Effectiveness of Boundary Structures in Limiting Residential Encroachment into Urban Forests

1

Effectiveness of boundary structures in limiting residential encroachment 1 into urban forests 2

______________________________________________________________________ 3 4 ABSTRACT /Urban forests provide essential functions in support of their human communities; 5

however, studies indicate adjacent residential land uses degrade these natural systems following 6

development. Local Ontario governments rely primarily on passive management, such as 7

boundary-focused measures (e.g. fences) to limit this degradation but, it is not known whether 8

these measures are effective for limiting the area of the forest floor covered by encroachment 9

impacts. Transect and quadrat sampling of 40 forest edges adjacent to 186 residential properties 10

were sampled in six Southern Ontario municipalities to determine the frequency and area cover 11

of encroachment impacts. The results indicated some boundary structures are effective in 12

reducing the frequency and area cover of some encroachment behaviours but, the area covered 13

by some behaviours of encroachment were not reduced, some were actually increased by 14

boundary measures, and substantial areas of encroachment continued to occur under even the 15

most effective boundary treatments. The most effective boundary treatment in this study was 16

ungated fencing in combination with a mown strip which was found to reduce the incidence of 17

yard extension and concentrated the encroachment closest to the forest edge. 18

Key words: Green infrastructure; Greenways; Residential encroachment; Urban natural systems 19 planning and management; Urban forest edges 20

1.0 Introduction 21

Urban natural systems are essential elements in the provision of urban ecological services 22

to human communities. They provide storm water management (Correll, 1999); filter pollutants 23

from the air (Scott et al., 1999); sequester carbon, reducing the rate of climate change (Nowak 24

and Crane, 2002), and moderate extreme weather conditions, improving human thermal comfort 25

2

(Brown and Gillespie, 1995). Furthermore, they provide key recreation facilities in support of 26

human health (Faber-Tayler et. al., 2001; Wells, 2000; Kaplan, 1995). 27

There is increasing evidence of adjacent residential land uses degrading urban natural 28

systems, particularly woodlands, following development within central Canada (Ouellet, 1996; 29

Taylor, 1992; McWilliam, 2009) and in the northeastern U.S. (Moran, 1984; Matlack, 1993). The 30

impacts of these adjacent land uses are collectively referred to as encroachment. Direct 31

encroachment occurs with residential activities in adjacent publicly-owned forest edges (e.g. 32

private garden extensions). Indirect forms occur with activities in residential yards (e.g. garden 33

plant invasions of forest edges from residential yards). 34

Ontario municipal planners and managers recognize that some impacts to natural systems 35

will occur with adjacent development. However, they want to ensure impacts are kept to 36

acceptable limits, and have developed planning and management policies and practices to ensure 37

their features and functions are protected for the long-term (McWilliam, 2009). They have relied 38

on boundary-focused measures in planning policies, such as requiring the installation of fences 39

and buffers, and infrequently implemented management measures, such as bylaw enforcement, 40

boundary monitoring and resident education (McWilliam, 2009). However, municipalities know 41

relatively little about the effectiveness of this strategy, or the relative effectiveness of boundary-42

focused planning or management measures for limiting encroachment impacts on their natural 43

systems. 44

Different types of encroachment results in different impacts. Exotic garden plant 45

extensions, and human trampling and disturbance within forest edges resulting from other 46

encroachment activities often result in a reduction in native shade-tolerant plant species, and an 47

increase in the number of exotic and native light and disturbance-tolerant plant species that can 48

3

in turn lead to reductions in local and regional native plant diversity (Bagnall, 1979; Florgard, 49

2000; Hoehne, 1981; Manning, 1979; McWilliam, 2009). 50

The impacts of other indirect forms of encroachment, such as alterations in noise, light, 51

hydrological and chemical regimes, or pet predation of wildlife within forested ecosystems are 52

less visually apparent , and therefore more difficult to measure. Individual impacts may take 53

years to accumulate before serious impacts become apparent and they can be measured (Murphy, 54

2006). Furthermore, impacts are often complex and species dependent. For example, studies 55

indicate that excessive chemical, particulate or bacteria flows from adjacent urban land uses into 56

natural system soils and water bodies can lead to the anatomic alteration of frogs (Mayer et al., 57

1999; Reeder et al., 2005; U.S. Environmental Protection Agency, 1995), and the alteration of 58

fungi and invertebrates within soils and water contaminated with these substances can in turn 59

lead to alterations in plants and animals (Cousins et al., 2003), and may impact human health. 60

Most direct behaviours of encroachment are much more obvious and measurable. The 61

impacts of some behaviours, such as the extension of residential lawns and gardens, are 62

particularly striking and can result in the long term loss of public forest area to private land uses 63

(McWilliam, 2009). Many direct forms of encroachment result in ecological and recreation 64

impacts similar to those experienced within forests of less developed landscapes. Recreation 65

ecology studies indicate that hiking and result in the trampling and disturbance of soils, 66

vegetation and wildlife, and degradation of recreational experience (Cole, 1987; Manning, 1979; 67

Farrell and Marion, 1998; and others). Similar, but more intensive, human activities within urban 68

forests, commonly referred to as passive recreation, result in similar impacts. Trampling can 69

negatively impact soil fertility, and reduce soil organism diversity (Malmmivaara-Lamsa and 70

Fritze, 2003 ). It also negatively impacts vegetation communities (Florgard, 2000). Furthermore, 71

