1 EFFECTS OF IRRIGATION CANALS ON STREAM ECOSYSTEMS IN A TROPICAL DRY FOREST REGION OF COSTA RICA By SUZANNE M. MOELLENDORF A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2009
2 2009 Suzanne M. Moellendorf
3 To my loving parents, Guy and Natalie, and my Costa Rican parents, Mayela and Silvano
4 ACKNOWLEDGMENTS I deeply thank m y advisor, Dr. Thomas Crisman, whose wisdom, mentoring, creative thinking, and strong vision made my graduate work possible and gr eatly aided my growth as a scientist. From his office in th e states to streambanks in Costa Rica, he offered tremendous time and energy into developing the ideas, cultivati ng the relationships, and providing the resources essential to the creation and ultimate su ccess of this research project. I thank Dr. Jorge Arturo Jimenez for suggesting a research project associated with the canals, helping develop my research questions, facilitating the unbelievabl e support I received in the field from the Organization of Tropical Studi es, and serving on my committee. I thank Dr. William Wise for his endless support and humor, a dvice with project development in Gainesville and Costa Rica, and helpful comments as a member of my committee. I thank Dr. Joann Mossa for her constant encouragement, fantastic insights, and participation on my committee. I am extremely thankful for the financia l support provided by: the National Science Foundation (NSF) Graduate Research Fellowshi p, International Research Experience for Graduate Students Award provided jointly by th e Organization for Tropical Studies (OTS) and NSF, Wetland Management and Restoration Aw ard provided jointly by OTS and US Fish &Wildlife Service, Howard T. Odum Graduate Fe llowship, University of Florida Department of Environmental Engineering Sciences Graduate Fellowship, personal research funds from Dr. Thomas Crisman, and the Storch Award from th e American Water Resources Association of Florida. I owe tremendous thanks to a wide array of pe ople in Florida, Costa Rica, and Oregon. I thank my hardworking assistants: Frances Las h, Greta Brom-Polkowski, Dan Auerbach, Farrel Ruiz Pacheco, Bernald Pacheco Chaves, and Adrian Valver de Gonzalez. I am very grateful to those who helped me with macroinvertebrate iden tification: Marcus Gr iswold, Pablo Gutierrez,
5 William Gerth, Richard Van Driesche, and Dr. Pa ul Hanson. I thank Monika Springer for facilitating my work at the University of Costa Rica, providing me with the opportunities to learn and teach, aiding with identifica tion, and helping make my time in San Jose so productive and enjoyable. To Dr. Eugenio Gonzalez, I send my gratitude for his amazing support that was critical to carrying out my research at Palo Ve rde successfully. His suppor t ranged from logistics and assistants to equipment and advice, and his drive, leadership, and positive attitude were inspiring. I thank Juan Serrano for his help w ith GIS map making and th e OTS Palo Verde staff for their kindness and support during my field seas ons. I heartily thank Dr. Sherri Johnson, Dr. Judy Li, Linda Ashkenas, Gail Achterman, and the Department of Fish & Wildlife for enabling me to eek every ounce of learning and enjoyment possible out of my Ph.D work through their support and facilitation of my res earch at Oregon State University. I also thank Dr. Alan Herlily for his support with statistical analyses. At the University of Florida, I thank Dr. Edward Phlips and his lab for their assistance with chlorophyll-a analysis and Dr. John Sansalone for use of his lab equipment. I thank Dr. Robert Knight for his sampling advice. I thank my friends and lab mates in Florida for opening up their hearts a nd homes to a vagabond. In Costa Rica, I thank Nora Pineda of SENARA for helping me obtain in formation for my study and Rita Vargas of the University of Costa Rica for help with shipping my samples. I also owe a special thanks to my Costa Rican parents, Mayela a nd Silvano Castro, for adopting me and giving me immeasurable support during my fieldwork in Costa Rica. Finally, I profoundly thank my extraordinary pa rents, Guy and Natalie, for all of their encouragement throughout my graduate studies. I would not be who I am and I would not have accomplished this work without their immense love and support.
6 TABLE OF CONTENTS page ACKNOWLEDGMENTS ............................................................................................................... 4 LIST OF TABLES ...........................................................................................................................9 LIST OF FIGURES .......................................................................................................................10 ABSTRACT ...................................................................................................................... .............12 CHAP TER 1 STUDY INTRODUCTION .................................................................................................... 14 Research Questions ............................................................................................................ .....14 Study Context .........................................................................................................................15 2 ECOHYDROLOGY STUDY: INTERMI TTENT AND PERENNIAL TROPICAL DRY FOREST STREAMS .....................................................................................................20 Introduction .................................................................................................................. ...........20 Background .................................................................................................................... .........24 Study Streams ..................................................................................................................24 Geology & Soils .............................................................................................................. 24 Rainfall ...................................................................................................................... ......25 Vegetation and Forest Phenology ....................................................................................25 Canal Oeste (Phase III) .................................................................................................... 26 Methods ..................................................................................................................................27 Sampling Strategy ...........................................................................................................27 Field Sampling ................................................................................................................ .28 Physical-chemical ..................................................................................................... 28 Biological ................................................................................................................. 29 Laboratory Methods ........................................................................................................ 29 Data Analysis ...................................................................................................................30 Results & Discussion ..............................................................................................................31 Hydrology ..................................................................................................................... ...31 Dry months ............................................................................................................... 31 Wet months .............................................................................................................. 31 Physical-Chemical Parameters ........................................................................................ 33 Dry months ............................................................................................................... 33 Wet months .............................................................................................................. 34 Canopy Cover ..................................................................................................................35 Aquatic Macroinvertebrates Results: Dry Months ......................................................... 36 Aquatic Macroinvertebrates Results: W et Months ......................................................... 39
