BUGS IN THE GOLF STREAM: BENTHIC MACROINVERTEBRATE COMMUNITY RESPONSE TO DISTURBANCE FROM A RESTORATION PROJECT IN FUNDY NATIONAL PARK, NEW BRUNSWICK.

Size: px
Start display at page:

Download "BUGS IN THE GOLF STREAM: BENTHIC MACROINVERTEBRATE COMMUNITY RESPONSE TO DISTURBANCE FROM A RESTORATION PROJECT IN FUNDY NATIONAL PARK, NEW BRUNSWICK."

Transcription

1 BUGS IN THE GOLF STREAM: BENTHIC MACROINVERTEBRATE COMMUNITY RESPONSE TO DISTURBANCE FROM A RESTORATION PROJECT IN FUNDY NATIONAL PARK, NEW BRUNSWICK. by Noel Richard Swain (u8j2t@unb.ca) A thesis submitted in partial fulfillment of the requirements for the degree of Bachelor of Science with Honours in Biology Department of Biology The University of New Brunswick April, 2007

2 Abstract Restoration projects in channelized streams using man-made structures to increase habitat heterogeneity and improve aquatic biodiversity are well documented in boreal regions in both Europe and North America. A great deal of associated research has been conducted in many of these streams to determine the effects or actual benefit of such restoration work on aquatic communities with a general focus on game fish conservation and to a lesser extent stream benthos. Such a study was conducted on Dickson Brook, a third order New Brunswick stream with reaches running through forested areas and channelized reaches running through a golf course. An ongoing restoration project was initiated in 2003 on channelized reaches within the golf course involving the removal of deteriorated bank structures and the creation of a more meandering riffle : pool channel profile similar to that of less impacted upstream reaches. The aim of this study was to assess the biological response to disturbances associated with the restoration by comparing benthic macroinvertebrate communities sampled from channelized and reconfigured reaches running through the golf course, and a forested less impacted reach. Benthic communities showed high resilience to disturbance; in June, overall invertebrate abundance was highest in an undisturbed, channelized reach and the older disturbed reach, roughly 2 X higher than that in the forested and recently disturbed reaches; September abundance levels were almost twice that of June and were characterized by similar abundance in the forested and two reconfigured reaches, but roughly 2 X higher in the undisturbed channelized reach. Large variation in relative abundance was observed between families of invertebrates, but was primarily driven by high chironomid and mayfly abundance. ii

3 Table of Contents LIST OF FIGURES......iv LIST OF TABLES vi LIST OF APPENDICES.....vii ACKNOWLEDGMENTS..... viii INTRODUCTION MATERIALS AND METHODS DESCRIPTION OF STUDY AREA FIELD SAMPLING...14 SAMPLING PROCEDURE.. 14 SUB-SAMPLING.. 15 SAMPLE PROCESSING STATISTICAL ANALYSIS ENVIRONMENTAL MEASUREMENTS...19 RESULTS. 19 OVERALL ABUNDANCE BENTHIC MACROINVERTEBRATE COMMUNITY STRUCTURE.. 20 JUNE SEPTEMBER.. 22 TEMPORAL CONSISTENCY IN MACROINVERTEBRATE COMMUNITY PATTERNS...24 ENVIRONMENTAL MEASUREMENTS: WATER TEMPERATURE AND FLOW DISCUSSION...26 REFERENCES...51 iii

4 List of Figures Figure 1. Map of Fundy Park, NB showing study reaches of Dickson Brook and surrounding area Figure 2. Modified aerial photograph of Dickson Brook running through the Fundy Park golf course, Study reaches are distinguished by colour, sampled riffles are shown as red dots, and temperature recorder locations are shown as purple stars. GPS was used to find coordinates of all boundaries, sample, and temperature locations shown here. Direction of flow is from left to right of photograph Figure 3. Photographs of different reach types in Dickson Brook, Fundy Park, NB. a) 2004 restored; b) 2005 restored; c) Control; d) Eroded gabion baskets on bank; e) Forested reference; f) Newly restored tributary Figure 4. Total pooled abundance of benthic macroinvertebrates from all riffle samples (24) in Dickson Brook, Fundy Park, NB for two sampling periods in 2006; June 8 th and September 14 th...37 Figure 5. Total abundance of common benthic macroinvertebrates from 24 riffle samples in Dickson Brook, Fundy Park, NB in two sampling periods in 2006: June 8 th and September 14 th...38 Figure 6. Average abundances of benthic macroinvertebrates in four reaches; reference and control reaches, and two restored reaches of Dickson Brook, Fundy Park on June 8 th From left to right in each reach: Total benthic macroinvertebrate abundance, total Ephemeroptera, Plecoptera, and Trichoptera (EPT) abundance, total Diptera abundance. Error bars = 1-SD..39 Figure 7. Average abundances of benthic macroinvertebrates in four reaches; reference and control reaches and two restored reaches of Dickson Brook, Fundy Park on September 14 th, From left to right in each reach: Total benthic macroinvertebrate abundance, total Ephemeroptera, Plecoptera, and Trichoptera (EPT) abundance, total Diptera abundance. Error bars = 1-SD...40 Figure 8. Multi-dimensional scaling (MDS) bubble plots comparing relative similarities and benthic macroinvertebrate abundances of 24 samples from each of two sampling periods: June 8 th and September 14 th, Plots show relative similarity (distance between samples) and total benthic invertebrate abundance. Each point represents a single benthic sample; the closer points are the more similar they are, and relative size of bubbles corresponding to abundance. 4 = 2004 restored, 5 = 2005 restored, C = control, and R = reference reaches. Arrow indicates highest abundance (1018 individuals) of all samples...41 iv

