. ...rtgrtxh;
. xv L .l
V . . shun auwfihfr
‘IV . , , ‘
n .. V .. V
‘
{In} . ,

V I .Vrc
rah“) . «1
Wang... é:
,

Wyxumas
H31? ....
Munomfimwmaufl.

,5...

V V in
mqwapaw

L . .,
911.933.;

mafififia

: A .... :4

V, :33,

3. 3?.

Han.“
an ,, . L...
23$.me

.3. . . ....wamwufififir

rav‘ (9.;

r u.

Wm...“ , .4:

u

. . (#:7f a
. u... «Gabi... , «...! ..V:
’29:. .

, ... .l.
V ennui“.
s... ...v}..1..9...:
all 33“. ...
in}... ... . .5
...JE)..3;
... .... (2',
all 2!. LI“.
..I 3...}! .21

 

 

 

 

 

  

. 1 I . I
, . t. .. V e .W 35,35; ... .
. Yup .I 'f .: ‘ r . ( I
. v £...r.?.::£...« . .
y i I. I .. 1.... if i. .5
» . ‘ ridgesihnxlt . 11.5.4.2}: (.V . ... .a u
. , : 55i§=J1fféeit .(1
. V . 2 ‘15:..uienu— ..naaixfnfi. 4 ... ... . .
fingernail .3" V V V r . ,

 
 

  

«l'

 

 

2004
4 403677

This is to certify that the
thesis entitled

THIRTY YEARS OF RECOVERY FROM VEHICLE DISTURBANCE
IN ALASKAN ARCTIC TUNDRA.

presented by

Steven P. Rewa

has been accepted towards fulfillment
of the requirements for the

 

 

 

 

MS. degree in Plant Biology
Wture

0\ DAWW @603

 

Date

memmmemw/Imm

 

LIBRARY
Michigan State
University

 

 

 

PLACE IN RETURN Box to remove this checkout from your record.
To AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 cJCIRC/Dateouepes-p. 15

THIRTY YEARS OF RECOVERY FROM VEHICLE DISTURBANCE IN
ALASKAN ARCTIC TUNDRA.

By

Steven P. Rewa

A THESIS
Submitted to
Michigan State University
in partial fulfillment for the requirements
for the degree of
MASTER OF SCIENCE
Department of Plant Biology

2003

ABSTRACT

THIRTY YEARS OF RECOVERY FROM VEHICLE DISTURBANCE IN
ALASKAN ARCTIC TUNDRA.

By

Steven P. Rewa

Recently, off-road travel over tundra has increased markedly around Barrow,
Alaska. Different types of vegetation respond differently to vehicle traffic and the effects
on vegetation may remain for many years with site moisture appearing to be a major
determinant of resilience. This study, examining the recovery of 30-year-old vehicle
disturbances, supports the hypothesis that wet sites recover more rapidly than dry sites.
Evidence is presented that suggests that sites of intermediate moisture are slowest to
recover as a result of only moderate resistance to disturbance and moderate resilience.

Lingering physical effects of vehicle disturbance may be manifested as
differences in microtopography, soil moisture and depth of thawed soil. A comparison of
the vegetation in impacted and non-impacted terrain using principal components analysis
showed that impacted areas had less variation in the recovered plant cover than the
controls. This suggests, on the assumption that recovery towards an advanced functional
state leads to a divergence in cover type, that impacted areas are still at early stages of
succession. Sites of moderate moisture show the most difference between control and
impact. This is further supported by the absence of species from impacted areas that are
present in the controls, which is most obvious in areas of moderate site moisture. Species

richness data also demonstrates spread of the disturbance beyond the initial impact.

ACKNOWLEDGMENTS

I would like to take this opportunity to thank Dr. Patrick Webber for his guidance
and encouragement, which was instrumental to completion of my project. I would also
like to recognize Dr. Robert Hollister for giving me my start in tundra research and
convincing me to continue my studies. I would also like to thank Dr. Craig Tweedie for
his constant reminders of what’s important and for helping to keep my head above water.
Thanks to Dustin Bronson for helping me with the legwork on this project. I am indebted
to the members of my thesis committee, Dr. Alan Prather and Dr. David Lusch for their
support and confidence. My sincerest gratitude goes to Dr. Jerry Brown for putting me in
touch with the people and materials necessary to complete this study, to Dr. Kim Peterson
for advice and information relating to his prior work, and to Dr. Alan Fryday for his help
with identifying species of lichen. I thank the various members of the International
Tundra Experiment whose fi'iendship and interest constantly remind me of the merit of
my work. I especially want to thank the loyal members of “Team Rewa” as well as the
Rewa and Williams families, whom I’ll always think of as the original members.
Especially I need to thank my parents, Bill and Barb Rewa, who dealt with life’s little

inconveniences in order for me to continue my travels and research.

iii

TABLE OF CONTENTS

LIST OF TABLES ............................................................................................................. vi
LIST OF FIGURES ......................................................................................................... viii
CHAPTER 1 ....................................................................................................................... l
1 Introduction ................................................................................................................. 1
1.1 Aims and Objectives ........................................................................................... 1
1 .2 Background ......................................................................................................... 2
1.3 Natural vs. Anthropogenic Disturbance In Tundra Systems .............................. 4
1.4 US. Army Corps of Engineers — Cold Regions Research and Engineering
Laboratory ........................................................................................................... 6
1.5 Definitions Of Disturbance, Recovery, Resistance, And Resilience .................. 9
1.6 Ecological Scale And Disturbance .................................................................... 10
1.6.1 Spatial Scale .............................................................................................. 11
1.6.2 Temporal Scale ......................................................................................... 11
1.6.3 Organizational Scale ................................................................................. 12
1.7 Tundra Ecology ........................................................... 13
1.8 Recovery of Tundra Following Off-road Vehicle Use ..................................... 15
1.8.1 Problems of Assessing Recovery From Vehicle Disturbance on Tundra
Systems ..................................................................................................... 17
1.8.2 Differential Response of Communities to Disturbance ............................ 18
1.9 The Study Focus ............................................................................................... 20
CHAPTER 2 ..................................................................................................................... 22
2 Methods ..................................................................................................................... 22
2.1 Site Descriptions ............................................................................................... 22
2.1.1 Site 1 — Pitelka’s Weasel trail ................................................................... 23
2.1.2 Site 2 — CRREL Test Track ...................................................................... 26
2.1.3 Sites 3 & 4 — CRREL Weasel road ........................................................... 28
2.2 Plot Layout ........................................................................................................ 30
2.3 Microtopography ............................................................................................... 3 l
2.4 Active Layer Depth ........................................................................................... 31
2.5 Soil Moisture Content, Bulk Density and pH ................................................... 32
2.6 Vegetation Data ................................................................................................ 32
2.7 Species Diversity .............................................................................................. 34
CHAPTER 3 ..................................................................................................................... 36
3 Results ....................................................................................................................... 36
3.1 Microtopography ............................................................................................... 36
3.2 Active Layer Depth ........................................................................................... 36
3.3 Soil Moisture Content ....................................................................................... 40
3.4 Soil Compaction ................................................................................................ 41

iv

3.5 Ordination ......................................................................................................... 42

3.5.1 Site 1 ......................................................................................................... 47
3.5.2 Site 2 ......................................................................................................... 49
3.5.3 Site 3 ......................................................................................................... 49
3.5.4 Site 4 ......................................................................................................... 51

3.6 Classification ..................................................................................................... 53
3.7 Species Losses .................................................................................................. 58
CHAPTER 4 ..................................................................................................................... 71
4 Discussion ................................................................................................................. 71
4.1 Effects Of Terrain Damage ............................................................................... 71
4.2 Importance of Scale .......................................................................................... 80
4.3 Relative Damage Of Each Vehicle ................................................................... 82
4.4 Signature Spread ............................................................................................... 83
4.5 Functional Recovery and Continuous Succession ............................................ 84
4.6 Summary And Conclusions .............................................................................. 85
APPENDIX A ................................................................................................................... 89
LITERATURE CITED ..................................................................................................... 94

LIST OF TABLES

Table 1: List of study sties with individual plot numbers and descriptions of
disturbances .................................................................................................................... 23

Table 2: Differences between disturbed and non-disturbed terrain at individual sites
for each sampling date. Degrees of Freedom and t statistics are given with an
indication of significance ............................................................................................ 37

Table 3: F statistics for variables related to thaw depth including numerator and
denominator degrees of freedom ................................................................................ 39

Table 4: Statistics for thaw depth comparisons between impacted and non-impacted
grid points for each site on each sampling date ........................................................ 40

Table 5: Comparison of species richness for Site 1, based on distance from impact.
Significance class indicates which classes are significantly different from each other.
Distance classes that share a letter are not significantly different. For example, a is
significantly different from b, but ab is significantly different from neither a nor b ........ 68

Table 6: Comparison of species richness for Site 2, based on distance from impact.
Significance class indicates which classes are significantly different from each other.
Distance classes that share a letter are not significantly different. For example, a is
significantly different from b, but ab is significantly different from neither a nor b ........ 69

Table 7: Comparison of species richness for Site 3, based on distance from impact.
Significance class indicates which classes are significantly different from each other.
Distance classes that share a letter are not significantly different. For example, a is
significantly different from b, but ab is significantly different from neither a nor b ........ 70

Table 8: Comparison of species richness for Site 4, based on distance from impact.
Significance class indicates which classes are significantly different from each other.
Distance classes that share a letter are not significantly different. For example, a is
significantly different from b, but ab is significantly different from neither a nor b ........ 70

Table A-1: Species list for Site 1 including the groupings used for analyses of
vegetation data in cases where the species could not be identified with certainty ........ 89

Table A-2: Species list for Site 2 including the groupings used for analyses of
vegetation data in cases where the species could not be identified with certainty ........ 90

Table A-3: Species list for Site 3 including the groupings used for analyses of
vegetation data in cases where the species could not be identified with certainty ........ 90

vi

Table A-4: Species list for Site 4 including the groupings used for analyses of
vegetation data in cases where the species could not be identified with certainty ........ 92

vii

LIST OF FIGURES

Figure 1: Coast Guard SK-S Air Cushion Vehicle used by CRREL. Photograph by

Atwood ca. 1971 .......................................................................................................... 7
Figure 2: Naval Arctic Research Laboratory M-29 Weasel. Photography by J. Brown,

ca. 1965 ...................................................................................................................... 8
Figure 3: A Rolligon, a low pressure tire vehicle. Photograph by Atwood, ca.

1974 .................................................................................................................................. 9
Figure 4. Typical polygon tundra with meandering streams ............................................ 15
Figure 5: The location of the study area, the City of Barrow and the National Arctic
Research Laboratory (N ARL) ............................................................................................ 22
Figure 6: Location of sites of observation and experiment ............................................ 24

Figure 7: Aerial photograph of Site 1, including Pitelka’s Trail and the [BF Power-
line trail (1965) ................................................... '. .................................................... 25

Figure 8: The IBP Power-line Weasel trail. Photograph by Webber, 1971 .................... 26

Figure 9: Aerial photograph of Site 2, the CRREL test site, taken in 1975. Horizontal
trails are access roads (AR). Actual test tracks run vertical between points. Some are
not readily visible in this photograph. Numbers indicate the number of passes made
over that track. The designation 258 indicates a track of 25 passes at a slower

speed ................................................................................................................................ 27

Figure 10: (A) The 15-pass Rolligon track at the CRREL test site in 1975 (photo by
Atwood). (B) The lS-pass Rolligon track at the CRREL test site 2001 (photo by

Rewa). See person (in 10A) for scale ................................................................................ 27
Figure 11: Aerial view of Sites 3 and 4 taken in 1964. Bars indicate locations of plots

at Site 4 .................................................................................................................... 28
Figure 12: The sampling layout, indicating subplots and grid points .................... 30
Figure 13: Plot 30 at Site 4 with lines drawn to indicate 0.5 x 0.5m subplots ........ 31
Figure 14: Results of the PCA for all plots in the study ............................................ 43

viii

Figure 15: Results of the PCA for those plots for which soil moisture data were
collected .................................................................................................................... 44

Figure 16: PCA results for control subplots from which soil moisture data were
collected .................................................................................................................... 46

Figure 17: PCA results for impacted subplots from which soil moisture data were
collected .................................................................................................................... 47

Figure 18: PCA results for plots at Site 1 from which soil moisture data were
collected .................................................................................................................... 48

Figure 19: PCA results for plots at Site 2 from which soil moisture data were
collected .................................................................................................................... 50

Figure 20: PCA results for plots at Site 3 from which soil moisture data were
collected .................................................................................................................... 51

Figure 21: PCA results for plots at Site 4 from which soil moisture data were
collected .................................................................................................................... 52

Figure 22: Classification of all plots based upon relative cover of vegetation. The
designations wet, moist and dry indicate subjective classifications made in the field.

Plots 0401 and 0402 (indicated by asterisks) are corresponding damaged and

undamaged plots, which are separated by the greatest distance possible in the
classification .................................................................................................................... 54

Figure 23: Classification, based on vegetation cover, of all plots from which soil
moisture data was taken. The plots are classified into two major groups which can be
subjectively called wet and dry ................................................................................ 56

Figure 24: Classification based on environmental variables of all plots from which

soil moisture data was taken. The plots are classified into two major groups that

generally separate Pitelka’s trail and the CRREL test trail from the old and new

CRREL trails .................................................................................................................... 57

Figure 25: Species-area curve for all plots from which soil moisture data were taken.
The coordinate listed indicates the species richness for the damaged subplots of the
study. The bashed lines indicate 1 standard deviation about the mean .................... 58

Figure 26: Species-area curve for all plots at Site 1, from which soil moisture data

was taken. The coordinate listed indicates the species richness for the damaged
subplots. The bashed lines indicate 1 standard deviation about the mean .................... 59

ix

Figure 27: Species-area curve for all plots at Site 2, from which soil moisture data
was taken. The coordinate listed indicates the species richness for the damaged
subplots. The bashed lines indicate 1 standard deviation about the mean .................... 60

Figure 28: Species-area curve for all plots at Site 3, from which soil moisture data
was taken. The coordinate listed indicates the species richness for the damaged
subplots. The bashed lines indicate 1 standard deviation about the mean .................... 61

Figure 29: Species-area curve for all plots at Site 4, from which soil moisture data
was taken. The coordinate listed indicates the species richness for the damaged
subplots. The bashed lines indicate 1 standard deviation about the mean .................... 62

Figure 30: (A) Species richness at a given distance from impact over the whole study.
(B) Mean species richness of 26 randomly chosen grid cells at a given distance from
impact over the whole study. A value of 26 grid cells was chosen to approximate the
inflection point of the species area curve for the impacted subplots. A negative

distance indicates subplots found between tracks. Error bars represent one standard
deviation about the mean ............................................................................................ 63

Figure 31: (A) Species richness at a given distance from impact for Site 1. (B) Mean
species richness of 5 randomly chosen grid cells at a given distance from impact for

Site 1. A value of 5 grid cells was chosen to include the inflection points of the

species area curves for all distances from the impact. A negative distance indicates
subplots found between tracks. Error bars represent one standard deviation about the
mean ................................................................................................................................ 64

Figure 32: (A) Species richness at a given distance from impact for Site 2. (B) Mean
species richness of 7 randomly chosen grid cells at a given distance from impact at

Site 2. A value of 7 grid cells was chosen to include the inflection points of the

species area curves for all distances from the impact. A negative distance indicates
subplots found between tracks. Error bars represent one standard deviation about the
mean ................................................................................................................................ 65

Figure 33: (A) Species richness at a given distance from impact at Site 3. (B) Mean
species richness of 9 randomly chosen grid cells at a given distance from impact at

Site 3. A value of 9 grid cells was chosen to include the inflection points for the

species area curves at each distance from the impact. A negative distance indicates
subplots found between tracks. Error bars represent one standard deviation about the
mean ................................................................................................................................ 66

Figure 34: (A) Species richness at a given distance from impact at Site 4. (B) Mean
species richness of 15 randomly chosen grid cells at a given distance from impact at

Site 4. A value of 15 grid cells was chosen to include the inflection points for the

species area curves at each distance from the impact. A negative distance indicates
subplots found between tracks. Error bars represent one standard deviation about the
mean ................................................................................................................................ 67

CHAPTER 1

1 Introduction
1.1 Aims and Objectives

Off-road vehicle (ORV) use has been widespread throughout the history of
exploration and expansion over the North Slope of Alaska. When the tundra is subjected
to ORV use it is easily and immediately marred and is often slow to recover, yet today
several former vehicle trails are difficult to trace. This study takes advantage of an
opportunity to examine the recovery of former tundra trails and to re-sample some
previous experiments that examined the effects of vehicle disturbances.

