher describes quantitative
data, interpret pheasant population trends, or estimate the number reasoning as “sophisticated reasoning
with elementary mathematics, rather
of person-hours required for removal of invasive shrubs in a forest, than elementary reasoning with so-
we are applying quantitative reasoning. An individual’s aptitude phisticated mathematics.” (Steen, L.,
2004, Achieving Quantitative Literacy: An
for quantitative reasoning is sometimes referred to as quantitative Urgent Challenge for Higher Education,
literacy. Washington, DC, MAA.)
A guiding premise of this book is that the lifetime value of strong
quantitative literacy is greater than that of pure mathematics for nat-
ural resource professionals. To be sure, basic mathematical skills are
40 iowa state university

fundamentally essential for quantitative reasoning. But for most pro-

fessionals outside of engineering and the physical sciences, these
skills are mostly learned in secondary school math. What often
isn’t learned is how to use those skills to make sense of things in
the world around us.

4.3 Units & Dimensions

Paying attention to units and dimen- Quantities with practical meaning to us will often have units. Con-
sions is not mere formality. In the sider a few random examples:
scaled quantities we use in science,
units and dimensions constrain what
mathematical operations are permissi-
• 71 foot tall tree
ble.
• 16.4 grams of soil

• 42 snow geese

• 39 breaths per minute

• 385 ppm carbon dioxide

Each of these examples would lose much of their meaning if we

only stated the number and not what the number corresponded
to: 71, 16.4, 42, 39, 385. We can certainly plug these numbers into
equations and work with them in a purely mathematical sense, but
they probably no longer have a meaning that we care about. Thus,
we need to keep track of units. Units can be any standard template or
scale according to which we measure a quantity. A foot, for example,
though originally loosely defined by the length of a Greek or Roman
man’s foot, is now defined with reference to a meter, which in turn
is defined according to the distance light travels in a vacuum during
a pre-defined time period. Likewise, grams and minutes are units
of mass and time defined according to internationally standardized
references.
Dimensions are sometimes conflated with units, but the term is
distinct and more broad. In casual usage, the term “dimensions” of-
ten implies that we wish to know how large something is - in other
words, its length, area or volume. In the physical sciences, dimen-
sions are a way to group different types of units that can be simply
scaled with one another. Inches and centimeters are both lengths.
No matter what object you measure – perhaps a young brook trout’s
total length – if it is an inch long, it is by simple unit conversion also
2.54 centimeters long. Both are units of length. But just because I
find that the fish is 1 inch or 2.54 centimeters long, that does not nec-
essarily tell me how heavy it is. So mass is a different fundamental
dimension than length, and whether I measure mass in grams or
slugs, it has dimensions of mass. In fact, most of the quantities that
 quantitative problem solving in natural resources 41

scientists deal with are composites of only three fundamental di-

mensions: mass, length, and time. Sometimes we use the symbols Most quantities in the natural sciences
[ M], [ L], and [ T ] in square brackets to denote when we are talking have dimensions involving mass [ M ],
length [ L] and time [ T ]
about dimensions. We can also have dimensionless quantities which
we can (for reasons that we’ll see below) denote with [1], and which
we’ll often encounter when we are dealing with nominal, ordinal, or
multitude scales.
A quantity made up of any combination of the fundamental di-
mensions can be called a derived quantity. Speed is an example of a
derived quantity because it implicitly contains two quantities, a dis-
tance or length traveled during a period of time, [ L T −1 ]. Note that
we haven’t said whether we are measuring length in meters or feet or
whatever, nor have we said that time is in seconds or hours or millen-
nia. Dimensions are more like categories of units, and we can convert
from one set of units to another within each category by performing
multiplication or division. Energy is also a derived quantity, and can
be expressed in units of Joules or Calories. But regardless of the units
employed, any quantity of energy has dimensions of [ ML2 T −2 ]. We
just need to be careful if we are working with equations that we don’t
mix units, or we’ll be left with gobbledygook.

4.4 Comparing apples and oranges

We all learn in junior high or high school science to pay attention to

units, and for good reason. Keeping track of units and making sure
that our units are consistent in any computation we do with phys-
ical quantities can prevent costly mistakes. I’m not terribly pleased
about it, but I probably spend a few hours a month painstakingly
performing unit conversions in my research – indeed it may be the
most frequent type of computation I do, but I do it because I know it
is important.
Let’s consider a simple contrived example, similar to one that we
will encounter later this semester: suppose you are told that the soil
in your back yard has 15.84 ounces of lead per metric ton of soil.
Should you worry? Well, according to EPA guidelines, remediation is
recommended if lead concentrations in residential yards exceed 400
parts per million. Parts per million is a dimensionless unit similar
to a percent but much smaller and it can correspond to a mass (or
volume) of one substance contained in a mass (or volume) of some
mixture of substances. Our measurements are already both masses,
so 15.84 ounces/1 ton is [ MM−1 ], and is already dimensionless, but
our units are not consistent. We could fix this in several different
ways, but one simple one might be to convert tons of soil to ounces,
so that we have ounces over ounces, yielding consistent units. We
42 iowa state university

could spend lots of time (not wisely, I would argue) setting up frac-
tions on paper to figure this out (how many ounces in a pound, how
many pounds in a kilogram, how many kilograms in a ton. . . ), but
as long as we have access to a computer we might as well use it.
Plenty of apps and websites will do basic unit conversions for you,
including google itself. Employing one of these, we see that there are
35274 ounces in a metric ton, and now we can express our measured
concentration as 15.84 ounces / 35274 ounces, or about 0.00045. To
convert from this decimal value to parts per million, we just need to
multiply by one million or 106 (much like you would multiply by 100
to get a percent), giving us roughly 450 ppm lead. So at 450 ppm, the
lead concentration in our soil exceeds the threshold that the EPA has
deemed safe, but is not cause for great alarm.

4.5 Problem solving without numbers

Dimensions are so useful in science and engineering that there are

entire subdisciplines devoted to dimensional analysis. While these
folks are usually using dimensional analysis as a technique for ex-
tracting relationships between variables in complex equations that
cannot be solved directly, we can also use dimensions to help solve
simpler problems and catch errors or inconsistencies. This is because
physically-meaningful equations that describe relationships between
9
Dimensional homogeneity requires quantities must be dimensionally homogeneous9 . That is, the di-
equations, if expressed only in terms of mensions on one side of the equation must be the same as those on
the units representing each quantity, to
remain equal the other side. If we know the units of every quantity in a mathemat-
ical relationship and can figure out the corresponding dimensions,
we can compare dimensions on either side of the equation to verify
that it is dimensionally homogeneous. This can feel alot like solv-
ing an algebraic equation without actually using any numbers, just
combinations of [ M ], [ L] and [ T ].Doing this after making some alge-
braic manipulations on an equation is a handy way of checking for
mistakes! Similarly, if we are uncertain of the dimensions of a con-
stant or variable in an equation, we can solve for that constant on one
side of the equation and the dimensional grouping on the other side,
through the requirement of dimensional homogeneity, will apply to
the unknown.

Heuristic: Dimensional Homogeneity

In constructing or manipulating algebraic relationships, enforc-
ing and verifying dimensional homogeneity can yield insights
and catch errors.

Rules for algebraic manipulation of dimensions are straightfor-

 quantitative problem solving in natural resources 43

ward. Multiplying, dividing, or raising to a power any quantities of

any dimensions is permissible, and the resulting quantity has dimen-
sions that are the appropriate product, quotient, or power of the orig-
inal dimensions. Adding and subtracting may only be done between
quantities with identical dimensions (and later you should verify
that the units are also identical). Let’s consider an example where
we wish to find the dimensions of the entities in a simple equation
describing the accumulation over time of insects (measured in mass)
in a pit trap:
m = m0 + kt (4.1)

where we know m is mass and so has dimensions [ M] and t is time

and so has dimensions [ T ]. What are the dimensions of the other two Rearranging equations algebraically
variables? Well, assuming the equation is dimensionally homoge- with only dimensions or units rather
than numbers is a useful way to find
neous and knowing that the left-hand side has dimensions [ M], the the units or dimensions of unknown
right-hand side must work out to have dimensions of [ M] as well. variables.
Furthermore, since added quantities must have identical dimensions,
we know that each of the two terms added together on the right hand
side must have dimensions of [ M ]. So m0 is a mass, and k must be
something that, when multiplied by a time with dimensions [ T ],
gives a mass. Therefore, k must have dimensions of [ MT −1 ], or mass
per time.

4.6 Communicating Quantitative Information

We all construct our understanding of quantitative information a

little bit differently. I personally grasp quantitative patterns or rela-
tionships best if I see it in a graph. Others might have an easier time
reading or listening to a verbal description of patterns and relation-
ships. Still others will be more moved by equations or lists of num-
graphs
bers. We can understand and communicate quantitative information
in all of these ways – and maybe more (we can encode quantitative
information in sound, right?)! But to be well-rounded scientists and
managers, we must be able to create and interpret each form. The
diagram in Figure 4.1 illustrates these four ways of communicating
quantitative information. words
Here’s what each of the nodes of the triangle mean:

• Graphs, images, figures: any visual display of quantities or re- numbers equations
lationships between them. Single-variable statistical charts (e.g.,
Figure 4.1: Ways of communicating and
histograms) and two-dimensional graphs (i.e., x-y scatter-plots)
understanding quantitative information.
will be the most frequently encountered, but maps are another
example. Often the most efficient way to demonstrate patterns or
large volumes of data.
44 iowa state university

• Numbers, in lists or tables: for numerical information, the most

direct, precise, and unambiguous way of communicating quanti-
ties that aren’t too numerous (i.e., a short list).

• Equations, inequalities, or proportionalities: a formal and precise

way to state hypothesized, derived, or observed relationships
between quantities.

• Words, concepts: descriptions and interpretations, either standing

alone or to accompany another form of expression.

In technical reports, it is good practice to employ at least two or three

forms of quantitative expression, where one form will always be
words. Words are in the center of the triangle because they must be
used to link the other forms conceptually, and without them we can-
not claim to be understanding and communicating effectively. Fur-
thermore, as professionals, we cannot simply provide charts or tables
of data and expect them to speak for themselves. Part of the role of
80
scientists and managers is to interpret quantitative information and
make decisions or recommendations based on our interpretations.
60
# trout

40
4.6.1 Example: brook trout recruitment (teaser Problem 3.7)
20

0 The data from electrofishing studies of fish population and age-

pass 1 pass 2 pass 3 pass 4 structure can be quite simple. A typical set-up would begin with
Figure 4.2: Bar chart showing the blocking the channel upstream and downstream with fences or nets
number of brook trout removed from
a stream reach during each of four
to prevent fish from entering or leaving the study reach. Then fish-
consecutive electrofishing passes. eries technicians would slowly traverse the study reach with one
person applying the electrodes through the water and a second col-
lecting the stunned fish into a bucket or live well. Captured fish are
then measured, weighed, aged (if desired) and returned to the water.
In depletion surveys, the fish are returned upstream or downstream
pass # trout removed effort
pass 1 86 35 min of the blocking nets, so as to avoid immediate re-capture. Subse-
pass 2 51 29 min quent passes would operate similarly, and the presumption is that
pass 3 32 28 min
pass 4 9 31 min with each removal, the number of fish remaining in the study reach
is diminished. Figure 4.2 and Table 4.1 both show data from four
Table 4.1: Brook trout removal from consecutive passes through a small trout stream.
four electrofishing passes in one stream
reach, including information about
The bar graph (Figure 4.2) provides a very simple visual indication
the number of minutes (effort) elapsed of the change in the number of fish captured in each pass. Graphs
during each pass. like this can be extremely valuable for efficiently conveying trends
or relationships between quantities. However, they often don’t allow
readers or viewers to know precisely what the values of the shown
quantities are. Furthermore, graphs with too many different vari-
ables can become overly complex. If we wish to communicate precise
values of quantities, particularly where there are multiple different
 quantitative problem solving in natural resources 45

kinds of variables, data tables like Table 4.1 are perhaps the best op-
tion.
Given only the information presented already about the brook
trout electrofishing catch, communicating through equations is prob-
ably unwarranted. However, a textual narrative is essential for com-
municating the nature and meaning of the data presented in this
example. Suppose you were unfamiliar with electrofishing and de-
pletion methods for fish population assessment, would the bars in
Figure 4.2 and the numbers in Table 4.1 tell a clear story? The figure
captions and the paragraphs above are essential for drawing meaning
from the figure and table. This is why words are at the center of the
triangle in Figure 4.1. And this is also why it quantitative problem
solving is a writing-intensive endeavor. If we are unable to communi-
cate clearly about our ideas, strategies, results, and conclusions, most
of the effort is for naught.

Exercises

1. From our bulleted list of examples at the beginning of section 4.3,

what are the units and dimensions for each quantity? Write them
with square-bracket notation.

2. In population studies, it is recognized that the number of individ-

uals captured depends strongly upon how much time (effort) is
spent in active pursuit. As a result, a better variable to quote than
the number of trout caught is the number of trout per unit effort,
or catch-per-unit-effort (CPUE). If one minute is defined as the ba-
sic unit of effort for the data in Table 4.1, convert the data to a new
variable, catch (trout removed) per unit effort (minute). Plot the
result in a bar graph similar to Figure 4.2 and create a table with
this additional variable alongside the other three.

3. What are the dimensions of the new variable created in #2?

4. What kind of quantity is shown on the horizontal axis of Fig-

ure 4.2? How do you think this constrains appropriate ways to
visualize these data graphically?
5
Working with Numbers

Among the most fundamental operations we do with quantities is

arithmetic. We can encounter the need for arithmetic in any phase Arithmetic according to Wikipedia: a
of problem solving, from making a ballpark estimate in the Under- branch of mathematics that consists
of the study of numbers, especially
stand phase to computing and double-checking a final result in the the properties of the traditional opera-
Execute and Check phases. Once we have a solid grasp of the op- tions on them – addition, subtraction,
multiplication and division.
erations that are allowable and those that aren’t – for example, is it
OK to add or subtract quantities expressed in different units or on
different scales? – we may get down to business with performing
basic operations.
Most of us probably feel comfortable with most of these opera-
tions, at least when they concern simple numbers. However, it be-
comes easy to make errors or overlook important steps when we’re
dealing with extremely large or small numbers, or when unit con-
versions become necessary. One setting in which we often encounter
such difficulties is in working with proportions, including concentra-
tions, ratios, and percentages. Though quantities like these are often
conceptually simple, working with them and converting among ways
of expressing them can be challenging. This chapter highlights some
concepts and techniques for working with these sorts of unwieldy
numbers so that we can work confidently, avoid simple mistakes, and
even catch more complex ones.
We begin with a method for doing arithmetic that can be used to
simplify computations, or to approximate solutions when a back-of-
the-envelope computation is all you need. The method is particularly
powerful when computations involve very large or very small num-
bers. As such, it can be useful for making ballpark estimates in the
early stages of problem-solving. Our method makes strategic use
of scientific notation, which you’ve probably encountered in sec-
ondary science classes. The philosophical basis of scientific notation
also leads to the notion of order of magnitude, a concept that can
be useful for comparing quantities as well as for judging the the ap-
propriateness of estimates. Along the way, we’ll compare some ways
48 iowa state university

of expressing normalized quantities like concentrations and propor-

tions, and review the rules for arithmetic with exponents.

5.1 Scientific Notation

In high school chemistry, we learn that there are more than 602 sex-
1
6.022×1023
602 sextillion, or is Avo- tillion molecules in a mole of a chemical substance1 . But we don’t
gadro’s constant, the number of normally see Avogadro’s constant written as some number of sextil-
molecules in one mole of a chemical
substance lions, nor do we see it elaborated with all of the 24 digits necessary
to write it in integer form: it is difficult to keep track of all those dig-
its when writing them, and even more difficult to keep track when
you’re reading or comparing different numbers. Instead of writing
the entire number out, we use the shorthand of scientific notation,
where Avogadro’s constant looks more like 6.022×1023 . In general,
scientific notation has the form:

a = 10b

where a and b are sometimes called the mantissa and power, respec-
tively. So Avogadro’s constant has a mantissa of about 6.022 and a
power of 23, which is equivalent to saying that the complete quantity
2
A related issue is that of significant has 23 digits after the mantissa2 . Obviously this is a very large num-
digits. Scientific notation allows us ber. We can just as easily express very small numbers with scientific
to clearly specify how precise we are
claiming to be through the number of notation. An e. coli bacterium is roughly 2 µm (micrometers) long,
digits included in the mantissa: in this which is 2×10−6 m. Here, the power of −6 indicates not that it’s a
case, 4.
negative number (it would be absurd to say something has a negative
length, because length is a ratio scale!), but that it is smaller than 1
and that there should be 6 digits to the right of a decimal point if we
wished to express it as a decimal number. So we could express this
equivalently in a few ways:

2µm = 0.000002m = 2 × 10−6 m.

Note that these equalities both amount to unit conversions, but the
second equality is specifically a conversion to scientific notation.
Negative exponents indicate numbers smaller than 1, and there
are occasions where it can be helpful at times to can write these as
fractions. When we have a quantity expressed in scientific notation
with a negative exponent 10−b , that is equivalent to the same quan-
tity divided by 10b . Therefore, another way to express the length of e.
coli is:
1
2 × 10−6 m = 2 × 6 m.
10
So dividing by 106 is the same as multiplying by 10−6 . Notice here
In the standard order of operations, that the order of operations is important. By convention, exponents
parentheses take precedence, then
exponents, then multiplication or
division, and finally addition and
subtraction.
 quantitative problem solving in natural resources 49

take precedence over multiplication, division, addition and subtrac-

tion. So we don’t divide by the mantissa (2) when we express this
quantity in fractional terms because the only thing that is raised to
the exponent is the base, in this case 10. We could, however, move
the mantissa to the denominator with its 106 by taking its reciprocal,
right? That’s another way of invoking the old grade-school rule: di-
viding by a number is the same as multiplying by it’s reciprocal. In
this case, we’d end up with an equivalent value for the length of e.
coli that looks like
1 1
2 × 10−6 m = m= m.
0.5 × 106 5.0 × 105
Notice that in the last step we’ve borrowed a “ten” from the power
to make the mantissa greater than 1: this is by convention3 . A gen- 3
Quantities expressed in scientific
eral rule for expressing a quantity in scientific notation is to have notation should have one nonzero digit
to the left of the decimal point.
one nonzero digit before the decimal point in the mantissa, and as
many significant figures as appropriate for the problem to the right
of the decimal. So we could express the e. coli length as 0.2×10−5 m
or 200×10−8 m, but in most cases that is bad form. We shall see be-
low, however, there are times when doing arithmetic by hand can be
simplified by temporarily expressing quantities in such an unconven-
tional way.
A useful concept in working with really large or really small num-
bers is the order of magnitude of a quantity. In obtaining a ballpark The order of magnitude of a quantity
estimate of a quantity or in computing something using only very is essentially the value of the exponent
when expressed in scientific notation.
rough approximations for the input values, it may be unnecessary
or inappropriate to worry about being off by a factor of 2 or so. We
might be satisfied knowing that the result is “a few thousand” or “a
coupe hundredths”. If we’re using scientific notation, this is equiva-
lent to ignoring the mantissa and just citing the base and power. So
instead of saying that an e. coli is 2×10−6 m, we can say it is on the
order of 10−6 m long. This kind of reasoning is particularly useful in
comparing multiple quantities. A grain of coarse sand, for example,
is on the order of 10−3 m in diameter, so it is three orders of mag-
nitude larger (−3 is three more than −6) than an e. coli bacterium.
Once we wrap our minds around what that means (three orders of
magnitude is a factor of 103 , or a thousand!), comparisons can be
enlightening in assigning quantitative “importance” to different vari-
ables in an equation.

The fact is, in normal communication about the length of

e. coli, we’d probably stick with 2 µm as a clear way to express it in
written text. Most of the alternative ways above are more clumsy
in writing, and certainly the last few equivalent expressions above
are not intuitive (we only went there to demonstrate the technique!).
50 iowa state university

However, in comparisons with other qantities or when performing

computations with other quantities that are expressed in different
units, it is usually smart to convert all quantities to a uniform system
of units, like the systeme internationale, or SI.

