(1)Comparison of the in-stream fauna and resources of Tasmanian river reaches lined with willows or with other riparian types.. by. Martin Read B.Sc. (Hons).. UTAS �ubmitted in fulfilment of the requirements for the degree of Doctor of Philosophy) University of Tasmania (July) 1999).

(2) This thesis contains no material which has been accepted for a degree Of diploma by the. University or any other institution, except by way of background information which is duly acknowledged in the text. To the best of my knowledge this thesis contains no material. previously published or written by another person, except where due acknowledgment is made in the text.. Access to this thesis. The tpesis copy held in the University Library shall be made available for loan and limited copy in accordance with the Copywright Act 1 968..

(3) Table of Contents. ABSTRACT. .................................................................................................... ..... ........... ..... .... .................. V. ACKNOWLEDGMENTS ........................................................................................................................ VII. CHAPTER 1. GENERAL INTRODUCTION I�JTRODUCTlON . .. .. .. ... .. .. .... ...... ... .. .. .... ... . . . ........... ........................ .... ..... .. .. ...... ..... .. .. .............................................. . .. .... ...... .. ... .... .. .. .. .... .. ..... . . .... .. .. ... . . . .. .. ... ... .. ... .. .. .. ..... . 1. ............... ... . ... .. . . . . 1 ... .. .. .. ... .. CHARACTER ISTICS OF SALiX SPP............................................................................................................ 2 E>USTJNG RESEARCH AND ANECDOTAL EVIDENCE OF IN-STREAM EFFECTS OF WILLOWS CORE RESEARCH AREAS . .. .... ..... .. .. .. .. ......... . .. . . . . . ... . . . . . .. .. .. ... ... .. ... .. .. .. ... PROJECT AIMS AND RESEARCH STRATEGY . ... . .... .. ... ... ... .... .. .. .. .. . ... ... .... .. . ... .. RATIONALE FOR lHE USE OF SURVEYS IN THIS STUDY . ...... ... .. .. ..... ..... .... .... ............ .. .. ...... ..... ..... .. . .. .. .. ..... ... .. . .. . . .. .. ... .. .. .. .. .. ... .. .... . .. ... ..... ....... . . ... ........ Approach adopted in this study. ........ .... .. .. .. .... ... .. .. ..... .. ..... .. ..... . .. . ... ... .... .. ... .. . .. ... ............ .. ... ... .. . . .. .. .. ... .. .. ... .. . . . ... .. ....... .. .. ... .. ... .... .. . ... .. ... ..... ..... ... .... .. ... . .... .. ...... ... .. ...... .. .6. ... .... .. . ... 7. . .. . .. .. .... .. OvetView of approaches in environmental impact assessment . ... .. .... . . .. ... .. .... .. . . . . ... ... . 4. .. .. .... .. . ... . . .. 8. ..... ........ .. .. .. . ... ..... ... .. .. . 8 ... .... 10. ........ 14. .......... 14. .... .... .. CHAPTER 2. SEASONAL COMPARISONS OF BENTHIC COMMUNITIES ADJACENT TO RIPARIAN NATIVE EUCALYPT AND INTRODUCED WILLOW VEGETATION. INTRODUCTION .. .... ...... . ... .. .. .. ... .. .. METHODS AND MATERIALS ..... . ... ... ..... .. ........ .. .. .. .. .. . ... Study sites . . ....... .. .. .. ... ..... .. .. .. .... . .. Habitat measurements. ... .... ..... .. .. ....... ... . . .. . ... ... ........ ... .... ...... ... .. .. .... . . .. . . .. .. ... .. ... .. .. .. .. .. . .... Benthic Macroinverlebrates. ... . .. ... .. ..... .. .... .. .. ... .... ...... .. ....... . .. .. ... .. . . . .. .. .. ... .. .... ... .. ..... .. ...... .. . . . . . .. .. ... ... .... ... .. ...... .... .. .. .. ... . . . . .. .. .. ... .. .. . .. . .. .. ... .. ........ ..... ........ .. .. .. ... .... .. . . . ... ... . .. ... .. .. .. .. .... .... ... .... . .... ... .. ... .. ....... ........ ..... ...... . .. . .. ... Habitat conditions. .. ... ........ .... .. ..... .... . . ..... .. .. .... ..... .. ...... ... . .. .. .. .. .. .. .... ..... . ... .. ..... .. ... .. .. . . ... ..... .. .......... ..... .. ..... ..... ... . . .. ... .. ... ....... .. ... ... .. .... .. . . ............. . . . . ... . .. . .. .. .. . .. .. .... ... .. .. .. .. .. .. .. .. .. .... .. . . . . .. . . .. . . .. ... .. ... .. .. ... ... .. The invertebrates .. ..... ...... . . . . .. .. ...... .... . .. ... ... ...... .. .. .... ... ... Biomass of food resources .. ..... .. ... .. .. .. ... ... .. ... ... ... .. . . . . . . .. . . .. .... .... .... ...... .. . .. ... ... .. ... ...... .. .... ... . . .. ...... .... .... ... .... .. .... ... .. .. ........ ........ .... . .. .. ... .... .. ..... ..... ..... .... ... . .. .. .. ... . .. . .. ... ... .. ........ .... ..... .. ... ... .. ... .. .. .... ... .... .... ....... ..... ... .. 16. ..... ... .. ... . . ..... .. .. .... .. .. ...... . . .... .. .. . .. .. . ..... . . ... .. .. .. . . . .... .. .. .. ............. .. ... .. ...... ... ..... ..... ... .. .. ........ . . . . .. . ... .. . .. ... .. .. .... 16 17. . 20. 23. .. ... .. .. . ... ..... ...........•...................................................................... ............................................................ ....... .... . . ..... .. ... .... ... ... .. ... . . . . . . . ....... . .. ... . .. ... .... .. . . .. . .. ... . . .. ... ... .. 22. ... ...... ... .. ........ .... .... ANA!.VSIS .. .. .. ... .... .. ... ... .. 20. .. .... ... .. .. ... ... RESUlTS. .... . . .. . ..... .. . .. �uTVey Design .. ... .. . . . ... ... ... .... .......................... .. ...... .. .. .. .. ..... .... ... ... .... 23. .... . 25. .... .. .. . ... ... .. 28.