7 Benthic macroinvertebrates ......................................................................................39 Drift macroinvertebrates .......................................................................................... 44 Structure of benthic versus drift communities ......................................................... 45 Aquatic Macroinvertebrates Discussion ..........................................................................46 Colonization ............................................................................................................. 47 Development times ................................................................................................... 51 Succession ................................................................................................................ 53 Functional feeding groups ........................................................................................54 Taxa richness and rankings ......................................................................................55 Fish Results & Discussion ...............................................................................................56 Dry months ............................................................................................................... 56 Wet months .............................................................................................................. 58 Conclusion .................................................................................................................... ..........63 3 RIPARIAN PATCH STUDY ................................................................................................. 92 Introduction .................................................................................................................. ...........92 Methods ..................................................................................................................................94 Study Area .......................................................................................................................94 Site Selection ...................................................................................................................95 Sampling Schedule ..........................................................................................................96 Field Sampling ................................................................................................................ .97 Physical-chemical ..................................................................................................... 97 Biological ................................................................................................................. 97 Laboratory Methods ........................................................................................................ 99 Data Analysis ...................................................................................................................99 Results ...................................................................................................................................100 Physical-Chemical Parameters ...................................................................................... 100 Biotic Parameters ...........................................................................................................106 Benthic macroinvertebrates ....................................................................................106 Stone macroinvertebrates ....................................................................................... 108 Drift macroinvertebrates ........................................................................................ 110 Taxa exclusivity .....................................................................................................113 Fish ......................................................................................................................... 113 Discussion .................................................................................................................... .........114 Riparian Canopy, Stream Ligh ting, and Prim ary Production ........................................ 115 Riparian Canopy, Physical-Chemical Condition s, and Streamflow .............................. 116 Aquatic Macroinvertebrates ..........................................................................................117 Benthic macroinvertebrates ....................................................................................118 Stone macroinvertebrates ....................................................................................... 123 Drift macroinvertebrates ........................................................................................ 125 Overall .................................................................................................................... 128 Fish .......................................................................................................................... ......131 Conclusion .................................................................................................................... ........135
8 4 CANAL CONTINUUM STUDY ......................................................................................... 164 Introduction .................................................................................................................. .........164 Methods ................................................................................................................................167 Study Area .....................................................................................................................167 Sampling Strategy .........................................................................................................168 Field Sampling ............................................................................................................... 168 Physical-chemical ................................................................................................... 168 Biological ............................................................................................................... 169 Laboratory Methods ...................................................................................................... 170 Data Analysis .................................................................................................................170 Results ...................................................................................................................................171 Physical-Chemical Parameters ...................................................................................... 171 Biotic Parameters ...........................................................................................................173 Algal biomass ......................................................................................................... 173 Aquatic macroinvertebrates ....................................................................................173 Discussion .................................................................................................................... .........176 Conclusion .................................................................................................................... ........185 5 STUDY CONCLUSION ......................................................................................................196 Ecohydrology .................................................................................................................. ......196 Riparian Patch Study ............................................................................................................197 Canal Continuum Study ........................................................................................................ 199 Holistic Perspective ..............................................................................................................200 LIST OF REFERENCES .............................................................................................................202 BIOGRAPHICAL SKETCH .......................................................................................................216
9 LIST OF TABLES Table page 2-1 Indicator species analysis by stream type over time. ........................................................ 83 2-2 Eight most abundant taxa in the benthos and drift, by stream type (16 taxa total per stream type), categorized by habitat occupied. .................................................................. 88 2-3 Taxon habitat exclusivity by stream type. ......................................................................... 89 2-4 Taxon exclusivity, by stream type and habitat. .................................................................. 90 3-1 Taxa richness, by zone: Be nthos. CI=Canal-Influenced. ............................................... 154 3-2 Taxa richness, by zone: St ones. CI=Canal-Influenced. ................................................. 155 3-3 Taxa richness, by zone: Dr ift. CI=Canal-Influenced. .................................................... 156 3-4 Top twelve most abundant taxa, by zone: Benthos. ........................................................157 3-5 Top twelve most abundant taxa, by zone: Stones. .......................................................... 158 3-6 Top twelve most abundant taxa, by zone: Drift. ............................................................. 159 3-7 Taxa exclusivity, by habitat. ............................................................................................160