5.2 Normalized quantities

In the sciences, normalization of quantities often refers to the process

of dividing some scaled quantity by a standard, total, or reference
value of the same quantity. Consider some schemes form normal-
ization that you are already very familiar with. A percentage is a
normalized quantity, determined by dividing some number that
represents a subset of a larger collection by the total number in the
collection and then multiplying by 100%. For example, suppose we
have tested 360 white-tail deer carcasses (from road-kill and hunter
harvest) for chronic wasting disease (CWD) and find that 83 are posi-
tive. Given this data, we can all agree that the percent of the sampled
population infected with CWD is:

83
× 100% = 23.0556% (5.1)
360
En route to computing this, we created the ratio 83 to 360, which is
around 0.23 if you simplify it with your calculator. As with many
such ratios, we can choose from a variety of different but equivalent
ways of expressing this quantity. We could just express it as the ratio
4
this is the way we usually express a of two whole numbers like 83:3604 , or as the fraction:
map scale, like 1:24,000. See Part III of
this book for more on that issue. 83
. (5.2)
360
Or as we’ve already seen, it is simple to write it as a decimal number
(0.230556). But since we encounter percentages frequently, we may
more readily appreciate it expressed as a percentage. For the present
purposes, we could describe a percentage as “parts per hundred”,
since it is just the same ratio scaled to an arbitrary reference value of
100. In other words, for every hundred deer in the sample, about 23
have CWD. Expressing a quantity in “per mil” is closely analogous,
except instead of multiplying by the factor 100% we’d multiply by
5
Although not common in many 1000h(that’s the per mil symbol)5 . In this case, we’d end up saying
disciplines, isotope concentrations that about 83/360 ×1000h= 231h(or 231 per thousand deer) are in-
are often expressed in h, where the
reference value is the isotopic ratio of a fected. To make this even more absurd, we could express the same
standard substance. information just as easily as parts per million (ppm) or parts per bil-
lion (ppb) following a similar tactic. Each of these ways of expressing
a normalized quantity is arithmetically equivalent, but implies a dif-
ferent realm of precision about the quantity of interest and the scope
 quantitative problem solving in natural resources 51

of its possible values. We’d likely never talk about deer in parts per
million, but we might talk about lead concentrations that way!
Other types of normalized quantities in science include frequen-
cies, concentrations, and probabilities, to name a few. The quantities
may be expressed somewhat differently, but in most cases there is
a comparison being made between values of the same dimensions
(and often the same units!). Indeed this is sometimes a simplifying
strategy: when you normalize a quantity to a standard of the same
units, details about the specific units by which the quantities were
measured can be discarded. Often this is a good thing. For exam-
ple, when we use a map that is scaled at, for example, 1:24,000, we
are not told what units that ratio was constructed with, because it
doesn’t matter! If you use a ruler to find that the map distance be-
tween two features on the map is 2 inches, that distance in the real
world is equal to 2 × 24, 000 = 48, 000 in. It doesn’t matter whether
your ruler is ruled in inches, centimeters, furlongs or rods, the quan-
tity you measure on the map only needs to be multiplied by the scale
factor (24,000) to find the true distance! As we’ve seen, however, ne-
glecting the specific units used to derive a normalized quantity can
also be the cause of some confusion (is the concentration of one sub-
stance mixed with another computed on the basis of their masses,
volumes, or something else?). It becomes a good thing if the proce-
dural statement for the quantity is either made clear or is known by
convention.
How do we use a normalized quantity to our advantage? Suppose
I extrapolate from our sample of CWD in deer carcasses to predict
that 23% of the deer in the entire county are infected with CWD. If
we take for granted that my science is good, all we need to know to
find out the number of CWD-infected deer in the county is the total
number of deer in the county, Ndeer . Once we recognize that the ratio
of infected deer to total deer is 0.23 (23% of the total population of
100%), we need only perform a simple multiplication:

NCWD 23%
= = 0.23 (5.3)
Ndeer 100%

NCWD = 0.23Ndeer (5.4)

Thus, the benefit of expressing the number of infected deer as a nor-

malized quantity (assuming our 23% assertion is accurate) is its gen-
erality. We can write a simple relationship like Equation 5.4 and, as
long as the relationship remains valid, apply it on any relevant scale6 . 6
Recognizing how far one can safely
The process of re-scaling a ratio (or other normalized quantity) scale up from a representative sample is
a rich, but complex issue.
is sometimes called proportional reasoning, and is one of the key
strategic processes in probability, and as we’ll see in the next chapter,
52 iowa state university

it is the foundation of much of trigonometry. The construction of

abstract triangles in the service of problem-solving is usually a means
of comparing the ratios of two lengths or distances.
There are some oddball normalized quantities in science that are
frequently expressed in inhomogeneous units, either as a conse-
quence of their very high or low intrinsic magnitudes or due to the
procedure used to measure them. One example is slope in the con-
text of river channels or footpaths, which are often less than 1%. Be-
cause typical channel slopes are so small, it is common to see slopes
expressed in units of “feet per mile” or “meters per kilometer”. They
are still normalized quantities, but the inhomogeneous units must be
stated explicitly. Similarly, concentrations of solutes or suspensions
are sometimes expressed in units like mg/L (milligrams per liter),
where the dimensions are a weight per volume. This is convenient
because of the relative simplicity of weighing a solid component
added to (or isolated from) a volume of liquid. On the other hand,
concentrations of substances like dilute hydrochloric acid (HCl; often
used in soil chemistry) are are often described as percentages: 5%
HCl usually means a mixture in which 5% of the total volume is pure
HCl and the remaining (100-5)% = 95% is pure water. Again, this
makes sense because when mixed, both components are liquids and
their volumes are simple to measure.

5.2.1 Example: maximized effluent, (Problem 3.6)

Phosphorus (P) is a limiting nutrient in many freshwater ecosys-

7
A great review of nutrients in ter- tems7 . That means that primary productivity is limited by the avail-
restrial ecosystems can be found in ability of P, and that excessive loads of P from fertilizer runoff or
Weather, K.C., D.L. Strayer, and G.E.
Likens, 2013. Fundamentals of Ecosystem municipal and industrial wastes can promote excessive productivity
Science, Academic Press, Elsevier Inc. and eutrophication. Thus, we are often seeking ways to reduce the
inputs of P into surface waters.
P concentrations in water are often expressed in mg/l, so they are
among those normalized quantities that are not dimensionless. A
given concentration in mg/l can be visualized as the mass of solute
that could be hypothetically extracted from a volume of water, if we
somehow had a perfect P-filter. No such filter exists, so not only do
8
In practice, measuring dissolved we need a different way of measuring P8 , we need more clever ways
P is most efficiently done using a to extract P from water if it does get in there.
“colorimetric” method wherein a
reagent is introduced to a dilute P The TMDL selected for P in surface water bodies depends on the
solution, resulting in the development designated uses (drinking water? swimming?) of the water bodies in
of a blue color in proportion to the P
concentration. question, but are often on the order of 0.1 mg/l. It’s worth remem-
bering that this means that for every one liter of water, we should have
no more than 0.1 mg of P. So if we happen to take a two-liter sam-
ple of water in a water body under this TMDL, we should find no
 quantitative problem solving in natural resources 53

more than 0.2 mg P in that sample, as that (0.2 mg divided by 2l)

corresponds to a concentration of 0.1 mg/l.

5.3 Tricks with scientific notation

As we’ve already discussed, simple order-of-magnitude computations

can be very informative, particularly in the early phases of problem-
solving. This is an occasion when scientific notation can really be
useful! To deftly manipulate expressions with scientific notation, it is Heuristic
helpful to remember some key rules for working with exponents. Get a ballpark or order-of-magnitude
estimate by hand using scientific
notation
Rules for Manipulating Exponents
1
x0 = 1 x1 = x x −1 =
x
xa
x a × x = x ( a +1) = x ( a −1)
x
xa
x a x b = x ( a+b) = x ( a−b)
xb
1 1
x−a = x a = −a
xa x
( x a )b = x ( a×b)

When confronted with problems where multiplication or division

of very large or very small numbers might be involved, we can set
the problem up in scientific notation to make things simpler. Con-
sider the simple example of determining how many milliliters (ml)
are in a cubic meter of water. One thing that is useful to know is that
a ml is the equivalent of a cubic centimeter (cm3 ). And we also know Numerical Benchmarks: Volume
that there are 100 (= 102 ) cm in a linear meter (m). So how do we de- 1 ml = 1 cm3
1 m3 = 1000 l
termine the number of cm3 in a m3 ? Recall from earlier that if we are
converting between, for example, one set of squared units to another
set of squared units, we need to square the conversion factor for the
linear units too! So for this problem, since there are 102 cm in every
m:
1 m3 = (102 )3 cm3 (5.5)

Using one of the above rules for exponents to modify the right-hand
side of this relationship, we can find that:

1 m3 = 10(2×3) cm3 (5.6)

1 m3 = 106 cm3 (5.7)

54 iowa state university

and we have our result. There are 106 cm3 , and therefore 106 ml
in a cubic meter! It would be just as easy to look up the conver-
sion online, but the same basic approach can be readily applied to
more complex problems with murkier solutions. In the next section
we’ll consider a more challenging and engaging example that can be
worked out with a similar strategy.

5.3.1 Example: Mercury in fish

9
Find more information about the EPA The RAFT program9 (Regional Ambient Fish Tissue) is an EPA effort
RAFT program by searching EPA raft to monitor concentrations of several harmful toxic substances in fish
on the web.
in the state of Iowa. This problem concerns the (slightly idealized
and modified) values of mercury (chemical symbol Hg) detected in
smallmounth bass sampled from two locations in Iowa. Samples of
fish tissue were obtained as "plugs", taken from live fish in a manner
Hg concentration advisory similar to a biopsy. Typical plug samples weigh 50 mg. The criteria
< 0.3 ppm no restrictions
> 0.3 to < 1.0 ppm 1 meal/week for issuing fish consumption advisories are shown in the table below.
≥ 1.0 ppm do not eat Plugs from smallmouth bass in Lake Wapello, IA contained on aver-
age 0.06 µg of Hg, while plugs from smallies in the Maquoketa River
contained 0.01 µg Hg. Should there be consumption advisories for
either waterbody?

A simple solution method

A useful first step is to identify the desired result. We’d like to find
a Hg concentration in each fish in the same units that the advisory
guidelines use: parts per million or ppm. This is a normalized and
dimensionless, derived quantity. A second helpful step is therefore
to express the key data in uniform units so that we can normalize
Heuristic: them in dimensionless form. Our Hg measurements are in µg, which
Convert to uniform system of units is 10−6 g, while our plug mass is in mg, which is 10−3 g. It doesn’t
really matter whether we convert everything to grams or something
else, but grams is straightforward. So now me construct the ratio
that expresses how much mercury there is, by mass, in our fish tissue
sample (using Lake Wapello values as an example):
0.06 × 10−6 g
(5.8)
50 × 10−3 g
Simplify this by cancelling units and expressing each quantity in
proper scientific notation:
6.0 × 10−8
(5.9)
5.0 × 10−2
Using rules for division in exponents with a common base, we can
simplify this:
6.0
× 10(−8)−(−2) (5.10)
5.0
 quantitative problem solving in natural resources 55

The exponent therefore becomes −6, which you recall is the base for
a “parts per million” ratio. We can simplify the fraction 6/5 either
directly on a calculator, in our heads10 , or by multiplying both nu- 10
One great benefit of using scientific
merator and denominator by two (= 12/10) and dividing by 10 to notation is that computations can be
approximated easily by hand!
get 1.2:
6/5 × 10−6 = 1.2 ppm (5.11)

So the result for Wapello is 1.2 ppm, which exceeds safe limits for
consumption. For the Maquoketa River, the Hg concentraion is only
0.2, so it is safe to eat and no advisory need be issued.

5.3.2 Example: forest fire losses (Problem 3.5)

Let’s use some of the above techniques and strategies to make some
ballpark estimates about the value of timber that could potentially be
lost in a forest fire, following the teaser problem in Section 3.5. This
could give us at least a starting point for imagining where the curve
NVC starts from on the left-hand side of Figure 3.2. Since no specific
information is given about the size of the property, we need to make
and explicitly state an assumption. Let’s suppose for now that the Heuristic: Not enough information
property has an area of 1000 hectares, since that number is both given? Make and state explicitly a
reasonable and potentially-scalable
reasonable for a single-ownership land parcel (this would be a bit less assumption. If appropriate, choose
than 4 square miles) and is easily scaled. Let’s also assume that this values that can easily be scaled, like 1
or 10.
forest in the absence of any fuel reduction effort is overstocked, with
perhaps 30 m2 ha−1 of basal area11 . Using timber cruising charts, this 11
basal area, usually given in ft2 ac−1
basal area would yield about 30,000 board feet per hectare12 . (square feet per acre) or m2 ha−1
(square meters per hectare), provides
To get a ballpark estimate of the value of this timber then, we need a quick glimpse of the amount of
to find the going price per board-foot of our timber and then scale standing timber on an area of land.
this up with the timber volume and property area. A reasonable
12
One board-foot is equal to about
0.00236 m3 of wood.
guess for the price for softwood saw-logs would be 0.20 US dollars
(USD) per board foot13 . So our computation becomes 13
A web search for “sawlog prices” can
give you some idea of how this varies
by place and time.
NVC (0) = 1000 ha × 30000 BF/ha × 0.20 USD/BF.

We can do this computation relatively quickly in a calculator, but

there is a risk of typing in the wrong number of zeros and making an
important error. However, if we convert these quantities to scientific
notation and rewrite the equation we can do the math in our heads.
The parcel area is 1 × 103 hectares, the wood volume is 3 × 104 board
feet per hectare, and the value is 2 × 10−1 USD per board foot. So we
may re-write the computation as

NVC (0) = 1 × 103 ha × 3 × 104 BF/ha × 2 × 10−1 USD/BF.

Since all these quantities are multiplied together, we can rearrange

(by the commutative principle for multiplication) and group the man-
56 iowa state university

tissas together, put the powers together, and put the units together.

NVC (0) = 1 × 3 × 2 × 103 × 104 × 10−1 ha BF/ha USD/BF.

Multiplying the mantissas through we get 6, and using the rules for
manipulating exponents (see the next section!) the exponents are
added together (3 + 4 + −1) = 6. Canceling units, we see that USD is
the only remaining unit. So our ballpark solution is that the value of
the standing timber in this 1000 ha parcel is 6 × 106 USD, or about $6
million.

Exercises

1. The Hg concentrations measured in the RAFT program problem

were taken from “keeper” size smallmouth bass, roughly 35 cm
long. A few scattered measurements from larger and smaller bass
indicated that there was some systematic relationship between Hg
concentrations and fish size at each site, but not across sites. What
systematic relationships would you predict to be present in fish of
different sizes? What quantities might be relevant to this problem?
Formulate a testable hypothesis for expected systematic variation
in smallmouth bass tissue Hg concentration as a function of fish
size.

14
Based on the article ”Prevalence of 2. 14 Leucism is partial albinism, manifested in penguins as a lack
leucism in Pygocelid penguins of the of (or substantial reduction in) pigment in plumage. A study of
Antarctic Peninsula” by Forrest and
Naveen, Waterbirds 23(2): 283-285, 2000. three species of penguin (Adélie, Gentoo and Chinstrap) in the
Antarcic peninsula sought to identify the prevalence of leucism
in these different species. The paper cited in the margin provides
species prevalence count the following information derived from detailed counts of penguin
Adélie 1:114,000 1,144,000
Chinstrap 1:146,000 293,800 breeding colonies made during the years 1994-1997:
Gentoo 1:20,000 41,550
Perform the following manipulations of the prevalence data for
each species:

(a) Express the prevalence as a fraction (a ratio of whole num-

bers).
(b) Convert the prevalence to a decimal number.
(c) Convert the decimal number to scientific notation.
(d) Express the prevalence as a percentage of the population.
(e) Express the prevalence in parts per million (ppm).
(f) Determine the number of leucistic penguins in each count.
 quantitative problem solving in natural resources 57

3. From our discussion of standing timber values (Section 5.3.2), how

would the result be different if we learned that the land parcel was
385 hectares instead of 1000?
6
Reasoning with Data

This chapter summarizes some of the key concepts and relationships

of single-variable statistics that we might find useful for characteriz-
ing measurements, particularly when we have measured a quantity
at multiple times, or we’ve measured many individual members of
a population or collection. This is not intended to be an exhaustive
introduction to statistics, and does not in any way substitute for a
proper statistics course. It does, however, point to some connections
that we can make between the measurement and characterization of
data and the scientific description of nature that we sometimes seek.

6.1 Measurement and Sampling

In the natural sciences we often need to estimate or measure a quan-

tity or set of quantities that is too large, too numerous, or too com-
plex to characterize completely in an efficient way. We can instead
characterize it approximately with a representative sample. A represen-
tative sample is a small subset of the whole that is measured in order
to characterize the whole.
Consider an example. In small headwater streams, many aspects
of biotic health are linked with the size of the substrate – the sand,
pebbles or boulders that compose the streambed. But it is impractical
to measure all the gajillions of particles scattered over the entire bed.
Instead, we attempt to get a smaller but representative sample of
the bed material. This may be done in a number of different ways,
but two common methods are: 1) to take one or more buckets full of Figure 6.1: Cobbles on the bed of the
Cub River, Idaho.
sediment from the streambed and do a detailed particle-size analysis
in a laboratory; and 2) measure the size of 100 randomly selected
particles from the bed. Both methods obtain a sample, but each may
represent the true streambed in a different way. The bucket method
requires us to choose sample sites on the streambed. Our choices
might be biased toward those places where sampling might be easier,
the bed more visible, or the water shallower. In this case, our results
60 iowa state university

might not be representative of the streambed as a whole.

The “pebble count” method, on the other hand, is intended to pro-
1
This method is sometimes called the duce a more random sample of the streambed1 . A person wading
“Wolman pebble count” method for in the stream steps diagonally across the channel, and at each step
Reds Wolman, the scientist who first
described and popularized it. places her index finger on the streambed immediately in front of the
toe of her boot. The diameter of the particle that her finger touches
first is measured, and then she repeats the process, zig-zagging
across the channel until she has measured 100 (or some larger pre-
determined number) particles. In principle, this random sample is
more representative of the streambed, particularly as the number of
particles in the sample is increased. Of course, increasing the number
of particles in the sample increases the time and effort used, but with
diminishing returns for improving the accuracy of the sample.
Hypothetically-speaking, an alternative pebble-count method
could be to stretch a tape measure across the stream and measure the
particle size at regular intervals, say every half meter. We can call this
strategy the “point count” method. This alternative is appealing since
it ensures that samples are distributed evenly across the channel and
that samples are not clustered in space. However, it is conceivable
2
Systematic sampling is sometimes that such systematic sampling could lead to a systematic bias2 . If for
an easier, more straight-forward ap- example the streambed had clusters or patterns of particles in it that
proach to sampling. However, if the
setting within which sampling is tak- had a wavelength of 0.5 m, you could be inadvertently sampling only
ing place might have some systematic a certain part of the top of each dune, which might skew your results
structure, systematic sampling could
inadvertently bias the sample. toward particle sizes that are concentrated on dune crests. Thus, a
random sample is usually preferable as it is less susceptible to this
kind of systematic bias.

Quantities derived from a random sample are unrelated to

one another in the same way that the size of one grain measured
during a pebble count has no influence on the size of the next one.
Part of our sequence of data might look like this:
12, 2, 5, 26, 4, 28, 19, 29, 3, 15, 31, 19, 24, 27, 7, 22, 28, 33, 21, 28, 13, 15,
25, 10, 14, 13, 16, 18, 33, 5

The random nature of this set of data allows us to use some of the
familiar ways of describing our data, while boosting our confidence
that we are also properly characterizing the larger system that we are
sampling.

6.1.1 Example: mark-recapture

A frequent concern of the wildlife ecologist is the abundance and
health of a particular species of interest. Ideally, we could count and
assess the health of every individual in a population, but that is usu-
ally not practical - heck, we have a tough enough time counting and
 quantitative problem solving in natural resources 61

assessing the health of all the humans in a small town! Instead of try-
ing to track down every individual though, we can do a decent job by
simply taking a random sample from the population and performing
the desired analysis on that random sample. As we have seen, if we
are sufficiently careful about avoiding bias in our sampling, we can
be reasonably confident that our sample will tell us something use-
ful (and not misleading) about the larger population that the sample
came from.
If our concern is mainly with the population of a target species in
a certain area, we can use a method called mark-recapture, or capture-
recapture. The basic premise is simple: we capture some number of
individuals in a population at one time, band, tag or mark them
in such a way that they can be recognized later as individuals that
were previously captured, then release them. Some time later, after
these individuals have dispersed into the population as a whole,
we capture another set. The proportion of the individuals in the
second capture who are marked should, in theory, be the same as the
proportion of the whole population that we marked to begin with. If
the number of individuals we marked the first time around is N1 , the
number we captured the second time around is N2 , and the number
in the second group that bore marks from the first capture is M, the
population P may be estimated most simply as:
N1 N2
P= (6.1)
M
This comes from the assumption that our sample each time is ran-
dom, and that the marked individuals have exactly the same like-
lihood of being in the second capture as they did in the first: 1/P.
Therefore, if we sampled and marked a fraction N1 /P the first time
around and sample N2 the second time around, then we should ex-
pect a fraction M/N2 of them to be marked.
Of course this whole plan can be foiled if some key assumptions
are not met. For example, we need the population to be “closed” –
that is, individuals do not enter and leave the population such that
our sample is not coming from the same set of individuals each time.
Problems could also ensue if our “random” sample isn’t random, if
somehow the process of marking individuals either harmed them
or made their likelihood of re-capture more or less likely, or if the
time we allowed for them to re-mix with their population was not
appropriate. On the last point, you can imagine that if we recapture
tortoises 10 minutes after releasing them from their first capture, our
second sample will not be very random. On the other hand, if we
recapture marked fish 20 years after they were first marked, many of
them may have died and been replaced by their offspring, and thus
our assumption of a “closed” population is violated. So in planning
62 iowa state university

a mark-recapture study, space and timescales need to be taken into

account.
It is worth noting that the method described here is about the most
stripped down version of mark-recapture. There are many modifi-
cations to the method and the equation used to compute population
that either account for immigration/emigration, multiple recaptures,
some possible re-recaptures, etc. There are also related methods us-
ing tagging and marking that can be used to explore the dispersal of
individuals, migration routes and alot more!

6.2 Describing measurements

Measurements, or “data”, can inform and influence much of a re-

source manager’s work objectives, since they convey information
about the systems of interest. Sometimes the data speak for them-
selves: raw numbers are sufficiently clear and compelling that noth-
ing more needs to be done to let the data speak. More commonly,
however, the data need to be summarized and characterized through
one or more processes of data processing and data reduction. Pro-
cessing might simply refer to a routine set of algorithms applied to
raw data to make it satisfy the objectives of the project or problem.
Data reduction usually summarizes a large set of data with a smaller
set of descritptive statistics. For a set of measurements of a simple
quantity, for example, we might wish to know:

Things we often want to know about our data

1. what is a typical observation?

2. how diverse are the data?

3. how should these properties of the data be characterized for

different types of quantities?

The first point suggests the use of our measures of central ten-
dency: mean, median and mode. The second goal relates to measures
of spread or dispersion in the data. For example, how close are most
values in the data set to the mean?