(4) DISCUSSION. .................. . .......... .. ... ... .. .......... .. .... •.. . .. . .... .. . ....... · ·. · · · · · · •· •· · · · ... . . . . · ·. . ·•· . . . . · · • · · • · · · · · · · · · . . · • •• .. . . . . · . . .. . . 36 ·. .. CHAPTER 3. EFFECTS OF RIPARIAN WILLOW REMOVAL ON MACROINVERTEBRATES AND FJSH. .................. .. ........................................ ..... .............................. .......... ............ ....... ....... ......... ......... ... 1NTRODUCfiON. .. . ... . . . . .... . .. . . . . . . . . . . . . . . ... . . . . . . . . . . .. .............. .. .. ..... .... .. . ............................. , ........................ 43 ... .. .. .. METHODS AND MATERIALS ...................... ... ... .. ......................................... .. SuNe y Design. ..... ... ... .. .. Study sites. ... ... . . . .. .. .. ... .. ... . ... ... .. .. .. . . . . . . . . . . .. .... . ....... . . . . . .. .. ··················. . . ··· ...... ·······. .. . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . ............................ .. 43. . . . . . . ... .. .. 46. .... ....... 46. ... .. .. .. . . . . ... ... ......... ..... ... ................ ... . .. . .. . . . 50. .... .. .. .. .. .. .. ... .. .. .. .. .. .. .. .. .. .. ... .. ... .. Habitat measurements . . . .... . ....... . . ............ . .... . .. . ........ .... .......... ........ ....... . .. ... ....... 50 .. .. .. .. .. .. .. .. .. .. Benthic Macroinvertebrates. . . . . .. .. .. .. . . . . . . . . . . . . . . . . .. .. .. ... . . . ... .. .. .. . .. . . . .. . . . . .. Fish...... . ........... . ............ ... . ...... ... ........ ....... .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. ... .. .. . . . . . . .. . . . . . . .. . . . . . . .. .. .. . . . . . .. . . . . . . .. . . .. . .. . . . . .. ... ....... .... .. . ...... ......... ...... ... .. .. .. . .. .. .. 50. .. .. .... ....... 51. ..... ... .. .. ANALYSIS ............ ...... ................ .............. . ... . ...................... .................... ...... ............. ............... 52 .. .. Habitat conditions. .. •. Macroinverlebrates. •. .. .. .. RESULTS. . . . . .. . . . . . . . . . . . . . . .. .... .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... Macroinvertebrates. Fish. .. .. .. .. .. .... .. .. .. .. . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 52. shallow rivers.. . . ........ . ........... . .. . . .......... . ............. ......... . . .. ....... 52 .. .. .. .. .. .. .. .. .. ... .. ... ... .. .. ... ... deep rivers............ . .... ...... ... .... . ...... ....... . .. .... ....... ......................... 55 ... .. .. .. .. .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. .. .. .. ... . . . . .. . . . . .. .. ... .. .. .. . . . . . . . . . . . . . . . . . . . . . .. . . .. ........ . . ... ................ 56. ................................ ... ............................................................ ........................................ 56 .. .. .. Habitat conditions· shallow reaches. .... . .... . . ...... ...... .. ...... .......... ... .. ..... ... . ......... ......... 56 .. Habitat conditions- deep reaches. . . . . . . . .. .. .. .. .. ... ..... .. .. .. .. .. .. .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... �iomass of food resources - shallow habitats ........ Macroinverlebrates. .. .... ...... .... .. ............. . . . .. . ... DISCUSSION ... ... .. .. .. . . . . . . . . . . .. . . . . . .. ... .. .. .. ... .. .. .. . . . . . . . . .. . .. . . . . . .. . . . . . . . . . . . . . . . . .. ......... .. . . . . .. . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . .. .. .. . .. .. .. 57 58. .. . . . . . . . . . . . . . . . . . . . . . . . .. .. .. . . . ...... . . ....... .... .. .. .. ... .. .. ... ........ . .. . . . . .. 64. .... 66. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. .. .. Fish.. .. 73. .... ...................... ........................... ................................................................... 79. . ... .. Shallow Reaches. . .... . Deep reaches. ... .. .. . . .. . . . . . . . . . . . . . . . . . .. .. ... . .. . . . . . . . . . . . . . . . . . . . . . . . .. Community similarity- deep and shallow water reaches .. . . ........ . . . .. .. .. Densities and community indices for shallow water habitats ..... .............. 63. •. Functional feeding groups- shallow water habitats. Fish. .. .. .. .. ... .. .... ..... . .. . ....... .... . .... . .. .. ..... ..... . ..... . ............... ..... ................... 79 .. . . .. . . . ... . . . . . . . . . . . . . . .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. . . . . .. . . . . .. . .. . . . . . . . .. . . . . . . .. . . . . . . . . . .. .. .. .. . . . . .. . ... ... .. .. .. . .. . . . . .. . . . . .. . . . . .. . . . . .. .. . . . ... ....... .. . . . .. 85. ... .... . .... .. . . .... .... ....... . .. . ... .... . ...... ... .... .. .... ............... ... . .. .... .... . .... ...... 85 .. .. ... .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. ... .. .. .. .. . .. .. .. Summary .... ........ .......... .... ...... .... . ........ ..... .. . ........... ....... ............ ..... .................. ............ 88 .. .. .. .. ... .. .. .. .. .. .. .. .. .. .. .. II.

(5) CHAPTER 4. THE ROLE OF LARGE WOODY DEBRIS OF DIFFERING RIPARIAN TYPES IN. TASMANIAN RIVERS. INTRODUCTION. ........................................................................................................................... 91. .................................................................................... ................... ................. . . . . . . . . . . . .. 91. SiTE SElECTION. .. . . . . ...... ............ .. .............. .. ............ Assessment of LWD standing stocks Assessment of LWD as habitat. METHODS AND MATERIAlS. .. ................. .. .. ............................. .. ................. .................. .. .......................................... .................... .................. . . . . . . . . . . . . . . . . . . .. .. .. Assessment of invertebrates Analysis of invertebrates. .. ....... .......................................... 94. ... ..... 96. .. .......................................... .. ...... 96. ........................ 99. .. ... 100. ............................................................................... ........ ....................... 101. ...... ........................ .. ................................................... .................... Assessment of fs i h .. 103. .. ..... 103. ........................................................................ .................................................................... 104. Analysis offish. ............ . . . . . . ..................................... ........ .. ............................. .. ................ .. .. ............................... .. ..... .......................................................... .. ........................... Invertebrates-willow vs. native wood . .. ....... . . . . . . ........... ....... .. .. ..... .................. ............ .. ................ .. 104 106. . 106. ....... ................ ....................... .......... 1 09. .................................................... 112. ................. Invertebrates on native wood under different riparian types. ...... .. .. .................................................................................... Invertebrates� benthic and wood habitats. .. ............................... ................................................................................... LWD standing stocks Invertebrates. .. .. ................. . .. .. ... RESUlTS. 93. . . . . . . . ........................... . . . . . . .. .................................................................... .. ........................ 93. .. ......... Assessment of LWD standing stocks . Analysis of LWD standing stocks. .. ......... .......................... ........................................................... ... .................................... .. ... ........ . . . . . . ....... ............ .. Relationships between fish and LWD.......................................................................................... 118. DISGUSSION. .......................................... .. ............. . . . . . . ............. ................... . . . . . . . . . . . .. LWD standing stocks. ......................... .. .......................... .. .. .............. . . . . . . . ................ .......... . . ... ................ .. ........................... 124. . 124. Relationships between invertebrates and LWD .......................................................................... 127 Relationships between fish and LWD.......................................................................................... 132 Conclusions. ....... .. ..... ............. .. .............. .. ...... .. ............. .. .. . . . . . . . ....................... ................................ .. ....... .. ... 134. Ill.

(6) €HAPTER 5. GENERAL DISCUSSION ···'":. .. INTRODUCTION. . ... .... .. . . . ..... .. .. ...... .. ........ .... ..... _, ..... �. ......... . M .......... M. .. .. ................................. . .. . .. ... . ...... .. .. .. ..... .. .. . .. ..................... . . ... . .. .. .... ..... ................. ......... . .. ... ........... ... .. ..... .. ... . ... 136. ....... ... ......... . 136 ... O VE RVIEW. .......................................................................................................................................... CAVEATS AND IMPliCATIONS OF THIS RESEARCH. ............... .............................................................. ..... 136 140. '. HOW ARE WillOWS BEING MANAGED NOw?. .. ................................................ ...... .. ................ .................. 143. FuruRE RESEARCH............................................................................................................................ 145. Research oriented toWards evaluating management practices. ............................................. ....... Other research CONCLUSIONS. .. ................................... .. .. ........... .. .... ....... 148. ..................... ......... 151. .............................. .......... . . . . .... ...................................................... .......................................... . . . . .. RE.FERENCES .. .. .................. ...... .. . ........ .. . ... ........ . ......... . .... . . .. .. . . ... ...... ... .. ... . . . . .. .... .. .. .... .. ... .... .. .. .... .. .. .. .. ... .. .. .. .. ... .. 145. ... ... .... .. .. ... .. ... ....... 152. .... APPENDIX A. ANOVA TABLES FOR BIOLOGICAL AND HABITAT DATA -CHAPTER 2. APPENDIX B. ANOVA TABLES FOR BIOLOGICAL AND HABITAT DATA. •. CHAPTER 3. APPENDIX C. SITE PHOTOGRAPHS OF REACHES SURVEYED. APPENDIX D. PAPER ACCEPTED FOR PUBLICATION IN FRESHWATER BIOLOGY. APPENDIX E. LIST OF ALL TAXA SAMPLED AND RATIONALE FOR THE ASSIGNMENT OF FEEDING GROUP. iv.

(7) The widespread di stributi on of wi llow trees(Salixfr agilis) has been thought to i mpact deleteriously oni n-stream faunasi n south-eastern Australi an ri vers. Thi s thesi s ai med to address some of the speculati oni n the li terature regardi ng thei mpacts of wi llows through three main research areas. Fi rstly, a survey was used to compare ri pari anfuncti on of wi llow vegetation to nati ve ri pari an vegetati on and associ atedi mpacts on macroi nvertebrate populati ons. Secondly, the same approach was used to exami ne di fferencesi n macroi nvertebrate andfish populati ons between wi llowed vegetati on and reaches where wi llows has been removed. Fi nally, the role of wi llow large woody debri s(L WD) i n Tasmanian ri vers wasi nvesti gated. Thi si nvolved a census of large woody debri s standi ng stocksi n 142 reaches onTasmani an ri vers. The ecologi cal role of wi llowL WD was i nvesti gated vi a a compari son ofi n-stream nati ve wood to wi llow wood and the associ ated effects on macroi nvertebrate andfish populati ons. In thi s thesi s, large woody debri s(L WD) refers to large organi c woody materi al defined conventi onally as greater than 1 .0 m i n length and0 . 1 m i n di ameter(Gi ppel, 1 995).. The pri nci pal effects of wi llow vegetati on on the bi ota occurredi n summ er and were due to a combi nati on of shadi ng effects and decreased waterquali ty and alterati ons to channel morphologyi n wi llowed reaches. Whi le reachesi n nati ve ri pari an zones supported hi gher densi ti es and numbers of taxa, these were si gnificantly loweri n wi llowed reaches. A sli ght effectwas observedi n autumn as macroi nvertebrate di versi tyi n wi llowed reaches was lower than nati ve reaches. I concluded that wi llows act as a poor surrogate for nati ve ri pari an vegetati on.. Compari sons between wi llowed reaches and reaches where wi llows had been removed revealed major di fferencesi n resources deri vedfr om ri pari an vegetati on. Wi llowed reaches had hi gh organi c matter standi ng stocks and usually low epi li thi c growth on the substrate. In contrast, removal reaches had lower organi c matter standi ng stocks and hi gher epi li thi c bi omass. The macroi nvertebrate populati ons refl ected these di fferences. Although no. v.