10 LIST OF FIGURES Figure page 1-1 Study sites. .........................................................................................................................192-1 Ecohydrology Study site maps........................................................................................... 642-2 Precipitation data from the Palo Verde Meterological Station (OTS). .............................. 652-3 Discharge. ................................................................................................................ ..........662-4 Physical-chemical average values, by str eam type for periods of streamflow (MayJune, September, and October) and pool s (Early May, July, Late-Nov, Jan). ................... 672-5 Isolated pool benthic macroinvertebrates having 5% or greater relative abundances. ...... 692-6 Benthic & drift macroinvertebrate orders having 5% or greater relative abundances, by dry & wet periods. .........................................................................................................712-7 Benthic macroinvertebrate total abundances. .................................................................... 732-8 Benthic macroinvertebrates having 5% or gr eater relative abundance, by stream type. ... 752-9 Benthic macroinvertebrate functional f eeding groups (FFG) having 5% or greater relative abundances, over time. .......................................................................................... 772-10 Relativized ordinations, all streamflow periods combined. ............................................... 782-11 Relativized Ordinati ons, by sampling month.. ................................................................... 802-12 Drift macroinvertebrates, by stream type. .......................................................................... 842-13 Drift macroinvertebrate rela tive abundance, over time. .................................................... 862-14 Fish relative abundance by stream type. ............................................................................ 913-1 Riparian Patch Study site map. ........................................................................................1373-2 Photos of Open Zone. ......................................................................................................1383-3 Aerial view of sampling site showin g sampling lengths and site dimensions. ................ 1393-4 Stream light and prim ary production averages. ............................................................... 1403-5 Physical-chemical characteristics over time, by zone. ..................................................... 1423-6 Total macroinvertebrate abundance over time. ................................................................ 144
11 3-7 Macroinvertebrate relative a bundance, entire study, by zone. .........................................1453-8 Benthic macroinvertebrate rela tive abundance, by sampling round. ............................... 1463-9 Stone macroinvertebrate rela tive abundance, by sampling round. ..................................1483-10 Functional feeding group relative abundance, by zone. .................................................. 1503-11 Drift macroinvertebrate relative abundance, entire study, by zone. ................................1513-12 Drift relative abundance, by order. .................................................................................. 1523-13 Relatived ordinations, by sampling round. ...................................................................... 1613-14 Fish abundance............................................................................................................ .....1634-1 River Continuum Concept versus Canal Continuum. ...................................................... 1884-2 Canal Continuum Study site map. ................................................................................... 1894-3 Canal dimensions, by canal size. ..................................................................................... 1904-4 Physical-chemical paramete r averages, by canal type. .................................................... 1914-5 Macroinvertebrates commun ity structure by canal size. .................................................. 1924-6 Macroinvertebrate functional feeding groups, by canal type. .......................................... 1944-7 Relativized Ordina tion, by canal type.. ............................................................................ 195
12 Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy EFFECTS OF IRRIGATION CANALS ON STREAM ECOSYSTEMS IN A TROPICAL DRY FOREST REGION OF COSTA RICA By Suzanne M. Moellendorf August 2009 Chair: Thomas L. Crisman Major: Environmental Engineering Sciences The interaction between irriga tion and stream systems near Palo Verde National Park, Costa Rica provided an opportunity to investig ate critical concepts in stream ecology: ecohydrology, riparian patch ecol ogy, and the river continuum c oncept, while addressing water management issues. This study integrated thes e components by exploring: 1) Effects of canals on stream community structure and function, and 2) Whether canals and streams have similar structure and function at the landscape level. The ecohydrology investigation compared interm ittent and perennial tropical dry forest streams. It provided details on their aquatic co mmunities and demonstrated the strong influence of streamflow on the colonization, development, and succession of aquatic biota. Furthermore, this investigation increased knowle dge of tropical dry forest stream ecosystems, which have been studied very little and exist within a threatened forest ecosystem. It also revealed that impacts of irrigation canals on tropical dry forest streams included increased habitat during the dry season, habitat fragmentation, channel scou ring, and water quality changes. The riparian patch investigation compared f our distinct stream e nvironments resulting from canal management at a site where an ir rigation canal crossing is maintained without riparian vegetation and canal water is directly di scharged into the stream. Some differences in
13 physical-chemical characteristics and the biotic community were found between the deforested stretch and the adjacent upstream and downstrea m forested stretches, but that small-scale riparian deforestation (< 35 m) may not have severe detrimental effects on the stream community and may actually provide a greater diversity of habitats and resources that enhances biotic richness. In addition, perennial discharge from the canal into th e stream may provide refugia in the dry season and a source of coloni zers for the seasonal stream reach. The canal continuum investiga tion analyzed the size classes of irrigation canals that diminish in size over their longitudinal gradient both to determine their structure and function and to examine whether canals follow prediction s of the River Continuum Concept (RCC), but in reverse order. As a whole, the canals did support an aquatic community whose structure and function did change over their longitudinal gradient, and the canals did follow some RCC predictions.
14 CHAPTER 1 STUDY INTRODUCTION Research Questions Understanding the structure and f unction of stream system s is critical for determining how water management strategies may be affecting fr eshwater resources. In regions where water is scarce, yet agricultural land available, wate r management often includes construction of irrigation systems. Introduction of a large quantity of additional water inevitably changes local hydrology, having implications on stream ecosystem s in the area. Streams near Palo Verde National Park, Costa Rica demonstrated this inte raction between irrigation and stream systems, providing an opportunity to inve stigate critical concepts in stream ecology: ecohydrology, riparian patch ecology, and the river continuum concept, while addressing water management issues. The objective of this en tire study was to investigate these concepts and local watershed management by addressing: 1) Effects of can als on stream community structure and function, and 2) Whether canals and streams have similar structure and function at the landscape level. To that end, examination of freshwater ecosystems altered or created by irrigation canals in a tropical dry forest region of Costa Rica occurred in three parts. The first part addressed whether or not canals affect stream co mmunity structure and function by investigating the effects of canal-induced changes in flow regime of tropica l dry forest streams, thereby exploring the concept of ecohydrology. The second part investigat ed the effects of riparian deforestation on a second-order tropical dry forest stream due to canal management practices, thereby exploring the concept of patch dynamics. The third part addr essed whether canals and streams have similar structure and function, explori ng the predictions and applicat ions of the River Continuum Concept.
15 Study Context Stream s near Palo Verde National Park are lo cated in a tropical dry forest. Tropical dry forests experience dry periods lasting about 6 months and are semi-deciduous, containing trees usually between 20-30 m tall, shrubs that often have thorns or spines, woody vines, and ground bromeliads (Holdridge 1967). This forest type originally compri sed 42% of the worlds tropical vegetation (Murphy and Lugo 1986), but is now consid ered one of the most threatened lowland forest ecosystems in the tropics (Janzen 1988). In the case of Central America, these forests once covered an area five times the size of Guatemala, but by the mid-1980s, there were no longer large areas of relatively pr istine dry forest, and less than 1% of the original forest had conservation status (Janzen 1986). Within Costa Rica, degradation, fragmentation, and elimination of tropical dry forest habitat have been largely attri buted to pasture and agricultural conversion (Janzen 1988). The only remaining tropical dry forest in Costa Rica is in the Guanacaste Province in the northwestern part of the country, specifically within the Tempisque River Basin (Jimenez et al. 2001). Palo Verde National Park and Lomas Ba rbudal National Park, located in the basin, remain among of the few protected areas with tropical dry forests in Costa Rica and the Neotropics in general (OTS 2009). Existence of these forests results from dis tinct seasonal patterns of precipitation in the region. Ninety-five percent of precipitation occurs between Ma y and November, with a short period of reduced rainfall in June, July, and A ugust called El Veranillo de San Juan (Jimenez et al. 2001). The marked seasona lity often results in drought dur ing the dry season and floods in the rainy season. While the preponde rance of sunny days encourages crop growth, lack of water in the dry season is a hindrance to growth (Jimenez et al. 2001).