Noble (1996) details a set of 4 necessary resources for quality management of
landscapes. They are;

Information on current land use practices,

Information on alternative land use practices,

Ability to predict consequences of particular practices, and
A methodology for implementing appropriate land uses.

This study concentrates on the first and third resources above. The aim of this study is to
document the extent to which an Alaskan Arctic tundra system is able to recover from
vehicle disturbance after extended periods of time. This requires collection of data
pertaining to the current land use practices in the Arctic. For the purposes of this study,
the alternative land use practice would be non-use, or preservation. Data on lingering
impacts from past disturbances that were left to natural recovery, would aid in the ability
to predict outcomes of both continued use and non-use. If residual effects of ORV traffic
are present even after a few decades, then it is likely that disturbance effects are longer
lasting than previously expected. If this is the case, further consideration needs to go into

the extent to which ORVs are used.

Information on land use goes band-in-band with information on land cover.
Variations in use within a cover type will likely have varying impacts. Likewise, similar
uses in different land cover types will also have varying impacts. Therefore, another
important consideration of this study is the relative impact endured by different
vegetation types, concentrating on the part played by moisture level of the areas in
question.

This study addresses the lingering effects of off-road vehicles after approximately
30 years of recovery, and how the moisture regime of a site affects its recovery rate.
Specifically, the lingering effects caused by tracked vehicles called Weasels and Low-

Pressure Tire vehicles called Rolligons will be examined (refer to section 1.2.1).

1.2 Background

Off-road vehicle (ORV) traffic has been extensive and necessary in Arctic
regions. With a lack of widespread roadway networks, transportation has been relatively
costly and inefficient, particularly in summer (Rickard and Brown 1974; Brown 1976;
Walker et al. 1977). It is estimated that by 1977 the most widespread cause of surface
disturbance was use of tracked vehicles (Van Cleve 1977; Milchunas, et al. 2000). The
same is true today. Transport of heavy equipment is limited to barge traffic and air freight
both of which are extremely reliant on good weather conditions. Traditional methods of
ground transport, such as the automobile, are simply not possible during large portions of
the year since many rural locations in the Arctic have no roadway system. Automobiles
are used within villages that have local road systems, but, in general, these villages are

isolated from each other (Brown 1976).

In winter, it is often possible to travel by automobile over ice roads composed of
densely packed snow and ice. The ice of the roads can be augmented by the addition of
water sprayed by tanker trucks. Ice roads produce little or no disturbance on tundra (Van
Cleve 1977), but in certain cases they have caused crushing or shearing of the shrub
vegetation under the snow (Bag 1973; Rickard & Brown 1974). These ice roads are not
without their difficulties. They are often hazardous as traction is limited and a slip off the
road leaves very little chance to retrieve the vehicle making it a permanent fixture on the
landscape. Like air and barge traffic, ice roads can only be used under appropriate
weather conditions since the stability of the road depends greatly on the temperature and
the consistency of the snow pack (Adam & Hernandez 1977). For these reasons, ice roads
cannot replace more damaging forms of off-road vehicle traffic.

Off-road vehicles are used to a great extent in the daily lives of both the native
Inupiat people as well as non-native settlers. The single-person, four-wheeled, all terrain
vehicle has become a widely used tool in a subsistence lifestyle for traffic across tundra
in search of game. In winter, the snowmobile is widely used as a replacement for the
traditional dog sled for hunting expeditions, and for daily travel between villages
(Rickard & Brown' 1974). These practices are widely adopted in the local communities.

In addition to the transportation needs of the local inhabitants, industry relies on
transport of heavy equipment and personnel for increasing development. Because of the
lack of expansive roadways, the construction related to development of the land for
purposes of oil drilling and urban expansion necessitates off-road traffic. Gravel
foundations for drilling pads and roadway expansion require heavy machinery to travel

across areas of tundra and these cause a great deal of disturbance to tundra systems. As

oil fields become abandoned it has been mandated that these gravel pads be removed to
allow the tundra to recover. These removals may require the continued use of ORVs. An
alternative method of gravel removal utilizes the roads that are to be removed. Gravel
could be removed from the extreme ends of the roads and hauled back toward the road
source while using the roads for traffic rather than driving off onto the tundra.

During the early development of the Prudboe Bay oil field, trains of sleds, called
tractor trains were hauled over the tundra by bulldozers (Klein 1970). These occasionally
operated in the snow free season, but were mostly a winter phenomenon. They have left
an indelible mark on the tundra.

Regulations related to ORV use have changed over the course of the years.
Initially, there were few regulations and one common practice was to remove the thawed
soil layer by blading to provide a smooth, solid track (Rickard & Brown 1974). It was by
this method that the previously mentioned tractor trains were able to travel during the
summer months. Regulations are an issue in other Arctic regions as well. In Siberia, ORV

use has been outlawed since 1989, but the laws are often broken (Khitun 1997).

1.3 Natural vs. Anthropogenic Disturbance In Tundra Systems

When considering disturbance it is often important to distinguish natural effects
from anthropogenic effects. Many consider natural effects to be part of a disturbance
regime that the ecosystem has adapted to recover from. Changes fiom the natural
disturbance regimes can have a dramatic effect on plant diversity or other properties of a
landscape (Pearson 1994). As a result, many consider anthropogenic effects to be more

serious, since they are not a natural occurrence and the recovery mechanisms may not be

present in the ecosystem. On the other hand, many see no difference at all between the
two types.

Many of the human-induced disturbances to Alaskan Arctic tundra are related to
vehicle traffic, both on and off roadways. Roadway construction brings with it a host of
disturbances. A great deal of tundra has been destroyed in order to lay the gravel for the
roads themselves. The Trans-Alaska Pipeline and associated road systems required the
disturbance of approximately 12,000 ha of land. Other problems with roads include oil
and diesel fuel spills related to vehicle traffic. In these cases, trampling and other
problems of clean up operations adversely affect the tundra as well (Bliss 1983). One of
the main secondary problems associated with roads is the accumulation of dust on
adjacent tundra. Accumulation of dust leads to changes in pH and nutrient levels of the
soil. A great deal of the dominant bryopbyte species that occur in more acidic tundra,
such as Sphagnum sp., are strongly affected by the increased pH levels resulting from
deposition of calcareous road dust (Walker & Everett 1987; Leadley et al. 1996). Dust
can also accumulate in the stomata of plants blocking gas exchange and seriously
inhibiting production (Auerbach et al. 1997).

Another significant effect of dust accumulation is the drastic change of the
thermal regime. The Dalton Highway is considered to be one of the most energy related
disturbances on the Alaskan North Slope (Leadley et al. 1996). Dust accumulation on the
surface of the snow pack causes a reduction in surface albedo, which results in a warming
of the snow pack and an earlier spring melt. Since the snow pack acts as an insulating
layer, the soil and vegetationmat are exposed to the cold early on in the season. The dust

also acts as an insulating layer in the summer time. On top of the dust effects, there can

be increased insulation due to a significant snow drifting effect related to the presence of
the road. These forces together cause drastic changes in active layer depth (Auerbach et
al. 1997).

One issue dealing with human-induced disturbance is the idea of cumulative
impacts. Cumulative impacts result from a combination of direct disturbance effects. As
roadways increase, the areas immediately adjacent to them are affected by the
disturbance in ways that are relatively predictable. However, as these areas become more
common other effects become apparent. Roadways may impede drainage leading to
drying or flooding of communities. As the unaffected patches become smaller and farther
apart, the effects on wildlife become more apparent. Edge-sensitive species become
scarce and migration corridors become more important. It has also been shown that many
animals are frightened off by increased traffic causing abandonment of their territory

(Bliss 1983).

1.4 US. Army Corps of Engineers - Cold Regions Research and Engineering
Laboratory

Following World War 11, there was an increased interest in military activity in the
northern regions. Much like commercial industry, the military had serious concern for the
transport of equipment, supplies and people. It was thought that future conflicts could
arise over northern territories. The US. Army Corps of Engineers — Cold Regions
Research and Engineering Laboratory (CRREL) was set up to examine the plausibility of
travel in Arctic and other cold regions. Of particular interest in the 19703 was the

development of air-cushion vehicles (ACV) for transport in arctic regions.

It was thought that ACVs would be ideal for Arctic travel considering the level
nature of the landscape and the ability of ACVs to travel from land to water with little
difficulty (Abele & Brown 1977). Researchers at CRREL were also concerned with the
amount of damage that would be sustained by the tundra as a result of traffic by ACVs.
For this purpose, a US. Coast Guard SK-5 ACV was driven multiple times over the

tundra near Barrow, Alaska in 1971 to document impact (Figure l).

 

Figure 1: Coast Guard SK-5 Air Cushion Vehicle used by CRREL. Photograph by
Atwood ca. 1971

The traditional method of off-road travel at the time used tracked vehicles.
Therefore, an M-29 Weasel was run concurrently with the SK-5 ACV for comparison of
the two vehicle types (Figure 2). At this same time, the Rolligon, a low-pressure tire

vehicle, began to see more widespread use for off-road traffic. In 1974, a Rolligon was

tested alongside the test tracks previously made by the SK-S and the Weasel (Figure 3).
These Rolligons were further tested in 1976 near Lonely, Alaska (Abele er al. 1977;
Abele er al. 1978) and at the Prudboe Bay oil field (Walker et al. 1977). These sites were
revisited during this study, 30 years later, to assess residual impact on the landscape.
According to studies by Abele and Brown (1977), the SK-5 hovercraft performed well
over tundra surfaces, however, it did not see widespread use following these tests. The
lack of mass production for hovercraft and its relatively high operational cost ($0.31

metric ton'llrm'1 compared with $0.19 for fixed wing aircraft and $0.02 for ships in 1974)

£351“: mum u
m as: m. as

 

Figure 2: Naval Arctic Research Laboratory M-29 Weasel. Photography by J. Brown, ca.
1965

 

Figure 3: A Rolligon, a low pressure tire vehicle. Photograph by Atwood, ca. 1974.

contributed to a decline in its use. Though no specific information on the operating costs
of the M-29 Weasel or the Rolligon were available, Rolligons were estimated to have an
even higher operational cost than the SK-5, but were cheaper and easier to construct. In
1977, Rolligons were used almost exclusively as the means of off-road travel for the

Prudboe Bay oil field (Walker et al. 1977).

1.5 Definitions Of Disturbance, Recovery, Resistance, And Resilience
The definitions of disturbance, recovery, resistance, and resilience are a matter of
some debate in the scientific community. Scientists in various fields define disturbance

based on relevance to their particular discipline. A disturbance is often defined by

whether it is natural or anthropogenic. Many view disturbance as a phase along a
successional pathway rather than an “unnatural occurrence.” Some would view hurricane
damage as a large-scale disturbance while others view it as a natural part of the
ecosystem that does not really affect the long-term equilibrium of a system (Paine et al.
1998).

In this study, I define disturbance as a change in vegetation or underlying
substrate caused by an external factor (Walker 1996). This definition encompasses all
forms of disturbance, whether natural or anthropogenic. I define recovery as a return to
the original ecosystem condition and function (Racine & Ahlstrand 1991). Areas that are
not immediately impacted, but are located near an impact may still feel the effects of the
disturbance and can therefore also be considered disturbed. In this study, I use the term
“impact” to refer to the areas directly crossed by the vehicle. The terms “non-impacted”
and “control” are used interchangeably and refer to any areas that were not directly
driven over. 1 define resistance as the ability of an ecosystem to withstand disturbance
and remain stable (W ardle et al. 2000). Resilience is defined as the rate at which an
ecosystem is able to rebound and recover from a disturbance (Seybold et al. 1999).
Resilience is the main issue in this study. Two major questions are addressed. Are
Alaskan Arctic tundra systems resilient enough to recover fully from vehicle disturbance

within 30 years? How does resilience differ between areas of differing site moisture?
1.6 Ecological Scale And Disturbance

Scale is an important factor in assessing the overall impact of a disturbance and

the recovery rate associated with that disturbance (Pearson 1994). Three scales must be

10

considered when evaluating a disturbance. Spatial scales are perhaps the easiest to
consider because of their physical nature. Also of importance are temporal scales, which
are of particular importance when considering the resilience or recovery rate of a system.
Finally, it is important to consider the organizational scale at which the extent of
disturbance and recovery is determined. For example, does a disturbance affect
individuals or populations, or will it affect whole ecosystems?

1.6.1 Spatial Scale

The dependence of plant biodiversity upon spatial scales is well documented. One
fairly prominent example is that of species area curves. Plot size can have a large effect
on the ability to detect associations between species (Kilburn 1966; Crawley & Harral
2001). The same is true in tundra systems as they pertain to disturbance recovery (Forbes
& Sumina 1999). When colonization becomes necessary, the speed at which an area
recovers is determined, in large part, by its distance to surviving patches of similar
vegetation (Romme er al. 1998). This is particularly true in areas with a poor viable seed
bank where dispersal and vegetative growth are necessary.

The spatial scale of the disturbance is important in determining the severity of the
impact on the ecosystem. Kevan et al. (1995) found strong evidence of lingering impact
of off-road vehicle traffic in high Arctic sites, but state that the impact was localized
around the vehicle tracks. In their case, the initial impact was limited to tractor treads of
approximately 42cm width. The current study deals with similar types of impacts.

1.6.2 Temporal Scale
When considering the extent of disturbance, it is important to consider the

temporal scale, as well as the spatial. A 30-year-long disturbance seems long lasting in

11

terms of a human lifetime, but it would be very short lived when one considers it on a
geological time scale. A disturbance that occurs at a single point in time would be very
important to an annual plant that depends upon a single season to propagate. To a large
clonal organism that reproduces vegetatively over many years, a. single disturbance may
be only a minor setback.

The temporal scale dictates the severity of a disturbance by placing importance on
the longevity of the disturbance effects. A widespread disturbance whose effects are only
present for a year may not be considered as important or as serious as a disturbance with
a smaller spatial scale with residual effects after 20 years (Milchunas et al. 2000). This
idea suggests a trade off between spatial extent and longevity of disturbance that can be
difficult to choose between. The choice can only be made after careful analysis of the
values of the public. One would like to preserve as-much undisturbed area as possible and
would, therefore, lean toward a smaller spatial extent. Theoretically, as small long-term
disturbances become more frequent, they become cumulative and initially undisturbed
areas become affected. It then becomes difficult to maintain undisturbed areas in which
case a widespread, but short-lived, disturbance may be favored. It is, therefore, important
to know how long a disturbance can be expected to last in order to accurately assess the
impacts likely to result from any management options. Studies such as this thesis are
useful for determining recovery rates.

1.6.3 Organizational Scale

The level of ecological organization at which a disturbance affects an area is

important for considering the overall impact of a disturbance. For instance, the clipping

of a leaf is a serious disturbance at the plant level or possibly even the population level.

12

However, at the level of the community and on up to the biosphere level, the clipping of a
leaf is not likely to have high importance. A serious, population-wide disturbance, on the
other hand, can effect interactions at the community level. As these community-level
disturbances become large and frequent enough, it can be a serious impact over the whole
biome.

Kevan et al. (1995) demonstrate that there is a significant, but highly localized,
effect on soil invertebrates due to off-road vehicle traffic after 20 years. Although
localized, the changes may become cumulative as the impacts become more frequent.
Eventually these differences may affect the nutrients of the soil and forage for animals at
higher levels of the food chain. This is a case of a localized impact causing changes to

CCOSYSIEITI processes.

1.7 Tundra Ecology

The ecology of the tundra is such that its behavior in response to impact is unique.
The annual average temperature in the Arctic is less than zero degrees Celsius (Brown et
al. 1980). As a result of the low temperatures, a layer of permafrost that never thaws lies
below the surface. During the summer, the thawed layer of soil at the surface is the only
portion available to plants for root growth and nutrient uptake. This thawed zone is
referred to as the “active layer.”

The layer of permafrost that underlies the tundra is impermeable to water, causing
poor drainage. Because the average temperature is near freezing for much of the year, the
soil, which is generally high in moisture, often succumbs to disturbance caused by the

expansion and contraction of the water contained within the soil as it repeatedly freezes

13

and thaws. This phenomenon is referred to as the “freeze-thaw cycle.” As the water
contained within the soil freezes its volume increases, and the soil is forced up and

outward. This process is known as “frost heave.” If the effect is localized within an area

of about 1.0 m2, it results in a structure called a “frost boil,” which resembles a patch of

bare, cracked soil. On a larger scale, or in cases where this occurs on a steep slope, the
cracked soil resulting from thermal contraction collapses and breaks off in large clumps.
This process, refened to as “thermokarst,” often causes large-scale disturbances and
leaves large patches of exposed soil whose thermal regime is drastically altered and it
disrupts the previously existing plant communities.

