ENVIRONMENTAL LIMITATIONS OF TWO RARE ATLANTIC

COASTAL PLAIN FLORA SPECIES AND THE IMPACT OF

HYDROLOGICAL ALTERATIONS

by

Jennifer Mary-Jane Lusk

Thesis

submitted in partial fulfillment of the requirements for

the Degree of Master of Science (Biology)

Acadia University

Spring Convocation 2006

© by Jennifer Mary-Jane Lusk, 2006 The examining committee for the thesis was:

Dr. John Roff, Chair

Dr. Sherman Boates, External Reader

Dr. Rodger Evans, Internal Reader

Dr. Ed Reekie, Supervisor

Dr. Phil Taylor, Department Head (delegate)

This thesis is accepted in its present form by the Division of Research and Graduate

Studies as satisfying the thesis requirements for the degree Master of Science (Biology).

ii Table of Contents

List of Tables ...... vi

List of Figures...... viii

Abstract...... xiv

Acknowledgements ...... xv

Chapter 1 - General introduction...... 1

1.0 Background ...... 1

1.1 The Atlantic Coastal Plain flora (ACPF)...... 2

1.2 Rarity ...... 6

1.3 Conservation initiatives...... 7

1.4 Natural water fluctuations and watershed alterations...... 8

2.0 Purpose of the study...... 11

3.0 Study species...... 12

4.0 Study sites ...... 13

Chapter 2 - The effect of growing season length...... 24

1.0 Introduction...... 24

2.0 Methods...... 27

2.1 Transplant preparation...... 27

2.2 Transplanting ...... 28

2.3 Measurements ...... 30

2.4 Data analysis...... 31

3.0 Results ...... 32

3.1 Effect of week planted ...... 32

iii 3.2 Effect of depth ...... 34

3.3 Effect of lake...... 36

4.0 Discussion...... 38

4.1 Effect of week planted ...... 38

4.2 Effect of depth ...... 42

4.3 Effect of lake...... 44

5.0 Conclusion and applications ...... 48

Chapter 3 - The effect of environmental variables ...... 73

1.0 Introduction...... 73

2.0 Methods...... 74

2.1 Transplants...... 74

2.2 Plant and environmental variable measurement...... 75

2.3 Data analysis...... 81

3.0 Results ...... 82

3.1 Environmental variables ...... 82

3.2 Effect of environmental variables...... 85

4.0 Discussion...... 91

5.0 Conclusion and applications ...... 97

General Conclusion...... 113

Appendix 1. Aspect of transplant transects...... 116

Appendix 2. The Unified Soil Classification System field classification technique for coarse-grained soils (USCS P13b version 2) for hand texturing soil in the field...... 117

iv Appendix 3. The loading pattern of PCA factors that summarize the environmental variables measured for each H. umbellata transplant; values closer to one indicate higher correlation of that environmental parameter with the factor...... 123

Appendix 4. The loading pattern of PCA factors that summarize the environmental variables measured for each C. rosea transplant; values closer to one indicate higher correlation of that environmental parameter with the factor...... 124

References...... 125

v List of Tables

Table 1. Results of ANOVA analyses for H. umbellata on each measurement day;

values are sums of squares for each term, significant terms are highlighted. In the day

of year row, the first four numbers of the number are the year, and the last three digits

are the Julian day of the year out of 365. Analyses from day 2004261 on only include

Raynard's and Wilson's Lake...... 51

Table 2. Results of ANOVA analyses for C. rosea on each measurement day; the

values are sums of squares for each term, significant terms are highlighted. In the day

of year row, the first four numbers of the number are the year, and the last three digits

are the Julian day of the year out of 365. Analyses from day 2004261 on only include

Raynard's and Wilson's Lake...... 52

Table 3. Periods for survival analysis...... 53

Table 4. Average precipitation (1971-2000) and actual precipitation in the summers

of 2004-5, and during the fall of 2004 for Yarmouth, Nova Scotia (20-30 km

southwest of the study lakes) (from Environment Canada http://www.climate.

weatheroffice. ec.gc.ca/climateData)...... 100

Table 5. Significant factors in stepwise regression of H. umbellata early growth (Ln

(# of leaves * maximum diameter)), measured July 10-13th, 2005...... 101

Table 6. Significant factors in stepwise regression of H. umbellata size at harvest

(Ln (# of leaves * maximum diameter), measured August 9-24th, 2005...... 101

Table 7. Significant factors in stepwise regression of H. umbellata final biomass...... 102

Table 8. Significant factors in stepwise regression of C. rosea early growth (Ln (# of

leaves * maximum height)), measured July 10-13th, 2005...... 102

vi Table 9. Significant factors in stepwise regression of C. rosea size at harvest (Ln (#

of leaves * maximum height), measured August 9-24th, 2005...... 103

Table 10. Significant factors in stepwise regression of C. rosea final biomass...... 103

Table 11. Significant factors in stepwise regression of number of flowers for C. rosea from 2005...... 104

vii List of Figures

Figure 1. The Atlantic Coastal Plain physiographic region, indicated in yellow, east and south of the black line, extending from Texas to Cape Cod. The main areas of disjunct populations are shown as circles; the asterix indicates a smaller disjunct of only a few species in Cape Breton, N.S. (adapted from http://tapestry.usgs.govna- info.html)...... 15

Figure 2. Changes in sea level in relation to a profile of Nantucket Island,

Massachusetts, to Cape Sable Island, Nova Scotia (adapted by Keddy and Wisheu

1989 from Odgen and Harvey 1975)...... 16

Figure 3. Factors affecting the distribution of Coastal Plain habitats along lakeshores

(adapted by Sweeney and Ogilvie 1993 from Keddy and Wisheu 1989)...... 17

Figure 4. The effects of eliminating water level fluctuations on a shoreline: the four vegetation zones in the top figure are converted to a simple two zone wetland in the bottom figure eliminating the marsh and wet meadow communities (Keddy 1991)...... 18

Figure 5. Images of forestry related dams in the early 1900's: photo on left is of a log drive on the Tusket River system, and the photo on the right is a sawmill and sawmill dam in operation on the Tusket River (photo courtesy of the Tusket Court House, from Eaton and Boates 2002)...... 19

Figure 6. Images of hydroelectric dams on the Tusket River at present: photo on left is the main dam at Tusket Falls in 2001, and the photo on the right is the Tusket Falls hydroelectric power station and dam in 2001, first built in 1929 by the Nova Scotia

Power Corporation (photo courtesy of Samara Eaton, from Eaton and Boates 2002)...... 19

viii Figure 7. Present and past locations of dams on the Tusket River system: the map on

left shows present day hydro dams, saw mill dams and unknown pre 1918 and older

dams, the map on the right names all hydro dams currently on the river system (Eaton

and Boates 2002)...... 20

Figure 8. Changes to the Tusket River system due to the installation of the Tusket

Falls and Carleton dams: the map on the right is a portion of the 1864 A. F. Church

map showing the Lake Vaughan area, and the map on the right is from the Nova

Scotia Department of Natural Resources 1989 GIS map of the Lake Vaughan area

(Eaton and Boates 2002). Note names changes: Bell Lake is now King's Lake,

Coven's Lake is now Bennet's Lake, and St. John Lake is now Wilson's Lake...... 21

Figure 9. Images and distributions maps of the study species; C. rosea locations are

indicated with black dots in the range map on the left, and H. umbellata range is south

of the black line and within the circles on the range map on the right. (Range maps

adapted from COSEWIC reports: Keddy and Keddy 1983, Wilson 1984, Newell

1998, Newell 1999.)...... 22

Figure 10. Tusket River Valley, Tusket Hydro-Station and study lakes in relation to

the province and to Yarmouth (adapted from http://www.mar.dfo-

mpo.gc.ca/masaro/english/ Species_Info/Atlantic_Whitefish)...... 23

Figure 11. Lifting and removing a portion of H. umbellata from the growth tray...... 54

Figure 12. Hydocotyle umbellata in sieve showing root and rhizomes exposed; soil

has been washed out into the base sieve...... 54

Figure 13. Standardized cuttings of C. rosea ready to be treated then planted...... 55

ix Figure 14. Treating the above ground portion of C. rosea transplants with 1:50

dilution of Safer's insecticidal soap...... 55

Figure 15. Measurement of vertical depths below the high water mark with a string

level to establish transplant locations...... 56

Figure 16. The flags in the image show the position of 30, 40, 50, 60, 70, 80, and 85 cm vertically below the high water mark during a very dry period...... 56

Figure 17. Arrangement of transplants: the first set of transplants are indicated by

the two outside rows of circles (yellow), the second round of transplants 4 weeks later

are indicated by the next set of circles toward the middle (blue), and last transplants

are indicated by the two innermost rows of circles (red). There are two circles at 50,

60 and 70 cm to show the overlap of C. rosea and H. umbellata planting elevations...... 57

Figure 18. Transplants and tags: a. Coreopsis rosea, b. Hydrocotyle umbellata...... 58

Figure 19. Transplanting specimens under water...... 58

Figure 20. Effect of week planted on growth measures (means +/- one standard error) for C. rosea and H. umbellata over time by lake. The growth index for C. rosea was Ln (# of leaves) and for H. umbellata it was Ln (# of leaves * maximum leaf diameter). The first four digits of the day of the year on the x-axis are the year and the last three are the Julian day of the year out of 365. The break in the x-axis indicates the winter of 2004/5...... 59

Figure 21. Total, above ground and below ground biomass (means +/- one standard error) of C. rosea and H. umbellata at the end of the two year experiment, by week planted and lake...... 61

x Figure 22. Effect of week planted on above/below ground biomass ratio of H. umbellata (means +/- one standard error)...... 62

Figure 23. Percent of transplants alive at the beginning of a period that survived to the end of that period (dates included in each period are shown in Table 3) by species, week planted and lake...... 63

Figure 24. Effect of depth on H. umbellata growth index (Ln (# leaves * maximum leaf diameter)) on different sampling dates, including interactions with lake or week when significant (means +/- one standard error). Graphs and analyses from day

2004261 on do not include Kempt-Back Lake because of small sample size...... 64

Figure 25. Effect of depth on C. rosea growth index (Ln (# leaves)) over time, including interactions with lake when significant (means +/- one standard error).

Graphs and analyses from day 2004261 on do not include Kempt-Back Lake because of small sample size...... 65

Figure 26. Effect of planting depth on total, above and below ground biomass

(means +/- one standard error)...... 66

Figure 27. Effect of planting depth on above to below ground biomass ratio for H. umbellata (means +/- one standard error)...... 67

Figure 28. Percent of transplants alive at the beginning of a period surviving to the end of that period (dates included in each period are shown in Table 3) by species, depth of planting and lake...... 68

Figure 29. Effect of depth on the average number of flowers for C. rosea by lake

(means +/- one standard error) in 2005...... 69

xi Figure 30. Water levels relative to the high water mark. The first four digits of the day of the year on the x-axis are the year, and the last three are the Julian day of the year out of 365. The break in the x-axis indicates the winter of 2004/5...... 70

Figure 31. Effect of lake on the above to below ground biomass ratio of H. umbellata (means +/- one standard error)...... 71

Figure 32. Effect of lake on the leaf to stem biomass ratio of C. rosea (means +/- one standard error)...... 72

Figure 33. Environmental variable experiment transplants are marked with flagging tape on the far side of the canoe, further in the background are the growing season experiment transplants and environmental sensors on Kempt-Back Lake. The white boxes house the data loggers which received information from the soil temperature and moisture probes...... 105

Figure 34. Soil moisture at all lakes at various depths below the high water mark in the summer, fall and early winter of 2004...... 106

Figure 35. Soil temperatures at Raynard's and Wilson's Lakes in November 2004...... 107

Figure 36. Soil and air temperatures at Raynard's Lake in the fall and winter of

2004/5...... 108

Figure 37. Actual (the top continuous blue line), extreme and normal temperature minimums for November 2004 to January 2005 for Yarmouth, Nova Scotia (20-30 km southwest of the study lakes) (from Environment Canada http://www.climate.weatheroffice.ec. gc.ca/climateData)...... 109

Figure 38. Chlorophyll a concentration at the three experimental lakes (mean +/- one standard error) in the summer of 2005...... 110

xii Figure 39. Depth of light penetration (mean +/- one standard error) in the summer of

2005...... 110

Figure 40. Apparent water colour at the three experimental lakes (0 is the lightest,

100 is the darkest) in the summer of 2005...... 111

Figure 41. Water pH at the three experimental lakes in the summer of 2005...... 112

Figure 42. Water alkalinity at the three experimental lakes in the summer of 2005. ....112

xiii Abstract

Nova Scotia is home to the northern-most disjunct populations of Atlantic Coastal Plain

flora species, 11 of which are listed as species at risk by the Committee on the Status of

Endangered Wildlife in Canada. This study examined variables that may be limiting the

rare elements of this flora in hopes of establishing why they have not expanded to utilize

other lakes that appear to be good habitat, including dam reservoir lakes, and how we

might alter the management of water levels at reservoir lakes to improve habitat quality there. Two transplant experiments were conducted to examine environmental variables

that may be limiting two of the at risk species, Coreopsis rosea Nutt. and Hydrocotyle

umbellata L. The first experiment was to determine the effect of growing season length

and involved three treatments: time of planting (June, July or August), depth on the

shoreline, and water fluctuation regime (three different lakes). The second experiment

examined the effect of environmental variables including substrate, vegetation cover, soil

moisture, temperature, and water quality. I found that establishment time, flooding, and soil nutrient holding capacity appear to limit both species. The effect of unseasonable flooding was similar to the effect of later establishment in reducing biomass and flowering. Growth and survival were poor under constant flooding. Earlier establishment resulted in slightly higher survival under poor conditions. C. rosea, the upper shoreline specie, though superior in its cold tolerance, had inferior mechanisms for adjusting to flooded conditions in comparison with H. umbellata, the lower shoreline species. At reservoirs, higher fall water levels would prevent cold damage, earlier spring draw down would reduce flood stress, and opening the dam during periods of high precipitation would avoid plant damage due to unseasonable flooding.

xiv Acknowledgements

I am very grateful to all those who have been involved in this project and to those who

have made my time in Nova Scotia so wonderful. Thank you to my supervisor Dr. Ed

Reekie who always made room in his busy schedule to meet, always had kind and

thoughtful input, and who taught me how to be a collaborator as well as a student. Thank

you to the readers of this thesis Dr. Sherman Boates, Dr. Rodger Evans, Dr. Phil Taylor

and Dr. John Roff, and to Ruth Newel, Dr. Soren Bondrup-Nielsen, for their time and

input. Many thanks also to the Atlantic Coastal Plain recovery team members who shared their knowledge and resources. Thanks also to Chris McCarthy at Kejimkujik

National Park for showing me around the water pennywort populations there and sharing his impressions of our sites. Thank you to Jim Gavels at the Tusket Hydro-station for access to data on water levels and for showing me the dams. Thank you to Mike

Brylinsky for his advice, equipment, and lab space to process my water samples. Thank you to Brent Lennox for his help with my LOI samples and Ian Spooner for the use of their muffle furnace and sedimentation columns. Thank you to Jane Harrington and

Melanie Priesnitz for helping me to keep the plants healthy in the greenhouse and

experimental gardens. Thank you to the faculty and support staff at Acadia who made

my time here warm, engaging, and productive.

One of the wonderful things about this project was getting to spend so much time in the

Tusket River area. I am very grateful to the people in those communities for their

interest, knowledge, and welcome, and to all the friends and family who came to the

Tusket to help me with my field work there. Thank you to Andrew Trant for generously

xv sharing ideas, papers, logistics, resources, and a drizzly home away from home. Thank

you to the intrepid Christine Dawe for her good humour and willingness to try new

things, and of course for the long days and hard work as my field assistant in 2004. I am

grateful to Geoff Kershaw, without whose good humoured help I would still be digging

up plants and picking off leaves. Thank you very, very much to Eric Kershaw, who

helped in many ways, going out to the field with me from April until December two years

straight, canoeing in the sleet and snow, enduring weeks of tedium weighing those tiny pennywort samples, cooking many of our meals, and making the two years in Nova

Scotia so exceptional.

To my brown-house mates, grad cohorts of 2002-5, and all the friends and peers who have shared this time, thank you for your enthusiasm and ideas.

Thank you to Joyce Gould for her contagious passion for rare plants and encouragement.

Thank you to Ellen MacDonald for getting me started and to NSERC for making it possible to keep going.

Finally, and first of all, thank you to my family.

xvi Chapter 1 - General introduction

"Altogether the list of southern Coastal Plain plants reported from Nova Scotia numbered between 30 and 40… the fact that several such species, unknown in adjacent New

Brunswick and eastern Maine, are obviously isolated on Nova Scotia as remnants of the flora which in the late Pleistocene or even later had lived on the then elevated but now submerged continental shelf made it evident that not only was there plenty of good botanizing left in peninsular Nova Scotia but that the region must hold some secrets of profound importance to a clear understanding of the history of life in eastern America."

Fernald (1921) of his 1920 expedition to Southwestern Nova Scotia and the Tusket River

1.0 Background

The Atlantic Coastal Plain flora (ACPF) is a unique suite of species found in the south- western region of Nova Scotia (Roland and Smith 1969, Roland 1991, Zinck 1998). This disjunct assemblage of plants counts in its ranks some of the province's rarest species

(Maher et al. 1978, Pronych and Wilson 1993) and occupies a habitat that is threatened

(Boates et al. 2005, Eaton and Boates 2002, Wisheu et al. 1994, Wisheu and Keddy

1989). Many initiatives have been taken by government and citizen groups to conserve both specific species and habitat (e.g. Lewis 1988, Francis and Munro 1994). One of the major obstacles faced by managers and conservation groups is the lack of research on the basic ecology and biology of these species (Keddy and Keddy 1983, Wilson 1984,

Newell 1998, Newell 1999). Since the first description of Coastal Plain species in the

1 province by Fernald (1921, 1922) watersheds have been dramatically altered by damming

and other development activities (Boates et al. 2005). After the hydro-electric dam was

put in place on the Tusket River (Yarmouth County) in the late 1920's, Coreopsis rosea

Nutt., one of the rarest Coastal Plain species, was extirpated from three sites; it is now

only known from eight sites in all of Canada (Keddy and Keddy 1983, Keddy 1985a,

Newell 1998). Other areas in the watershed appear appropriate for colonization by rare

elements of the Coastal Plain flora, including reservoirs feeding into dam headwater

lakes. The mechanisms restricting the rare ACPF species from expanding their range to

occupy these areas are poorly understood (Hazel 2004). Greater understanding of the

variables that limit the rare elements of this flora could help us to better understand the

dynamics of present populations, and thereby better manage and maintain them,

potentially moving toward restoring numbers of the rarest species and restoring habitats.

Evaluation of variables governing the occupation of potential habitat could also expand

our understanding of the plant zonation and growth differences among and within lakes

(Hill and Keddy 1992, Keddy 1984, Keddy 1985b, Keddy 1989, Sharp and Keddy 1985,

Shipley et al. 1991a, Shipley et al. 1991b) and the role of natural disturbance regimes in

the maintenance of biodiversity (Keddy 1984, Keddy 1985b, Morris 1994, Morris et al.

2002).

1.1 The Atlantic Coastal Plain flora (ACPF)

Grouped by their common biogeographical history and occurrence primarily in

oligotrophic wetland habitats (including seasonal ponds, lake shores, peatlands, wet

flatwoods and some river shores), the Atlantic Coastal Plain flora (ACPF) is a

2 taxonomically unrelated group of species from the Atlantic Coastal Plain physiographic

region (Figure 1) (Rawinski and Price 1994). The main range extends from Texas to

Cape Cod, though with habitat loss due to urbanization on the eastern seaboard many of

the species are becoming increasingly rare within this range (Keddy 1994, Schneider

1994, Sorrie 1994, Sutter and Kral 1994, Wisheu and Keddy 1989). Disjunct populations

of certain species of the ACPF exist in the Great Lakes region and in south-western Nova

Scotia (Figure 1). The Great Lakes disjunct populations include pockets of ACPF south

of Lake Michigan, on the northeast of Georgian Bay, and off southern Lake Erie

(Reznicek 1994). The Nova Scotia disjunct populations of ACPF are the northern-most

occurrences for sixty-six of the ACPF species (Sweeney and Ogilvie 1993). This study focuses on the rare ACPF in Nova Scotia.