4

human disturbance of wildlife and alteration of wildlife habitat can reduce native wildlife 72

communities and diversity (Sauvajot et al., 1998). Human activities such as waste disposal, tree 73

and plant damage, fire ring and unauthorized trail creation can also degrade the quality of 74

recreational experience (Lynn and Brown, 2003). Many direct forms of encroachment result in 75

similar impacts (Matlack, 1993; Ouellet, 1996; Taylor, 1992; McWilliam, 2009; McWilliam et 76

al., 2010). 77

Extensive recreation ecology studies have identified the factors contributing to the soil, 78

plant, wildlife and recreation experience impacts of human activities that result in trampling and 79

disturbance, and managers of these landscapes use their understanding of these factors to 80

develop effective mitigation strategies. Factors include the characteristics of the impact (its areal 81

extent, longevity, and the intensity of impact) and the attributes of the ecosystem (i.e. the human 82

value placed on the natural system in which the impact takes place (Cole, 2003). Intensity of 83

impact is in turn determined by the frequency of use, the type of use and how it occurs, the 84

season in which it occurs and the sensitivity of the ecosystem to the impact (Cole, 2003). 85

Managers of some of the larger U.S. and Canadian national parks have developed 86

strategies that manipulate many of these factors through time, and at different spatial scales (for 87

example, Leung and Marion, 1999; USDI National Park Service, 1997), focusing their resources 88

on factors that have the greatest potential to reduce impacts. However, in general, the results of 89

recreation ecology studies looking at the relative efficacy of the different factors contributing to 90

the significance of human activity impacts, indicate that reducing the spatial extent over which 91

impacts occur is particularly important in reducing the significance of human activity impacts 92

(Cole, 2003, and others). Applied to the reduction of similar encroachment-related impacts, this 93

5

suggests that strategies that controlling the area in which encroachment occurs are likely to be 94

more effective than those seeking to reduce the intensity (e.g. frequency) of impacts. 95

In terms of areal extent of encroachment, at coarse spatial scales, studies indicate 96

residential encroachment is highly prevalent within Southern Ontario municipal forests, 97

occurring in 100% of municipalities surveyed and adjacent to 95% of residential yards 98

(McWilliam et al., 2010). At the scale of the individual forest, encroachment tends to be 99

concentrated within forest edges rather than uniformly distributed throughout the forest 100

(Matlack, 1993; McWilliam et. al., 2010). For example, Matlack (1993) surveyed 95% of both 101

yard and recreation related impacts within 67 metres of forest borders in forests without internal 102

roads; while McWilliam et al. (2010) surveyed 95% of residential impacts along within 34 103

metres of forest borders adjacent to abutting and contiguous single-family housing (with gross 104

densities of 5 to 19 houses per hectare). Studies also indicate some boundary structures that 105

control physical access, delineate private from public land uses and encourage boundary 106

monitoring significantly reduce the maximum distance from forest borders in which 107

encroachment occurs. However, the area of the forest floor covered by encroachment within this 108

distance and the effectiveness of boundary measures for reducing this encroachment area are 109

unknown. 110

This paper investigated the relationship between boundary treatment and the 111

encroachment area of forest edge behind residential properties. The results will assist planners 112

and managers in developing more effective approaches for limiting encroachment impacts within 113

municipal forested natural systems. 114

115 2.0 Methods 116 117 2.1 Study areas 118

6

119 Human impacts were sampled behind 186 residential properties within 40 forest 120

fragments within six municipalities of Southern Ontario, including Cambridge, Guelph, 121

Kitchener, Mississauga, Oakville and Waterloo. Study municipality populations range between 122

100,000 to 200,000, with the exception of Mississauga, with a population of 700,000, and are 123

situated within the Greater Toronto Area (Figure 1). All are currently implementing boundary 124

treatments as a passive management measure to limit encroachment, and are infrequently 125

implementing active management measures including boundary monitoring, bylaw enforcement 126

and resident education (McWilliam, 2009). 127

128 Figure 1. Study municipalities in Southern Ontario, Canada 129

(Source: (Ontario Ministry of Natural Resources, 2002)). 130 131

Most study forests were remnants of upland deciduous eastern forests. A few were mixed 132

deciduous/coniferous upland eastern forest fragments, or were lowland deciduous eastern forest 133

corridors. Fragments were between 1 and 50 ha in area, without internal roads, and had minimum 134

widths of 20 metres (if there was no development on the other side of the forest). If there was 135

opposing development, then the minimum width was 40 metres. This site selection criterion 136

reduced the risk of sampling opposing resident encroachment and responds to results of a pilot 137

study (McWilliam, 2009) indicating a majority of edge-resident encroachment occurred within 138

20 metres of the property boundary. To avoid overlapping encroachment associated with 139

community recreation authorized recreational trails had to be located a minimum of five metres 140

away from study areas. Research indicates that the area of impact associated with recreational 141

trail use is approximately five metres from the trail edge (Cole, 1987). Furthermore, sites 142

immediately adjacent to park entry points were not sampled to avoid the sampling of 143

encroachment activities arriving from other residences in the community. 144

7

Forest borders were without significant natural barriers to human entry (such as dense, 145

prickly, or poisonous vegetation, steep slopes, or poorly drained soils). Adjacent housing was 146

typical of post World War Two suburban housing within Southern Ontario, consisting of single-147

family, contiguous and detached or semi-detached, with backyards. Gross housing densities were 148

between 5 and 19 houses per hectare. Housing was at least 10 years old. Research had indicated 149

that maximum intensities of the most visible effects of recreational activities, such as loss of 150

organic soils, erosion, and changes in vegetation communities occurred in forests within two to 151

five years of the occurrence of recreation activities resulting in trampling (Cole, 1987). Back 152

yards were between 10 and 40 metres in width and directly abutted municipal natural area 153

boundaries. 154

2.2 Measures for limiting encroachment 155 156 2.2.1 Passive management measures 157 158 Municipal environmental planners and forest managers within the study municipalities 159

indicated they had implemented two boundary-focused passive management treatments within 160

sites meeting the above site selection criteria: 1) fence (installed on municipal land) and 2) 161

boundary demarcation post (installed on municipal land). However, the pilot study (McWilliam, 162