(8) differences were observedin summar y variables such as density or taxon number, differences were found betweenfunctional feeding groups. Groupings generally reflected the food sources available in either a vegetated reach with a high organic input and a dense canopy or a non­ vegetated reach with no canopy, higher incidental sunlight and therefore a denser epilithic cover. A separate study revealed that in extreme situations willowed reaches are severely impacted with a large decline in waterquality and high organic standing stocks eliminating most intolerant taxa. Fish populations at these sites were also depauperate, while at remaining sitesfish species showed a strong relationship with their preferred habitat.. Census estimates of woody debris revealed that rainforest vegetation has the highest standing stock ofL WD across a spectrum of ripariap. types. Usually removal of woody native vegetation oftenin concert with active removal of in-streamL WD accounts for lower wood loadings in theTasmanian rivers surveyed. WillowL WD is not common in rivers in Tasmania and is a poor ecological substitute for the more complex native debris, which supported higher densities and richness of macroinvertebrate taxa than willow wood; however, both wood types supported similar community composition. L WD provided important habitat for thefish populations surveyed and reduced or negligible standing stocks ofL WD corresponded to a reductionin the number and size of particular fish species.. The findings confirm some of the speculations regarding the impact of willows on rivers in south-eastern Australia. Wiflows were foundto be a poor surrogate for native vegetation althoug h they provided important riparian resources in the absence of any vegetation atall. The restoration of riparian zones and selective and strategic removal of willowed vegetation over the long term and replacement with endemic vegetation should minimise the ecological impacts of riparian vegetation removal on macroinvertebrates and fish.. vi.

(9) Acknowledgments I. would like to thank my supervisorDrL eonBarmuta for his advice, assistance and continual. support throughout the course of this thesis. His constructive criticism was always appreciated and his support very much valued.. I. would also like to thankDr PeterDavies for his advice and encouragement.. I. would like to thank theL and andWaterResourcesResearch andDevelopmentCorporation. forfinancial assistance in the form of aPostgraduateResearch Scholarship and for top up funding under theNationalRiparian Program( Sub- ProgramBl- Sources ofEnergy) for work onL WD.. I. would also like to thank various volunteers who provided technical andfield assistance. throughout the course of the project. These areRomaRead, BelindaRobson, Ed Chester, GrahamO'Meagher, AngelaMcGaffin, DerekTurnbull, PaulL ewis, L ouisJaclido, David Oldmeadow, HenryMaxwell andJonWaters.. Thanks toGlenMcPherson andDavidRatkowsky for assistance with design and analysis.. Special thanks to workmate$, BryceGraham for assistance with analysis and graphical presentation andDavidFuller( my boss) for support during the latter stages of this study.. To thenumerous property owners who allowed property access to various rivers in the study and the manyL andcare groups who provided valuable information and anecdotal evidence on sites studied and willow removal issues.. Finally none of this would have been possible without the continual emotional support and love of my wifeGail to whomI owe the completion of this thesis to. Her ability to single handedly care for our two childrenMalcolm andElizabeth is something whichI will always treasure in our commitment and love for one another. I cannot thank her enough.. vii.

(10) 1. General Introduction. Riparian vegetation is recognised as having a key influence on in-stream biological function through shading and inputs of litter (Cumm ins, 1993), and the relationship between different types of riparian vegetation and the impacts of human disturbance on riparian vegetation has been well documented (Hawkins et al., 1982; Dudgeon, 1989; Quinn et al., 1992b; Townsend. et al., 1997). By contrast, far less is known about riparian-stream linkages in Australian lotic systems (Bunn, 1994). In particular, the in-stream impacts of a number of invasive exotic riparian species in rivers have been the subject of much speculation but remain poorly documented with any empirical data, and this situation is exemplified by the widespread introduction of willows (Salix spp.) in many temperate lowland rivers in Australia.. Willows were first introduced to Australasia in the 191h century and are now the dominant riparian tree in many lowland rivers in south-eastern Australia (Mitchell & Frankenberg, 1993; Cremer et al., 1995) and New Zealand (Collier, 1993; Glova& Sagar, 1994; Lester et a/., 1994a). Their expansion along rivers is contentiousl with willows being promoted for their value in bank stabilisation and "soft" river engineering works by some (e.g. Strom, 1962; Nanninga et al., 1994)) or reviled by others because of the hydraulic problems they sometimes '. cause and their putative impacts on in-stream fauna (e.g. Standing Consultative Committee on river improvement, 1983; Frankenberg, 1995; Ladson et al., 1997). Despite the controversy, there have been very few formal investigations of their in-stream ecological impact (Schulze & Walker, 1997), while the few investigations that have taken place have generally been inconsistent in their findings (Latta, 1974; Besley, 1992; Glova &Sagar, 1994; Lester et a!., 1994a). This is probably due to site specificity, with all of the published studies being restricted to a few (generally <3) sites, which are usually located in the same river system. This narrow empirical base prompted this study, where I sought to find general patterns across a variety of small to medium-sized rivers in Tasmania.. 1 IIJ.

(11) In the remainder of this introduction, I will provide a brief exposition of the characteristics of willows that have led to their expansion to the extent that they are now regarded as the dominant riparian tree species in many lowland riparian environments in Australia and New Zealand. Next, I will briefly review both anecdotal evidence and past research concerning the influence of willow riparian vegetation on in-stream fauna, and identify core research areas which arise. I then identify and elaborate the general hypotheses relating to the core research questions which I cover in each chapter of the remainder of this thesis. Finally, I provide the rationale and justification for the survey-based nature of the research presented in this thesis.. Characteristics of Salix spp.. Willows are not native to Australasia and all species have been introduced during the last years (Thompson & Reeves,. 150. 1993; Cremer et a!., 1995), although Radcliffe (1 990) maintains. that the major expansions ofriparian willows only developed in Tasmanian rivers systems over the last 50 years . Their establishment and expansion has been most successful where cuttings were first used to control bank erosion following clearing of fertile riverflats, where native species had been removed to maximise agriculture (Bobbi,. Some 28. 1999).. Salix taxa are now naturalised in Australia (Ladson et al., 1 997), although Salix. fr agilis L . , Salix babylonica L.(weeping willow) an d Salix alba L.(white willow) most common species in Australia, while in New Zealand, S. fr agilis and Salix (grey willow) are the more abundant of the. are. the three. cinereaL.. 1 7 naturalised species recorded (Partridge, 1993).. S. fr agilis, in particular, spreads aggressively along rivers, especially in Tasmania (Cremer et at. , 1 995) and New Zealand (Thompson &Reeves, 1 993) so that they are now the most dominant willow species along these lowland rivers (Cremer. et al., 1 995).. Under natural conditions in the Northem Hemisphere, the large species of Salix grow in the fertile soil of river valleys (Meikle,. 1 984). Almost all species of willows are deciduous (losing. their leaves in autumn) and dioecous , but many species can spread vegetatively (Cremer. et a!. ,. 1995). In the Northern Hemisphere they are regarded as early colonists or pioneering species , usually being among the first woody shrubs to colonise areas following natural disturbances or soil erosion associated with flooding (Meikle,. 1 984; Cumm ins, 1 993). 2.