The biogeographical story of the ACPF is a fascinating one. Speculations on the revegetation of Nova Scotia after the Wisconsin ice sheets retreated has ignited an interesting academic debate, as reviewed in Holland (1981). The vegetation history of the Maritimes has been studied by Livingstone (1968), Mott (1975a, b) and Ogden (1960,

1965). Refugia in the southern states, on exposed continental shelf and within glaciated regions, and deciduous forests marginal to the glaciers have all been proposed to explain the revegetation of Nova Scotia (Holland 1981). The theory most often presented to explain the introduction of ACPF into Nova Scotia is an expansion via exposed continental shelf (Figure 2). Keddy and Wisheu (1989) present that approximately

14,000 years ago the sea level was over 100 m below present levels, exposing both

George's and Brown's Banks (portions of the continental shelf that are, at present under

3 the Atlantic Ocean) between Cape Cod (Nantucket Island) and Nova Scotia. It is posited that ACPF species could have colonized this exposed continental shelf, slowly advancing north to Nova Scotia. This land would have been submerged again as overland flow from melting ice sheets flowed into the ocean, raising sea levels. Jackson and Singer (1997) suggest a similar colonization of newly exposed materials following glacial retreat up through river valleys, as an explanation for the origins of the ACPF Great Lakes disjunct populations.

The ACPF have interesting attributes as a group that may have contributed to their expansion northward and for their rarity today. Wisheu and Keddy (1994) have shown that the ACPF species have low competitive ability; they disappear from areas where taller more aggressive herbaceous and shrub species establish and out-compete them.

This restricts the ACPF to areas where other species cannot survive, such as areas disturbed by wind, waves, ice scour and changing water levels, and those with low nutrient availability. The ACPF, even the rare elements, can be locally very abundant, and it has been suggested that they are what Grime (1973, 1977, 1979) described as stress tolerators: slow growing perennials able to survive in areas of low nutrients and low competition (Gaudet and Keddy 1988). What is interesting is that these species exist in areas that are both low in nutrients and high in disturbance, something Grime thought to be impossible. The adaptations that allow the ACPF to tolerate stress from both disturbance and low nutrient levels may have also allowed them to colonize exposed continental shelf and newly deglaciated areas. There would have been greater expanses of sandy to rocky wet areas in a postglacial landscape, in which the ACPF could have

4 potentially flourished. The proportion of these habitats has likely shrunken and

connectivity decreased as soils have developed, features eroded and water tables dropped.

In Nova Scotia the hot spots for Coastal Plain flora diversity and rare species occurrence

are now large watershed lakes in the south-western portion of the province, in particular

on the Tusket (Hill and Keddy 1992), Mersey and Medway River systems (Keddy and

Wisheu 1989, Wisheu et al. 1994, Roland 1991). The large watershed lakes situated at

the end of the catchments within these river systems receive the greatest levels of water

fluctuation, within and between years, which creates the necessary amount of natural disturbance and wide shorelines necessary to support ACPF (Holt et al. 1995). Keddy

(1989) showed that when shrubs are experimentally removed, ACPF will successfully colonize the exposed shoreline. Flooding appears to be an important mechanism to keep the shoreline free of shrubs, however we still do not know what flooding frequency is necessary to keep shrubs at bay and maintain ACPF (Keddy and Wisheu 1989). The local bedrock type appears to be important too; areas underlain by "red till" make the best

ACPF habitat as they result in gently sloping fine gravel or sandy shores (Keddy 1984,

Keddy and Wisheu 1989). When these gently sloping fine gravel and sandy shores occur on large catchment area lakes, in particular when they occur in areas exposed to dominant winds, they are kept free of shrubs and dense permanent mats of vegetation (Morris 1994,

Morris et al. 2002, Keddy 1989, Gaudet and Keddy 1995) by periodic flooding and ice scouring (Keddy and Wisheu 1989). Nutrient levels are kept low on these exposed shores by repeated wind and wave disturbance which strip the shore of finer sediment such as silt and clay and remove most organic material (Wisheu and Keddy 1994). It

5 appears that only with the correct combination of substrate, exposure and water fluctuation regime can ACPF flourish, particularly the rare elements (Figure 3).

1.2 Rarity

According to Rabinowitz (1981), there are seven forms of rarity based on the combination of their local abundance, geographic range, and broadness of their habitat specificity. The rare ACPF in Nova Scotia are generally locally abundant, occurring over a large geographic range but only in a specific habitat. The ACPF in Nova Scotia are also at the northern-most extent of their range and are disjunct from other populations.

Effort has been invested in establishing the location and size of populations of ACPF species in Nova Scotia, particularly within the Tusket, Medway and Mersey River

Systems (Keddy and Wisheu 1989, Wisheu et al. 1994, Roland 1991). Surveys have mostly focused on the presence or absence of ACPF populations, with little systematic surveying of presence or absence of unoccupied but appropriate ACPF habitat. It may be that ACPF habitat is itself rare. To date, no transplant experiments have been carried out to determine if the plants could establish elsewhere in south-western Nova Scotia, or even in other locations within the watersheds or lakes that they occupy at present. Keddy and

Wisheu (1989) suggest transplant experiments could help determine if it is dispersal limitations or lack of suitable habitat that produce the observed pattern of rare ACPF in

Nova Scotia. Wisheu and Keddy (1994) also note that transplant experiments would help to reveal what mechanisms restrict Coastal Plain species to infertile and exposed shorelines. Determining the rarity of suitable habitat for the rare ACPF would greatly

6 improve our capacity to ensure that the rare elements of the ACPF are maintained as part

of the biodiversity of the Nova Scotia and eastern North America.

1.3 Conservation initiatives

Throughout their range, the ACPF are facing increasing pressure. The eastern seaboard

is among the most densely populated areas in North America, habitat loss to urbanization

and eutrophication has lead to a deterioration of the status of ACPF throughout its main range (Wisheu and Keddy 1989, Keddy 1994). Nova Scotia has relatively undisturbed

ACPF habitat in comparison (Sweeney and Ogilvie 1993), but even here damming, cottage development, associated shoreline alteration, and ATV use pose significant and ongoing threats to ACPF and the integrity of their habitat (Boates et al. 2005). Dams are of particular concern as their placement often coincides with best habitat, the large watershed lakes at the end of catchments (Hill and Keddy 1992).

A Coastal Plain Recovery team, consisting of government agency representatives, various stakeholders and citizens groups, has been formed to address these threats and other conservation issues relating to the ACPF. The Nova Scotia Department of Natural

Resources has been very active in studies on threats to the ACPF here, including water quality and invasive species, and with many other projects such as ACPF information sessions for municipal governments and planners. They have also been involved in ongoing mapping, field surveys and compilation of ACPF information in conjunction with the Atlantic Canada Conservation Data Centre. The Nova Scotia Nature Trust has taken initiatives such as public information sessions and information packages to cottage

7 owners, and land acquisition, stewardship and conservation easement programs. In 1988,

a nature reserve was established on Wilson's Lake, one of the hotspots for species

richness in the Tusket River Valley (Lewis 1988). A citizens group called the Tusket

River Environmental Protection Association was active in this initiative and in public

education. Some ACPF are also protected within Kejimkujik National Park, where

conservation, research and education programs for the rare species have been put in

place.

Both in the COSEWIC reports and in the Multiple Species Recovery and Conservation

Strategy and Action Plan for the ACPF in Canada (2005), a lack of knowledge of the

basic biology and ecology of many ACPF species has been identified as an obstacle to

conservation and management. This is of particular concern for the rarest species such as

Coreopsis rosea Nutt. (pink tickseed) and Hydrocotyle umbellata L. (water pennywort)

as there are only a few populations of each of these species in the province and they are

thus, on the whole, more vulnerable. The recovery plan (Boates et al. 2005) also indicates that transplant trials would be beneficial to assess possibilities for population restoration.

1.4 Natural water fluctuations and watershed alterations

Natural water fluctuations play an important role in maintaining species richness

(Dynesius and Nilsson 1994), maintaining habitat (Keddy 1991), and establishing zonation patterns within lakes and vertically on shorelines (Wisheu and Keddy 1989,

Keddy 1984). Keddy (1991) explains that water level fluctuations in wetlands play a

8 similar role to that of fire in forests in that they destroy vegetation, and allow regeneration from seed. At more moderate levels of water fluctuation, only certain species may be destroyed or excluded, such as shrubs (Wisheu and Keddy 1994) as woody species are particularly sensitive to flooding (Kozlowski 1984). With ACPF, this disturbance restricts shrubs and allows ACPF herbaceous species to persist (Keddy

1989). Stabilization of water fluctuations simplifies the shoreline, allowing shrubs to encroach into and replace the marsh and wet meadow portion of the shoreline (Figure 4).

In other systems, water fluctuations are important for recruitment as well as habitat maintenance. Rood et al. (1995, 2003a, b) showed the role of periodic flooding in recruitment of Populus deltoides Bartr. ex Marsh. (cottonwood), the keystone species in that system, and the importance of recreating flooding disturbance in order to restore the community. There is increasing recognition of the importance of restoring disturbance dynamics and flood pulsing in wetland restoration and maintenance efforts (Middleton

1999).

The flow of many large river systems has been altered by human interventions such as damming for hydroelectric generation (Dynesius and Nilsson 1994). Damming, diversions, draining, and other such watershed alterations have contributed to the loss of

50% of wetlands in the US and Canada (Moore et al. 1989). Appropriate levels of water fluctuation appear to be key in maintaining ACPF habitat. For ACPF, alteration of natural patterns of water flow can be disastrous, as seen in the Tusket watershed (Keddy

1985a) where rare elements of the ACPF were extirpated from three locations following stabilization of water levels with dams.

9 Alterations to the Tusket River watershed include cottage development, road

construction, forestry, and agriculture, but by far the most dramatic effects have been due

to damming (Eaton and Boates 2002). Small scale, temporary dams were used to control flow for moving logs in forestry operations in Tusket River Valley during the early part

of the century (Figure 5); larger scale dams such as Tusket falls were put in place for

hydroelectric power generation in the late 1920's (Figures 6 and 7) (Eaton and Boates

2002). When the hydroelectric dam was constructed, Lake Vaughan, Gavelton Lake and

Kings Lake became one head water lake emptying through the hydro-station dam

turbines. A secondary, smaller dam was located where the Tusket River entered Lake

Vaughan, controlling the amount of water going into Lake Vaughan (see Figure 7). This caused an enlargement of the river and the formation of Raynard's Lake (Figure 8). The rise and stabilization of water levels in Lake Vaughan, Gavelton and King's Lake, and the alteration of flow down stream from the dam are the reasons cited for extirpations of some of the rarest ACPF, Sabatia kennedyana Fern. (Plymouth gentian) and Coreopsis

rosea Nutt. (pink tickseed), from these areas (Keddy 1985a). Even after seventy-five years, the historic populations of these rare elements of the ACPF have not reestablished

at those sites (Keddy 1985a). Morris et al. (2002) estimated that half of the large

watershed lake shoreline has been lost as Coastal Plain flora habitat to development and

damming.

An interesting twist in the story of the ACPF is that many of the reservoir lakes have

been colonized by the more common ACPF species, but not by as many of the rare

species as you would expect (Hill et al. 1998). One exception is Kempt-Back Lake,

10 which despite being a reservoir for many years, supported two rare species, which is high

for its size and location (Hill and Keddy 1992, Hill et al. 1998). This suggests that if

properly managed, reservoir lakes could support populations of rare ACPF. In contrast to

Kempt-Back Lake, areas such as Raynard's Lake possess what looks like appropriate

habitat and even have some ACPF species such as Rhexia virginica L. (meadow beauty),

but conspicuously lack other rarer elements of the ACPF despite their presence in nearby

lakes, and presence upstream in the case of S. kennedyana and C. rosea (Eaton and

Boates 2002). As discussed by Keddy (1999), research on how water fluctuation regime

controls plant distribution could help us to better manage and potentially restore these

systems.

2.0 Purpose of the study

The aim of this study was to examine environmental variables limiting the rare elements of the Atlantic Coastal Plain Flora in Nova Scotia, and to increase our understanding of

the biology and ecology of these plants, therefore expanding our toolbox for conservation and management.

There are many mechanisms that may be limiting the rare elements of the ACPF in Nova

Scotia. Dispersal may be restricting colonization of appropriate habitat. This can in part

be addressed by monitoring the success of experimental transplants. On the other hand, it

may be that our perception of what constitutes appropriate habitat is incorrect or

incomplete. This too could be tested with experimental transplants by measuring

environmental variables and the response of transplants to those variables.

11 This study involved two transplant experiments: the first and second were respectively designed to assess the importance of growing season length, and physical environmental variables, in limiting the growth and survival of the study species. The growing season experiment involved three treatments to explore the importance of time of establishment, duration of spring flooding, and water fluctuation regime, and their relationship to growing season length and their impact on transplant health and survival (Chapter 2).

The second experiment focused on the measurement of environmental variables at the plant, shoreline depth and lake levels and their effect on transplant growth and survival

(Chapter 3).

3.0 Study species

The species chosen for the transplant experiments were Coreopsis rosea Nutt. (nationally

endangered and globally rare) and Hydrocotyle umbellata L. (nationally endangered and

nationally rare) (Keddy and Keddy 1983, Wilson 1984, Newell 1998, Newell 1999)

(Figure 9). These species were selected among the ACPF due to their provincial and

national significance and the availability of transplant material. Given that these are rare

species, care was taken to utilize material that had already been taken from the field for

other projects, and thus reduce impact by reducing the amount of material being removed

from natural populations. Plants for this experiment were propagated from H. umbellata

and C. rosea growing at the Harriet Irving Botanical Garden at Acadia University. This

material had been collected in the fall of 2000 from Wilson's Lake within the Tusket

River Reserve; the collected material consisted of vegetative slips of H. umbellata and

seeds from C. rosea. The plants propagated from these slips and seeds were maintained

12 in plastic flats filled with a mixture of 2 parts peat based potting soil to 3 parts sand. In

the winter these flats were kept in a heated greenhouse, and then moved outdoors in the

summer. This work was done under permit (pursuant to section 14 (1)(a) of the Nova

Scotia Endangered Species Act 1998 and section 16 of the Nova Scotia Special Places

Protection Act 1989).

4.0 Study sites

Study sites were chosen from lakes within the Tusket River system in south-western

Nova Scotia, near the town of Yarmouth (Figure 10). Though some of the Atlantic

Coastal Flora species are present at other locations in south-western Nova Scotia, the

Tusket River watershed has the highest species richness of ACPF in Atlantic Canada, and

is the only place where my two study species are found together.

Three lakes were chosen that have common ACPF at present and reflect different hydrologic regimes: Wilson's, Raynard's and Kempt-Back Lake. Wilson's Lake was chosen as the control lake as it is the only lake in south-western Nova Scotia that supports natural populations of both H. umbellata and C. rosea. This was important as it allowed

us to test the effect of transplanting on a shoreline known to be appropriate habitat for

both species. Wilson's Lake is undammed and has the highest species richness and

diversity of rare ACPF in Nova Scotia (Hill et al. 1998). The other two lakes were

chosen from the same watershed, as close as possible to Wilson's Lake, to keep substrate

and climate as consistent as possible. Raynard's Lake is a reservoir with a dam which is

periodically opened to allow more water to flow into Lake Vaughan, the Tusket Falls

13 power generating station's head pond (Figure 6). It was chosen as it was known to

support fewer rare ACPF than predicted by models by Hill et al. (1998) though it appears

to be good habitat. Kempt-Back Lake was chosen as it was known to support a high

number of rare species despite having being a dam reservoir lake; water levels were

allowed to drop sooner in the spring at this reservoir lake than at others, and it was

suggested by Hill et al. (1998) that this may be the reason that it supported so many rare

species. Once the study began I discovered that the lake was no longer being managed

by Nova Scotia Power as a reservoir lake, and that the weir that had previously been used

to regulate water flows had been boarded up, likely by locals, keeping the water levels

fairly uniform (Jim Gavels, personal communication).

Sites within each lake were chosen on the basis of known ACPF habitat preferences.

These plants are known to inhabit gently sloping, rocky to sandy, exposed shorelines with little competition from shrubs (Keddy and Wisheu 1989, Wisheu et al. 1994). Three sites satisfying these criteria were chosen at each of the study lakes. Sites were selected along the eastern to south-eastern shores of the lakes to keep sunlight, wind and wave exposure consistent. Individual sites on a given lake were at least 100 m apart.

14

*

Figure 1. The Atlantic Coastal Plain physiographic region, indicated in yellow, east and south of the black line, extending from Texas to Cape Cod. The main areas of disjunct populations are shown as circles; the asterix indicates a smaller disjunct of only a few species in Cape Breton, N.S. (adapted from http://tapestry.usgs.govna-info.html).

15

Figure 2. Changes in sea level in relation to a profile of Nantucket Island,

Massachusetts, to Cape Sable Island, Nova Scotia (adapted by Keddy and Wisheu 1989 from Odgen and Harvey 1975).

16

Figure 3. Factors affecting the distribution of Coastal Plain habitats along lakeshores

(adapted by Sweeney and Ogilvie 1993 from Keddy and Wisheu 1989).

17

Figure 4. The effects of eliminating water level fluctuations on a shoreline: the four vegetation zones in the top figure are converted to a simple two zone wetland in the bottom figure eliminating the marsh and wet meadow communities (Keddy 1991).

18

Figure 5. Images of forestry related dams in the early 1900's: photo on left is of a log drive on the Tusket River system, and the photo on the right is a sawmill and sawmill dam in operation on the Tusket River (photo courtesy of the Tusket Court House, from

Eaton and Boates 2002).

Figure 6. Images of hydroelectric dams on the Tusket River at present: photo on left is the main dam at Tusket Falls in 2001, and the photo on the right is the Tusket Falls

hydroelectric power station and dam in 2001, first built in 1929 by the Nova Scotia

Power Corporation (photo courtesy of Samara Eaton, from Eaton and Boates 2002).

19

Figure 7. Present and past locations of dams on the Tusket River system: the map on left shows present day hydro dams, saw mill dams and unknown pre 1918 and older dams, the map on the right names all hydro dams currently on the river system (Eaton and

Boates 2002).

20

Figure 8. Changes to the Tusket River system due to the installation of the Tusket Falls and Carleton dams: the map on the right is a portion of the 1864 A. F. Church map showing the Lake Vaughan area, and the map on the right is from the Nova Scotia Department of Natural Resources 1989 GIS map of the Lake Vaughan area (Eaton and Boates 2002). Note names changes: Bell Lake is now

King's Lake, Coven's Lake is now Bennet's Lake, and St. John Lake is now Wilson's Lake.

21

Figure 9. Images and distributions maps of the study species; C. rosea locations are

indicated with black dots in the range map on the left, and H. umbellata range is south of

the black line and within the circles on the range map on the right. (Range maps adapted from COSEWIC reports: Keddy and Keddy 1983, Wilson 1984, Newell 1998, Newell

1999.)

22 Kempt-Back Lake

Raynard's Lake Wilson's Lake

Tusket Hydro- Station Dam

Yarmouth

Figure 10. Tusket River Valley, Tusket Hydro-Station and study lakes in relation to the province and to Yarmouth (adapted from http://www.mar.dfo-mpo.gc.ca/masaro/english/

Species_Info/Atlantic_Whitefish).

23 Chapter 2 - The effect of growing season length

1.0 Introduction

The Atlantic Coastal Plain flora (ACPF) is at the northern-most edge of its range in Nova

Scotia (Sweeney and Ogilvie 1993). Length of growing season may be important in limiting the growth and survival of ACPF and in controlling their distribution in Nova

Scotia. If the rare ACPF in Nova Scotia are already at the limit of their climatic tolerance, a change in hydrologic regime affecting growing season length could hamper their growth and survival. Work by Hill et al. (1998) in south-western Nova Scotia suggests that the length of the growing season there may be drastically shorter at dam reservoir lakes because the lake levels are kept artificially high through the spring and summer. Exposure of the shoreline may be important for triggering seed germination and regrowth from below ground material, and will likely enhance growth rates. Exposed soil will reach higher daytime temperatures, and have higher light penetration and oxygen levels than flooded shoreline. For perennial species, the extended flooded period at dam reservoirs may deplete their carbohydrate reserves, reducing their vigour and potentially their survival.

Building on Hill's work, Hazel (2004) characterized the hydrologic regime of a reservoir and a natural lake on the Tusket River system in south-western Nova Scotia, and experimentally tested the effect of a shortened growing season in a garden experiment on rare ACPF and common associated species. The common species were chosen as they were known to tolerate reservoir conditions, unlike many rare ACPF. Hazel (2004) compared the responses of the common and rare plants to a shortened growing season to

24 see if that was what prevented the rare species from successfully colonizing reservoirs.