2009) indicated that a majority of forests with adjacent residential neighbourhoods within 163

Southern Ontario municipalities had a wide variety of both residential and/or municipality-164

installed boundary treatments. Encroachments occurring under eight of these additional 165

boundary treatments were sampled. These boundary treatments consisting of: 1) no or minimal 166

boundary demarcation (e.g. a flower bed), 2) grass strip, 3) grass strip and path, 4) fence and 167

grass strip, 5) fence, grass strip and path, 6) fence with gate, 7) fence with gate and grass strip, 168

and 8) fence with gate, grass strip and path. Sites adjacent to resident-installed fences as well as 169

8

municipal fences were sampled in order to determine whether there was a significant difference 170

in encroachment between resident and municipality-installed fences. Some of the municipal 171

components within boundary treatments were not implemented to reduce encroachment, but to 172

provide recreational facilities or maintain utility corridors. Table 1 summarizes the boundary 173

treatments surveyed, their ownership and whether they were implemented to address 174

encroachment. 175

Table 1. Evaluated boundary treatments 176

Grass strips were a maximum of 50 metres in width (the average width was 20 metres). 177

Only sites with mown grass strips were selected for sampling because the pilot study 178

(McWilliam, 2009) indicated long grass or shrubs within the grass strip could alter encroachment 179

behaviours. Paths were located adjacent to, or within, grass strips, and ran roughly parallel to 180

residential property boundaries. 181

182 2.2.2 Active management measures 183 184 Municipal environmental planners and forest managers within the study municipalities 185

indicated all study municipalities implemented management measures to limit encroachment, but 186

infrequently and informally (McWilliam, 2009). Within study sites measures consisted of the 187

enforcement of bylaws prohibiting encroachment activities, the informal municipal and/or 188

community monitoring of boundaries and the installation of signs prohibiting the dumping of 189

resident waste. Municipal staff and available records indicated bylaws had been implemented 190

within 11 of the 186 (6%), study sites. Signs at park entries prohibiting the dumping of waste 191

were primarily installed within park entrances of 16 of the 40 (40%), study forests. There were 192

no records of municipal monitoring of boundaries; however, staff from five out of six 193

municipalities indicated that municipal monitoring of boundaries occurred infrequently, and not 194

9

more than once per year. The exception was Oakville where municipal staff indicated regular 195

municipal monitoring of boundaries occurred within sites with boundaries consisting of a fence, 196

grass strip and a path. 197

198 2.3 Sampling methods 199 200 The frequency and percentage area of the forest floor covered by encroachment activities 201

were sampled using a quadrat/transect sampling method. The use of quadrats and transects are 202

commonly used to sample forest vegetation (Kent and Coker, 1992). Recreation ecology research 203

also commonly measures the impacts of recreational trampling on vegetation communities using 204

this method (For example, Cole and Marion, 1988). The quadrat/transect sampling method 205

requires the researcher to visually estimate the percentage of a sample area occupied by different 206

components of the forest floor. In urban forest edges, component categories include native, 207

naturally occurring elements (such as native plants, soils or woody debris), exotic vegetation 208

(arriving via other means than from adjacent residential yards), and encroachment elements or 209

traces (e.g. construction waste, lawn extensions, unauthorized pathways or plants originating 210

from residential yards). Within each ½ metre2 quadrat sampled, we recorded the percentage 211

cover of the quadrat of each of these components visible at 30 cm above the ground, according to 212

the Braun-Blanquet (1932) abundance-cover scale. The scale assigns a number, or code, to each 213

forest floor component, ranging from 0 to 6 (Table 2). 214

215 Table 2. Braun-Blanquet (1932) abundance-cover scale 216

217 The frequency in which encroachment occurred was calculated by counting the number 218

of times one type of encroachment appeared within the quadrat, no matter what percentage of the 219

quadrat it covered. The sample design was perfected during the pilot study to capture the breath 220

10

and depth of encroachment activities under different boundary conditions (McWilliam, 2009). 221

Eleven 0.5m by 0.5m quadrats were sampled at two-metre intervals along five transects placed 222

perpendicular to, and equal distance along, property boundaries. From the property boundary, 223

transects extended 20 metres into forest edges. The first and last transects were placed one metre 224

from neighbouring property boundaries to reduce the risk of recording neighbouring 225

encroachment. This design resulted in 55 samples per site, for a total sampling intensity of 930 226

transects and 10,225 samples (Figure 2). 227

228 Figure 2. Quadrat/Transect sampling design for one study site 229

230 Care was taken to record only types of encroachment associated with adjacent study 231

residents. For example, cut trees were not recorded since resident-cut trees could not be 232

distinguished from municipally cut, or hazardous, trees. In addition, evidence of old dumps, rock 233

piles, and old fencing, were not recorded as they may be from previous agricultural land uses. 234