(12) MostSalir.spp. tolerate long periods of drought and their prolific seed production and highly _. ble distances effective adaptation to wind dispersal ensures their distribution over considera and a wide variety of lowland and wetland types (Meikle, Reeves,. 1 984; Partridge, 1 993; Thompson &. 1993). As most taxa in Australia are represented by male clones, seed production is. not the primary means by which willows have proliferated in Australian rivers, although seed set is now being recognised as a more frequent occurrence because of records of various hybrids in a number of localities (e.g.. Salix matsudana Koidz (tortured willow) in Tasmania. and Sal ix humboldt iana Willd (pencil willow) in NSW (Cremer et al .,. 1 995)).. Thus willows have all o f the features that Ricciardi & Rasmussen (1998) identified as general attributes of invasive aquatic species: (i) wide environmental tolerance, (ii) high reproductive capacity, (iii) rapid dispersal using natural mechanisms, and (iv) rapid growth. All of these attributes have been responsible for the rapid expansion of Salix spp. along rivers in both Australia and New Zealand (Cremer et al.,. 1 995; Frankenberg, 1 995; Askey-Doran et al .,. 1 999).. This expansion has been facilitated by vegetative reproduction resulting from the dispersal by rivers and streams of fragments of twigs and branches which readily establish on banks and shoals downstream; this generally occurs during storms or floods (Frankenberg,. 1 985).. S. fr agil is, in particular, i s prone to fragmentation (Hathaway, 1 986; Mitchell &Frankenberg, 1 993) as branches are more brittle or "fragile" than other willow species, and this characteristic, along with its vigorous growth (Thompson &Reeves, '. 1 993) accounts for the. greater expansion of this particular species along rivers and streams. Willows are also thought to outcompete native riparian species by overshading. Both Askey­ Doran et al .. (1 999) and Frankenberg (1 985) suggested that willow shade excludes all but a. few shade tolerant species and Partridge. (1 993) reported that willow species have the capacity. to out shade pre-existing native wetland vegetation by "overtopping".. These characteristics have led to the extensive and ubiquitous distribution of S. fr agil is in many rivers in south-eastern Australia and New Zealand and its classification as an environmental weed in both countries (Partridge,. 1 993; Askey-Doran et a!. , 1 999). In addition,. 3.

(13) the major characteristics of willow species such as deciduous leaf fall, dense shading and i. prodigious root growth have been suggested as the principal factors contributing to impacts on in-stream fauna in Australian and New Zealand rivers (Collier, 1 993; Frankenberg, 1 995).. Existing research and anecdotal evidence of in-stream effects of willows �-. Despite reviews by various authors, both in Australia (Mitchell &Frankenberg, 1 993; Ladson. et a!. , 1 997) and New Zealand (Collier, 1 993), little research has directly examined or quantified the impacts of willow species on in-stream fauna. Historically studies have compared leaf breakdown rates and conducted feeding preference trials for various macroinvertebrate taxa and concluded that willow leaves break down faster than some native Australian and New Zealand plant species (Pidgeon & Cairns, 1 9 8 1 ; Collier & Winterbourn, 1 9 8 6; Yeates, 1994; Parkyn & Winterboum, 1 997; S chulze &Walker, 1 997). Many of these studies have speculated that the combination of fast leaf breakdown rates for willow leaves and their abundant input over a short period of time in autrnnn will be detrimental to stream faunas which might be adapted to slower breakdown rates and more continuous supply of leaf litter provided by many native riparian species in-stream (Mitchell &Frankenberg, 1 993).. Only a few studies have directly compared in-stream fauna in willowed riparian zones with native riparian zones in Australia (Pidgeon &Cairns, 1 981; Besley, 1 992; S chulze & Walker, 1 997). Schulze & Walker (1 997) found few differences between invertebrate communities under willowed vegetation and native riparian vegetation in some sites on the large River Murray,-South Australia while Besley (1 992), in the Murrumbidgee River in N.S.W., and Pidgeon & Cairns ( 1 9 8 1 ) , in upland northern N.S.W., concluded that willows affected stream faunas by a combination of shading and litter inputs respectively. The majority of studies, however, have originated in New Zealand where comparisons have been made between reaches with and without willows (Latta, 1 974; Lester, 1992), and between unshaded reaches and those varying densities of willows (Glova &Sagar, 1 994). As with the Australian studies, only a small number of streams have been included in any one study, and the results have been inconsistent. Both Latta ( 1 974)and Glova & Sagar (1 994) concluded that willowed vegetation was beneficial to in-stream fauna in moderate densities, which they attributed to the increased habitat diversity under willows, particularly for brown trout. (Salmo tru tta). In contrast, Lester 4.

(14) 1¢,J ':. l. ;f � "��� i{�: i:�o.ff� & (. \.if� ��61992). �. ... ��· �. . found that willows did affect invertebrate populations, and attributed th1s to a. )t� combination of shading differences and changes in the substrate due to willow root '�- -. �� �;· penetration; he also suggested the possibility of toxic compounds leaching from roots leading to elimination of some invertebrate taxa. Clearly, further research is warranted to quantify these patterns as suggested in reviews by other ecologists (Bunn, 1993; Campbell, 1993; Collier, 1993; Mitchell &Frankenberg, 1993; Walker, 1993). There are, however, two other avenues of research related to impacts of willows and their management that urgently require research in Australia.. Firstly, large scale willow removal programs are currently underway on many Australian rivers under initiatives funded by Landcare and the Natural Heritage Trust (Bobbi, 1999), and the ecological impacts of these clearance operations remain unknown. It is possible that willows act as a surrogate for native trees and shrubs in the absence of any woody vegetation at all. Many willow removal operations replace willows with pasture grasses and, if the linkages between riparian and in-stream processes are as tight as some maintain, this would have clear implications for the supply of detritus and in-stream primary productivity. Research is required to ascertain whether sudden, wholesale removal operations are justified, or more gradual or even selective removal over a longer time period is a more ecologically sensitive procedure.. Secondly, as part of many ri�er management activities, LWD is also removed along with in'. stream willow blockages; historically this "desnagging" has been a common practice along many Australian rivers (Gippel etal., 1996b). Native LWD has been found to be an important .. habitat for macroinvertebrates (O'Connor, 1991; McKie & Cranston, 199 8; McKie & Cranston, in prep) and also for some native Australian fish species (Davies, 1989; Koehn et al., 1994), but, with the exception of Lloyd etal. (1991), no studies have compared the relative benefits of willow and native LWD for macroinvertebrates or fish. In addition, some ecologists (Lake etal., 1985; Campbell et al., 1992b) have noted that temperate Australian streams have a larger component of woody detritus (both twigs and branches and seasonally-shed decorticated bark) than comparable northern hemisphere streams and have speculated that native EucalyptUS'"dominated riparian zones would contribute a greater and more sustained load of woody material. Mitchell & Frankenberg (1993) and Lloyd et al. (1991) suggest. 5.

(15) ..�t�· · �,�·· j;�.·· •' ...�,_J.( ". W'· ' j�( ���· i;' :IU!tlier. that.willow wood may be an inferior habitat and may break dovvn more quickly than \.�, i� :��:-: na. tive woody debris. Again these speculations need to be tested with empirical data. ��. .. .�f� .--. -J... ��. '�iti'. it -�. -. .. ·­. �.,.. ��. ...... �. Core research areas The primary requirements for further research lie in three core areas. These are:. •. Determining the impacts of will owed riparian zones on in-stream biota in comparison to native riparian vegetation.. •. •. Identifying the impacts of willow removal operations on in-stream biota.. Assessing the ecological value of willow wood debris compared with wood debris originating from native vegetation.. These core areas form the basis for the three investigations carried out in each chapter. Chapter 2 uses a survey approach to compare differences in macroinvertebrate communities between willow-lined and native-lined reaches in nine rivers in south-eastern Tasmania. Chapter 3 uses the same survey approach in seven rivers to compare differences in macroinve1tebrate and fish communities between willow-lined reaches and reaches where willows have been removed in shallow rivers and more degraded deep rivers. Chapter 4 investigates the role of willow LWD in Tasmanian rivers using an extensive survey to develop a predictive model of the relationship between LWD type and quantity and adjacent riparian vegetation type. A smaller scale survey focused on the effect of wood type and riparian vegetation on fish and macroinvertebrate communities in two rivers in northern Tasmania. The final chapter is a synthesis of the findings of this research and suggests further work required to substantiate the ecological role of willows in rivers.. 6.