16 Both to reduce vulnerability of agriculture to droughts and to increase the number of possible harvests annually, the Arenal-Tempisque Irrigation Project (PRA T) was initiated in 1980. This large irrigation canal project conveys water from Lake Arenal in the highlands to agricultural lowlands of Guanacaste (Pacific 2002). PRAT helps support the regions agricultural industry, whic h produces approximately 45% of the rice, 50% of the sugar cane, and 100% of the melon in Costa Rica (Jimenez et al. 2001). One of the primary canals of PRAT is Canal Oeste, which a co-operative of companies, COPEVICA, recently paid to extend approximately 21 km to irrigate their fields. This extension was completed in 2003 (Pineda 2007). Along a por tion of this extension, Canal Oeste crosses a series of parallel streams coming off foothills covere d in tropical dry forest and affected some of the streams in the process. Lomas Barbudal Rese rve covers part of these foothills and Canal Oeste crosses a wildlife corridor connecting the reserve to Palo Verde National Park. This corridor contains a second-order stream, Quebra da Barbudal, which Cana l Oeste also affected. Canal construction practices varied along the extension length. The initial section has concrete lining and culverts, while the remainder of the canal has walls composed of rock and earth that are nearly devoid of concrete lining an d has no culverts. As a result of the culverts and lining in the initial section, seve ral streams that the canal crosse s are able to maintain their intermittent flow regime, which is characteristic of the region due to the distinct wet and dry seasons (Fig. 1-1). Meanwhile, the canal leaks at some locations, primarily in the unlined segment, providing perennial flow to other stream s nearby, thus altering the typically intermittent hydrologic regime of the streams. Canal construction also involved the use of an inverted siphon to allow the canal to pass unde r a second-order seasonal stream called Quebrada Barbudal. At this location of canal crossing, th e riparian zone of Quebrada Ba rbudal is kept deforested, which
17 creates an open patch with high light exposure al ong an otherwise forested stream (Fig. 1-1). These alterations in streamflow and riparian can opy cover would be expect ed to affect stream community structure by changing food web comple xity, colonization and succession patterns, and nutrient cycling. Some water from the Canal Oeste extension is diverted into a comparatively smaller irrigation canal system that provides water to rice and sugar cane fields between Palo Verde National Park and the Lomas Barbuda l Reserve (Fig. 1-1). This canal system creates freshwater habitat that harbors fish, cr ayfish, and aquatic macroinvert ebrates (Rizo-Patrn 2003) and it decreases in channel width and depth over its longitudinal gradient, the reverse size sequence of a typical stream. Consequently, these canals w ould be expected to follow stream community changes related to stream size th at were predicted in the Rive r Continuum Concept (Vannote et al. 1980), but in reverse order. Research into the structure and function of tropical streams in Costa Rica has largely occurred in tropical wet fore sts (Pringle and Ramirez 1998, Ra mirez and Pringle 1998b, Boyero and Bosch 2004). Previous research on Costa Rican dry forest streams is sparse, but has provided information on aquatic macroinvertebrate taxonom y (Pritchard 1996) and development (Jackson and Sweeney 1995), population dynamics of the fish Poecilia gillii (Kner & Steindachner 1863) (Chapman et al. 1991), reproduction of the fish Archocentrus nigrofasciatus (Gunter 1869) (Wisenden 1994), and water level fluctuations in seasonal pools (Chapman and Kramer 1991b). Due to the threatened status of tropical dr y forests and the lack of understanding of how irrigation projects may affect them, an investiga tion of natural and alte red tropical dry forest streams is especially pressing. In addition, st rongly seasonal rainfall patterns and irrigation canals passing through the region combine to of fer unique opportunities to investigate the
18 dynamics of ecohydrology and riparian patch ecology, and the presence of irrigation canals offers an opportunity to test predictions of the River Continuum Concept within an irrigation canal system.
19 Figure 1-1. Study sites. Upper right-ha nd box indicates the Ecohydrology Study sampling points in the lined canal section with culverts. Upper left-hand box indicates Ecohydrology Study sampling sites in the unlin ed canal section without culverts and the Riparian Patch Study sampling sites at the crossing of Quebrada Barbudal and an inverted siphon. Center box indicates Canal Continuum Study sampling points.
20 CHAPTER 2 ECOHYDROLOGY STUDY: INTERMITTEN T AND PERENNIAL TROPICAL DRY FOREST STREAMS Introduction Most scientific literature on stream ecology has focused on temperate systems (Allan 1995, Merritt and Cummins 1996, Thorp and Covich 2001), but tropical sy stems have begun to receive more attention recently (Dudgeon 2008), including Ma dagascar (Benstead et al. 2003), Southeast Asia (Dudgeon 2006), Puerto Rico (Cross et al. 2008), Brazil (Callisto a nd Goulart 2005), and Costa Rica. Within Costa Rica, most stream research has been in wet forests (Pringle and Hamazaki 1997, Ramirez and Pringle 1998a, Boyero and Bosch 2004) while few studies have occurred in dry forests, such as those in the Guanacaste region. Tropical dry forests historically comprised 42% of global tropical vegetation (Murphy and Lugo 1986), but they are currently one of the most threatened lowland fore st ecosystems in the tropics (Janzen 1988) and little is known about their streams. Tropical dry forests are characterized by dry periods lasting about 6 mont hs and semi-deciduous forests that contain trees usually between 20-30 m tall, shrubs that often have thorns or spines, woody vines, and ground bromeliads (Holdridge 1967). Dr y forests in Central America on ce covered an area five times larger than Guatemala. However, by the mid-1980s vast areas of relatively pristine tropical dry forest in the region had disappeared and less th an one percent of the original forest had conservation status (Janzen 1986). Degradation, fragmentation, and elimination of dry forests in Costa Rica have been largely attributed to th eir conversion to cropla nds and pasture (Janzen 1988). The remaining dry forest in Costa Rica is mostly in the Tempisque River Basin in the northwestern province of Guanacaste (Jimenez et al. 2001). Palo Verde and Lomas Barbudal National Parks, located within the basin, are among the few protec ted areas for these forests in