Categories of encroachment were developed in order to analyze the significance of 235

different encroachment behaviours within the forest edge in response to boundary treatments. 236

Categories of encroachment were developed from the types and spatial patterns of encroachment 237

in forest edges in relation to yard activities and patterns revealed in the pilot study (McWilliam, 238

2009). Municipal forest managers, bylaw enforcement officers, and casual conversations with 239

residents verified these behaviours. 240

Behaviours identified include: 1) waste disposal, 2) yard extension, 3) forest-recreation, 241

4) response to forest encroachment and 5) yard plant invasions. Disposal behaviour results in 242

yard or house-related waste being dumped within forest edges, including construction debris, 243

grass clippings, or miscellaneous organic waste. The yard extension category of encroachment 244

includes any extensions of private yard-related land uses within the public forest edge, including 245

11

lawn extensions, or the installation of recreational equipment such as trampolines or swimming 246

pools. The forest recreation category of encroachment includes types of encroachment that result 247

in private, forest recreation-related facilities within the forest edge, such as children's forts, fire 248

pits or private entry pathways. Yard plant invasion encroachments result in garden plants 249

growing within the forest edge. Exotic plants growing in the forest edge, but not typically grown 250

in residential gardens, are not included within this encroachment category. Residential 251

encroachment in response to forest encroachment refers to encroachment activities that respond 252

to forest components (e.g. vegetation or wildlife) encroaching into their yards. An example of 253

encroachment resulting from this behaviour is the resident removal of forest border vegetation 254

extending over property boundaries. 255

256 3.0 Data Analysis 257 258

Null hypotheses of uniform distribution within the forest of all encroachment traces, 259

encroachment traces by behaviour, and encroachment traces by boundary treatment were tested 260

using a Kolomogorov-Smirnov test. 261

The frequency of traces and the percentage of the sampled forest floor covered by traces, 262

was calculated for all encroachment occurring, as well as by type and behaviour of 263

encroachment. In addition, an indicator of the relative significance of encroachment was 264

calculated by multiplying the mean frequency of traces (number of areas of encroachment) by 265

the mean Braun-Blanquet abundance-cover scale representing the percentage area covered within 266

the sample area. For example, if a yard extension encroachment trace was sampled three times in 267

a site (i.e. in three different quadrats) and covered 75 to 100% of the quadrat (a Braun-Blanquet 268

abundance cover scale of 6), then the indicator of encroachment significance for this site 269

(heretofore referred to as indicator) would be 18. 270

12

Frequencies of traces and significance of encroachment by type, behaviour and with 271

distance from forest borders were also calculated for each boundary treatment to determine the 272

relative efficacy of boundary treatments for limiting encroachment. Kolomogorov-Smirnov two-273

sample tests were conducted to determine the extent to which boundary treatments significantly 274

altered mean frequencies and indicators of encroachment. A Kruskal-Wallis test was conducted 275

to determine whether the ownership of the boundary treatments (i.e. municipality or resident 276

owned) significantly altered mean frequencies and indicators of encroachment. We also 277

conducted a Spearman correlation test to determine whether fence height significantly altered the 278

frequency and percentage cover of encroachment. 279

4.0. Results 280 281 4.1 What is the mean frequency and percentage cover of encroachment under different 282 boundary treatments? 283 284 Encroachment activities occurred within sites under all boundary treatments; however, 285

there was a significant difference in mean frequency and mean indicators between boundary 286

types (P<.05). Boundary treatments discouraging resident access to forest borders had 287

significantly lower frequencies and indicators than treatments allowing access. Fences were 288

effective in discouraging access to forest edges, functioning to decrease the area cover of the 289

forest floor covered by encroachment. Boundary treatments with fences (no gates) had 290

significantly lower mean frequencies and indicators than boundary treatments without fences, or 291

those having gates (P < .05). For example, the forest floor covered by encroachment traces 292

increased from 26 -50% in fenced sites to 51 -75% in fenced sites with gates. There was no 293

significant difference in the mean frequency or mean indicator between sites having municipal 294

fences and sites having resident fences (P >.05). 295

13

Sites having no boundary demarcation had significantly higher frequencies and 296

indicators than all other boundary treatments, except sites with grass strips (P < .05). In addition 297

to their lack of elements deterring resident entry to forest edges, many sites with no boundary 298

demarcation lacked any official property line demarcation. For example, only 30% of sites with 299

no boundary treatment had a survey stake. However, in many cases survey stakes did not appear 300

to reduce encroachment activities. For example, Figure 3 illustrates a garden extension 301

encroachment within a site subject to a municipal boundary post and survey stake. 302

303 Figure 3. Yard extension encroachment (highlighted in light green) 304

despite a municipal boundary post (marked with an 'x') and survey stake (bottom left) 305 (Photo: Sugar bush Park, Waterloo. Source: McWilliam, 2004) 306