(16) J1 f ' !1. �• . �� - �· t :i :t -��. )if 0. .. 1. �� �reroject Aims and research strategy ' Jifi:t.1" ·; -f"r\ s.... ' ;, ·. i. �. � .. -:�,. �t··· · 1� �� r:::. .. �. �·:. f�. s.,. · .,. •. '. I·.formed three broad hypotheses that form the central investigations of each chapter. Firstly, I suggest that willow riparian vegetation adversely impact in-stream faunas in comparison to. endemic native vegetation by shading in summer and willow leaf accession in autumn. In these seasons� I suspect that in willow-lined reaches the macroinvertebrate fauna will be different from that found in reaches under native riparian vegetation. I also suspected that infiltration of the river bed by willow roots of these smaller rivers would affect the composition of macroinvertebrates by denying them access to interstitial spaces.. Secondly, I hypothesised that reaches where willows have been removed will have different faunas compared to wiHow-lined reaches and these differences will be predominantly a result of the removal of shading and reduction of allochthonous inputs. Reaches where willows have been removed will most likely be characterised by a fauna that are tolerant of unshaded conditions and also best able to exploit the expected abundance of autochthonous material brought about by an increase in primary production in an unshaded environment.. Thirdly, given the suspected differences in wood substrate quality between willow LWD and native wood types, I hypothesised that willow wood is a poor surrogate for macroinvertebrate habitat in comparison to the more complex native wood substrates and that the macroinvertebrate communities on willow wood will be different from those communities ,. found on native wood substrates. I also suggest that macroinvertebrate community composition will also be different under different riparian types given differences found by .. other ecologists for benthic macroinvertebrate populations under different riparian vegetation types (Dudgeon, 1989; Quinn et al., 1993; Glova &Sagar, 1994; Reed et al., 1994; Townsend et al., 1997). Furthermore, given the vegetative characteristics of willow species and the anecdotal evidence surrounding quantity of willow wood accession into rivers (Frankenberg, 1995; Ladson et al., 1997), I suggest that the standing stocks of willow wood will be comparatively less than native LWD standing stocks in rivers.. All these general hypotheses imply that willowed riparian zones impact in-stream communities and are a poor substitute for native Australian riparian vegetation in providing natural riparian. 7.

(17) J:t.,pf·:�;-.' � ;,. !f. . .f l. ..:� ' i·ft .. '•. :.p. "Y'�-... 1. :- �ifi'n kaoes �� :� ���. Ailstralian native plant species. I suspect that willowed riparian zones and riparian zones -. �. environments, although careful consideration needs to be given to the to aquatic � .l/,.t1r�b. . }: 1�'l' -r �r ��latLve benefits or impacts of removing willowed vegetation without replanting of endemic .;J:'.i ·. .;. .. -·· ·. •I �. J. '. <. .. •. -without shading by woody species (i.e. unshaded reaches) both impact river ecosystems and that neither vegetation type is a good substitute for the Australian riparian vegetation which in comparison is both structurally and floristically diverse.. Beyond hypothesis testing, a major objective of the study is to provide ecological information and recommendations to catchment managers and community groups in regard to ecological prescriptions for riparian restoration and rehabilitation practices. Rehabilitation of degraded riparian vegetation should not only be carried out for aesthetic or economic goals but primarily aimed at restoring riparian linkages to river community structure and function.. Rationale for the use of surveys in this study. Overview of approaches in environmental impact assessment. Designs for investigating impacts should have three essential features (Green, 1979). Firstly, they should clear, testable questions using measurable variables (Mapstone, 1995). Secondly, they should be statistically tractable and robust without violating the assumptions of the proposed statistical test (Keough & Mapstone, 1995), and finally, they should be carried out over appropriate spatial and temporal scales with sufficient precision and power to test the specific questions that were originally identified (Underwood, 1994b).. Three main approaches have been used to quantify impacts on ecosystems (Underwood, 1991a): (i) surveys and descriptive studies, (ii) before-after-control-impact (BACT) designs, and(iii) field experiments. The relative merits of these different approaches have been well reviewed, and there has been much activity over the last decade on improvements to the BACI design originally proposed by, among others, Green (1979) and Underwood (1991), as well as increased interest in incorporating field experiments into environmental assessments (Cooper & Barmuta, 1993). Each general approach has its strengths and weaknesses in practice.. 8.

(18) -�' � '""!?. ;' :�� f· . j ;f{. · .·� ·:>-. "". I. \�):.��. eys were used in all investigations in this study, and this approach has been commonplace. �-�: �similar studies investigating effects of riparian vegetation on in-stream fauna (Dudgeon, �I" : ¥ 1 989; Lester et al., 1994a; Reed et al., 1994). Surveys examine relationships among biotic and .. �� .. ". �.... .. '". abiotic factors across a number of sites or times (Cooper&Barmuta, 1993) and, in rivers ,. '. usually involve comparisons of unaffected upstream areas to putatively impacted downstream areas. Such designs are problematic because it is possible that some factor other than that being tested may affect the downstream impact area but not the upstream control (Underwood, 1990; Norris& Georges, 1993; Underwood, 1994a). Many surveys have additional problems. in that environmental conditions at the upstream and downstream sites are likely to be different anyway, and upstream processes may affect downstream results (Cooper&Barmuta, 1993). This has been called variously "pseudoreplication" (Hurlbert, 1984), "pseudodesign" (Eberhardt& Thomas, 1991)or, more simply, "confounding" (Underwood, 1994a).. Potentially before-after-control-impact (BACI) designs can deal with the lack of spatial and temporal replication inherent in surveys, and thus overcome false conclusions resulting from confounding as suggested in (e.g. Underwood, 1991a; Osenberg et al., 1994). Early formulations such as that offered by Green ( 1979) still suffered from a lack of true replication in both space and time at appropriate spatial scales. Stewart-Oaten and co-workers (1986; 1992; 1993) and Osenberg et al. (1994) extended Green's idea by examining a single control and single impact location at multiple times before and after a putative impact; Underwood has subsequently argued that this design is still open to spatial confounding and that multiple .. control (and where possible and appropriate) multiple impact locations should be included as well (Underwood, 1991a; Underwood, 1991b; Underwood, 1 992; Underwood, 1993; Underwood, 1994a; Underwood, 1994b). Keough& Mapstone ( 199 5; 1997) broadly support Underwood's position, but note that some of his more complex proposed designs may be difficult to optimis e and implement, and earlier research has also noted some practical difficulties in meeting the assumptions of analysis of variance when the available time scales are short (e.g. Millard et al. ( 1985; 1986)). Despite some deep differences held by some of these researchers there are three major areas of agreement. Firstly, for BACI procedures to be valid, a long time series of data needs to be collected both before and after the putative impact occurs to ensure independence between sampling times. Even for invertebrates with univoltine life-histories, this means that rivers must be sampled for many years to establish patterns of. 9.

(19) .lilfr .,. �j'3� . J! {ti. fP �· �ii$� �� �r �annual.variability to act as a baseline comparison to post manipulation changes. �-:t�i�-J;-t '���condly, for causation to be linked to any statistical design, the investigator must be able to •. �.. ��.. ... ,�. �. ��:��-�· allocate ·. �. 'f'o•·. experimental units (in this case independent reaches of a river) at random to. "treatments" (i.e. to ''control" or "impact", or "willow-lined" v. "native"); investigators of environmental impacts are clearly limited in their capacity to achieve this, and so even the best BACI design is hampered from inferring causation in the way that a fully randomised experiment can (Eberhardt &Thomas, 1991; Keough&Mapstone, 1995; Keough &Mapstone,. 1997).. The use offield experiments in environmental assessment has also attracted much interest over the last 20 years, and the advantages and disadvantages were clearly identified by Cooper & Barmut a (1993). The chief consideration relevant to the current study is that, although such experiments circumvent the issue of identifying causation in a statistical sense" experimental manipu lations are often unrealistically small spatial or temporal scales so that extrapolation to the field is usually problematic.. Ideally, therefore, it is probably most desirable to adopt a process that uses a variety of lines of evidence to infer or detect impact in rivers. Hall and co-workers (reviewed in Hall (1978)) arrived a comprehensive appraisal of the effect s of forestry activities on fish in north-western U.S.A. by conducting a combination of surveys, small- and large-scale manipulations, and some before-and-after studies over a number of years. The surveys served to identify patterns .. and suggest mechanisms that might be behind the changes. Some of these mechanisms were amenable to experimental test, and the before-and-after studies served to test predictions made . from survey and experimental results. A similar combination of approaches have been employed in investigating the effects of acidification and remedial liming in Wales (Merrett et. al., 1991; Neal, 1992; Reynolds et al., 1992; Weatherley& Ormerod, 1992).. Approach adopted in this study. In contrast with point-source contaminants such as sewage effluent and heavy metals from mining operations, there are little data about t h e patterns of macroinvertebrate and fish faunas. 10.

