Landscape and Ecological Engineering

, Volume 6, Issue 2, pp 235–245

Landscape scale assessment of stream channel and riparian habitat restoration needs

Authors

    • Department of Natural ResourcesCornell University
  • Mark B. Bain
    • Department of Natural ResourcesCornell University
Original Paper

DOI: 10.1007/s11355-010-0103-6

Cite this article as:
Meixler, M.S. & Bain, M.B. Landscape Ecol Eng (2010) 6: 235. doi:10.1007/s11355-010-0103-6

Abstract

Human modifications of streams and rivers have caused extensive stream channel and riparian degradation. Cost-effective, rapid assessment tools can be used to better manage such areas by identifying the status of habitats for restoration planning and protection. We used a spatially explicit, reach-scale geographic information system modeling strategy to examine stream channel and riparian condition and prioritize restoration actions. The stream channel condition index uses information on land use, road and railroad density, and sinuosity. The riparian condition index uses calculations of percent forest, patch density, and convexity based on land cover in the floodplain. Reaches were classified into restoration categories based on stream channel and riparian condition model results, land ownership, slope, position in the subwatershed, and adjacency to high-quality habitat. We compared modeled restoration priority rankings with those in the management plan for the East Credit subwatershed in Ontario, Canada. Predicted stream channel restoration priority rankings matched field-based classifications for 86% of the reaches in the East Credit subwatershed. Predicted riparian restoration priority rankings matched field-based classifications for 81% of the reaches. Our methods replicate with fairly good accuracy the results obtained using intensive field surveys and stakeholder input. Managers can use these cost-effective strategy development tools to identify candidate reaches for further study and prioritize stream channel and riparian restoration actions over large regions.

Keywords

Rapid assessmentSpatial analysisReach scaleAnthropogenic disturbanceGeographic information systemsDegradation

Introduction

Human modification of streams and rivers is pervasive (Benke 1990; Allan and Flecker 1993). Forestry, municipal, agricultural, and industrial practices caused the removal of riparian vegetation (Petersen 1992; National Research Council 2002) and channelization of lowlands (Petersen 1992), which further resulted in degraded aquatic ecosystem function and altered community structure (Dudgeon et al. 2006; Richardson and Danehy 2007). The negative effects of stream and riparian alteration on water quality, hydrology, benthic invertebrate communities, fisheries resources, and the recreational value of streams are well-documented (Arner et al. 1976; Karr and Schlosser 1978; Poff et al. 1997; Wang et al. 2001; Meador and Goldstein 2003; Wheeler et al. 2005; Meixler et al. 2009).

The pervasive negative impacts on the world’s flowing waters have prompted a greater focus on stream management, conservation, and restoration (Millennium Ecosystem Assessment 2005; Revenga et al. 2005; Dudgeon et al. 2006). Calls have been made for cost-effective, rapid assessment tools that will provide information on habitat status (Goodwin et al. 1997; Tiner 2004).

We know that managing the landscape around streams is vital to maintaining high-quality flowing waterways and restoring degraded ones (Tiner 2004). However, identifying where landscape management could be most effectively applied has been a challenge given the large sizes of the areas of concern (Mollot and Bilby 2008). This problem could be reduced if potential restoration sites could be located using spatial analysis tools to efficiently evaluate stream quality and set priorities for conservation management (Tiner 2004; Mollot and Bilby 2008).

Geographic information systems (GIS) can be used to provide rapid, quantitative, landscape-scale assessments of physical changes in stream and riparian health (Snyder et al. 2005) and can be a more cost-effective way to gather data than traditional field methods. In addition, data can be readily updated, making them effective long-term monitoring tools for natural resource planners (Tiner 2004).

Past studies have used GIS techniques to map riparian buffers (Muller 1997; Narumalani et al. 1997; Hunter et al. 1999; Goetz et al. 2003) and assess in-stream habitat (Hardy 1998; Gilvear 1999; Legleiter 2003; Legleiter and Goodchild 2005; Legleiter and Roberts 2005). However, none of the proposed methods have categorized in-stream and riparian habitat and prioritized restoration actions at the reach scale. This scale is especially relevant in practice because most restoration is conducted on short stream reaches (Roni et al. 2002).