She chose a reduction in growing season of 30 days, roughly reflecting the delay in

exposure experienced by plants at the upper shoreline at Raynard's reservoir in the Tusket

system. She found that shortening the growing season by 30 days equally reduced

growth of the rare and common plants. It was concluded that for upper shoreline plants,

growing season length was likely not a more important limiting element for the rare

ACPF than the common plants. This research suggests though that the shortening of the

growing season at lower shoreline positions on the reservoir may be much more

important given the greater reductions in growing season there might result in a more

dramatic limitation of the rare species than the common species. The length of the

growing season (days above water and above 0°C) at the lower shoreline positions at

Raynard's Lake reservoir was calculated to be 50 versus 120 days at Wilson's Lake (a

natural lake), a potential reduction of 70 days. Hazel (2004) assumed that growth below

water is insignificant, an assumption tested under field conditions in this research.

The main objective of this experiment was to test the effect of growing season length on

rare ACPF species along a vertical shoreline gradient under field conditions. We had three treatments: transplanting at different dates, transplanting at different shoreline

depths, and transplanting at lakes with different hydrologic regimes. We used two rare

ACPF as our target species: Coreopsis rosea Nutt. (pink tickseed), an upper shoreline

plant, and Hydrocotyle umbellata L. (water pennywort), a lower shoreline plant. For the

first treatment the target species were transplanted at three successive dates: June 26-28th,

July 21-23rd and August 18-20th. This allowed direct testing the effect of a reduced

25 growing season. These transplants were placed at different depths below the high water mark on the shoreline (the second treatment) within their normal range. This second treatment allowed evaluation of the effect of shoreline position and how it related to growing season length. Planting at these different shoreline depths also allowed evaluation of the importance of growth below water for these species, as the transplants at different depths on the shoreline were below water for varying amounts of time. The third treatment was to replicate the first two treatments on lakes with different water fluctuation regimes so that the effect of time of shoreline exposure could be tested.

Together, the treatments in this experiment should determine if the length of growing season is the primary explanation for differences in the success of these species at different lakes and at different depths along the shoreline.

I hypothesized that growing season would be important in influencing growth and survival of these species in Nova Scotia as they are at their northern-most location, and that the importance of growing season length would be reflected in all treatments. I expected plants put out later to be smaller and have poorer survival. Based on Hazel's work (2004), I anticipated the plants at lower shoreline positions to do more poorly. I also expected that plants at lakes that had high water levels for longer in the spring would have shorter growing seasons, resulting in smaller plants and lower survival than at lakes where water levels dropped earlier.

26 2.0 Methods

2.1 Transplant preparation

Standardized transplant plugs were prepared from H. umbellata and C. rosea maintained

at the Harriet Irving Botanical Garden at Acadia University from a previous experiment

(see section 3.0 of the Introduction). Transplants were prepared from this stock material

through a process of cleaning, cutting and treating the vegetative slip for any potential

pests. Trays of H. umbellata and C. rosea were first washed over sieves to remove soil

and expose the roots (Figures 11 and 12). Once soil was removed by rinsing, the

individual strings of rhizomes were untangled and standardized vegetative slips cut

(Figure 13). For H. umbellata this consisted of a 4 cm length of rhizome with 1 node and

a leaf with a diameter of 1-1.5 cm. Roots were trimmed to a maximum length of 1 cm.

For C. rosea the slips consisted of a 2 cm length of rhizome with one ramet that was 3-5

cm in height. Roots were trimmed to a maximum length of 3 cm. These transplants were

prepared April 28-29th, 2004 for C. rosea and May 4-5th, 2004 for H. umbellata.

Once cut, the standardized transplant slips were kept under moist paper towel to reduce

moisture loss. The above ground portions of transplants were dipped in a 1:50 dilution of

Safer's insecticidal soap to kill any pests if present (Figure 14). Plant material was then planted in a mixture of 2 parts peat based potting soil to 3 parts sand. Plug cell size for the transplants was 5.5 cm deep, 4 by 4 cm square plugs for H. umbellata, and 5.5 cm deep, 4.5 by 5.5 cm plugs for C. rosea.

27 Once prepared, the transplants were placed in growth chambers for a period of 7 weeks.

The initial conditions were set at 80% humidity, 15°C day and 10°C night temperatures,

with 8 hours of light at 400 µmol m-2 s-1, to recreate late summer conditions. These

conditions were maintained for a week, after which the temperatures were dropped to

12°C during the day and 7°C at night, followed by 10°C during the day and 5°C at night

the following week. After three weeks the root production was judged insufficient to

produce successful outdoor transplants, so the conditions in the growth chambers were

increased to 20°C during the day and 15°C at night, with 12 hours of light at 680 µmol

m-2 s-1 to stimulate more vigorous growth. These conditions were maintained for 3

weeks, followed by one week cool down to harden off the plants at 12°C during the day

and 7°C at night, with the same light conditions. At that point transplants were randomly divided into three groups for the three transplant dates, two of which were placed in a walk-in fridge at 4°C and 80% humidity with no light, to arrest development and keep the plants at a uniform stage, while the remaining group was brought out to the field for transplanting. Plants kept in the fridge were checked daily and watered as needed to keep the soil moist.

2.2 Transplanting

There were three transplant sessions: the initial (week 0) transplants were done June 26-

28th, the second session (week 4) July 21-23rd and the third (week 8) August 18-20th.

According to Hazel (2004) upper shoreline plants at Raynard's Lake reservoir can experience a four week delay in exposure, while lower shoreline plants experience up to a

28 70 day delay. We chose to do plantings that were four and eight weeks after the original

planting to capture most of this range.

Two replicate transects were planted at each site per transplant session. Transects were

placed at 90° to the shoreline, running in straight lines perpendicular to the shore (the

aspects of each transect can be found in Appendix 1). Along the transects, transplants

were placed at equivalent depths relative to the high water mark so that the timing of exposure of plants would be the same across sites at a given lake as water levels dropped.

The high water mark was assessed as the average shrub line at the middle experimental site; depths below the high water mark there were established using a string level and a meter stick, then marked with flags (Figure 15, 16, and 17). At subsequent sites, the water level was used as the reference point to ensure that the planted positions at each separate site would be vertically equidistant to the water table and reflect the same moisture conditions. C. rosea was planted at 30, 40, 50, 60, and 70 cm below high water mark and H. umbellata was planted at 50, 60, 70, 80, and 85 cm below high water mark based on work by Hazel (2004) where she inventoried the shoreline and described the elevation range of these two species. We chose to plant in the middle portion of the range she reported, allowing for any between year changes in the lake levels and shoreline environment that may have rendered the uppermost or lowermost portion of the range unsuitable in a given year. In total, there were two replicate transplants per transplanting session at each of the five planting depths at each site and three sites per lake; thus replication of 6 plants per planting depth at a given lake and 30 plants total at

29 each lake for each of the species per transplant session. In all there were 90 plants per

species at each lake; a total of 540 transplants for this experiment.

To prevent confusion of transplants with natural plants of the same species at Wilson's

Lake, a 10 cm diameter plastic tube (from a clean 2 litre clear plastic pop bottle) was first installed with the top flush to the ground and any plant material removed from the soil before the transplant was inserted. This was not deemed necessary at the other two lakes as there were no reports of our study species found there. Figure 18 shows the size and root mass of transplants and Figure 19 shows transplanting under water.

2.3 Measurements

The transplants were visited from planting until November 27th, 2004, then from the

beginning of April 2005 when the ice broke up until the end of August when the

transplants were harvested, and the different measurements taken at different intervals.

Growth was measured at two week intervals through the first growing season, and three times in the second year (April 26th, July 10-13th and August 9-24th when the plants were

harvested). Growth was assessed using maximum and cumulative height, leaf number,

branch number and ramet number for C. rosea; leaf number, maximum height, and

minimum and maximum leaf width for H. umbellata. Flowering was recorded through

the two summers and first fall. Survival was noted every two weeks through the spring,

two summers and first fall. At the end of the experiment the transplants were harvested,

washed and separated into leaf, stem and below ground components and oven dried for

48 hours at 100°C before weighing.

30 2.4 Data analysis

All statistical analyses were performed using SAS for windows version 8.02 (SAS

Institute Inc., Cary, NC, USA). Analysis of variance (GLM procedure) was used to

determine if week of planting, vertical position on the shoreline, lake, or their interactions had a significant effect on final biomass and on growth indices over time. Week planted, depth along the shoreline of planting and lake were all treated as fixed factors; as a result, any inferences made from the analysis were applied only to the time frame, range of depths, and lakes which we tested. One growth indicator was chosen per species through a correlation analysis of final growth measures with final biomass. The best index of final H. umbellata biomass was Ln (number of leaves * maximum leaf diameter) (r2=

0.731 with the Ln of biomass), and the best index of final C. rosea biomass were Ln of cumulative height (r2= 0.701 with the Ln of biomass) and Ln of number of leaves (r2=

0.651 with the Ln of biomass). We used the Ln of the number of leaves as the index of

C. rosea as the cumulative height was not available for all sample dates. Due to a storm

between September 1st and September 17th, 2004, that killed the majority of plants at

Kempt-Back Lake, we could not include that lake in analyses from September 17th (day

2004261, the first four digits are year, the remaining three digits are Julian day) on; analyses from day 2004261 on only included Raynard's and Wilson's Lake. The residuals of the data used in these analyses were normal or showed only slight departures from normality, as tested with the Shapiro-Wilk test, stem and leaf plots or histograms, box plots, and normal probability plots (UNIVARIATE procedure). The Shapiro-Wilk test is very sensitive to slight departures from normality when samples sizes of over 50 are used; our sample sizes were from 75-270. The Shapiro-Wilk scores were all quite high,

31 even for the data with slight departures from normality (values of 0.83 to 0.98, 1 being normal), as verified with the various plotting techniques. These data were all used without further transformation in the analyses of variance as Sokal and Rohlf (2000) indicate that means will follow the normal distribution more closely than the distribution of the variates themselves, and that slight variations from normality have little impact on the significance of the F-test or the conclusions from analysis of variance. Survival analyses were done with categorical models (CATMOD procedure) in SAS.

3.0 Results

3.1 Effect of week planted

During the course of the experiment, time of establishment (week planted) had a significant effect on the size of both species (Tables 1 and 2). The difference in size of

H. umbellata (Ln (number of leaves * maximum leaf diameter)) between plants transplanted at different times became apparent in August of both years and was influenced by lake in early August 2004 and in the summer of 2005 (Table 1). Generally, earlier H. umbellata transplants were smaller in the first year, though interestingly, week

0 (the first transplant session) plants were the largest in the second year at Wilson's Lake

(Figure 20).

The difference in size (Ln (number of leaves)) of C. rosea among transplant dates throughout the experiment was significant (Table 2). The pattern for C. rosea was opposite to that of H. umbellata, in that the C. rosea transplanted earlier were generally larger than the ones planted out later; this difference in size was maintained throughout

32 the experiment (Figure 20). The effect of time of establishment (week planted) on the

size of C. rosea was dependent on lake throughout the experiment (Table 2). The week 0

C. rosea transplants were smaller than the other transplants at Wilson's Lake (Figure 20),

contrary to the pattern we generally saw at other lakes and at other transplant dates at

Wilson's Lake, suggesting that there was something unique to that transplant date at that

lake.

Time of establishment (week planted) was important for the final biomass of H.

umbellata on Wilson's Lake and C. rosea on Raynard's Lake (Figure 21). In both cases the first planting (week 0) did much better in terms of total, below ground and, in the case of C. rosea, above ground biomass (Tables 1 and 2, Figure 21). For H. umbellata, the

week planted had an effect on the above to below ground biomass ratio (Table 1); the

latest planting had a lower above to below ground biomass ratio (Figure 22). Kempt-

Back Lake displayed the same effect of week planted on above to below ground biomass

ratio as at the other lakes, though the means were not significantly different due to the

low sample size (data not presented).

Effect of time of establishment (week planted) on survival of the transplants was divided

into periods (Table 3), so that the timing and cause of mortality could be more easily

judged. The first full set of census data for the second year was collected July 10-13th,

2005, which was later than the first year because of high water levels and poor visibility in the spring of 2005. Generally survival was high, and week planted had little effect on survival (Figure 23). The only exception was at Kempt-Back Lake during the fall of

33 2004 (period 2); week planted was significant in explaining mortality for H. umbellata (p

= 0.0014). The earlier the H. umbellata transplants, the better their survival at Kempt-

Back Lake during period 2 (Figure 23): survival of week 0 transplants was greater than

that of week 4 transplants, which was in turn greater than that of week 8 transplants.

Only C. rosea produced flowers during the experiment, and week of planting was not

significant in influencing flowering.

3.2 Effect of depth

Depth of planting had a significant effect on plant size (Ln (# leaves * maximum leaf

diameter) for H. umbellata and Ln (# leaves) for C. rosea) throughout the two years of

the experiment for both species (Tables 1 and 2). Generally, plants of both species were

smaller lower on the shoreline (Figure 24 and 25). For H. umbellata, there was an

interaction between depth and lake on two days: on July 7th, 2004 (day 2004189)

transplants at Wilson's Lake at 80 cm were smaller than expected, and on August 18th,

2005 (day 2005230) transplants at 70 cm at Wilson's Lake were smaller than expected

(Table 1 and Figure 24). The effect of planting depth on the size of H. umbellata was influenced by time of establishment in August of the first year (Table 1 and Figure 24); week 0 plants were larger than week 4 plants at 50 cm depth but smaller at all greater depths. In the early fall of 2004 (day 2004261), week 0 H. umbellata transplants at 60 cm and 70 cm at Raynard's Lake, and at 80 cm at Wilson's Lake were also smaller than expected, resulting a three way interaction between depth, time of planting and lake

(Table 1 and Figure 24). For C. rosea, there was an interaction between depth of planting

34 and lake in late summer of both years (Table 2). Throughout, the effect of planting depth was most pronounced at Kempt-Back Lake and least at Wilson's Lake for C. rosea

(Figure 25).

The planting depth of the transplants had a significant effect on the final biomass of both species (Tables 1 and 2). As seen with measures of plant size, generally plants lower down on the shoreline had less biomass (Figure 26). For C. rosea, each increment of 10 cm vertical depth consistently decreased biomass (Figure 26). For H. umbellata, plants at the 50 cm position performed much better than all plants at greater depth; there was no difference in biomass between plants at 60 cm and deeper (Figure 26). Depth was particularly important in influencing amount of below ground biomass in H. umbellata

(Table 1); below ground biomass became a more important component of total biomass for H. umbellata higher on the shoreline (Figure 26). In H. umbellata the ratio of above/below ground biomass increased with vertical depth up until 80 cm after which there was no further change (Figure 27). At Kempt-Back Lake H. umbellata showed no variation in biomass with depth of planting (data not shown). Though the means were smaller than at the other lakes, C. rosea at Kempt-Back Lake followed the trend that plants were bigger higher on the shoreline particularly at 30 and 40 cm depth (data not shown).

The effect of depth on survival was not significant, likely because survival rates were so high, but there were some patterns worth noting (Figure 28). In the fall (period 2) at

Kempt-Back Lake, percent survival for both species decreased with depth (Figure 28). A

35 similar pattern was seen for H. umbellata at Kempt-Back during the winter (period 3)

(Figure 28). The survival of C. rosea at 40 cm below high water mark at Kempt-Back during the first summer (period 1) was abnormally low (Figure 28).

Depth had an important effect on flowering in C. rosea (Table 2); the number of flowers for C. rosea decreased with depth, particularly at Raynard's Lake (Figure 29).

3.3 Effect of lake

Each of the lakes had a different water fluctuation regime based on whether or not it was dammed and where it was in the watershed. Wilson's and Raynard's Lake are both low in the watershed and have the potential for high water level fluctuations, however Raynard's is dammed and the flooding is controlled. Wilson's Lake water levels were lower in the spring and much of the two summers than the other lakes, with the exception of a flood in

July 2004 that was much more dramatic at Wilson's Lake than at the others (Figure 30).

Raynard's Lake levels did not drop as quickly as Wilson's Lake in both years of the study because the dam was kept closed through spring (Figure 30). Raynard's Lake did not have a flood peak in July 2004 (day 2004205), because the power company opened the dam during the period of heavy precipitation (Figure 30). Raynard's Lake also had a dramatic drop in water levels in the fall of 2004 because of the dam; water levels at

Raynard's reservoir were much lower through the fall than at Wilson's Lake (Figure 30).

Kempt-Back is high in the watershed, and so receives smaller amounts of runoff and has smaller flood peaks. There is a weir on Kempt-Back which prior to 2001 was opened periodically to supply water to the hydro station head pond (Jim Gavels, personal

36 communication). Though it is no longer opened and closed by the power company, it has

been boarded up, likely by locals to keep water levels high for fishing (Jim Gavels,

personal communication). As a result, the water levels at Kempt-Back only dropped to

30 cm below the high water mark during the summers (Julian day ~175 to 275, the end of

June to the end of September), unlike the other lakes where levels dropped much lower

(Figure 30).

The planting location (lake) had a significant effect on both species throughout the

experiment. Hydrocotyle umbellata performed well at Raynard's Lake in the first year,

but were smaller in the spring of in the second year and did not reach the same size as at

Wilson's Lake by August when they were harvested (Figure 20). This was reflected in

the lower overall biomass of H. umbellata at Raynard's Lake (Figure 21). Hydrocotyle

umbellata at Kempt-Back Lake did poorly in both years; it was consistently much smaller

than plants at the other two lakes (Figure 20) and finished with a much smaller final

biomass (Figure 21). The ratio of above to below ground biomass for H. umbellata was

higher at Raynard's and Kempt-Back Lake than at Wilson's Lake (Figure 31).

Coreopsis rosea growth at Wilson's and Raynard's Lake was similar in the first year, but

in year two the growth (Figure 20) and above ground biomass for C. rosea (Figure 21)

were greater at Raynard's Lake. At Kempt-Back Lake C. rosea had poor growth in both

years, so the low final biomass is not surprising (Figures 20 and 21). Lake influenced the

ratio of leaf to stem biomass; Wilson's Lake had a higher leaf to stem ratio than at the other lakes (Figure 32).

37 Survival was generally high across all lakes, but in period 2, survival of both species was much lower at Kempt-Back Lake than the other lakes (Figure 23).

Lake had a significant effect on the flowering of C. rosea (Table 2), which flowered more at Raynard's Lake than at Wilson's Lake, and in turn more at Wilson's than at Kempt-

Back Lake (Figure 29). Lake also had a significant interaction effect with depth on the flowering of C. rosea (Table 2). There was no flowering at any of the lakes at 60 and 70 cm; at Raynard's lake the flowering began at 50 cm and was greater than at the other lakes at 30 and 40 cm, while Wilson's Lake only had a few flowers at 30 and 40 cm (none at 50 cm) and Kempt-Back had on average less than one flower per plant and only at 30 cm (Figure 29).

4.0 Discussion

4.1 Effect of week planted

When water conditions were favourable, earlier establishment had a positive effect on the performance of the two study species. A longer growing season did confer a significant size advantage to C. rosea plants at Raynard's and Wilson's Lake, with the exception of week 0 plants at Wilson's Lake. A mid-season flood just before the second round of transplants damaged the week 0 C. rosea transplants. Raynard's and Kempt-Back Lake levels were only marginally affected by the rain event that triggered the flood at Wilson's

Lake as the Raynard's Lake dam was open during that period, and Kempt-Back is higher in the watershed and on a smaller branch of the drainage area, and so, subject to much smaller flood peaks. Water levels only increased by 10 cm at Kempt-Back Lake and 5

38 cm at Raynard's Lake, versus 65 cm at Wilsons Lake. Plants that evolved with frequent

enough flooding have developed physiological, morphological, anatomical and life

history characteristics to avoid damage from the low oxygen levels associated with

flooding (Crawford 1989, Gurevitch et al. 2002). Some examples include differentiation

of leaf morphology between aerial, floating and submersed leaves to increase gas exchange, production of cytoxhrome oxidases which react with oxygen at low

concentrations, and enlargement of intercellular spaces or cell differentiation to form aerenchyma tissue which allow gas diffusion to and from the roots (Crawford 1989,

Gurevitch et al. 2002). Hydrocotyle umbellata is able to quickly alter its morphology

following a rise in water level by extending the length of its petioles so that the leaves are

at the water surface (personal observation). Coreopsis rosea however, is not able to

quickly change its morphology to adapt to flood conditions (personal observation). As a

result, the week 0 transplants of C. rosea were damaged by the flood, after which they did not grow as quickly or as much as the week 4 transplants that year. The following summer they were as small as the week 8 transplants with similar final above, below and total biomass. The similar size of the week 0 and week 8 C. rosea transplants a year after the flood suggests that the damage caused by that flood was roughly equivalent to an 8 week reduction in growing season. This may be due in part to a depletion of carbohydrate reserves of week 0 plants during the flooded period, and also to the lower metabolic efficiency of the flood damaged tissue (Crawford 1989). Had there not been a flood that year affecting the week 0 transplants at Wilson's Lake, we would have likely seen the same pattern at Wilson's Lake as Raynard's Lake where the older plants were bigger and more successful.