6.3 Central tendency

The central tendency of a data set is a characteristic central value

that may be the mean, median, or mode. Which of these measures
of central tendency best characterizes the data set depends on the
nature of the data and what we wish to characterize about it.
 quantitative problem solving in natural resources 63

Most of us are already familiar with the concept of a mean, or av-

erage value of a set of numbers. We normally just add together all of
the observed values and divide by the number of values to get the
mean. Actually, this is the arithmetic mean, and there are many alter-
native ways of computing different kinds of means that are useful in
particular circumstances, but we won’t worry about these now. For
our purposes, the arithmetic mean is the mean we mean when we say
mean or average. It would be mean to say otherwise.
Before continuing, lets briefly discuss the different kinds of nota-
tion what we might use when talking about data. To define some-
thing like the mean with an equation, we’d like to make the defi-
nition as general as possible, i.e., applicable to all cases rather than
just one. So we need notation that, for example, does not specify the
number of data points in the data set but allows that to vary. If we
want to find the mean (call it x̄) of a set of 6 data points (x1 , x2 , and
so on), one correct formula might look like this:
x1 + x2 + x3 + x4 + x5 + x6
x̄ = (6.2)
6
and of course this is correct. But we can’t use the same formula for a
dataset that has 7 or 8 values, or anything other than 6 values. Fur-
thermore, it is not very convenient to have to write out each term in
the numerator if the data set is really large. So we need a shorthand
that is both brief and not specific to a certain number of data points.
One approach is to write:
x1 + x2 + ... + xn
x̄ = (6.3)
n
where we understand that n is the number of observations in the
data set. The ellipsis in the numerator denotes all the missing values
between x2 and xn , the last value to be included in the average. Using
this type of equation to define the mean is much more general than
the first example, and is more compact as long as there are 4 or more
values to be averaged.
One additional way you might see the mean defined is using so-
called “sigma notation”3 , where it looks like this: 3
This symbol is a handy shorthand
for the process of adding a bunch of
n quantities together, but also serves the
1
x̄ =
n ∑ xi (6.4) purpose of scaring many poor students
i =1 away. Once you realize that it’s just an
abbreviation for listing all the the terms
where the big Σ is the summation symbol. If you’ve never encoun- to be added (x1 + x2 + ...) and some of
tered this before, here’s how to interpret it: the “summand”, the stuff the rules for doing so, it becomes a tad
less fearsome.
after the Σ, is to be interpreted as a list of values (in this case xi ) that
need to be added together, and i starts at 1 and increases until you
get to n. You can see the rules for what i means by looking at the
text below and above the Σ. Below where it says i = 1 that means
64 iowa state university

that i begins with a value of 1 and increases with each added term
until i = n, which is the last term. So in the end, you can interpret
this to have a meaning identical to the equivalent expressions above,
but in some cases this notation can be more compact and explicit. It
also looks fancier and more intimidating, so people will sometimes
use this notation to scare you off, even though it gives you the same
result as the second equation above.

6.3.1 Mean versus Median

For some data sets, the mean can be a misleading way to describe the
central tendency. If your creel after a day of fishing includes 5 half-
pound crappies, a 3/4-pound walleye, 4 one-pound smallmouths and
one 16-pound muskie, it would be correct but misleading to say that
the average size of the fish you caught was 2.1 pounds. The distribu-
species weight (lbs.)
crappie 0.5 tion of weights includes one distant outlier, the muskie, that greatly
crappie 0.5 distorts the mean, but all of the other fish you caught weighed one
crappie 0.5
crappie 0.5
pound or less. We might say in this case that the mean is sensitive to
crappie 0.5 outliers.
walleye 0.75 The median is an alternative measure of central tendency that is
smallmouth 1.0
smallmouth 1.0 not sensitive to outliers. It is simply the value for which half the ob-
smallmouth 1.0 servations are greater and half are smaller. From your fishing catch,
smallmouth 1.0
muskie 16
the 0.75 pound walleye represents the median value, since 5 fish (the
mean 2.1 crappies) were smaller and 5 fish (the smallies and the muskie) were
median 0.75 larger. The median may also be thought of as the middle value in
a sorted list of values, although there is really only a distinct mid-
Table 6.1: A decent day’s catch on the
lake. dle value when you have an odd number of observations. In the
event that you’ve got an even number of observations, the median is
halfway between the two middle observations.

6.3.2 Mode
The mode is the value or range of values that occurs most frequently
in a data set. Since you caught 5 half-pound fish and fewer of every
other weight value in the dataset, the mode of this distribution is
0.5 pounds. Now if the weights we’ve reported above are actually
rounded from true measured weights that differ slightly, this defini-
tion becomes less satisfactory. For example, suppose the half-pound
crappies actually weighed 0.46, 0.49, 0.5, 0.55 and 0.61 pounds. None
of these are actually the same value, so can we say that this is still a
mode? Indeed we can if we choose to discretize or bin these data. We
might say that our fish weights fall into bins that range from 0.375
to 0.625, 0.625 to 0.875, 0.875 to 1.125, and so on. In this case, since
all of our crappies fall in the range 0.375 to 0.625 (which is 5 ± 1/8
lbs), this size range remains the mode of the data set. We can see this
 quantitative problem solving in natural resources 65

visually in a histogram, which is just a bar-chart showing how often

measurements fall within each bin in a range (Figure 6.2).
It is permissible to identify multiple modes in a data set if it im-
proves the description. The first mode is the data bin that appears
most frequently, but second and third and additional modes can be
used as well. A second mode in our fish sample is in the 1-pound
bin, which included 4 smallmouth bass. It is particularly useful in
the case of multimodal data sets to report the modes because the
multimodal nature of the data set cannot be represented by the
mean or the median. In fact, if you were only presented with the Figure 6.2: A histogram showing
the frequency of observations of fish
list of weights, you might still have a hunch that there were multiple weights. The height of each bar corre-
species or multiple age-classes present in the creel due to the multi- sponds to the number of fish in each
modal weights. of the weight bins along the horizontal
axis. Values are scrunched to the edges
In practice, reporting all of these measures of central tendency because the large dispersion of data.
may deliver the most complete picture of data, but as we’ve seen In some cases, multi-modal data can be
each is particularly useful in some cases and can be misleading in suggestive of a mixed sample; that is,
there is more than just one type of thing
others. That said, we can actually infer additional properties of the or from more than just one source in
dataset by noting, for example, the difference between the mean and the sample.
median.

6.4 Spread

As mentioned previously, one way to quantify dispersion of a data

set is to find the difference between any given observation and the
expected value or sample mean. If we write this:

xi − x̄, (6.5)

we can call each such difference a residual. A could be used to de-

scribe the relationship between individual data points and the sample
mean, but doesn’t by itself characterize the spread of the entire data
set. But what if we add together all of these residuals and divide
by the number of data points? Well, this should just give us zero,
according to the definition of the mean! But suppose instead that
we squared the residuals before adding them together. The formula
would look like:
1 n
n i∑
( xi − x̄ )2 (6.6)
=1
This expression is defined as the variance and is strangely denoted
by σ2 , but you’ll see why in a minute. Squaring the residuals made
most of them larger and made negative residuals positive. It also ac-
centuated those outlier data points that were farther from the mean.
Now if we take the square root of the variance, we’re left with a fi-
nite positive value that very well represents how far data typically
are from the mean: the standard deviation of the sample, or σ4 . The 4
If you keep track of the units of these
different measures of spread, you’ll
notice that the standard deviation
should have the same units that the
original data, xi does.
66 iowa state university

formal definition of standard deviation looks like this:

s
1 n
n i∑
σ= ( xi − x̄ )2 (6.7)
=1

The gives us a good sense for how far from the mean a typical mea-
surement lies. We can now characterize a sample as having a mean
value of x̄ and standard deviation of σ, or saying that typical values
are x̄ ± σ. But in reality, if we computed x̄ and σ, the bounds set by
x̄ − σ and x̄ + σ only contain about 68% of the data points. If we want
to include more of the data, we could use two standard deviations
above and below the mean, in which case we’ve bounded more than
95% of the data.

6.5 Error & Uncertainty

One piece of information we have thus far omitted from our list of
properties that fully define a quantity’s value is uncertainty. This is
particularly important when we are quantifying something that has
been measured directly or derived from measurements. Thus, to even
more completely define the value of a measured quantity, we should
include some estimate of the uncertainty associated with the number
assigned to it. This will often look like:

x = xbest ± δx, (6.8)

where x is the thing we are trying to quantify, xbest is our best guess
of its value, and δx is our estimate of the uncertainty. Though it will
depend on the quantity in question, our best estimate will often be
the result of a single measurement or – better yet – the mean of a
The preferred value xbest for a quantity number of repeated measurements.
of interest will often be the mean of
repeated measurements of that quantity.
6.5.1 Uncertainty in measured quantities
All measurements are subject to some degree of uncertainty, arising
from the limited resolution of the instrument or scale used to make
the measurement, or from random or systematic errors resulting
from the method or circumstances of measurement. Let’s consider an
example:
Suppose two fisheries biologists each measured the lengths of
ten of the brook trout captured during the electrofishing traverse
from Problem 3.7. Both used boards with identical scales printed
on them, graduated to half of a centimeter. They then plan to put
their measurements together to get a data set of 20 fish. One of them
was trained to pinch together the tail fins to make this measurement,
while the other was not. In addition, because they wished not to
 quantitative problem solving in natural resources 67

harm the fish, they made their measurements quickly, even if the fish
flopped and wiggled during the measurement. What are the poten-
tial sources of error and how big are they relative to one another?
For starters, implicit in the graduations on this board is that the
user cannot confidently read any better than half-centimeters off the
scale. He or she can, however, visually interpolate between two adja-
cent graduations to improve precision (see below). However, this step
is inherently subjective and limits the certainty of the measurement.
We might call this instrumental error because its magnitude is set by
the instrument or device use to make the measurement. One way to
reduce this source of error is to use a more finely-graduated scale. Instrumental error is fixed by the
A second source of error arises from the hasty measurements and resolution of the device used to make
a measurement, and can usually only
the fact that the fish were not necessarily cooperative. Perhaps the be reduced by using a more precise
mouth was sometimes not pressed up all the way against the stop, or instrument.
the fish wasn’t well aligned with the scale. Some lengths may have
been too large or small as a result, yielding a source of error that was
essentially random. Indeed, we can call this random error since its
sign and magnitude are largely unrelated from one measurement
to the next. Reducing this source of error in this case would require Random measurement errors may be
either more careful and deliberate effort at aligning and immobilizing mitigated by repeating measurements.

the fish, or making multiple measurements of the same fish. Both

of these solutions could endanger the fish and may therefore not be
desirable.
A third source of error is associated with the difference in the
way the two scientists dealt with the tail fin. Length measurements
made with the fins pinched together will usually be longer than those
without. Had they measured the same group of ten fish, one set of
measurements would have yielded lengths consistently smaller than
the other. This is a systematic error, and can often be troublesome
and difficult to detect. This highlights the need for a procedural state- Systematic errors result in data that
ment that establishes clear guidelines for measurements wherever deviate systematically from the true
values. These errors may often be more
such sources of systematic error can arise. difficult to detect and correct, and
Each of these types of error can affect the results of the measure- data collection efforts should make
great pains to eliminate any sources of
ments, and should be quantified and included in the description of systematic error.
the best estimate of fish length. But errors can affect the best estimate
in different ways. Instrumental error, as described above, can itself
either be random or systematic. The printed scale on one of the fish
measurement boards could be stretched by a factor of 3% compared
to the other, resulting in a systematic error. Likewise one board might
be made from plastic that is more slippery than the other and thus
more difficult to align the fish on. This could result in additional
random error associated with that device. But what are the relation-
ships between these types of errors and the best estimate that we are
seeking?
68 iowa state university

Error or variation? Questions to ask yourself

1. What were possible sources of error in your measurements?

Are they random or systematic?

2. How can you tell the difference between error in measure-

ment and natural variability?

6.5.2 Real variability

Not all deviations from the mean are errors. For real quantities in
nature, there is no good reason to assume that, for example, all age-0
brook trout will be the same length. Indeed we expect that there are
real variations among fish of a single age cohort due to differences in
genetics, feeding patterns, and other real factors. If we’re measuring
a group of age-0 fish to get a handle on how those fish vary in size,
then at least some of the variation in our data reflects real variation
in the length of those fish. How do we tease out the variation that is
due to errors from the variation that is due to real variability?
Often a good approach is to try to independently estimate the
magnitude of the measurement errors. If those measurement errors
are about the same magnitude as the variations (residuals) within the
data, then it may not be possible to identify real variability. However,
in the more likely event that our measurements are reasonably ac-
curate and have small measurement errors compared to their spread
about the mean, then the indicated variations probably reflect true
variability.
This observation returns us to our earlier question: when we seek
to characterize some quantity how should we identify our best es-
timate and our degree of uncertainty in that estimate. If we wish to
characterize a single quantity and our certainty that our best estimate
is close to or equal to the true value, we should use the mean of re-
peated measurements of this value and the standard error of those
measurements. The standard error can be readily estimated by divid-
ing the standard deviation of the repeated measures by the number
of measurements n:
σ
SE = √ . (6.9)
n
This should be equivalent to the standard deviation of a number of
estimates of the mean x̄, if several samples were taken from the full
population of measurements. Like the standard deviation, we can
be about about 68% confident that the range xbest + SE to xbest − SE
includes the true value we wish to characterize, but if we use 1.96 SE
5
Note that we are currently assuming instead, we can have 95% confidence5 . A complete statement, then, of
that our measurements are normally
distributed.
 quantitative problem solving in natural resources 69

our best estimate with 95% certainty in this context is to say:

x = xbest ± 1.96 SE, (6.10)

If instead we desire a characterization of a typical value and range

for something that has real variability among individuals in a popu-
lation, we will usually describe it with the mean and standard devia-
tion.
x = xbest ± 1.96 σ, (6.11)

6.6 Distributions

The kind of data we’ve been talking about thus far is univariate: a
single quantity with variable values like the diameter of a stream-
bed particle, or the length of a fish. As we know, not all age-0 brook
trout are the same size. In a first-pass capture of 50 fish, for example,
we should expect some variability in length that might reflect age,
genetics, social structure, or any other factor that might influence de-
velopment. The variation may be visualized graphically in a number
of ways. We’ll start with a histogram.
A histogram shows the distribution of a set of discrete measure-
ments – that is the range of values and the number of data points
falling into each of a number of bins, which are just ranges of val-
ues (112.5 to 117.5 is one bin, 117.5 to 122.5 another. . . ). This can be
Figure 6.3: A histogram showing the
called a frequency distribution, and a histogram is one of the best distribution of simulated (random)
ways to visualize a frequency distribution (Figure 6.3). measurements of the length of 100
snakes.
But what if we had uniformly distributed data? A uniform dis-
tribution means that it is equally likely that we’ll find an individual
with a length on the low end (97.5-102.5 mm) of the range as any
other. That would look quite different – there would be no hump
in the middle of the histogram, but rather a similar number of mea-
surements of each possible length. The uniform distribution is great:
in fact, we count on uniformity sometimes. If you are at the casino
and rolling the dice, you probably assume (unless you’re dishonest)
that there is an equal probability that you’ll roll a 6 as there is that
you’ll roll a 1 on any given die. We can call that a uniform probabil-
ity distribution for a single roll of a die. But What if the game you
are playing counts the sum of the numbers on 5 dice? Is there still a
uniform probability of getting any total value from 5 to 30?
We could actually simulate that pretty easily by randomly choos-
ing (with a computer program like R6 or Excel) five integers between
1 and 6 and adding them together. Figure 6.4 shows the plot that
comes out. Looks sorta like a bell curve, right? Well, how likely is
it that you’ll get five 1’s or five 6’s? Not very, right? You’re no more Figure 6.4: Sum of the values of five
likely to get one each of 1,2,3,4 and 5 either, right? However, there dice, rolled 100 times each.
6
R is a top choice software for general-
purpose data analysis and modeling.
It is free software, works on most com-
puter platforms, and has nearly infinite
capabilities due to the user-contributed
package repository. Learn more about R
at https://cran.r-project.org/
70 iowa state university

are multiple ways to get a 1,2,3,4 and 5 with different dice showing
each of the possible numbers, whereas there is only one way to get
all sixes and one way to get all ones. So there are better chances that
you’ll get a random assortment of numbers, some higher and some
lower, and their sum will tend toward a central value, the mean of
the possible values. So, since your collection of rolls of the dice rep-
resent a random sample from a uniform distribution, the sum of
several rolls will be normally distributed.
What’s it got to do with fish? If we sample brook trout randomly
from one stream reach and measure their lengths, we might expect
them to be normally distributed. Describing such a normal distribu-
tion with quantities like the mean and standard deviation gives us
the power to compare different populations, or to decide whether
some individuals are outliers. The nuts and bolts of those compar-
isons depend on how the type of distribution represented by the
population. An ideal normal distribution is defined by this equation:

( x − µ )2
 
1
f (x) = √ exp (6.12)
2πσ 2σ2

and it’s graph, in the context of our original hypothetical distribu-

tion of fish lengths, looks like the red line in Figure 6.5. In order to
compare the continuous and discrete distributions, we’ve divided the
counts in each bin by the total number in the sample (50), to yield
a density distribution. The blue line is just a smoothed interpolation
of the top centers of each bar in the discrete distribution, so it gen-
erally reflects the density of data within each bin. As you can see,
the discrete distribution density and the continuous normal distribu-
tion functions are similar, but there are some bumps in the discrete
distribution that don’t quite match the continuous curve. As you
can imagine though, that difference would become less pronounced
as your dataset grows larger. Related to this, then, is the idea that
your confidence in the central tendency and spread derived from your
dataset should get better with more data.

Figure 6.5: Superimposed discrete

distribution density (bars), interpolated Exercises
continuous density from the discrete
distribution (blue line), and an ideal
continuous distribution function with 1. Download the data from Derek Ogle’s InchLake2 dataset from
the same mean and standard deviation. the fishR data website. Using either a spreadsheet or data analysis
package, isolate the bluegill from the dataset and identify the
following:

(a) Mean bluegill length.

 quantitative problem solving in natural resources 71

(b) Standard deviation of bluegill length.

(c) Mean bluegill weight.
(d) Standard deviation of bluegill weight.

2. The graph and data table below and right show measurements of
brook trout lengths from pass #1 of the electrofishing campaign
described in Problem 3.7. Use these resources to answer the fol-
lowing questions:

(a) Judging from the histogram in Figure 2, does the dataset con-
tain just one mode or more than one? What might be the reason
for this?
(b) What is the mean and standard-deviation for the (presumed)
age-0 portion of this sample?

Figure 6.6: A complete frequency

distribution of brook trout lengths from
electrofishing pass #1 from Problem 3.7.
72 iowa state university

index 1 2 3 4 5 6 7 8 9 10 Table 6.2: Brook trout length data from

electrofishing pass #1. All lengths in
1 313 135 342 297 137 112 379 116 142 154 mm.
2 288 322 241 364 360 348 265 127 297 143
3 355 110 152 107 157 338 135 345 251 110
4 127 372 164 417 364 358 113 329 83 366
5 305 343 129 378 298 245 392 121 371 394
6 256 397 114 292 146 147 243 320 294 154
7 406 301 156 294 396 132 296 349 247 313
8 261 406 332 381 329 250 233 316 130 104
9 248 294 427 295 316 339 328 255 344 121
10 312 339 271 323 272 259 120 123 316 301
11 401 114 279 160 293 321 217 301 240 133
12 135 370 275 137 139 130 276 299 296 111
13 323 250 414 308 317 362 336 332 429 114
14 141 163 264 325 151 167 380 100 138 120
15 160 321 246 351 369 146 284 108 131 136
16 263 131 376 374 419 310 431 121 321 326
17 125 410 312 347 113 297 89 96 294 134
18 342 356 110 131 139 296 285 99 313 372
19 361 428 344 301 365 347 283 158 331 397
20 149 155 307 165 321 224 137 333 132 231
21 329 133 305 388 319 120 389 330 411 143
22 306 261 359 126 143 386 338 179 319 140
23 273 320 122 144 384 112 408 316 344 303
24 122 324 137 331 92 113 341 399 353 305
25 287 117 354 332 376 282 244 335 157 144
7
Interlude: Collecting and managing data

Data is information. Data is the result of somebody’s efforts to record There are some great resources for
and store information, often to provide an opportunity for insight. It data management out there in various
forms, including some geared toward
can be used to discover patterns, test hypotheses, and support argu- biologists. A few great ones are:
ments, among other things. But the numbers themselves often cannot Data Carpentry
www.datacarpentry.org.
convey much meaning – it is through manipulation and interpreta- Wickham, H., 2014, Tidy Data. Journal
tion of the data that those uses can be realized. We therefore need of Statistical Software 59(10).
to be conscious about how data are structured and managed, so that Saltz, J.H. and J.M. Stanton, 2017,
Introduction to Data Science, Sage publ.
they can be manipulated and interpreted to reveal insights. If data
are badly structured and managed, the risks can be great. Optimisti-
cally, we risk wasting lots of time trying to restructure data to allow
the kind of analysis we wish to perform; worse yet, we risk compro-
mising the integrity of or, heaven-forbid, the complete loss of data
through poor structuring and mismanagement.
First, lets be clear about what is meant by data structure and data
management:

• Data structure refers to the organization and layout of data as

it is stored. Whether it is handwritten in notebooks or stored in
spreadsheets or text files, data usually has an architecture that
reflects the intentions (or ignorance) of the data manager or some-
times the protocols of his/her organization or institution.

• Data management is the set of practices aimed at preserving the

quality, integrity, and accessibility of the data. This can include all
phases of data usage from collection and manipulation to storage,
sharing, and archiving.