307 When grass strips, with and without paths, were added to fences, the frequency and 308

indicator were significantly less than those of sites having just a fence (P<.05). This suggests that 309

grass strips and paths combined with fences provide more barriers to encroachment behaviour 310

than fences alone. 311

Municipal boundary posts were not highly visible to residents and did not physically 312

deter residents from entering forest borders. Yet, sites with municipal boundary posts had 313

significantly lower mean frequency and indicator than sites with no municipal boundary 314

demarcation policy, fenced sites with gates, and fenced sites (P<.05). These findings may have 315

been due to site-specific factors, since only 12 sites with this boundary type were available for 316

sampling, and most were located in one municipality. Within many of these sites, the adjacent 317

houses were built without removing forest vegetation from resident yards, and some residents 318

retained the forest floor vegetation following development (Figure 4). In addition, shaded 319

conditions, created by overhanging forest canopy trees, may have deterred residents from 320

establishing lawns thereby discouraging yard extension encroachments. 321

14

322 Figure 4. Low levels of encroachment where residents retained native forest floor 323

cover despite a subtle boundary treatment that did not deter residents 324 from entering the forest border 325

(Photo: Sparrow Park, Waterloo. Source: McWilliam, 2005) 326 327

328 4.2 Was there a significant difference in the frequency and indicator of significance of 329 encroachment with respect to the distance from the forest border? 330 331 The frequency and indicator of encroachment intensity occurring within sites with fence 332

components (without gates) appeared to be distributed closer to forest borders than that occurring 333

within sites without fences, and within sites having fences and gates. For example, 95% of 334

encroachment traces were located within 16 metres of fenced sites, but were located within 18 335

metres in fenced sites with gates, and 20 metres within sites with no boundary treatment. 336

However, the difference in distribution was not statistically significant. 337

338 4.3 What was the mean frequency and intensity of different encroachment behaviours 339 under different boundary treatments? 340

341 4.3.1 Yard extension encroachment 342 343

Boundary treatments that included a fence (but not a gate) had significantly lower 344

frequencies and indicators of yard extension encroachment than sites without fences or fenced 345

sites with gates (P < .05). For example, the mean indicator of yard extension increased from 10 346

in fenced sites to 35 in fenced sites with gates. The percentage of the forest floor covered by yard 347

extensions increased from a mean of 51 to 75% in fenced sites, to 76 to 100% in fenced sites 348

with gates. Furthermore, the frequency at which yard extension encroachment was sampled 349

increased three fold between the two boundary treatments. Figures 5 and 6 illustrate the 350

differences in the mean indicators of encroachment behaviours between fenced and fenced sites 351

15

with gates with respect to forest borders. Notice the dramatic increase in mean indicator of yard 352

extension intensity between the two boundary treatments. 353

354 Figure 5. Distribution of mean indicators of significance of encroachment 355

behaviours with respect to distance from forest borders in fenced sites. 356 357 358

Figure 6. Distribution of mean indicators of significance of encroachment 359 behaviours with respect to distance from forest borders in fenced sites with gates. 360

361 Sites with grass strips (and no gate) appeared to have lower mean frequency of yard 362

extension encroachment and indicators of yard extension intensity than boundary treatments 363

without grass strips. In fact, many of these boundary treatments, such as fences, grass strips and 364

paths had no yard extension traces within the forest edge. However, relatively few sites with 365

grass strips were sampled due to lack of availability, and these relationships were not statistically 366

significant. Nevertheless, grass strips appeared to deter yard extensions within forest borders and 367

encourage them within grass strips (Figure 7). 368

369 Figure 7. Grass strips appear to encourage yard extensions within grass strips 370

(flowerbed to left of measuring tape) rather than forest edges 371 (Photo: Anndale Park, Waterloo. Source: McWilliam, 2005) 372 373 4.3.2 Waste disposal encroachment 374 375

None of the boundary treatments eliminated waste disposal encroachment behaviour. 376

Sites with grass strips significantly reduced waste disposal mean frequency and indicators 377

relative to sites without grass strips (as long as there were no gates) (P < .05). The appearance of 378

gates within boundary treatments with grass strips led to a significant increase in waste disposal 379

encroachment relative to sites having grass strips, but no gates (P < .05). Figure 8 illustrates the 380

distribution of waste disposal encroachment behaviour with respect to distance from forest 381

16

borders within sites with fences, grass strips and paths. Notice the lower waste disposal 382

encroachment relative to those within sites with just fences or fences and gates (Figures 5 and 6). 383

384 Figure 8. Distribution of mean indicators of significance of encroachment behaviours 385

with respect to distance from forest borders within fenced sites with grass strips and paths. 386 387 Fences did not significantly reduce waste disposal frequency or indicator relative to any 388

other boundary type (P > .05). In fact, fenced sites had significantly higher frequencies and 389

indicators than sites with grass strips, with and without paths, and higher frequencies of waste 390

disposal than sites with municipal boundary posts (P < .05). Notice the relatively high indicators 391

of waste disposal in fenced sites (Figure 5) relative to fenced sites with grass strips and paths 392

(Figure 8). 393

The heights of the fences may have been too low to discourage residents within their 394

yards from dumping over fences. The mean height of study site fences was 1.5 metres measured 395

from the municipal forest side of the fence. The height of the fence on the resident's side; 396

however, was often appreciably less than this since many residents installed raised flowerbeds, 397

patios and pool decks. These structures raised the grade of yards above that of the adjacent forest 398

edge, and effectively reduced the height of the fence on the side of the residence. While 399

increasing the height of the fence may result in reductions in waste disposal encroachment; there 400

was no correlation between fence heights (ranging from 91 to 163 cm) and waste disposal mean 401

frequency and indicator of intensity (P>.05). 402

403 4.3.3 Forest recreation encroachment 404 405

Fenced sites reduced the mean frequency and indicator of forest recreation-related 406

encroachment relative to that occurring in fenced sites with gates (Figures 5 and 6), and fenced 407

sites with grass strips and paths (Figure 8) (P < .05). For example, all fenced sites with gates had 408