Unlike permafrost, which is frozen soil, ice also exists in pure form as long veins
that crisscross under the soil surface. These ice veins are generally triangular in cross
section and are refened to as “ice wedges”. As these ice wedges expand, they create
trenches and troughs in the tundra surface. These troughs increase in size as water runs
into them accentuating the freeze-thaw cycle that forms them. Because these ice wedges
are plentiful and extend for long distances, the tundra often takes on a reticulate
appearance and is considered to be “polygonized tundra” or “patterned ground”(see
Figure 4). .

Most of the natural disturbances mentioned above, such as thermokarst and
patterned ground development, result from the thermal properties of the soil. Off-road
vehicle traffic can affect these thermal properties and can compound the effects of the
freeze thaw cycles, possibly causing these natural disturbances to become more frequent

and more severe.

l4

 

Figure 4: Typical polygon tundra with meandering streams.

1.8 Recovery of Tundra Following Off-road Vehicle Use

Tundra systems can be easily impacted by off-road traffic (Gersper & Challinor
1975). The traffic signature (the visible impact of the traffic) is generally the most
obvious effect. Signature effects range from being purely esthetic, to a physical
disruption of the ground surface, such as the removal of the vegetation. Along the
gradient between these two extremes lie effects such as change in surface albedo, heat
content of the soil, and ice content of the soil (Kevan et al. 1995). Because the soils of
tundra systems generally have high moisture contents they are easily compressed by the
weight of many ORVs during the summer months. These ORVs generally leave distinct
ruts (Brown 1976). Pressure by ORV traffic can also affect the thermal regime of the soil

caused by changing the surface albedo or by compacting the soil which can change the

15

amount of heat conducted through the soil thereby affecting the depth of the ice-rich
permafrost layer that underlies tundra soils. The combination of these two effects can
lead to a trenching effect as the tracks fill with water and succumb to the freeze thaw
cycle (Abele and Brown 1977).

Other effects may be more subtle. Traffic by ORVs can lead to a removal of
standing dead vegetation or compression of dead vegetation into the soil causing a “green
belt” effect where the affected area looks greener due to the removal of brown, dead
vegetation (Rickard & Brown 1974). Additionally, the compression of dead matter into
the soil system can accelerate decomposition and lead to an increase in nutrient
availability and a true increase in greenness. These forces can affect plant succession in a
number of ways. With the removal of the dead vegetation, space previously occupied by
leaf litter may become available for plants to spread out. Increased sun exposure may
lead to an increase in the vigor of shade intolerant species. Because the standing dead
vegetation acts as an insulating layer, its removal would cause soil temperatures to be
more extreme, possibly causing frost heave, allowing disturbance adapted species to
colonize in the newly exposed soil.

Anthropogenic and natural disturbances are often similar in the ways in which
recovery takes place. Species that occur the most in sites recovering from human-induced
activities are those that would inhabit naturally disturbed areas such as riverbanks
(Auerbach et al. 1997). Generally, pioneer species take over once the natural dominant
species have been removed due to disturbance.

Historically, one of the largest and most important sources of anthropogenic

surface disturbance has been the use of off-road vehicles (ORV) for oil exploration and

16

development in the Arctic (Abele et al. 1977). For example, vehicle trails associated with
Petroleum Reserve No. 4 have remained detectable 20 to 30 years after disturbance and
roadways constructed in wet terrain can often require decades to revegetate (Van Cleve
1977). Disturbances like these can, therefore, be considered long lasting and potentially
serious.

A fairly obvious result of vehicle disturbance on tundra is the depression of the
ground surface resulting in increased bulk density, lower soil moisture, and lower soil
aeration (Van Cleve 1977). It has been shown that a reduction in microtopography in
tundra, which often results from vehicle disturbance, causes a reduction in species
richness (Forbes 1992; Auerbach et al. 1997 ). Increased active layer thickness has been
shown to cause an increase in plant productivity (Van Cleve 1977), possibly changing the
plant community. It is also believed that these effects on the soil may be accentuated over
time before they become ameliorated (Gersper & Challinor 1975).

The available seed bank is also affected by the alteration of the topography and
microtopography of a site. Because much of the seed rain is wind bourn, depressions
often collect seeds while hills may act as barriers. Higher densities of seeds and seedlings
are often found in depressions (Grulke & Bliss 1983). Vehicular disturbances that change
the topography of the sites would also change the distribution of seedlings impacting
succession and other community level plant interactions.

1.8.1 Problems of Assessing Recovery From Vehicle Disturbance on Tundra
Systems

There are several problems with assessing vehicle disturbance and recovery of
tundra systems. The extent to which an area has been disturbed can be difficult to discern

since the effects of a disturbance are not always visually obvious. Impacts on the thermal

l7

regime and moisture content may go unnoticed before causing serious problems. The
effects of disturbances in tundra systems also tend to spread beyond the original
boundaries of the disturbance due to heat conductance and changes in water content or
water holding capacity, affecting areas adjacent to the initial impact.

The idea of disturbance is somewhat confused because landscapes and vegetation
types are continuously changing and disturbance can often be looked at as a phase along
the natural gradient rather than a unique event that is detrimental to the system. The idea
of recovery is similarly confused. Defining recovery based on the condition of the
landscape can be problematic based on the degree to which succession has occurred in
neighboring regions. If recovery occurs over long periods of time, return to the original
state of the landscape would leave the community lagging successionally. The
community may require a great deal of time to catch up again from the lag and residual
effects of the disturbance would be present, even if they are not very noticeable. In
regions of high spatial heterogeneity, comparing an area to a neighboring region in order
to assess recovery can be difficult. In effect, we may attempt to judge recovery based on a
community that would naturally be different due to differing successional pathways of its
particular region. '

1.8.2 Differential Response of Communities to Disturbance

Both the resistance and resilience of the tundra surface depend on several factors.
The climate and environment of the region, the topography and microtOpography of the
landscape and the‘season of the year all play strong roles in the extent of disturbance. Soil
conditions such as bulk density, permafrost, and moisture content have a strong effect on

resistance and resilience (Slaughter et al. 1990; Pearson 1994; Khitun 1997).

18

Soil moisture can be one of the most important factors determining the
productivity of an Arctic system (Leadley et al. 1996). The various amounts of saturation
in the soils lead to different communities ranging from dry heaths dominated by
ericaceous shrubs to aquatic communities dominated by sedges, grasses or forbs. The
various growth forms dominating is these diverse moisture regimes will have quite
different responses to impact and, as such, will have different successional trajectories
and different recovery rates.

Moisture plays a role in the thermal regime of a community, but plays a large part
in the nutrient availability. Chapin et al. (1988) found that in areas where water was
being channeled and collected, the productivity of vascular plants was higher than
normal. This was attributed to the increased flow of nitrogen owing to the presence of
moving water. This situation may be analogous to vehicle tracks like those found in the
sites investigated in this study. Artificially increased productivity in this way could lead
to different communities and, therefore, greatly hinder recovery to a pre-disturbance
community.

Various authors have proposed differing hypotheses relating the recovery rates of
landscapes to their moisture levels. Walker et al. (1977) found in their study of Rolligon
damage in the Prudboe Bay area that the wettest sites were the most easily impacted and
they concluded that the dry sites would be the quickest to recover. Ebersole (1985)
hypothesized, based on his observations and those of Hernandez (1973), that it would be
the drier sites with lower resilience that would take the longest amount of recovery time
to be identical to the undisturbed landscape. A synthesis conducted by Forbes et al.

(2001) of studies concerning 20-75 year old disturbances showed similar results. Gersper

l9

and Challinor (1975) have shown that the greatest amount of physical change due to off-
road vehicles occurred in the wet soils while the dry soils showed a greater change in
color. They found soil compaction to be important and noted that the wettest and driest
areas had higher initial bulk densities and that the magnitude of change was actually
greatest in areas of intermediate moisture regime. Similar results were noted by Abele

and Brown (1977).

1.9 The Study Focus

This study capitalizes on the opportunity to examine the recovery of a number of
20-40 year old vehicle trails in the vicinity of Barrow, Alaska. For example, trails left by
Brown and Johnson (1965), Pitelka (Peterson 1978), and Abele and Brown (1977).

For these trails, no detailed quantitative measures of initial impact on vegetation
or environment by the vehicles was recorded, although some photographic evidence
exists and some personal interviews were possible with witnesses, such as Dr. Jerry
Brown and Dr. Kim Peterson. Thus, this thesis relates primarily to recovery after a long
time interval and allows interpretation of system resilience to be made. Only inferences
can be made about initial impact and resistance to impact.

Because off-road vehicle traffic has been so extensive in the past and continues in
some areas, long lasting effects become more important. If effects of the disturbance
remain for 30 years or more, it is likely that in many areas, the frequency of disturbance
is too high to allow adequate recovery time. In this case, management practices may need

to be modified to further protect these areas. In order to document the extent to which the

20

tundra is able to recover, this study examines historic vehicle disturbances of known age
and duration.

The various trails examined in this study have been made over a number of terrain
types and vegetation types. The literature documents and general observations suggest
that different communities respond differently to disturbances. Many of the disturbances
that frequently affect the tundra are related to the thermal properties of the soil. These
thermal properties are modulated, in part, by the moisture content of the soil. Vehicle
disturbances affect these thermal properties differently based on these soil properties.
This study aims at testing the general hypothesis emerging from the earlier work of
Hernandez (1973), Walker et al. (1977), Ebersole (1985) and Forbes et a1. (2001) that

drier sites have less resilience than wet sites.

21

CHAPTER 2

2 Methods
2.1 Site Descriptions

The study area is in northernmost Alaska in the vicinity of the City of Barrow.
Barrow is situated in the littoral zone of the Arctic coastal plain (Figure 5). The climate is
heavily influenced by the presence of the Arctic Ocean, which creates a more moderated
climate. Mean July temperature is about 37°C. The annual mean temperature is
approximately —12.6°C. The region annually receives about 124.1 mm of precipitation

(Brown et al. 1980).

. Alaska

  

3 .

Figure 5: The location of the study area, the City of Barrow and the National Arctic
Research Laboratory (NARL).

22

The Arctic coastal plain is a relatively flat landscape with low amounts of large-
scale relief. The terrain is generally quite wet with thousands of oriented thaw lakes and
ponds, as well as meandering streams. At a smaller scale, there is a great deal of variation
in microtopography with ice wedges forming a reticulate pattern of troughs delineating
polygons on the surface (Figure 4). The underlying soil of the coastal plain is frozen year
round and the depth of the active layer is often no more than 1.0m (Gersper & Challinor
1975). Specific details for the individual sites are given in the following section.

2.1.1 Site 1 - Pitelka’s Weasel trail

The test sites vary in location, moisture regimes, and vegetation types. In many
cases, the vehicle causing the disturbance and the age of that disturbance also differ. The
major differences between the sites are outlined below and also listed in Table 1. See

Figure 6 for the locations of the sites.

Table 1: List of study sites with individual plot numbers and descriptions of

 

 

disturbances.

Site Name Disturbance Description Plots
1. Pitelka’s Trail Periodic traffic by *M-29 Weasel c. 1955-1964 1-5
2. CRREL Test Site Multiple passes by "Rolligon 1974 6-11
3. Oldest CRREL Trail Periodic traffic by *M-29 Weasel c. 1962-1963 12-22

4. Youngest CRREL Trail Periodic traffic by *M-29 Weasel c. 1963-1964 2339

*Small tracked vehicle
a""‘Large low-pressure tire vehicle

The vegetation of this site ranges from dry, polygonized tundra at the northeast
end to a more wet meadow community at the southwest end. In the dryer areas, the
dominant vascular species are Eriophorum angustrfolium, Eriaphorum scheuchzeri, and

Dupontia fischerr'. Dryer areas also have an abundance of Luzula confirsa, while the

23

wetter areas contain an abundance Arctophila fulva. The dominant bryophytes are
Calliergon sp. and Polytrichum juniperinum. Many lichens are present including
Dactylina arctica and Cetraria islandica. A complete species list for this site (Table A-1)

is given in Appendix A.

 

 

 

 

 

Figure 6: Location of sites of observation and experiment.

Within the site is a trail that was used 4 or 5 times per year as a route by scientists
to check on experiments via Weasels. Frank Pitelka first used the track in 1955, and use
of the track was stopped in 1964 following the creation of a gravel road located nearby

the experiments. The track was later studied by Peterson (1978), who described five 5m

24

by 20cm quadrats across the track and one control off the track. Figure 7 shows Pitelka’s
trail along with a Weasel trail used by the International Biological Programme (IBP). The
IBP trail was relocated during this study, but not examined. A ground photograph of the

IBP power-line trail is shown in Figure 8.

 

Figure 7: Aerial photograph of Site 1, including Pitelka’s Trail and the IBP Power-line
trail (1965).

25

 

Figure 8: The IBP Power-line Weasel trail. Photograph by Webber, 1971.

2.1.2 Site 2 - CRREL Test Track

The site is situated in a drained lake basin and is considered a wet meadow
community. Site 2 was the most species-poor site in the study, containing only four
species of vascular-plants: Eriophorum scheuchzeri, Eriophorum angustrfolium, Dupontia
fischeri, and Saxr'fraga cemua. Bryophytes were more prevalent at this site and were
dominated by Sphagnum sp., Calliergon sp., Mnium sp., and Polyrrichum juniperinum.
The only common lichen at this site was Peltigera sp. A complete species list for this site
(Table A-2) is given in Appendix A.

Within the site are several vehicle tracks of differing intensities made by different
vehicles. Of particular interest is a track left by 15 passes of a Rolligon causing complete

destruction of the vegetation mat (Abele er al. 1974). See Figures 9 and 10.

26

a A Weasel
“ 0 Rolligon

 

Figure 9: Aerial photograph of Site 2, the CRREL test site, taken in 1975. Horizontal
trails are access roads (AR). Actual test tracks run vertical between points. Some are not
readily visible in this photograph. Numbers indicate the number of passes made over that
track. The designation 25s indicates a track of 25 passes at a slower speed.

 

Figure 10: (A) The 15-pass Rolligon track at the CRREL test site in 1975 (photo by
Atwood). (B) The 15-pass Rolligon track at the CRREL test site in 2001 (photo by
Rewa). See person (in 10A) for scale.

27

2.1.3 Sites 3 & 4 — CRREL Weasel road

These sites cross a ridge and represent a gradient from weakly polygonized tundra
in the wet basin to strongly polygonized tundra on the dry ridge top (Figure 11). At Site
3, the dominant vascular plants are Salix rotundifolia, Arctagrosn's latifolr'a, Luzula
confusa, Luzula arctica and Alopecurus alpinus. The diversity of bryophytes and lichens
at these sites is higher than at the previous two sites. The dominant bryophytes at Site 3
are Pohlr'a nutans, Tomenthypnum nitens, Calliergon sp., Drepanocladus brevifolius, and
Polytrichum juniperinum. Dominant lichens include Cerraria islandica, Thamnolia
subuliformr's, Dactylina arctica, and Ochrolechr’afrigida. A complete species list for this

site (Table A-3) is given in Appendix A.

 

Figure 11: Aerial view of Sites 3 and 4 taken in 1964. Bars indicate locations of plots
at site 4.

28

Site 4 ranges from a wet meadow community at its lowest point to a dry ridge at
its highest. The dominant vascular plants in the wet meadow are Poa arctica,
Eriophorum scheuchzeri, Dupontia fischeri, Carex aquatilis, and Saxifraga cemua.
Dominant bryophytes include Campylium stellatum, Pohlia nutans, Drepanocladus
brevrfolius, and Calliergon sp. Common lichens include Peltigera sp., Cetraria islandica,
and Lobaria linita.

In the dryer areas of Site 4, the dominant vascular plant species are P. arctic, Salix
rotundrfolia, Alopecurus alpinus, Arctagrostis lanfolia, and Petasitesfrigida. Dominant
bryophytes include Pohlia nutans, Warnstorfia exannulata, Polytrichum juniperinum,
Calliergon sp., and Campylium stellatum. Common lichens include Dactylina arctica,
Peltigera sp., Lobaria linita, and Thamnolia subuliformis. A complete species list for this
site (Table A4) is given in Appendix A.

Like Pitelka’s trail, these tracks served as Weasel roads for CRREL scientists
conducting cryopedological studies. The older trail, hereafter known as Site 3, was used
in the summers of 1962 and 63 when it was abandoned due to overuse. A new trail, Site
4, was then adopted in 1963 and used through the end of the 'study in 1964 (Brown &

J obnson 1965). Therefore, the disturbance at Site 4 is slightly more recent and more
severe than that at Site 3. The Site 3 trail is more or less confined to a single set of tracks,
while at Site 4, two and sometimes three sets of tracks are discemable. This necessitated
the use of entirely no-impact control plots rather than the use of the ends of plots as

controls. The plot design is presented in section 2.2.