39 Hydrocotyle umbellata appeared to be negatively affected by earlier establishment in the

first year, but biomass measures in the second year showed that at Wilson's Lake, earlier

establishment did result in greater biomass, particularly below ground. In the first year earlier H. umbellata transplants looked smaller than ones planted later, particularly at

Kempt-Back Lake (causing a lake*week interaction on day 2004217, Table 1 and Figure

20), likely because the shoreline range where we planted H. umbellata was almost entirely below water at most lakes and the earlier transplants would have had a different growth form in response to the flooded conditions than the newly transplanted plants.

Older plants would have been adapted to the flooded conditions and would have had few leaves but long petioles so that those leaves could reach the surface, while new transplants would still have had many leaves on short petioles that were not as efficient as they were entirely below water. The growth index we used included both the diameter of leaves, which increases slightly in flood adapted leaves, as well as the number of leaves; the diameter of flood adapted leaves would not affect the index value as much as number of leaves which varied over a broader range. As the growth index we used was more strongly affected by number of leaves, these older plants would have appeared smaller than newly transplanted plants though the flood adapted older transplants would have been more active and building up more below ground biomass as we saw with the biomass results for week 0 plants at Wilson's Lake. The greater size of the week 0 H.

umbellata transplants in 2005 at Wilson's Lake was seen in the significant lake*week

interactions through that summer (Table 1, Figure 20). In the end, earlier establishment

conferred a significant size advantage, largely below ground. The lack of effect of

establishment time on H. umbellata at the other lakes is likely due to their water

40 fluctuation regime. Raynard's Lake had consistently high water levels for most of the growing season, while Wilson's Lake had greater fluctuations and longer periods of low water during the growing season. H. umbellata seems to prefer the latter conditions, as it is able survive flooding but grew better and sequestered more resources into below ground biomass when above the water line, as we saw in the increase in biomass of H. umbellata at 50 cm as compared to the deeper plantings (Figure 26). Kempt-Back and

Raynard's Lake both had consistent water levels through the growing season, though

Kempt-Back's water levels stayed around 30 cm below the high water mark while

Raynard's Lake stayed around 50 cm in the first year. Thus, earlier establishment could confer an advantage to H. umbellata if water levels fluctuate low enough during the growing season.

A secondary effect of week planted was a change in morphology for H. umbellata. The later transplants of this species had a lower final ratio of above to below ground biomass.

However, both the above and below ground biomass of week 0 plants were greater than those of week 4, which in turn were greater than those of week 8. Perhaps the larger size of the earlier plants allowed them to expend proportionally more energy on above ground material while still having a larger absolute below ground size because they had a larger energy budget.

Earlier establishment had a positive effect on survival of transplants in poor conditions.

Plants from all transplant dates at Kempt-Back Lake had poor growth, lower final biomass, and almost no flowering; however the older transplants there did have better

41 survival rates. Hydrocotyle umbellata planted earlier at Kempt-Back had a higher rate of survival through the fall storms in 2004 than the later plantings (week 0 survival > week

4 survival > week 8 survival). This was likely because plants that were not out for as long had a less developed root system and were more easily washed out. At Wilson's and

Raynard's Lakes, survival was so high that earlier establishment did not confer any noticeable advantage; time of establishment (week planted) only had an effect on survival at Kempt-Back Lake where water levels were always high (Figure 30), resulting in poor conditions (Figure 23). The flood stress at Kempt-Back Lake was a chronic source of damage, which when coupled with acute disturbance events in the form of fall storms

(wave damage) resulted in high mortality.

4.2 Effect of depth

Plants at higher positions on the shoreline had better growth, greater above, below and total biomass, and, in the case of C. rosea, flowered more. This is likely because they experienced less flood stress at these higher positions, were exposed to higher daytime temperatures, had better light access, and more efficient metabolism as they were above water more often. The effect of planting position on survival was not significant, likely because survival was generally very high, but some patterns were noteworthy. At

Kempt-Back Lake where plants were constantly flooded and conditions were the poorest, both species had higher survival during the fall (and winter for H. umbellata), at the upper shoreline positions where flooding depth was the smallest and the plants the least restricted and weakened by flood stress. The exception was C. rosea at 40 cm depth at

Kempt-Back, which had only 61% survival during the first summer while plants at other

42 positions were close to 100% survival. This was likely due to waves uprooting the plants

as they broke on the shoreline; the wave wash area would have been around 40 cm as the

water level was generally around 30 cm depth on Kempt-Back Lake.

The morphology of H. umbellata was affected by its position on the shoreline. It had a

higher ratio of above to below ground biomass at lower elevations. Plants at the lower

positions produced longer petioles and larger leaves in response to the greater depth of

flooding. Hydrocotyle umbellata leaves and petioles were easily broken and rarely lasted

longer than two weeks (personal observation). As a result, plants at lower shoreline

positions would have had a greater energetic expense when producing new leaves as the

petioles had to be much longer (sometimes over a meter) to reach the water surface. The

higher investment in leaves and petioles for plants at greater depths may have resulted in

less below ground investment, and lower total biomass at harvest. Conversely, plants at

the upper shoreline positions had a lower ratio of above to below ground biomass (Figure

27), and this proportionally lower investment in above ground biomass allowed plants to store more in their below ground structures. Liette Vasseur (personal communication) noted that H. umbellata at Kejimkujik Lake only produced below ground tuber-like starch rich deposits at nodes on its rhizomes in the uppermost shoreline positions where it was not flooded.

Our data show that these species were able to survive moderate levels of flooding, but grew better under emergent conditions. Wisheu and Keddy (1994) showed that the low competitive ability of the rare ACPF may limit the upper extent of their range on the

43 shoreline. The range these species occupy may not then be the optimal portion of the

shoreline, but rather bounded by where they can survive and not be out competed.

Though lower shoreline positions do not have shorter absolute growing seasons as the

plants begin to grow below water, the plants behave the same as plants transplanted at

later dates likely due to flooding stress which reduces growth rates and depletes

carbohydrate reserves. Extended flooding at lower shoreline positions hampered growth

and affected survival under poor conditions, as seen at Kempt-Back Lake. Reduced

duration of flooding at the lower shoreline positions would likely improve the vigour,

size and flowering rates of plants there.

4.3 Effect of lake

The water fluctuation regime at Raynard's Lake was better than Wilson's in the summer

as it had fewer floods, but worse than Wilson's Lake during the rest of the year for the lower shoreline species due to low fall and early winter water levels and longer high spring levels. Kempt-Back Lake was poor habitat due to high water levels during all the

seasons.

The water fluctuation regime at Kempt-Back was not favourable to either species, as the water levels were high and stable through both years. Though both species appear to be

able to tolerate a certain amount of flooding, the regime at Kempt-Back may be too

stressful for the plants as they must tolerate flooding through the entire growing season.

Due to poor growth during the summer, plants were small and easily washed away in fall

storms; overwintering survival was likely compromised by their small below ground

44 biomass. It is unlikely that rare ACPF dispersing there, or to a similar lake, could persist over a long period of time.

The water fluctuation regime at Raynard's Lake seemed to favour C. rosea, but not H. umbellata. In the first year H. umbellata did well at Raynard's Lake but came back more slowly in the second year and failed to grow as big as plants at Wilson's Lake in 2005.

The later start of plants at Raynard's Lake in 2005 and subsequent smaller size suggest that conditions outside of the growing season may be limiting for H. umbellata there. In freezing tolerance experiments (data not presented here), we found that H. umbellata had a much lower cold tolerance (-4oC) than C. rosea (>-12oC). Similarly, Hazel (2004) found that H. umbellata did not survive the winter in a raised bed garden experiment, while C. rosea did. Given that water levels on Raynard's Lake were such that these plants were exposed to air until January that year, it is possible that the plants were damaged by cold exposure. The smaller size of H. umbellata at Raynard's Lake in the spring of 2005 may also be due to deeper and longer flooding that spring relative to the year before, and also relative to Wilson's Lake. Deeper and longer flooding would result in lower light penetration, lower temperatures and greater flood stress (Crawford 1989), all of which could have slowed spring growth of H. umbellata at Raynard's Lake in spring 2005.

The higher ratio of above to below ground biomass for H. umbellata at Raynard's and

Kempt-Back Lakes was likely due to the high water levels at these lakes during the growing season. Similar to what I found with H. umbellata at lower shoreline positions,

45 H. umbellata must put proportionally more energy into above ground biomass when the

overall depth of flooding is greater.

At Raynard's Lake, C. rosea was higher on the shoreline and not flooded as deeply as H.

umbellata. It did not seem to be negatively affected by the cold exposure in the fall. It also did not seem strongly affected by the longer spring flooding except in its

morphology. The leaf to stem ratio biomass of C. rosea was lower at Raynard's and

Kempt-Back Lake than at Wilson's Lake, suggesting that C. rosea can elongate its

internode length in response to longer periods of flooding. Ruth Newell (personal

communication) noticed this same stem elongation in Rhexia virginica L. (meadow

beauty), another rare ACPF species, at Raynard's Lake on September 12, 2001. The

difference in above ground biomass between plants at Raynard's and Wilson's Lake may

in part be due to the stem elongation at Raynard's Lake. The difference in week 0

biomass between the two lakes is most likely due to the flood at Wilson's Lake that

damaged the week 0 transplants there, as the biomass of week 4 and 8 transplants were not different between Raynard's and Wilson's Lakes. Coreopsis rosea at Raynard's did

flower earlier in both years and had more buds per ramet than at the other lakes. The

plants at Raynard's Lake did not have a longer growing season, as plants at both lakes started growing in late April below water, and they were below water for longer in the spring than at Wilson's Lake, so perhaps the higher flowering rate at Raynard's Lake is due to some physical characteristic of the shoreline relating to fertility, such as higher organic matter, or due to the absence of mid season floods. It should be noted that despite the higher flowering rates at Raynard's Lake, the overall seed dispersal is not

46 likely to be high; many stems and seed heads are broken off before seeds have matured

by waves during fall storms. Raynard's Lake is deeper and has larger bays than Wilson's

Lake and so the fall storms were more destructive.

Overall, as it is managed at present, Raynard's Lake reservoir may be better habitat for upper shoreline species such as C. rosea than lower shoreline species like H. umbellata.

The presence of Rhexia virginica at Raynard's Lake supports this as does our discovery

of a few small patches of C. rosea near our study sites on the north-western portion of

Raynard's Lake early in the summer of 2004. Though these upper shoreline plants may

survive the cold and exposure in the early winters and longer flooding in the spring, and

do well in the summers, fall storms seem to damage flowering stalks before seed matures,

restricting dispersal by seed and persistence via the seed bank of these species there.

Earlier filling of the reservoir in the fall could allow less cold tolerant upper shoreline

species and lower shoreline species such as H. umbellata to grow at Raynard's Lake.

Earlier draw down in the spring could benefit all species there.

Wilson's Lake represents a combination of the right ranges of environmental variables to

allow a large number of diverse ACPF to have persisted over the long term, though it

does not always provide ideal conditions. The plants there appear to be subjected to

periodic growing season floods that are quite damaging, they were also affected by

erosion in the fall and floods in the spring.

47 5.0 Conclusion and applications

For most species at the northern-most extension of their range, length of the growing

season likely limits their growth and survival. At our study lakes in south-western Nova

Scotia, growing season length does appear to limit the rare ACPF. We found that a

longer growing season can increase growth substantially, but only if growth conditions

are favourable, and that later establishment can result in lower survival if conditions are

poor. Thus later establishment could certainly be important in limiting the spread and

persistence of these species. Lower positions on the shoreline did not directly equate to

shorter growing season, as we found that our study species were able to start growing

below water, but did hamper growth during the flooded portion of the growing season

with effects still apparent a year later. Plants at greater depths had reduced growth,

biomass and flowering. The importance of planting depth and timing of exposure

between lakes is likely due to the stress and lower growth rate associated with longer

flooding. Management of this watershed affecting timing and duration of flooding is

therefore very important to the success of these rare and at risk species.

Slight alterations to the management of reservoir lakes to reduce the length of spring

flooding by opening the dam at an earlier date, and opening the dam during periods of

high precipitation to reduce the impact of growing-season floods would improve the

vigour of plants there. Also, an earlier rise in water levels in the fall could allow less cold tolerant upper shoreline species, as well as lower shoreline species such as H. umbellata

to persist on reservoirs. Management of watersheds to reduce flood peaks would also

benefit these species; minimizing areas of impermeable surfaces, minimizing clear

48 cutting and land clearing for development or agriculture, enforcing the use of shoreline

vegetation buffers, and use of policies to reduce small water course alteration or infilling,

could help to retard and reduce land runoff after large precipitation events that cause the

extreme flood peaks that are damaging to these species. However, a certain amount of

natural water fluctuation is necessary to keep shrubs and other terrestrial plants from

encroaching down to the waterline. Complete stabilization of water levels would also be harmful to these plants as it would increase competition from other more aggressive species (Wisheu and Keddy 1994) and results in poor growing conditions as seen at

Kempt-Back Lake.

Conservation initiatives to date have focused on habitat maintenance; however altered management of dammed lakes may make it possible to restore or create ACPF habitat. If transplants were considered for such a restoration project, I would recommend larger

transplants (as big or bigger than what we used in our experiment) with well developed below ground reserves, as we have seen the importance of below ground material in helping plants survive flooded periods and keeping plants from being washed away.

Earlier transplanting, even if it meant planting material below water, would be better provided that the shoreline was appropriate and the lake's water fluctuation regime favourable.

The variables limiting C. rosea and H. umbellata and contributing to their rarity in Nova

Scotia, are likely also impacting other ACPF species along the shoreline, thus we

potentially improve ACPF habitat for all species in that community by managing lakes

49 and the watershed to accommodate the rarest plants' needs. The next step is to take what we have learned from C. rosea and see if other more common upper shoreline ACPF species have similar responses and mechanisms to deal with flooding and cold, and do the same with H. umbellata and other more common lower shoreline ACPF species; using our understanding of these study species to better understand and manage the

ACPF community as a whole.

50 Table 1. Results of ANOVA analyses for H. umbellata on each measurement day; values are sums of squares for each term, significant terms are highlighted. In the day of year row, the first four numbers of the number are the year, and the last three digits are the Julian day of the year out of 365. Analyses from day 2004261 on only include Raynard's and Wilson's Lake.

2004189 2004204 2004217 2004231 2004244 2004261 2005192 2005230 2005230 2005230 2005230 2005230 2005230 (July 7, (July 22, (Aug 4, (Aug 18, (Aug 31, (Sept 17, (July 11, (Aug 18, (Aug 18, (Aug 18, (Aug 18, (Aug 18, (Aug 18, Day of year 2004) 2004) 2004) 2004) 2004) 2004) 2005) 2005) 2005) 2005) 2005) 2005) 2005) Above/ Ln (# lv Ln (# lv Ln (# lv Ln (# lv Ln (# lv Ln (# lv Ln (# lv Ln (# lv Above Below Below Leaf / * max * max * max * max * max * max * max * max Total ground ground ground Stem Variable diam) diam) diam) diam) diam) diam) diam) diam) biomass biomass biomass biomass biomass Lake 7.77 10.39 5.10 6.59 8.90 27.31 58.81 0.07 0.023 0.0004 0.017 5.88 5.09

Depth 1.68 3.20 6.05 9.06 5.62 8.76 8.24 14.30 0.090 0.0077 0.048 6.38 11.50 Week 0.42 2.22 9.80 4.51 1.30 3.59 0.038 0.0056 0.015 1.23 1.15 Lake*Week 6.46 0.30 3.83 2.20 6.32 7.33 0.039 0.0061 0.015 0.73 2.25 Lake*Depth 4.87 2.43 3.88 5.62 4.48 3.82 3.05 6.32 0.005 0.0020 0.001 0.28 4.63 Week*Depth 5.29 5.03 8.68 4.12 5.98 3.97 0.030 0.0032 0.015 2.11 3.58 Lake* Week* Depth 3.64 3.38 10.15 9.09 7.12 4.30 0.008 0.0015 0.004 1.56 3.74 Error SS 11.89 20.01 58.15 71.82 119.06 69.39 78.19 70.54 0.21 0.037 0.08 18.78 45.38 Total SS 26.05 38.07 88.47 103.04 169.97 128.92 161.60 109.42 0.43 0.063 0.19 35.91 80.38 r2 0.54 0.47 0.34 0.30 0.30 0.46 0.52 0.36 0.52 0.42 0.56 0.48 0.44

Significance Level ≤ 0.0001 ≤ 0.01 ≤ 0.05

51 Table 2. Results of ANOVA analyses for C. rosea on each measurement day; the values are sums of squares for each term, significant terms are highlighted. In the day of year row, the first four numbers of the number are the year, and the last three digits are the Julian day of the year out of 365. Analyses from day 2004261 on only include Raynard's and Wilson's Lake.

2004189 2004204 2004217 2004231 2004244 2004261 2005192 2005230 2005230 2005230 2005230 2005230 2005230 2005230 Day of (July 7, (July 22, (Aug 4, (Aug 18, (Aug 31, (Sept 17, (July 11, (Aug 18, (Aug 18, (Aug 18, (Aug 18, (Aug 18, (Aug 18, (Aug 18, year 2004) 2004) 2004) 2004) 2004) 2004) 2005) 2005) 2005) 2005) 2005) 2005) 2005) 2005) Above/ Above Below Below Leaf / Ln # Ln # Ln # Ln # Ln # Ln # Ln # Ln # Total ground ground ground Stem Variable Leaves Leaves Leaves Leaves Leaves Leaves Leaves Leaves biomass biomass biomass biomass biomass Flowering Lake 0.40 5.65 0.19 8.42 12.12 1.92 3.50 34.53 19.13 6.60 3.43 1.98 0.83 30.05 Depth 3.68 5.71 5.01 8.68 9.66 3.24 4.69 3.51 23.22 2.36 10.08 16.61 0.78 154.55 Week 6.27 0.89 28.03 15.40 8.39 5.94 13.94 1.81 5.61 2.78 0.10 13.51 Lake * Week 3.99 4.64 7.84 9.00 7.02 6.95 14.28 2.69 4.28 26.23 0.16 4.48 Lake * Depth 2.26 4.91 0.74 1.75 4.52 1.88 5.12 3.53 2.25 0.78 1.05 51.86 0.68 25.92 Week * Depth 0.53 0.25 2.65 1.13 1.67 2.64 4.71 0.73 2.59 70.00 1.79 16.13 Lake * Week * Depth 1.64 1.48 4.86 2.71 3.02 2.93 3.10 0.57 1.29 49.90 2.04 28.25 Error SS 22.52 22.76 33.62 32.85 55.12 36.60 46.87 32.60 89.25 19.23 32.51 1201.80 25.72 351.08 Total SS 28.80 37.14 52.30 58.45 128.09 71.16 78.60 91.69 163.92 34.29 58.22 1431.62 31.81 632.61 r2 0.22 0.39 0.36 0.59 0.57 0.49 0.40 0.64 0.46 0.44 0.44 0.16 0.19 0.45

Significance Level ≤ 0.0001 ≤ 0.01 ≤ 0.05

52 Table 3. Periods for survival analysis.

Period Season Dates Day of year

1 1st Summer June 26-Aug 30, 2004 2004178 - 2004245 2 Fall September 1 - November 27, 2004 2004245 - 2004333

3 Winter/ Spring November 28, 2004 - July 11, 2005 2004333 - 2005192

4 2nd Summer July 11 - August 18, 2005 2005192 - 2005230

53

Figure 11. Lifting and removing a portion of H. umbellata from the growth tray.

Figure 12. Hydocotyle umbellata in sieve showing root and rhizomes exposed; soil has been washed out into the base sieve.