(20) �� ;�\$. .�JA4f1 P '': · -�11 ,�. " s¥ •t ·F , : · . ifit, � � , l }.�\ .·. ,. i troduced willows as their riparian vegetation. As outlined above, ;� . .�iliat are:fomid in rivers with n 1, ·· · :· .·Mi"'at data exist are inconsistent, and this may be a reflection of the limited spatial scope of the ' l ':' � \=: ,. .�. �. .. ·. .. -1. ��:. -. •. ; . .. ... .,. .�1i1f studies conducted thus far. Thus my main justification for using a survey-based approach for. � ... ol• ·. this study was that I wanted to establish whether there really were any patterns that were. common across separate rivers within the same geographical region. If such patterns do exist, then these surveys can then be used to generate more process-based hypotheses that might be amenable to testing via field or laboratory experiments. The surveys would also provide n i sights into what variables would be most useful to measure in any future BACI-type studies.. My reasons for not using BACI designs were principally logisticaL Firstly, the time span available for this project was not sufficient for such studies. BACI designs would only be remotely feasible for studying willow removal (even if plantings were still being made, willows do not grow quickly enough even for a Ph.D. project !). Thus, for example, ifl wanted to examine the impact of removing willows on densities of shredders in autumn, I would need to have measured these densities for several autumns both before and after willow removal to avoid some of the problems of confounded effects in time (and I would also need to have monitored densities in several control sites where willows were left intact to avoid spatial confounding).. An alternative, and less stringent BACI-type design might have been to visit many willowed sites just once before and once after some of the sites had had willows removed. However, I '. fom1d it virtually impossible to frame a monitoring program around the activities or objectives ofthe commun ity groups and landowners responsible for willow removal in Tasmania. During the first year of my study, there was a moratorium on willow removal , and in subsequent years removal plans were often changed at short notice. Latta's. (1974) experiences in New. Zealand were salutary. He attempted to employ a BACI-style approach to investigate the effects ofwillows on brown trout populations (S. trutta) in six rivers. Latta. (1974) required. three sections on each river consisting of two experimental sections in which the density of willows would be changed and one control section (completely clear of willow cover) in which the density of willow would be left unchanged. Of the six r v i ers originally chosen for study, three experimental sections were abandoned within a year due to large decreases in trout population density between pre-removal sampling occasions and another experimental section. 11.

(21) .\. ,�t ; ' •i ·:� w· .. Y ) f� �!1 ¢" ,••r• ,. \. I. 1.r-: 't .� ·� ii). >{f: -�� :;� ���-�aslost due.to the spraying of willows. Four alternative experimental sections were then .•. l. :\ : ·. r-. �.iJ�it;.Iected. Two ofthese sections were subsequently abandoned due to willow clearance and .,. -�-� : � stream bed alteration by bulldozing. Latta (1974) chose another experimental section and despite further willow clearance and large decreases in pre-removal trout densities ( overcome by stocking experimental sections) finally carried out the study to completion on three rivers.. \ �. .... ,. .. �"' �'. �:-!' "">.­ �. Finally, any post-removal monitoring I could have implemented would have been restricted in its temporal extent. Casual observations ofremoval operations over the last few years suggests. that it may take >2 years for the bed and banks ofthe river to stabilise after the operation, and. even longer if the beneficial effects of any riparian re-planting program are to manifest themselves.. The experimental unit that I wanted to generalise to was a "river reach" that was long enough to encompass several riffle or run and pool sequences. This was the appropriate spatial scale to test the hypotheses that I have identified (Cooper &Barmuta, 1993; Downes et al., 1993) since willows are usually planted or removed at "reach scales" (i.e. 101 to 102 of metres). I sought to. reduce spatial confounding by ensuring that there was just one of each reach type (native or willow-lined in Chapter 2 and willow-lined or willows removed in Chapter 3) on each independent, separate river used. Thus, in design terms, rivers were serving as "blocks" for the analyses in these two phases of this project. Thus to increase the power of the contrast of most interest (i.e. between reach types), I sought to maximise the number of rivers in each of these surveys (cf Underwood, 1994a). As variation in abundance of animals occurs across a range of spatial scales (Downes et al. , 1993; Johnson & Gage, 1997), I also included a range of spatial scales and appropriate stratification in the surveys in Chapters 2 and 3. Although "rivers" are the unit of replication for these surveys, small�scale, within-reach variation influences the precision of estimates of variables, which can indirectly affect the power ofthe test for the contrast of most interest (Downes et al. , 1993; Keough &Mapstone, 1995; Keough &Mapstone, 1997). Thus, the surveys were nested, multi-stratum designs (McPherson, 1990), with sample units stratified into fast· and slow�flowing habitat types where appropriate (Norris &Georges, 1 993; Resh & McElravy, 1993). (This was not always possible, particularly in Chapter 4 where, after extensive searching, only three rivers could be found that satisfied the criteria for a portion of 12.

(22) · � : � ·tf �-Sl���t�. ���!�, . . ··t�. -\. � .�fJ:.· � '�i.ffl � t phJSe ofthe investigation; this was further compromised when one ofthe rivers was " '� " ' '\ f. 1( . :-"' �·· ��ubjccted to a 1 : 1 00 year flood event.). To determine the number of sample units per reach, ten �f:_. iivcrreache s were sampled in a pilot study in December 1993; following the procedures ; outlined in Elliott (1983), 10 sample units per reach were sufficient to estimate the densities of. t. ''"�. ·. �. ... ·:.-. !. ·�· ··· v. the most abundant invertebrate taxa to ± 20% ofmean density with. 95% confidence.. Additional steps were taken to reduce differences between sites other than those being examined (riparian vegetation) and to minimise upstream downstream difference due to natural processes. This typically involved habitat assessments at each site for inclusion into the mnin survey and map based assessment to ensure that geomorphological and hydrological characteristics of each reach were similar enough to warrant further comparison. These precautionary measures are elaborated upon in each chapter.. �. I. ·.c--=·======��..::;:::=. :;:: _� .. . ... ��..... �---._... . .:..= ;.;..: __. .. .. .... ""'-" ::::. . ,_ _ _ ---. -. --. - ._. ..J. 13.

(23) ..,. .. Introduction Replacement of native species of riparian vegetation by exotic species is a world-wide phenomenon. In some parts of Europe, for example, Eucalyptus plantations are causing concern because in-stream fauna seems less likely to use this material than native species (Busaguren & Pozo, 1994; Pozo et a/. , 1 997). Conversely in temperate parts of Australia and New Zealand replacement of native species by willows (Salix spp.) raises similar concerns that riparian functions (sensu Cummins, 1993) will be altered.. Clearance of river banks in agricultural areas in south-eastern Australia and New Zealand has resulted in many banks in lower catchments being dominated by willows (Pidgeon &Cairns, 1981; Cremer et a/. , 1995). These introduced willow species are distributed widely along rivers in Victoria (Ladson et a/., 1997), New South Wales (Cremer et al., 1 995), Tasmania (Askey - Doran, 1993) and New Zealand (Collier &Winterboum, 1986; Glova &Sagar, 1 994). As a result, many ecologists have speculated that the differences between deciduous willow species and the native evergreen Australian and New Zealand vegetation will affect the stream biota (Campbell, 1993; Collier, 1993 ). Native vegetation throughout this thesis is defined as endemic Australian woody tree species.. Willows are quite distinct from native Australian trees and shrubs on several grounds. Willows are deciduous, dropping all their leaves in autumn (March - May) over a relatively short period of time (Frankenberg, 1 995). Willow leaves also break down faster than many native species (Pidgeon &Cairns, 1 98 1 ; Yeates, 1 994; Frankenberg, 1995; Schulze &Walker, 1997) which suggests that willow detritus represents an abundant food source that is available for a relatively short period of time that may not be fully utilised by macroinvertebrates within a given reach (Lester et a/. , 1 994a). By contrast, native vegetation is evergreen and has a continuous leaf fall throughout the year, usually with a slight peak around late summer (December- February) (Campbell et al. , 1 992b; Thomas et al., 1 992; Swain et a/. , 1993).. 14.

(24) ·�). .��;, .� �- .. if. ��. '· <F. jj. .. ! :?. tit i{ ¥ rhis·means that there is likely to be a more continuous food source for macroinvertebrates :.i .� ·��g-,yhich have presumably adapted to this pattern of litter accession. Jf. .I$ f-.,_. :1 {(. ·:. Another potentially deleterious effect from willows is a change in the light climate of the. · .. stream bed. Shade is an important factor affecting water temperature and the level of primary production in streams and rivers (Dawson & Haslam, 1983), and several authors have found that algal growth has been absent from densely willowed streams when compared with native sites (Pidgeon, 1 978; Glova &Sagar, 1 994). They suggest that the dense canopy of willows ( Plate C12, Appendix C) alters primary productivity in rivers and that this may be reflected in invertebrate assemblages (Besley, 1 992; Lester et al. , 1 996).. In addition there are the physical effects on the habitat wrought by the denser root mats of willow species. At the reach scale, willow trees can invade river beds leading to blockages and channel diversion (Lester et a!. , 1994a). At smaller spatial scales willow roots can infiltrate interstitial spaces between rocks, particularly in riffles (Lake & Marchant, 1 990) which may lead to decreased habitat availabilit y and fewer refugia for macroinvertebrates (Lester et al. , 1 994a), decreased heterogeneity of habitat for fish and aquatic plants, and a simplified community on the bank itself (Collier, 1993; Walker, 1993).. Despite these differences between native and willow riparian vegetation, the impacts of willows on in-stream fauna have yet to be quantified comprehensively, despite a few localised studies (Pidgeon &Cairns, 1 9 8 1 ; Besley, 1 992� Hardwick & Wood, 1 997). The majority of research in Australia and New Zealand has focused on willow leaf breakdown rates in aquatic ecosystems (Pidgeon &Cairns, 1 9 8 1 ; Collier &Winterboum, 1986; Yeates, 1994; Parkyn &Winterboum, 1 997; Schulze &Walker, 1 997) and correspondingly for Australian native riparian species (Pidgeon &Cairns, 1 9 8 1 ; Bunn, 1988; Campbell et al. , 1 992a; Yeates, 1 994). Studies of the in-stream fauna have been comparisons between willowed reaches and unvegetated banks (Latta, 1 974; Lester et al. , 1 994a) or differing densities of willows along river banks (Glova &Sagar, 1 994) in New Zealand, while Pidgeon & Cairns ( 1 9 8 1 ) documented differences in the fauna colonising leaf packs in native and willow-lined reaches of one stream in northern New South Wales; at most only 3 separate rivers have been included. 15.