29

2.2 Plot Layout

The plots were set up in a 2x10m grid by placing bamboo rods at 0.5m intervals
along the length and width of the plot (Figure 12). Each plot contained 105 sample points
to be used in point-based measurements such as microtopography and active layer
sampling, and 80 subplots for sampling done on an area basis such as percent cover of
vegetation. The plots were oriented perpendicular to and centered on the track (Figure

13).

 

 

 

Subplot 0.5x0.5m Grid point n=105
=80

Figure 12: The sampling layout, indicating subplots and grid points.

Each plot was intended to include its own control by including sample points
from non-impacted areas near the ends of the plots. As mentioned above, this was not
practical in some particularly disturbed plots. In these circumstances, additional control
plots were set up in nearby adjacent tundra that was considered representative of the
appropriate undisturbed community. Plots were established at each of the 4 sites.
Wherever possible, 3 replicate plots were established over as many vegetation types as
were present, for a total of 39 plots. The primary benefit from this sample design was the

ability to sample in increments of increasing distance away from the initial impact.

30

 

' ~.

 

Figure 13: Plot 30 at site with lines drawn to indicate 0.5 x 0.5m subplots. ‘

2.3 Microtopography

Surface depression caused by vehicle traffic was assessed by measurements of
microtopography using a telescopic level. The height of the ground surface at each point
within the plot was recorded to the nearest cm above or below the average height of the
plot, to determine spatial distribution of the topography within the plot.
2.4 Active Layer Depth

Measurements of active layer were taken to the nearest cm below the ground
surface by thrusting a 1cm diameter metal rod into the earth at the grid points. This was
done 4 times over the course of the summer at approximately lO-day intervals in order to
determine seasonal variation. The dates of sampling were chosen in order to coincide

with similar measurements recorded for the Circumpolar Active Layer Measurement

31

(CALM) project. The dates sampled included July 13 and 23, and August 2 and 12. An
initial sampling was also conducted at Site 2 only on June 28.

In plots located over vehicle tracks, comparisons were made on an individual plot
basis. The average thaw depth for non-impacted areas within the plot was compared with
the average thaw depth for the impacted areas. In addition, all of the thaw depths taken
throughout the whole study were analyzed together to test for the effects of time and site

interactions with the thaw depth.

2.5 Soil Moisture Content, Bulk Density and pH

Measurements of soil moisture were taken for impacted plots by collecting a
known volume of soil using 325ml soil cans. Three random soil samples were collected
from areas under tracks, outside of tracks and between tracks. Wet weights were
measured. Soil was dried overnight using a drying oven and dry weights were then
measured and compared to wet weights. All weight measurements were recorded to the
nearest 0.1g. The dry weights were also used to calculate the bulk density of the soil in
order to compare soil compaction. A small amount of wet soil from each sample was

used to determine the soil pH.

2.6 Vegetation Data

Percent cover of all vascular plants, as well as the dominant lichen and bryophyte
species, was determined by visual estimation for the individual subplots (n=80). The
measurements were conducted during the peak of the growing season to ensure maximum

percent cover for the season. Estimation of percent cover of surface water and bare soil

32

was also conducted at this time. Plants were identified to the species level whenever
possible. In all cases, the plants were identified to the genus level. These groupings were
considered species for purposes of species richness analysis and are hereafter referred to
as such. The species and groupings used in the analyses are listed in appendix A and
taxonomy matches that used in Webber (197 8).

‘ The plant cover data were converted to relative cover and relative frequency
within the subplots. The variation in the relative cover data was examined using
ordination by Principal Components Analysis (PCA) (McCune & Mefford 1995). Each
plot was separated into two groups comprised of control and impacted subplots. These
subplots were then arranged by the PCA in an n-dimensional hyperspace based on the
similarities of their vegetation. All the variation is collected along the n axes of the
hyperspace. Subplots with similar vegetation will be arranged near each other. The
ordination was conducted once for all plots in the study and again using only those plots
from which soil moisture data were recorded. For those plots with associated soil
moisture data, separate ordinations were conducted for control and impacted subplots.

Correlations of the ordination axes with the microtopography, active layer depth,
soil moisture, pH, and amounts of surface water and bare soil were calculated to estimate
the relationships between the environmental factors and the variation in the vegetation.
Ordinations for the individual sites were also conducted to determine if differences
between impact and control are more or less pronounced at the site level than the whole
study level. The individual site ordinations also indicate if different environmental factors

are important at different sites.

33

Classification of the plots based on the relative cover data was done at the whole
study level by cluster analysis using Euclidean distance measures and Ward’s group
linkage method (McCune & Mefford 1995). This was done for all plots and again for
only those where soil moisture was measured. Additionally, a classification of the sites

was conducted based on the environmental factors rather than the vegetation data

2.7 Species Diversity

The vegetation data were used to calculate species richness for each treatment
within each site from which soil moisture data were collected. Because the impacted
terrain was represented by fewer samples than the non-impacted terrain, it was possible
that different species diversity in the impacted areas could be attributed to a smaller
sample area. Species area curves were constructed for each site based on the vegetation
data. These species area curves were created using PC-Ord (ver. 4.0) and calculated using
the Sorenson method. They represent the mean number of species present in random
selections from the total of a given number of subplots. This mean value is used to
estimate the expected number of species for a given area. Up‘ to 500 different random
selections are used to generate these means. In cases when there are not 500 different
random combinations of the appropriate number of subplots, all combinations of subplots
are used. A range of one standard deviation about the mean is also included.

These species area curves were also used to investigate a possible change in
species richness relative to the distance from the impact. All subplots within a single plot
whose distance from the edge of impact were equal, were grouped together and their

combined species richness was calculated. An analysis of variance was conducted using

the general linear models procedure in SAS (ver. 8.0) to determine if there was a
significant effect of distance from impact on species richness. Pair-wise comparisons
were conducted between the richnesses of the different distance classes using Tukey’s
test. Additionally, species area curves were constructed for each distance class to indicate
the number of species that are likely to be present in a given area a specific distance away

from the impact.

35

CHAPTER 3

3 Results
3.1 Microtopography

Differences in microtopography between impacted and non-impacted areas within
each of the 25 plots containing tracks were determined by comparison of means using t-
tests. Each plot contained 105 sample points allowing 103 degrees of freedom.
Microtopography differences were found in 20 of 25 plots indicating that the track ruts
are still present along most of the trails considered in this study. Four of the plots with the
least significant difference between impacted and non-impacted states (plots 6, 23, 24,
and 25) are located in wet meadow communities. The fifth such plot (plot 22) was from a
mesic community at the very end of the track at Site 3 where the disturbance was likely
less severe. Even in this case the difference was nearly significant (tcdc=l.976, tcfiF1.984-,

or:0.05).

3.2 Active Layer Depth

Differences between thaw depth of impacted and non-impacted areas within a plot
were determined by comparison of means using t-tests. Each plot contained 105 sample
points allowing 103 degrees of freedom. Out of the 25 comparisons made, 22 plots
showed significant differences between impacted and non-impacted subplots on at least
one sampling date. Seven plots showed significant differences between the two
conditions at all sampling dates. Plots with the most significant differences were those

where track depressions have lead to ponding. The plots with the least significant

36

differences between impacted and non—impacted areas were typically those that were
uniformly wet.

Using the average thaw depth within a plot as a single data point, a comparison
between whole plots containing disturbances and undisturbed plots within a particular
moisture regime was made for each site. This was done with the wet meadow and dry
slope plots at Sites 2 through 4. Site 1 was not used due to a lack of undisturbed plots.
These comparisons were made for each of the 4 sampling dates, as well as the first test
sampling at Site 2 on June 28. The thaw depth of disturbed plots was significantly deeper
in 5 of the 17 comparisons made. In no comparison was the thaw significantly shallower
for disturbed plots. None of these comparisons done at Site 3 showed significant
differences between plots containing disturbed terrain and those that did not. The wet
meadow plots at Sites 2 and 4 showed the most significant differences on the July 23rd
sampling date (Site 2: Tcalc = 3.803, Tait: 2.132; Site 4: Tcalc = 2.608, Tait: 2.132), while
Site 2 also showed significant differences on the August 12th sampling date (Tmc = 2.771,
Tm; 2.132) The dry heath plots at Site 4 showed the most significant differences at the
plot level and steadily increased in significance until the last sampling on August 12th
(August 2: Tcalc = 4.342, Tait: 2.132; August 12: Tcalc = 6.500, Ten-t: 2.132). See Table 2
for statistics of the comparisons.

All of the data from the 4 complete sampling dates were analyzed together using a
repeated measures model using SAS (ver. 8.0) to look at the effect of time on the thaw
depth progression and to compare the differences in the effect over the different sites. All
of the tests for fixed effects showed significance in the study (p <0.0005). The

Kohnogorov-Smirnov, Cramer-Von Mises, and Anderson-Darling tests for normality

37

showed that the data were significantly different from normal (p values <0.01, (0.005
and <0.005, respectively). It should be noted that with data sets as large as this the
statistical power alone will often lead to problems with normality tests. Therefore, the

results of the test will be reported here.

Table 2: Differences between disturbed and non-disturbed terrain at individual sites for
each sampling date. Degrees of freedom and t statistics are given with an indication of

significance.

 

 

 

Site Moisture Date dof tgL 19m Significant
2 wet meadow 06/28/01 4 0.3072 2.132 No
2 wet meadow 07/13/01 4 0.2708 2.132 No
2 wet meadow 07/23/01 4 3.8034 2.132 Yes
2 wet meadow 08/02/01 4 1 .9796 2.132 No
2 wet meadow 08/12/01 4 2.7710 2.132 Yes
3 dry slope 07/13/01 4 0.0319 2.132 No
3 dry slope 07/23/01 4 1.5368 2.132 No
3 dry slope 08/02/01 4 1.1 190 2.132 No
3 dry slope 08/12/01 4 0.6637 2.132 No
4 dry slope 07/1 53/01 4 0.0720 2.132 No
4 dry slope 07/23/01 4 1 .7430 2.132 No
4 dry slope 08/02/01 4 4.3427 2.132 Yes
4 dry slope 08/12/01 4 6.5002 2.132 Yes
4 wet meadow 07/13/01 4 0.5661 2.132 No
4 wet meadow 07/23/01 4 2.6082 2.132 Yes
4 wet meadow 08/02/01 4 1.0715 2.132 No
4 08/12/01 4 0.5401 2.132 No

wet meadow

 

Sample date had the highest F statistic (4046.74) of any effect on thaw depth
measurement. The condition of the terrain, disturbed or undisturbed, bad the second
highest F statistic of 161.86. The lowest F statistic (5.94) was for the effect of the site on

thaw depth, but this was still highly significant (p<0.0005). The interactions between site

38

and condition, date and condition, and date and site were all highly significant as well.

See Table 3 for degrees of freedom and statistics.

Table 3: F statistics for variables related to thaw depth including numerator and
denominator degrees of freedom.

 

Effect Num DF Den DF F Value Pr > F

 

Site 3 4052 5.94 0.0005
Cond 1 4052 161.86 <0.0001
Site'Cond 3 4052 23.73 <0.0001
Date 3 12000 4046.74 <0.0001
Date‘Cond 3 12000 82.2 <0.0001
Date‘Site 9 12000 97.95 <0.0001

 

Because the initial test indicated that there were significant differences between
the thaw depths of sites and dates, and interactions between the effects, separate
statistical tests were’run for each site on each sampling date to further examine these
differences. Again the data were significantly different from normal (Kohnogorov-
Smimov, Cramer-Von Mises, and Anderson-Darling p values: <0.01, <0.005 and <0.005,
respectively). Wet sites behaved differently than dry sites throughout the study. Site 2,
the wettest site, did not show a significant difference in thaw depth between impacted and
non-impacted points until the last sampling on August 12*. Site 4, which has many wet
plots as well, did not show a significant difference between impacted and non-impacted
points until the second sampling on July 23'“. The difference at both of these sites seems
to increase throughout the summer until reaching significance. Sites 1 and 3, which are
the driest sites, showed significant differences from the beginning, and the differences

seem to be unaffected by date of sampling over the course of the study period. The

39

difference at Site 3 seems to begin to decline at the last sampling and the differences are
likely to disappear beyond the end of the sampling period. It is interesting to note that the
most significant differences are found at Site 1, which is intermediate in moisture when

compared to the other sites. See Table 4 for statistics.

Table 4: Statistics for thaw depth comparisons between impacted and non-impacted grid
points for each site on each sampling date.

 

Site Date Num DF Den DF F Statistic p value Significance

1 7/13/01 1 519 71.90 <0.0001 Yes
1 7/23/01 1 519 122.38 <0.0001 Yes
1 8/2/01 1 519 122.25 <0.0001 Yes
1 8/12/01 1 519 127.22 <0.0001 Yes
2 7/13/01 1 623 0.00 0.9793 No

2 7/23/01 1 623 0.02 0.8967 No

2 8/2/01 1 623 0.33 0.5645 No

2 8/12/01 1 623 4.83 0.0283 Yes
3 7/13/01 1 1143 25.29 ' <0.0001 Yes
3 7/23/01 1 1 143 23.36 <0.0001 Yes
3 8/2/01 1 1143 21.15 <0.0001 Yes
3 8/12/01 1 1143 13.87 0.0002 Yes
4 7/13/01 1 1767 0.06 0.8127 No

4 7/23/01 1 1767 52.20 <0.0001 Yes
4 8/2/01 1 1767 51.15 <0.0001 Yes
4 8/12/01 1

1767 39.62 <0.0001 Yes

 

3.3 Soil Moisture Content

A comparison was conducted for the whole study area of the percent soil moisture
for impacted areas, areas outside of tracks, and areas between tracks of each plot for
which these data are available. These data were analyzed in SAS (ver. 8.0). Again, the
data were non-normally distributed (Kolmogorov-Srnirnov, Cramer-Von Mises, and
Anderson-Darling p values: <0.01, <0.005 and <0.005, respectively), but the results are

interesting. There was a significant difference between the soil moisture of each site

40

(numerator degrees of freedom (df) =3, denominator df=170, p value <0.0001). However,
the effect of the position of the samples relative to the tracks was not significant
(numerator df =2, denominator df=170, p value =0.6596). The interaction between site
and position was also insignificant (numerator df =6, denominator df=170, p value
=0.8641), suggesting that the differences in soil moisture between positions would not be
significant regardless of the site.

This was confirmed by analyses of the data for each site separately. Even at Site 1
where there was the greatest difference in surface water content between impacted and
non-impacted areas, there was no significant difference in soil moisture content
(p=0.1746). The differences at the remaining sites were not nearly so close to significant
(Site 2: p=0.9223, Site 3: p=0.9335, Site 4: p=0.9375).

A simple linear regression was run to test the effect of soil moisture on thaw
depth using the average thaw depth in meters on 8/12/01 (y) and percent soil moisture (x)
for impacted and non-impacted areas of the plots. The analysis of variance showed a
weak, but highly significant, trend (p<0.0001). As the soil moisture content increased, the
thaw depth decreased. The equation for the trend is y=-.0031x + 0.5498, with an r2 value

of 0.2679.

3.4 Soil Compaction

In addition to the soil moisture, a comparison was conducted for the whole study
area of the soil compaction for impacted areas, areas outside of tracks, and areas between
tracks of each plot for which these data were available. Soil bulk density was used for

this comparison. The data were also non-normally distributed (Kolrnogorov-Smirnov,

41

Cramer-Von Mises, and Anderson-Darling p values: <0.01, <0.005 and <0.005,
respectively), and the results are similar to those from the soil moisture data. There was a
significant difference between the soil bulk densities of each site (numerator degrees of
freedom (df) =3, denominator df=170, p value <0.0001). The effect of the position of the
samples relative to the tracks was not significant (numerator df =2, denominator df=170,
p value =0.5943). The interaction between site and position was also insignificant
(numerator df =6, denominator df=170, p value =0.9613), suggesting that the differences

in soil moisture between positions would not be significant regardless of the site.

3.5 Ordination

The following results are based on principal components analysis (PCA) of the
relative plant cover data. Graphing the PCA of the relative cover of all species in all
subplots demonstrates a greater variation within the control subplots than the impacted
subplots. Axis I of the ordination captured 16.9% of the variation and was most closely
correlated with average percent of bare soil (r2=0.379). Axis 2 captured 8.1% of the
variation and was most closely correlated with the average active layer depth on August
12 (r2=0.116). Axis 3 captured 7.1% of the variation and was most closely correlated with
active layer depth on August 2 (r2=0.061). Correlations with soil moisture and pH were

not conducted for this ordination (Figure 14).