54

Figure 13. Standardized cuttings of C. rosea ready to be treated then planted.

Figure 14. Treating the above ground portion of C. rosea transplants with 1:50 dilution

of Safer's insecticidal soap.

55

Stringo level 30 cm 40 cm Shrub line/ 50 cm High water 60 cm mark Water 70 cm 80 cm 85 cm

Figure 15. Measurement of vertical depths below the high water mark with a string level to establish transplant locations.

Figure 16. The flags in the image show the position of 30, 40, 50, 60, 70, 80, and 85 cm

vertically below the high water mark during a very dry period.

56

85 cm

80 cm

70 cm

60 cm

50 cm

40 cm

Figure 17. Arrangement of transplants: the first set of transplants are indicated by the two outside rows of circles (yellow), the second round of transplants 4 weeks later are indicated by the next set of circles toward the middle (blue), and last transplants are indicated by the two innermost rows of circles (red). There are two circles at 50, 60 and

70 cm to show the overlap of C. rosea and H. umbellata planting elevations.

57 a.

b.

Figure 18. Transplants and tags: a. Coreopsis rosea, b. Hydrocotyle umbellata.

Figure 19. Transplanting specimens under water.

58 Figure 20. Effect of week planted on growth measures (means +/- one standard error)

for C. rosea and H. umbellata over time by lake. The growth index for C. rosea was Ln

(# of leaves) and for H. umbellata it was Ln (# of leaves * maximum leaf diameter). The

first four digits of the day of the year on the x-axis are the year and the last three are the

Julian day of the year out of 365. The break in the x-axis indicates the winter of 2004/5.

59 Wilson's Lake Raynard's Lake Kempt-Back Lake

Week 0 5 Week 4 Week 8 4

3 growth index index growth

2 C. rosea

1 5

4 growth index index growth

3

H. umbellata H. umbellata 2

2004 2 20 20 200 2005 200 2 2 20 200 200 200 2 2 20 20 200 0042 0042 0051 0042 0051 051 051 051 051 052 52 42 52 52 42 52 2 2 00 4 20 60 00 40 00 4 2 60 00 40 00 4 2 60 00 4 0 0 0 0 0 0 Day of Year

60 C. rosea H. umbellata

Kempt-Back 2.5 0.12 Raynard's 2.0 0.10 Wilson's 0.08 1.5 0.06 1.0 0.04

Total biomass (g) 0.5 0.02 0.0 0.00

2.5 0.12 2.0 0.10 0.08 1.5 0.06 1.0 0.04 0.5 0.02

Above groundAbove (g) biomass 0.0 0.00

2.5 0.12 2.0 0.10 0.08 1.5 0.06 1.0 0.04 0.5 0.02

Below ground biomass (g) 0.0 0.00

048 048 Week Planted

Figure 21. Total, above ground and below ground biomass (means +/- one standard error) of C. rosea and H. umbellata at the end of the two year experiment, by week planted and lake.

61 0.90

0.85

0.80

0.75

0.70

0.65 Above/Below ground biomass (g) biomass ground Above/Below

0.60

0.55 048

Week Planted

Figure 22. Effect of week planted on above/below ground biomass ratio of H. umbellata

(means +/- one standard error).

62 Week 0 Week 4 Week 8

100

80

survival 60

40 Raynard's C. rosea 20

% Wilson's 0 Kempt-Back

100

80 survival 60

40

20 H. umbellata H. umbellata % 0

1234 1234 1234

Period

Figure 23. Percent of transplants alive at the beginning of a period that survived to the end of that period (dates included in each period are shown in Table 3) by species, week planted and lake.

63 Day 2004189 Day 2004217 5

4

3 Kempt-Back Week 0 2 Raynard's Week 4 Wilson's 1 Day 2004231 Day 2004244 5

4

3

Week 0 2 All lakes Week 4

1 Raynard's Lake day 2004261 Wilson's Lake day 2004261 5

4

growth index: Ln (# leaves diameter) (# leaves index:* max Ln growth 3

Week 0 Week 0 2 Week 4 Week 4 Week 8 Week 8 1 H. umbellata Day 2005192 Day 2005230 5

4

3

2 Raynard's Raynard's and Wilson's Wilson's 1 50 60 70 80 50 60 70 80 Vertical position on shoreline (cm below high water mark/ shrubline)

Figure 24. Effect of depth on H. umbellata growth index (Ln (# leaves * maximum leaf diameter)) on different sampling dates, including interactions with lake or week when significant (means +/- one standard error). Graphs and analyses from day 2004261 on do not include Kempt-Back Lake because of small sample size.

64 5 Day 2004189 Day 2004204

4

3

2 All Lakes All Lakes

5 Day 2004217 Day 2004231

4

3

2 All Lakes All Lakes

5 Day 2004244 Day 2004261

4 growth index: Ln (# of leaves) (# Ln index: growth

3

C. rosea Kempt-Back Raynard's 2 Raynard's and Wilson's Wilson's

5 Raynard's Day 2005192 Wilson's Day 2005230 4

3 Raynard's 2 Wilson's

30 40 50 60 70 30 40 50 60 70 Vertical Height (cm below high water mark/shrub line)

Figure 25. Effect of depth on C. rosea growth index (Ln (# leaves)) over time, including interactions with lake when significant (means +/- one standard error). Graphs and analyses from day 2004261 on do not include Kempt-Back Lake because of small sample size.

65 Total biomass Above and below ground biomass

2.0 Above ground Below ground 1.5

1.0 Biomass (g) Biomass

0.5 C. rosea 0.0 30 40 50 60 70 30 40 50 60 70

0.12

0.10

0.08 Biomass (g) 0.06

0.04

0.02 H. umbellata umbellata H. 0.00 50 60 70 80 50 60 70 80 Vertical possition on shoreline (cm below high water mark/shrubline)

Figure 26. Effect of planting depth on total, above and below ground biomass (means

+/- one standard error).

66 1.1

1.0

0.9

0.8

0.7

0.6 Ratio of above/below ground ground biomass of above/below Ratio

0.5

0.4 50 60 70 80 90

Vertical position on shoreline (cm below high water mark)

Figure 27. Effect of planting depth on above to below ground biomass ratio for H. umbellata (means +/- one standard error).

67 Period 3 - Winter Period 1 - Summer 2004 Period 2 - Fall 2004 2004/5 and Spring 2005 Period 4 - Summer 2005

100 80

Survival 60

40 Raynard's . rosea 20 Wilson's

% C 0 Kempt-Back 30 40 50 60 70 30 40 50 60 70 30 40 50 60 70 30 40 50 60 70

100 80 Survival 60 40 20 H. umbellata 0 % 50 60 70 80 50 60 70 80 50 60 70 80 50 60 70 80

Vertical position on shoreline (cm below shrub line/high water mark)

Figure 28. Percent of transplants alive at the beginning of a period surviving to the end of that period (dates included in each period are shown in Table 3) by species, depth of planting and lake.

68 4

Kempt-Back Lake Raynard's Lake 3 Wilson's Lake

2

1 Average number of flowers

0

-1 30 40 50 60 70

Vertical position on shoreline (cm below shrub line/high water mark)

Figure 29. Effect of depth on the average number of flowers for C. rosea by lake (means

+/- one standard error) in 2005.

69 100

50

0

-50

-100

-150

-200

-250 Kempt-Back Water level relative to the high water to mark (cm) relative the Water level high water Raynards -300 Wilson's

-350 2004200 2004250 2004300 2005125 2005200 2005275

Day of year

Figure 30. Water levels relative to the high water mark. The first four digits of the day of the year on the x-axis are the year, and the last three are the Julian day of the year out of 365. The break in the x-axis indicates the winter of 2004/5.

70 1.2

1.0

0.8

0.6

0.4 Above/Below ground biomass (g) 0.2

0.0 Kempt-Back Raynard's Wilson's

Lake

Figure 31. Effect of lake on the above to below ground biomass ratio of H. umbellata

(means +/- one standard error).

71 0.5

0.4

0.3

0.2 Leaf/stem biomass ratio biomass Leaf/stem

0.1

0.0 Kempt-Back Raynard's Wilson's

Lake

Figure 32. Effect of lake on the leaf to stem biomass ratio of C. rosea (means +/- one standard error).

72 Chapter 3 - The effect of environmental variables

1.0 Introduction

Additional environmental variables may limit rare Atlantic Coastal Plain flora (ACPF) other than those associated with growing season length. In my first experiment on water level fluctuations and effect of growing season, the three treatments (establishment time, planting depth/duration of flooding, and lake/water fluctuation regime) accounted for only about 50% of the variation in final biomass of each study species. Similarly, a study in Northern Sweden found that the duration of flooding was the most important variable in predicting vegetation pattern and change in shoreline vegetation, but that only 41% of vegetation changes could be predicted by the water level fluctuations (Nilsson and Keddy

1988). What variables contribute to the remaining variation? This experiment was designed to explore additional environmental variables that may be affecting the growth of two rare ACPF, Hydrocotyle umbellata L. (water pennywort) and Coreopsis rosea

Nutt. (pink tickseed).

Growth and final biomass of transplanted specimens were measured at different depths along the shore within their natural range at three lakes, as with the first experiment.

Transplants were done from August 30th to September 5th, 2004, allowing time for establishment but ensuring that all transplants went into the winter with similar reserves and health. Environmental variables were measured at the plant level, shoreline elevation level and lake level. Vegetation cover, clast size distribution, soil depth, soil texture, and organic matter content were assessed for each plant. Soil temperature and moisture were measured at each depth at each lake. Water quality was measured at each lake. Sites

73 were chosen that had similar slope, aspect, wave and wind exposure, substrate and

vegetation.

These environmental variables were modeled to determine how variables other than those

associated with water level fluctuations and growing season affect the growth of the rare

ACPF on shorelines in south-western Nova Scotia. Knowing what environmental

variables limit our study species will help to indicate what components are essential to

good ACPF habitat. With this information, we should be able to choose the best areas of shoreline for protection, and understand how our activities affecting the physical environment impact the health and persistence of these at risk species.

2.0 Methods

2.1 Transplants

Transplants were prepared in the same manner and from the same material as described in section 2.1 of Chapter 2. The transplant slips for this experiment required longer lengths of rhizome than those for the first experiment as we were transplanting the material in the early fall and I wanted to ensure that transplants would have sufficient reserves to be in good health going into the winter. The transplant slips for H. umbellata consisted of a 30 cm length of rhizome with 10 to 14 nodes and 3 to 5 leaves. Roots were trimmed to a maximum length of 1 cm. The transplant slips for C. rosea consisted of a

10 cm length of rhizome with one ramet that was 1 to 4 cm in height. Roots were trimmed to a maximum length of 3 cm. These transplants were prepared July 27-28th,

2004 for H. umbellata and August 9th, 2004 for C. rosea. Once cut, the slips were treated

74 with Safer's insecticidal soap to kill any pests present and planted in a mixture of 2 parts

peat based potting soil to 3 parts sand as with the other transplants. To accommodate the

larger size of these transplants, the plug cell size was 5.5 cm deep, 4.5 by 4 cm plugs for

H. umbellata and 5.5 cm deep, 4.5 by 6 cm plugs for C. rosea.

Once prepared, the transplants for this experiment were placed outside in the K. C. Irving

Environmental Science Centre experimental garden and watered regularly. Transplanting was done in the field from August 30th to September 5th, 2004 in a grid adjacent to the first experiment transplants at the same depths along the shoreline (Figure 33). As in the first experiment, C. rosea was transplanted at 30, 40, 50, 60 and 70 cm below the high water mark and H. umbellata was transplanted at 50, 60, 70, 80 and 85 cm below the high water mark (these depths were chosen as the mid portion of the natural elevation range of these plants surveyed by Hazel (2004)). As with the first experiment, transplants were planted into 10 cm diameter plastic tubes (2 litre pop bottles) at Wilson's Lake to avoid confusion of the transplants with plants from the natural population. There were six replicate transplants at each of the five planting depths per site and three sites per lake; thus replication of 18 plants per planting depth at a given lake and 90 plants total at each

lake for each of the species. There were in total 540 transplants used in this experiment.

2.2 Plant and environmental variable measurement

Growth of the transplants was assessed July 10-13th and August 9-24th, 2005.

Measurements included maximum and cumulative height, leaf number, branch number

and ramet number for C. rosea, and leaf number, maximum height and maximum leaf

75 diameter for H. umbellata. The transplants were harvested from August 9th to 24th, 2005, cleaned then separated into leaf, stem and below ground components and oven dried for

48 hours before weighing.

Environmental variables that were measured for each transplant included vegetation cover, clast cover (boulder, stone, pebble, cobble and sand), organic soil cover, soil depth, soil texture, and soil organic matter content. The non-destructive measurements for each transplant were done within a 30 by 30 cm square plot around the transplant from July 21st to August 8th, 2005. Percent cover of all other species was estimated

within the plot. Percent cover of all clast sizes from boulder to sand, and surface organic

soil were estimated within the plot as well. Clast size groupings followed Morris (1994):

diameter ≥ 25 cm = boulder, 25 - 12.5 cm = large stone, 12.5 - 6.2 cm = medium stone,

6.2 - 1 cm = cobble, 1 - 0.2 cm = pebble and ≤ 0.1 cm = sand. A Dickey-John soil

compaction tester (Spectrum technologies, Plainfield, IL, USA) was used at each of the

corners of the 30 by 30 cm plot and at the center beside the transplant, to get an idea of

the depth of soil and stoniness in a third dimension of the rooting area available to the transplants. An even pressure of 200 psi (seen by reading the scale on the probe) was applied for a count of 10 seconds, then the penetration depth was noted.

Soil texture was assessed for each plant from August 10th to August 24th as the plants

were harvested by hand texturing soil from around the plant roots using the Unified Soil

Classification System field classification technique for coarse-grained soils (USCS P13b

76 version 2, Appendix 2). Organic soils were classed as either straight organic soil or as an intermediate between well sorted sand and organic soil.

Three samples of each hand texture class per lake were brought back to the lab to assess the sand, silt and clay content of their fine fraction (sediments less than 2 mm diameter).

Particle size analysis of these samples was done using a modified Bouyoucos hydrometer method following Gavlak et al. (1994), adapted from Gee and Bauder (1986). I used this

method as it has a detection limit of 2% for sand, silt and clay, and is reproducible with

+/- 8%, rather than the traditional method by Bouyoucos (1962) which will give a 10% or greater error for the sand content. I treated the samples for organic matter with hydrogen peroxide (Day 1965) as some of the samples exceeded the 3.5% limit (Gavlak et al.

1994); organic matter interferes with sedimentation of individual particles as it binds the particles (Rowell 1994). I then dispersed the samples with 250 ml of 5% Calgon water

softener, a complex of sodium phosphate and sodium sesquicarbonate (Gee and Bauder

1986, Gavlak et al. 1994, Rowell 1994). A control column was also prepared by filling

one of the sedimentation columns with the dispersing solution. This control was

measured at the same time as the soil/dispersing solution mixture so that the

measurements could be adjusted to remove the effect of the dispersing solution on the

hydrometer readings. Following Gavlak et al. (1994) and Loveland and Whalley (2000) I

used amyl alcohol to reduce the amount of foaming while mixing. The soil and

dispersing solution were mixed with a plunger for 15 strokes then allowed to settle. At

40 seconds, after the sand had settled, a hydrometer reading was taken for all

sedimentation columns (Rsand), and the control (Rc1). Settlement rate for fine particles is

77 noticeably affected by temperature, so the time of the second reading had to be adjusted

for temperature of the solution (Gee and Bauder 1986). The second readings for the

sedimentation column (Rclay) and control (Rc2) were taken at 7 hours and 45 minutes as

the solution was at 20°C (correction for other temperatures found in Gee and Bauder

1986). Percent sand, silt and clay were calculated with the following formulae:

% sand = (((oven dry soil mass) - (Rsand-Rc1)) / (oven dry soil mass)) * 100

% clay = ((Rclay-Rc2) / (oven dry soil mass)) * 100

% silt = 100 - (% sand) - (% clay)

To assess soil organic matter content, each soil sample was assigned a colour value from

2 to 5 based on the 10YR Munsel colour chart, under the assumption that greater soil

organic content resulted in darker soil. Four samples from each colour value designation

were taken from each lake and brought back to the lab. These samples were sieved with

a # 10, 2 mm sieve, then oven dried overnight at 105°C to remove any residual moisture.

Organic matter content was assessed through the loss on ignition method (Heiri et al.

2001); 40 g of this oven dried soil was measured and transferred to crucibles, then left in

a muffle furnace at 500°C for 4 hours then reweighed.

Measurements at each depth increment below the high water mark included soil moisture

and soil temperature; these were measured at the central site at each lake. The soil

moisture sensors (Watermark soil moisture sensors, Spectrum Technologies, Plainfield,

IL, USA) were installed at the same vertical position on the shoreline as the transplants:

78 30, 40, 50, 60, 70, 80, and 85 cm below the high water mark in late June 2004. Soil

moisture sensors were installed vertically so that the top of the sensor was 2 cm below the

soil surface. A hole was created just larger than the sensor; then a slurry of soil and water

poured in to ensure good contact between the soil and sensor. The sensor was inserted

into the hole, the extra slurry wiped away, and the top covered with soil to avoid

preferential drainage into the sensor area. In the fall (October 28-29th, 2004) as water

levels rose, temperature sensors (External Soil Temperature Sensor, Spectrum

Technologies, Plainfield, IL, USA) were installed 1 cm below the surface, and soil

moisture sensors removed below water level as soil moisture was unlikely to be at a

deficit and soil temperature became a more important variable for plants. Information

from all sensors was relayed to WatchDog Dataloggers (Spectrum Technologies,

Plainfield, IL, USA) installed within radiation shields on poles along the transects (Figure

33), and downloaded at two week intervals.

Water quality was measured at each lake on a monthly basis from early May until August

2005. Light penetration was taken with a secchi disk by lowering it into the water and

recording the depth when it disappeared, then raising the disk and recording the depth when the disk reappeared. This was repeated three times and an average taken. A 1:1 composite water sample was taken from 0.5 m depth and from twice the secchi disk depth

(and at least 1 m off the lake bottom to avoid disturbing bottom sediments). Water samples were kept cool and dark, and brought back to the lab within 24 hours.

79 In the lab, the water samples were processed for chlorophyll a, pH, alkalinity and colour.

For the chlorophyll a extraction, one litre of the composite sample from each lake was

vacuum filtered through a Whatman glass fiber filter. The filters and residue from the

water sample were transferred to vials containing 20 ml of 90% acetone solution and left

refrigerated in the dark for 24 hours. The samples were then transferred to centrifuge

tubes and centrifuged for 5 minutes. The liquid portion was poured into glass cuvettes

and mounted into the spectrophotometer slots. Absorbance was measured at 665 nm and

750 nm, acidified with 3 drops of 1 N HCl then measured again at the same wavelengths.

The concentration of chlorophyll a was calculated as:

Chlorophyll a (µg l-1) = (26.7*((E665 - E750) - (E665A - E750A)))*EV)/(VF*CL)

Where:

E665 = absorbance at 665 nm before acidification

E750 = absorbance at 750 nm before acidification

E665A = absorbance at 665 nm after acidification

E750A = absorbance at 750 nm after acidification

EV = elution volume (in ml), in this case 20 ml

VF = volume of sample filtered (in litres), in this case one litre

CL = cuvette length (in cm), in this case 5 cm

The pH of the water samples was measured in the lab with a pH meter at 20°C that had

first been calibrated with pH 7 solution, rinsed with distilled water then calibrated with

pH 4 solution, and rinsed again. Samples were gently stirred while readings were being

80 taken. Alkalinity of water samples was also measured in the lab by adding a Bromcresol

Green and Methyl Red indicator to 200 ml of composite sample water and titrating with standard sulfuric acid until the colour changed from green to pink. The amount of titrant used was noted and the alkalinity calculated from the formula:

Alkalinity (ppm Calcium Carbonate) = 5 * Standard Sulfuric Acid (ml) used in titration

Apparent colour was determined by comparing 50 ml of sample water in a Nessler tube with the set of standards by looking down the tubes toward the white surface, then noting the standard colour that most resembled the water sample. Total phosphorus was determined by the Capital District Health Authority Division of Environmental Services in Halifax using ultraviolet radiation in the presence of acid potassium persulphate

(standard method 4500PF).