17

unauthorized pathways leading from residential yards into the forest edge, while few fenced sites 409

had these pathways. The grass strip appeared to increase access to the forest border for this type 410

of encroachment, and many pathways were used for waste disposal, sometimes in addition to 411

forest recreation. Closed side canopies did not appear to discourage unauthorized pathway 412

creation. Unauthorized pathways were observed throughout many of the study sites, particularly 413

within forests with a sparsely growing understory. 414

415 4.3.4 Residential encroachment in response to forest encroachment 416 417

None of the boundary treatments eliminated residential encroachment in response to 418

forest encroachment. Although the frequency and indicator of encroachment in response to 419

forest encroachment within residential yards was higher in fenced sites than in sites with grass 420

strips, the difference was not significant (P > .05). A grass strip may separate the residential 421

property from the forest border, thereby removing the encroaching forest vegetation from the 422

property boundary; however, further proof is required to support this theory. 423

424 4.3.5 Garden plant invasions 425 426

None of the boundary treatments eliminated garden plant invasions, although grass strips 427

significantly reduced their occurrence relative to sites without grass strips. Sites with grass strips 428

had a significantly lower mean frequency and mean indicator of garden plant invasion than sites 429

without grass strips (P <.05). Grass strips and paths may serve as a barrier to the dispersal of 430

garden plants that spread through vegetative reproduction. 431

Fences were ineffective as barriers to this type of encroachment. Garden plant invasions 432

were just as frequent and covered just as much area in fenced sites as sites without fences (P 433

>.05). Boundaries with high frequencies and indicators of waste disposal encroachment may also 434

18

increase garden plant invasions that enter forest edges through the disposal of viable plant 435

propagules in waste. Similarly, boundaries that abut residential yards with forest borders (e.g. 436

fenced sites) may also encourage garden plant invasions planted by residents. Informal 437

conversations with residents during sampling indicated some residents planted forest edges to 438

improve the aesthetic appearance of the forest edge. 439

440 5.0 Discussion 441 442 Boundary-focused passive management measures can significantly limit the area of 443

encroachment within forest edges. For example, those containing components that reduce 444

resident access to forest edges (e.g. fences, grass strips with and without pathways) led to 445

significantly lower areas of encroachment than those with no restrictions, or that encouraged 446

access to forest borders (e.g. no boundary demarcation, or fences with gates). Fences (with no 447

gates) appeared to concentrate the frequency and area cover of encroachment closer to forest 448

borders. This finding is supported by a study that found fences significantly reduce the maximum 449

extent of encroachment within forest edges (McWilliam et al., 2010). Other recreation ecology 450

studies have demonstrated similar relationships between human activity impacts and access 451

(Matlack, 1993; Roggenbuck and Lucas, 1987; Loeb, 1989; Furuseth and Altman, 1991). 452

Reducing access to urban natural systems, or to those sensitive to human activity impacts, may 453

reduce some social and economic values of urban natural systems, such as recreational 454

opportunities or real estate values, and residential areas and urban natural systems are commonly 455

planned to maximize the exposure of natural systems to residential land uses in order to 456

maximize these latter values. However in doing so, ecological services and some recreation-457

related values may be compromised. Furthermore, forest management resources required to 458

protect these latter values may have to be increased if these values are to be maintained for the 459

19

long term. Key ecological, social and economic functions of natural systems need to be clearly 460

identified and communicated to surrounding communities so that planning and management 461

decisions can be made in support of these functions prior to and following adjacent land use 462

development. 463

Different boundary components were effective in reducing the area cover of different 464

encroachment behaviours with no one boundary component effective in reducing all behaviours. 465

Therefore, to maximize the effectiveness of boundary-focused passive management measures, 466

multiple boundary components need to be designed into a boundary treatment. Boundary 467

treatments should include fences (without gates) to reduce the incidence of yard extension 468

encroachment both within forest edges and adjacent mown grass strips, and to concentrate 469

encroachment as close to the forest border as possible. Although differences between boundary 470

treatments for reducing the extent of encroachment were not statistically significant within this 471

study, the evidence from this and other studies (example, McWilliam et al., 2010; Matlack, 472

1993) suggest that fences and other structures that effectively limit access to forest borders 473

significantly reduce the extent of encroachment activities. Fences significantly reduced 474

encroachment whether residents installed them on their own land, or the municipality installed 475

them on municipal land. However, if the fence is installed on municipal land, the municipality 476

has the right to remove resident-installed gates from fences. They do not have this right if the 477

fence is installed on resident-owned land. Furthermore, if the fence is installed on resident land, 478

the resident may remove the fence. A survey of boundary treatments within the pilot study 479

indicated that in absence of municipal fences, residents preferred no boundary treatment, or 480

installed fences with gates. Under these boundary treatments, the results indicate encroachments 481

will significantly increase. A few residents did install gates within municipal fences. This led to 482

20

higher levels of encroachment within forest edges, particularly yard extensions (Figure 9). 483

Therefore, to minimize encroachment within forest edges, municipalities should ensure fences 484

are installed on municipal land, gates are not authorized, and fences are periodically monitored to 485

ensure the removal of gates. 486

Figure 9 Some residents place illegal gates (indicated by arrow) in municipal fences 487 leading to a significant increase in the area of encroachment. 488 (Photo: Village Wood Park, Oakville. Source: McWilliam, 2005) 489