42

Axis 3

 

 

 

 

 

 

 

 

 

 

A AA
AAA A
AA A
A
lmpacted‘
0. ControlA
AA
A
Ta is A ’ “
80i -A A
AA A A“ A A
A AM A :3.“ A
09A . AAA A
A A 93 A
40‘ r. . 2. . A
A A
O“ v I V l 1
0 40 0 40 80
Axis1 Axis2

 

 

Figure 14: Results'of the PCA for all plots in the study.

The results of the PCA for plots with moisture data show a greater variation in the

control subplots than is present in the impacted subplots (Figure 15). Axis I of the

ordination corresponds most closely with percent of bare soil (r2 = 0.371), axis 2 with

average soil pH (r2 = 0.111), and axis 3 with late season thaw depth (r2 = 0.243). Percent

soil moisture has a lower correlation with these axes (r2=0.244, 0.002, 0.182,

43

respectively). Axis 1 captured 17.9% of the variation, axis 2 captured 9.0%, and axis 3

captured 7.5%. The cumulative variance captured by these three axes is 34.4%.

 

 

 

 

 

 

 

 

 

 

 

 

A A
80‘ A ’5
A
A
N A
.9 A; A
40 A .
5.5 A A. t ‘ It
A A ‘ A
A A ‘ A
A A Impacted A
Control A
0 ~ - - -
4A A
A A
80‘
AA ‘ ‘l A
00 A A A A AA ‘fi . A flea A: A A
.3 ‘2 ‘ ‘A A
A
40- A‘ A ‘ A. 5
A A
A A
0 40 80 0 40 80
Axis 1 Axis 2

Figure 15: Results of the PCA for those plots for which soil moisture data were
collected.

Separate PCAs were conducted for the control and impacted subplots using those

plots from which soil moisture data were collected. The impacted subplots show slightly

higher correlations with the environmental variables than do the controls. Axis I of the
ordination for the controls corresponds most closely with early season thaw (r2 = 0.576),
axis 2 with average soil pH (r2 = 0.172), and axis 3 with soil moisture content (r2 =
0.244). Percent soil moisture has correlations with the first two axes of r2=0.342 and
0.004, respectively. Axis I captured 19.4% of the variation, followed by axis 2 with
11.4%, and axis 3 with 8.4% (Figure 16). The cumulative variance explained by these
three axes is 39.3%.

Axis I of the ordination for impacted subplots corresponds most closely with late
season thaw (r2 = 0.625), axis 2 with thaw depth on week 2 (r2 = 0.309), and axis 3 with
early season thaw depth (r2 = 0.243). Percent soil moisture has correlation with these

axes of r2=0.161, 0.041, 0.098, respectively). Axis I collected 25.3% of the variation,
followed by axis 2 with 10.2%, and axis 3 with 9.5% (Figure 17). The cumulative
variance captured by these three axes is 45.0%.

The individual sites showed varying patterns in the ordination results. Separate
Principal Components Analyses were conducted based on relative plant cover data to
identify which factors were potentially controlling in the different sites. These ordinations

only deal with data collected from plots where soil moisture data is available.

45

 

 

 

 

 

 

 

 

 

or
.9
:2 40‘. , .
A A
A A A
A
0‘, -
90‘ A A
A
co 70‘
.2 A A
2 . .
A
50.
A
30‘? V .
0 4O
Axis1

Figure 16: PCA reSults for control subplots from which soil moisture data were

collected.

46

Axis 2

so

 

 

 

 

 

 

 

 

 

 

 

 

 

90‘ A
AA
70‘
N 50<
.92
31‘ ‘ a
A A AA ‘ A
30* A A
l A A
10 , A
100‘ A A
80‘ A A
AAA AA
2 60 AAA ‘ A A
A
g 4 A Ang A ARA;
40.
A A
A A
20 - 1 j A A
0 40 80 10 30 50 7O 90
Axis1 AxisZ

Figure 17: PCA results for impacted subplots from which soil moisture data were
collected.

3.5.1 Site 1
Axis I of the ordination for Site I captured 41.1% of the variation in the relative

plant cover data and corresponds most closely with average soil pH (r2 = 0.437). Axis 2

captured 27.5% of the variation and correlates most closely with the average percent of

surface water in the plot (r2 = 0.342). Axis 3 captured an additional 11.9 % of the

47

variation and correlates well with the average microtopographic height of the plots (r2 =

0.829). Percent soil moisture has correlation with these axes of r2=0.040, 0.080, and

0.020, respectively (Figure 18). The first 3 axes explain a total of 80.5% of the variation.

 

60‘

4o«

Axis 2

20‘ A

i Impacted A
Control A

 

 

 

 

70‘

Axis 3
0|
9

 

 

 

30 .
- . - A . . -

 

 

o 40 80 o éo ' 4‘0 ' so
Axis1] Axis 2

Figure 18: PCA results for plots at Site 1 from which soil moisture data were collected.

48

3.5.2 Site 2

Axis I of the ordination for'Site 2 captured 44.5% of the variation in the relative
plant cover data and has its highest correlation with the average percent of surface water
in the plot (r2 = 0.719). Axis 2 captured 32.3% of the variation and correlates most
closely with the average soil pH (r2 = 0.613). Axis 3 captured 11.4% of the variation and
correlates best with the average active layer depth on July 23 (r2 = 0.419). Percent soil
moisture has correlation with these axes of r2=0.315, 0.097, and 0.119, respectively
(Figure 19). The first 3 axes captured a total of 88.2% of the variation.
3.5.3 Site 3

Axis I of the ordination for Site 3 captured 23.8% of the variation in the relative
plant cover data and corresponds most closely with the average microtopographic height
of the plots (r2 = 0.626). Axis 2 captured 21.6% of the variation and correlates most
closely with the active layer depth on August 12 (r2 = 0.695). Axis 3 captured 12.1 % of
the variation and has its highest correlation with the average percent of bare soil (r2 =
0.207). Percent soil moisture has correlation with these axes of r2=0.046, 0.027, and

0.136, respectively (Figure 20). The first 3 axes capture a total of 57.6% of the variation.

49

Axis 2

Axis 3

Figure 19: PCA results for plots at Site 2 from which soil moisture data were collected.

60‘

40'

20

 

 

Impacted A
Control A

 

 

 

 

 

4o
Axis 1

O

80

50

20

Axis 2

 

 

 

 

 

 

 

 

 

 

80“ A
. ‘ A
60‘ g A
A
J A
N 40‘
.92
:2 u.
20‘
A
A lmpacted‘
01' A i f ControlA
A A
90‘
A A
m 701 4
.2 A A
SOAA‘ ‘ ‘ ‘ ‘ A
A A
30-.-.r’3u-.,-.°efi.
0 40 80 0 2O 40 60 80
Axis1 ’ AxisZ

Figure 20: PCA results for plots at Site 3 from which soil moisture data were collected.

3.5.4 Site 4

Axis I of the ordination for Site 4 collected 25.6% of the variation in the relative

plant cover data and has its highest correlation with the average thaw depth on July 13 (r2

= 0.387). Axis 2 collected 13.0% of the variation and also correlates most closely with

the average thaw depth on July 13, though the correlation is weaker than for axis 1 (r2 =

51

0.077). Axis 3 captured 11.1% of the variation and correlates best with the average active

layer depth on August 12 (r2 = 0.576). Percent soil moisture has correlation with these

axes of r2=0.088, 0.047, and 0.021, respectively (Figure 21). The first 3 axes explain a

 

 

 

 

 

 

 

 

total of 49.7% of the variation.
A
80‘ A
“‘ A
(D
3 4° A .
A A‘ “A
A
A Impacted 1*
0 Control A
A
A A
80* A ‘A
A
A A
60.
(0 AAA AA A
.9 A A A ‘
5°
é A MA A A A
40‘ A A
A A
20 - - . j g _ - - .
0 4O 80 O 40 80
Axis 1 Axis 2

Figure 21: PCA results for plots at Site 4 from which soil moisture data were collected.

52

 

3.6 Classification

The statistical classification of the sites based on the relative cover of plants
within the plots separated the plots into three fairly distinct groups. For purposes of the
classifications each plot is separated into its control impacted subplots, designated by
their plot number followed by 01 or 02, respectively. For example, the impacted portion
of plot 5 is designated 0502. For the most part, nearby sites are grouped together in the
classification. All of the plots at Site 2 are separated into a group by themselves, along
with three of the plots at Site 1. The second and third groups are more closely related to
each other than to the first. One of these groups is made up almost entirely of dry slope
plots from Sites 3 and 4 and includes no plots from Sites 1 or 2. The last group is more
variable and contains most of the plots from Site 1, as well as most of the plots from Sites
3 and 4 that were not located on the dry slope. Figure 22 shows these relationships and
also includes a subjective classification into groups that roughly correspond to “wet”,
“dry” and “moist” tundra. This subjective classification was done using simple
observation in the field during the establishment of the plots. Although these two
classifications are fairly consistent, plots 23 to 28 at Site 4 were not placed in the same
group by the statistical classifier compared to the subjective method. Also interesting is
the separation of the impacted portion of plot 4, designated 0402, from its non-impacted
counterpart 0401.

In a similar classification using vegetation cover data from only the plots where
soil moisture data. were available, the plots separate naturally into two groups
subjectively designated “wet” and “dry” (Figure 23). There is a small group that

corresponds most closely to the “moist” group from the full study classification, but it

53

Distance Objsctive Function)
2.1§+01 . 2.7.804 L .5§+ 8.2.E+04 .1E+0_5

4

lnforrnation Remaining (‘70)
100 75 50 2 .

I I I I I I I

wet

   
        
 
 
   

  

0101-0502: Pitelka
0601-1101: CRREL TEST
1201-2202: New CRREL
2301-3902: Old CRREL

moist

dry

r-{

Worst
damage:

Figure 22: Classification of all plots based upon relative cover of vegetation. The
designations wet, moist and dry indicate subjective classifications made in the field.
Plots 0401 and 0402 (indicated by asterisks) are corresponding damaged and

undamaged plots which are separated by the greatest distance possible in the
classification.

54

contains only plots 37 through 39. These plots are located on an elevated flat just above
the dry slope at Site 4. Because of the small size of this group it may be more useful to
consider them a part of the dry classification to which they are most closely related.

The plots that were subjectively designated “wet” and “dry” during the initial site
setup, are again in distinct groups much as they were in the whole-study classification.
Those plots that had been designated “moist” are distributed fairly evenly into both the
“wet” and “dry” groups in this second statistical classification, rather than being in their
own group as they were in the full-study classification. Some relationships are closer in
this second classification and others are farther away. The two parts of plot 4, which had
the lowest possible similarity in the first classification, are fairly closely related in this
second one. Conversely, although still all within the same major grouping, the plots of
Site 2 are nearly as closely related in this classification as they were in the first.

A The plots for which there was moisture data available were also classified based
on their environmental characteristics rather than their vegetation composition. Again,
two major groups were formed (Figure 24). The major environmental factor determining
the clusters appears to be soil moisture. Within the major groupings, percent surface
water seems to dictate the sub-groupings. In this case, the plots from Sites 1 and 2 made
up the majority of the entries in the first group and within the group these two sites were
separated quite distinctly. Plot 2402 was the only plot within this group that was not
found at Sites 1 or 2. Similarly, plot 0302 was the only plot in the second group that isn’t
located at Sites 3 or 4. These two plots had the least possible similarity with their non—
irnpacted counterparts, which classified as expected. Unlike Sites 1 and 2, Sites 3 and 4

are not distinct from one another within the major grouping to which they belong.

55

3.5E+O1

Distance (Objective Function)

2.1.E-l-04

Information

75

4.35104

5.0

I

 

 

 

 

 

 

 

 

 

 

 

 

6.4.E+04

I

Remaining (°/o)
25

8.5E;I-04

 

 

 

 

 

 

 

 

0101-0502: Pitelka

0601 -1 1 01: CRREL TEST
1201 -2202: New CRREL
2301 -3902: Old CRREL

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1

Figure 23: Classification of all plots from which soil moisture data was taken. The
plots are classified into two major groups which can be subjectively called wet and

dry.

56

Distance (Objective Function)

 

1.5.501 J 7.215103 . 1.4.E+04 . 2.2.E+04 . 2£E+04
Information Remaining (%)
25 0

1 00 75 50

 

 

 

 

2

(9 060
080% 0101-0502: Pitelka
070

070 0601 -1 101: CRREL TEST
1201-2202: Old CRREL
2301-3902: New CRREL

 

 

 

iiiiii:

33§§§M§§

0°C 00

 

 

 

Group 2
”Aw—A
Amߤ

 

ii.

 

 

 

 

 

 

 

Figure 24: Classification based on environmental variables of all plots from which
soil moisture data was taken. The plots are classified into two major groups that
generally separate Pitelka’s trail and the CRREL test trail from the old and new
CRREL trails.

57

3.7 Species Losses

Over the whole study site there were species which do not occur in impacted
subplots, but which are present in the corresponding non-impacted areas. 70 different
species were encountered in soil moisture plots throughout the study. Only 58 were found
in both impacted and non-impacted areas. Of the rest, 11 species were only found in non-
irnpacted subplots, and one species, Bryum cyclophyllum, which was not found in any
non—impacted subplot, was found in one impacted subplot. Species area curves were used
to determine if these losses were simply due to a smaller sample area and relatively rare
species. 278 subplots were impacted and contained 59 species. 278 subplots sampled at
random from the 2078 total subplots for generation of the species area curves had a mean
number of species of approximately 64. Graphs of the species area curve with the

standard deviation about the mean are shown in Figure 25.

 

 

 

 

 

 

a)

.9

§

a)

”5

5

.0

E ,

3

Z 20 -

0)

U)

S

g I

<

-20 I I I I I I I I I I
525 1050 1575 2100
Number of Subplots

Figure 25: Species-area curve for all plots from which soil moisture data was taken.
The coordinate listed indicates the species richness for the damaged subplots of the
study. The hashed lines indicate 1 standard deviation about the mean.

58

The species losses in the impacted areas were not uniform across the study. The 4
sites had different diversities and the effects of the disturbance were more or less
pronounced.

At Site 1, Pitelka’s trail, there was a total of 36 impacted subplots containing 17
species. 36 subplots sampled at random from the 320 total subplots for generation of the
species area curves had a mean number of species of approximately 28. Graphs of the

species area curve with the standard deviation about the mean are shown in Figure 26.

 

Average Number of Species

 

 

 

 

0 1 00 200 300
Number of Subplots
Figure 26: Species-area curve for all plots at site 1, from which soil moisture data was
taken. The coordinate listed indicates the species richness for the damaged subplots.
The hashed lines indicate 1 standard deviation about the mean.

59

At Site 2, the CRREL test site, there was a total of 52 impacted subplots
containing 10 species. 52 subplots sampled at random from the 240 total subplots for
generation of the species area curves had a mean number of species of approximately 11.
Although the impacted subplots contained one fewer species, this was still within the
expected range of species richness. Graphs of the species area curve with the standard

deviation about the mean are shown in Figure 27.

 

-----
---------
‘_"',--UW‘
..-‘ola‘

   

,,,,,

 
 

_ fl. «156,3 0)

Average Number of Species

 

 

 

o 100 200
Number of Subplots

Figure 27: Species—area curve for all plots at site 2, from which soil moisture data was
taken. The coordinate listed indicates the species richness for the damaged subplots.
The hashed lines indicate 1 standard deviation about the mean.

At Site 3, the oldest of the CRREL trails, there was a total of 64 impacted
subplots containing 38 species. 64 subplots sampled at random from the 560 total

subplots for generation of the species area curves had a mean number of species of

approximately 48. Graphs of the species area curve with the standard deviation about the

mean are shown in Figure 28.

 

 

 

Average Number of Species

 

 

 

 

I I I I ‘ 1 I j

o 200 400 600
Number of Subplots

Figure 28: Species-area curve for all plots at site 3, from which soil moisture data was
taken. The coordinate listed indicates the species richness for the damaged subplots.
The bashed lines indicate 1 standard deviation about the mean.

At Site 4, the younger of the CRREL trails, there was a total of 126 impacted
subplots containing 53 species. 126 subplots sampled at random from the 958 total
subplots for generation of the species area curves had a mean number of species of
approximately 58. Graphs of the species area curve with the standard deviation about the

mean are shown in Figure 29.

61

 

 

 

Average Number of Species

 

 

-10.

 

I I I I I I I I I

1 240 480 720 960
Number of Subplots

Figure 29: Species-area curve for all plots at site 4, from which soil moisture data was
taken. The coordinate listed indicates the species richness for the damaged subplots.
The hashed lines indicate 1 standard deviation about the mean.