5 Figure 9. Water temperatures for Dickson Brook taken from four temperature recorders in golf course reaches of the stream between June 1 st and October 18 th, Min., max., and Mean water temperatures for entire period were 7.0, 17.5, and 9.2 C respectively. Mean water temperature for June 8 th sampling period was 8.7 C and that for September sampling period was 12.3 C.42 Figure 10. August 2002 water temperatures for three tributaries and the upstream reach of Dickson Brook, Fundy Park NB. Canopied upstream reach and tributary one show relatively constant temperature ranging around 6 C and 8 C respectively, while tributary 2 and 3, both running through the golf course, show much higher and more variable temperatures (ranging between C), similar to those of the main stem reaches running through the golf course Figure 11. Hydrograph showing mean daily discharge (m 3 s -1 ) for Dickson Brook for the periods, of and between two sampling periods in Flow was measured at a hydrometric gauge station located below the Wolfe Point Road at the lower margin of the golf course...44 v

6 List of Tables Table 1. Total abundance of invertebrate families from sub-sampled riffle grabs from four reaches in two sampling periods, June 8 th and September 14 th..45 Table 2. Analysis of similarity (1-way ANOSIM) for benthic macroinvertebrate communities between reaches in June. Values obtained by ANOSIM are global R statistic, pair-wise comparison R statistic and significance level (p-value)..46 Table 3. Average abundance of benthic macroinvertebrate families contributing > 5% to pair-wise dissimilarity percentages analysis (SIMPER) for June using reach as sample grouping factor. Percent contribution is the relative influence of invertebrate families to the overall differences in community structure between reaches. Average abundance is for 15cm 2 substrate area. Higher abundances and greater differences in relative abundance increase the contribution of families to the overall dissimilarity between pair-wise groups Table 4. Analysis of similarity (1-way ANOSIM) for benthic macroinvertebrate communities between reaches in September. Values obtained by ANOSIM are global R statistic, pair-wise comparison R statistic and significance level (pvalue)...48 Table 5. Average abundance of benthic macroinvertebrate families contributing > 5% to pair-wise dissimilarity percentages analysis (SIMPER) for September using reach as sample grouping factor. Percent contribution is the relative influence of invertebrate families to the overall differences in community structure between reaches. Average abundance is for 15cm 2 substrate area. Higher abundances and greater differences in relative abundance increase the contribution of families to the overall dissimilarity between pair-wise groups Table 6. Average abundance of benthic macroinvertebrate families contributing > 5% to pair-wise dissimilarity percentages analysis (SIMPER) between June and September using date as sample grouping factor. Percent contribution is the relative influence of invertebrate families to the overall differences in community structure between reaches. Average abundance is for 15cm 2 substrate area. Higher abundances and greater differences in relative abundance increase the contribution of families to the overall dissimilarity between pair-wise groups...50 vi

7 List of Appendices Appendix 1. Relative differences in benthic macroinvertebrate community structure between sampling periods and stream reaches within sampling periods compared using dissimilarity percentages analysis (SIMPER). Average dissimilarity percentage values represent the relative dissimilarity between pair-wise sample groups (date or reach) where difference increases with higher values. 53 Appendix 2. Percent contribution and average abundance of benthic macroinvertebrate families from pair-wise dissimilarity percentages analysis (SIMPER) for June (a) and September (b) sampling periods using reach as the grouping factor, and for the entire data set using date as the grouping factor (c). Percent contribution is the relative influence of invertebrate families to the overall differences in community structure between reaches. Average abundance is for 15cm 2 substrate area. Higher abundances and greater differences in relative abundance increase the contribution of families to the overall dissimilarity between pair-wise groups Appendix 3. Multi-dimensional scaling (MDS) bubble plots comparing relative similarities and benthic macroinvertebrate abundances of 24 samples from each of two sampling periods: June 8 th and September 14 th, Plots show relative similarity (distance between samples) and total benthic invertebrate abundance. Each point represents a single benthic sample. Points that are closer together are more similar in benthic macroinvertebrate abundance and family composition. Bubble size indicates abundance. 4 = 2004 restored, 5 = 2005 restored, C = control, and R = reference reaches: a, b) total BI abundance [EPT + Diptera]; c, d) total EPT abundance; e, f) total Ephemeroptera abundance; g, h) total Chironomidae abundance. MDS level of stress, sample, and number of invertebrates represented by largest bubbles are shown in upper right hand corner of each plot 58 Appendix 4. GPS coordinates for all sampled sites and temperature recording stations in Dickson Brook. 1) GPS coordinates for all sampled sites in Dickson Brook, Fundy Park, NB. 2) GPS coordinates for four temperature recording stations in the golf course reaches of Dickson Brook.59 vii

8 Acknowledgements I would first like to thank my supervisor, Rick Cunjak for his good-natured guidance, support and patience, the opportunity to work in his lab and conduct this research, and for reinforcing my interest in ecology. I would also like to thank all those in the Environment Canada lab who so generously lent me their equipment and support. In particular, I would like to thank Kristy Heard and Alexa Alexander for showing me the ropes of benthic invertebrate processing, their guidance, advice, and above all patience for my barrage of questions and concerns throughout the project. Also, thankyou to Alexa and Douglas Swain for help with my statistical analysis and interpretation. To all the Fredericton CRI students and the UNB biology department, thank you for your interest, help, and friendship throughout my project. Thanks also to Bria Crouch for keeping me grounded and putting up with me throughout the process. I would also like to thank Jane Watts, Kim Wen, and all those in Fundy Park, as well as Bob Newbury of Newbury Hydraulics for all their help, and for providing the opportunity to conduct this research. Thanks also to my examiners Joseph Culp and Simon Courtenay for their support. Finally, I honour all those invertebrates that so valiantly gave up their lives in the name of science and higher education. viii