7.1 Who is data for?

Unless you work with highly classified or proprietary information

and are required to protect and encode your data, you likely need
data to be readily understood and usable, not just to you but to oth-
ers you work with or the broader public1 . But we also need to realize 1
Many government funding agencies
such as the National Science Founda-
tion, US Department of Agriculture,
and National Institutes of Health now
require that researchers develop a
data management plan that includes
strategies for structuring, archiving and
indexing data in publicly-accessible
repositories.
74 iowa state university

that the humans who need to make sense of the data will be using
tools like computers to facilitate this approach. Therefore the data
structure needs to also accommodate the demands of the computer
hardware and software that it is used on as well as the humans.
Thus, data should be organized in a logical, self-consistent way and
it should be accompanied by documentation that helps to explain
the content and and context of the data. Similarly, accessible data
archiving, in principle, allows colleagues and competitors to test, ver-
ify, reproduce, and/or compare results with their own, ensuring that
scientific advances that you make with the help of your data can also
lead to advancement of science and management more broadly.

Figure 7.1: A typical workflow for data.

After Grolemund and Wickham, 2017, R
for Data Science, O’Reilly.

Consider the schematic workflow illustrated in Figure 7.1. Once

collected, data must be organized and formatted in a way that facil-
itates their analysis on a computer. A popular term for data that are
formatted to simplify computer manipulation is tidy data (more on
this below). When this process is complete, the data may be analyzed
as needed to address the problem or hypothesis at hand. This process
of making sense of the data may then produce a result that needs to
be communicated back to humans. When data is presented for com-
sumption by the human eye and brain, the organization and structure
should reflect the expectations and attention span of the humans.
Unless the data set is small, the raw or transformed data may not be
appropriate to display. Instead, summary data are more appropriate,
either in narrative, table, or graphical format.

7.2 Tidy data

To understand the significance of tidyness, it is perhaps helpful to

consider untidy or kludgy data. Below is a portion of a data table
2
This is a deliberately disorga-
nized snippet based on a dataset containing the weights and lengths of small fish captured during a
from Derek Ogle’s neat website population survey of Inch Lake, Wisconsin2 . Let’s unpack this data
http://www.derekogle.com/fishR/data
set. There are two different fish species listed, one observed both
in 2007 and 2008, and one only in 2007. Lengths and widths are
provided for all fish measured, but there are different numbers of
 quantitative problem solving in natural resources 75

Table 7.1: A portion of an untidy number bluegill (2008) bluegill (2007) Iowa Darter (2007)
dataset.
1 L:1.5 W:0.7 L:1.9 W:1.3 L:2.1 W:0.9
2 L:1.0 W:0.7 L:1.6 W:1.3 L:2.0 W:1.3
3 L:2.6 W:1.5 L:2.4 W:1.7 L:1.7 W:0.7
4 L:1.1 W:0.6

fish in each column. There is also an index value in the first column
that facilitates counting the number of fish of each species captured
in each year. This is reasonably straight-forward for a human to
interpret, particularly if we are told that L corresponds to a length
in inches and W corresponds to weight in grams. However a larger
dataset organized like this table would be miserable to analyze for a
variety of reasons, including:

• columns don’t have the same number of values

• the same species has data across multiple columns

• two variables (length and width) are listed together within each
column, with numbers and letters mixed

One instructive question to ask is how many variables there are here.
We notice that data span multiple years, so year could be a variable.
There are also multiple species here, so species could be viewed as a
variable. Then length and width should each be variables. Finally, if
we wish to have an index or ID number for each fish, that might be 3
If there are multiple datasets derived
a fifth variable3 . In general, tidy data is organized in a rectangular from the same group of fish, assigning a
fish ID number would be a simple way
array in which each column represents a variable and each row an to connect these datasets using database
observation. In most cases, the first row contains descriptive but sim- methods.
ple column headers. This simple prescription seems unthreatening,
but it is often surprising how pervasive untidy data is. Tidy data has:
So with five variables, how many observations do we have? Each • one column for each variable
fish represents an observation, with a unique ID number, year of cap- • one row for each observation
ture, species, weight and length. From Table 7.2, there appear to be • a header row
ten fish listed among the three columns. so since each fish is an ob-
servation. According to the principles of tidy data, then, there should
be ten rows of data with values in each of the five columns. Below is
a tidy representation of this dataset. This table is now organized in a
way that can easily be sorted, filtered, and summarized in common
statistics and computational software packages.

7.3 Data management

Because data is often the hard-won result of costly and time-consuming

observations and measurements, its management should be deliber-
76 iowa state university

fishID year species weight (g) length (in) Table 7.2: The data from Table 7.2
transformed to a tidy dataset.
1 2008 bluegill 1.5 0.7
2 2007 bluegill 1.9 1.3
3 2007 Iowa darter 2.1 0.9
4 2008 bluegill 1.0 0.7
5 2007 bluegill 1.6 1.3
6 2007 Iowa darter 2.0 1.3
7 2008 bluegill 2.6 1.5
8 2007 bluegill 2.4 1.7
9 2007 Iowa darter 1.7 0.7
10 2007 bluegill 1.1 0.6

ate and careful. Well-managed data can be stored, retrieved, ana-

lyzed, and used to develop insights or aid in management decisions
without compromising the data itself, and without spending un-
necessary time and energy in decoding and interpreting the raw
data. Thus, proper management entails not only careful structure as
described above, but also well-organized storage and full documenta-
tion.
First, important raw data should be stored redundantly. If it’s
only in hard-copy (e.g., in field notebooks), consider scanning or
transcribing the hard-copy data to preserve a digital version that can
be backed-up regularly. When the data results from original research
4
Data repositories like the LTER Data that can be shared publically, it can be uploaded to data repositories4
Portal require formatted data and meta- A complete set of data should be archived and never modified, while
data to ensure long-term accessibility
and adequate documentation. data reduction and analysis are done on copies of the formatted raw
data.
When analysis or reduction occurs much later than the time of
collection and storage, or is done by a different person or group than
the researcher who collected the data, adequate documentation or
metadata is essential. Metadata can include narrative descriptions of
where the data was collected, when and how it was collected, and
should include references or links to any published or publically-
available research or information stemming from the data. The meta-
data should always include a data dictionary, fully describing the
quantities represented by each of the variables collected (i.e., variable
name, symbol, units, and procedural statement). These guidelines
ensure that data remain safe, useful, and accessible.
 Part III

SPATIAL REASONING
8
Geometry and Geography

One of the fundamental types of quantities that we use as natural

resource scientists and professionals is a measure of distance or size.
Whether we’re describing the board-feet of merchantable lumber
in a ponderosa pine, the fork length of a trout, the size of a white-
tail deer’s home range, or the storage capacity of a flood-control
reservoir, we are concerned with spatial quantities that ultimately
manifest from linear measurements in space. Much of this may feel
familiar to you, but there are important messages to take home from
working with both simple and compound spatial quantities that
will serve you well in working with maps, photos, design plans, and
other tools that professionals use.

8.1 Length, Area, and Volume

Consider the wetland shown in the map below (Figure 8.1). How
might we characterize its size? Perhaps the answer depends on the
context of the question. Are we interested in how far it is to cross
it in a canoe? How long the shoreline habitat is for waterfowl? The
number of acres it occupies? What about the amount of water it
holds? In turn these questions point to linear distance, curvilinear
distance, area and volume, respectively. Each of these types of quan-
tities can be expressed in a variety of ways1 according to the setting, 1
For example, we can express all the
the application, or the purpose of communication. quantities in terms of SI units of m, m2 ,
and m3 , or we can use more traditional
The distinctions between length, area, and volume are more than U.S. agricultural units like feet, acres,
just trivia. They clearly reflect not only different ways of estimat- and acre-feet.

ing size, but different numbers of spatial dimensions. In common

parlance, length is one-dimensional or 1D, area is two-dimensional
or 2D, and volume is three-dimensional or 3D. The units can be a
clue to how many dimensions are indicated in a spatial description.
From Chapter 4, recall that we can generalize units in terms of the
fundamental dimensions they entail. From this perspective then, a
1D length has dimensions of [ L], a 2D area [ L2 ], and a 3D volume
80 iowa state university

Figure 8.1: Map of a wetland in Iowa.

The blue polygon shows the extent of
seasonal open water overlain on shaded
relief.

[ L3 ]. This is true regardless of the specific units used to describe the

quantity of interest, though sometimes the dimensionality can be
obscured by the use of compound units. For example, if we’re told
that a woodlot is 820,000 ft2 , that is a straight-forward 2D measure of
2
One board-foot is equivalent to 144 area. If that same woodlot is described as 126,000 board feet2 , now
cubic inches of merchantable timber; we’re talking about a volume of wood expressed in a unit that is not
can be visualized as a 12 inch long and
12 inch wide board that is 1 inch thick. particularly transparent to outsiders, though it is customary among
foresters. If we need to do computations involving quantities of this
sort we need to be certain that we understand what makes sense and
what doesn’t make sense to do; what’s permissible and what isn’t.
Note that some of our commonly-used spatial units are compound
by definition. An acre, for example, is a unit of area even though
it is not expressed in a squared-length form. Originally defined as
the area of land that could be plowed with oxen in a day, an acre is
43,560 ft2 or roughly 4,047 m2 . If you measured out a square 208.7 ft
on a side, that would be approximately an acre. The acre’s cousin in
SI units, the hectare, is also a unit of area with a simpler definition:
3
Neither the acre nor the hectare need an area of land 100 m wide and 100 m long, or 10,000 m2 .3
be any particular shape, nor do they We’re also familiar with several alternative ways of expressing vol-
necessarily need to be contiguous,
though they usually are. ume with derived units, particularly when talking about liquids. It’s
not unheard-of to talk about fluid volumes in cubic meters or cubic
feet (particularly if we’re referring to volume per unit time as we do
when describing river flows in cubic feet per secod or cfs), but it is
more common to hear fluid volumes expressed in gallons, millileters
 quantitative problem solving in natural resources 81

or liters. These are all legitimate expressions of fluid (gas or liquid)

volumes and some have relatively straight-forward relationships to
length-cubed volumes: for example, 1 ml is the same as 1 cm3 or cc.
However, if we wish to perform computations more complex than
addition or subtraction on such quantities, it can be advantageous
to convert them into more fundamental units like cubic meters. One
interesting unit of volume mentioned above is the acre-foot, which
(as you might guess) is the volume corresponding to a one acre area
of something that is one foot deep. This means its dimensions are
an area [ L2 ] times a depth [ L], therefore its a volume [ L3 ]. We en-
counter this unit of volume sometimes in descriptions of ponds or
stormwater-basins because it may be easier to visualize, but this can
also make it more difficult to perform computations.

8.1.1 Unit conversions in space

Here’s an common exercise that American students often need to
perform in earth science, geography or natural resource courses4 : 4
perhaps this exercise is going the
measure a rectangular land area on a USGS map with a scale in feet way of the paper map itself as people
increasingly interact with only digital
and miles, and convert it to square meters or square kilometers. maps these days!

5000 ft. Figure 8.2: A generic map scalebar

showing map distances in feet and
0.5 1 mile miles.
scale = 1:31,680

This exercise will typically begin with each student begrudgingly

making tickmarks on the edge of a sheet of paper that is lined up
with the map scale. For purposes of illustration, we’ll follow the
hypothetical (but not uncommon) path of a student who is prone
to making some common mistakes. Our student uses the marked
paper to estimate the length of each side of the rectangular land area
using the scaled map units5 . Perhaps the values are 6.2 miles and 5
see the next section for more on
2.1 miles. He then proceeds to multiply them together, because he’s scaling

aware that the area of a rectangle is the product of its sides. So he

punches 6.2 × 2.1 into his calculator and gets 13.02. When asked to
supply units for his answer, he reasons that since the units he was
measuring distances in were miles, the answer is also in miles. His Common Mistake #1: leaving the
instructor points out that miles are a unit of length not area, and that units out of computations can lead to
errors in unit assignment for solutions.
he should write the equation out complete with units to ensure that
his result comes out in units of area. So he writes:

6.2 mi × 2.1 mi = 13.02 mi2

and the instructor nods approval but says, “and now we need the
area in square kilometers”. Our downtrodden student proceeds to
82 iowa state university

look up the conversion factor between miles and kilometers: 1mi '
1.609 km. Great. The calculator buttons click away until the student,
exasperated, inquires “so the area in square km is 13.02 × 1.609,
Common Mistake #2: using length which equals about 20.95 right?” The ever-patient instructor shakes
conversion factor for area conversion. her head: “there are 1.609 km in a mile, but how many square km
in a square mile?”. Our student stares at the map and feigns interest
in the question. On a whim, he asks “do I need to square the 1.609
too?” The instructor pats him on the shoulder and remarks “yep,
write it all out, and don’t forget the units” as she walks away. Our
student, relieved at having guessed correctly, writes:

6.2 mi × 2.1 mi = 13.02 mi2

 2
km
13.02 mi2 × 1.609 = 33.71 km2
mi
To see why we need to square the conversion factor like our student
ultimately did, let’s write out the conversion equation the way he did
it at first, but using only the units (this is a variation of the dimen-
sional homogeneity heuristic from Chapter 4):
km
mi2 × = km2
mi
If we cancel common units, we should be able to show that the units
of the left-hand side are equal to the units on the right-hand side,
but here we can only cancel miles in the numerator from miles in
the denominator on the left-hand side, leaving a meaningless unit
equation: mi km = km2 . That can’t be true.
If instead we reason that our conversion factor needs to allow us to
cancel through to make the units equivalent on both sides, we square
Heuristic: Unit conversions can be the conversion factor and its units to yield:
written as equations with the current
quantity and units on the left-hand
 2
km
side and the quantity in desired units
2
mi × = km2
on the right-hand side. All conversion mi
factors should include units and be
subject to operations such that the ex- This approach can be generalized for other types of spatial unit
pression satisfies unit and dimensional conversions as well, provided that our original and desired units are
homogeneity.
not compound. Suppose we are measuring the size of something in
units based on the length unit U1 and we need to convert it into units
based on the length unit U2 . If the conversion factor between U1 and
U2 is C1→2 , the conversion equation can be written:

U1d × C1d→2 = U2d (8.1)

In this equation, d is the spatial dimensionality of the quantity, so its

1 for lengths, 2 for areas, and 3 for volumes. It’s important to note
that the conversion factor C1→2 should correspond to the number of
unit 2’s per unit 1, as we did for the map area conversion above.
 quantitative problem solving in natural resources 83

Note that unit conversion factors between compound units like

acres and hectares are not subject to these concerns. There is no such
thing as a square acre because an acre is already a unit of area, so
nothing needs to be done to the conversion factor if you are convert-
ing from, for example, acres to hectares: there are 0.4047 hectares
in an acre, period. In this way these compound units can make life
easier, but if you are simultaneously working with other quantities in
meters, this convenience comes at a price.

It might have occurred to you to approach the map problem in

a slightly different way. Suppose that instead of computing the area
in square miles immediately after measuring the sides of the rect-
angle, our student had converted the sides from miles to kilometers
first. Does this make any difference?
In this case, the map distances are 6.2 × 1.609 = 9.976km and
2.1 × 1.609 = 3.379km. Their product is 33.71km2 , which is the same
result as before. That should come as no surprise, since the only
difference is that the unit conversions from miles to km occurred
before finding the area rather than after. Indeed this is one way to
make the problem a bit simpler to think about, but thanks to the
commutative property of multiplication there is no real difference
between the approaches.

8.1.2 Scales and Scaling

The map scalebar in Figure 8.1.1 shows not only a graphical scale
that can be used to directly measure real-world distances from the
(scaled) map representation, but also indicates a ratio: 1:31,680. What
does this scale mean? Does it have units?
Map scales are like most other dimensionless proportions, as we
discussed in Chapter 4. The beauty of many dimensionless propor-
tions is that we can use any units we want in them provided that both
parts of the scale ratio or proportion have the same units. So for the
map scale, we can say that 1 inch on the map is equal to 31,680 inches
in the world that the map represents6 . If we measure out a path on 6
For better or worse, the US still per-
the map that is x inch long, the distance we would cover walking sists with using inches, feet, and miles
as conventional measures of distance
along that path in the real world is x × 31680 inches. That’s not a in official maps and documents. Even
very easy distance to envision because there aren’t any very familiar though most scientists adopted the met-
ric system long ago, we retain imperial
benchmarks near that quantity of inches, but we could convert the units here to recognize the persistence
latter to feet or miles to make it simpler, and then we can re-express of legacy units in our maps.
the scale in as a dimensional ratio: 31680 in. × 1 ft./12 in. = 2640 ft.. Recall that benchmarking is a process
of conceptualizing the size of a quantity
We can go one step further still: 2640 ft. × 1 mi./5280 ft. = 0.5 mi.. by comparing it with a known reference
That works out pretty nicely, and it often does so by design! So we quantity.
can restate the map scale as 1 inch = 0.5 miles, or equivalently 2 1 mile = 5280 feet
84 iowa state university

inches per mile. This means the same thing as the scale ratio 1:31,680
but is more specific because we have already chosen the units that we
wish to measure with. Note that we cannot say that the map scale is
2:1 or 1:0.5 because in converting the second number from inches to
miles we’ve made the scale statement unit-specific.
Maps aren’t the only thing that we encounter that are scaled repre-
sentations of reality. When we are learning about microscopic prop-
erties of molecules or cells, we often look at diagrams or physical
models of things that are too small to see. When looking through
a microscope, we perceive a much enlarged version of the object of
study. In each case, we are seeing representations of reality scaled
7
This number comes from: Trans- to size that is easier for us to grasp. Importantly, we are also (usu-
portation Research Board and Na- ally) seeing things scaled isometrically, meaning that all dimensions
tional Research Council, 2005. As- are enlarged or shrunk by the same constant factor. We’ll see in later
sessing and Managing the Ecological
Impacts of Paved Roads. Washington, chapters some interesting problems associated with scaling that is not
DC: The National Academies Press. isometric.
https://doi.org/10.17226/11535.

8.1.3 Example: the area of roads in a county (Problem 3.3)

One reasonable sub-problem to address in the issue of deer-car col-

lisions is how widespread are roads in the area of interest? As with
several of the other teaser problems in this book, no specific county
is cited, so as a first approximation I’ll just estimate numbers for my
own home county: Story County, Iowa. According to Wikipedia,
Story County has an area of 574 mi2 . The extent of roads in Story
County or elsewhere in the US is something that could be readily
assessed with a GIS system, and that would certainly be among the
most accurate and efficient ways to obtain this value for specific
counties. However, for the sake of a first approximation let’s try
something easier. Literature about road systems in the US suggests
that there is about 1.2 miles of road per square mile of land, on av-
erage7 . Clearly this is an underestimate in urban areas, and perhaps
Figure 8.3: In rural farm country, roads an overestimate in remote, rural areas. In the not-so-remote grid-
are often arranged in an almost-regular ded farmscape of central Iowa (Figure 8.3), a road density closer to
grid with spacing of 1 mile. In these
settings, we can estimate the “road
2 mi./mi.2 is perhaps more appropriate. By this estimate, my county
density” by imagining a square-mile would have approximately 2 × 574 = 1148 miles of roads.
box centered on a road intersection. Road density is informative, but it only gets us partway to the
Within the box shown above, there
are 2 miles of road, suggesting a road notion of area. What we need to know now, given that we have road
density of 2 miles per square mile. length, is the average width of a road. Let’s assume this is 20 feet.
To be a meaningful area, we either need to decide to convert length
8
This is a seemingly-trivial but still to feet or width to miles8 . Since the county area was estimated in
significant decision. In Schoenfeld’s miles, it would be wise for comparison purposes to convert road
framework for problem-solving, making
this kind of decision deliberately with width into miles as well. Thus, the average road is about 20/5280 =
the broader goals and practical issues in 0.003788 miles wide. Multiplying that road width (in miles) by the
mind is an example of control.
 quantitative problem solving in natural resources 85

total road length (in miles) yields about 4.35 square miles. That’s
about 4.35/574 × 100% = 0.76% of the county area!

8.2 Geometric Idealization or Approximation

Some problems require spatial measurements or computations that

are complex, time-consuming, or difficult to visualize. In some of
these cases, a rough estimate may be adequate for the type of solu-
tions we seek; in other cases, we may wish to establish quick ballpark
estimates before we dig too deeply into the complex computations,
much like we just finished doing in the previous section. For these
spatial problems, it can be helpful to idealize the spatial information
we have in terms of simple geometric figures that we know some-
thing about. For example, suppose we wish to estimate roughly how
much an 40cm-long snake might weigh, and we have no prior infor-
mation or experience upon which to base this estimate. If we are able
to estimate its diameter, we may make some progress by idealizing
the snake as a long cylinder. Consulting the table below, we find that
a cylinder’s volume is expressed as:

V = πr2 h (8.2)

where r is half of the diameter and h is the length of the snake. If

the largest r is around 1cm and h is 40cm, a first guess for the to-
tal volume is about 126 cm3 . Now if this radius corresponds to the
largest part of the snake, this volume might be an upper bound.
Since the snake’s body tapers a bit, perhaps a mean radius is better –
say 0.7cm. Now the volume is 61.6 cm3 .
Next, given that many non-avian animals have densities close to
that of water9 , we can estimate the mass of the snake using density 9
in so-called cgs units, the density of
and volume. To see how we should do that, we might use a strategy water is 1.0 g/cm3 .

described in an earlier chapter: write out the problem with just di-
mensions. We’re looking for how much it weighs, but really what we
want is a mass. If we list the dimensions of the variables we have and
want, they look like this:

mass:[ M] volume:[ L3 ] density:[ M L−3 ] (8.3)

we see that the mass we are looking to solve for appears in the den-
sity term. The volume term appears as a negative 3rd power in the
density term and a positive 3rd power in the volume itself, so when
multiplied together, volume (V ) and density (ρ) should yield dimen-
sions of mass: m = ρV. If we remind ourselves that the definition
of density is indeed ρ = m/V, this makes sense as a simple alge-
braic modification of that definition. Thus, our estimated mass for the
86 iowa state university

snake would be:

61.6cm3 × 1.0g/cm3 = 61.6g. (8.4)

Is this close enough? Perhaps, but that depends on the nature of the
problem: why do we wish to know how much the snake weighs, and
what will we do with that information?