21 Costa Rica and the neotropics in general (OTS 2009) By default, streams of tropical dry forests have become scarce and ra rely have protection. Low-order, tropical dry forest streams typically display intermittent flow due to distinct wet and dry seasons. Ninety-five percent of preci pitation occurs between May and November in Guanacaste, with a short period of reduced rainfall from late June to mid-August, called El Veranillo de San Juan (Jimenez et al. 2001). The marked seasonality of rainfall results in loworder streams flowing only during the wet season, intermittently or continuously depending on precipitation patterns, and desiccating duri ng the dry season. Intermittent streams are characterized by higher vari ability in flow than perennial st reams, periods of no surface flow, and no low flow except following high flow events resulting from rainfall. In contrast, perennial streams typically have continuous low flows deri ved from watershed storages and experience periodic high flows of varying magnitude and duration due to rainfall events (Hughes 2005). Flow variability (spates or drying events) is a characteristic disturbance in streams within seasonally dry or semi-arid regions, including Guanacaste, the Mediterranean basin (Acuna et al. 2005), Australia (Closs and Lake 1994) and prairies of the United States (Fritz and Dodds 2004) While a notable body of scientific literature exists for streams in semi-arid regions, such as in Brazil (Nolte et al. 1997, Medeiros and Maltchik 2001, Maltchik a nd Medeiros 2006) and Australia (Boulton 2003, Hose et al. 2005, Robson et al. 2008), literature specifically addressing tropical dry forest streams in Central America is sparse and comes primarily from Costa Rica. However, the limited information that does exist from these Costa Rican dry forest streams has begun to illuminate some effects of the hydrol ogic cycle on aquatic ecosystems. Low-order streams in Costa Rica flow intermittently, leaving pools lasting into the dry season that fluctuate in size and physical chemical charac teristics (Chapman and Kramer 1991b). Poecilia gillii (Kner
22 & Steindachner, 1863), a common live-bearing poeci lid, suffers major losses after spates, but those that survive find refuge in residual, isolated pools upon drying despite low dissolved oxygen (Chapman and Kramer 1991a). Genera l life history and geographic distribution information on the fishes in Guanacaste have also been compiled (Bussing 2002). The few existing aquatic macroinvertebrate studies have revealed rapid re productive cycles and lack of diapause (Jackson and Sweeney 1995) and have provided taxonomic information (Pritchard 1996, Avila and Flowers 2005). Reduced water availability in the dry season in Guanacaste hinders agricultural productivity in a region with a preponderance of sunny days and fertile soil that are ideal for crops (Jimenez et al. 2001). To increase its agricultural productivity and decrease its vulnerability to drought, the Costa Rican governme nt initiated the Arenal -Tempisque Irrigation Project (PRAT) in 1980. This la rge-scale irrigation canal proj ect conveys water from Lake Arenal in the central highlands westward to th e agricultural lowlands of Guanacaste (Pacific 2002), which has permitted conversion of lowland, dry forests and wetlands to rice and sugar cane agriculture. By 2001, PRAT supported produc tion of approximately half of the rice and sugar cane grown in Costa Rica (Jimenez et al. 2001). Beyond agricultural production, PRAT has influenced stream hydrology in tracts of dry forests through which it passes. One of the pr imary canals of PRAT, Canal Oeste, crosses a series of parallel headwater streams in a dry fo rest adjacent to Palo Verde and Lomas Barbudal National Parks. Canal construc tion occurred in two phases that utilized different construction practices. The initial segment is concrete lined with culverts to convey streams under the canal while the later segment generally lacks concrete lining, instead having canal walls composed of rock and earth, and lacks culverts (Pineda 2007). Several streams that the canal crosses have
23 maintained their characteristic intermittent flows in the lined section of the canal. However, the canal leaks at several locations, especially in the unlined sectio n, providing perennial flow to typically intermittent streams. Extremes in stream hydrology, like spates and dr oughts, cause disturbances that affect the structure and function of biotic communitie s (Lake 2003), especially the composition and function at the community level, population densities, and biom ass and growth of taxa and individuals (Biggs et al. 2005). Biota respond to these dist urbances through resistance or resilience mechanisms (Lake 2000). Furthermore, drying reduces av ailable habitat, causes fine sediments to settle out, stops export of detritus and nut rients, and traps aquatic biota (Lake 2000). Isolated pools can serve as crit ical refugia for mainta ining populations, with survival dependent on size of the refuge, rate of drying, and phys ical and chemical conditions (Humphries and Baldwin 2003). Canal Oeste not only represents regional water management in Costa Rica, it also more broadly represents the general expansion of irrigation schemes throughout semi-arid regions of the world (Hooke 2006). The inter-basin water transf er from highlands to lowlands displayed by Canal Oeste is an increasingly common practi ce of these irrigation plans (Hughes 2005), which can alter community structure and function (Poff et al. 1997), such as through a shift from intermittent to perennial hydrology (Uys and O'Kee ffe 1997). The purpose of this study is to address the knowledge gap in tropical dry forest stream ecology and to increase understanding of the effects of canal management practices by inve stigating: 1) the ecohydrology of intermittent and perennial tropical dry forest streams, and 2) the impacts of canal s on tropical dry forest stream structure and function. This involved comparing abiotic and biotic components of
24 naturally intermittent streams (control) to can al-influenced streams with perennial flow (experimental) throughout the wet season and the seasonal transition periods. Background Study Streams Study stream s were within a tropical dry forest near Palo Verde National Park, Costa Rica along a range of foothills partially within th e Lomas Barbudal Reserve and Palo Verde-Lomas Barbudal Reserve Wildlife Corridor and partially on private land (Fig. 2-1). A series of parallel, first-order streams flow from thes e foothills and are concentrated in two locations. Streams at the eastern location inter act with a lined section of the main irrigation can al of the region, Canal Oeste (Fig. 2-1A), while those at the western location interact with the predominantly unlined canal section (Fig. 2-1B). A canal length of approximately 13 km separates the Rio Piedras siphon at the eastern sampling lo cation and Quebrada Barbudal siphon at the western sampling location. Land use in the unlined canal section is exclusively fore st and elevation ranges from approximately 15 m to 30 m above mean sea leve l (msl) along the stream lengths sampled. Land use in the lined canal section is forest for pere nnial and intermittent streams sections below the canal and a combination of forest and pasture for stream sections above the canal. Elevation in the lined canal section ranges from approxima tely 20 m to 40 m above msl along the stream lengths sampled. Water from the canal serves agricultural fiel ds in a lowland plain within several hundred meters of the lined and unlined sections. Ri ce and sugar cane are the dominant crops, with acreage under cultivation still expanding to a small degree. Geology & Soils The foothills are of volcanic origin (Qv) and comprised of calcareous and siliceous igneous rock from pyroclastic flows. The adjacent lowla nd plain has quarternary alluvial deposits (Qal),