9 Introduction Growing public concern for environmental health over the last two decades has resulted in increasing efforts to restore degraded systems and to mitigate human impacts on the natural environment (Edwards et al. 1984; Lepori et al. 2006). In particular, stream and river restoration has received considerable attention. A widespread impact on stream systems that has received much attention from restoration ecologists and managers is channelization; the straightening, dredging and artificial banking of stream channels for the redirection, drainage or transport of water, and flood control (Hynes 1960; Laasonen et al. 1998; Lepori et al. 2006). This impact has been particularly common in western boreal regions where running water systems have been channelized by removal of flow obstructions and the placement of embankments and retaining walls to alter stream direction to accommodate land use in North America and England, or to facilitate log transport in areas of northern Europe such as Denmark, Sweden and Finland (Harrison et al. 2004; Moutka and Laasonen 2002). Despite such diverse purposes for channelization, the consequences for stream ecosystems are similar; loss of structural complexity, more uniform flow patterns, a decrease in habitat variety and an associated decline in richness and abundance of benthic invertebrates (Hynes 1960; Muotka et al. 2002). Many of these channelized streams in both North America and Europe have undergone restoration projects. These projects usually consist of channel enlargement and the construction of instream structures such as riffles, boulder clusters, flow deflectors and pools in order to increase heterogeneity and mimic the configuration of natural meandering streams. The goal of such projects is usually to improve habitat to better support the biodiversity and abundance of fish and macroinvertebrates (Laasonen et al. 1998). A large amount of research has been conducted to assess the ecological benefits of stream restoration and its effects on benthic organisms (Lepouri et al. 2006). Much of 9

10 this research has been conducted on streams in minimally impacted forest catchments of northern Europe (eg. Korsu 2004; Laasonen et al. 1998; Lepori et al. 2006; Muotka et al. 2002). Such work has focused on the importance of detritus retention and presence of bryophytes in chiefly allochthonous systems (those with food-webs primarily driven by carbon/energy sources outside the aquatic system) as well as the shift in benthic macroinvertebrate communities in response to the physical disturbance and increased stream heterogeneity resulting from restoration work. Studies in this area span a wide range of temporal scales from seasonal patterns to those spanning several years. Research in North America and England has also attempted to evaluate the benefit of rehabilitation structures to benthic communities (eg. Edwards et al. 1984; Harrison et al. 2004). These studies were typically conducted in more autochthonously driven streams (those with food-webs primarily driven by instream carbon/energy sources) running through generally open agricultural areas and have focused on differences in community structure around rehabilitation structures and that in unrestored streams. The present study was conducted in Dickson Brook, a third order coastal stream running through Fundy National Park in New Brunswick, Canada (Figure 1). Stream configuration in lower reaches of Dickson was severely altered through channelization as part of golf course construction in Numerous flood events and associated erosion problems resulted in repeated repair and replacement of channelization structures in the 1960s and 1990s, and eventually led to a recent rehabilitation work involving channel reconfiguration through the construction of riffle : pool sequences analogous to upstream unchannelized reaches. The stream offered an excellent opportunity to assess the effects of stream restoration work in northeastern Canada, an area that has received little consideration but might well augment research on streams in western and southern North America and western Europe. The objectives of this study were to measure the response of benthic macroinvertebrate communities to channel restoration within the golf course by 10

11 comparing abundance and community structure in restored reaches of Dickson Brook to that of a channelized, unrestored reach, here after referred to as the control reach. It was also a goal within this context to see how these communities responded over time by sampling a two-year old restored reach as well as a one-year old restored reach, here after referred to respectively as the 2004 restored and 2005 restored reaches. To account for seasonal variability and to see if observed patterns were generally consistent across seasons, the stream reaches were sampled in early June and again in early September of Both the control and restored reaches were located within the altered land area of a golf course, thus a secondary objective was to compare the benthic macroinvertebrate communities of these reaches to that of a minimally impacted upstream reach running through a forested area, essentially unaltered since the construction of the Wolfe Point Road prior to opening of the park, here after referred to the reference reach. Several predictions were made: 1) That benthic macroinvertebrate communities would rebound quickly after disturbances from restoration with abundances at least approaching those in the control reach within two years of disturbance. 2) As a result of the increased habitat heterogeneity, abundances in the 2004 restored reach may exceed that of the control. 3) Despite the confounding effects of seasonal variability in community structure, general patterns observed in June would also be apparent in September. 4) Benthic macroinvertebrate communities in the forested reference reach will differ from those in all reaches within the golf course due to differing stream habitats. 11

12 Materials and Methods Description of Study Area Dickson Brook is approximately 4 km in length with a drainage basin of 8.8 km 2 and mean annual flow of m 3 s -1 at the downstream margin of the golf course reaches of the brook. Its upper reaches flow through mid-aged mixed forest while a 900- m section of the brook flows through a golf course in Fundy National Park (Figure 1, 2). (The following description is based on pers. comm. and an unpublished report by Jane Watts, Fundy National Park). Between 1949 and 2006, the brook and two of its tributaries have been greatly manipulated by the construction and maintenance of a golf course. The original course of the brook was altered and culverts, ponds, and bridges were constructed. In the past half-century, numerous embankment protection structures were installed for flow containment and flood control. The original cedar and stone retaining walls were washed out in the early 1960s and replaced with rock-filled gabion baskets. These structures were also washed out in floods in the 1990s and replaced with additional gabion baskets and armour rock. By 2003, the brook was highly degraded, eroded banks were extensive, and instream structures were failing throughout the reach running through the golf course (Figure 3c). The stream was nearly uniform in depth and lacked pools, riffles, large rocks or woody debris characteristic of its minimally impacted upstream reaches. Over the summers of , the main stem and two of its three tributaries running through the golf course underwent restoration work to return them to conditions more closely resembling a natural stream configuration (Figure 3). In 2004, construction began at the downstream end of the degraded brook. Gabion baskets and armour rock were removed from stream banks, the stream channel 12