(25) �· ����. If;·� t�t'-�.'�·: 1;�f nf � "!!ti�-. j��� � t '�4q -P.J.\:�� ·"' "'�:. }}1. f". �. ·. ;. �· \.. •�·�. � �of these studies. There have been substantial differences in the results and many of the . : uestions surrounding the impacts of willows on in-stream fauna remam unresolved.. ·-.&�·'·!",... ·'. _.. l. .. 'Ji'o.examine some of the hypothesised differences between native and willow-lined reaches, I conducted a survey of nine small to medium sized rivers in south-eastern T�mania. These ,. rivers were suited to this comparison because they had similar in-stream habitats, similar flow regimes and land uses in their catchments (Table 1 ) I used a survey-based approach to document differences in the fauna and key habitat variables in order to determine whether the .. observed differences were consistent with the predictions of the aforementioned speculations. By examining these differences across a number of rivers, I sought to find \Vhich patterns were clear and consistent While a survey of this nature does not provide strong evidence of causal links between vegetation type and in-stream fauna, I regard it as an important first step in generating plausible hypotheses for later testing. Specifically I hypothesised that differences between macroinvertebrate communities would be most apparent in seasons when relative litterinputs and shading ofwillowed and native reaches are different, i.e.. in autumn (a high leaf fall in willowed reaches), spring (negligible. leaf fall and a bare canopy in willowed reaches) and summer (a dense canopy in willowed. reaches and a slight peak in native leaf fall). I also examined the hypothesis that willow roots. alter the faunal composition of riffle habitats by comparing portions of stream bed free of and. invaded by willow roots in willow-lined sites. in summer.. Methods an d Materials Survey Design. The survey design consisted of three strata: 9 rivers within each of which 2 reaches were. selected (one lined with willows, the other with native vegetation), and within each reach two habitats (fast- and slow-moving water) were distinguished;. sample units were allocated at random to these habitats, with the number of units dependent on the variable being measured (see below). Essentially the design resembles a randomised complete block design with rivers serving as blocks, and with plots (reaches) subdivided to form sub-plots (habitat types) and a two factor arrangement of treatments with vegetation type "allocated" bet\veen plots within 16.

(26) . f,�; :! -�; ·f. ;��J�· -. L \.. l. i. 'i. .� ,.*.; f"l·,:;.· ·-Ay;. ·Jfi � ·'iB cks .. .rr�·��rusmg. !. .i< ••• '· [t;...,•. I. ��;_!'/,:t "" '' . •. ... ". and.habitat type "allocated" between sub-plots within plots. The data were analysed. analysis of variance, with the appropriate error tenns for each stratum chosen according. fo·the standard procedures (McPherson, 1990), Because I was not able to allocate vegetation. or habitat type at random, significant differences for these terms in the analyses need to be interpreted with caution. A significant difference for vegetation type, for example, may not indicate that the observed difference was caused by the differences in vegetation type because other factors may covary (and hence be confounded) with the change in vegetation type (McKie &Cranston, 1998). Unfortunately, in south-eastern Tasmania, the lower parts of river catchments were settled and cleared first, so that all my willow-lined sites were downstream ofthe native sites.. To minimise the potential for confounded effects between willow- and native-lined reaches, I ensured that the sites were sufficiently close to each other (<1 0 km and <120 m difference in altitude; Table 1) so that any differences in the fauna resulting from longitudinal or altitudinal changes were minimised as judged from zonation studies elsewhere (e.g. Brussock & Brown, 1991; Corkum, 1991) and from faunal information collected on sequential sites in similar. riparian vegetation in northern Tasmania (Bobbi et al., 1996). Sites were also separated by at least 1.2km to minimise the effects of downstream transport of native litter and invertebrates. In all other respects, the sites were as similar as possible. They all had rocky-bottomed. habitats, supported similar in-stream floras and faunas, drained catchments with the same geology and geomorphology (Davies, 1988), and there were no point source discharges or other obvious sources of pollution between the sites. Those rivers that have been gauged show the same flow patterns (Hughes, 1 988), with low flows during summer and autumn (January­ l\Aay) and high flows in late winter (July-October). Willowed reaches were surrounded by pasture, although on some reaches, apple orchards were set back at least 50 m from the river (Table 1).. Study sites. Nine rivers were chosen for study. The approximate length of each reach was lOOm. Eight rivers had native riparian reaches dominated by open wet sclerophyll forests dominated by 17.

(27) j.. I'. t�.. !l �J." ..... -,· ;.���fl.. �. ( �. !. :�. �. �. �. � ' ;Z �{w;,.,k (Eucalyptus ob/iqua L'Herit), swamp gum (Eucalyptus regnans F.Muell.) with an. -�s-� ---· ·.' h�iY.( �d�rsto rey ofblackwood (Acacia melanoxylon R.Br.), dogwood (Pommaderris apetala. ,� 1;. � ". �·v. · ·. -:- •'i.•. .. ... �y;. :.� ,.. tabill.), woolly tea tree (Leptospermum lanigerum (Ait) Sm.) and silver wattle (Acacia. �. ;' ·. dealbata Link). GroWldcover was a mixture of fishbone water fern (Blechnum nudum),. bracken(Pteridium esculentum) and blackberry (Rubusfruticosus L.). The native reach on the Kermandie River was cool temperate rainforest, but was still dominated by stringybark (E.. obliqua) with an Wlderstory of myrtle (Nothofagus cunninghamii (Hook.) Oerst) and sassafras. (Atherosperma moschatum Labill.). Willow sites on all nine rivers were dominated by crack. willow (Salixfragilis L.) with sparse cover of other exotics such as poplar (Populus nigra L.). and hawthorn (Crataegus monogyna Jacq.) with a groundcover of blackberry (Rubus. fruticosus L.) and some remnant native fern species (Blechnum nudum (Labill.) Met.ex. Luerss.). Site photographs are provided in Appendix C (Plates Cl - C 1 8).. In all other respects the sites within each river had similar in-stream habitats. All sites have a dominant substrate of boulder and cobble with underlying gravels, although in willowed. reaches, some riffles were partially overgrown by willow root mats. Riffles overgrown by root. mats were treated as. a separate habitat and were sampled only once in summer to explore the. effects of willow roots on the in-stream faWla.. 18.

(28) Table 1 Physical characteristics of native and willowed reaches in each ofthe 9 rivers in south-eastern Tasmania. Land use code: WS = Wet sclerophyll forest; P = Pasture; R. =. Rainforest; 0. =. Orchard.. River. Reach. Catchment Area (km2). Stream order. Altitude (m). Mean width (m). Latitude/Longitude. Bedslope. Land use Distance between sites(km). Browns. Native Willow Native Willow Native Willow Native Willow Native Willow Native Willow Native Willow Native Willow Native Willow. 14 50 13 41 20 57 17 19 34 172 27 71 15 16 25 32 35 93. 2 4 2 3 2 4 3 3 2 4 4 4 2 3 4 4 2 3. 120 20 180 160 140 60 60 20 140 15 120 20 180 160 80 50 120 20. 4.9 3.8 4.9 7.9 7.8 6.5 10.5 5.6 13.6 6.5 4.1 4.5 2.1 2.0 5.6 5.3 15.2 15.0. 147"15'20"E , 42"57'12"S 147"19 ' 5 1 "E , 42"5 8'00"S 147"01'30"E , 42"5 1 '06"S 147"0 1 '44"E , 42"5 1 '3 3"S 146"52 '40' 'E , 43" 1 1 ' 10' 'S 146"55 '08"E , 43"1 0'30"S 147"14'09"E , 43"04'20"S 147"1 5'20"E , 43"03'40"S 147"07'55"E , 42"56'28"S 147"03'57"E , 42"59' 15"S 14 7"05' 1 6"E , 43"03 ' 55 "S 147"03'30"E , 43"04'22"S 147"04'55"E , 42"43 ' 15"S 147"04'55"E , 42"43 '33"S 147"09'00"E , 43"08'40"S 147"09'00"E , 43"09'00"S 147"12'00"E , 42"5 9'00"S 147"15'50"E , 43"01' 16"S. 0.05 0.0083 0.025 0.024 0.016 0.0067 0.05 0.01 0.016 0.0067 0.05 0.008 0.07 0.04 0.012 0.008 0.01 0.005. ws. Lachlan Kermandie Snug Mountain Kellaways Back Nicholls North West Bay. 4.2. p ws p. 1.2. R. 4.0. 0 ws p. 2.0. ws. 9.0. 0 ws. 6.5. 0 ws. 1.2. p ws p. 1.5. ws. 8.0. p. 19.