We tested a rapidly applied, practical, and quantitative geospatial method to assess the condition of stream channel and riparian habitats. Our models used easily available GIS data and variables known as major sources of catchment disturbance with potential to alter river processes (Stein et al. 2002). Where human land uses impinged on the stream, we assumed that channel control measures and riparian modifications were likely and that these practices negatively impacted aquatic ecosystems (Roth et al. 1996; Forman 2000; Karwan et al. 2001; Stein et al. 2002). We present a method for prioritizing restoration actions using model results with additional geospatial data and compare modeled stream channel and riparian priority rankings against those in the 2007 management plan for the East Credit subwatershed in Ontario, Canada.

Methods

Study site

The East Credit subwatershed (51 km2) in Ontario, Canada is a small headwater basin of the Credit River watershed, which flows into Lake Ontario. This subwatershed is located approximately 56 km northwest of Toronto. No towns exist within the subwatershed, although the villages of Caledon East, Inglewood, and Caledon are all within 2 km of the subwatershed boundary. The subwatershed is characterized by broad floodplains, flat channel gradients, and high recharge. Overall water quality is considered good (Credit Valley Conservation 2007b) despite the intensive and nonintensive agricultural (45%) and urban/rural development (7%) activities present in the area (Credit Valley Conservation 2007a; Fig. 1). The East Credit subwatershed has a significant number of forest patches with interior habitat (40%) and many wetlands (8%), including one Provincially Significant Wetland and one Earth Science Area of Natural and Scientific Interest. This subwatershed is dominated by two significant intersecting topographical features, the Oak Ridges Moraine and the Niagara Escarpment, both of which support a diverse and healthy terrestrial wildlife system.
https://static-content.springer.com/image/art%3A10.1007%2Fs11355-010-0103-6/MediaObjects/11355_2010_103_Fig1_HTML.gif
Fig. 1

Land cover, roads, and location of the East Credit subwatershed, Ontario, Canada (Sources: Credit Valley Conservation 2007a and Google Maps)

The East Credit subwatershed is a relatively healthy and productive system. However, in some areas, the impacts of human land uses are visible and the environmental features, functions, and linkages are showing early warning signs of instability, including impaired water quality and degraded stream channels. Further, Credit Valley Conservation (CVC) noted the potential for increasing developmental pressure in the subwatershed (Credit Valley Conservation 2007a). Consequently, this subwatershed has been targeted to receive a range of restoration activities to improve subwatershed health and prevent further deterioration of the ecosystem. Intensive characterization and management plans were developed for this subwatershed by the CVC to identify stream restoration opportunities and guide implementation efforts. We chose this area for model application and verification due to the abundance of readily available data and the potential usefulness of the model results.

Stream channel and riparian condition assessment

We developed our multimetric assessment tool to evaluate stream channel and riparian condition based on a set of reach-level environmental properties. Each measure of condition is comprised of a set of indices meeting the following criteria: (1) applicable over large geographic areas, (2) derived from widely available geospatial data, (3) rapidly assessed, and (4) efficiently and cost effectively updated with newer data. Such indices are by nature coarse filters suitable for assessing the overall condition of stream channel and riparian habitats at the reach scale. As such, they were intended to augment, not supplant, more rigorous, fine-filter approaches for describing stream channel and riparian condition such as indices of biotic integrity and field studies (Tiner 2004).

Two separate indices were developed using information in recent literature for stream channel and riparian condition. The stream channel condition index (SCCI) uses land use, infrastructure, and sinuosity data as indicators of stream channel degradation. The riparian condition index (RCI) uses calculations of percent forest, patch density, and convexity to assess the quality of riparian habitat. Selection of variables to assess stream channel and riparian condition was based on their utility for landscape assessment and their sensitivity to the influences of human activities on fish and wildlife habitat and water quality (Trani and Giles 1999; Jones et al. 2001; Tiner 2004; Mattson and Angermeier 2007). Both indices have a maximum value of one and a minimum value of zero, with higher values indicating greater disturbance.