13 was widened to twice its bankfull width (from 6 to 12 m), and shallow sloping stream banks were constructed upon which natural riparian vegetation were planted. The channel radius of curvature was also adjusted to improve the natural meander pattern of the stream and seven riffles were constructed to create a riffle : pool channel profile to reduce stream grade, erosion from flood flow, and to increase stream heterogeneity. In 2005, gabion baskets and armour rock were removed from a reach upstream of that restored in 2004, the channel was widened to twice its channelized width, and seven riffles were constructed. During the summer of 2005, boulder clusters were also added to all pools created in 2004 and 2005, and vegetation was planted along the riparian zone of the newly restored stream reach. In 2006, a section of approximately 180 meters of channel was restored in the upper golf course reach of the brook. This work included 140 meters of degraded channel in tributary 3 and 40 meters of the main channel of the brook above and below the outfall of the restored tributary at the upper region of the golf course stretch of stream. Gabion baskets were removed from the sides and streambed of the restored reach of tributary 3 and the main stem and stream banks were re-sloped and planted with natural riparian vegetation. Seven rock riffles were also constructed: five in the tributary and two in the main stem, one above and one below the tributary s outfall. For the purpose of this study, Dickson Brook was divided into four reaches which were sampled for benthic invertebrates (Figure 2): 1) The 2004 restored reach, wherein a major disturbance (construction) took place two summers prior to the start of this study (Figure 3a) and located at the lower end of the golf course. 2) The 2005 restored reach that was disturbed one summer prior to the study (Figure 3b) and located upstream of the 13

14 2004 restored reach. 3) An older (> 7 years) unrestored, channelized control reach not subjected to the construction disturbance characteristic of the stream restoration (Figure 3c) located just upstream of the restored reaches. 4) A near-natural reference reach upstream of the golf course with a forested canopy and mature riparian vegetation. This reach has not been physically disturbed, but is artificially confined by the construction of an adjacent road (Figure 3d). These four somewhat distinct reaches provide the opportunity to conduct research in a dynamic system with largely different ecosystem composites resulting partially from differences in historic land use and partially from recent disturbances from restoration work. Field sampling Two replicate riffles were sampled for benthic macroinvertebrates from each of the four study reaches. Each riffle was sampled at three locations for a total of 24 samples per sampling period. Downstream riffles were sampled first to avoid affecting subsequent samples. The only exception to this protocol was that only one riffle was sampled in the reference reach during the June sampling period due to time constraints. Photographs of each site were taken and adjacent riparian vegetation was flagged for later identification and description of site characteristics. Sampling procedure Benthic macroinvertebrate samples were obtained with a U-net (0.06 m 2 ; 250 µm mesh) using a modification of the procedures outlined by Culp and Halliwell (1999). Within the wetted-width of each riffle, three replicate samples were collected along a diagonal transect spanning the stream from the downstream left bank to the upstream right bank at 1/4, 1/2, and 3/4, of the wetted width so that a representative composite 14

15 sample could be created but still allow for analysis of community composition across the width of the stream channel. U-nets were placed flat on the substrate so that no gaps were present between the net frame and the substrate. The net was positioned with the open end faced up stream and the holding jar at the bottom of the net trailed down stream in the direction of water flow. Being careful not to disturb substrate in front of the sampling area, the substrate within the area of the U-net was disturbed; all rock surfaces were rubbed and underlying sediment was agitated to a depth of ~5 cm. Once the net area had been disturbed for one minute, the net was lifted out of the water in a forward motion and all clinging invertebrates and sediment were washed into the holding jar by splashing the outside of the net with stream water. All large rocks and debris were rubbed and washed to dislodge clinging invertebrates and then discarded. The contents of the holding jar were then emptied into a jar with 75% ethanol for preservation and storage. Sub-sampling Due to time and personnel constraints, benthic macroinvertebrate samples were sub-sampled in the laboratory. Each sample was dumped into a swirl bucket (6 L) roughly 1/4 full (~1.5 L) of water. Remaining large rocks and organic debris were carefully inspected and rinsed, then removed. Organic and inorganic materials were separated by swirling the bucket to suspend any organic debris and invertebrates, while leaving the inorganic substrate on the bottom. The water and suspended material was then poured off through a 250 µm sieve and rinsed. The bucket was refilled with water and the process was repeated 4-6 times until all organic matter was removed. The remaining inorganic substrate was carefully examined for remaining invertebrates such as 15

16 large caddisfly cases, which were picked out using forceps. The contents of the sieve were then rinsed into a sub-sampler. Invertebrate samples were separated into quarters using a 250-µm sieve that had been divided into four even wedge pieces as in Culp et al. (2000). The divided sieve was fitted in a bucket and invertebrates were suspended and mixed by inserting a plunger into the fitted sub-sampler, and raising and lowering it 10 times with a twisting motion. The sub-sampler was then removed from the bucket and the divided invertebrates sample separated. The 1/4 portion of the sample to be analyzed was then rinsed into petri dishes with any clinging detritus and invertebrates removed with forceps and added to the dishes. The remaining three sub-samples were rinsed into storage jars with fresh 75% ethanol and stored together. Sample processing Petri dishes containing sub-samples were examined under a dissecting microscope at X magnifications, and all observed invertebrates were removed using fine tipped forceps. The Ephemeroptera, Plecoptera, and Trichoptera (EPTs) larvae were sorted to family level, with individuals too damaged, deteriorated, small or otherwise unidentifiable, grouped together and classified as other within that order. Due to the rarity of Trichoptera pupae and identification difficulties, these were also grouped into the other group for this order. Diptera larva and pupa were sorted into two groups; Chironomids and other Dipterans (Tipulidae, Tabanidae, Simuliidae, etc). Counting was done using a manual counter and identified samples were placed in 75% ethanol and stored in vials according to family, date, sub-sample, composite, and replicate sample. 16

17 Statistical analysis All invertebrate families contributing < 3% to the abundance of any sample were excluded from statistical analysis; the only family fitting this criterion was the family Perlidae (Order Plecoptera). Multivariate community analysis was conducted using Primer v5 software (2001 PRIMER-E Ltd, Plymouth, UK). Square root transformations were applied to the abundance data in order to reduce the influence of exceptionally large counts of some taxa. A Bray-Curtis similarity matrix was generated using the transformed abundance data. This matrix contains similarity coefficients (S) calculated between each pair of samples based on their family composition and abundance. Similarity coefficient values are defined in a range from 0 to 100%; when S = 0, two samples are completely dissimilar and when S= 100%, two samples are totally similar (Clarke and Warwick 2001). Macroinvertebrate communities were compared between samples using nonmetric multidimensional scaling (MDS) based on the Bray-Curtis similarity matrix for the entire data set and separately for each sampling period. An MDS ordination visually represents relative similarity between samples through spatial relationships. It is a 2- dimensional (or 3-dimensional) map of samples that attempts to satisfy conditions imposed by the ranked similarity matrix. The relative distance between samples represents a measure of their similarity; points that are close together signify samples that are similar in community structure while samples that are very different in taxa composition and abundance are represented by points that are far apart (Clarke and Warwick 2001). Bubble plot graphics were applied to the MDS plots with the size of 17