25 including gravel, sand, lime, quartz, and clay of volcanic origin. Stream sections above the canal have the former geology, while those below disp lay a mixture of upland and lowland geology. Soils in the lined canal section are a mixture of mollisols and entisols, while those in the unlined canal section are entisols and vertisols. Rainfall This study o ccurred in the wet season and its transition periods (May 2007 to January 2008). Total rainfall in 2007 was 1195.1 mm, which was slightly above the 984.6 mm average for 2000-2006, excluding 2005, which had incomplete data. In 2007, August and October had the highest total rainfall, 265 mm and 262.8 mm, respectively (Fig. 2-2A). June and September also had relatively high rainfall, with 206.9 mm and 180.4 mm, respectively. Average monthly rainfall (2000-2006) was greates t in September (244.8 mm), Oc tober (206.8 mm), June (152.8 mm), and May (127.1 mm). Peaks in daily rain fall (30 cm or greater in one day) in 2007 occurred on 10 April (52 mm), 13 June (31.8 mm) 22 June (38.2 mm), 5 August (90.4 mm), 16 September (32.8 mm), and 11 October (30.4 mm) (Fig. 2-2B). October had five days with more than 20 mm of rainfall, the most of any mont h, while April and late May had two, June and August had three, and July and September had one. Vegetation and Forest Phenology The foothills are covered in savanna, deciduous fo rests, and riparian forests, with cacti in extrem ely dry sites (OTS 2009). Tr ees greater than 3 m height in tropical dry forests leaf out at the beginning of the wet season (M ay-June) (Frankie et al. 1974). Leaf fall begins in October but is most pronounced between January and April, peaking in March. Approximately 75% of tree species partially or completely drop their leav es, and many riparian tr ee species drop all their leaves at one time during leaf fall, but new leaves replace them immediately. Immature trees and shrubs on hills leaf out at the beginning of th e wet season, while riparian communities wait until
26 the end of the wet season, though leafing out in secondary forest cont inues throughout the wet season. Most immature trees and shrubs are bare and dormant during the dry season, with greatest leaf loss occu rring in February (O pler et al. 1980). Canal Oeste (Phase III) The Arenal-Tem pisque Irrigation Project (PRAT) is a large irrigation canal system that conveys water from Lake Arenal in the central highlands to agricultural fields in the lowlands of Guanacaste. Several agricultural companies and the Servicio Nacional de Aguas Subterrneas, Riego y Avenamiento (SENARA) collaborated to ex tend one of its principl e canals, Canal Oeste, to serve lands adjacent to Palo Verde National Park and the Lo mas Barbudal Reserve. The 21 km extension between Rio Piedras and Rio Cabuyo, Canal Oeste Phase 3 (COP3), began in November 2000 and was completed in August 2003. Within this extension, approximately 1 km is lined and 20 km is unlined. This canal has a 7 m wide base and is 4.2 m high with an extra 0.2 m of height to compensate for extreme water fl uctuations. Water depth averages about 3.62 m, and the canal drops 1.5 m in elevation per 10 km can al length. Concrete tiles used in the lined portion are 1 x 1 m. Engineers did not use concrete lining along most of COP3 because of sufficiently small flows and stable earth. Culverts or tubes, were installed at some significant creek crossings and inverted siphons at river or large str eam crossings (Pineda 2007). Canal Oeste Phase 3 can convey water up to 15 m^3/s. Discharge typically ranges between 10-15 m^3/s in the dry and 5-7 m^3/s in the wet season when demand for irrigation drops (Pineda 2007), but it can drop as low as 3 m^3/s (Santana 2007 Personal Communication). Water transit time from Lake Aren al to COP3 can be as fast as 1.5 days (Santana 2007 Personal Communication), and it allows irrigation of nearly 12,600 hectares representing 382 users (Pineda 2007).
27 Methods Sampling Strategy This stream study focused on the 2007 wet s eason and was framed by single sampling events both at the end of th e preceding dry season (Early May 2007) just before heavy, consistent rainfall began and in the early mont hs of the following dry season (Late November 2007 and January 2008). This strategy was meant to characterize baseline conditions of the aquatic communities of intermittent streams during dry periods, against which the structure and function of both intermittent and perennial stream s during wet periods could be compared. The annual short dry period El Veranillo de San Juan the Little Summer (J uly), provided further insight into the responses of both st ream types to rainfall reduction. First-order, headwater streams were sampled and characterized as intermittent or perennial based on presence or absence of flow at the end of the dry season in early May. Intermittent streams were classified as control streams and pe rennial streams as experimental, since the latter are atypical of headwater streams in the region. Control streams were concentrated along the lined canal section, and experimental streams were generally associated with the unlined canal section. Samples were taken from streams both above and below the lined canal section with culverts and taken only below the canal in the unlined canal section with no culverts since streams above the canal did not have any reco rded flow during the 2006 wet season pilot study. Data were collected at three points 50 m apart along a 100 m stream secti on with a beginning or end point within meters of the canal. These st ream lengths downstream of the canal measured several hundred meters, except for on e stream in the lined canal se ction that was less than 100 m long, such that sampling points were 10 m apart al ong a 20 m section above and below the canal.
28 Sampling occurred from May 2007 to January 2008, covering the entire wet season, including transitions into and out of it. Four to six weeks passed between sampling events, with one eight week stretch between the July and September samplings. Field Sampling Physical-chemical Bankfull width and height, channel width a nd height, incline, and bed texture were record ed once for each stream at the end of th e dry season. Stream gradient was determined using a laser level. Canopy cover was determin ed using a convex spherical densiometer and was measured at each point during early May, MayJune, October, November, and January. Canopy cover data were collected 60 times in perennial streams and 45 times each in intermittent stream sections above and below the canal. Physical-chemical and streamflow data we re collected each sa mpling period in the daytime. Dissolved oxygen (% and mg/L), salinity, to tal dissolved solids (g/L), temperature (C), and conductivity (mS/cm-adjusted for 25 C) were measured using a YSI 556 Handheld Multiparameter Instrument (YSI, Inc. Yellow Springs, OH, U.S.A.). Stream velocity was measured using a Marsh-McBirney flowmeter at three equidistant points across the width of the stream channel. In stream depths less than 20 cm, velocity was measured at 60% of the depth from the stream surface. When depth was 20 cm or greater, velocity was measured at 20% and 80% of the depth from the stream surface and then averaged. Discharge was calculated using the velocity, water depth, and channel width measurem ents. The total number of samples taken was 36 in perennial streams, 27 in intermittent str eams below the canal, 23 in intermittent streams above the canal during periods of flow (May-J une, September, and October), and 12 from residual pools during drier periods Rainfall data were based on a rain gauge at Palo Verde National Park operated by the Organization for Tropical Studies.