(29) Habitat measurements A suite of habitat and physicochemical variables were measured in each reach. Physicochemical variables were measured in each season. Three replicate readings of temperature (OC; WTW Microprocessor ConductMeter LF1 96) and conductivity ().ls/cm3 ; WTW Microprocessor ConductMeter LF 196), pH (WTW Microprocessor model pH95), turbidity (NTU; Hach 2 1 00P Turbidimeter) and dissolved oxygen (mg/L; WTW Oximeter Microprocessor Model OXI96) were measured along each reach. These were averaged to obtain mean values for the reach. The measurement of the above physicochemical parameters were chosen as I expected that these would show the greatest differences between the two vegetation types. The top, middle and bottom widths of each reach were measured in summ er and were averaged to give a mean river width for each reach. Catchment area, distance between reaches and stream order for each reach at the point of sampling were determined from 1 : 1 00 000 maps (TASMAP). Estimation of altitude and bed slope for each reach were determined from 1 :25 000 maps (T ASMAP). Physical characteristics of the 9 rivers are presented in Table 1 .. Benthic Macroinvertebrates. � conducted in December 1 993 sampled macroinvertebrates in all rivers. Using. A pilot stud. these data, I determined that 1 0 sample units were required in each reach to obtain density estimates ± 20% of the mean (Elliott, 1 983). The term "density" throughout this thesis refers _ to the number of individuals per unit area ofthe quadrat.. Macroinvertebrate sample units were taken with a suction sampler (Brooks, 1 994) (area 0.23 m2 ; mesh size 250).lm) at random locations within the reach. Five sample units were taken in fast flowing habitats (0. 3 5 - 1 . 7 m/s) and five in slow flowing habitats (0-0.3 5m/s). Prior to each sample unit being taken, velocity (m/s) (Schiltknecht Mini Water 2 anemometer flow meter; propellor diameter 1 em), and depth (mm) were recorded, and percentage composition of seven substrate categories (ranging from mud through to bedrock) was estimated within the quadrat. The sample unit was then taken over a three minute interval and preserved in 70%. 20.

(30) � ·.. ..J · \ \: .* -;j 1 ::� • � t,..: 6;'-; . �-,:�( 1 t. r,Qi ("cthanol In sum.mer R �·· by willow roots. �· •. !. r.. . •·. ,. .. ,.. f. '. an. additional five sample units were taken in fast flowing areas infiltrated. In the laboratory, samples were separated by elutriation into inorganic and organic. �. components. The organic component was further sieved into coarse (> limn) particulate. >. organic matter (CPOM) and fine (<lmm and > 250J-Lm) particulate organic matter (FPOM).. t. Both inorganic and organic components were sorted for macroinvertebrates. These were. identified to family level with the exception of Oligochaeta and Acarina Family level identification has been shown to be sufficient to allow consistent site classification in relation to disturbance (Marchant, 1 990) and, in the present study, identification to family level was deemed sufficient to determine broad scale impacts. After sorting, CPOM and FPOM were dried to constant weight at 600C, and ash-free dry weights obtained using standard methods (Allen, 1974).. Preliminary gut content analysis indicated that macroinvertebrates were feeding not just on algae, but on other epilithic material as welL Rather than estimate chlorophyll a, a more appropriate measure was total epilithic biomass. This was estimated by randomly selecting five rocks greater than 1 0 em in diameter from each reach in fast flowing sections; these were labelled and sealed in plastic bags and transported back to the laboratory on ice. Epilithon (i.e. algae, bacteria, fimgi and associated organic matter) was removed by scrubbing thoroughly with a wire brush, and the resulting material was then dried and ashed at 5500C to obtain ash free dry weight The surface area of each rock was measured using Shelly's (1 979) method.. Analysis. Univariate analyses of variance were carried out on the full designs for the total number of individuals, the total number of taxa, taxon diversity, evenness and functional feeding groups (shredders, collectors, filterers and scrapers) and the habitat variables measured for each sample unit (CPOM and FPO M). For physicochemical variables and epilithic biomass, ANOVA was also used but the design had one less stratum owing to the lack of differentiation between habitat types within the sites. After each ANOVA, residuals were checked for. 21.

(31) ·l 'fi!it.'1��. .{ -f�·- �. ,{J� t�­·.:_. � � i.:J i j ;/. -� } .t f, �(.. �h�rmality. $1. · , ��. ' "� �. ;r. , ;� HI. t�. �J tf �'}:.�. W, �. .. ;t �. were. of error terms and homogeneity of variances to ensure that assumptions of ANOVA. satisfied. If necessary the data were loge transformed and ANOVA models re-run and. assumptions re-checked. ANOVA tables for each analysis except those tables presented in this chapter are provided in Appendix A.. To determine whether there were any differences in community composition between the. willow and native reaches, Analysis of Similarity (ANOSIM Clark & Green, 1988) was. employed using 1000 randomisations on Bray-Curtis dissimilarity coefficient computed from log-transformed density data. The ANOSIM routine was that implemented in PATN (Belbin, 1993). Where there were significant differences between the vegetation types, the taxa that. differed most between the vegetation types were identified by computing the Kruskal-Wallis. statistic for each taxon using the GSTA sub-routine of PATN (Belbin, 1993 ). Those taxa with. the highest values of this statistic are those with the greatest difference in density between the :. vegetation types.. Because the rivers varied in size (Table 1) I investigated the possibility that the differences in. the densities of invertebrates between native and willowed reaches might be more pronounced in smaller rivers, where the impacts of willows (via shading and invasion of substrate by root. mats) were likely to be greater. I used general linear models to determine whether any. relationship existed between river size (as measured by catchment area and mean width of. downstream reach) with differences in the density of the total number of animals and for the 7. most abundant taxa between the reaches (i.e. density in willowed reach - density in native I. reach). Residuals were inspected, and transformations undertaken where necessary as for the other univariate ANOVAs.. Results. Habitat conditions Conductivity and pH were similar between willowed and native reaches in all seasons and no. significant differences were detected among the nine rivers studied. Conductivity covered a. 22.

Show more

New documents

In the First Attempt, Un- familiar partners who repeat each other to a significant degree summing across levels of linguistic representations, and thus align to their partner, have

Our results showed that a high concentration 100 μM of RAS inhibitors suppressed the proliferation of pancreatic cancer cell lines Panc-1, Bxpc- 3, and CFPAC-1 when used alone or in

4.4 Re-ranking models using trees Since the aim of concept annotation re-ranking is to exploit innovative and effective source of infor- mation, we can use, in addition to sequence ker-

We will deconstruct what appears to be the implicit ontology in Borooah et al.’s paper: that there is a community of Irish speakers whose children attend Gaelscoileanna Irish-language

All systems use development set accuracy to select their meta-parameters, such as the number of perceptron iterations, the size of the source- context window, and the n-gram length used

Linguistic Complexity and Argumentative Unity: A Lvov–Warsaw School Supplement.. Published in Katarzyna Budzynska, Frans

Therefore, the aim of the pr- esent study was to investigate whether HHD attenuated DOX-induced cardiotoxicity, in vitro H9c2 cells and in vivo in mice, and to exam- ine whether any

The second experi- ment shows that compact inference scales well to a full-blown RTE setting with several large-scale rule bases, where up to hundreds of rules are ap- plied for a text

For applications which require high gain and carrier densities, InGaNAs/GaAs quantum wells have a much lower linewidth enhancement factor over a temperature range 300–400 K than

Some examples are clustering co-click patterns to learn semantically related queries Beeferman and Berger, 2000, combining co-click patterns with hitting times Mei et al., 2008,