18 bubbles around data points corresponding to relative total benthic macroinvertebrate abundance of that sample. The accuracy of the MDS plot in representing the similarity between samples is given by its level of stress, the distortion between the similarity rankings and the distance rankings in the MDS plot. For 2-D graphs, an acceptable level of stress is < 0.2, which indicates that an MDS plot is a fair representation of sample similarities with a low risk of misinterpretation (Clarke and Gorley 2001). Using the relative ranking of similarity coefficients (from 0 to 1) in the Bray- Curtis similarity matrix, analysis of similarity (ANOSIM) was conducted to test for significant differences in benthic macroinvertebrate abundance and composition between sampling periods and among reaches. The test statistic (R) was: [R= (rb rw)/(0.5m)] where rw is the average rank similarity among samples within a group (sampling period; reach), rb is the average rank similarity among samples between groups, M = 0.5 n (n-1) and n is the total number of samples. This Global R-value indicates the level of overall dissimilarity (0 = totally similar; 1 = totally dissimilar) of samples in a data set. Significance values (p-values) are obtained by randomly re-assigning samples to groups and calculating R for each of 999 random permutations, then comparing this permutation R distribution to that of the real data. The significance level was (N + 1)/1000, where N is the number of random permutations with an R greater than or equal to that of the observed data. Pair-wise comparisons between groups (sampling period or reaches) were also tested using similar randomization tests. In such analysis pairs are deemed significantly different when R > than 0.2 and p < 0.05 (Clarke and Green 1988; Clarke and Warwick 2001). A two-way ANOSIM was conducted using the entire data set with 18

19 sampling period and reach as grouping factors, and separate one-way ANOSIMs were also conducted within each sampling period with reach as the grouping factor. The relative contribution of each family to the dissimilarity between reaches was determined using dissimilarity percentages analysis, calculated with the SIMPER routine of Primer v5 (Clarke and Gorley 2001). This test disaggregates the dissimilarity between samples summarized in the Bray-Curtis similarity matrix to identify and rank the relative contributions of particular families of invertebrates (defined as a % contribution) to the dissimilarity within and between groups (date and reach) (Clarke and Warwick 2001). Environmental measurements Hourly water temperatures recorded by park staff using Vemco temperature loggers were obtained for the period between June 1 st and October 30 th, Four loggers were placed in the stream adjacent to the two restored and the control reaches (Figure 2). No 2006 temperature data were available for the reference reach but August 2002 data were available for comparison. Mean daily discharge for and between June and September sampling periods were also obtained from an Environment Canada gauge station located at the lower margin of golf course reaches of Dickson Brook. Results: Overall abundance A total of 8,689 benthic macroinvertebrates representing four orders and > 20 families were counted from grab sub-samples (15 cm 2 ) from four reaches of Dickson Brook on June 8 th and September 14 th, 2006 (Table 1). Benthic macroinvertebrate abundance was 1.6 X higher in September relative to June (Figure 4). Ephemeroptera, 19

20 Plecoptera, and Trichoptera (EPT) contributed 52%, and Diptera contributed 47% to the total average June abundance. In September, EPT contributed 43% and Diptera contributed 57% of the total average abundance. Chironomidae (order Diptera) was the most abundant family representing ~ 47% and ~29% of the total abundance in June and September samples respectively. The second most abundant families were Ephemerellidae in June and Baetidae in September, and the least abundant were the various families of Trichopterans such as Rhyacophilidae (Table 1; Figure 5). Benthic macroinvertebrate community structure: June In June, total benthic macroinvertebrate abundance was generally highest in the three golf course reaches and lowest in the reference reach (Figure 6). Analysis of similarity (1-way ANOSIM) indicated significant differences (global R > 0.2; p =0.005) in benthic macroinvertebrate abundance and distribution among reaches (Table 2). The golf course reaches generally had higher total and similar relative invertebrate abundance compared to the reference reach (Figure 6). Subsequent pair-wise comparison tests found that restored reaches in the golf course were significantly different from each other (R = 0.29) but not from the control reach (R< 0.13), and all golf course reaches were significantly different from the forested reference reach, particularly restored reaches (R > 0.60) (Table 2). The highest total abundance was observed in the 2004 restored reach, which was 1.3 X, 2.7 X, and 4.2 X higher than that of the control, 2005 restored and reference reaches respectively. The patterns in abundance of the most common families, in particular Chironomidae (order: Diptera) and Ephemerellidae (order: Ephemeroptera) were similar 20

21 to that of total benthic macroinvertebrate abundance (Table 1, 3). Chironomids tended to be the most abundant family in June, especially in the control and 2004 restored reaches, in which their abundances were higher than that of all Ephemeroptera, Plecoptera, and Trichoptera (EPT) families combined. In general, chironomids were approximately seven times more abundant in golf course reaches than in the reference reach. Chironomids also had the highest percent contribution to dissimilarity of reaches in all pair-wise comparisons except for that of 2005 restored-reference reaches as indicated by dissimilarity percentages analyses (SIMPER) (Table 3). The next most abundant benthic macroinvertebrates in June were from the order Ephemeroptera: Ephemerellidae, Heptageniidae, Baetidae, and Leptophlebidae (Table 1, 3). In particular, the ephemerellids were the second most abundant family and the second highest contributors to differences between reaches (15-18%) in all pair-wise comparisons but that between the 2005 restored and reference reach. The pattern in ephemerellid abundance followed that of total benthic macroinvertebrate abundance more closely than the chironomids, with average abundances in 2004, 2005 restored and control reaches 11 X, 6 X, and 8 X greater than in the reference respectively (Table 3). Heptageniidae was also abundant but had less variability between reaches and thus contributed less ( %) to the dissimilarity of reaches (Table 3). Baetidae and Leptophlebidae also had low variability and were the least abundant of the Ephemeroptera among reaches. In the order Plecoptera, only Chloroperlidae, which had patterns in abundance similar to that of Ephemerellidae, was relatively abundant or had notable contributions to reach dissimilarity (Table 3). Trichoptera had the smallest average abundance of the EPTs but this was partly due to very patchy and clumped distributions within reaches relative to 21