2.3 Data analysis

All statistical analyses were performed using SAS for windows version 8.02 (SAS

Institute, Inc. Cary, NC, USA). Analyses were conducted separately for each species for growth indices measured July 10-13th, 2005 and August 9-24th, 2005, for final biomass,

and for flower number for C. rosea. A principal components analysis (FACTOR

procedure) was used to find uncorrelated, orthogonal axes made up of the environmental variables measured for each transplant. These factors were then used in a backward stepwise regression (STEPWISE procedure) to isolate which factors were important in

determining the size and final biomass of each species, and the flower number for C.

rosea. I followed the standard practice of allowing the stepwise regression to progress

81 until all variables included in the model were significant at the p<0.1 level and discussing factors significant at the p<0.05 level (Hocking 1976, Kennedy and Bancroft 1976, SAS

Institute Inc. 1985, Sokal and Rohlf 2000). The effect of lake on the size and final biomass of each species, and the flower number for C. rosea was analysed alone with an analysis of variance (GLM procedure), then the significant factors were analysed in an analysis of covariance with and without lake (GLM procedure) to determine the extent differences in size, and biomass for both species, and flower number for C. rosea, at different lakes could be accounted for by environmental variables. Growth indices were calculated through a correlation analysis (CORR procedure) of final growth measures with final biomass. The best index for H. umbellata growth was Ln (number of leaves * maximum leaf diameter) (r2=0.608 with the Ln of biomass), and the best index for C.

rosea Ln (number of leaves * maximum ramet height) (r2=0.588 with the Ln of biomass).

A fall storm in mid September 2005 killed the majority of transplants at Kempt-Back

Lake, only 14 H. umbellata and 4 C. rosea survived; the remaining plants at Kempt-Back were included in the analyses.

3.0 Results

3.1 Environmental variables

Loss on ignition results showed that soil organic values were not tightly correlated with soil colour across lakes. The organic soil category had on average 5.3% organic matter by

weight, and the sand/organic soil intermediate category had on average 3.7% organic

matter by weight; there was no differentiation among all the remaining mineral soil

categories, their average was 1.6% organic matter. These three values were assigned

82 back to each transplant based on their soil texture category. When averaged by lake,

mineral soils at Raynard's Lake had slightly higher average soil organic matter content

(2.49%) than Wilson's Lake (2.06%), followed by Kempt-Back Lake (1.08%).

Soil moisture was at saturation (0 kPa) for the entire summer of 2004 at all lakes. By the time the plants for this experiment were transplanted in early September 2004, the soil moisture levels were beginning to drop at Raynard's and Wilson's Lakes, though soils there were still quite moist (Figure 34). Soil moisture remained at saturation for the entire fall at Kempt-Back Lake as the water levels there were at 45 cm below high water mark or higher in the fall and also in the summer of 2005, with significant flooding in the spring until early July (Figure 30). Soil moisture at Wilson's Lake was at saturation for all depths greater than 40 cm below the high water mark for the entire fall, but experienced more drying at 30 cm. Plants at 30 cm at Wilson's Lake did not experience matric potentials as low as -30 kPa until November 2004, and then only briefly before they were flooded for the winter. Lake levels at Wilson's Lake were a bit lower in late

June 2005 but only because there was no mid season flood in 2005; the water level was at

65 cm but quickly returned to 40 cm below high water mark (Figure 30). Plants at

Raynard's Lake were all exposed to slight drying of the soil in the fall of 2004 after they were transplanted though soil potential never went below -30 kPa that fall (Figure 34).

Lake levels at Raynard's Lake were higher on average by 10-15 cm in the summer of

2005 than in 2004 (Figure 30), ranging between 30 and 50 cm below high water mark

(covering all the H. umbellata and all but the two upper most C. rosea plantings).

83 Precipitation values for 2005 were lower in July, but comparable to August 2004 (Table

4). Precipitation was above average in the summers of 2004 and 2005, below average for

September and October 2004, and above average in November and December 2004, in

part due to higher than average snowfall that November.

Soil temperature was measured in the fall of 2004 only at Raynard's and Wilson's Lake as

portions of those shorelines were still exposed to the air, while the Kempt-Back Lake

shoreline was submerged (Figure 30). The soil temperature profiles at the two lakes were

similar for exposed portions of the shoreline (Figure 35). All soil depths at Raynard's

experienced the same soil temperature conditions as they were all equally exposed

(Figure 35). All depths at Raynard's Lake continued to be exposed until early January

2005. You can see the effect of submersion as the 60 cm depth at Wilson's Lake was submerged in mid November, followed by the 50 cm depth a day or two later, then 40; the soil temperatures slowly go up after each depth was submerged (Figure 35). All heights were submerged at Wilson's Lake at the end of November (where Figure 35 ends), but the soil temperatures continued to drop at Raynard's Lake and reached 0°C before the plants were submerged in early January (Figure 36). Unfortunately, the logger malfunctioned and stopped collecting data before water levels rose at Raynard's Lake in early January 2005.

Air temperatures mostly oscillated around normal in the fall of 2004, but did drop below normal for the early-mid portion of November (the first below zero period of 2004) and at the end of January 2005 (Figure 37). Temperatures reached the extreme minimum

84 temperatures on record (1940-2005) on November 10th, November 16th, and in late

January after water levels had come up on all lakes (Figure 37). Minimum air

temperatures for December 2004 were well above the extreme temperatures seen in the

past for that period (Figure 37).

Chlorophyll a was low at all three lakes (<2.5 µg l-1 mean and <8 µg l-1 max Chlorophyll

a) (Figure 38). The total phosphorus was 37 µg L-1 at Wilson's Lake, 41 µg L-1 at

Raynard's Lake, and 20 µg L-1 at Kempt-Back Lake. The depth of light penetration was

low at Raynard's and Wilson's Lakes and deeper at Kempt-Back (Figure 39). This

difference was reflected in the water colour (Figure 40), Wilson's Lake's water was the

darkest, followed by Raynard's Lake, then Kempt-Back which was much lower than the

other two lakes. The lake with the lowest light penetration and darkest colour was also

the lake with the lowest pH (Figure 41); Wilson's Lake pH ranged from 3.35 to 3.7, while

Raynard's and Kempt-Back ranged from ~4.0 to 4.2. Conversely, the alkalinity was

lowest at Wilson's Lake, and higher at Raynard's and Kempt-Back Lakes (Figure 42).

The alkalinity at Raynard's Lake fluctuated during the course of the summer, unlike the

other two lakes, with a peak in July 2005.

3.2 Effect of environmental variables

Lake did not have a significant effect on the growth or biomass of H. umbellata

transplants, however, in a separate analysis, environmental variables summarized into

PCA factors did have an effect on growth and biomass of H. umbellata.

85 Early H. umbellata size, measured July 10-13th, 2005, was positively related to factor 7

(r2=0.0828, Table 5). Factor 7 was negatively correlated with the percent cover of

organic soil (-0.49) and positively correlated with percent cover of pebbles (0.44),

penetration depth near the transplant (0.44), and percent cover of large stones (0.30). The

remaining variables had relatively little effect on this factor. Growth early in the season,

therefore, was better in those areas with good rooting depth but with a surface cover of

pebbles or stones. The complete H. umbellata environmental factor loading patterns can

be found in Appendix 3.

When the late summer H. umbellata size was modeled against the environmental factors,

factors 1 and 2 negatively related to size, and factor 4 positively related to size

(r2=0.0674, Table 6). The factors more or less equally accounted for variation in H.

umbellata size (Table 6, sums of squares 3.8-4.3). Factor 1 was most strongly related to

the percent sand in the soil (-0.87), percent silt in the soil (0.86), percent cover of sand on

the surface (0.68), percent clay in the soil (0.61) and percent cover of large stones (-0.57).

Factor 2 was most strongly related to depth (0.72) and percent cover of cobbles (-0.72), and also related to the percent cover of sand (0.59). Factor 4 was most strongly related to

the percent organic matter in the soil (0.80), percent cover of organic soil on the surface

(0.69), and to a lesser extent, penetration depth (0.32). These results indicate that by the

end of the growing season, H. umbellata performed well in more than one type of site. A

large plant size was achieved on sites with a high sand content in the soil but with large

stones on the surface, as well as on sites with organic soils. Hydrocotyle umbellata also

did better at shallower depths.

86 Like the size index, plant biomass was positively related to factor 4 (r2=0.0851, Table 7);

factor 13 also contributed marginally to explaining variation in H. umbellata biomass but

was not significant at the p<0.05 level. As discussed above, factor 4 was most strongly

related to the percent organic matter in the soil (0.80), percent cover of organic soil on the

surface (0.69), and to penetration depth (0.32).

Lake did have a significant effect on the growth and biomass of C. rosea after the

spring/early summer, but generally once the environmental factors were analysed with

lake, a portion of the variation attributed to lake could be explained by the environmental

variables. At the depths where C. rosea was planted, Raynard's lake had greater organic

matter in the soil than the other two lakes (2.49% at Raynard's, 2.09% at Wilson's, and

1.08% at Kempt-Back Lake). Raynard's and Wilson's Lake were similar in their average soil texture (75-80% sand), which was finer than at Kempt-Back Lake (90% sand), and average soil depth (6.2-6.7 cm), which was deeper than Kempt-Back (3.0 cm). Raynard's

Lake had the lowest percent cover of pebbles (7.5%, vs. 10.8% at Wilson's and 13.9% at

Kempt-Back Lake).

Early C. rosea size, measured July 10-13th, 2005, was not affected by lake. In a separate

analysis, environmental variables summarized into PCA factors did have an effect on early growth. Only factors 1, 10, and 12 were significant at the p<0.05 level; all three were positively related to C. rosea size (r2=0.2138, Table 8). Factor 12 accounted for the

most variation (Table 8, sums of squares (SS) of 9.257), followed by factor 1 (SS of

8.427); factor 10 also contributed, though much less (SS of 4.539). Factor 12 was related

87 to percent clay in the soil (-0.21), percent organic matter in the soil (0.16) and percent silt

in the soil (0.10). Factor 1 was most strongly related to percent clay in the soil (0.92),

percent sand in the soil (-0.91), percent organic matter in the soil (0.80), average

penetration (0.72), and percent silt in the soil (0.67). Factor 10 was related again to the

average penetration (0.52). In summary, early season growth in C. rosea was better on

fairly deep soil with either high organic matter content or high content of fine soil

particles (clay or silt). The complete C. rosea environmental factor loading patterns can

be found in Appendix 4.

Differences in C. rosea growth among lakes developed by mid August 2005; the largest

plants were at Raynard's Lake, followed by Wilson's Lake, then Kempt-Back Lake. Lake

alone accounted for 34.6% of the variation in C. rosea growth in mid August (r2=0.3460); when the environmental factors were analysed with lake in a separate analysis, the model improved to 41.4% of the variance explained (r2=0.4143). The sum of squares attributed

to lake went down from 68.61 when it was added to the model first, to 21.01 when it was

added after the significant environmental factors, thus the environmental variables

explained 69.4% of the variation attributed to lake as well as an additional 6.8% of the

unassigned variation (when the r-squared value of the model with lake alone was

compared to the model with lake and environmental factors). When the late summer

plant size was modeled against the environmental factors alone, factors 2, 4, 6 and 12

were positively correlated with growth in mid August, and factor 5 was negatively

correlated (r2=0.3090, Table 9). Factor 2 explained the most variation (SS 24.55, Table

9), followed by factor 12, 4 and 5 which all explained about the same amount of variation

88 (SS 11.87, 9.81, 9.42 respectively). Factor 6 also contributed a small amount to the

explained variance (6.05). Factor 2 was related mainly to the percent cover of sand (-

0.89), but also to depth (-0.59) and the percent cover of cobbles (0.57). Factor 12 was

related to percent clay in the soil (-0.21), percent organic matter in the soil (0.16) and

percent silt in the soil (0.10). Factor 4 was most strongly related to the percent cover of

large stones (0.59) and the percent cover of boulders (0.46). Factor 5 was mainly related

to the percent cover of boulders (0.70), but also to the percent cover of medium stones (-

0.55). Factor 6 was related to the percent cover of pebbles (-0.48), depth (0.46), and

percent cover of cobbles (0.45). The net effect of these sometimes contrasting

correlations was that late season growth was better on fine textured soils (i.e. low sand

and high silt content) with high organic matter content at shallow depths. There was also

some indication that sites with a surface cover of cobbles or stones had better growth,

while cover of sand and pebbles had a negative effect on growth.

The differences in C. rosea size among the lakes translated into a difference in plant

biomass, with the largest plants again at Raynard's Lake followed by Wilson's then

Kempt-Back. Lake alone accounted for 5.5% of the variation in C. rosea biomass

(r2=0.055); when the significant environmental factors were analysed with lake in a

separate analysis, the model improved to 32.9% of the variance explained (r2=0.3288).

The sum of squares attributed to lake went down from 1.068 when it was added to the model first, to 0.475 when it was added after the environmental factors, thus the

environmental variables explained 55.5% of the variation attributed to lake as well as an

additional 27.3% of the unassigned variation (when the r-squared value of the model with

89 lake alone was compared to the model with lake and environmental factors). When C.

rosea biomass was modeled against the environmental factors alone, factors 1, 2, 8, 10

and 12 were significant (r2=0.3049, Table 10). Factors 1, 2, 10 and 12 were positively related to biomass and factor 8 was negatively related to biomass. Factors 2, 12, and 1 explained the most variation (Table 10, SS of 1.951, 1.687, and 1.346 respectively), followed by factor 10 (SS of 0.608) and factor 8 (SS of 0.457). Factor 2 was related mainly to the percent cover of sand (-0.89), but also to depth (-0.59) and the percent cover of cobbles (0.57). Factor 12 was related to percent clay in the soil (-0.21), percent organic matter in the soil (0.16) and percent silt in the soil (0.10). Factor 1 was most strongly related to percent clay in the soil (0.92), percent sand in the soil (-0.91), percent organic matter in the soil (0.80), average penetration (0.72), and percent silt in the soil

(0.67). Factor 10 was related to the average penetration (0.52). Factor 8 was most strongly related to percent cover of pebbles (0.46), percent cover of sand (-0.35), and organic matter in the soil (-0.32). In summary, final biomass was higher sites with a deep soil, fine texture (low sand and high silt/clay content), and high organic matter content at shallow depths. There was also some indication that sites with higher coverage of cobbles, and lower percent cover of pebbles and sand had greater biomass.

Coreopsis rosea flowering was not affected by lake. In an analysis with environmental factors alone, factors 1 and 2 were positively related and factor 6 negatively related to flowering (r2=0.1860, Table 11). Factor 11 also contributed marginally to explaining

variation in C. rosea flowering, but was not significant at the p<0.05 level. Factors 2 and

1 explained the most variation (Table 11, SS of 10.604 and 9.560 respectively), followed

90 by factor 6 (SS of 3.749). Factor 2 was related mainly to the percent cover of sand (-

0.89), but also to depth (-0.59) and the percent cover of cobbles (0.57). Factor 1 was most strongly related to percent clay in the soil (0.92), percent sand in the soil (-0.91), percent organic matter in the soil (0.80), average penetration (0.72), and percent silt in the soil (0.67). Factor 6 was related to the percent cover of pebbles (-0.48), depth (0.46) and the percent cover of cobbles (0.45). The net result of these correlations suggests that flowering was greater on sites with a deep soil, fine texture (low sand and high silt/clay content), and high organic matter content at shallow depths. There was also some indication that sites with lower percent cover of sand, and perhaps higher coverage of cobbles, had greater flowering.

4.0 Discussion

Lake was not significant in explaining variation in growth or biomass of H. umbellata in

2005, and environmental models explained only 6-9% of the total variation. Natural populations also exhibit a wide range of variability at a given height on the shoreline, and within a site (personal observation). When H. umbellata was flooded, the leaves had a short lifespan (personal observation); two plants under the same conditions may have had a different number of leaves based on whether or not old leaves had been shed. Flooded

H. umbellata also had a tendency to fragment and have portions float away (personal observation), which resulted in the remaining biomass of a plant that had lost a fragment being much smaller than an unfragmented plant grown under the same conditions.

91 Of the variation we could explain, differences in the growth and final biomass of H.

umbellata were influenced by the amount of organic matter in the soil, percent cover of

organic soil on the surface, and soil depth. Soil organic matter acts as a source of

macronutrients, stabilizes soil structure, increases water retention, darkens the soil

altering its thermal properties, increases cation exchange capacity, buffers soil pH, and

chelates metals and trace elements reducing the loss of soil micronutrients and toxicity of

metals, and enhancing the availability of phosphorus (Summer 2000). Hydrocotyle

umbellata has been investigated for its superior nutrient uptake relative to other

macrophytes, and for its reduction of nitrogen content of sewage (Moorhead and Reddy

1990, Reddy and De Busk 1985, Stephen and Tanner 1986, Ogwada et al. 1984).

Hydrocotyle umbellata was likely able to benefit from patches of higher organic matter as

a result of its high nutrient uptake efficiency, measurably increasing in size and biomass.

The results for the H. umbellata growth index in August suggested that both sites with high sand content in the soil and large stone cover, and sites with high organic matter produced large plants. The growth index only gave us information on what factors were affecting the above ground portion of the plants; plants at greater depth would have had a large above ground size, but could have had a low below ground biomass. Depth was

only significant for this late above ground measure of size, and was correlated with high

sand cover (which was in turn correlated with cover of large stones). The large size of

these plants at sites with high sand cover and large stones then could be attributed to the

large above ground size that H. umbellata had in response to flooding at greater depths.

This would explain why high sand sites were no longer significant for total biomass, as

92 plants at high organic matter sites had higher above and below ground biomass, but plants at sites with high sand cover would only have had high above ground biomass based on depth, and variable below ground biomass based on other variables such as organic matter content.

The distribution of organic matter in the soil was not affected by depth, and as organic matter content was the most important variable affecting H. umbellata biomass, this could explain why we did not see a depth effect on final H. umbellata biomass.

Lake was only important for C. rosea at harvest in August; lake did not effect early growth or flowering. None of the variation in size of C. rosea in early July was accounted for by lake, though 21% was accounted for by soil texture, soil depth, and organic matter content. Plants at sites with finer textures and higher organic matter content would have had access to more nutrients (Summer 2000) and may have developed larger reserves in the fall of 2005, as conditions were still warm in September and plants were actively growing until mid October. The transplants were all still submerged when the early growth measurements were taken, and thus were likely relying on these carbohydrate reserves in their rhizomes (Crawford 1989). If plants at spots with finer soil texture and higher organic matter had developed greater reserves going into the winter, then those plants may have been able to produce more above ground tissue in the spring. Soil texture, soil depth, and soil organic matter content were significant in explaining later growth, biomass and flowering as well, explaining most of the variation attributed to lake. As with H. umbellata, this suggests that C. rosea was nutrient limited.

93 There was also evidence that sites with higher cover of cobbles (and in one case stones),

and lower cover of sand and pebbles produced bigger C. rosea plants with more flowers.

As these variables were correlated with depth, it may have been that depth was more important, and the better, higher sites tended to have lower cover of sand and pebbles, and higher cover of cobbles and stones.

The negative relationship of C. rosea plant size, biomass and flowering rates with depth in this experiment reflected Chapter 2 results; plants lower on the shoreline experienced more flood stress, probably had reduced carbohydrate reserves as a result of depletion during periods of flooding, and a reduced growth rate when submerged. This difference had not evolved yet for the first growth measurement, as all plants had been flooded from the time they broke dormancy until that point, and depth on the shoreline had not yet resulted in a difference in conditions. Once spring flooding had receded, plants higher on the shoreline would have experienced less flood stress, which translated into greater biomass and more flowers for C. rosea in upper shoreline positions. In the fall of 2004, the upper shoreline plants were also above water for longer, particularly at Wilson's

Lake; plants at these higher positions could have had greater reserves going into the winter as they were flooded for a shorter period of time before going dormant in the fall

(Crawford 1989).

Environmental variables that we measured and determined not likely to be limiting during our study may in fact be limiting in other years, such as soil moisture, soil temperature and water quality.