8.2.1 Example: herbicide purchase (Problem 3.2)

We can use a similar approach to get a start with the herbicide vol-
ume needed to eliminate woody shrubs from our city greenspace. A
reasonable assumption is that, when cut with a saw or lopper, the ex-
posed stem cross-section of a woody plant is approximately circular.
If the goal is to cover the stumps completely with a coating of herbi-
cide, each stump will have a volume (from Equation 8.2) equal to its
cross-sectional area πr2 times the thickness h of the herbicide coating
applied to the stump.

8.3 Measuring polygon area

Not all spatial bodies of interest to us are easily measured using the
simple idealized shapes reviewed above. Alternatively, geometric
idealizations introduce too much error for certain applications. Non-
ideal shapes can, however, be approximated by irregular polygons in
some settings. Here, we describe a method for computing the area
of an arbitray two-dimensional polygon using a clever trick that is
frequently built into CAD, GIS, and other geospatial software pack-
ages. The primary requirement we must meet to use this method is
to have coordinate pairs for each vertex in the polygon in a Cartesian
(i.e., a plane with two orthogonal, linear axes; a.k.a. an x-y plane) co-
ordinate system. This algorithm may be implemented easily enough
by hand, but finding the area of a more complex shape is better left
up to a computer. Let’s have a look at how this algorithm works.
 quantitative problem solving in natural resources 87

Table 8.1: Geometric relationships for Properties of simple geometric forms

common shapes.
shape property

circle
circumference: 2πr
r
area: πr2

rectangle h
perimeter: 2b + 2h
area: bh b

a
triangle h b
perimeter: a + b + c
c
area: 12 ch

sphere r
surface area: 4πr2
volume: 43 πr3

h
rectangular prism
surface area: 2bw + 2bh + 2hw
volume: bwh w
b

cylinder h
surface area: 2πrh
volume: πr2 h
r
88 iowa state university

First, recall that a trapezoid is a four-sided shape in which only

two sides are parallel, as in Figure 8.4. At first glance, it might seem
that the area of the trapezoid would be challenging to estimate, but
when we realize that it should be the same as the area of a rectangle
that’s as tall as the average “height” of two vertical sizes of the trape-
zoid, we may see some hope10 . We know that the area of a rectangle
is just its height times its width: Arect = hw. For a trapezoid whose
vertical sides have heights hl and hr for left side height and right side
height, respectively, we can restate the formula for area in terms of
Figure 8.4: A trapezoid (in gray) with the average of those heights:
dashed line indicating the rectangle
with the same area. 1
Atrap = ( h + hr ) w (8.5)
2 l
10
In Figure 8.4 you can imagine snip-
ping the tope of the trapezoid on the Now suppose that instead of defining heights and widths in terms
dashed line, flipping it over and filling of h and w, we have the vertices of our trapezoid in Cartesian coor-
the void in the upper right. Alterna-
dinates11 . Each point at a vertex (corner) of the trapezoid therefore
tively, we could imagine splitting the
trapezoid into a shorter rectangle and has an x, y coordinate, where x refers to the horizontal coordinate
a full-width triangle and compute the direction and y is vertical. In this system, note that the width of the
area as the sum of those two areas.
With a bit of algebra, we find that the trapezoid will be defined by the difference in two x coordinates. So
result is the same. if we take the upper left corner in Figure 8.4 to have an x coordinate
of x1 and the upper right corner to be at x2 , the width of the trape-
11
Cartesian just means that we are
specifying the location of points in a zoid is x2 − x1 . Similarly, if our trapezoid height extends to zero12
two-dimensional coordinate system in the y coordinate direction, the “average” height of our trapezoid
where the coordinate directions are
can be re-written by substituting y1 and y2 for hl and hr . If we make
perpendicular to one another. We will
see below that the UTM coordinate these changes to the area formula above, any given trapezoid in our
system for maps is Cartesian, whereas coordinate system has an area:
latitude and longitude are not.
1
Atrap = (y + y2 )( x2 − x1 ) (8.6)
12
in fact we could show that it needn’t 2 1
go all the way to zero provided that
all of the y values in the trapezoid Now consider the polygon depicted in Figure 8.5. Each pair of
collection are positive all are negative, adjacent vertices can be viewed as the upper corners of a trapezoid. If
but this is left as an exercise.
we apply the formula above to each pair of adjacent vertices and add
the areas together, what do we get?

1
A= [( x2 − x1 )(y2 + y1 ) + ( x3 − x2 )(y3 + y2 ) + ...
2
... + ( x1 − xn )(y1 + yn )] (8.7)

Here, n is the number of vertices and the ellipsis “. . . ” means that

we’ve left out some number of terms in the equation, though in this
case we’ve only got five vertices so we’ve only left out two terms.
Notice that if we systematically label the vertices in our polygon in
a clockwise manner, about half of the trapezoids will have negative
areas and half will have positive areas, though the negative areas will
be somewhat smaller. This is because as we come around the bottom
 quantitative problem solving in natural resources 89

of the polygon as we march clockwise from one pair of vertices to 7 3

2
the next pair, our x coordinates are becoming smaller as we go from
6
right to left. This is good! The result is that the width value (and con-
4
sequently the area) computes as negative, and as a result this lower 5
1
trapezoid is subtracted from the total area of the polygon, which is 4
exactly what we want! If we order the vertices counterclockwise we’d
3
get the opposite result, but the resulting (negative) area would still be 5
correct. 2
As mentioned earlier, this algorithm is readily implemented on a
1
computer, using either spreadsheet software like Excel, or computa-
tional/statistical software like R. Likewise, this method is built into
0 1 2 3 4 5 6
other software tools that natural resource students and professionals
use, including most GIS packages. Figure 8.5: An arbitrary polygon.
Using the trapezoidal algorithm,
areas for individual trapezoids are
8.3.1 Example: open-water waterfowl habitat (Problem 3.1) computed one-by-one in clockwise
order. In the lower part of the polygon
One important variable that influences waterfowl abundance is the (where trapezoid fill is darker gray), the
computed areas are negative, trimming
presence of different types of habitat. Most waterfowl feed exten-
the unwanted trapezoid area from
sively in open water, so the area of open wetland is a key habitat below the polygon.
variable. In Section 8.3 we identified the trapezoidal algorithm as a
tool for estimating the area of arbitrary shapes. We also acknowl-
edged that, while it is possible to do the required computations by
hand, automating the algorithm improves computational efficiency
by orders of magnitude. The algorithm can be implemented as a for-
mula in a spreadsheet containing the coordinates of the polygon but,
as we’ve discussed, this operation is common enough that it is incor-
porated in most GIS software. Therefore, comparing watershed areas
between the parcels described in Problem 3.1 is a geospatial prob-
lem. We nevertheless provide an opportunity in the chapter Exercises
below to work with this algorithm directly.
90 iowa state university

Exercises

1. What is the longest distance across the wetland in figure 8.1?

2. Estimate the dimensionless map scale ratio from the scalebar in

figure 8.1.

3. Would the map scale be the same or different if you made an

enlarged photocopy of a map?

4. Use geometric approximation to estimate the area of the wetland

in figure 8.1.

5. Use the trapezoidal algorithm to make a more precise estimate of

the wetland area. The data table below contains UTM coordinates
of the wetland perimeter.
 quantitative problem solving in natural resources 91

easting, m northing, m
1 477698 4651450
2 477720 4651488
3 477738 4651524
4 477746 4651549
5 477768 4651554
6 477776 4651574
7 477784 4651607
8 477792 4651630
9 477789 4651652
10 477794 4651668
11 477825 4651670
12 477855 4651675
13 477882 4651671
14 477906 4651678
15 477923 4651701
16 477946 4651714
17 477971 4651721
18 477994 4651723
19 478002 4651737
20 478009 4651761
21 478015 4651779
22 478035 4651794
23 478053 4651792
24 478066 4651792
25 478083 4651794
26 478099 4651794
27 478116 4651789
28 478131 4651780
29 478136 4651757
30 478132 4651736
31 478114 4651714
32 478099 4651694
33 478081 4651678
34 478066 4651661
35 478070 4651633
36 478068 4651604
37 478065 4651587
38 478065 4651562
39 478065 4651539
40 478061 4651516
41 478053 4651500
42 478040 4651491
43 478033 4651472
44 478023 4651449
45 478010 4651425
46 477986 4651406
47 477953 4651394
48 477923 4651379
49 477905 4651371
50 477882 4651364
51 477857 4651356
52 477837 4651359
53 477814 4651368
54 477797 4651363
55 477781 4651368
56 477761 4651368
57 477740 4651368
58 477716 4651376
59 477705 4651391
60 477698 4651412
61 477690 4651432
62 477698 4651450
9
Triangles

9.1 Measuring with Triangles

It is fair to ask why we should bother learning about triangles, since

their relevance to ecology and natural resources isn’t immediately ap-
parent. Indeed, natural materials tend to approximate more tabular
or rounded shapes, and true triangles are harder to find in nature by
comparison. But the real power of triangles is not in where we can B
see them, but where we can imagine them. Believe it or not, imag-
inary triangles can help us measure properties of a landscape or
organism, and that fact is firmly incorporated into many of the tools
and technologies that researchers and professionals use. In particular, c
a
the ratios between the lengths of triangle sides is one of their key as-
sets. In this chapter, we’ll review some of the properties of triangles
and see how these properties can be leveraged to measure things we
care about.
A C
b

9.2 Trigonometry primer Figure 9.1: A simple triangle. Note that

the corners (or vertices) are labeled
Trigonometry is the study of triangles, specifically the relationships with capital letters, and that the sides
opposite those vertices are labeled with
between the lengths of their sides and the angles between them. At
the lower-case version of the same
first glance, that might not seem like it is very relevant to the natural letter. This is partly by convention, and
sciences, but a few examples might convince you otherwise: to facilitate some of the techniques we’ll
review below.
• Determining the distance “as the crow flies” between two geo-
graphic points is often most easily done with the help of triangles.

• Measuring the height of a tree or a mountain can employ triangles.

• Telemetry often uses triangulation to determine the geographic

location of collared animals.

If you’ve had a trigonometry class, you might associate the dis-

cipline with manipulating equations with sec 2θ and cot(1 + π/2).
Outside of math class, did you ever find yourself needing to find
94 iowa state university

the secant of an angle? Not likely. But it isn’t too uncommon to en-
counter the likes of sine and cosine, which are often written sin and
1
At this point you may just need to take cos, respectively. That’s because they are really, really useful1 . And
my word for it, but I hope that you’ll it turns out, nearly all the other trig functions you may have learned
appreciate this fact by the time you
finish this chapter. about are readily defined using sines and cosines! For example, the tan-
gent of an angle can be defined as the ratio of the angle’s sine to its
cosine, but it is so useful that we should recognize it as well.
As far as I’m concerned, on the off chance that you ever need to
manipulate an equation containing the hyperbolic cosine (cosh) of
something, you can look it up or type it into an internet tool like
Wolfram|Alpha is a web-based tool Wolfram|Alpha. If this is your first experience with trigonometry,
developed by mathematician and en- have no fear, we’ll take it slow!
trepreneur Stephen Wolfram. It is based
on the same underlying computational
engine as the math software Mathe-
matica, but can be used (for free with
slightly-limited functionality) from
any web browser. In addition to per-
forming computation and algebraic
simplification, it can attempt to com-
prehend simple written questions and
can retrieve data from a few established
databases, concerning for example Similarity
weather, fincance, and sports.

Similarity is a concept that may not boast enough sophistication to

warrant its inclusion in a trigonometry class. However, it is an intu-
itively easy idea to grasp and its utility can be great. And fortunately
for modern scientists, the formal application of similarity allows us to
design tools for measuring things efficiently in the field.
The principle of geometric similarity2 is straight-forward. If we
2
In some fields this concept is endowed
with a more sophisticated sounding have a given shape with known side lengths and/or known angles
name: geometric similitude. formed between the sides, we can say that another shape is similar
if it has the same number of sides and a relationship between those
sides and angles that is the same as our reference shape. The two
shapes can still be similar even if they are not the same size or orien-
tation. If any combination of translation (moving the shape), rotation,
3
The word isometric in this context reflection (a mirror image), or isometric3 scaling could allow you
means that any change in one spatial
to overlay one shape on the other to find them to be identical, the
dimension of a shape (e.g., length) is
matched by a proportional change in all shapes are similar.
other dimensions. For triangles, the qualifying criteria for similarity are simple, since
there are only three sides and one internal angle at each of the three
vertices. For shorthand, when two triangles have one angle that is
identical, we’ll call that A. When two triangles have all three angles
equal, we’ll refer to that is AAA. Likewise, if the side-length of one
side of two triangles is equal, we’ll describe that with S. With these
definitions, we’ll make the following claims, as yet unproven, about
the criteria for determining similarity:
 quantitative problem solving in natural resources 95

Triangle Similarity
Two triangles are similar if any one of the following can be es-
tablished:

• AAA. The angles of one triangle are equal to the angles of the
second.

• SSS. The side-lengths of one triangle are equal to those of

the second. Side lengths may be scaled by a constant C if that
constant is the same for each side.

• SAS. Two side-lengths and one angle of one triangle are equal
to those of the second. Side lengths may be scaled by a con-
stant C if that constant is the same for each side.

When you’ve reached the end of this chapter, you should be able to
show how each of these criteria for similarity could be derived from
one of the others. That is left as an exercise for you to work on, one
that can build your intuition for using triangle properties for problem
solving.

9.2.1 The right triangle and sohcahtoa

The mnemonic SOHCAHTOA is one of a small number of things
that most trig students remember years after taking the class. Indeed
this is really helpful way to remember the algorithms relating the
basic trig functions to ratios between the sides of a right triangle4 . 4
As you may know, a right triangle is
But it reveals nothing about the ways that triangles can be employed defined as a triangle with one right, or
90◦ angle.
for practical purposes. So before we deal with these functions, let’s
B
revisit what a right triangle is, where we might encounter one, and a
few terms and rules regarding these beasts. Figure 9.2 shows a nice,
well-behaved one.
Notice that each vertex (a corner point) joins two of the three sides c

a
and doesn’t touch the side that is opposite it. For reasons that might
be apparent in a moment, we choose names for the sides and vertices
that imply a relationship between a vertex and the side opposite A C
b
it. So for example, side a is opposite vertex A (meaning the vertex
A is not one of the endpoints of side a). It probably satisfies your Figure 9.2: A right triangle.
intuition that the size of the angle at a vertex might have some simple
relationships to the length of its opposite side – at least there is a
more intuitive relationship there than between vertex A and one of
the other two sides. Imagine keeping vertices A and C stationary,
but allowing the angle at A to grow. It is plain to see that if the angle
∠ A increases, the vertex B must move up and the length of side a
increases accordingly. This thought experiment produces similar
96 iowa state university

results for the other opposing side/angle pairs as well, and we’ll use
it to our advantage shortly in dealing with triangulation.
One of the most fundamental properties of all triangles is that the
three vertex angles sum to 180◦ (∠ A + ∠ B + ∠C = 180◦ ). For the
special case of a right triangle, the right angle by definition is 90◦ , so
B
the other two angles must be smaller than 90◦ . This seems obvious,
but it has an important consequence: the longest side of a triangle is
se the one that is opposite the largest angle. Therefore, since the right

opposite
te nu
po angle is the largest angle in a right triangle, the longest side (which
hy
we call the hypotenuse) is across from the right angle (Figure 9.3).
In addition to the rule that angles must sum to 180◦ , one of the
A C most powerful properties of right triangles is their adherence to the
adjacent
Pythagorean theorem:
Figure 9.3: A right triangle. a2 + b2 = c2 . (9.1)
This is always true provided that c is the hypotenuse of a right trian-
gle. It turns out that there is a simple modification that can be made
to this if we are dealing with any arbitrary triangle. But before we go
much farther, consider an example setting where a right triangle can
be a useful aid to measurement.

9.2.2 Example: overland distances

5
Universal Transverse Mercator, or If you are visiting GPS waypoints stored as UTM coordinates5 , the
UTM, refers to a projected geographic distance on-the-ground between two points might not be obvious
coordinate system wherein locations
are given coordinates (meters easting, from the coordinate sets. For example, what is the distance between
meters northing) according to their (452632,4660214) and (452991,4660580), the marked locations of
distance in meters east and north of a
predefined datum. The benefit of UTM two observed dickcissel nests? Once we recognize these ordered
coordinates compared with latitude pairs as the geographical equivalent of (x,y) pairs, it is pretty easy
and longitude is that it is an orthogonal to see that the second nest is 452991 − 452632 = 359 m east and
coordinate system analogous to the
Cartesian x-y coordinate system we 4660580 − 4660214 = 366 m north of the first. But neither a dickcissel
sometimes use for abstraction in math. nor an ornithologist would likely go from one nest to another by first
going 366 m north and then 359 m east. Both would be more likely to
go in an approximately straight line. Since east and north are perpen-
dicular, we can construct a triangle like Figure 9.4, with a 359 m long
easting side and a 366 m long northing side to represent the coordi-
nate distances. The as-the-crow-flies distance is the hypotenuse of the
triangle since it is opposite the right angle. Therefore we can use the
pythagorean theorem to find that distance, which we can call d:

d2 = (easting)2 + (northing)2 (9.2)

q
d= (easting)2 + (northing)2 (9.3)
q
d= (359 m)2 + (366 m)2 = 513 m (9.4)
 quantitative problem solving in natural resources 97

q r
Knowing that the second nest is about 513 m away from the first is
great. But if you were to give instructions to a field assistant to walk
513 m from the first nest to find the second nest, that alone is insuffi-
cient information to get to the correct place. Which direction does she
have to go? You could, of course, simply have her walk north 366 m
and then east 359 m, but that wouldn’t be terribly efficient. What is
missing is obviously the direction. If she’s carrying a compass, you
could give her a bearing or compass direction to follow, but what is
that bearing, and do we have enough information to determine it? p
Figure 9.4: Dickcissel nest distances.
North is toward the top of the page.
9.2.3 Angles and azimuths
At this point, we need to draw more important distinctions regard-
ing coordinate systems and conventions. When we need to be more
accurate than simply saying “northeast”, compass bearings are of-
ten given as angles in degrees. Some people prefer to use quadrant 0
30
bearings, where directions are given with respect to deviations from 330
north or south. For example, due northeast might be expressed as

300

60
“north 45 east”, or equivalently N45◦ E. That can be interpreted
as 45◦ east of due north. Similarly, southeast could be S45◦ E and

270
southwest is S45◦ W. This can sometimes be easier to understand

90
in conversation, but bearings expressed in azimuth are less prone
to mis-interpretation. Azimuth is the compass direction in degrees

120
240
clockwise from north, increasing continuously from 0◦ to 360◦ . In this
system, north is both 0◦ and 360◦ , east is 90◦ , south is 180◦ and west
150
210
is 270◦ (Figure 9.5).
180

In the world of mathematics, angles are usually measured counter-

clockwise from the x-axis6 . Not only does this mean that the starting Figure 9.5: The azimuth coordinate
point (0◦ ) is in a different place, but it increases in a different di- system in a compass. Graphic modified
from D. Orescanin.
rection. We will occasionally employ this convention, since it is so 6
In math and physics, angles are also
prevalent in quantitative topics unrelated to geography. But for the frequently measured in radians rather
current problem, we’ll stick with azimuth. than degrees. Note also that many
computer programs that are able to do
Returning to the problem of finding the dickcissel nest, how can trigonometric computations assume
we decide what bearing to give to the field assistant? Since the east- that your argument (or desired result)
will be in radians. Radians make
ing and northing distances are similar, we can be fairly confident
alot of sense for geometry because,
that it will be near 45◦ (NE), but probably not exactly that. But how by definition a radian is the angle
do you determine the unknown angles in a triangle when you only traversed when you traverse a length
along the perimeter of the circle that
know one angle (the right angle) and all of the side lengths? Aha! is equal to the radius of the circle. But
sine and cosine to the rescue!! for practical use in the field, degrees
are easier to work with. So here’s a
quick rule of thumb in case you need
9.3 Angles, circles and sines to convert: once around a circle is 360◦
and 2π radians.

Before we proceed with finding the azimuth, let’s more formally

define a few trigonometric quantities. We’ll do this initially in a
98 iowa state university

mathematics framework, using an x-y coordinate system with an-

gles increasing counterclockwise from the x-axis. Figure 9.6 shows
what we might call the “unit circle”, a circle centered on the point
(0,0) (also called the “origin”) with a radius of 1 unit. If we choose
y any point on the circle, call it p, it lies a distance of 1 unit from the
(0, 1) origin. But its coordinates are not immediately obvious. As with the
case of the dickcissel nests, however, we can break the path from the
p
origin to p into a component in the x direction and a component in
H
O the y direction. Connecting each of those component paths with the
(−1, 0) (1, 0)
θ
x direct path (the radius line), we end up with a right triangle 4OAH,
A
as illustrated in yellow in Figure 9.6.
Since we have a right triangle, we could use the Pythagorean the-
orem again to find an unknown side length, provided that we know
(0, −1) two of the sides. But in this case, we do not know two sides. Instead
we know the angle θ between the x-axis and the line connecting the
Figure 9.6: Right triangle inscribed
origin with point p. We also know that this line, call it H for hy-
inside the unit circle. potenuse, has length 1 by definition. The other two sides, O and A
for opposite and adjacent, are unknown but can be found from the
fundamental trigonometric functions.

Trigonometric Ratios
The sine of an angle is the ratio of the length of the opposite
side (O) to the length of the hypotenuse side (H).
The cosine of an angle is the ratio of the length of the adjacent
side (A) to the length of the hypotenuse side (H).
The tangent of an angle is the ratio of the length of the opposite
side (O) to the length of the adjacent side (A).