22 other EPTs (pers. obs.). Only Rhyacophilidae and Uenoidae made considerable contributions (> 5%) to the dissimilarity between reaches. While rhyacophilids had more even distribution among reaches, most Trichopterans were lower in abundance in the golf course reaches. With the exception of the reference reach, total macroinvertebrate abundance was highly variable within stream reaches as indicated by the error bars in Figure 6. Most benthic macroinvertebrate families also showed large variability in their distribution both among and within reaches (pers. obs.). This pattern was more evident in the golf course reaches than in the reference reach and most notable in the chironomids and trichopterans, which appeared to have very patchy and clumped distributions (pers. obs.). September Abundance patterns apparent in June were less obvious in September. In September among the golf course reaches, benthic macroinvertebrate abundance was highest in the control reach; approximately 4.2 X higher than that of the 2004 restored reach and three times higher than that of the 2005 restored reach (Figure 7). Total abundance in samples from the reference reach was also much higher in September, greater than that of the 2004 restored reach and approximately equal to that of the 2005 restored reach. Differences among samples in September were statistically significant and relatively large (global R > 0.5; p = 0.001) (Table 4). Subsequent pair-wise comparison tests indicated that restored reaches in the golf course were significantly different from the control reach (R > 0.5; p < 0.02), but not each other, and all golf course reaches differed significantly from the forested reference reach (R > 0.5; p = 0.02) (Table 4). This was largely due to substantial differences in relative abundance of dominant 22

23 taxa, (eg. chironomids and baetids) both in relation to each other and between reaches (Table 5; Figure 7). Chironomids were the most common family in macroinvertebrate samples from the control and reference reaches, contributing > 50% of the total average macroinvertebrate abundance in these samples, but were second most common in the two restored reaches (Table 5). Baetids were the most common EPT in September, having the highest abundance in restored reaches, and second highest in the reference and control reaches. Chironomids still had the highest percent contribution to dissimilarity (between 15 and 28%) in pair-wise comparison tests (Table 5); resulting from their high relative abundance and between reach variability in all but the 2005 restored-reference reach pairwise test. Baetids were the second greatest contributors to the dissimilarities between reaches in all pair-wise comparison tests except that for the 2005 restored and control reach (Table 5). Heptageniids were again the most evenly distributed Ephemeroptera, while overall, leptoplebids and ephemerellids were the least abundant (Table 1, 5). Chloroperlids were again the most common family of Plecoptera and followed total abundance patterns. Perlodids were the second most common Plecopteran while nemourids, present in June, were absent. The average Trichopteran abundance was highest in samples from the control reach; 2.6 X higher than that for samples from the reference and 2004 restored reaches, and more than four times greater than that of the 2005 restored reach (Table 1, 5). Lepidostomatidae had the highest average abundance in all reaches except the reference and 2004 restored reaches, followed by Rhyacophilidae and Hydropsychidae in September. 23

24 Temporal consistency in macroinvertebrate community patterns Differences among reaches were more apparent in September than in June as indicated by a higher Global R, pair-wise R values, and significance levels from 1-way analysis of similarities for the two sampling periods (Table 2, 4). Differences in community structure between the two sampling periods were more pronounced than those among reaches within each period. This was apparent in the MDS plots where June and September samples were clearly separated while samples from common reaches did not clearly aggregate in either period (Figure 8). High variability in relative abundance can also be seen in the bubble plot superimposed on this MDS plot in which higher abundance was observed in September, particularly in the control reach which was driven by one exceptionally large sample (indicated by an arrow). This high variability can also be seen when comparing error bars for average abundance histograms for the two sample periods (Figures 6, 7). Average abundance of sampled invertebrates also differed greatly for reaches between June and September (Figures 6, 7). In June, samples from the three golf course reaches were similar in abundance, considerably higher than that of the reference reach, with the highest observed in samples from the 2004 restored reach. In contrast, this reach was lowest in abundance in September, while an exceptional increase in abundance was observed in the control reach, and to a lesser extent, the reference reach. While EPT abundance was similar in samples from restored reaches in both June and September, it almost doubled in the control and reference reaches in September (figures 6, 7). Differences in the abundance and family composition of benthic macroinvertebrates between June and September were highly significant (R = 0.687, p = 24

25 0.001, 2-way ANOSIM). Changes in the abundance of chironomids, baetids, and ephemerellids contributed most to these differences (Table 6). The most apparent example of this was the switch in dominant Ephemeroptera from ephemerellids in June, to baetids in September contributing 12% and 16 % to the dissimilarity between June and September respectively. Chironomids had a similar contribution as baetids to the dissimilarity between June and September samples with higher chironomid abundance in September, particularly in samples from the control and reference reaches. Plecopteran abundances also changed between sample periods; chloroperlids became over twice as common in the control and reference reaches but declined in the restored reaches, while perlodids, almost absent in June, became 9-40 X more abundant in September and nemourids, present in June, were absent in September. In September, the average abundance of sampled trichopterans more than doubled in all reaches, increasing six-toeight -fold in the 2005 restored and control reaches (Table 1). The average size of larvae, particularly that of the EPTs, was much smaller and they were typically earlier instars in September compared to June, to the point of making identification difficult among many orders, particularly Ephemeroptera (pers. obs.). Environmental measurements: water temperature and flow Water temperatures averaged from four temperature loggers in golf course adjacent to the two restored reaches and the control reach of Dickson Brook varied from 6.8 C to 23.7 C with a mean temperature of 9.34 C (Figure 9). Though temperature readings were not recorded for the reference reach, water temperature data from August, 2002 were available to show temperatures in the reference compared to golf course reaches which showed the typically higher and more variable temperature of golf course 25