The following sources of digital geospatial data were used in the development of the indices: (1) 1999 polygon land cover data, (2) 10 m digital elevation models (DEMs), (3) roads, (4) railroads, (5) 1:100,000-scale streams, and (6) drainage basins delineated for each reach. All spatial data were obtained from the Ontario Ministry of Natural Resources. Land cover classes were aggregated and reclassified to improve the accuracy of individual land-cover categories (Zhu et al. 2000) into: (1) urban land, (2) agricultural land, (3) forests, (4) wetlands, (5) transitional land (moving toward some type of development or agricultural use, but future status unknown), and (6) water (Anderson et al. 1976). Model assessments were performed for each reach defined as the section of stream between two confluences or from headwaters downstream to the first confluence.

Stream channel condition index

We calculated the SCCI following a modification of Mattson and Angermeier (2007) and Stein et al. (2002) using an Arc Macro Language (AML) program in ArcGIS 9.2 (Fig. 2). We used this index as a surrogate for the pervasiveness of human alteration in the valley and floodplain of each reach, as stream channel modifications and sedimentation are commonly associated with development (Mattson and Angermeier 2007) and result in an alteration in habitat dynamics and creation of new conditions to which the native biota may be poorly adapted (Poff et al. 1997). The SCCI is:
$$ {\text{SCCI}} = \left( {14({\text{UA}}) + 14({\text{AA}}) + 9({\text{RD}}) + 9({\text{RR}}) + 10({\text{S}})} \right)/148 $$
where UA and AA represent the percent urban and percent agricultural areas in the drainage basin (Roth et al. 1996; Allan et al. 1997) classified into one of four (0–3) degradation categories using the following thresholds for urban (<1.9, 1.9–10, 10–30, >30%) and agricultural (<1.9, 1.9–40, 40–50, >50%) areas (Allan 2004; Mattson and Angermeier 2007). RD and RR represent road and railroad density (km/km2) within a 200-m buffer of the stream classified into one of four (0–3) categories used to denote the degree of infrastructure-related habitat degradation: 0, <0.1068, 0.1069–0.1622, >0.1622 (Mattson and Angermeier 2007). Sinuosity, S, is the ratio of the length of a reach divided by the straight-line distance between its start and end, thus a value of one indicates a straight, likely channelized, stream. Sinuosity values <1.05 are associated with poor-quality habitat (Wang et al. 1998), so we used this threshold to classify sinuosity measures into one of two categories (0, 1). This threshold is less than the average sinuosity of the subwatershed (1.12; Credit Valley Conservation 2007a). The values 14, 9, and 10 weight each factor based on the levels of severity associated with each threat (Mattson and Angermeier 2007). We divided the final score by 148 to place the metric on a scale from 0 to 1, with higher values indicating greater severity. Using equal intervals, resulting scores were classified as poor, fair, or good.
https://static-content.springer.com/image/art%3A10.1007%2Fs11355-010-0103-6/MediaObjects/11355_2010_103_Fig2_HTML.gif
Fig. 2

Flowchart for stream channel condition index. Dotted boxes contain weights

Riparian condition index

Riparian vegetation provides several critical functions to streams by filtering pollutants, detaining flows, contributing organic material for habitat and food, moderating water temperature, slowing or preventing excessive stream-bank erosion, and supplying important habitat areas and movement corridors for many terrestrial flora and fauna species. Higher forest fragmentation has been shown to closely correlate with a shift in the fish community toward species that are more able to tolerate warmer temperatures and higher silt levels (Brazner et al. 2005).