94 The soil moisture sensor producers (Spectrum Technologies) suggest that crop plants should be watered when moisture sensors read -30 to -60 kPa. Coreopsis rosea and H. umbellata are likely slightly more sensitive to moisture stress than most mesophytic crop plants given that they are semi-aquatic (Gurevitch et al. 2002), as a result I have used the lower end of the suggested irrigation range as a conservative approximation of the lower end of safe (no water stress) moisture levels for these plants. Most plants can continue to extract water from the soil until the potential reaches -1500 kPa (permanent wilting point)

(Summer 2000). Soil moisture was not likely limiting for these species during the fall of

2004 and summer of 2005. Soils at Kempt-Back were saturated for the entire fall of

2004, soils at Raynard's were never below -30 kPa, and only soil at 30 cm at Wilson's

Lake experienced soil moisture potentials of less than -30 kPa briefly in November 2004, after the plants were dormant and thus were likely unaffected. Plants at Kempt-Back

Lake were likely saturated through the summer of 2005 as only the 30 cm position was occasionally above water. It is unlikely that plants at Raynard's Lake experienced moisture stress in the summer of 2005, given that the shoreline was saturated at the same time of year in 2004 with similar minimum precipitation (in August), and despite lower water levels in 2004. Plants at Wilson's Lake may have experienced slightly drier conditions in the summer of 2005 than 2004, but it is unlikely that they experienced moisture stress as the water table was close to or above plants for most of the summer and precipitation levels were similar to those in August 2004 (under which conditions soil moisture had been at saturation). Soil moisture was not likely below critical thresholds during the growing seasons study, but precipitation levels were also generally above average. Should there ever be a sequence of several very dry years, lake levels

95 could recede and soil moisture levels could drop to stressful levels, particularly in the upper shoreline positions.

Soil temperature could also become limiting in the fall and early summer at reservoir lakes if water levels were low for as long as they were in 2004. Raynard's Lake appeared to have been close to, but not at the damage threshold for H. umbellata (-4°C, data not presented here). Air temperatures in the fall of 2004 were far above the extreme lows recorded for those areas, particularly in December when air temperatures can go as low as -20°C. A late rise in water may have an impact on the health and survival of H. umbellata in other years if the weather was colder than it was in 2004. Allowing water levels to rise sooner in the fall at reservoirs such as Raynard's Lake would make these shorelines more suitable habitat for H. umbellata and other ACPF species that may be susceptible to cold and freezing damage.

The water quality measurements were used to assess the trophic level of the study lakes.

Based on the chlorophyll a criteria of the OECD (Organization for Economic Co-

Operation and Development) fixed trophic classification system (from Ryding and Rast

1989, after OECD 1982), all three lakes were oligotrophic. Eaton and Boates (2003) also found that most lakes on the Tusket watershed were oligotrophic in 2002. The total phosphorus was higher than the mean annual criteria of <4 µg L-1, as it was measured on

May 3rd, 2005 after spring turn over; the annual averages would be much lower. The secchi disk means were lower than the criteria (>6 m mean and >3 m max), but that is common in oligotrophic lakes high in organic acids which darken the water (Mike

96 Brylinsky, personal communication). Water quality was not likely limiting during this

summer but could become a problem if allowed to decline. Eaton and Boates (2003) cited lakeshore development as a potential risk to water quality in this area, in a report on the threat of water quality issues to the ACPF. In a research report for Kejimkujik Park,

Vasseur (2005) noted that water quality was a concern for H. umbellata there and should be monitored. Eutrophication of wetlands often results in alteration and simplification of the species composition and replacement of macrophytes with phytoplankton in extreme cases (Keddy 2000). Hydrocotyle umbellata may not persist if enrichment of the water column resulted in the establishment of more competitive macrophytes or increased phytoplankton. Even in the oligotrophic study lakes, damage was observed to both study species from algal loading as periphyton clung to the stems increasing drag, contributing to stem damage from waves, and more importantly, covering leaves and weighing down stems to the point of breaking them when water levels dropped (personal observation). If algal loading were increased due to higher nutrient levels in the water column, these problems could increase and become limiting.

5.0 Conclusion and applications

In general, the addition of environmental variables helped us to better understand the differences between lakes that were not related to growing season length and water fluctuation regime, and explained additional variation across lakes. However, there was still much variation left unexplained. Part of this unexplained variation may be inherent to the species. Other studies indicate that C. rosea and H. umbellata have low genetic

diversity in Nova Scotia (Vasseur 2005, and Wood and Good-Avila, unpublished data),

97 however we cannot rule out the possibility that different genotypes present may respond differently to the same conditions. A large source of variation between plants was likely related to their growth form and shedding of leaves, in particular in the case of H. umbellata as already discussed. The tendency of flooded H. umbellata to fragment and disperse (personal observation) would also contribute to differences in size between neighbouring plants. Coreopsis rosea was also seen to fragment and to root from broken off sections of stem (personal observation). We could better compare and measure the success of these species if we knew more about the timing and triggers of these forms of vegetative reproduction, as strict leaf counts and even biomass measurements are not capturing all of the plant's output. A study following the longevity of individual leaf and stem portions, and fate of fragments, would help us to estimate the importance of these processes and their contribution to the variation seen.

A study of the effects of herbivory, insect interaction (such as effect of egg-laying in stems observed for C. rosea), periphyton and algal loading of underwater plant material, damage (e.g. from waves), and strand line deposits may account for further unexplained variation. Given that these species demonstrated nutrient limitation in this study, the effect of below ground competition or mutualisms with other organisms could have important effects on growth and biomass. Mycorrhizae are relatively uncommon in wetlands, less than 50% of wetland species have mycorrhizae (Keddy 2000), declining to

1% in areas that are frequently flooded, but though they are uncommon they are not unheard of and could be present for the upper shoreline species such as C. rosea.

98 The limitation of these species primarily by soil organic matter and soil texture should

not be addressed with management, as altering these physical properties or altering their

associated fertility could change the plant community and have unwanted effects for the

rare ACPF. Conservation of a range of sites on the shoreline will ensure that nutrient rich

pockets will continue to exist in the otherwise nutrient poor habitat. It is important that

the majority of the shoreline remain nutrient poor and that the water quality remain good

so that ACPF continue to have shoreline and shallow water habitat where competition

from other herbaceous species and algae are low. Effects of cottages and development

could be mitigated by reducing shoreline alteration including compaction, removal of

organic matter and infilling with sand for beaches. As discussed, other variables such as

soil moisture, temperature, and water quality that were not found to be limiting in this

study, may be limiting in other years. Earlier filling of reservoirs in the fall could help

avoid fall temperature stress, and regulation of development and use of areas within the

watershed to reduce nutrient inputs into these lakes could protect water quality.

Knowledge of these current and potential environmental limitations can help guide our management and conservation decisions to the benefit of this plant community and its rare elements.

99 Table 4. Average precipitation (1971-2000) and actual precipitation in the summers of 2004-5, and during the fall of 2004 for

Yarmouth, Nova Scotia (20-30 km southwest of the study lakes) (from Environment Canada http://www.climate.weatheroffice. ec.gc.ca/climateData).

Jul Aug Sep Oct Nov Dec

Average Rainfall (mm) 84.5 74.4 99.1 107.9 123.2 98.7 Average Snowfall (cm) 0.0 0.0 0.0 1.6 7.3 42.9 Average Precipitation (mm) 84.5 74.4 99.1 109.6 129.9 134.7

2004 Rainfall (mm) 127.4 87.9 63.6 67.8 132.0 102.0 2004 Snowfall (cm) 0.0 0.0 0.0 0.0 62.5 52.2 2004 Precipitation (mm) 127.4 87.9 63.6 67.8 194.5 156.8

Difference between average 42.9 13.5 -35.5 -41.8 64.6 22.1 and 2004 precipitation

2005 Rainfall (mm) 85 85.8 2005 Snowfall (cm) 0.0 0.0 2005 Precipitation (mm) 85 85.8 Difference between average 0.5 11.4 and 2005 precipitation

100 Table 5. Significant factors in stepwise regression of H. umbellata early growth (Ln (#

of leaves * maximum diameter)), measured July 10-13th, 2005.

Sum of Mean DF Squares Square F Value Pr > F Model 1 26.623 26.623 16.79 <0.0001 Error 186 294.979 1.586 Corrected Total 187 321.602 R-Square 0.0828 n = 188

Parameter Standard Estimate Error Type II SS F Value Pr > F Intercept 3.676 0.092 2540.153 1601.70 <0.0001 Factor 7 0.377 0.092 26.623 16.79 <0.0001

Table 6. Significant factors in stepwise regression of H. umbellata size at harvest (Ln (# of leaves * maximum diameter), measured August 9-24th, 2005.

Sum of Mean DF Squares Square F Value Pr > F Model 3 11.925 3.975 4.43 0.0049 Error 184 164.949 0.896 Corrected Total 187 176.874 R-Square 0.0674 n = 188

Parameter Standard Estimate Error Type II SS F Value Pr > F Intercept 4.800 0.069 4330.858 4831.04 <0.0001 Factor 1 -0.142 0.069 3.777 4.21 0.0415 Factor 2 -0.152 0.069 4.302 4.80 0.0297 Factor 4 0.143 0.069 3.846 4.29 0.0397

101 Table 7. Significant factors in stepwise regression of H. umbellata final biomass.

Sum of Mean DF Squares Square F Value Pr > F Model 2 0.168 0.084 8.60 0.0003 Error 185 1.808 0.010 Corrected Total 187 1.976 R-Square 0.0851 n = 188

Parameter Standard Estimate Error Type II SS F Value Pr > F Intercept 0.159 0.00721 4.776 488.75 <0.0001 Factor 4 0.026 0.00723 0.131 13.43 0.0003 Factor 13 0.014 0.00723 0.037 3.77 0.0536

Table 8. Significant factors in stepwise regression of C. rosea early growth (Ln (# of leaves * maximum height)), measured July 10-13th, 2005.

Sum of Mean DF Squares Square F Value Pr > F Model 6 28.742 4.790 7.07 <0.0001 Error 156 105.688 0.677 Corrected Total 162 134.430 R-Square 0.2138 n = 163

Parameter Standard Estimate Error Type II SS F Value Pr > F Intercept 7.768 0.065 9750.396 14392.00 <0.0001 Factor 1 0.229 0.065 8.427 12.44 0.0006 Factor 5 -0.110 0.064 2.000 2.95 0.0878 Factor 8 -0.115 0.065 2.098 3.10 0.0804 Factor 10 0.169 0.065 4.539 6.70 0.0106 Factor 11 0.121 0.064 2.409 3.56 0.0612 Factor 12 0.246 0.067 9.257 13.66 0.0003

102 Table 9. Significant factors in stepwise regression of C. rosea size at harvest (Ln (# of leaves * maximum height), measured August 9-24th, 2005.

Sum of Mean DF Squares Square F Value Pr > F Model 5 61.678 12.336 14.67 <0.0001 Error 164 137.906 0.841 Corrected Total 169 199.584 R-Square 0.3090 n = 170

Parameter Standard Estimate Error Type II SS F Value Pr > F Intercept 9.227 0.070 14473.000 17211.50 <0.0001 Factor 2 0.381 0.071 24.546 29.19 <0.0001 Factor 4 0.241 0.071 9.805 11.66 0.0008 Factor 5 -0.236 0.071 9.415 11.20 0.0010 Factor 6 0.189 0.071 6.045 7.19 0.0081 Factor 12 0.265 0.071 11.867 14.11 0.0002

Table 10. Significant factors in stepwise regression of C. rosea final biomass.

Sum of Mean DF Squares Square F Value Pr > F Model 5 6.049 1.210 14.39 <0.0001 Error 164 13.792 0.084 Corrected Total 169 19.841 R-Square 0.3049 n = 170

Parameter Standard Estimate Error Type II SS F Value Pr > F Intercept 0.572 0.022 55.551 660.55 <0.0001 Factor 1 0.089 0.022 1.346 16.00 <0.0001 Factor 2 0.107 0.022 1.951 23.20 <0.0001 Factor 8 -0.052 0.022 0.457 5.44 0.0209 Factor 10 0.060 0.022 0.608 7.23 0.0079 Factor 12 0.100 0.022 1.687 20.06 <0.0001

103 Table 11. Significant factors in stepwise regression of number of flowers for C. rosea from 2005.

Sum of Mean DF Squares Square F Value Pr > F Model 4 26.136 6.534 9.43 <0.0001 Error 165 114.364 0.693 Corrected Total 169 140.500 R-Square 0.1860 n = 170

Parameter Standard Estimate Error Type II SS F Value Pr > F Intercept 0.500 0.064 42.500 61.32 <.0001 Factor 1 0.238 0.064 9.560 13.79 0.0003 Factor 2 0.250 0.064 10.604 15.30 0.0001 Factor 6 -0.149 0.064 3.749 5.41 0.0213 Factor 11 -0.115 0.064 2.223 3.21 0.0751

104

Figure 33. Environmental variable experiment transplants are marked with flagging tape on the far side of the canoe, further in the background are the growing season experiment transplants and environmental sensors on Kempt-Back Lake. The white boxes house the data loggers which received information from the soil temperature and moisture probes.

105 -50 30 cm 40 cm 50 cm and deeper -40 -30 -20 -10 Wilson's Lake

Soil Moisture (kPa) Moisture Soil 0

Jul Aug Sep Oct Nov Dec Jul Aug Sep Oct Nov Dec Jul Aug Sep Oct Nov Dec

-50 30 cm and deeper -40 -30 -20 -10 Kempt-Back Lake Kempt-Back Soil Moisture (kPa) Moisture Soil 0

Jul Aug Sep Oct Nov

-50 30 cm 40 cm 70 cm 85 cm -40 -30 -20 -10 Raynard's Lake

Soil Moisture (kPa) Moisture Soil 0

Jul Aug Sep Oct Nov Dec Jul Aug Sep Oct Nov Dec Jul Aug Sep Oct Nov Dec Jul Aug Sep Oct Nov Dec

Figure 34. Soil moisture at all lakes at various depths below the high water mark in the summer, fall and early winter of 2004.

106 Raynard's Lake Wilson's Lake 15 30 15 30 10 10

5 5

0 0

-5 -5 15 40 15 40 10 10

5 5

0 0

-5 -5 15 50 15 50 10 10

5 5

0 0

-5 -5 C) o 15 60 15 60 10 10

5 5

0 0

Soil Temperature( -5 -5 15 70 Nov-1 Nov-8 Nov-15 Nov-22 Nov-29 10

5

0

-5 15 80 10

5

0

-5 15 85 10

5

0

-5 Nov-1 Nov-8 Nov-15 Nov-22 Nov-29

Figure 35. Soil temperatures at Raynard's and Wilson's Lakes in November 2004.

107 20

10 C)

o 0

-10 Temperature (

-20

Air Temperature Soil Temperature -30 Nov 29 Dec 13 Dec 27 Jan 10 Jan 24 Feb 7

Figure 36. Soil and air temperatures at Raynard's Lake in the fall and winter of 2004/5.

108

Figure 37. Actual (the top continuous blue line), extreme and normal temperature minimums for November 2004 to January 2005 for

Yarmouth, Nova Scotia (20-30 km southwest of the study lakes) (from Environment Canada http://www.climate.weatheroffice.ec. gc.ca/climateData).

109 0.010 Kempt-Back Raynard's Wilson's 0.008

0.006

0.004 Chlorophyll a (ug/L) Chlorophyll 0.002

0.000

May Jun Jul Aug

Figure 38. Chlorophyll a concentration at the three experimental lakes (mean +/- one standard error) in the summer of 2005.

4.5 Kempt-Back 4.0 Raynard's Wilson's 3.5

3.0

2.5

2.0

1.5

Average secchi disk depth (m) 1.0

0.5 May Jun Jul Aug

Figure 39. Depth of light penetration (mean +/- one standard error) in the summer of

2005.

110 100

80

60

40 Water colour

20 Kempt-Back Lake Raynard's Lake Wilson's Lake 0 May Jun Jul Aug

Figure 40. Apparent water colour at the three experimental lakes (0 is the lightest, 100 is the darkest) in the summer of 2005.

111 4.4

4.2

4.0

3.8

Average pH 3.6

3.4 Kempt-Back Raynard's Wilson's 3.2 May Jun Jul Aug

Figure 41. Water pH at the three experimental lakes in the summer of 2005.

) 10 - 1 Kempt-Back 8 Raynard's Wilson's 6

concentration mg l 4 3

2

0 Alkalinity (CaCO May Jun Jul Aug

Figure 42. Water alkalinity at the three experimental lakes in the summer of 2005.

112 General Conclusion

The results of this study show that establishment time, flooding, and nutrient levels limit

the growth and survival of rare ACPF in Nova Scotia. In Chapter 2, I found that significant growth did occur below water in the spring while plants were flooded, and as such the species that I studied, C. rosea and H. umbellata, were not limited by actual number of growing days, but rather were hampered by greater energetic costs of growth and maintenance of tissue, and likely reduced photosynthetic ability, under prolonged

flooded conditions. As a result, the effect of prolonged or unseasonable flooding was

similar to the effect of later establishment in reducing biomass and flowering. Prolonged

flooding changed the morphology of H. umbellata; plants that were flooded for longer

had higher above to below ground ratios. Under constant flooding, plants of both species had greatly reduced below ground investment, were poorly anchored and highly susceptible to being washed out in fall storms. Results from Chapter 3 suggest that nutrient availability and soil depth limit the growth of both species. Depth on the shoreline was also important for the final size of both species, and flowering of C. rosea.

The total amount of variation that was accounted for by environmental variables was low

(~8% for H. umbellata and ~30% for C. rosea), in part that may be due to the inherent variability in the study species. The remaining variability may also be due to variables not quantified such as the effect of herbivory, wave damage, algal loading, and below ground interactions.

The survival rate of the transplants was a good overall indicator of site suitability.

Transplants at Raynard's and Wilson's Lakes had very low mortality, while there was

113 almost complete mortality at Kempt-Back Lake, where flooding was constant. Raynard's

Lake may provide suitable habitat at present for the upper shoreline species, as is

supported by our discovery of natural C. rosea patches there, and with slight alterations

to timing of flooding, could potentially support rare lower shoreline ACPF species as well. Dammed lakes have, in the past, represented the loss of ACPF habitat, but with

altered management of reservoir lakes, they could provide the opportunity to restore and

create habitat. The success of the two rare species at Kempt-Back Lake while it was

being managed prior to 2001 demonstrates that this is possible to support rare ACPF at

dammed lakes.

The success of the transplant experiments has shown that these rare ACPF species could

do well at other locations, at least in the short-term, and suggest that part of the cause of

the rarity of these species is related to dispersal limitation. Dispersal limitation is not

surprising given the low rate of sexual reproduction in the two study species, the

generally limited range of dispersal by vegetative reproduction, and low probability a

fragment settling in appropriate habitat and successfully establishing. Appropriate

habitat appears also to be limited. The results from this study further suggest that habitat

offering appropriate environmental conditions is itself rare, as the ACPF in our study

demonstrated limitation by several environmental variables during a period of fairly

average climatic conditions. The appropriate combination of suitable water fluctuation

regime and range of physical environmental attributes that would allow a population to

persist through extreme climatic occurrences and stochastic events, such as the Wilson's

Lake reserve shoreline, must be quite rare. The loss of nearly all the transplants at

114 Kempt-Back Lake showed that in less than ideal environments, a single event could wipe

out the an entire population.

Conservation and stewardship of present habitat should remain our main routes of action,

as we are still far from understanding all the variables that influence the long-term

success and persistence of rare ACPF. Nova Scotia is considered to have some of the last relatively undisturbed ACPF habitat (Sweeney and Ogilvie 1993). The results of this study support the conclusion that this habitat is truly rare, and reinforce that it is a unique

and important component of Nova Scotia's biodiversity. Understanding the variables that limited the rare ACPF species in this study should help us to better manage the shoreline, and mitigate potential impacts from human use and past alteration of the hydrologic regime.

115 Appendix 1. Aspect of transplant transects.

Aspect (in Lake Site degrees)

Wilsons A 122 Wilsons B 122 Wilsons C 130 Raynards A 85 Raynards B 123 Raynards C 139 Kempt-Back A 82 Kempt-Back B 90 Kempt-Back C 85

116 Appendix 2. The Unified Soil Classification System field classification technique for coarse-grained soils (USCS P13b version 2) for hand texturing soil in the field.

117 118 119 120 121 122 Appendix 3. The loading pattern of PCA factors that summarize the environmental variables measured for each H. umbellata transplant; values closer to one indicate higher correlation of that environmental parameter with the factor.