26 reaches (tributaries 2, 3) relative to the forested reference reach which maintained a temperature between 6-7 degrees (figure 10). Stream discharge within Dickson Brook was variable between the early summer and fall. Flow was generally higher during the June sampling period (mean daily = m 3 s -1 ), than in the September sampling period (mean daily = m 3 s -1 ) (Figure 11). Discussion This project aimed to assess the response of the benthic macroinvertebrate community to the disturbance and habitat change associated with the restoration work in Dickson Brook. At a basic level, benthic communities appeared to have rebounded in abundance with levels similar to that of the channelized control reach within 1-2 years after the disturbance. This is consistent with results noted by Korsu (2004) and Laasonen et al. (1998) who found that benthos in restored Finnish streams tended to rebound to predisturbance abundance levels after several months to one year. Despite large variability within and among reaches in June, differences in benthic macroinvertebrate community between restored reaches and the control reach were minimal, with similar total invertebrate abundances and relative abundances of dominant families across all reaches within the golf course. That the greatest abundance was found in the 2004 restored reach gives some indication of community recovery in this reach beyond that in the control reach, due to increased heterogeneity resulting from the restoration. However, because this difference was not significant (possibly a reflection of high variability between samples and relatively low sample size), it may be that more than two years are needed for significant increases in abundance in restored reaches relative to the control reach to become apparent. This is consistent with findings by Edwards et al. (1984) who noted general increases in abundance and richness of stream macrobenthos in restored streams compared with channelized streams in the eastern US, in which that of restored streams 26

27 approached numbers typical of unaltered reference streams over longer timescales. Alternatively, Harrison et al. (2004) noted only marginal differences in abundance between restored and channelized reaches among lowland streams in the UK over similar time periods. Previous studies have found that stream benthos are highly resilient to physical disturbance, with invertebrates quickly recolonizing areas subjected to large scale disturbances from restoration work. However, it may take months to several years for abundance levels to approach or exceed pre-disturbance levels, varying among taxa and depending on the presence and proximity of refugia (eg. Korsu 2004; Laasonen et al. 1998; Muotka et al. 2002; Smock 2006). Thus, the lower abundance observed in the 2005 restored reach is not surprising. As the disturbance from restoration took place during the prior summer, the benthic community had only the fall, winter, and early spring to recolonize from other reaches. This may not have corresponded well with aspects of life history such as growth periods and emergence times of many taxa. Chironomids were the most abundant family of invertebrate in all reaches of the golf course, which would be predicted, as they are the most widespread and abundant invertebrates in most temperate fresh waters, include many environmentally tolerant genera typical of degraded systems, and are among the first colonizers of disturbed substrates (Giller, and Malmqvist 1998; Laasonen et al. 1998). Hence, it was not surprising that chironomids had highest relative abundance in the control reach; oddly, however, their relative abundance was much lower in the 2005 restored reach than in the 2004 restored reach. As expected, families of Ephemeroptera, Plecoptera, and Trichoptera (EPT) appeared to be more correlated with channel heterogeneity, with highest abundances relative to chironomids in the upstream reference reach, second highest in the 2005 restored, and comparable in abundance to the chironomids in the 2004 restored reach. Of the EPTs the mayflies were the dominant taxa while caddisflies 27

28 and stoneflies were relatively scarce. Large differences in community structure were observed between the two sampling periods, complicating the interpretation of results. The statistical difference between the two sampling periods was notably higher than among the various reaches of Dickson Brook within either period. Such confounding effects were anticipated and can largely be explained by normal seasonal variation in benthic invertebrate lifecycles (eg. Hynes 1970; Carter et al. 2006). The general increase in abundance of invertebrates in September was most pronounced in the control reach and was characterized by an over two-fold increase in abundance of chironomids and most EPTs. There is a clear succession of benthic invertebrates, and changes in relative abundance and size of individuals over time (eg. Hynes 1970; Giller and Malmqvist 2002), especially in temperate streams such as Dickson Brook. Laasonen et al. (1998) reported similar results where invertebrate abundance in early June samples were approximately an order of magnitude lower than that in samples collected in early October, with extensive changes in the relative abundance of various taxa between the two sampling periods. Furthermore, the average size of larvae sampled in September were considerably smaller and of earlier instars than those in June. The relative abundance of several insect families in Dickson Brook changed markedly between June and September. Among the mayflies and stoneflies in particular, the increases in some families (eg. baetids, perlodids) replaced declines in others (Ephemerellids, nemourids). Increased abundance in the control relative to restored reaches in September could be more apparent than real as it was driven largely by one sample with an exceptionally large count of chironomids. However, taxa with summer 28

29 growth periods such as many chironomids, would be more severely impacted by disturbances in restored reaches during the previous summer than overwintering taxa such as many families of mayflies and stoneflies that emerge in late spring and early summer (Hynes 1970; Giller and Malmqvist 2002). Korsu (2004) noted that overwintering taxa might be better able to rebound from summer disturbances over the fall and winter than those with lifecycles tied to summer months. This would be consistent with the general trend of much smaller size and earlier instars of many taxa observed in September compared to larger mature larval stages characteristic of insects in June samples. Alternatively, those trends apparent in June may be merely the result of differing emergence and growth times. The impacts of seasonal variability on interpretations of responses to restoration require further work. Other studies that have evaluated the benefits of stream restoration on benthic communities have generally applied more detailed benthos analysis, with identification to genus or even species level, as tolerance of benthic taxa differ within family and even between genera (Carter et al. 2006). Such analyses often measure abundance levels, richness, and eveness and focus on the interaction between community assemblages and a number of environmental parameters. Because the present study did not conduct such detailed analyses, or measure environmental variables such as primary production, depth, or water velocity, only general trends between benthic communities and environmental factors associated with restoration work in Dickson Brook can be detected. Many studies have highlighted the importance of allochthonous (detritus input from surrounding terrestrial environment) verses autochthonous (algal primary production) food sources in determining the structure of benthic communities and their response to 29