Again using an AML in ArcGIS 9.2 (Fig. 3), we computed the RCI for each reach by calculating the percent forest in the drainage basin, patch density, and convexity fragmentation metrics following Trani and Giles (1999). We used a variable width buffer (VWB) to define the extent of the riparian zone from which we computed these metrics. VWBs have the advantage of accounting for regional differences in physical and ecological conditions and in overcoming the difficulty in reaching consensus on a politically acceptable standard minimum buffer distance (Xiang 1993). VWBs also define an area large enough to attain expected runoff water filtration without incorporating unnecessary land (Lin et al. 2002). We modeled the extent of the VWB by estimating the area of the active 50-year flood inundation zone from DEMs while maintaining a 30-m minimum buffer around all streams (modified from Wenger 1999). The RCI, calculated in the VWB is:
$$ {\text{RCI = }}(I_{\text{pd}} + I_{\text{conv}} )/2 $$
where Ipd, the index of patch density, was calculated as the number of patches within the VWB divided by the total landscape area (ha) of the VWB surrounding each reach, and Iconv, the index of convexity, was calculated as the total perimeter of forest patches in the VWB divided by the total area of forest patches in the VWB. Patches represent discrete areas of relatively homogeneous environmental conditions at a particular scale. If no forested patches existed in the VWB of each reach, the RCI was based solely on the Ipd. Both patch density and convexity are variables with no upper limit. To place these variables on a scale from 0 to 1, we classified all reaches with Ipd >0.451 or Iconv >0.056 [values associated with fragmentation (Trani and Giles 1999)] as 1, then summed the counts and divided by 2. Thus, higher scores indicate greater fragmentation. Resulting scores of 0, 0.5, and 1 were classified as good, fair, and poor, respectively.
https://static-content.springer.com/image/art%3A10.1007%2Fs11355-010-0103-6/MediaObjects/11355_2010_103_Fig3_HTML.gif
Fig. 3

Flowchart for riparian condition index

Using results to prioritize management actions

For this screening tool to be most useful to managers, reaches need to be prioritized in terms of suitability for restoration (Beechie et al. 2008). Restoration to a completely natural state is not expected; however, moderate restoration activities such as riparian buffer installations, road removal, and natural channel design (NCD) have proven effective at improving water quality and clarity (Osborne and Kovacic 1993; Barling and Moore 1994; Parkyn et al. 2003), enhancing channel stability (Parkyn et al. 2003), reducing erosion (Switalski et al. 2004), and rehabilitating the structure, function, and health of resident fish populations (Zika and Peter 2002; Baldigo et al. 2008). Restoration efforts have also proven effective at increasing fish and macroinvertebrate habitat and community structure through reconnection of isolated habitats and in-stream habitat improvements (Muotka et al. 2002; Zika and Peter 2002; Roni et al. 2008). Even moderately degraded urban streams can achieve partial recovery of physical integrity if riparian buffers can be reforested and road crossings eliminated (McBride and Booth 2005).

We followed a set of guidelines based on current river restoration literature to identify and prioritize reaches for restoration. Recent research indicates that restoration should focus on reconnecting isolated high-quality habitats (Roni et al. 2002), as long buffer lengths are linked to reduced water temperature and improvements in invertebrate community health, as is rehabilitation of headwater tributaries (Parkyn et al. 2003). Only riparian areas with slopes of 1–10% are considered suitable for vegetative buffer strips, as runoff from slopes >10% will tend to flow through the buffers too quickly, reducing trapping efficiency, and runoff on slopes <1% will likely pool (Hayes and Dillaha 1992). Publicly owned spaces should be prioritized over privately owned areas, as research has shown that despite people’s preferences for heavily vegetated riparian habitats, there is a discrepancy between what people prefer and how they manage their own land. This discrepancy is due to lack of knowledge about what actions are permissible along streams, fear of potentially negative impacts from making changes to their properties, cost, and time (Shandas 2007).

Table 1 shows how the above criteria can be used as a guide to identify and rank reaches for restoration. Criteria in Table 1 assume the user has the model results—termed stream channel condition and riparian condition—and spatial overlays of public lands (e.g., national and state parks, fish and wildlife refuges) and slope (available from DEMs). Adjacency to high-quality habitats (e.g., intact riparian habitat) and position in subwatershed (e.g., headwater reach) can be determined visually from maps. Reaches with stream channel and riparian habitats in fair condition were identified as potential restoration sites with high priority. Fair quality stream channels on public land, adjacent to high-quality habitat or in the tributary headwaters were ranked to receive special priority, with extra special priority given to reaches meeting two or more of these criteria. Reaches with fair-quality riparian condition with slopes between 1% and 10% were ranked to receive special priority, with extra special priority given to reaches meeting one of more of the following criteria: on public land, adjacent to high-quality habitat, or in the tributary headwaters of the subwatershed. Reaches with stream channel and riparian habitats in good condition were identified as low priority given their already superior condition. Reaches with poor stream channel and riparian condition were listed as medium priority, though riparian areas with slopes between 1% and 10% were given high priority for their suitability to receive vegetative buffer strips.
Table 1