Factor 1 2 3 4 5 6 7 8 9 10 11 12 13 14

Depth -0.17 0.72 0.16 -0.17 -0.26 0.28 -0.03 -0.23 0.04 -0.08 0.41 0.12 0.00 0.0000 % Cover of boulders -0.33 0.13 -0.22 -0.04 0.41 0.68 -0.04 0.36 -0.21 0.14 -0.01 0.06 0.01 0.0000 % Cover of large stones -0.57 0.23 0.29 0.08 0.43 0.10 0.30 0.00 0.41 -0.27 -0.13 0.00 0.03 0.0000 % Cover of medium stones -0.47 -0.18 0.45 0.00 0.46 -0.43 0.07 0.00 -0.22 0.16 0.26 0.02 0.04 0.0000 % Cover of cobbles -0.31 -0.72 -0.03 -0.05 -0.44 0.10 -0.18 0.21 0.02 -0.28 0.09 0.11 0.06 0.0000 % Cover of pebbles 0.27 -0.49 -0.52 0.02 -0.02 0.15 0.44 -0.16 0.25 0.27 0.18 0.00 0.02 0.0000 % Cover of sand 0.68 0.59 -0.35 -0.11 0.00 -0.09 -0.05 -0.08 -0.10 -0.03 -0.11 -0.10 0.09 0.0000 Penetration depth near plant 0.23 0.39 0.17 0.32 -0.40 -0.16 0.44 0.53 -0.03 0.03 0.06 -0.01 0.00 0.0000 % Cover vegetation -0.28 0.00 0.66 0.25 -0.41 0.26 -0.01 -0.21 0.01 0.31 -0.23 0.04 0.03 0.0000 % Cover organic soil 0.29 0.11 -0.05 0.69 0.17 -0.07 -0.49 0.15 0.32 0.13 0.13 0.03 0.01 0.0000 % Sand in the soil -0.87 0.13 -0.35 0.17 -0.16 -0.07 -0.03 -0.03 -0.03 0.03 0.01 -0.18 0.00 0.0019 % Silt in the soil 0.86 -0.10 0.32 -0.24 0.16 0.02 0.02 0.05 0.05 0.00 -0.03 0.24 0.00 0.0017 % Clay in the soil 0.61 -0.29 0.49 0.07 0.08 0.34 0.00 -0.01 -0.04 -0.11 0.16 -0.38 0.00 0.0002 % Organic matter in the soil 0.20 -0.10 -0.15 0.80 0.11 0.11 0.19 -0.28 -0.30 -0.22 -0.01 0.13 0.00 0.0001

123 Appendix 4. The loading pattern of PCA factors that summarize the environmental variables measured for each C. rosea transplant; values closer to one indicate higher correlation of that environmental parameter with the factor.

Factor 1 2 3 4 5 6 7 8 9 10 11 12 13 14

Depth -0.41 -0.59 -0.12 -0.02 -0.16 0.46 -0.01 0.20 0.27 0.11 0.32 0.01 0.00 0.0000 % Cover of boulders -0.06 0.10 -0.07 0.46 0.70 0.19 0.48 -0.06 0.02 0.10 0.03 0.00 0.01 0.0000 % Cover of large stones -0.26 0.29 -0.15 0.59 0.16 -0.20 -0.56 0.21 0.19 0.10 -0.04 0.00 0.03 0.0000 % Cover of medium stones -0.27 0.37 -0.40 0.33 -0.55 -0.21 0.37 -0.12 -0.01 0.03 0.12 0.01 0.05 0.0000 % Cover of cobbles -0.11 0.57 0.63 -0.09 0.00 0.45 -0.07 0.03 -0.01 -0.20 0.02 0.01 0.07 0.0000 % Cover of pebbles -0.07 -0.17 0.62 -0.15 0.03 -0.48 0.30 0.46 0.11 0.08 0.00 0.01 0.02 0.0000 % Cover of sand -0.09 -0.89 0.14 0.01 0.06 -0.08 -0.10 -0.35 -0.01 0.09 -0.15 -0.02 0.07 0.0000 Average corner penetration 0.72 0.24 0.16 -0.03 -0.18 0.24 -0.04 0.06 -0.16 0.52 -0.07 0.02 0.00 0.0000 % Cover vegetation 0.53 0.23 -0.47 -0.39 0.03 0.10 0.14 0.05 0.44 -0.01 -0.25 -0.01 0.02 0.0000 % Cover organic soil 0.48 -0.03 -0.58 -0.37 0.31 -0.10 -0.11 0.24 -0.28 -0.03 0.19 0.01 0.05 0.0000 % Sand in the soil -0.91 0.21 -0.06 -0.30 0.10 -0.05 -0.03 -0.08 -0.01 0.12 -0.01 0.01 0.00 0.0013 % Silt in the soil 0.67 -0.40 -0.09 0.41 -0.17 0.18 0.09 0.26 -0.09 -0.18 -0.16 0.10 0.00 0.0007 % Clay in the soil 0.92 0.00 0.18 0.16 -0.04 -0.06 -0.03 -0.07 0.08 -0.05 0.16 -0.21 0.00 0.0006 % Organic matter in the soil 0.80 0.11 0.22 -0.05 0.11 -0.22 -0.08 -0.32 0.20 0.00 0.24 0.16 0.00 0.0002

124 References

Boates, J. S., S. T. Eaton, M. F. Elderkin, D. Bayne, S. Blaney, C. Burgess, C. Drysdale,

N. Hill, L. Helmer, P. MacDonald, D. MacKinnon, J. Mills, M. Munro, R.

Newell, D. Sam, L. Vasseur, and A. Wilson. 2005. National Recovery of the

Atlantic Coastal Plain flora, Volume 1: The Multiple Species Recovery and

Conservation Strategy and Action Plan. Nova Scotia, Canada.

Bouyoucos, G. J. 1962. Hydrometer method improved for making particle-size analysis

of soils. Agronomy Journal 54:464-465.

Crawford, R. M. M. 1989. Studies in plant survival: ecological case histories of plant

adaptation to adversity. Blackwell Scientific Publications, Oxford, England.

Day, P. R. 1965. Particle fractionation and particle size analysis. Agronomy 9:545-567.

Dynesius, M., and C. Nilsson. 1994. Fragmentation and flow regulation of river systems

in the northern third of the world. Science 266:753-762.

Eaton, S. T., and J. S. Boates. 2002. Assessing Threats to Coastal Plain Flora in the

Tusket River Watershed. Centre for Wildlife and Conservation Biology at Acadia

University, Wildlife Division of the Nova Scotia Department of Natural

Resources and the Federal Habitat Stewardship Program for Species at Risk.

Eaton, S., and J. S. Boates. 2003. Securing the Science Foundation for Responsible

Stewardship and Recovery of Atlantic Coastal Plain Flora (ACPF) Species at Risk

- Threats to ACPF: Water Quality and Alien Invasive Species Benchmarks,

Assessment and Stewardship. Centre for Wildlife and Conservation Biology at

Acadia University and Nova Scotia Department of Natural Resources, Wildlife

Division and the Federal Habitat Stewardship Program for Species at Risk.

125 Environment Canada. 2006. http://www.climate.weatheroffice.ec.gc.ca/climateData.

Fernald, M. L. 1921. The Gray Herbarium Expedition to Nova Scotia 1920. Rhodora

23:89-111, 130-152, 153-171, 184-195, 223-245, 257-278, 284-300.

Fernald, M. L. 1922. Notes on the flora of western Nova Scotia 1921. Rhodora 24:157-

181, 201-208.

Francis, G., and N. Munro. 1994. A biosphere reserve for Atlantic Coastal Plain flora,

south-western Nova Scotia. Biological Conservation 68:275-279.

Gaudet, C. L., and P. A. Keddy. 1988. A comparative approach to predicting competitive

ability from plant traits. Nature 334:242-243.

Gaudet, C. L., and P. A. Keddy. 1995. Competitive performance and species distribution

in shoreline plant communities: a comparative approach. Ecology 76:280-291.

Gavlak, R. G., D. A. Hornek, and R. O. Miller. 1994. Plant, soil and water reference for

the Western Region. Western Regional Extension Publication 125, University of

Alaska, Fairbanks.

Gee, G. W., and J. W. Bauder. 1986. Particle-size analysis. Pages 3383-3411 in A. Klute,

editor. Methods of Soil Analysis, Part 1. American Society of Agronomy,

Madison, Wisconsin.

Grime, J. P. 1973. Competitive exclusion in herbaceous vegetation. Nature 242:344-347.

Grime, J. P. 1977. Evidence for the existence of three primary strategies in plants and its

relevance to ecological and evolutionary theory. American Naturalist 111:1169-

1194.

Grime, J. P. 1979. Plant strategies and vegetation processes. John Wiley and Sons,

Chichester, England.

126 Gurevitch, J., S. M. Scheiner, and G. A. Fox. 2002. The Ecology of Plants. Sinauer

Associates Inc., Sunderland, Massachusetts.

Hazel, S. 2004. Tolerance limits of Atlantic Coastal Plain plants: what prevents

colonization on a reservoir? Thesis, Acadia, Wolfville.

Heiri, O., A. Lotter, and G. Lemcke. 2001. Loss on ignition as a method for estimating

organic and carbonate content in sediments: reproducibility and comparibility of

results. Journal of Paleolimnology 25:101-110.

Hill, N. M., and P. A. Keddy. 1992. Prediction of rarity from habitat variables: Coastal

Plain plants of Nova Scotian lakeshores. Ecology 73:1852-1859.

Hill, N., P. Keddy, and I. Wishue. 1998. A hydrological model for predicting the effects

of dams on the shoreline vegetation of lakes and reservoirs. Environmental

Management 22: 723-36.

Hocking, R. R. 1976. The analysis and selection of variables in linear regression.

Biometrics 32:1-49.

Holland, P. G. 1981. Pleistocene refuge areas, and the revegetation of Nova Scotia,

Canada. Progress in Physical Geography 5:535-562.

Holt, T. D., I. Blum, and N. M. Hill. 1995. A watershed level analysis of the lakeshore

plant community. Canadian Journal of Botany 73:598-607.

Jackson, S. T., and D. K. Singer. 1997. Climate change and the development of Coastal

Plain disjunctions in the central great lakes region. Rhodora 99:101-117.

Keddy, P. A. 1984. Quantifying a within-lake gradient of wave energy in Gillfillan Lake,

Nova Scotia. Canadian Journal of Botany 62:301-309.

Keddy, P. A. 1985a. Lakeshores in the Tusket River Valley, Nova Scotia: distribution

127 and status of some rare species, including Coreopsis rosea Nutt. and Sabatia

kennedyana Fern. Rhodora 87:309-320.

Keddy, P. A. 1985b. Wave disturbance on lakeshores and the within-lake distribution of

Ontario's Atlantic Coastal Plain flora. Canadian Journal of Botany 63: 656-660.

Keddy, P. A. 1989. Effect of competition from shrubs on herbaceous wetland plants: a 4

year field experiment. Canadian Journal of Botany 67:708-716.

Keddy, P. A., editor. 1991. Water level fluctuation and wetland conservation. Association

of State Wetland Managers Inc., Niagra, New York.

Keddy, P. A. 1994. Introduction: Why a special issue of Biological Conservation devoted

to Coastal Plain wetlands? Biological Conservation 68:199-200.

Keddy, P. A. 1999. Wetlands restoration: the potential for assembly rules in the service of

conservation. Wetlands 19:716-732.

Keddy, P. A. 2000. Wetland Ecology - Principles and Conservation. Cambridge

University Press, Cambridge.

Keddy, C. J., and P. A. Keddy. 1983. Status Report on the Pink Coreopsis (Coreopsis

rosea). Committee on the Status of Endangered Wildlife in Canada (COSEWIC).

Keddy, P. A., and I. C. Wisheu. 1989. Ecology, biogeography, and conservation of

Coastal Plain plants: some general principles from the study of Nova Scotian

wetlands. Rhodora 91:72-94.

Kennedy, W. J., and T. A. Bancroft. 1971. Model-building for prediction in regression

based on repeated significance tests. Annals of Mathematical Statistics 42:1273-

1284.

Kozlowski, T. T. 1984. Flooding and Plant Growth. Academic Press, Orlando, Florida.

128 Lewis, C. 1988. Nova Scotia's special places. Nature Canada 17:13-18.

Livingstone, D. A. 1968. Some interstadial and postglacial pollen diagrams from eastern

Canada. Ecological Monographs 38:87-125.

Loveland, P. J., and W. R. Whalley. 2000. Particle size analysis. Pages 271-328 in K. A.

Smith, editor. Soil and Environmental Analysis: Physical Methods. Marcel

Dekker Inc., New York, New York.

Maher, R. V., D. J. White, G. W. Argus, and P. A. Keddy. 1978. The Rare Vascular

Plants of Nova Scotia. National Museums of Canada, Ottawa.

Middleton, B. 1999. Wetland Restoration: Flood Pulsing and Disturbance Dynamics.

John Wiley and Sons, Inc., New York, New York.

Moore, D. R. J., P. A. Keddy, C. L. Gaudet, and I. C. Wisheu. 1989. Conservation of

wetlands: do infertile wetlands deserve a higher priority? Biological Conservation

47:203-217.

Morris, P. A., N. M. Hill, E. G. Reekie, and H. L. Hewlin. 2002. Lakeshore diversity and

rarity relationships along interacting disturbance gradients: catchment area, wave

action and depth. Biological Conservation 106:79-90.

Morris, P. D. 1994. The relationship between disturbance and the distribution of rare

Coastal Plain plants in Queens County, Nova Scotia. Thesis, Acadia, Wolfville,

Nova Scotia.

Moorhead, K. K., and K. R. Reddy. 1990. Carbon and nitrogen transformations in

wastewater during treatment with Hydrocotyle umbellata L. Aquatic Botany 37:

153-161.

Mott, R. J. 1975a. Palynological studies of lake sediment profiles from south-western

129 New Brunswick. Canadian Journal of Earth Science 12:273-287.

Mott, R. J. 1975b. Postglacial history and environments in south-western New

Brunswick. Pages 67-82 in J. G. Ogden and M. J. Harvey, editors. Environmental

change in the Maritimes, Halifax. Nova Scotia Institute of Science, Halifax.

Newell, R. 1998. Update Status Report for Pink Coreopsis (Coreopsis rosea). Committee

on the Status of Endangered Wildlife in Canada (COSEWIC), Wolfville, Nova

Scotia.

Newell, R. 1999. Update Status Report for Water Pennywort (Hydrocotyle umbellata).

Committee on the Status of Endangered Wildlife in Canada (COSEWIC),

Wolfville, Nova Scotia.

Nova Scotia Endangered Species Act. 1998. Bill 65, Chapter 11 of the Acts of 1998.

Halifax, Nova Scotia.

Nova Scotia Special Places Protection Act. 1989. Section 1, Chapter 438 of the Revised

Statutes of 1989. Halifax, Nova Scotia.

Ogden, J. G. 1960. Recurrence surfaces and pollen stratigraphy of a post glacial raised

bog, Kings County, Nova Scotia. American Journal of Science 258:341-353.

Ogden, J. G. 1965. Pleistocene pollen records from eastern North America. Botanical

Review 31:481-504.

Ogden, J. G., and M. J. Harvey. 1975. Environmental Change in the Maritimes. Nova

Scotian Institute of Science, Lawson Graphics Atlantic Ltd., Halifax, Nova

Scotia.

130 Ogwada, R.A., K.R. Reddy, and D.A. Graetz. 1984. Effects of aeration and temperature

on nutrient regeneration from selected aquatic macrophytes. Journal of

Environmental Quality 13:239-243.

Pronych, G., and A. Wilson. 1993. An Atlas of Rare Vascular Plants in Nova Scotia.

Nova Scotia Museum (Curatorial Report #71), Halifax, Nova Scotia.

Rabinowitz, D. 1981. Seven forms of rarity. in H. Synge, editor. The biological aspects of

rare plant conservation. John Wiley and Sons Ltd., New York, New York.

Rawinski, T. J., and S. Price. 1994. Conclusion: An action plan for Coastal Plain wetland

conservation: toward a continental conservation strategy. Biological Conservation

68:281-284.

Reddy, K.R. and W.F. De Busk. 1985. Nutrient removal potential of selected aquatic

macrophytes. Journal of Environmental Quality 14(4): 459-462.

Reznicek, A. A. 1994. The disjunct Coastal Plain flora in the great lakes region.

Biological Conservation 68:203-215.

Roland, A. E. 1991. Coastal Plain plants in inland Nova Scotia. Rhodora 93:291-298.

Roland, A. E., and E. C. Smith. 1969. The flora of Nova Scotia. Proceedings of the Nova

Scotia Institute of Science 26:5-238, 277-743.

Rood, S. B., J. M. Mahoney, D. E. Reid, and L. Zilm. 1995. Instream flows and the

decline of riparian cottonwoods along the St-Mary River, Alberta. Canadian

Journal of Botany 73:1250-1260.

Rood, S. B., J. H. Braatne, and F. M. R. Hughes. 2003a. Ecophysiology of riparian

cottonwoods: stream flow dependency, water relations and restoration. Tree

Physiology 23:1113-1124.

131 Rood, S. B., C. R. Gourley, E. M. Ammon, L. G. Heki, J. R. Klotz, M. L. Morrison, D.

Mosley, G. G. Scoppettone, S. Swanson, and P. L. Wagner. 2003b. Flows for

floodplain forests: A successful riparian restoration. Bioscience 53:647-656.

Rowell, D. L. 1994. Soil Science: Methods and Applications. Pearson Education Limited,

Essex, England.

Ryding, S., and W. Rast. 1989. Control of Eutrophication of Lakes and Reservoirs.

Parthenon Publishing Group, New Jersey.

SAS Institute Inc. 1985. SAS User's Guide: Statistics, Version 5 Edition. SAS Institute

Inc., Cary, North Carolina.

Schneider, R. 1994. The role of hydrologic regime in maintaining rare plant communities

of New York's Coastal Plain pondshores. Biological Conservation 68:253-260.

Sharp, M. J., and P. A. Keddy. 1985. Biomass accumulation by Rhexia virginica and

Triadenum fraseri along two lakeshore gradients: a field experiment. Canadian

Journal of Botany 63:1806-1810.

Shipley, B., P. A. Keddy, C. Gaudet, and D. R. J. Moore. 1991a. A model of species

density in shoreline vegetation. Ecology 72:1658-1667.

Shipley, B., P. A. Keddy, and L. P. Lefkovitch. 1991b. Mechanisms producing plant

zonation along a water depth gradient: a comparison with exposure gradient.

Canadian Journal of Botany 69:1420-1424.

Sokal, R., and F. J. Rohlf. 2000. Biometry. W. H. Freeman and Company, New York.

Sorrie, B. A. 1994. Coastal Plain ponds in New England. Biological Conservation

68:225-233.

132 Stephen, T., and G. Tanner. 1986. Nitrogen and phosphorus concentrations within

freshwater marsh plant species. Journal of Freshwater Ecology 3(3):347-358.

Summer, M., editor. 2000. Soil Organic Matter. CRC Press, Washington, D. C.

Sutter, R. D., and R. Kral. 1994. The ecology, status, and conservation of two non-

alluvial wetland communities in the south Atlantic and eastern gulf Coastal Plain,

USA. Biological Conservation 68:235-243.

Sweeney, S., and R. Ogilvie. 1993. The conservation of Coastal Plain Flora in Nova

Scotia. Maine Naturalist 1:131-144.

Unified Soil Classification System. Field classification technique for coarse-grained soils

(USCS P13b Version 2). www.dlwc.nsw.gov.au/care/soil/soil_pubs/soil_tests/

pdfs/usc.pdf

Vasseur, L. 2005. Monitoring populations of Hydrocotyle umbellata (water pennywort)

in Kejimkujik National Park and National Historic Site. Research Report Parks

Canada, Kejimkujik National Park.

Wilson, S. D. 1984. Status Report on Water-pennywort Hydrocotyle umbellata: an

endangered species in Canada. Committee on the Status of Endangered Wildlife

in Canada (COSEWIC).

Wisheu, I. C., C. J. Keddy, P. A. Keddy, and N. M. Hill. 1994. Disjunct Atlantic Coastal

Plain species in Nova Scotia: Distribution, habitat and conservation priorities.

Biological Conservation 68:217-224.

Wisheu, I. C., and P. A. Keddy. 1989. The conservation and management of a threatened

Coastal Plain plant community in eastern North America (Nova Scotia, Canada).

Biological Conservation 48:229-238.

133 Wisheu, I. C., and P. A. Keddy. 1994. The low competitive ability of Canada's Atlantic

Coastal Plain shoreline flora: implications for conservation. Biological

Conservation 68:247-252.

Zinck, M. 1998. Roland's Flora of Nova Scotia, 3rd edition. Nimbus Publishing and Nova

Scotia Museum.

134