The tangent is identical to the ratio of the sine to cosine of an

angle, which you can see is equivalent to the above definition if you
cancel the hypotenuse terms in the ratio of ratios. These are loaded
definitions, so let’s take a few moments to ponder what we’ve just
seen.
• The trig functions are functions in the formal sense: they convert an
input (angle) to a unique output (ratio of side lengths).

• The opposite and adjacent sides are defined relative to the angle
that is the argument of the trig function.

• The output of each trig function is a dimensionless quantity that

represents the ratio of two side lengths.

• If we know one angle (other than the right angle) and one side
length, we can find the two remaining side lengths using the
trigonometric functions.
 quantitative problem solving in natural resources 99

In equations, we don’t spell the full name of these functions, but

instead use sin, cos, and tan as shorthand. With this shorthand and
the definitions above, we can construct a few simple equations that
can help us find the unknown side lengths O and A in Figure 9.6:

O
sin θ = (9.5)
H

A
cos θ = (9.6)
H
Since we know θ and H on the unit circle, we can rearrange these
equations to solve for the unknowns. Multiplying both sides of each
equation by H, we get:
H sin θ = O (9.7)

H cos θ = A (9.8)

Now the reason we have done this in the unit circle is that H = 1,
so we essentially end with the definitions sin θ = O and cos θ = A.
An important thing to realize, then, is that we can scale up to any
arbitrary side lengths. Suppose we were interested in not a triangle
with hypotenuse 1 unit, but one with hypotenuse 55 meters. Defining
f = 55 and isometrically scaling all the triangle sides by that factor,
we can see for example that:
 
fO m
sin θ = . (9.9)
fH m

Of course the f ’s simply lengthen H and O by the same common

factor, and could be easily canceled. But this illustrates the fact that
sine and other trig functions describe dimensionless side-length
ratios, and that those ratios can scale proportionally without changing
the sine, cosine, and tangent of the angles! An obvious implication is
that if one vertex of a right triangle has the same cosine and sine as
another triangle, the triangles can be shown to be similar.
Actually, we already knew that, since by knowing one angle other
than the right angle in a right triangle we can easily find the third.
Recall, then, that one of the criteria for identifying similarity in trian-
gles is AAA, or equality of all three angles regardless of side length.
For any pair of similar right triangles, the only difference in side
lengths is a constant scaling factor f . Thus, any right triangle can be
scaled down to a similar one on the unit circle by dividing all side
lengths by the length of the hypotenuse, so that H/ f = 1!
In addition to simplifying triangle scaling, the unit circle also
allows us to imagine moving point p along the perimeter and ob-
serving how the lengths of A and O change accordingly. In fact, the
100 iowa state university

software Geogebra is perfectly suited for doing this, and I highly

recommend playing around with it to boost your intuition. The key
thing to notice is that the denominator of the side ratios defined as
the sine and cosine is the hypotenuse, or 1 on the unit circle. So the
lengths of the opposite (sine) and adjacent (cosine) sides are the out-
put of those respective functions. For angles between 0◦ and 90◦ ,
both functions range from 0 to 1. If you allow x and y to take on neg-
ative values as point p goes down or to the left of the origin, you’ll
see that both functions remain between −1 and 1, inclusive.
But tangent is a different story. Recall that one definition for the
tangent of an angle is O/A. You can see that for a small angle θ, O
is quite small and A is pretty close to 1, so the ratio of the two will
be nearly 0. O and A are equal and tan θ = 1 when θ is 45◦ , and as
θ approaches 90◦ , A approaches 0 and O approaches infinity, so tan θ
approaches infinity as well. What do you suppose happens as θ gets
larger than 90◦ ?

9.3.1 Example: tree clinometry

One of the most common ways to measure the height of a tree is

with a clinometer. This is a small handheld device with a sighting
lens and crosshair and one of a variety of different mechanisms for
measuring the inclination angle (either positive or negative) of the
sight-line from horizontal. Figure 9.7 illustrates the hypothetical
ne
ht li triangles constructed with vertices at the observer’s eye, and the base
r sig Hu
pe and crown of the target tree. Obtaining the height distribution of
tree

up
merchantable timber in our forest parcel in Problem 3.5 could include
a set of representative height measurements with a clinometer.
θu D
E In measuring the height of a tree, two readings are often taken
θ Hl
lower sig l
ht line with the clinometer: one to the crown θu and one to the base or
stump θl . The observation point E is a pre-determined distance D
Figure 9.7: Clinometry measurement of from the tree itself. From this information, what is the height of the
tree height.
tree?
We identify H as the quantity of interest, and observe that H =
Hu + Hl . As a first step, we must therefore determine Hu and Hl .
We assume the geometry of the problem allows us to construct two
imaginary right triangles as illustrated and that our clinometer gives
us angles in degrees from the horizontal. Since we know the hori-
zontal distance to the tree D and have measured the angles to the top
(θu ) and bottom (θl ) of the tree, we know the adjacent side length (D)
and an angle for both triangles. The target unknowns are the oppo-
site sides of each triangle, and from sohcahtoa we know that we can
use tan to find the opposite sides when we know the adjacent sides.
 quantitative problem solving in natural resources 101

Thus:
Hu
tan θu = (9.10)
D

D tan θu = Hu (9.11)
and
Hl
tan θl = (9.12)
D

D tan θl = Hl (9.13)
and since H = Hu + Hl ,

H = D tan θu + D tan θl (9.14)

H = D (tan θu + tan θl ). (9.15)

Thus, we can use an elementary trigonometric function (tan) and
a bit of algebra to produce a formula that relates clinometer angle
measurements to tree height.

9.4 Arbitrary triangles

While some problems may be approached profitably with imaginary

right triangles, others present triangles without right angles. We’ll
call these arbitrary or general triangles. Triangulation is a typical B
task in which ecologists or wildlife managers might encounter arbi-
trary triangles. When radio-tracking a collared animal, for example,
one method for determining the animal’s location at a given time is
by triangulation from multiple directional antennae. Figure 9.4 is the c
h a
same as Figure 9.1, except that now we are considering the vertices to
be radio transmitters (animals) and receivers (ecologists).
Thus far, we only have two tools that are safe to use with triangles
that lack a right angle: the rule that all the interior angles sum to e g
A C
180◦ (for which we may use the shorthand (Σ∠180) and the general b D

criteria for and implications of similar triangles. These might be of

Figure 9.8: Triangulation in telemetry. B
limited use if our goal is to determine the distance to or location of a is a target radio-collared animal and A
collared animal. But if we know the length of one side (say side b in and C are directional antennae.
Figure 9.4), and a few angles, we can make some progress.

9.4.1 Law of sines

The law of sines is valid for general triangles, including right trian-
gles. If we are careful to define our sides and vertices as we have
(with vertex angle A opposite side a and so on), we can state the law
of sines as follows:
102 iowa state university

Law of Sines

sin A sin B sin C

= = . (9.16)
a b c

Note that this equation is strange in that there are two equals
signs. Don’t get worried, this is just a shorthand that allows us to
say on one line that each ratio between the sine of an angle and
the length of its opposite side is equal to the other corresponding
sine/side ratios. If we wrote each equation with a single ‘=’, there
would be three of them and it would simply take more space. But in
actual implementation you can use any of the implied equalities in
Equation 9.16, such as:
sin A sin C
= . (9.17)
a c
The key concept here is that there is a simple and consistent relation-
ship between the sine of each angle and the length of its opposite
side, and that this applies to all triangles, no matter what size or shape.
That makes plenty of sense right? If you imagine taking a triangle
formed by three knotted rubber bands and lengthening one side
(without changing the length of the other sides), what happens to the
angle opposite that stretched side? It grows right? But to accommo-
date the growth of that angle, the other two angles must get smaller.
The law of sines is especially helpful if you know two sides and one
angle (SSA) or two angles and one side (AAS).

9.4.2 Law of cosines

Another tool that is useful for general triangles is called the “Law
of cosines”. In many ways, it provides the same information that
you can easily find from other tools we have already discussed, so
we won’t derive it or discuss it in great detail. But mathematician
Paul Lockhart makes the case that the law of cosines might be a
misleading name, and that the relationship might be better described
as a modified version of the Pythagorean theorem that is good for all
general triangles. Check it out:

c2 = a2 + b2 − 2ab cos C. (9.18)

As you can see, it is identical to the Pythagorean theorem except that

there is a correction factor 2ab cos C that is subtracted to account for
deviations from a right triangle. As with the Pythagorean theorem,
the law of cosines gives you the third side length if you know the
other two, but you also need to know the angle between the known
sides (SSA). As such, it’s function overlaps that of the law of sines.
 quantitative problem solving in natural resources 103

9.5 Triangle tools: a summary

Trigonometry is a large subdiscipline of mathematics, and can and

does fill more than a semester in math classes. Our treatment here
has focused on the tools that are most commonly encountered in
practical field settings in the natural sciences. Many additional func-
tions, relationships and skills can become important in specific, more
technical applications, but most of these can be derived from the
basic functions discussed here. These functions and properties are
summarized in the table that follows.
Rule name Relationship right general use for
Angle sum ∑ ∠ = 180◦ X X AA known, want A
Pythagorean theorem c2 = a2 + b2 X SS known, want S
Similarity X1 = CX2 X X C known, want X (A,S or other)
sine sin θ = O/H X AS(O or H) known, want S(H or O)
cosine cos θ = A/H X AS(A or H) known, want S(H or A)
tangent tan θ = O/A X AS(O or A) known, want S(A or O)
sin A sin B sin C
Law of sines a = b = c X X AAS or SSA known, want S
Law of cosines c2 = a2 + b2 − 2ab cos C X X SSA known, want S

9.5.1 Example: shoreline waterfowl habitat (Problem 3.1)

Some dabbling duck species like Mallards seem to prefer very shal-
low water. This means that small, shallow wetlands can fit the bill,
but even the shallow shoreline areas of larger and deeper wetlands
may be adequate. Shorelines are also the interface between feeding
and nesting areas for many species, and often support diverse flora
and fauna across the ecotone.
One way we could estimate the extent of shoreline habitat is to
find the length of wetland perimeters, or outlines. If, as described in
the last chapter, we have digitized (or obtained existing data for) the
outlines of wetlands in our candidate parcels, we should have easting
and northing coordinates for these outlines. As with polygon area,
most GIS software will automatically compute the perimeter of any
shape. However, it is instructive to see how this follows from our
earlier discussion of triangle-assisted spatial reasoning.
Recall that when we were traversing between dickcissel nests,
we used an implementation of the Pythagorean theorem to find the
straight-line distance from the coordinates of both end points. We can
write this relationship in an x, y coordinate system as follows:
q
l1→2 = [( x2 − x1 )2 + (y2 − y1 )2 ], (9.19)

where l1→2 is the length of the straight-line distance from point 1

(with coordinates x1 , y1 ) and point 2. Notice that the result is always
104 iowa state university

a real, non-negative number because the differences are squared.

If we have a series of n points describing the wetland outlines, the
sum of all n of the lengths forming a closed polygon approximate the
7
How good is this sort of approxima- perimeter P of the polygon7 . We can generalize this as follows:
tion of the perimeter? This is a simple
question with a not-so-simple answer. q q
For our purposes here, the more points P= [( x2 − x1 )2 + (y2 − y1 )2 ] + [( x3 − x2 )2 + (y3 − y2 )2 ] + ...
we have, the better – particularly if q
we’ve got an automated way to do ... [( x1 − xn )2 + (y1 − yn )2 ]. (9.20)
the computations. However, in a more
philosophical sense this is the crux of
the coastline paradox, first popularized As with the trapezoidal area formula, this can be implemented by
by mathematician Benoit Mandelbrot. hand, in a spreadsheet, or with GIS software.

Exercises

1. Use Figure 9.4 and the right triangles formed by dropping the
perpendicular h, to derive the law of sines from the trig functions
you already know.

2. Explain how you could find the telemetered location of B if you

know the locations of A and C and their internal angles.

3. What happens if you try to apply the law of sines to a right trian-
gle?

4. What happens if you use the law of cosines on a right triangle?

Assume angle C is the right angle.

5. Triangles can be used to measure distant objects, even if we can’t

get to them. This can be used to estimate the height of an object
(e.g., a mountain peak or tree) where we cannot access the base,
and therefore cannot measure the full horizontal distance that
separates us from the object we wish to measure. The general
premise of the remote method is illustrated in figure 9.9. If we
can use a clinometer (a device for determining the angle between
p horizontal and a sight-line to an object of interest) to determine
the sight-line angles to the object of interest p from two different
places b1 and b2, and we know the distance between those places
H
l, we can use trigonometry and algebra to determine the desired
height H.
α β
b1 l b2 Devise a strategy for measuring H from the information gathered
at b1 and b2. Derive and justify a formula that can be used for this
Figure 9.9: Triangle abstraction of the task.
mountain problem.
 Part IV

ALGEBRAIC REASONING
10
Generalizing Relationships

We have encountered many instances in this book where solving a

problem numerically required numbers that we didn’t have. We often
don’t know alll the numbers needed to solve real problems. In some
cases, the simplest way to overcome this issue is to estimate or guess
a value. However, in many other cases, the value of an important
quantity isn’t constant in space or time – our lack of knowledge is
not a reflection of uncertainty in measurement. Instead, there is a
systematic variation in the real value of a quantity and we need to
allow for those changes. Under these circumstances we need to treat
these quantities as variables that have unknown numerical values.
Perhaps we have some idea of how large or small the numberical
values can get1 , but within these limits, the variable can take on any 1
For example, if we’ve used ballpark
value. estimates and deliberately chosen
high-end estimates of some of the
In this new world of uncertainty, we have the tools of algebra at parameters, this could provide us with
our disposal. At least in parts of the problem-solving process, this an approximate upper limit on the
value of the variable of interest.
can be disorienting as we have to carry symbols rather than reducible
values through any operations that we find necessary. However as
we’ll see shortly, performing symbolic manipulations as a means to
solving problems can lead to versatile and reusable solutions. What
we have done prior to now can be called specializing, where we seek
particular numerical values in every calculation when possible. The
alternative, which you’ll recognize as a stepping-stone for algebra, is
to generalize. Heuristic: Express variable quantities
needed for solving a problem as sym-
bolic variables and manipulate them
according to the rules of algebra to
yield general relationships.
Writing algebraic relationships can seem to be hocus pocus
at first. However, the mystique fades a bit when we remind ourselves
that mathematical relationships are little more than formal logical
statements. By carefully assembling what we know about quantities
of interest, striving to sustain generality, and following a few tips, we
can begin to use algebraic reasoning as a powerful tool for creativity
and sense-making.
108 iowa state university

Writing algebraic expressions

• Identify the relevant variables and constants

• Introduce descriptive notation for each quantity

• List what is already known about each variable, using expres-

sions with symbolic notation when possible

• Look for ways to set expressions equal to one another based

on what you know; are there two ways to define the same
quantity using the variables of interest?

• Guess or infer unknown relationships

• Write and simplify equated expressions as a symbolic equa-

tion

• Check for dimensional or unit consistency

When we express and manipulate equations with symbolic vari-

ables, we are doing algebra. When we state systematic relationships
between symbolic variables, we’re using functions. Functions can
describe derived, hypothetical, or observed relationships, depending
upon how we arrive at them.

10.1 Families of Functions

In the natural and environmental sciences, a few families of functions

can be used to describe relationships between key variables of inter-
est. We will explore the most prevalent of these kinds of functions,
examining their algebraic composition and the characteristics of their
graphs. In the chapters that follow, we wil see how functions can be
used to describe relationships between measured variables and how
they can be used to devise mathematical models.

10.1.1 Linear functions

The simplest relationship between two variables – let’s call them x
and y – is perhaps something like y = x. This relationship is indeed a
linear relationship, stating only that y is equal to x without any mod-
ification, or that any change in the variable x results in an identical
change in y. In reality, we will rarely encounter any relationships like
this that are worth describing in an equation. Instead we may often
find that the variables of interest are related through a constant of
proportionality, call it m (might as well stick with the notation we
 quantitative problem solving in natural resources 109

may have seen elsewhere!). In this case, y = mx is still a linear rela-

tionship, but now for any change in x, we expect a change in y that
is m times as large as the change in x. That is what this function does y
y=x
for us: it converts any proposed value of x into a corresponding y
according to the definition of the function. Indeed, the definition of
a function in mathematics is an operation that takes a value of an
y = 0.5x
explanatory or independent variable as input and produces a value
for a unique response or dependent variable as output.
Here’s an example: an elephant’s tusks grow continuously with
age, beginning a bit less than a year after birth. Although there is y = − 14 x + 3
2
x
likely some variability within the population, this relationship allows
biologists to estimate elephant age. Thus, a mathematical description
of this relationship can be written as a linear equation: Figure 10.1: Various linear functions.

l = ra (10.1)

where we are calling tusk length l and age a. Notice that this way of
writing the relationship implies that we are treating a as an indepen-
dent variable (that is, we can think of it as sort of a cause) and l is the
dependent variable (an effect that depends on the cause). Depend-
ing on the circumstances, these roles could be switched. Indeed, it
is easier to measure tusk length than age for a given elephant, so we
might wish to use measurements of tusks to help determine the age
distribution in a wild elephant population.
Also important to remember is that when we are doing science
instead of just math, the variables usually have units and dimen-
sions, which we discussed previously. If we expect our equation to
be meaningful, the dimensions on the left- and right-hand sides of
the equation need to be consistent (i.e. equal). So in the elephant tusk
example r = l/a is a growth rate, must have dimensions of length
per time, or [ L T −1 ] (see how we get that? would that be the same if
we swapped our independent and dependent variables?). If we have
been measuring length in inches and age in years, our value for r
should be in inches per year.
Great! But as we said above, we might wish instead to know the
age as a function of tusk length. So we need to rearrange things a
bit. Let’s now define m as the number of years of age per inch of tusk
length, which is just the reciprocal of r. In other words, m = 1/r,
which also means r = 1/m. Since we’ve just taken a reciprocal here,
the dimensions of m are just the reciprocal of the dimensions of r,
[ T L −1 ] .
Now we can re-state our new linear relationship as l = a/m, or
a = ml. In this form, we have the dependent variable (age) on the
left-hand side of the equation and the independent variable (tusk
length) on the right hand side, as is convention. But at this point,
110 iowa state university

what is implied about an elephant’s age if it’s tusk length is zero (l =

0)? Regardless of the value of m, plugging zero into this equation
yields a = 0. Of course, as mentioned above, adult tusks do not
begin to develop until several months after birth. So our equation is
probably not very good at representing reality (particularly for young
elephants), and is therefore not yet useful. But suppose we were to
change what we mean by “age” on the left-hand side. It makes sense
that what we’re measuring is growth from the age when the tusks
first appeared, so let’s call that age a0 , which is close to 0.5 years. So
the elephant age that we wish to determine is more than we would
have predicted before by the an amount corresponding to the age
when the tusks first appeared, a0 . So our new equation, modified to
account for this correction, reads:

a = a0 + ml (10.2)

In the abstract but precise terms of mathematics, we say that a is a

linear function of l with a slope of m and offset (or y-intercept) of a0 .
Although it seems obvious in the context of this example, the offset
a0 must have dimensions of time for this equation to be meaningful.
Note that the value of the growth rate, or slope m, can be determined
age, a
algebraically by solving the linear equation above for m:

a − a0
m= , (10.3)
l
a1 ( l1 , a 1 )
which we might recognize as the “rise” of the function, ( a − a0 )
divided by the “run” l.
In some cases it is unnecessary, but in others we may need to spec-
a0 length, l
ify something about the domain (a set of upper and lower constraints
l1
on the values of the independent variable) over which a proposed
Figure 10.2: Plot of tusk length l as a
relationship is valid. For example, it doesn’t make sense for an ele-
function of age a. phant to have a negative age any more than it does to have a negative
tusk length. There is probably an upper limit to tusk length as well,
though it is hard to be confident what that might be. To be complete
but conservative, we may specify that the domain of the function as
0 < l < 160 inches. The range for our linear function is the spread
of minimum to maximum values of a corresponding to the minimum
and maximum values in the domain. Note that this last comment
applies to linear functions (though sometimes the signs are reversed),
but for some non-linear functions of interest, maximum and min-
imum values in the range may not correspond to maximum and
minimum values bounding the domain. We’ll see examples of this
later.
Functional relationships that are approximately linear are very
common in the sciences. Indeed, a routine procedure in the analysis
 quantitative problem solving in natural resources 111

of multivariate data is linear regression, wherein the coefficients

(slope and intercept) of a linear function that best fits the data are
sought. Linear functions – or nonlinear ones for that matter, as we’ll $
see below – can also be postulated hypothetically in the construction
C
of mathematical models.

10.1.2 Example: fire suppression costs (Problem 3.5)

The issue of how much suppression effort to use is at the heart of this
person hours
problem, so it’s clear that suppression effort should be considered a
Figure 10.3: Schematic illustration of a
variable. As is routine with variable quantities, we should assign a
hypothetical linear relationship between
symbol to the variable and decide, at least for now, what units it will the cost of fire suppression and the
be quantified with. A single symbol is preferable (though subscripts number of person hours of suppression
effort.
are permissible if necessary) to prevent any ambiguity. Therefore,
let’s choose the symbol E for effort, and provisionally assign the
units of person-hours. A person hour has dimensions of [1 T ]. Like
acre-feet and other similar units, person hours is compound unit that
we should understand as the number of hours worked per person
multiplied by the number of people. For example, if two people work
8 hours each, that effort represents 16 person hours.
Now we need to deal with the other variable that is implied in this
part of the problem: cost. First, suppose we define C as the symbol
we’ll use, and US dollars as the unit of cost. Relating the cost of
suppression to the effort requires some way of assigning a cost per
unit of effort. Recalling our choice of units, this cost-per-unit-effort
will have units of dollars per person hour, which sure sounds like an
hourly wage. In fact, that’s exactly what it is! So let’s call it w. We
can now state the algebraic equivalent of the sentence “suppression
cost equals the number of person hours of effort times the hourly
wage”. In symbols:
C = wE (10.4)

This equation is illustrated in Figure 10.3 as a straight line increas-

ing from left to right. The slope of the line, analogous to m in our
abstract concept of the prototypical linear function, is w, and the
y-intercept is zero. This latter observation simply articulates the
(hopefully obvious) notion that the cost of zero person hours of labor
should be $0.