Prioritization criteria for restoration of reaches

Restoration priority level

Criteria

Related references

High

Stream channel condition = fair

 

 Priority (a) for 1 condition met,

 

 Extra priority (b) for 2+ conditions met if also:

 

   Along a headwater reach

Parkyn et al. (2003)

   Adjacent to intact riparian habitat

Roni et al. (2002)

   On public land

 

High

Riparian condition = fair

 

 Priority (a) if slope is 1–10%

Hayes and Dillaha (1992)

 Extra priority (b) if also:

 

   Along headwater reach

Parkyn et al. (2003)

   Adjacent to intact riparian habitat

Roni et al. (2002)

   On public land

 

Medium

Stream channel condition = poor

 

Medium

Riparian condition = poor

 

 High priority if slope is 1–10%

Hayes and Dillaha (1992)

Low

Stream channel condition = good

 

Low

Riparian condition = good

 

Model verification

We compared our modeled restoration priority rankings with recommendations from the CVC Phase III Management Plan for the East Credit Subwatershed (Credit Valley Conservation 2007b) to determine whether these two independent methods would prioritize the same locations for restoration. Comparisons were performed for both stream channel and riparian condition.

The CVC evaluated subwatershed condition through field investigation in the summer and fall of 2002 following procedures set forth in the Credit Watershed Natural Heritage Project Detailed Methodology: Identifying, Mapping and Collecting Field Data at Watershed and Subwatershed Scales, Version 3 (Credit Valley Conservation 1998). Using these procedures, the CVC evaluated hydrogeology, hydrology, terrestrial and aquatic biological communities, stream geomorphology, and water quality. Specific to stream channel condition, 13 reaches were evaluated through the application of two channel-assessment techniques: rapid geomorphic assessment (RGA) and rapid stream assessment technique (RSAT). The Ontario Ministry of Environment (1999) developed the RGA to assess reaches in urban channels. RGAs were undertaken for each reach and involved recording evidence of channel instability, such as exposed tree roots, undercutting, and presence of chutes. Any observations regarding the presence or absence of these indicators were divided into the following categories of adjustment: aggradation, degradation, channel widening, and planimetric form. The instability features for each category were tallied and used to calculate a reach stability index, which corresponds to a stability classification.

The second assessment involved application of the RSAT (Galli 1996), a broader, more qualitative, assessment of the overall health and function of the reach. The RSAT ranks each reach based on six indicators: channel stability, scour/deposition, in-stream habitat, water quality, riparian conditions, and biological indicators, such as the abundance of benthic invertebrates. Once each indicator had been ranked numerically, scores were totaled to give an overall rank. Lower values indicate a stream with relatively poor health, whereas higher values represent a relatively rich, healthy, functional stream. Also included in the RSAT are general observations of channel dimensions, such as bankfull width and depth, substrate size, bank height, vegetation cover, channel hardening, and other disturbances.

The CVC then used integration to examine the system and structure and arrive at an understanding of its function. In this process, the major structural elements (geology, terrain, major water bodies) at work in this subwatershed were analyzed. The structural elements were then used to develop a conceptual model of the water movements (recharge, discharge, groundwater flow, and surface-water flow) and the implications to resources in the watershed. This conceptual understanding then used an analysis of results from other disciplines to create an understanding of how this particular subwatershed functions. The key functions were then used to determine appropriate management strategies to improve any degraded areas of the watershed.

Final restoration prioritization recommendations from these analyses were compiled in the Phase III Management Plan for the East Credit Subwatershed (Credit Valley Conservation 2007b). We compared our model restoration prioritization recommendations for stream channel and riparian habitats to the management recommendations in the Phase III report by counting the number of matches and mismatches. Reaches were modeled at a finer resolution than those in the field-based prioritization recommendations. Reaches were matched at the resolution used by the model; thus, some cases exist in which a single field-based recommendation was matched with several different modeled reach recommendations. A sign test was used to compute the probability of obtaining results by chance, and these results were used to judge our confidence in our findings.