In the examination of species richness based on the distance from the impact there
is considerable overlap in the standard deviation about the mean of the curves, indicating
that the species richness at any given distance will be fairly similar to that of any other
distance. In both the whole-study case and the cases of the individual sites, the impacted
subplots and the subplots between tracks that did not experience a physical impact,
consistently have the lowest species richness. The species richness generally increases
with distance away from the impact (Figures 30-34). One exception occurred at Site 4

where the lowest species richness was found at the three greatest distances away from the

impact (Figure 34).

62

 

‘4'"

I ”‘1'“

 

   

— Damaged

 

 

 

 

 

 

U)
.9
0
ID
0.
a)
“5 50-
:5 Between track
D - — ' 0-0.5m
g "' 0.5-1m
E 30. -'- 1—1.5m
g —— 1.5-2m
Er ' — - - 2-2.5m
< —' ' 2.5-3m
10' 3-3.5m
- IIIII 3.5-4m
0 100 200 300
B Number of Subplots
60
50
3 f l i i i i r i
r t 4or
to
Q-
2 30
8
g 20
2
10
, 0
-1 0 0-0.5 0.5-1 1-1.5 1.5-2 2-2.5 2.5-3 3-3.5 3.5-4
Dlstanee from Disturbance (m)

Figure 30: (A) Species richness at a given distance from impact over the whole study.
(B) Mean species richness of 26 randomly chosen grid cells at a given distance from
impact over the whole study. A value of 26 grid cells was chosen to approximate the
inflection point of the species area curve for the impacted subplots. A negative
distance indicates subplots found between tracks. Error bars represent one standard
deviation about the mean.

63

i9

 

 

    
 

    

 

 

m I
.9
8 30- ___
C?)- .-¢""-fl..acflo""'u°°'.'.’.:_'f
0‘. w. p:'__.""1°—‘—-T HI—
? u . ogw‘;:':ét':‘T—;-
m O.../.’o’-"' ‘-
.0 / o’ ”' ’
E 20" ‘9' .: I .— fl
5 . .-/-' I
on
or
‘3 r
g 10 Between track —- - 2-2.5m
< . — - 0-0.5m —-- 2.5-3m
--- 0.5-1m "°" 3-3.5m
0 ‘°' 1-1.5m ""' 3.5-4m
10 20 30
B Number of Subplots
25'
20- { { { { {
(I)
.9
§. 151 I
(D
”5 ..
g i 101»
E ..
E
51
.0

 

 

0 0-0.5 0.5-1 1-1.5 1.5-2 2-2.5 2.5-3 3—3.5 3.5-4
Distance from Disturbance (m)

-1

Figure 31: (A) Species richness at a given distance from impact for Site 1. (B) Mean
species richness of 5 randomly chosen grid cells at a given distance from impact for
site 1. A value of 5 grid cells was chosen to include the inflection points of the species
area curves for all distances from the impact. A negative distance indicates subplots
found between tracks. Error bars represent one standard deviation about the mean.

 

   

 

 

 

 

m 12 '
.9 .0 ...-3“"
o o "
o
a
..5 — Damaged
5 Between track
.0 - ' 0 — 0.5m
g ‘ " " 0.5-1m
f, - - - 1-1.5m
g’ —' - 1.5-2m
§ — - - 2-2.5m
< . -' ' ° 2.5-3m
. . . . . 3-3.5m
o . I I I I I I I
0 20 40 60
B Number of Subplots
12'
10‘ i i
U)
o
g: l 8:. i i i i i
(I) d0
‘5 64
g
E 4'
3
Z 2
0

 

 

-1 0 0-0.5 0.5-1 1-1.5 1.5-2 2-2.5 2.5-3 3—3.5
Distance from Disturbance (m)

Figure 32: (A) Species richness at a given distance from impact for Site 2. (B) Mean
species richness of 7 randomly chosen grid cells at a given distance from impact at
site 2. A value of 7 grid cells was chosen to include the inflection points of the species
area curves for all distances from the impact. A negative distance indicates subplots
found between tracks. Error bars represent one standard deviation about the mean.

65

a»

 

 

 

 

 

 

ul
§ 40-
so.) - I — Damaged
g 30‘ I, Between track
.0 — " 0 - 0.5m
g ' " - ' 0.5-1m
E 20.1 - ° - l-l.5m
g ‘ —' - 1.5-2m
1:: — - - 2-2.5m
< 10‘ — ' ° 2.5-3m
. ...... 3-3.5m
0 . III-II 3.5-4m
0 20 40 60
B Number of Subplots
40.
.g i 30 .1»
a 4)
(D 25"
”E 20-
8
g 15"
Z 10q
5 .l

 

 

31 0 0:05 0.5-1 151.5 1.5-2 252.5 2.5-3 323.5 3.5-4
Distance from Disturbance (m)

Figure 33: (A) Species richness at a given distance from impact at Site 3. (B) Mean
species richness of 9 randomly chosen grid cells at a given distance from impact at
site 3. A value of 9 grid cells was chosen to include the inflection points for the
species area curves at each distance from the impact. A negative distance indicates
subplots found between tracks. Error bars represent one standard deviation about the
mean.

 

 

 

 

 

 

 

a) .. —
.9 " "'
fig 50-
%.: — Damaged
5 .1 Between track
.2 - ' 0 - 0.5m
3 30' - - ' 0.5-1m _
g --- 1-15m
g - — - 1.5-2m
1;) — - - 2-2.5m
<2 10. — ' ' 2.5-3m
. . . . . 3-3.5m
I I I I I 3.5-4m
0 40 80 120
B Number of Subplots
45'
40 i } f {
g i 3511 { .
'6 30-’ f
o
8 25- i i
3 201
3 .
g 15
Z 10‘
5 I

 

 

-1 0 005 0.5-1 1-1.5 1.5-2 2-2.5 2.5-3 3—3.5 3.5-4
Distance from Disturbance (m)

Figure 34: (A) Species richness at a given distance from impact at Site 4. (B) Mean
species richness of 15 randomly chosen grid cells at a given distance from impact at
site 4. A value of 15 grid cells was chosen to include the inflection points for the
species area curves at each distance from the impact. A negative distance indicates
subplots found between tracks. Error bars represent one standard deviation about the
mean.

67

The results of the analysis of variance that tested for the effects of distance from
impact on species richness and the comparisons made with Tukey’s test vary between
sites. At Site 1, there was a significant effect of distance class on richness (p=0.002, error
df=27, or=0.05). The species richness of the impacted subplots and the subplots located
between the tracks was significantly lower than the species richness of all the other
distance classes. Areas between 1.5 and 2m away from the impact and areas between 3
and 3.5 meters from impact had significantly higher species richness than all other

distance classes (see Table 5).

Table 5: Comparison of species richness for Site 1, based on distance from impact.
Significance class indicates which classes are significantly different from each other.
Distance classes that share a letter are not significantly different. For example, a is
significantly different from b, but ab is significantly different from neither a nor b.

 

Distance Class (m) Upper 95% Cl Lower 95% Cl ggnificance Class

 

between 3.54 10.96 a
impacted 4.54 1 1.96 a
0-0.5 7.04 14.46 ab
0.5-1 8.79 16.21 ab
1-1.5 10.79 18.21 ab
1.5-2 12.79 20.21 b
2-2.5 11.04 18.46 ab
2.5-3 1 1.79 19.21 ab
’3-3.5 14.29 21.71 b

 

At Site 2 there was no significant effect of distance on species richness (p=0.811,
error df=16, a=0.05). The pair-wise comparisons showed that no distance class had

significantly different species richness from another (see Table 6).

68

Table 6: Comparison of species richness for Site 2, based on distance from impact.
Significance class indicates which classes are significantly different from each other.
Distance classes that share a letter are not significantly different. For example, a is
significantly different from b, but ab is significantly different from neither a nor b.

 

Distance Class (m) Upper 95% Cl Lower 95% Cl Significance Class

between 5.43 9.90 a
impacted 4.43 8.90 a
0-0.5 5.43 9.90 a
0.5-1 5.10 9.57 . a
1-1 .5 4.43 8.90 a
1 .5-2 6.10 10.57 a
2-2.5 6.43 10.90 a
2.5-3 6.10 10.57 a

 

At Site 3, there was a significant effect of distance class on richness (p<0.001,
error df=54, a=0.05). The species richness of the subplots located between the tracks was
significantly lower than the species richness of all the other non-impacted subplots. The
species richness of the impacted subplots was not significantly different from either the
areas between tracks or the areas immediately adjacent to the impact. It was, however,
significantly lower than that of all other distance classes. All of the distance classes
beyond 0.5 meters away from the disturbance had species richnesses significantly higher
than that of the impacted subplots. The species richness of the area between 2.5 and 3
meters away from impact is also significantly higher than that of the areas immediately
adjacent to the impact (see Table 7).

At Site 4 there was no significant effect of distance on species richness (p=0.993,
error df=48, or=0.05). The pair-wise comparisons showed that no distance class had

significantly different species richness from another (see Table 8).

69

Table 7: Comparison of species richness for Site 3, based on distance from impact.
Significance class indicates which classes are significantly different from each other.
Distance classes that share a letter are not significantly different. For example, a is
significantly different from b, but ab is significantly different from neither a nor b.

 

Distance Class (m) Upper 95% Cl Lower 95% Cl Significance Class

between 12.49 18.65 a

impacted 15.78 21 .93 ab
00.5 20.21 26.36 bc
0.5-1 23.35 29.51 cd
1-1 .5 24.78 30.93 cd
1 .5-2 25.78 31 .93 cd
2-2.5 24.92 31 .08 cd
2.5-3 26.64 32.79 (1
33.5 23.92 30.08 cd

 

Table 8: Comparison of species richness for Site 4, based on distance from impact.
Significance class indicates which classes are significantly different from each other.
Distance classes that share a letter are not significantly different. For example, a is
significantly different from b, but ab is significantly different from neither a nor b.

 

(Distance Class (mLUpper 95% Cl Lower 95% CI Significance Class

between 12.07 25.93 a
impacted 1 1 .18 25.04 a
005 10.96 24.82 a
0.5-1 13.41 27.26 a
1-1.5 10.52 24.37 a
1.5-2 12.07 25.93 a

 

70

CHAPTER 4

4 Discussion
4.1 Effects Of Terrain Damage

Signatures of vehicle trails are still present after 30 years or more of natural
recovery. There are differences in the appearance of the trails, the thermal regime of the
soil and in the plant communities themselves. Because there is a significant difference in
the topography of the plots, there is a possibility that drainage and moisture levels within
the tracks and adjacent areas of tundra also differ. This could have an effect on the plant
community in these areas in the long run (Pearson 1994).

It should be noted that the differences, although statistically significant are not
large. Because the landscape inherently has a large degree of microscale heterogeneity,
the degree to which this difference will be ecologically significant is determined, in large
part, by the characteristics of the site. On a dry slope where the soil is well drained, the
depressions caused by vehicle traffic will not likely affect the surface or soil moisture
levels. Similarly, in wet meadow areas that are typically submerged for a large part of the
growing season, the change in the depth of standing water caused by off-road traffic is
not likely to have an effect on the plants.

There can be ecologically significant differences in areas of moderate drainage.
At Pitelka’s trail, Site 1, the depression of the ground due to vehicle tracks has led to
ponds forming in several locations. The plant communities in these ponds are drastically
different from those outside the ponds. The result is small patches of wet meadow-like
communities within an area dominated by lichen-heath type plants. This site is the most
polygonized of the four sites and so we may expect to see different plants within polygon

troughs, but the difference between a trough community and a polygon center community

71

typical of the site, is not as distinct as the differences seen between the impacted and non-
impacted areas found here. It is likely that the succession to a wet community is
accentuated a great deal by the microtopographic differences caused by the off-road
traffic at this site.

There seems to be no difference between the impacted and non-impacted areas in
terms of soil moisture and soil compaction. This is encouraging because it suggests that
the soil environment has recovered to the point where the root systems of the plants will
not be affected by their location in relation to the tracks. There are, of course, other
differences in the soil. No nutrient analyses were done in this study. If differences were
observed in the nutrient levels of the soil it would be important for the recovery of the
plants.

Though soil moisture and compaction were similar between impacted and non-
impacted areas, one soil property, which differed significantly, was the extent to which
the soil was able to thaw. In the most drastic case, an increase in active layer thickness
allows plant roots to grow deeper and have access to more nutrients. This has the
potential to effect plant vigor leading to larger, more robust plants with higher fitness.
The effects would likely differ between species and the competition between species
would be affected. Over the long-term, successional pathways could be changed by
effects like this. The case presented here is an extreme one, but may not be terribly
unrealistic given that the differences in thaw depth between impacted and non-impacted
terrain are still significant after as much as 40 years of natural recovery. Long-term
exposure to altered conditions can be an important driver of community change.

Increased microbial activity in a thicker active layer over such a long period of time

72

could greatly affect the nutrient levels in the soil. There is concern that carbon dioxide
flux from the soil would be altered as well by these types of conditions.

Other evidence for successional change is found in the vegetation ordinations.
The graphs of the Principal Components Analysis (PCA) of the vegetation data for the
whole study demonstrate a greater variation in the relative cover of non-impacted
subplots than that of impacted subplots. A possible explanation for this is that the
disturbance caused by the vehicles has reset succession back to some baseline in the
impacted plots.

If one considers that succession is not unidirectional, it is logical to think that
undisturbed areas will diverge in their vegetation communities within the range of those
communities able to cope with the particular environmental restraints present. This
divergence would be based on the chance arrival of species and the fortunes of those
species in dealing with the microenvironment of the place where the propagules end up.

If a vehicle drives over the tundra causing the destruction of the plant base, the
rarer species, or even just the slow growing species, the succession in that particular spot
would be reset to a base of ruderal species or in many cases the faster growing clonal
plants in the surrounding undisturbed areas. These baseline communities would be more
similar to each other than would be the undisturbed communities that have diverged from
each other unhindered by the damage caused by the disturbance. Possible results would
be lags in succession and likely changes in the direction of the successional pathways
leading to different communities than would otherwise be present. The magnitude of the

change would be related to the severity of the impact.

73

It is the intent of this study to test some hypotheses about the controlling factors
involved in the recovery of tundra from disturbance, and to generate some others. The
author’s initial thoughts were that the moisture of the soil would have a strong impact on
the resistance to damage and possibly also on the resilience of the system. Several other
environmental factors were also examined to get a better feel for the total picture rather
than to look at one variable, however important it may be. Interestingly enough, the PCA
results suggest that many of these other variables may be more closely related to the
distribution of the vegetation than the variable of primary interest. It is important to
consider that many of the variables examined in this study may be fairly strongly
correlated to each other.

The variable that correlated most highly with the ordination axes, in both the
whole-study ordination, and the ordination using only those plots from which soil
moisture data were collected, was the amount of bare soil present in the plot. Bare soil
was fairly rare in this study and nearly all subplots had a total relative plant cover of
100%. The presence of bare soil is often representative of a continuously disturbed site,
such as a frost boil or a grazing site. This suggests that the vegetation present in a plot
containing bare soil may be similar to that found in a site disturbed by vehicles. Consider
that the bare soil condition was only found in plots located on the dry ridge at Sites 3 and
4. It is possible that the similarities of plant communities found in areas with bare soil are
based on the well-drained soils as much as on the actual presence of bare soil.

Another of the variables most highly correlated with the ordination axes was the
depth of the active layer. As stated earlier, there were significant differences in the active

layer depth between impacted and non-impacted areas. Because the active layer is

74

significantly affected by disturbance and is implicated as a possible controlling factor for
the vegetation, this demonstrates a considerable link between disturbance and vegetation
cover that continues beyond the initial impact. It should be noted, however, that
correlation does not imply causality.

Another variable that is potentially important is the average soil pH of the plot.
The soil pH was collected along with the soil moisture data and as such it is only
available for the plots from which soil moisture data were taken. In the ordination for
these specific plots, soil pH was most highly correlated with the second ordination axis.
This suggests that the pH controls to a degree the vegetation present in a location.
Therefore, knowing the pH of the soil in an area could give an indication of the likely
species to invade in a newly disturbed patch. In turn, knowing the growth rates and
colonization rates of these plants may give an indication of the speed in which an area is
likely to recover.

The soil pH recorded in this study ranged from 5 to slightly above 7, indicating
that the site is primarily weakly acidic. It would have been interesting to also test areas of
alkaline soil. However, this is an uncommon condition in the study area and, therefore,
poses a difficulty in finding comparable vehicle trails on alkaline sites.