29 Biological Macroinv ertebrate samples were collected during each sampling round from streams with flow or residual pools. Benthic macroinvert ebrates were collected using a 243 um Surber sampler. Sediment, small rocks, and debris within the 0.093 m^2 area of the sampler were disturbed by hand, and organic debr is and macroinvertebrates were collected in the attached net. Macroinvertebrates in drift were collected at only one point per stream using a 363 um drift net that was placed at a single location in the stream for 20 hours, spanning dusk, night, and dawn. A 20-hour sampling period was chosen due to lo gistical constraints in the field. All macroinvertebrate samples were pres erved with ethanol in the field. Fish were sampled at each sampling point us ing bread-baited minnow traps, which were placed in the deepest part of the stream and le ft for 20 hours, spanning dusk, night, and dawn. A 20-hour sampling period was chosen due to logistical constraints in the field. Captured fish were measured for standard and tota l length, identified, and released back into the stream. The total number of samples taken during periods of flow (May-June, September, and October) for benthos was 35 in perennial streams, 27 in intermittent streams below the canal, and 20 in intermittent streams above the canal. The number of samples taken from drift was 12 in perennial streams, 11 in intermittent streams belo w the canal, and 9 in intermittent streams above the canal. The number of fish samples taken was 24 in perennial streams and 18 in intermittent streams below the canal. No fish were observed in stream secti ons above the canal. For residual pools, 15 samples were collected for bent hic macroinvertebrates and 6 for fish. Laboratory Methods All aquatic m acroinvertebrate samples were passed through 1 mm and 250 um sieves in the laboratory. Both size classes were examined under a dissecting microscope, and macroinvertebrates were removed and placed in to dated and labeled vials containing 70%
30 ethanol. Samples were then shipped to Florid a for identification. Macroinvertebrates were generally identified to Family or Genus, ex cept the families Chironomidae and Ceratopogonidae, which were identified to Sub-Family. Non-in sect macroinvertebrates (Oligochaeta, Nematoda, Hydracarina, and Hirudinea) were identified to Order. Iden tification of tropical aquatic macroinvertebrates to species is difficult due to a paucity of taxonomic tropical literature (Jacobsen et al. 2008). Identification utilized Merritt & Cummins (1996), assistance from M. Springer and P. Hanson at the University of Costa Rica, and Roldan (1996). For mass of leaves in drift, all organic matter (leaves, sticks) retained in the 1 mm sieve in the drift samples was checked for macroinvertebra tes, and then stored in 70% ethanol in a separate plastic bag for drying, then weighing. E ach sample of drift organic matter was placed on a labeled aluminum foil tray. Samples were th en put in a rudimentary dryer consisting of a box with open ventilation around the base about 3 cm high, a cardboard lid, and a series of light bulbs installed underneath a metal rack. These samples were left to dry for 24-48 hours to evaporate the ethanol. Next, samples were put into a drying oven for approximately 17 hours at 103 + 0.2C, and then weighed to 0.01 grams. Data Analysis To test for differences in variables betw een stream types and over tim e, Repeated Measures ANOVAS using PROC MIXED were run in the program SAS (SAS Institute, Inc 2006). For correlations between different variab les, Pearsons Correlations using PROC CORR were tested in the program SAS. Variables we re log-normalized before running statistics when appropriate. When differences we re significant, post-hoc analysis was conducted using Tukeys test and Bonferroni corrections. Taxa richness was used rather than species richness since not all identifications were made to the same taxonom ic level. Consequently, taxa richness was calculated by summing the number of different taxa found to the level of identification chosen in
31 this study. To investigate relations hips of abiotic and biotic str eam components with stream type and time, non-metric multidimensional scaling ordinations using Sorensons metrics (i.e. BrayCurtis) and Monte Carlo tests we re created in PC-Ord (McCune and Grace 2002). Taxa data were logged and relativized. Biplots with a cutoff of P=0.2 were used to define associations between the sample and taxa distributions. Indi cator species analysis using Monte Carlo tests was conducted in PC-Ord. Results & Discussion Hydrology Dry months Interm ittent (T) streams had no flow in May, July, November, and January, months of the dry season or El Veranillo de San Juan, except fo r one stream that had very low but measurable flow in November. This stream lacked flow above the canal, and slow leakage was evident below the canal from the culvert, which suggested that leaks in the lined canal controlled this discharge rather than rainfall. However, discharge from leaks had ceased by the January sampling, and the stream was completely dry. Although, no streamflow existed as a whole in T streams in dry months, some isolated pools persiste d. All four perennial (P ) streams had flow at the end of the dry season in early May 2007. In November, the start of the next dry season, one P stream in the unlined canal section was dry. In January, that P str eam and another in the unlined section were dry, while the remaining P streams maintained flow. Wet months Flow occurred in both T and P stream s in three sampling periods: late May-early June, September, and October, which coincided with periods of heavy rainfa ll during the wet season (Fig. 2-3A).