Results

Model results predicted that all but two of the 35 reaches in the East Credit subwatershed had fair-quality stream channels; thus, the majority of reaches were classified by the model as high-priority candidates for stream channel restoration (Table 2; Fig. 4). Sixteen of the high-priority reaches were classified by the model to receive special priority, and two were recommended for extra special priority given their location on public land, adjacency to high-quality habitat, or placement in the tributary headwaters. The remaining two reaches with predicted good stream channel condition were classified as low priority.
Table 2

Restoration priority ranking of reaches in the East Credit subwatershed, Ontario, Canada

Stream ID

Stream channel condition

Riparian condition

On public land

Headwater reach

Adjacent to high-quality habitat

Slope (%)

Model stream channel priority ranking

Model riparian priority ranking

CVC stream channel priority ranking

CVC riparian priority ranking

1

Fair

Poor

N

Y

N

1–10

Higha

High

High

High

2

Fair

Fair

N

Y

N

>10

Higha

High

High

High

3

Fair

Fair

N

Y

N

1–10

Higha

Highb

High

High

4

Fair

Fair

N

Y

N

1–10

Higha

Highb

High

High

5

Fair

Fair

N

Y

N

>10

Higha

High

High

High

6

Fair

Fair

N

N

N

1–10

High

Highb

High

High

7

Fair

Fair

N

N

N

>10

High

High

High

High

8

Fair

Fair

N

Y

N

1–10

Higha

Highb

High

Low

9

Fair

Fair

N

Y

N

1–10

Higha

Highb

High

High

10

Fair

Fair

N

N

N

<1

High

High

High

High

11

Fair

Good

N

Y

N

<1

Higha

Low

High

N/A

12

Fair

Poor

N

Y

N

1–10

Higha

High

High

High

13

Fair

Fair

N

N

N

1–10

High

Higha

High

High

14

Fair

Fair

N

Y

Y

1–10

Highb

Highb

High

Medium

15

Fair

Fair

N

Y

N

1–10

Higha

Highb

High

High

16

Fair

Fair

N

N

N

1–10

High

Higha

High

High

17

Fair

Fair

N

N

N

1–10

High

Higha

High

High

18

Fair

Good

N

N

N

<1

High

Low

High

High

19

Fair

Fair

N

N

N

1–10

High

Higha

High

High

20

Fair

Poor

N

Y

N

1–10

Higha

High

High

High

21

Fair

Poor

N

Y

N

<1

Higha

Medium

High

High

22

Fair

Fair

N

N

N

1–10

High

Higha

High

High

23

Fair

Fair

N

N

N

1–10

High

Higha

High

High

24

Fair

Fair

N

N

N

<1

High

High

High

High

25

Fair

Fair

N

Y

N

<1

Higha

High

Medium

N/A

26

Fair

Poor

N

Y

N

1–10

Higha

High

High

High

27

Fair

Fair

N

Y

N

<1

Higha

High

High

High

28

Fair

Fair

N

N

N

1–10

High

Higha

High

High

29

Fair

Fair

N

Y

N

<1

Higha

High

Medium

N/A

30

Fair

Fair

N

N

N

<1

High

High

High

High

31

Fair

Fair

N

N

N

1–10

High

Higha

High

High

32

Fair

Fair

N

Y

Y

<1

Highb

High

Medium

N/A

33

Fair

Fair

N

N

N

<1

High

High

High

High

34

Good

Poor

N

N

N

>10

Low

Medium

High

High

35

Good

Poor

N

N

N

<1

Low

Medium

High

High

Stream channel and riparian condition were calculated using the stream channel condition and the riparian condition indices. Model stream channel and riparian priority rankings were based on columns 2–7 and follow the algorithm in Table 1. Credit Valley Conservation (CVC) priority rankings were from the East Credit subwatershed study: Phase III Management Plan and Implementation Report, Fig. 4.2.9 (Credit Valley Conservation 2007b)

aPriority candidates for restoration

bExtra priority candidates for restoration

https://static-content.springer.com/image/art%3A10.1007%2Fs11355-010-0103-6/MediaObjects/11355_2010_103_Fig4_HTML.gif
Fig. 4