Although soil moisture had lower correlations with the ordination axes than other
variables, it still bears consideration. It should be kept in mind that the variables that have
higher correlations typically are themselves correlated with available soil moisture. As
previously mentioned, bare soil is found in predominantly drier areas. Data collected in
this study, as well as the general observation, shows that, near Barrow, Alaska, sites with

high soil moisture content typically have shallower thaw depths than areas of dry soil

75

(Komarkova & Webber 1980). It appears that the factors which directly affect the
vegetation present in a plot are affected themselves by the amount of soil moisture
present in a site.

Looking at the ordinations for the individual sites tells a slightly different story. _
Generally, the ordination axes explain a greater amount of the variation in the data and
the variables correlate better with the axes. This is likely due to the smaller amount of
variation contained within an individual site compared to the whole study.

One of the differences seen in these ordinations is that, at the level of the
individual site, the effect of the amount of bare soil is overshadowed by the other
variables that may have a higher site-specific effect. There was no bare soil present at
either Sites 1 or 2. At Site 3, bare soil had the highest correlation with axis 3 of the PCA
rather than axis 1 as it did in the whole study PCA. At Site 4, other variables had higher
correlations with all three of the primary ordination axes.

At the wetter sites, the first two axes were most correlated with average pH of the
soil and the average amount of standing water present in the plot. The average pH
correlation is no surprise considering that it had the highest correlation with axis 2 in the
whole study moisture ordination. Average amount of surface water did not figure
prominently in the whole study ordinations but in these individual sites, particularly Site
2, it does. Axis I of the PCA for Site I captured nearly 45% of the variation in the data,
which is the highest of any axis in any of the ordinations carried out in this study. The
correlation between this axis and the amount of standing water present is the highest of

any variable with any axis in this study (r2=0.7 19).

76

At the drier sites, active layer depth was more important than at the wet sites. All
three primary axes at Site 4 had their highest correlations with thaw depth. This is also
the case for axis 3 of Site 2 and axis 2 of Site 3.

At Site 3 the highest correlation with axis 1 was with the average
microtopographic height of the plots. The average microtopographic height is best
thought of as an indication of the difference in height of non-impacted areas and the
corresponding impacted areas, since all points were recorded as height above a baseline.
If wheel ruts were present from traffic, the impacted portion of the plots would be lower
on average. The PCA of Site 3 separates the plots distinctly into a group "of impacted
subplots and another of non-impacted subplots indicating that there is different vegetation
present at lower areas than at higher areas. Microtopography is also most closely
correlated with axis 3 at Site 1. In wetter areas such as Site 1, microtopography is often
the cause of differences in the amount of surface water present, which is also quite
important at Site 1. It should be noted that at Site 1 the depressions caused by Weasel
traffic have since become ponds in which species are found that are not present within
other parts of the site.

One interesting point is the lack of a strong correlation of any of the
environmental variables with the second axis in the whole-study PCA of the vegetation
data. The highest correlation with this axis in the ordinations of either the whole-study or
of those plots with moisture data was r2=0.116. This is weak in comparison to the other
axes. None of the environmental variables recorded in this seem to adequately explain
this large portion of the variation in data. In further studies of this kind, nutrient

availability should also be considered. This study focuses on the physical impact of off-

77

road vehicle disturbance, but if there is also a chemical impact, it may go a long way to
explaining the differences observed in this study.

Perhaps the most striking evidence of lingering disturbance from vehicle traffic
comes from the loss of species from impacted subplots that are present in the surrounding
area. There is a possible bias involved in comparing the species richness of the impacted
and non-impacted areas within a plot. A great deal more area was sampled in the non-
impacted areas surrounding the tracks and so it is more likely to encounter rare species in
the non-impacted parts of the plot. In an attempt to remove this bias, a variation of the
species area curve was used. A typical species area curve estimates the number of species
that one would likely encounter in a plot of given size based on the total number of
available species in the area. It is assumed that the species are distributed randomly
throughout the area. If the species were, in fact, randomly distributed throughout the
landscape, one would be just as likely to find rare species on an impacted area as a non-
impacted one. Therefore, a designated area of impacted ground should yield the same
species richness as the same area of un-impacted ground randomly chosen. If, on the
other hand, randomly chosen, un-impacted areas yield more Or fewer species than the
same area of impacted land, it can be concluded that the species are not randomly
distributed and any loss or gain could be attributed to the presence of the disturbance.

A species area curve of this sort was generated using the data from all plots for
which soil moisture data were collected. There were a total of 70 species in this data set,
only 59 of which were found in the impacted areas. If the distribution of species were
random, there should be approximately 64 species present in the impacted area. Fifty-

nine species is more than one'standard deviation below the mean value of 64, which

78

indicates that the species distribution is not random throughout the landscape and that the
loss of species richness in the impacted areas is indeed related to the disturbance.
Negative effects of the vehicle disturbance are, therefore, still present after 25 to nearly
40 years of recovery.

Information from these species area curves is interesting in another light and
supports the hypothesis of this study that the recovery rate varies based upon the moisture
of the sites. The results of the species area curves generated for the individual sites all
show a reduction in species in the impacted areas over the non-impacted. However, at
Site 2, the CRREL test site, the number of species present falls well within one standard
deviation of the mean species richness showing that the species loss is not significant.
This is the wettest of the three sites and is expected to have the highest resilience of all
the sites.

The species loss at Site 4 is greater than one standard deviation from the mean,
indicating a significant loss of species. It is, however, close to being within the acceptable
range. This site combines some of the wettest plots and some of the driest plots. Site 3
has a similar community composition to the drier portions of Site 4. At Site 3, the species
richness is well below the acceptable range about the mean. Because Site 3 has had
approximately 2 years longer to recover from a similarly sever impact, it could be
expected that Site 3’s species richness would be closer to the mean than at Site 4, but this
is not the case. This is probably due to the increased resilience of the wetter plots at Site
4.

Site 1 demonstrates the greatest difference between the species richness of the

impacted subplots and the mean random species richness. This site experienced the most

79

infrequent and lightest disturbances of any of the four, though over a longer duration.
Even considering the longer time span over which the trail was used, the number of
passes over this particular trail is likely to be fewer than was the case at Sites 3 or 4.
Some of the test tracks at Site 2 which had recovered to the point where they were not
able to be located for this study were documented to have had a greater number of passes
than was estimated for Pitelka’s trail. Though possibly having the least impact and the
greatest recovery time of the four sites, Site 1 shows the greatest contrast between the
species richness of it’s impacted subplots and the predicted species richness of a
randomly selected area within the site.

Site 1 had soil moisture levels intermediate between Site 2 on the high end of the
spectrum and Sites 3 and 4 at the low end. It appears that Site 1 suffers from the reduced
resistance commonly found at wet sites, as well as the reduced resilience commonly

found at dry sites. This site is, therefore, easily impacted and slow to recover.

4.2 Importance of Scale

It is important to consider the scale of the disturbance to make predictions about
the severity of the impact over longer time scales. The disturbances considered in this
study are relatively small in size and short in duration of impact. On a typical Weasel trail
considered in this study, none of the impacted points were more than one or two meters
away from non-impacted areas. In rare cases where the disturbance was more severe, the
distance to non-impacted areas could be as much as three or four meters, but would not
likely be more. Considering the clonal nature of many of the plants found in the study

sites, it could be expected that the root systems of many of the plants cross the boundary

80

between impacted and non-impacted terrain and would dampen the effects of the .
impacted condition. Recovery can be expected to progress fairly quickly in this case.

The vehicles used to make the trails examined in this study were, for the most
part, confined to established trails and only used a few times in a season. In cases where
drivers of off-road vehicles don’t adhere to limitations like these, we can expect
disturbance over a broader area. The farther a disturbed area is from the nearest pristine
tundra, the longer it is likely to take for recovery to proceed. Likewise, the more frequent
the disturbance, the higher the degree of dissimilarity to the undisturbed tundra and the
longer recovery can be expected to take.

In the examination of species diversity in an area, the scale of the study plays
several roles. This particular study measured the diversity at several points to generate an
estimate of the stand diversity for thestudy sites in question. It is necessary to consider
this in a study of this sort since diversity at these two scales is not likely to respond
similarly to a disturbance. For example, Chaneton and Facelli (1991) found that in
response to flooding in Argentinean grasslands, the patch diversity behaved differently
from stand diversity. Their study also examined the effects of grazing in combination
with flooding and they found that the relationship between the effect of flooding on patch
and stand diversities was not consistent or predictable between the grazing treatments.
Therefore, it is necessary to consider the scale at which a disturbance is affecting an area
in order to make the appropriate measurements to detect it. This is directly applicable to
studies like this where the disturbed area is influenced by vehicle disturbance, grazing by

animals and other tundra-specific disturbances such as thermokarst processes.

81

In this case, if we treat an entire site as a stand and the disturbance as a patch then
we are at a loss to calculate changes in diversity without a pre-disturbance measurement
of stand diversity, though we can compare the disturbed patch to an undisturbed patch
and examine the patch diversity. When considering a disturbance and its undisturbed
counterpart as distinct stands from each other and individual subplots as patches, it is
possible to detect a reduction in diversity at this smaller scale due to disturbance and
relate it back to the whole diversity of a vehicle track or the surrounding undisturbed

land.

4.3 Relative Damage Of Each Vehicle

Of the three vehicles examined in the initial CRREL study, the M-29 Weasel is
the one which has made the majority of the tracks identified and examined in this study.
This is a good reflection of the trend in off-road vehicles from 1955 to 1975. This
particular vehicle no longer sees use on the tundra. There are a few similar tracked
vehicles still present in Barrow, but these are rarely, if ever, used. Nearly all large, off-
road vehicle traffic is conducted using Rolligons.

The Rolligon is a large, low-pressure tire vehicle that was designed for use on
fragile surfaces and has since replaced the M-29 Weasel as the primary mode of large
off-road vehicle traffic. What is perhaps ironic is that the Rolligon is not as gentle on the
landscape as was the Weasel. The initial result of the Second CRREL study run at Site 2
by Abele and Atwood (1976) indicated that in wet meadow tundra, 15 passes of a

Rolligon was as devastating, if not more so, than 50 passes of a Weasel. There are a

82

number of Rolligons of various sizes and types currently in use in Barrow and at the
Prudhoe Bay oil field.

The vehicle that had the best test results was the SK-S hovercraft. After 4 years of
recovery, the track signature was nearly invisible to the eye and Abele and Atwood
(1976) had reported an almost complete recovery. Although the SK-5 performed very
well in the tests, it was deemed unusable for off-road transport. Hovercrafts are very

difficult to maneuver over tundra and are not practical for carrying large loads.

4.4 Signature Spread

An interesting phenomenon in the recovery of terrain from disturbance is the
effect of the disturbance on adjacent areas that were not impacted. These adjacent areas
were unafi‘ected‘by the direct disturbance, but secondary effects spreading from the
impacted areas to their non-impacted neighbors begin after the initial disturbance has
ended. Changes in drainage or heat conductivity of adjacent soil will have lingering
effects on non-impacted terrain as long as the signature of the initial impact remains. In
this way, the disturbance is likely to extend to areas not initially impacted and possibly
even degrade further before recovery takes place.

The species richness data gathered from subplots at set distances from the impact
demonstrates evidence for this. Though not directly impacted by the wheel or track of a
vehicle, the areas caught between the rats had nearly as low species richness as the
impacted areas. Likewise, the areas immediately outside of the ruts have low species
richness compared to subplots that are a greater distance away. This suggests degradation

in areas that were not physically impacted by the initial disturbance.

83

Again it is important to consider the scale of exarrrination for comparisons of this
kind. In this particular case, the subplots were a set width of 0.5 m. This scale is fine
enough to register the differences in richness as distance increases and seems to be
appropriate. It also makes sense because the wheel ruts themselves are generally on a
similar scale ranging from 0.3 — 2.0 m. If the subplot size were even 1m wide instead of
0.5m, it is likely that any differences would be masked by including areas that are beyond
the signature spread with areas within it.

Signature spread can be a problem in studies like this one that seek to locate
disturbances left by others in the past, which were, in most cases, not well documented
since it was not the focus of the activity that caused them. It becomes more difficult to
distinguish the actual boundaries of the initial impact from the signature spread and
leaves it to the judgment of the investigator to determine what should be included in the
study. The ideal way to conduct a study of this nature would be to set up the disturbance
and immediately mark off any boundaries in a fairly permanent manner for study in the

long term.

4.5 Functional Recovery and Continuous Succession

The effects of many vehicle disturbances are still present on the tundra, including,
but certainly not limited to, those which are the subject of this study. However, the results
are positive in some aspects. The vegetation has returned to areas from which it had been
totally removed as a result of vehicle disturbance. In cases such as Pitelka’s trail, Site 1,
the vegetation that has colonized the impacted terrain is different from the initial

community present. This appears to be a case where a new equilibrium has been formed

84

with the soil environment so that it is unlikely that the impacted area will ever return to
initial conditions. This can be considered a functional recovery since a different, but
healthy, community is in place as a result of the disturbance. Peterson felt during his
study in 1975 that this particular area would never be more than a patch of bare ground
and was impressed at the amount of recovery at the time of this study (Peterson, K.M.
personal communication).

It should also be considered that because succession is always ongoing, it is likely
that the communities found on formerly disturbed areas will never match identically with
their undisturbed counterparts. Succession may have advanced the recovery of the
community to the point where it mirrors its pre-disturbance condition, but the
neighboring undisturbed communities have not been static during those years of recovery
and might therefore have diverged. The disturbed community will either lag behind the .

undisturbed community or change successional pathways.

4.6 Summary And Conclusions

The initial general hypothesis of this study is that site moisture plays a strong role
in the resistance and resilience of the tundra vegetation to damage by vehicles. The
prevailing literature supports the notion that wet sites recover quickly due to a high
resilience and dry sites recover slowly but have high resistance to impact. This study
conforms to the general hypothesis, however, a finding of this study, for this region and
for these types of disturbances, is that landscapes with intermediate moisture levels,
appear to show the most residual effect between impacted and non-impacted areas after

30 years. This may be a result of the combined effects of having a lower resistance than

85

dry sites and lower resilience than wet sites. It is impossible to speak directly about
recovery rate since there is no real quantitative data to describe the initial conditions after
impact. We are able to determine similarity to the undisturbed condition, but the degree
of similarity is a combination of resistance and resilience and we can not separate these
two factors without knowledge of the initial impact. However, the photographic and
anecdotal evidence suggests enough about the initial impacts that we can make inferences
about them even if they can not be directly measured.

The vegetation ordinations indicate that it is not generally the soil moisture that
dictates the variation in the vegetation. Generally, thaw depth, the amount of bare soil,
and soil pH correlate more closely with the ordination axes indicating that they may be
controlling factors in the variation in vegetation in the areas. However, these effects are
often related to the soil moisture and standing water at a site, and are in keeping with the
observations of Webber (1978) that the direct controls of vegetation at Barrow are site
relief and water drainage.

The strongest evidence for moisture controlling the recovery of vegetation at a
site is demonstrated by the losses of species in impacted subplots. At the wettest site,
there is not a significant loss of species in the impacted subplots. At the driest sites there
are fewer species in the impacted subplots than in the non-impacted subplots, but the
differences are not as pronounced as they are at the intermediate moisture site where less
than half of the species are present Within impacted subplots.

Additionally, there is evidence that the disturbance has spread beyond the initial
impact of the vehicle traffic. The species richness increases as the distance from the

initial impact increases demonstrating that the areas adjacent to, but not immediately

86

impacted by the vehicles have suffered a subsequent decline in species richness. The
landscape has suffered further degradation prior to any recovery and the lingering effects
of the impact are present after 30 years even on terrain that was not initially impacted.
This secondary expansion of species loss has not been previously recognized.

Residual effects of off-road vehicle traffic are still evident at the study sites after
30 years. Differences in microtopography are still present allowing for identification of
the tracks. Active layer depth and vegetation composition of impacted areas still differ
from their non-impacted counterparts. Nevertheless, considerable plant community
function has been restored. The impacted areas have been re-colonized by the native
plants and almost no bare soil remains where the vegetation had been completely
removed by the disturbance. At a glance, the plant communities within the immediately
impacted subplots resemble those that were not impacted.

Most ongoing disturbances will likely require more than 30 years to recover
depending on the type of terrain. It is certain that some effects of disturbance will never
wholly be eliminated. Future off-road vehicle traffic should be kept to a minimum to
protect past disturbances from further damage and aid in their functional recovery, as
well as to prevent the development of new disturbances. Regulations are already in place
in many areas for the protection of the systems, but more attention should be paid to their
enforcement. It is important that the public considers the ramifications of continued off-
road vehicle use and avoids it when possible.

In cases where off-road vehicle use is unavoidable, the extent and duration of the
traffic should be considered before locations for trails are chosen and permits granted.