10.1.3 Polynomial functions

A polynomial function is one in which the dependent variable also
depends on the independent variable raised to an exponent. Poly-
nomials are among those functions that can have multiple ups and
downs in the dependent variable over the domain of the function. To
112 iowa state university

refresh your memory, let’s write an abstract polynomial equation for

starters:
y = b + mx + lx2 + kx3 + ... (10.5)
Here we have a function in which the quantities added together on
the right-hand side have dependence on increasing powers of x, and
the way we’ve written it we imply that the equation could go on in-
definitely, incorporating ever-growing powers of x as we go. One way
to describe a polynomial is by its order, which is nothing more than
the integer values of the exponents of x included. If we take away the
“. . . ” from the equation above and just stop the equation after kx3 ,
y this would be a third-order polynomial, since 3 is the highest expo-
y = x2
nent of the independent variable x. Sometimes you will see the term
cubic for third-order polynomials, while the term quadratic is used for
y = 31 x2
second-order polynomials. To be complete, we can even pretend that
the first term on the right-hand side b is really bx0 , representing a
“zeroth order” term.
Suppose that we got rid of the l and k terms in the equation above,
y = − 81 x2 + 2
which we could simply do by saying that l = k = 0. What’s left is
x
just the linear function we had above, and we see now that the linear
function is really a special case of a polynomial function, a “first-
Figure 10.4: Polynomial functions. order polynomial”. Likewise, if b = m = k = 0, we’re left with just
y = lx2 , a second order or quadratic polynomial. It is still a second-
order polynomial if m and b are nonzero. So you see that writing the
equation as we did above allows us to imagine a polynomial of any
order that we choose, and keeping or discarding any terms we wish
by adjusting the lettered coefficients.
So how do these polynomial functions differ from linear functions?
Take as an example the formula for the surface area of a sphere –
perhaps representing a raindrop: As = 4πr2 . A simple linear func-
tion as described above has a single independent variable and the
values of the dependent variable depend only on the first power of
the independent variable and a constant of proportionality. We can
write the surface area equation as A = (4πr )r, and now it looks
like we only have the first power of r. Great, but now our constant of
proportionality contains r, so it is not a constant at all but a variable
itself. So nonlinear functions are those that cannot be written as a
relationship between the dependent variable and the first power of
the independent variable times a constant constant of proportionality.
Before we waste too much more time talking about polynomials,
I need to be clear on one thing: When we encounter polynomials in
most undergraduate mathematics classes, we are only considering
functions where the powers of the independent variable are whole in-
tegers. With this in mind, it is worth thinking about whether they are
really useful for us. What relationships depend on integer powers of
 quantitative problem solving in natural resources 113

the independent variable? One area where these polynomials are use-
ful in natural science is spatial measurement. you probably remem-
ber that the areas of squares and circles each depend on the second
power of a characteristic length (side or radius). Likewise, volumes of
spheres and cubes depend on the 3rd power (coincidence?). While we
may never encounter perfect spheres and cubes in the natural world,
we may find occasion to idealize the size and shape of something
(like a sand grain, egg or raindrop as a sphere, a tree root or snake as
a cylinder, etc.) in a simple model so that we can better understand
something about it.
Likewise, some physical phenomena can be described with equa-
tions that depend on a whole number power (often 2) of time or
position. In more complicated problems in the real world, it can also
be advantageous to approximate an unruly function using a so-called
series expansion of the function, which often amounts to a polynomial.
These examples notwithstanding, true polynomial functions do
not arise as commonly in the natural sciences as linear and some
other non-linear functions do2 . An important exception, which we’ll 2
But as we’ll see below, there are
grapple with quite a bit later this term, is the so-called logistic or certainly relationships in the natu-
ral sciences where the relationships
density-dependent growth function. In ecology, this function de- between variables are best described
scribes the theoretical growth of populations constrained by limited with functions that have non-integer
exponents.
space or resources. We can write the basic relationship as:
r 2
G = rN − N , (10.6)
K
where the dependent variable G is the population growth rate y = x3/2
[1 T −1 ], the independent variable N is the number of individuals,
r is a growth constant and K is the carrying capacity. This function is y

quadratic because N 2 is the highest power term.

10.1.4 Power functions y = x1/2

In addition to linear and polynomial functions, it is relatively com-

mon to encounter at least four other classes of functions in the nat-
ural sciences. Power functions arise commonly in ecology and ge-
x
ography, especially in scaling properties of organisms and habitats
in space. Power functions may include any function in which the
independent variable is raised to an arbitrary exponent, of the form: Figure 10.5: Examples of power func-
tions.

y = ax b (10.7)

The power function differs from a polynomial in that the exponent

on the independent variable is not constrained to be an integer. Fig-
ure 10.5 compares the appearance of power functions with exponents
greater than and smaller than 1.
114 iowa state university

In the biological subdiscipline of island biogeography, a relation-

ship between island area Ai and species diversity S has often been
described with a power law:

S = cAiz (10.8)

where c is a fitting parameter and z is an exponent that is usually less

than 1.
Another example of a power function appears in the description
of what hydrologists call a stream’s “hydraulic geometry”, which de-
scribes how the width, depth and average velocity of a river change
3
One of the scientists who developed in time and space3 . Channel width w, for example, typically increases
and popularized this concept was Luna downstream in a way that can be described as w = aQb , where Q,
Leopold (1915-2006), the second son of
Aldo Leopold. the water discharge, is our independent variable and a and b are
empirical constants. Channel depth and velocity are described with
y = ex
analogous relationships.
y

10.1.5 Exponential functions

Exponential functions are different from power functions in that the
independent variable appears as part of the exponent, rather than the
base. A generic exponential function might look something like this:

y = e− x y = abx . (10.9)
x

The base may often be e, which is an important (but irrational like π)

Figure 10.6: Examples of simple expo- number close to 2.718, but needn’t be. Exponential functions describe
nential functions. ever-increasing or ever-decreasing change, and appear in contexts
like the decay of radioactive substances or unrestrained growth of
populations. The radioactive decay equation might look a bit like
this:
N = N0 e−λt (10.10)

A similar form describes the extinction (attenuation) of sunlight with

depth in a water column or forest canopy according to the Beer-
Lambert law:
I = I0 e−kd (10.11)

where d is the independent variable. In addition to natural growth

and decay phenomena, exponential functions appear extensively in
economic analysis.
A somewhat more complicated form of exponential function is
sometimes used to describe growth of individuals (fish, trees, etc.)
over time. The Von Bertalanffy growth function (VBGF) can be writ-
ten:
L t = L ∞ [1 − e − K ( t − t0 ) ]. (10.12)
 quantitative problem solving in natural resources 115

In general, when the exponential argument is a negative number,

these functions describe decay or asymptotic approach toward a
limiting value. However, when the argument is positive, exponential
functions describe explosive growth.

10.1.6 Logarithmic functions

Closely related to exponential functions are logarithmic functions.
The natural logarithm, sometimes written ln, is the inverse func-
tion of e, meaning that ln (e x ) = x. The base-10 logarithm, written
log10 or simply log, behaves in a similar way but for exponential
functions with base 10. So log10 (10x ) = x. Both types of logs, and
logarithms with any other base, are functions that increase rapidly
for low values of the independent variable, but increase ever slower
thereafter. We will find logarithms especially useful in transforming
data that we suspect might be a power or exponential function, and
must therefore have a basic command of the algebraic rules that ap-
ply to them. Outside of transformations and inverting exponentials,
however, we won’t encounter logarithms extensively.

Exercises

1. Given the description of species-area relationships given in Sec-

tion 10.1.4 and the notion that the exponent z in Equation 10.8 is
less than 1, describe what this means conceptually. How does the
species diversity change with island area, and how does an incre-
ment of area change affect small islands differently than larger
islands?

2. Using only symbolic variables and constants, write an expression

that defines that time necessary for 95% of a radioactive isotope to
decay. Hint: interpret this to mean that we seek an expression for t when
N/N0 = 0.05.

3. Review the description of Problem 3.2. Write a hypothetical, but

well-justified, algebraic equation relating the volume of herbicide
needed to eliminate woody shrubs, and the basal area per unit
land area of those shrubs. Consider all quantities to be variables,
so use symbols rather than numbers for this.

4. Review the description of Problem 3.1. Using reasonable geomet-

ric idealizations (not computer algorithms), can you write a simple
algebraic equation that relates the length of a wetland’s perimeter
habitat to the wetland’s area?
11
Relationships Between Variables

In the previous chapter, our discussion of variables and functions

largely assumed that relationships were known or developed inde-
pendent of any measurement or data. However, functional relation-
ships between variables can also be derived from data. Here, we
explore two concepts that help us understand the strength and nature
of systematic relationships between variables.

11.1 Correlation

In common parlance, the word correlation suggests that two events

or observations are linked with one another. In the analysis of data,
the definition is much the same, but we can even be more specific
about the manner in which events or observations are linked. The
most straight-forward measure of correlation is the linear correlation
coefficient, which is usually written r (and is, indeed, related to the r2
that we cite in assessing the fit of a regression equation). The value
of r may range from -1 to 1, and the closer it is to the ends of this
range (i.e., | r |→ 1), the stronger the correlation. We may say that Figure 11.1: Correlation between the
two variables are positively correlated if r is close to +1, and nega- maximum lifespan and gestation period
of various mammals, r = 0.73.
tively correlated if r is close to -1. Poorly correlated or uncorrelated
variables will have r closer to 0.
In the margin are two plots comparing life-history and repro-
ductive traits of various mammals. In the first one, Figure 11.1, the
arrangement of points in a band from lower left to upper right on
the graph is relatively strong, corresponding to a relatively high r of
0.73. In contrast, the correlation between litter number per year and
litter size in Figure 11.2 is (surprisingly?) weak, producing more of a
shotgun pattern and r a modest 0.36.
In the abstract, the mathematical formula for the correlation be-
tween two variables, x and y, can be written:
1 n xi − x̄ yi − ȳ
  
r= ∑ , (11.1)
n i =1 σx σy
Figure 11.2: Correlation between the
number of litters per year and the litter
size of various mammals, r = 0.36.
118 iowa state university

where the subscript i corresponds to the ith observation, the overbar

indicates mean values, and σx and σy are standard deviations. The
specifics of this formula are not of great interest to us. The important
thing to understand is that when positive changes in one variable
are clearly linked with positive changes in a second variable, this
indicates a good, positive correlation, r > 0. The same is true if
negative changes in one variable correspond to negative changes in
the other. However, positive changes in one variable corresponding
to negative changes in another indicate negative correlation, r < 0.
It is also important to note that this is a good measure of correlation
only for linear relationships, and even if two variables are closely
interdependent, if their functional dependence is not linear, the r
value will not be particularly helpful.
Nevertheless, correlation can still help us identify key relation-
ships when we first encounter a dataset. Consider the changes in
weather variables measured at a meteorological station as a function
of time. Weather data can be very overwhelming due to the num-
ber of variables and the sheer volume of data. One handy way to
isolate some of the strongest interdependencies among variables of
interest is to look for correlations. A correlation matrix plot is es-
sentially a grid of plots where each variable is plotted against all the
other variables in a square array of panels. Relationships with strong
positive or negative correlations immediately jump out, suggesting
which relatinships we might wish to investigate further. For exam-
ple, let’s look at a month-long weather dataset downloaded from
www.wunderground.com.
There is alot of information in these plots, so let’s look at them
piece by piece. Notice that the panels on the diagonal from upper left
to lower right would be a variable plotted against itself (r = 1), and
they are therefore replaced by a density distribution for each variable.
Also notice that since the upper right half would be a mirror image
of the lower left, there is just a number in each of those panels rather
than a plot. In any case, here we have just selected four variables of
potential interest, and you can immediately see that there is a strong
positive correlation between mean temperature and mean dew point,
with r = 0.962. The strength of the correlations from these plots
(the six in the lower left) is indicated by the correlation coefficient
in the (mirrored) corresponding panel in the upper right. There are
also relatively strong negative correlations between temperature
and pressure, and dew point and pressure. In contrast, we see weak
correlations between humidity and temperature and humidity and
pressure, as indicated by the low r values.
The important thing to remember from all of this is what the cor-
relation coefficient can tell us: a high, positive correlation between
 quantitative problem solving in natural resources 119

Figure 11.3: Correlation plots for

weather data from Ames, IA, April
2014.
120 iowa state university

two variables indicates that when one goes up, so does the other. A
high negative correlation indicates that as one goes up, the other goes
down. Low correlation coefficients indicate that a consistent linear
relationship cannot be established. If correlation is established, how-
ever, this analysis doesn’t yet provide details about the functional
relationships present.

11.2 Regression

Regression is the process of fitting a mathematical function to a set of

data points using some criterion for judging “goodness-of-fit”. The
resulting “best-fit” function may then be used to predict unknown
values, to forecast future values, or to evaluate the dependence of
one variable upon another. Goodness-of-fit can be determined by one
of many statistical techniques that determine how well a function
describes the variations in the data used to generate it. The most
common criterion for goodness-of-fit is called “least-squares”, so
you might sometimes see the whole process called least-squares
regression. Least squares means what it sounds like, sort of. When
a function (let’s write it y = f ( x )) is tested for goodness-of-fit, the
difference between the y-values predicted by the test function, each
of which we can call ȳi , at a given xi , and the yi -values in the data
set are found, squared, and added together for the entire dataset.
The best-fit line is then the one for which the sum of the squares of
the residuals are minimized (least). This is very commonly done for
linear equations, but we can use the same techniques for nonlinear
equations as well.
Figure 11.4: Schematic representation of Some data sets that we may encounter just don’t appear to have
the quantities involved in finding best-
fit functions by least-squares regression. linear trends though. In these cases, we can try transforming one or
1
Some common data transformations both variables1 or we can attempt to perform nonlinear regression.
include logarithmic, exponential, and As with many of the statistical and spatial methods discussed in this
reciprocal. In these transformations,
a modified variable is created by book, the heavy lifting for most of these options can – and probably
performing the selected operation should – be done with computer software. However, we should still
on the original variable values.
cumulative catch catch/effort be aware of what is happening
86 2.46
137 1.76
169 1.14 11.2.1 Example: brook trout electrofishing (Problem 3.7)
178 0.29
Having isolated the age-0 brook trout from each electrofishing tra-
verse and computed the catch per unit effort cue of that subset, we
may now employ the Leslie method to estimate the total population
of age-0 brook trout in the study reach. In this method, we create a
dependent variable ccumul. corresponding to the cumulative num-
ber of fish removed in each pass, the “cumulative catch”. We then
plot and perform a linear regression of the catch per unit effort as a
 quantitative problem solving in natural resources 121

function of cumulative catch, as illustrated in Figure 11.5.

By the Leslie method if we extrapolate the best-fit line to a vertical-
axis value cue = 0, the cumulative catch value where that occurs
is the estimated total population. This value can be estimated from
the graph itself, but the result is better if we solve the for the value
directly from the best-fit line. The equation of the best-fit line for this
regression is:
cue = −0.0208ccumul. + 4.38. (11.2)
Note that the slope of this line (−0.0208), consistent with intuition, is
a negative number. The y-intercept 4.38 corresponds to the hypothet-
ical initial catch per unit effort at the very start of the first traverse. Figure 11.5: Catch per unit effort as a
Rearranging and solving for ccumul. gives function of total catch for age-0 brook
trout, from Table 11.2.1.
ccumul. = 210.6 − 48.1cue (11.3)

and we find that the estimated total population is 210.

Exercises

1. Discuss in a paragraph the benefits and drawbacks of deciding,

prior to any data analysis, what type of function to seek best-fit
parameters for.

2. In Section 11.2.1, we skipped several steps in the algebraic ma-

nipulation that allowed solution for ccumul. . Carry out all the in-
termediate steps, showing your work completely, and determine
whether the solution cited above is acceptable.

3. Find a dataset that interests you within a public ecological or nat-

ural resource data repository2 , identify variables within a dataset 2
For example, browse the Global
that may be related, and perform a regression to see the nature of Registry of Biodiversity Repositories.

that relationship.
 Part V

MODELING
12
Modeling

12.1 What is a model?

A model is a representation of reality that allows us to understand

something better. There are many types of models, including con-
ceptual, mathematical, and physical models. A physical model is a
physical object or set of objects intended to represent something else
that is too large, small, complex or otherwise inaccessible for direct
investigation. A conceptual model is a collection of hypothesized
relationships between different objects or variables, and is usually
described in narrative. From an early age, we learn how to construct
both physical and conceptual models. Children create conceptual
models to help them understand cause and effect relationships that
lead to either desirable or unwanted outcomes (‘if I jump down one
or two steps, it’s fun, but if I jump down three or more steps it hurts
my legs: jumping farther hurts more’). When my gradeschool son
builds a spaceship from Legos, he is creating a physical model of a
spaceship he has seen in a movie or book. These are not particularly
sophisticated models, but they are nevertheless ways of representing
some aspect of reality (or imagined reality).
As with Legos, mathematical models can serve mostly a desire
for creative play. Like Lego models, it is perfectly possible to create
a mathematical model that represents reality poorly, and is therefore
not very useful. Perhaps we claim to have created a model of a car,
but if we’ve only stacked rectangular bricks together and failed to
add wheels, it is not a particularly good or useful model of a car.
Thus, model construction and use should be done with the broader
problem context in mind. The means should justify the desired ends. Heuristic: Mathematical models are
In this book, we are interested in mathematical and conceptual only as useful as the conceptual models
on which they are based.
models and the connections between them. Ultimately, our goal
isn’t necessarily to become mathematical modelers, but rather to be
able to construct, use, and understand models that can assist with
problem-solving. Indeed, many mathematical models originate from
126 iowa state university

a desire to quantify the relationships in a conceptual model devised

to address a problem. Several possible approaches to quantification
lead to a handful of varieties of mathematical models. We’ll focus
our discussion on three distinct but related types of mathematical
models that differ in their origins and implementation. The first two
are grounded in theory, while the third often arises from statistical
data analysis.

• Analytical models are usually developed from theory based on

fundamental physical, chemical or biological principles. A hypoth-
esis that a tree’s height should scale with it’s trunk diameter raised
to the 2/3 power in order to retain structural integrity is such a
model. These models are often the most general and abstract, and
can sometimes be solved with paper and pencil. However, they
can become hopelessly complex and un-solvable when one tries to
incorporate realistic details and context. The idealizations neces-
sary to make an analytical model solvable can also sometimes limit
its utility.
• Numerical models may be created and motivated in the same
manner as analytical models, but employ techniques for mathe-
matical approximation that permit relaxation of analytical ideal-
izations and introduction of detail without making the equations
too difficult to solve. Numerical models can be solved by hand for
very small systems, but are more appropriately implemented in
computer programs.
• Empirical models may have analytical or numerical components,
but contain parameters that must be quantified by experiment or
systematic observation. Data must be incorporated and usually
analyzed statistically in order to define parameter values. In some
cases, regression is used to constrain the functional relationships
between variables or to identify the value of coefficients. Thus, a
fully empirical model is data-driven or data-calibrated.

We have already seen or worked with a few examples of mod-

els. The Logistic population growth model that we discussed briefly
in Section 10.1.3 is a theoretical model that can be implemented ei-
ther in numerical or analytical form. Even that model, however, has
empirical components, since it’s use in practical problem-solving re-
quires some observational constraints on r and K. When we solved
for total brook trout population in Section 11.2.1, we employed an
empirical model known as the Leslie method, which is based on a
conceptual model of the change in catch probability under declining
population.
 quantitative problem solving in natural resources 127

12.1.1 Example: The Universal Soil Loss Equation (USLE)

The widely-used Universal Soil Loss Equation (USLE) is an example
of an empirical model. The master equation for USLE is:

A = RKLSCP (12.1)

where A is the soil loss (usually in tons/acre/year), R is a rainfall-

erosivity factor, K is a soil erodibility factor, L and S are the slope
length and angle factors, C is a ground-cover factor and P is a param-
eter that accounts for soil conservation practices or structures.
The factors in USLE are quantities whose values cannot be mea-
sured directly. Instead, the numerical values are each derived from
a combination of carefully-designed field experiments where all but
one factor is held constant. The factor values are then derived from
measured differences in soil loss.
The great value of the USLE and it’s kin is that it is sufficiently
easy to use that farmers with little formal training in math or com-
puting can easily get satisfactory results. Most factor values can
either be looked up in tables or measured on the ground or from
maps.
The ease of use comes at a cost, however. Because factor values
are derived from experiments, they are strictly valid only within
the range of conditions considered within the experiments. In other
words, if applied in settings where – for example – rainfall intensity
is twice as large as the largest observed in experiments, the reliability
of results is uncertain. Fully empirical models can therefore some-
times be unreliable in conditions outside the range of the conditions
under which factor values were determined.