490

Other boundary components (or other measures) are needed to complement fences, to 491

reduce waste disposal encroachment, yard plant invasions, and encroachments related to resident 492

responses to forest encroachment. Relative to fences, grass strips were effective in reducing, 493

waste disposal, yard plant invasions and encroachment in reaction to forest encroachment. 494

However, grass strips need to be coupled with fences to achieve these results. For example, 495

without fences, mown grass strips tend to encourage yard extensions within mown grass strips 496

meant to provide other functions, such as public recreation facilities, stormwater management or 497

utility corridor maintenance. 498

Despite the benefits of mown grass strips for reducing these forms of encroachment, they 499

do not provide supplemental wildlife habitat to urban forests, their maintenance generates air and 500

noise pollution, and absorbs municipal management budgets that could otherwise be used in 501

monitoring or education programs to further protect forests from encroachment. 502

Many remaining forest fragments within Southern Ontario tend to be too small in area 503

and convoluted in shape to provide habitat to sensitive, less common species requiring larger 504

areas or separation from human disturbance. Many also have a high edge to interior ratio that 505

leaves them highly exposed to adjacent land use impacts. Official or Secondary plan policies 506

within Southern Ontario municipalities often require forested buffers with development adjacent 507

21

to some natural area designations in order to improve their wildlife values and resistance to 508

construction impacts (McWilliam, 2009). While forested buffers could segregate encroachment 509

activities from occurring within designated forest areas, they do not limit encroachment that lead 510

to the loss or degradation of forested habitat within buffers. It is possible that mown grass strips 511

do not have to be wide to limit encroachment, and therefore, only a small portion of the forested 512

buffer would have to be dedicated to encroachment protection. However further research is 513

required to determine the minimum effective width of mown grass strips for limiting these 514

encroachment behaviours. 515

The addition of pathways could further improve the performance of mown grass strips for 516

limiting encroachment by encourage municipal and informal community monitoring of 517

boundaries, and by clearly delineating municipal from resident-owned property. Furthermore, 518

they direct recreation-related impacts on forests away from the more sensitive interior forest 519

area, in addition to facilitating municipal edge management activities, such as hazardous tree and 520

encroachment removal. However, there was an insufficient number of sites with pathways 521

existing within study municipalities to determine the efficacy of pathways, independent of grass 522

strips, for limiting encroachment. 523

The results of this study have contributed to a better understanding of the relationship 524

between boundary treatment and the area of forest edge behind residential properties that can be 525

affected by encroachment. Landscape architects and planners can use the results to design and 526

implement residential developments that will have the advantages of living near natural areas, 527

but will also protect the integrity of those natural areas. 528

529 Acknowledgements 530 531

22

We would like to thank the Canadian Mortgage and Housing Corporation for their financial 532 support of this research. In addition, we would like to thank the Municipalities of Cambridge, 533 Guelph, Kitchener, Mississauga, Oakville and Waterloo for their participation and assistance. 534 535 536 References 537 538 Bagnall, R.G. (1979) A study of the human impact on an urban forest remnant, New Zealand 539 Journal of Botany, 17, pp. 117 - 126. 540 541 Braun-Blanquet, J. (1932) Plant Sociology: the Study of Plant Communities (New York: 542 McGraw-Hill). 543 544 Brown, R.D., Gillespie, T.J. (1995) Microclimatic Landscape Design (New York: John Wiley & 545 Sons). 546 547 Cole, D. N. (1987) Effects of three seasons of experimental trampling on five montane forest 548 communities and a grassland in western Montana, USA, Biological Conservation, 40, pp. 219-549 244. 550 551 Cole, D. N. and Marion, J. L (1988) Recreation impacts in some riparian forests of the eastern 552 United States, Environmental Management, 12, pp. 99-107. 553 554 Cole, D. N. (2003) Backcountry impact management: lessons from research, Trends, 31(3), pp. 555 10-14. 556 557 Correll, D.L. (1999) Vegetated stream riparian zones: their effects on stream nutrients, 558 sediments and toxic substances (Maryland, USA: Smithsonian Environmental Research Center). 559 available at http://www.serc.si.edu/SERC_web_html/pub_ripzone.htm 560 Accessed 18th October 2002. 561 562 Cousins, J.R., Hope, D., Gries, C. and Stutz, J.C. (2003). Preliminary assessment of arbuscular 563 mycorrhizal fungal diversity and community structure in an urban ecosystem, Mycorrhiza, 13, 564 pp. 319-326. 565 566 Faber-Tayler, A., Kuo, F. E. and Sullivan, W.C. (2001) Coping with ADD: The surprising 567 connection to green play settings, Environmental Behaviour, 33, pp. 54-77. 568 569 Farrell, T. A. and Marion, J. L. (1998). An Evaluation of Camping Impacts and their 570 Management at Isle Royale National Park, (Houghton, MI: USDI National Park Service). 571 572 Florgard, C. (2000) Long-term changes in indigenous vegetation preserved in urban areas, 573 Landscape and Urban Planning, 52, pp. 101-116. 574 575 Furuseth, O.J. and Altman, R.E. (1991) Who's on the greenway: socioeconomic, demographic, 576 and locational characteristics of greenway users, Environmental Management, 15, pp. 329-336. 577