Trails which will be used only a very few times should be made over dry ground where

87

resistance is high enough to withstand the disturbance. Trails to be used repeatedly would
be best created over wet areas unless the terrain will not support the weight of the
vehicle. The impact will be more severe, but the increased resilience of the area will
allow for quicker recovery. Areas of intermediate moisture should be avoided whenever

possible as they are easily impacted and very slow to recover.

88

APPENDIX A

SPECIES LISTS

Table A-1: Species list for Site 1 including the groupings used for analyses of vegetation
data in cases where the species could not be identified with certainty.

 

 

Species Name Type

Unknown Algae Algae
mum WMch.) Kamefelt Lichen
and: my Lichen
cats/b “m(L.) Ach. Lichen
cure/ImmSandst.) Hustich Lichen
aromasp Lichen
Wammichardson) Nyl. Lichen
mwhlfiAch.) Rabenh. Lichen
mm rift/dranflHook.) Kamefelt Lichen
Power/asp Lichen
WM Lichen
Warfare-WWW) Vainio Lichen
WWEhm.) Culb. Lichen
Unidentified leafy liverwort Liverwort
WWMahlenb.) Schwaegr. Moss
Gila/pump Moss
WWM(Hedw.) C. Jens. Moss
mmmmmndb.) Wamst. Moss
[Msp Moss
WWHedw. Moss
Wimp ' Moss
Unidentified bryophyte Moss
WIMS"). Vascular
Ammo/wanna) Anderss. Vascular
awoqlnde'Wahlenb. Vascular
MM.BT. Vascular
WeWHonck. Vascular
WWW-tome Vascular
WWL Vascular
Mlm Blytt Vascular
mmumw. Vascular
Within/Durand Vascular
Petal-2.9mm) Franch. Vascular
Pm ardmRBr. Vascular
Mom/”Matte Vascular
WWbL. Vascular
Sch/Wham. Vascular

 

89

Table A-1: Continued.

 

 

Species Name Type
Saxifraga cemua L. Vascular
Saxifraga hirculus L. Vascular
Stellaria Iaeta Richards. Vascular

 

Table A-2: Species list for Site 2 including the groupings used for analyses of vegetation
data in cases where the species could not be identified with certainty.

 

 

Species Name Type
mam MmMch.) Ka'mefelt Lichen
Wsp Lichen
Unidentified leafy liverwort Liverwort
WWMahlenb.) Schwaegr. Moss
Com» Moss
mmmmMMndb.) Wamst. Moss
”nimsp Moss
PWWHedw. Moss
Mmsp Moss

_ ”WM/H.311 Vascular
WeerHonck. Vascular
WWW-tome Vascular
ngL. Vascular

 

Table A-3: Species list for Site 3 including the groupings used for analyses of vegetation
data in cases where the species could not be identified with certainty.

 

 

Species Name Type

Unidentified Algae Algae

mum WMch.) Kamefelt Lichen
Can/b may Lichen
can/re enema.) Ach. Lichen
adv/7148(Sandst.) Hustich Lichen
Wsp Lichen
Unidentified Crustose lichen Lichen
Wadbfiichardson) Nyl. Lichen
[1mm ammyl.) Raes. , Lichen
lab/bMMCh.) Rabenh. Lichen
mm [bar/wanflHook.) Kémefelt Lichen
WW(Sw.) Lynge Lichen
Wsp Lichen
WW Lichen
WWWahl) S. Gray Lichen
Mama.) Ach. Lichen

 

90

Table A-3: Continued.

 

Species Name Type

WWWUdson) Vainio Lichen
Wuhad aanaurer ex Funck Lichen

 

 

WMEhm.) Culb. Lichen
Unidentified lichen Lichen
WWMahlenb.) Schwaegr. Moss
Gab/gmsp Moss
amammww.) C. Jens. Moss
Dbl/u» WMSchleich. ex Schwaegr. Moss
MMMMMndb.) Wamst. Moss
llaimsp Moss
mum-(Hem) Lindb. Moss
WWHedw. Moss
Whfimfiedw.) Brid. Moss
Wsp Moss
meMMedw.) Loeske Moss
”WW Moss
Wilkins-Sm. Vascular
MMbMMBr.) Griseb. Vascular
ammmmt. Vascular
ammeahlenb. Vascular
Camel/Ir MbL. Vascular
WMBI’. Vascular
WJWHonck. Vascular
WWW-lows Vascular
WMBlytt Vascular
mmumw. Vascular
Within/Durand Vascular
MW“) French. Vascular
Pat/war R.Br. Vascular
Part/mm Malte ' Vascular
filmmbL. Vascular
fiammewwmlenb. Vascular
SJRW Anderss. Vascular
Self/Wham. Vascular
WWL. Vascular
WWRBL Vascular
WgML. Vascular
Wmn L. Vascqu
Shh/b bob Richards. Vascular
Unknown vascular plant Vascular

 

91

 

Table A-4: Species list for Site 4 including the groupings used for analyses of vegetation
data in cases where the species could not be identified with certainty.

 

 

Species Name Type

Unknown Algae Algae
mmmmm.) Kemefelt Lichen
Con/b my Lichen
Con/b enema.) Ach. Lichen
chm Lima.) Willd. Lichen
Msp Lichen
WJW(Rmhamm) Nyl. Lichen
”Mm ammmyl.) Raes. Lichen
tab/biflNAch.) Rabenh. Lichen
my: run/mmflHook.) Kamefelt Lichen
WW(SW.) Lynge Lichen
hmbsp Lichen
Wsp Lichen
WW Lichen
WWWahl) S. Gray Lichen
Mama.) Ach. Lichen
WMMudson) Vainio Lichen
“vanguard aflmLaurer ex Funck Lichen
WWEhm.) Culb. Lichen
Unidentified lichen Lichen
Unidentified leafy liverwort Liverwort
WWMahlenb.) Schwaegr. Moss
wmwartm.) Kindb. Moss
WW(Schwaegr.) BSG Moss
Gamsp Moss
“mama-19m.) C. Jens. Moss
Dara/m WMSchleich. ex Schwaegr. Moss
mmmmmmndb.) Wamst. Moss
”Wasp Moss
Mummedw.) Lindb. Moss
WWHedw. Moss
fiamtimbmmmedw.) Brid. Moss
Spinmsp Moss
anbmWedw.) Loeske Moss
Unidentified bryophyte Moss
Wm mnmbtflSchlmp. in 8.8.6.) Loeske Moss
WopimSm. Vascular
MMbflhmer.) Griseb. Vascular
Arm mam.) Anderss. Vascular
ammmmt. Vascular
“magawWahlenb. Vascular
omen/e MbL. Vascular
mam Vascular
WW.BL Vascular

 

92

Table A-4: Continued.

 

 

Species Name Type
Wow Honck. Vascular
W» Mam-lame Vascular
WWL. Vascular
new ”we: Blytt Vascular
m WLindeb. Vascular
0m ma.) Hill Vascular
PaprwrWnaben Vascular
ABM/b bin/Durand Vascular
hbmW(L) French. Vascular
Par am R.Br. Vascular
Poe/Mir 0mm Malte Vascular
flames niabL. Vascular
flammmwahlenb. Vascular
.‘hkowihfiTrautv. Vascular
Slkp‘xbm Anderss. Vascular
fikpuh‘anham. Vascular
San/WTrautv. Vascular
WWL. Vascular
My: W R.Br. Vascular
My MMWaldst. & Kit. Vascular
My! ML. Vascular
5:45}:be L. Vascular
Momma») Fedtsch. Vascular
Shh/immRottb. Vascular
sub/re bob Richards. Vascular

 

93

LITERATURE CITED

Abele G, Atwood DM. 1976. Effects of SK-S Air Cushion Vehicle Operations on
Organic Terrains After Four Years. Hanover, New Hampshire: U.S. Army Corps
of Engineers - Cold Regions Research and Engineering Laboratory; Report nr IR
494. 29 p.

Abele G, Atwood DM, Gould ID. 1974. Effects of SK-5 Air Cushion Vehicle Operations
on Organic Terrains After Two and Three Years. Hanover, New Hampshire: U.S.
Army Corps of Engineers - Cold Regions Research and Engineering Laboratory;
Report nr IR 425. 77 p.

 

 

Abele G, Brown J. 1977. Arctic transportation: Operational and environmental evaluation
of an air cushion vehicle in Northern Alaska. Journal of Pressure Vessel
Technology 99: 176-82.

Abele G, Brown J, Brewer MC, Atwood DM. 1977. Effects of Low Ground Pressure
Vehicle Traffic on Tundra at Lonely, Alaska Hanover, New Hampshire: U.S.
Army Corps of Engineers - Cold Regions Research and Engineering Laboratory;
Report nr SR 77-31. 32 p.

 

Abele G, Walker DA, Brown J, Brewer MC, Atwood DM. 1978. Effects of Low Ground
Pressure Vehicle Traffic on Tundra at Lonely, Alaska Hanover, New Hampshire:
U.S. Army Corps of Engineers - Cold Regions Research and Engineering
Laboratory; Report nr SR 78-16. 63 p.

Adam KM, Hernandez H. 1977. Snow and rce roads: Ability to support traffic and effects
on vegetation. Arctic 30:13-27.

Auerbach NA, Walker MD, Walker DA. 1997. Effects of roadside disturbance on
substrate and vegetation properties in Arctic tundra. Ecological Applications
7(1):218-35.

Bliss DC. 1983. Modern human impact in the Arctic. InzHolzner W, Werger MJA,
Ikusima I, editors. Man's Impact on Vegetation. The Hague/Boston/London: Dr.
W. Junk Publishers; p 213-225.

Brown J, Everett KR, Webber PJ, MacLean SF, Murray DF. 1980. The coastal tundra at
Barrow. InzBrown J, Miller PC, Tieszen LL, Bunnell FL, editors. An Arctic
Ecosystem: The Coastal Tundra at Barrow, Alaska. Stroudsburg, PA: Dowden,
Hutchinson, & Ross inc.; p 1-29.

94

Brown J, Johnson PL. 1965. Pedo-ecological Investigations - Barrow, Alaska Hanover,
New Hampshire: U.S. Army Material Command - Cold Regions Research and
Engineering Laboratory; Report nr TR 159. 32 p.

Chaneton EJ, Facelli J M. 1991. Disturbance effects on plant community diversity: spatial
scales and dominance hierarchies. Vegetatio 93: 143-55.

Crawley MJ, Harral J E. 2001. Scale dependence in plant biodiversity. Science 291 :864-8.

Ebersole JJ. 1985. Vegetation Disturbance and Recovery, Oumalik, Alaska [dissertation].
University of Colorado. 408 p.

Forbes BC. 1992. Tundra disturbance studies, 1: Long-term effects of vehicles on species
richness and biomass. Environmental Conservation 19(1):48-58.

Forbes BC, Ebersole JJ, Strandberg B. 2001. Anthropogenic disturbance and patch
dynamics in Circumpolar Arctic ecosystems. Conservation Biology 15(4):954-69.

Forbes BC, Sumina OI. 1999. Comparative ordination of Low Arctic vegetation
recovering from disturbance: Recording two contrasting approaches for field data
collection. Arctic, Antarctic, and Alpine Research 31(4):389-99.

Gersper PL, Challinor J L. 1975. Vehicle perturbation effects upon a tundra soil-plant
system: I. Effects on morphological and physical environmental properties of the
soils. Soil Science Society of America Proceedings 39:737-44.

Grulke NE, Bliss LC. 1983. A note on winter seed rain in the High Arctic. Arctic and
Alpine Research 15(2):261-5.

Haag RW. 1973. Energy budget changes following surface disturbance to two northern
vegetation types Environmental-Social Committee Northern Pipelines, Task Force
on Northern Oil Development; Report nr 73-43. 26 p.

Hernandez H. 1973. Natural plant recolonization of surficial disturbances, Tuktoyaktuk
Peninsula Region, N.W.T. Canadian Journal of Botany 51:2177-96

Kevan PG, Forbes BC, Kevan SM, Behan-Pelletier V. 1995. Vehicle tracks on high
Arctic tundra: their effects on the soil, vegetation, and soil arthropods. Journal of
Applied Ecology 32:655-67.

Khitun O. 1997. Self-recovery After Technogenic and Natural Disturbances in the
Central Part of the Yamal Peninsula (Western Siberian Arctic). In:Crawford

95

RMM, editor. Disturbance and Recovery in Arctic Lands. Netherlands: Kluwer
Academic Press; p 531-562.

Kilbum PD. 1966. Analysis of the species-area relation. Ecology 47(5):831-43.

Klein DR. 1970. The impact of oil development in Alaska (a photo essay). InzFuller WA,
Kevan PG, editors. Productivity and Conservation in Northern Circumpolar
Lands. Morges, Switzerland: International Union for Conservation of Nature and
Natural Resources; p 209-242.

Komarkova V, Webber PJ. 1980. Two low Arctic vegetation maps near Atkasook,
Alaska. Arctic and Alpine Research 12:447-472.

Leadley PW, Li H, Ostendorf B, Reynolds J F. 1996. Road-related disturbances in an
Arctic watershed: Analysis by a spatially explicit model of vegetation and
ecosystem processes. Ianeynolds J F , Tenhunen JD, editors. Landscape Function
and Disturbance in Arctic Tundra. Berlin Heidelberg: Springer-Verlag; p 387-
415.

McCune B, Mefford MJ. 1995. PC-ORD. Multivariate Analysis of Ecological Data,
version 2.0. Gleneden Beach, Oregon, USA: MjM Software Design. 126 p.

Milchunas DG, Schultz KA, Shaw RB. 2000. Plant community structure in relation to
long-term disturbance by mechanized military maneuvers in a semiarid region.
Environmental Management 25(5):525-39.

Noble IR. 1996. Linking the human dimension to landscape dynamics. In:Walker B,
Steffen W, editors. Global Change and Terrestrial Ecosystems. Cambridge:
Cambridge University Press; p 173-183.

Paine RT, Tegner MJ. Johnson EA. 1998. Compound perturbations yield ecological
surprises. Ecosystems 1:535-45.

Pearson SM. 1994. Landscape-level processes and wetland-conservation in the southern
Appalachian Mountains. Water, Air and Soil Pollution 77:321-32.

Peterson KM. 1978. Vegetational Successions and Other Ecosystemic Changes in Two
Arctic Tundras [dissertation]. Duke University. 305 p.

Racine CH, Ahlstrand GM. 1991. Thaw response of tussock-shrub tundra to experimental
all-terrain vehicle disturbances in South-Central Alaska. Arctic 44(1):31-7.

96

 

Rickard WE, Brown J. 1974. Effects of vehicles on Arctic Tundra. Environmental
Conservation 1(1):55-62.

Romme WH, Everham EH, Frelich LE, Moritz MA, Sparks RE. 1998. Are large,
infrequent disturbances qualitatively different from small fiequent disturbances?
Ecosystems 1:524-34.

Seybold CA, Herrick JE, Brejda JJ. 1999. Soil resilience: A fundamental component of
soil quality. Soil Science 164(4):224-34.

Slaughter CW, Racine CH, Walker DA, Johnson LA, Abele G. 1990. Use of off-road
vehicles and mitigation of effects in Alaska permafrost environments: A review.
Environmental Management l4(1):63-72.

Van Cleve K. 1977. Recovery of Disturbed Tundra and Taiga Surfaces in Alaska.
In:Caimes JJr, Dickson KL, Herricks EE, editors. Recovery and Restoration of
Damaged Ecosystems. Charlotsville: University Press of Virginia; p 422-455.

Walker DA. 1996. Disturbance and Recovery of Arctic Alaskan Vegetation. Ianeynolds
J F, Tenhunen JD, editors. Ecological Studies vol. 120. Berlin: Springer-Verlag; p
35-71.

Walker DA, Everett KR. 1987. Road dust and its environmental impact on Alaskan taiga
and tundra. Arctic and Alpine Research 19(4):479-89.

Walker DA, Webber PJ, Everett KR, Brown J. 1977. The Effects of Low-Pressure
Wheeled Vehicles on Plant Communities and Soils at Prudhoe Bay, Alaska
Hanover, New Hampshire: US Army Corps of Engineers Cold Regions Research
and Engineering Laboratory; Report nr SR 77-17. 49 p.

Wardle DA, Bonner KI, Barker GM. 2000. Stability of ecosystem properties in response
to above-ground functional group richness and composition. Oikos 89: 1 1-23.

Webber, PJ. 1978. Spatial and Temporal Variation of the Vegetation and Its Production,
Barrow, Alaska. InzTieszen, LL, editor. Vegetation and Production Ecology of an
Alaskan Arctic Tundra. New York: Springer-Verlag; p37-112.

97

 

IIIIIIIIIIIIIIIIIIIIIIIIIII

|||||||||||||||||||||||||||||||||||||||||||||||||