12.1.2 Example: probability of deer-automobile encounters (Problem 3.3)

As we have already seen, simple theoretical models can sometimes
be sufficient to explore a range of system behaviors, even when func-
tional relationships are uncertain. These models will inevitable by
limited in power by the simplifying assumptions or idealizations
used, but when the science or management problem permit a solu-
tion with substantial uncertainty, this approach is still warranted.
Let’s assume that deer in our county are randomly distributed in
space, and that they have no particular reason to either avoid or seek
out roads. Call the total area of the county Ac and the proportion of
the area occupied by roads f , so that the area of roads Ar = f Ac .
Let’s assume that there are N0 deer in the county. It follows that – if
the deer are randomly distributed – there will be approximately f N0
deer on the road at any moment. What is that number according to
the numbers we produced earlier for Story County, IA? The value of
128 iowa state university

f was estimated to be approximately 0.0076, so if there are say 1000

deer in the county, we should expect either 7 or 8 of them on the road
at any given time. That seems reasonable, but that isn’t what we’re
after. We’d like to know about how likely collisions are between
deer and automobiles. So we need to work in something about the
number and distance of car trips through the road system, right?
This is left as an exercise for the student, as there are many possible
ways to approach this.

12.2 Dealing with higher mathematics

Many powerful mathematical models have been devised to explore

and describe phenomena in nature. Some of the most powerful are
those that allow predictions of unobserved or future events or pat-
terns. These can directly inform management decisions provided
that managers trust and understand their results. Unfortunately,
many of these powerful models employ mathematical concepts and
methods that are beyond the typical undergraduate training in math.
Does that mean that most people are doomed to never understand
or use these models? Absolutely not! There isn’t any inherent rea-
son that students need to take calculus, linear algebra, or differential
equations courses before they can comprehend the gist of a model
constructed with those skills. It certainly helps to have at least a
conceptual grasp of some key concepts in calculus, but that doesn’t
translate to a pre-requisite.

12.2.1 Example: prairie dog plague (Problem 3.4)

Since this problem deals with hypothetical future events, it may

not be possible to glean the answer directly from past work or from
observation. Instead, we can construct a simple model of the prairie
dog community with random, probabilistic interactions among well-
mixed individuals.
A common way to model disease transmission is with a compart-
ment model often called SIR. We consider individuals in a popula-
tion to be in one of three (or four) states: Susceptible (S), Infected (I),
and Removed (R) or Recovered. Individuals move from compartment
S to compartment I by disease transmission. Infected individuals in
compartment I then either recover and move to compartment R, or
are removed from the population by death or isolation. These trans-
fers between compartments are often described with a system of
 quantitative problem solving in natural resources 129

differential equations:
dS
= − βSI (12.2)
dt
dI
= βSI − γI (12.3)
dt
dR
= γI (12.4)
dt
These differential equations are not easily solved in most cases,
but we can use them as a basis for a numerical simulation of disease
dynamics if we are able to estimate the parameters β and γ. A nu-
merical representation of the first equation might look something like
this, for example:

St+1 = St − βSt It (12.5)

It+1 = It + βSt It − γIt (12.6)
Rt+1 = Rt + γIt (12.7)

This says that in a given time increment, susceptible individuals are

moved from the S compartment to the I (infected) compartment at a
rate that is proportional to the product of the numbers of individuals
in each compartment and the transmission rate constant β. You can
see in the first and second equations above that when a number
of idividuals infected according to the βSI term in lost from the S
compartment (because it is negative), it is gained (positive) in the I
compartment. All individuals are accounted for in moving into or
out of the I compartment. Similarly, individuals move from the I
compartment to the R compartment at a rate governed by the rate
constant γ. Selection of these rate constants to a large extent governs
the behavior of the model, and thus the predicted fate of the prairie
dog colony. But implementing management options informed by
positive model outputs is where the biggest challenge arises.

12.3 Power-Law Scaling

Consider this seemingly innocuous question: are larger

animals heavier than smaller animals?
You: Hmmm, well, yeah I think so?! An adult bear weights more than
a snowshoe hare, for instance.

OK, great, but how would we know if this is true more generally?
And what exactly do we mean by larger? Does that mean taller?
Larger volume? This brings up a few issues that become important
when we’re talking about real quantities rather than abstract vari-
ables. Unambiguously defining quantities can be an important first
130 iowa state university

step in communicating quantitative information. In the next section

we’ll be specific about what information is required to fully define
a quantity. For now let’s agree that we’re satisfied with relating the
mass of an animal to its volume. Do animals that take up more space
(i.e., have greater volume) also weigh more? Maybe we can say it
another way: is the weight or mass of an animal proportional to its
body size? We could write this in symbols:

M ∝ V? (12.8)

The symbol ‘∝’ between M (body mass) and V (volume) means

“proportional to”. So this isn’t an equation yet because we’re not
sure anything is equal. And of course it’s nonsense that an animal’s
weight is equal to its volume. There must be some other parameter
that transforms an animal’s volume into a mass. Let’s call it c, and
try it out in an equation:
M = cV (12.9)

But what is c? As we said above, we’d prefer to have some meaning

for the symbols we throw around in equations. Let’s use one of our
old algebraic tools for manipulating equations and “solve the equa-
tion for c”. By that we mean get c onto one side of the equation all by
itself. To get there, we just need to divide both sides of the equation
by V, yielding:
Mass

M
=c (12.10)
V
Now recall that the definition of density is mass per unit volume.
Volume
That’s exactly what we have on the left-hand side of the equation! So
Figure 12.1: Plot of some hypothetical our equation now says that c, the parameter we used to transform
measurements of animal mass and volume into mass, is the same as density! So for an individual ani-
volume.
mal, the parameter that relates mass to volume is density. As we have
done previously, let’s assume that most animals have a density close
to that of water so this proportionality parameter c doesn’t vary sig-
nificantly among species. So to the extent that it is correct to say that
most animal’s body density is close to that of water, we can argue
that larger animals do indeed weigh more, in general.
1
An entertaining and well-composed This is probably not a very profound revelation to you1 . But with
article on some not-so-obvious conse-
only a few more small leaps in logic, we can get somewhere consider-
quences of size differences in animals
is On Being the Right Size, byt J.B.S. Hal- ably more interesting. For more than a century, biologists have been
dane, published in Harper’s Magazine, intrigued by a remarkable relationship between the basal metabolic
March 1926.
rate and body mass for animals of a wide range of sizes and shapes.
Amazingly, if one assembles a large set of data and plots it on a
graph with a logarithmic scale, mice, humans and elephants and
most of the rest fall along a straight line! An equation that describes
 quantitative problem solving in natural resources 131

this relationship and the line on the graph looks like this:

B = B0 Mb (12.11)
where B is the basal metabolic rate, M is body mass as before, and
B0 and b are constants (we’ll see what they mean later!). This equa-
tion is yet another power law, and equations with this form pop up
surprisingly often in ecology once you start looking. We’ll get more
into functions and power laws later on. But for now, some important
points should be made:

• The argument that there should be a proportionality between body

mass and metabolic rate was originally conceived theoretically on
the basis that energy given off by an animal to its surroundings
might depend mostly on the animal’s surface area, while its mass
scales with volume.

• Measurements by many researchers over more than a century have

been compared against this theoretical prediction, with varying
degrees of success. In most cases however, the power-law relation-
ship holds.

• By comparing theoretical predictions with real data, one can dis-

cover truly novel and interesting things about physiological sim-
ilarities or differences between different organisms – insights we
might not have ever developed without the quantitative analyses.

We’ll look into this in more detail a bit later.

13
Models of growth and decay

Some of the most well-known applications of quantitative analy-

sis in the life sciences relate to describing changes in processes or
ecosystem properties with time. Among the most important exam-
ples is population change, where the number of individuals N in a
population is expressed as a function of the independent variable t: y = ex
N = f (t). In this chapter we will explore two types of exponential y
functions and a polynomial function that form the basis for describ-
ing and predicting population change and a lot more.

13.1 Exponential functions & population models

An exponential function is one in which the independent variable ap-

pears in the exponent, or power, of some other quantity. The equation y = e− x
y = a x is an example of a simple exponential function if x is the in- x

dependent variable and y is the dependent variable. In this case, the

constant a can be called the base, since it is the quantity that is raised Figure 13.1: The typical ever-changing
to a power. From our high school math classes, we learned about growth and decay of the exponential
function.
exponential and logarithmic (the inverse of exponential) functions
mostly with bases of 10 and e, where e is Euler’s number (∼ 2.718)
and is sometimes written exp(something). But we can have an expo-
nential function with any arbitrary base.
Exponential functions arise frequently in economics, physics, and
in some contexts in ecology. Imagine, for example, a population of
marbled murrelets in a coastal bay in the Pacific Northwest1 . At 1
Why murrelets you might ask? As
some time, suppose their population was 100 individuals. With time, you’ll see shortly, it is convenient to
begin with “simple” populations,
this can change as individuals die or reproduce. If we assume no where the causes of population changes
murrelets emigrate or immigrate (are added to or subtracted from the estimated from visual surveys are
limited.
population), changes in population with time are controlled only by
birth and death rates, and we can say the population N after one year
is:
N1 = N0 + B − D (13.1)
In this equation, we take N0 to be a constant, initial population. The
134 iowa state university

birth and death rates may scale with the population, such that they
can be represented like this:

B = b × N, D = d×N (13.2)

where b and d are birth and death rates per individual. So, for exam-
ple, if the birth rate is approximately 0.15 individuals per murrelet
2
Note that this birth rate is given per per year2 , and death rate is 0.05 individuals per murrelet per year, we
individual. Obviously males cannot can write our equation for population as:
give birth to offspring, so a better way
to express fertility or fecundity is in
terms of birth rates per female; however N = N0 + 0.15N0 − 0.05N0 (13.3)
the per individual or per capita birth rate
is easier to work with. If we simplify the right-hand side of this, we have N after one year as
a simple function of N0 :

N = (1 + 0.15 − 0.05) N0 (13.4)

N = 1.1N0 (13.5)
If you plug in 100 for N0 , this gives us an unsurprising result that
population is 110 murrelets. This makes sense, since we get 0.15 ×
100 = 15 births and 0.05 × 100 = 5 deaths during that year.
Now if we project into future years (where t is the number of
years after our initial measurement of population N0 ) with the same
relationship, we’ll see that after another year of births and deaths,
we’ll get:
Nt=2 = 1.1(1.1N0 ) (13.6)
where the quantity in parentheses is the population after one year,
now incremented by another series of births and deaths. We can
rearrange that equation slightly to yield:

Nt=2 = N0 × 1.12 (13.7)

After another year, we’ll get:

Nt=3 = N0 × 1.13 (13.8)

And by now you probably see the pattern. If t is the number of years
after an initial population census N0 , our projection of population is:

Nt = N0 × 1.1t (13.9)

Interpreted as N as a function of t, this is an exponential function

with a base of 1.1 and a constant N0 . Note that a very similar func-
tion could describe compounding interest on a loan, savings account
or credit card balance, if the principal (the amount saved or bor-
rowed) remains unchanged over time.
 quantitative problem solving in natural resources 135

We could have written our equation above a bit differently. In-

stead of keeping a constant reference to N0 , we could have said that
population next year depends only on the population this year and
the birth and death rates this year. This alteration would give us:
Nt+1 = Nt + B − D = Nt + Nt (b − d) = Nt (1 + r ) (13.10)
where r = b − d can be defined as the population’s intrinsic growth
rate. There is no difference in the result of this equation if we apply
the same assumptions and constraints as we did in the first version,
but this form of the equation is a bit more versatile. It will also be-
come useful to us in a few days. We can call it a discrete difference
equation.
Before we move on, notice a few things about our population
model. First, population is unrestrained. The only factors influencing
the growth rate are birth and death rate, and these are considered
constants. In reality, these might not be constant as individuals com-
pete for limited resources. Alterations to this model to account for
this fact will be introduced next time. Also, notice that the intrinsic
growth rate r is positive because we have said that the birth rate is
higher than the death rate. It is, of course, possible for the reverse to
be true: death rate could be larger than the birth rate, and the result-
ing r would be negative. As you can see from the above equations, a
negative r would result in an exponential decrease in population with
time.
When r = 0, we may say that the growth rate is zero and births
balance deaths. The birth rate that balances death rate is sometimes
called “replacement”, since it replaces each death with a birth.

13.1.1 More exponentials

One place where exponential functions appear in the natural sciences
is in animal physiology, particularly where processes are regulated
by temperature. The “surface area” theory for metabolic scaling dis-
cussed above suggests that basal metabolic rate scales allometrically
with the mass of the animal. As we hinted at above, this hypothesis
stems from the postulate that metabolic rate scales with the surface
area (through which heat can be lost), which is in turn a function of
[ L2 ], where [ L] is a characteristic length of the animal. Mass, how-
ever, scales with the volume of the animal, which is a function of [ L3 ].
If we combine the two relationships to express metabolic rate as a
function of mass, we get the allometric relationship:
B ∝ Mb (13.11)
where B is metabolic rate, M is body mass, and b is the scaling expo-
nent, which is equal to 2/3 according to the surface area theory. We
136 iowa state university

briefly acknowledged that several studies in the 20th century suggest

that the 2/3-power scaling is not correct, and that a 3/4-power scal-
ing might be more appropriate. Nevertheless, the general form of the
relationship is reasonable. To transform this proportionality into an
equation, we could introduce a constant B0 , so that we have

B = B0 Mb (13.12)

If we interpret B as the dependent variable and M as the indepen-

dent variable, this is clearly a power function because M is the base.
Contrast this type of equation with the population equation above,
where the independent variable t was the exponent.
The simple power-law equation for metabolic rate has some sim-
ple applications for which it is useful, but it fails to describe many
important phenomena that are seen by animal physiologists. One is
the fact that metabolic rate is also very sensitive to temperature. A
modification to the simple power law was proposed not too long ago
in this Science paper. The modification supposes that metabolic rate
depends on the kinetics of biochemical reactions on a cellular scale,
which are in turn temperature dependent. In chemistry, the temper-
ature dependence of reactions is often expressed as an exponential
function of temperature through the Arrhenius relationship:
E
R ∝ e− kT (13.13)

where R is a reaction rate constant and E/k is an energy-related

constant for a given reaction, and T is temperature. While this looks
a bit ugly, it is an incredibly important relationship for chemistry,
physics, and now biology, because it does a surprisingly good job of
describing how temperature affects physical and chemical processes.
Let’s look for a moment at the general form of this equation by
imagining a similar function

R = e−1/T (13.14)

where we consider temperature T to be the independent variable. As

you can see, as temperature increases, the exponent becomes smaller
and approaches zero. Since x0 = 1 for all x, this function approaches
1 as temperature increases, but becomes very small for small T. Of
course, we cannot compute 1/T for T = 0, and for that reason the
Arrhenius equation is written for T in Kelvin rather than Celsius.
In any case, a much improved relationship for the basal metabolic
rate of animals that includes both a dependence on body mass and
temperature can be written:
E
B ≈ B0 Mb e− kT (13.15)
 quantitative problem solving in natural resources 137

This is a more complex function because it contains two independent

variables (mass and temperature), but can be visualized by treating
one of them as a constant while the other varies. If we imagine how
metabolic rate changes for a single ectothermic organism of a given
mass as body temperature changes, it might have a pattern that looks
similar to the plot above, but that approaches a value of B0 with
increasing temperature.3 3
If you’re interested in more on this
topic, revisit this neat article written
for the Nature Education Project, and
the references therein, or check out this
13.2 Adding complexity summary of the paper that examined
this function.
Our first population growth model was a simple exponential one.
We assumed unrestrained growth with a constant per-capita (per
individual) rate parameter r = b − d, where b and d are per capita
birth and death rates. Our year-to-year prediction of population N
with this growth model is

N1 = N0 (1 + r ) (13.16)

Given an initial population N0 , the population after t years was

N = N0 (1 + r )t (13.17)

While we arrived at this result with just some reason and algebra,
a more general solution can be found using calculus. We won’t
worry too much with how this solution is obtained, nor will you
be expected to reproduce it, but it is always nice to see how more
advanced topics can help us with the problem at hand. So here is a
quick summary of how the calculus version works:
If we re-write our first incremental population change equation
above
N1 = N0 + rN0 (13.18)

N1 − N0 = rN0 (13.19)

Notice that the left-hand side is now just the population change over
one year. One of the strategies of calculus that allows elegant solu-
tion of complex problems is to imagine “smooth” changes, where
the increment over which those changes are measured in vanish-
ingly small. While this is obviously an oversimplification of popu-
lation dynamics (i.e., many animals have discrete breeding seasons
so that births are clustered during a relatively small period of time,
and no births occur during the remainder of the year), but in many
cases we don’t need to worry too much about this. We express these
vanishingly-small change increments with derivatives, where the
138 iowa state university

derivative of N with respect to t can be translated as the instanta-

neous rate of population change as a function of time, i.e., the popu-
lation growth rate. With this strategy, the above equation is written:

dN
= rN (13.20)
dt
Applying some second semester calculus, we’d come up with the
following solution, which works at all t:

N = N0 ert (13.21)

Compare this equation with the one above, N = N0 (1 + r )t , which

we developed with discrete differences. Graph both functions and
see if they match reasonably well. They should be close, but not
exactly the same. The discrete model is, in fact, subtly different, and
is often called the geometric model for population growth, while the
exponential version is the classical Malthusian model.
Calculus aside, the above unrestrained population models are use-
ful as a starting point, but they neglect any mechanisms of slowing
population growth. In most settings, resource limitation slows or
reverses growth rates as population increases. If you’re not familiar
with the story of St. Matthews Island reindeer, it is an interesting
illustration of this effect taken to an extreme.
A fairly simple way to account for resource limitation, and to
thereby restrain population growth according to some carrying ca-
pacity K, is to include an “interaction” term for our growth rate.
Using the same notation as above, an increment of growth in this
new population model is:

(1 + r ) N02
N1 = (1 + r ) N0 − (13.22)
K
This looks a bit clunky, but we can clean it up with a little bit of alge-
bra and by making the same kinds of calculus-oriented modifications
that we made above:
 
dN N
= rN 1 − (13.23)
dt K

As above, the derivative term on the left hand side is the rate of pop-
ulation change as a function of time, or the population growth rate. If
we write the equation with G for growth rate on the left-hand side, it
looks a bit more manageable:
 
N
G = rN 1 − (13.24)
K

r 2
G = rN − N (13.25)
K
 quantitative problem solving in natural resources 139

As you can see, the growth rate is just a second-order polynomial

equation. As such, it’s graph might be a bit familar to us: it is a
downward-opening parabola that crosses the x-axis at x = 0 and
x = K. This is the logistic population growth model, perhaps the
simplest way of incorporating density dependence and carrying ca-
pacity into the description of population changes in a place with
finite resources.
Solving this differential equation is not particularly easy, but for-
tunately for us, smart people have found useful solutions. The most
straight-forward solution for N as a function of t is:

N0 K
N= (13.26)
N0 + (K − N0 )e−rt

Here is an example of a case where we can defer to the experts who

came before us and simply borrow their result for our own use. The
fact is, even with the above solution, there is plenty of complexity in
the logistic population model since we must define, for any particular
scenario, several of the parameters before we can use it to any avail:
K, N0 , and r.

13.2.1 Example: minimizing suppression and loss costs (Problem 3.5)

The hypothetical functions we have proposed for the suppression

cost C and net value change Vnc were simple idealizations and would
need to be modified according to better understandings of cost-effort
relationships. Nevertheless, our cost-plus-net-value-change function
can still allow an instructive optimization. Our function reads:

C + Vnc = wE + V0 e−kE , (13.27)

where the first term on the right-hand side is the cost of suppres-
sion activities, while the second term is the net value change in case
of fire. The lowest-cost state is clearly the bottom of the dip in Fig-
ure 3.2, but can we identify that point algebraically? If we use a little
calculus, we can indeed.
In first-semester calculus, we learn that the maxima and minima of
functions can be found by setting the derivative equal to zero. In this
case:
d
(C + Vnc ) = w − kV0 e−kE = 0. (13.28)
dE
For our purposes here, I won’t explain how we arrive at this, but
suffice it to say that when we solve the right-hand equality for E, we
retrieve the effort corresponding to the minimum total C + Vnc . We’ll
140 iowa state university

follow the algebraic manipulations through here:

w − kV0 e−kE = 0 (13.29)

−kE
w = kV0 e (13.30)
w
= e−kE (13.31)
kV0
 
w
ln = −kE (13.32)
kV0
 
1 w
E = − ln (13.33)
k kV0

This result isn’t necessarily pretty, but it provides a robust analytical

solution that depends only on the coefficients we assigned to the trial
functions, and that can be easily modified for different coefficient
values.

Exercises

1. Review Section 13.2.1. In the equation for cost plus net value
change, there is a constant k. What are it’s units?

2. Propose some reasonable values for the constants and coefficients

for the fire-suppression problem in Section 13.2.1 and determine
the optimal effort and its cost.

3. Review Section 12.1.2 and ensure that you are comfortable with
the analysis presented there – or that you have developed and
justified your own approach to achieving an analogous solution.
Propose and execute a strategy for incorporating car trips through
the county road network in order to estimate the probable number
of collisions in a given span of time.

4. Review Section 12.2.1. Construct and evaluate a spreadsheet

model to solve the numerical approximation of the SIR system
of equations.
Index

algebraic expressions, 24 Iowa Roadside Pheasant survey, 12 roots, quadratic equation, 11

algorithm, 11 isometric scaling, 84
arithmetic, 47 sample, 13
Avogadro’s constant, 48 Leslie method, 120 scientific notation, 48
Lilavati, 14 significant digits, 48
ballpark estimate, 24 list all cases, 25 simpler problem, 25
basal area, 55 standard deviation, 66
belief, 26 metabolism, 131 strategies, 20
brook trout, 44 sub-problems, 24
order of magnitude, 49
control, 26 order of operations, 48
trapezoidal algorithm, 86

domain, 110 Pólya, George, 20

UPEC, 21
draw a picture, 15 Pòlya, 8
UTM coordinates, 96
problem-solving process, 20
exercises, versus problems, 19 problems, versus exercises, 19
variables, 107
guess-and-check, 24 quadratic formula, 11 variance, 65
vertex, 95
heuristics, 20, 26 range, 110 visualize the data, 25
residual, 65
I suck at math, 25 resources, 26 work backwards, 25