Modeled stream channel and riparian restoration priority rankings in East Credit subwatershed, Ontario, Canada

More variation existed in modeled riparian condition, with seven reaches predicted to be in poor condition, 26 in fair condition, and two in good condition. All reaches in fair condition and four in poor predicted riparian condition were classified as high-priority candidates for riparian restoration (30 in total). Nine of the high-priority reaches were classified by the model to receive special priority given their advantageous slope for vegetative buffer strips. Six additional high-priority reaches were classified to receive extra special priority given their advantageous slope and location on public land, adjacency to high-quality habitat, or placement in the tributary headwaters. The remaining three reaches in poor predicted riparian condition were classified as medium priority and those with good predicted riparian condition were classified as low priority. None of the reaches with predictions of high-quality stream channels had low predicted riparian health and vice versa.

The model recommendations for stream channel restoration matched those of the CVC for all but five reaches (86%; Table 2). The probability of obtaining 30 matching recommendations out of 35 reach comparisons was <0.0001, indicating that the high rate of matches is a significant and highly confident result. The CVC classified three reaches as medium priority, whereas the model recommended high priority; and in two cases, the CVC classified reaches as high priority although the model classified them as low priority. Thus, in only two cases were the recommendations of the two approaches off by two category steps.

The riparian restoration recommendations of the model matched those of the CVC for all but six reaches (81%; four reaches were unclassified by the CVC; Table 2). The probability of obtaining 25 matching recommendations out of 31 reach comparisons was 0.0009, indicating the high rate of matches is a significant and highly confident result. The CVC classified four reaches as high priority, but the model recommended medium or low priority. In two cases, the model classified reaches as high priority although the CVC classified them as low or medium priority. Thus in only two cases were the recommendations of the two approaches off by two category steps.

Discussion

Human modifications of streams and rivers have caused extensive stream channel and riparian degradation and prompted a greater need for cost-effective, landscape-scale, rapid assessment tools to provide information on the status of habitats for restoration prioritization. We described GIS models of stream channel and riparian condition, ranked reaches using our prioritization method, and compared our modeled rankings against those in the 2007 management plan for the East Credit subwatershed in Ontario, Canada. Our methods replicate with fairly good accuracy the results obtained using intensive field surveys and stakeholder input.

Modeled stream channel and riparian restoration prioritization results indicate that most reaches in the East Credit subwatershed are in need of restoration. This modeled result is consistent with CVC recommendations. Agriculture is the predominant land use practice in the East Credit subwatershed, and early settlers cleared much of the area during the nineteenth century (Credit Valley Conservation 2007a). Today, 45% of the land in the subwatershed is used for agriculture, and the riparian corridors in these areas are quite minimal (Credit Valley Conservation 2007a). Such agricultural practices, coupled with road and railroad infrastructure bisecting or following the stream channels, have resulted in channel control structures and a dearth of forested riparian habitat throughout the subwatershed, as reflected in the modeled and field-based restoration recommendations.

We attempted to locate subwatersheds with more variation in field-based restoration recommendations but found that recommendations were similarly high in all subwatersheds. Intensive characterization and management plans are likely developed primarily for subwatersheds experiencing development pressure and with funding allocated to undertake future restoration activities, thereby biasing plans toward abundant high-priority sites needing restoration.

Restoration efforts should focus first on restoring hydrologic, geologic (sediment delivery and routing), and riparian processes through road decommissioning and restoration of riparian areas before in-stream habitat enhancement, such as channel reconfiguration, is attempted (Roni et al. 2002; McBride and Booth 2005). Conversely, heavily disturbed systems should be given lowest priority, as they may not recover without massive human intervention (Goodwin et al. 1997). Restoration actions include the use of riparian vegetative buffer strips (Osborne and Kovacic 1993; Barling and Moore 1994; Parkyn et al. 2003), reducing patch perimeter–area ratio (Lee and Thompson 2005), road decommissioning (Roni et al. 2002; Switalski et al. 2004), NCD, and channel reconfiguration (Roni et al. 2002; Parkyn et al. 2003).