23

578 Hoehne, L. M. (1981) The ground layer vegetation of forest islands in an urban-suburban matrix, 579 in: Burgess, R.L. and Sharpe, D.M. (Eds) Forest Island Dynamics in Man-dominated 580 Landscapes, pp. 41-54 (New York: Springer-Verlag). 581 582 Kaplan, S. (1995) The urban forest as a source of psychological well-being, in: Bradley, G.A. 583 (Ed) Urban Forest Landscapes: Integrating Multidisciplinary Perspectives, (Seattle: University 584 of Washington Press). 585 586 Kent, M. and Coker, P. (1992) The description of vegetation in the field, in: Vegetation 587 Description and Analysis: a Practical Approach, pp. 28-76 (London: Belhaven Press). 588 589 Leung, Y. and Marion, J.L. (1999) Spatial strategies for managing visitor impacts in national 590 parks, Journal of Park and Recreation Administration, 17(4), pp. 20-38. 591 592 Loeb, R.E. (1989) The ecological history of an urban park, Journal of Forest History, 33, pp. 593 134-143. 594 595 Lynn, N. and Brown, R.D. (2003) Effects of recreational use impacts on hiking experiences in 596 natural areas. Landscape and Urban Planning, 64, pp. 77-87. 597 598 Malmivaara-Lamsa, M. and Fritze, H. (2003) Effects of wear and above ground forest site type 599 characteristics on the soil microbial community structure in an urban setting, Plant and Soil, 256, 600 pp. 187-203. 601 602 Manning, R E. (1979) Strategies for managing recreational use of national parks, Parks, 4(1), pp. 603 13-15. 604 605 Matlack, G. R. (1993) Sociological edge effects: spatial distribution of human impact in 606 suburban forest fragments, Environmental Management, 17, pp. 829-835. 607 608 Mayer, T., Snodgrass, W.J. and Morin, D. (1999) Spatial characteristics of the occurrence of 609 road salts and their environmental concentrations as chlorides in Canadian surface waters, Water 610 Quality Research Journal of Canada, 34, pp. 545-574. 611 612 McWilliam, W. J. (2009) The Housing/Forest Interface: Structural Approaches for Protecting 613 Suburban Natural Areas Post Development (Saarbrucken, Germany, VDM Verlag Dr. Muller). 614 615 McWilliam, W.J., Eagles, P., Seasons, M. and Brown, R. (2010) The housing/forest interface: 616 testing structural approaches for protecting suburban natural systems following development, 617 Urban Forestry and Urban Greening, 9, pp. 149-159. 618 619 Moran, M A. (1984) Influence of adjacent land use on understory vegetation in New York 620 forests, Urban Ecology, 8, pp. 329-340. 621 622

24

Murphy, S.D. (2006) Why micro-scale urban ecology matters, in: Bunting, T. and Filion, P. 623 (Eds) Canadian Cities in Transition: From the Local to the Global, 3rd. Ed., pp. 379-392 (Don 624 Mills, Ontario: Oxford University Press). 625 626 Nowak, D.J., Crane, D.E. (2002) Carbon storage and sequestration by urban trees in the USA, 627 Environmental Pollution. 116, pp. 381-389. 628 Ouellet, P. (1996) Environmentally sensitive policy areas as tools for environmental protection, 629 in: Filion, P., Bunting, T., and Moss, M.R. (Eds) The Dynamics of the Dispersed City: 630 Geographic and Planning Perspectives on Waterloo Region (Publication Series No. 47), 631 (Waterloo, Ontario: University of Waterloo). 632 633 Reeder, A.L., Ruiz, M.O., Pessier, A., Brown, L.E., Levengood, J.M., Phillips, C.A., Wheeler, 634 M.B., Warner, R.E. and Beasley, V.R. (2005) Intersexuality and the cricket frog decline: historic 635 and geographic trends, Environmental Health Perspectives, 113, pp. 261-265. 636 637 Roggenbuck, J. W. and Lucas, R. C. (1987) Wilderness use and user characteristics: a state of 638 knowledge review, in: Luca, R.C. (comp) Proceedings, National Wilderness Research 639 Conference: Perspectives, State of Knowledge and Future Directions, pp. 204-205, USDA 640 Forest Service General Technical Report INT-220 (Ogden, Utah: Intermountain Research 641 Station). 642 643 Sauvajot, R. M., Buechner, M., Kamradt, D. A. and Schonewald-Cox, C. M. (1998) Patterns of 644 human disturbance and response by small mammals and birds in chaparral near urban 645 development, Urban Ecology, 8, pp. 329-340. 646 647 Scott, K. I., Simpson, J.R., and McPherson, E.G. (1999) Effects of tree cover on parking lot 648 microclimate and vehicle emissions, Journal of Arboriculture, 25(3), pp. 129-142. 649 650 Taylor G. (1992) A Study of Encroachment into and Around Selected Urban ESPAs, (Region of 651 Waterloo Ecological and Environmental Advisory Committee, Waterloo, Canada). 652 653 U.S. Environmental Protection Agency (1995) Riparian Buffer Strategies for Urban Watersheds 654 (Rep. No. 95703), (Washington, D.C: Department of Environment Programs, Metropolitan 655 Washington Council of Governments). 656 657 USDI National Park Service (1997) The Visitor Experience and Resource Protection (VERP) 658 Framework: A Handbook for Planners and Managers (Publication NPS D-1215), ( Denver, Co: 659 USDI National Park Service, Denver Service Center). 660 661 Wells, N.M. (2000) At home with nature: Effects of "greenness" on children's cognitive 662 functioning, Environment and Behavior, 32(6), pp. 775-795. 663