30 stream restoration (Lepori et al. 2006). Research in Finnish woodland streams by Muotka et al. (2002a, b) and Korsu (2004) indicated that bryophytes are important as refugia during disturbances and increase benthos richness, and that channel heterogeneity and time since disturbance affect relative detritus retention and accumulation among natural, channelized, and restored streams. While benthic communities were found to rebound quickly after disturbance, their structure changed substantially in restored reaches with increased temporal variability and declines in detritivores, replaced by increased scraper abundance. In contrast, subsequent research in Finnish streams by Lepori et al. (2006) suggests low correlation between detritus retention and benthic community structure in comparing channelized and restored stream reaches, pointing to relative patchiness of food sources within reaches in dictating invertebrate distributions. The factors affecting benthic community structure in Dickson Brook likely differ from those identified in the Finnish studies of woodland streams. The golf course reaches of Dickson have very little riparian vegetation and are surrounded by relatively open greens. Lamberti et al. (2006) noted that instream primary production usually represents a major energy source for benthic food webs and may dominate annual energy budgets in streams with limited riparian shading and input of terrestrial vegetation. Thus, it is more likely that leaf litter retention played, at most, a subsidiary role in dictating community structure among the control and restored reaches of Dickson. While bryophytes were of importance in Finnish research, they typically grow in forested streams with stable substrate and thus would have little presence in the open canopied unstable golf course reaches before or after restoration. In open canopy lowland streams in England, Harrison et al. (2004) found that rehabilitation structures in channelized 30

31 streams appeared to have little overall effect on benthos, or that their effects were overshadowed by larger scale environmental parameters, many of which, Winters et al. (2004) note, are associated with golf courses. Sampling in the forested reference reach provided an opportunity to compare the patterns in invertebrate community structure in an historically impacted reach of stream (golf course) to a near-natural forested reach that likely represented the stream system prior to building of the golf course and channelization of the stream within its borders. The reference reach may also be more similar to the setting in which studies were conducted on the restoration of streams in forested areas of northwestern Europe channelized to accommodate log transport (eg. Muotka and Lepori et al. 2006); while reaches in the golf course likely resembled streams in studies on stream restoration in open lowland areas in England and United States (eg. Edwards et al. 1984; Harrison et al. 2004). In contrast to the results in these studies, the reference reach of Dickson was relatively depopulated, particularly in June, relative to the channelized control, and to a lesser extent, the two restored reaches. This could be due to complete canopy cover limiting primary production in the reference relative to the open reaches in the golf course. Reaches in the golf course experienced higher water temperatures and likely had higher solar influx resulting in higher algal growth (Giller and Malmqvist 2002). Winter et al. (2004) note amplified primary production in golf courses due to input from runoff of limiting nutrients from fertilizers applied to course greens. This is not likely the case in golf course reaches of Dickson however, as fertilizers are not applied to areas of the course adjacent to the stream (J. Watts, Parks Canada, pers. comm.). The role of limited 31

32 nutrients cannot be determined without some measure of primary production or nutrient levels in the various reaches. Furthermore, the effect of limiting nutrients on the differences in community structure between the reference and golf course reaches would likely be small compared to effects of the starkly different surrounding drainage areas (Winter et al. 2004). Still, this puts into context the fact that restoring systems to pristine or even pre-channelization conditions is usually unrealistic, especially in areas of significant human impacts, and that restoration projects should focus on improving stream heterogeneity and stability using adjacent, more natural streams or reaches as models in order to provide the habitat necessary for the recovery of community structure and function (Laasonen et al. 1998; Lepori et al. 2006; Newbury and Gaboury 1993; Ormerod 2003). This study showed that invertebrate abundance recovered from disturbance associated with stream restoration activities within one to two years. However, because of the limited sampling effort and analyses employed in this study, interpretations are tenuous and subject to speculation. While many volunteer based monitoring programs identify stream invertebrates down to a mix of order and family level to infer water and ecosytem quality, most assessments on the effects of restoration have employed more detailed identification to genus and species level (Carter et al. 2006; Lepori et al. 2006). This was not feasible in the present study, yet it would have possibly given more meaningful results. The effects on benthos both in regard to the disturbance and habitat change associated with stream restoration appear to be more physical in nature (eg. retention, microhabitat structure, flow patterns organic debris levels) with variable correlation to food web dynamics (Muotka et al a, b; Lepori et al. 2006). Thus 32

33 classification of taxa into functional feeding groups and incorporation of indicator taxa with known habitat specifications would be useful if not necessary to accurately interpret results. Other studies aiming to evaluate the impact on benthic communities by restoration of channelized streams have conducted a wide range of environmental measurements, usually physical, chemical, and biological, in order to rule out confounding factors such as water quality or temperature that may not be directly associated with disturbances from restoration. Despite these more rigorous methods employed in such research, results have been varied and often ambiguous. Though the practice of stream restoration has long been employed throughout North America, and more recently Europe, actual evaluation of its benefits on stream ecosystems was, until recently, very limited (Harrison et al. 2004). Before increasing attention in the last two decades to macroinvertebrate responses, such assessments focused predominantly on improving salmonid and other game fish habitat with nonvertebrate components of the stream biota having only secondary consideration (Laasonen et al. 1998; Lepori et al. 2006). However, as Hynes (1960; 1970) and Carter et al. (2006) note, benthic macroinvertebrates have often proven to be the most effective and commonly used bioindicators in many kinds of aquatic environmental and ecological assessments. As frameworks for aquatic systems continue to be refined, it is likely that they will become of similar importance as indices for predicting the benefits of restoring degraded stream systems. 33

34 Figures Fundy Park Park Study area Figure 1. Map of Fundy Park, NB showing study reaches of Dickson Brook and surrounding area. 34