All model-predicted candidate sites for restoration need to be field-checked and must account for the degree of community support, quality of surroundings and upland areas, and the problem being mitigated. It is well known that personal values based on historical, social, and cultural beliefs affect conservation practices (Shandas 2007). Further, maintenance time and labor, local government land use regulations, and cost may act as constraints on some restoration actions (Shandas 2007). Therefore, understanding what residents value and accounting for common obstacles may help to increase restoration success.

In several places, restoration recommendations differed between modeled and field-based methodologies. This may have resulted in part from discrepancies we found between the digital 1999 land use/land cover metrics and those cited in the Phase I characterization report for the East Credit subwatershed (Credit Valley Conservation 2007a). The report cites that agriculture makes up 45% of the subwatershed, urban/rural development 7%, forest 40%, and wetlands 8%, whereas the older digital land-use data show that agriculture makes up 74% of the subwatershed, urban/rural development <1%, forest 25%, and wetlands 1%. In addition, the report shows intensive agriculture concentrated more heavily in the southern portion of the subwatershed, whereas digital land-use data show agriculture dispersed throughout. These discrepancies may in part be the result of changes in population and land-use management between 1999 and 2007. The population in the area increased between the late 1990s and early 2000s (Credit Valley Conservation 2007a), causing a shift in land use toward urban/rural development. Additionally, reforestation efforts practiced in the area have been responsible for converting intensive agricultural lands back into forests (Credit Valley Conservation 2007a). This may account for part of the difference between the high agricultural and low forested percentages in the 1999 digital data and the lower agricultural and higher forested percentages cited in the Phase I characterization report for the East Credit subwatershed. The discrepancies may also be the result of the quality of digital data and handling of field-based recommendations. Digital data are subject to attribution error, coarseness of resolution, and temporal aging. Further, field verification of land use was performed on data used for the report, although not for modeled data. The differences between modeled and field-based prioritization recommendations may also have been the result of stakeholder feedback at meetings. Although no instances of stakeholder influence were mentioned in the reports, such feedback could have affected the final recommendations. All these factors would have a substantial impact on the model’s ability to correctly match the field-based restoration recommendations.

Though field surveys are necessary before restoration actions are undertaken, there are inherent drawbacks in their practice. Survey techniques are often subjective, requiring the field technician to make judgments about the extent of degradation. Further, the field of view at the site is often limited, possibly obscuring details such as degree of fragmentation and extent of forested region more easily detected from the air. Stakeholder input can be used to overcome some of these drawbacks, but participation is sometimes difficult to obtain, and personal interest may interfere with objectivity.

The methodological framework used here offers several advantages over these approaches to restoration-site selection and planning. First, the models were built using known relationships between physical properties, such as land use and channel condition, instead of location-based empirical data, enabling them to be broadly applied at the landscape scale. Second, variables in the indices of the models can be tailored or updated to fit specific uses or newer data (e.g., addition of gradient to assess fish habitat, as in Mollot and Bilby 2008). Third, the quantitative nature of the models may allow for greater objectivity than other restoration prioritization approaches. Finally, model-based stream and riparian assessments can be accomplished with lower cost, time, and labor requirements over other commonly used assessment procedures.

It is clear that the cost-effective, landscape-scale modeling procedures presented here can be used by managers, with appropriate field verification of top candidates, to evaluate stream channel and riparian condition and identify candidate reaches for restoration prioritization.

Acknowledgments

This study was supported by the Great Lakes Protection Fund under Grant No. 766. Model verification was accomplished using information from Credit Valley Conservation, and we thank them for their willingness to share their data and resources. We also specifically thank Adrienne Ockenden and Francoise Vermeylen for generously contributing their time and effort in support of this project. All experiments described in this paper comply with the current laws of the country in which they were performed.

Copyright information

© International Consortium of Landscape and Ecological Engineering and Springer 2010