32 Discharge was particularly high for one T stream in May-June due to a heavy rainfall event that differentially increased flow due to its slightly larger watershed. This outlier was responsible for T and P streams displaying signi ficantly different flow s in May-June (t=-2.34, p=0.31). When it was removed, discharge averages between stream types were similar (Fig. 24D). No significant difference in discharge was found between T streams sections above and below the canal. All four P streams had flow during all sampling periods from the end of the dry season in early May to the end of the wet season in Oct ober. However, the one P stream maintained by pronounced seepage from the lined canal nearly lo st flow in September despite relatively heavy rains that month. Averages of P stream discharges and Canal Oe ste water levels showed similar trends (Fig. 2-3A and 2-3B). As water level in the canal increased from May-J une to July, so did streamflow in P streams despite the onset of drier climate conditions from El Veranillo de San Juan. Flow in P streams dropped considerably during the most intense periods of rainfall in September and October, contrary to the expect ation that increased rainfall w ould increase stream discharge. However, water level in the canal dropped to it s lowest levels during these two months, which likely accounts for the drop in P stream discharge, particularly in the one P stream in the lined section during September. Discharge in P stream s then increased again in November during the early dry season, as did canal water level. The drying of some P streams in the unlined section in November and January could reflect changes in the storage a nd movement of leaked canal water, especially in the unlined canal due to its earthen construc tion and greater leakag e potential. Overall, the existence of streamflow before onset of the wet season in May and continued flow in most streams into
33 November and January indicate that P streams are able to maintain flow in the dry season when typical dry forest streams are dr y due to leakage from the canal. Physical-Chemical Parameters Dry months Interm ittent streams in the dry season and the little summer lacked flow, but did occasionally contain isolated pools, particular ly immediately downstream of canal culverts, which were sampled for physical-chemical properties, aquatic macroinverteb rates, and fish. Isolated pools were encountered in T streams in early May (N=1), July (N=5), November (N=4), and January (N=2). Average water temperature was 26.2 + 0.7C, which fell within the range of T streams with flow (Fig. 2-4A), and average dissolved oxygen was 1.8 + 1.1 mg/l, lower than T streams with flow (Fig. 2-4B). The averag e conductivity (257 + 80 us/cm) was higher than T streams with flow (Fig. 2-4C), likely due to leachates from the decomposition of wetted organic matter or nutrient inputs from water passing through the hyporheic zone (Williams 1996). Physical-chemical values fall within the ranges of isolated pools examined in a limnological study of tropical dry forest streams in Santa Rosa National Park (SRNP) in Guanacaste, Costa Rica (Chapman and Kramer 19 91b). The SNRP streams also had isolated pools in May, July, and November. Mean wate r temperature of the SNRP pools was 24.7 + 0.9C in the morning and 25.1 + 1C in the afternoon, s lightly lower than the pools in this study. Little variation in water temperature occurr ed among SNRP pools or over time. Daily mean dissolved oxygen concentration in the SNRP pools varied between 1.3 to 5.5 mg/l, with low values before June floods, increased concentrat ions over the course of the wet season, then a return to low values by February (Chapman and Kramer 1991b).
34 Thus, low dissolved oxygen and high conductivit y are typical characteristics of isolated pools in dry forests in Costa Ri ca. Habitat contraction from drying and subsequent pooling can result in variability in dissolved oxygen due to such factors as elevat ed temperatures, algal production, microbial respiration, and reduced aeration. Furthermor e, cessation of streamflow generally reduces dilution of nutrients, increases deposition of fine sediments, and increases salinity and conductivity in residual waters (Gasith and Resh 1999). Wet months Water tem perature was significantly higher in P streams than T streams during each streamflow period (May-June t=3.43, p=0.003; September t=2.25, p=0.036; & October t=2.93, p=0.009) (Fig. 2-4A). Canal Oeste had higher aver age water temperatures than the P Streams, ranging between 26.928.0C. The significantly higher water temperatures in P streams could reflect a warming effect from leaked canal water influencing P streams. T stream temperatures in reaches above and below the canal were not signifi cantly different. No significant difference in dissolved oxyge n was found between stream types when both had flow in September & October, but stream types were significantly diffe rent in May-June (t=2.49, p=0.022), likely due to the one T stream outlier with particularly high discharge. Dissolved oxygen ranged from 4.6 + 2.0 mg/l to 6.8 + 1.0 mg/l among both stream t ypes in the periods with flow (Fig. 2-4B), and oxygen saturation ranged between 57 + 25% and 84 + 13%. In both stream types for all sampling periods combined, dissolv ed oxygen was positively correlated to stream discharge (T: F=0.6, p=0.001; P: F=0.384, p=0.025). No significant difference in dissolved oxygen was found between T stream reaches above and below the canal. In contrast, Canal Oeste oxygen saturation ranged between 97-133% during the same sampling periods, indicating high photosynthetic productivity, attri buted in part to full exposure of canal water to sunlight. However, this build up of dissolved oxygen in the water was lost before contributing to
35 streamflow in the P streams, probably due to the water passing through the earthen canal and ground before entering the stream. A significant difference in conductivity was found between stream types during each period of streamflow (May-June t=3.73, p=0.001; September t=2.25, p=0.036; & October t=2.93, p=0.009). Conductivity was always gr eater in P streams than T st reams, with a range of 90 + 28 uS/cm to 184 + 14 uS/cm in T streams versus 516 + 647 uS/cm to 399 + 265 uS/cm in P streams (Fig. 2-4C). P streams also ha d greater conductivity than Canal Oeste, which had a range of 106131 uS/cm during the same time periods. In T streams for all sampling periods combined, conductivity was negatively corr elated with dissolved oxygen and stream discharge (F=-0.426, p=0.021; F=-0.759, p=<0.0001). Elevated conductivity in P streams could be attributed to the water picking up solutes as it leached through the earthen canal walls. Perennial streams also have consistent flows at velo cities slow enough to allow accumula tion of organic matter and in situ decomposition, which could release more solute s or leachates into the water (Wantzen et al. 2008). No significant difference in conductivity wa s found between the T stream reaches above and below the canal. Canopy Cover Canopy cover was neither significantly differe nt between stream types nor between T stream reaches above and below the canal. The T streams averaged 82 + 9% and P streams averaged 81 + 9%. Consequently, light availability for primary production was similar between stream types and similar above and below the canal. However, canopy cover was significantly different over time (F=16.01, p<.0001), decreasing as the wet season prog ressed into the dry. Canopy cover was high in early May over T (90 + 6.5%) and P (88.6 + 3.6%) streams, which was unexpected given that the consistent, heavy rains of the wet season had not yet begun. This extensive canopy cover during the e nd of the dry season likely resu lted from an atypical period
36 of heavy rainfall in mid-April that induced early leaf out (Fig. 2-2B). Canopy cover was lowest in the early dry season and approximately 80% by November (T=80.1 + 6.1%, P=80.3 + 7.1%) and 70% by January (T=75.9 + 6.8%, P=70.3 + 6.4%), which is lower than May but still substantial cover. This trend in canopy cover corresponds with the observed vegetation phenology general pattern of leaf out in the wet season and leaf fall in the dry season in the region (Frankie et al. 1974, Ople r et al. 1980). The vegetation phenology studies found no clear difference between early and late wet season leaf out for trees, but had observed a difference in hill and riparian treelet/shrub communities. Average leaf matter mass in dr ift was greater in T streams be low the canal than P streams and T streams above the canal, but highly variable, and larger sticks were occasionally collected that likely skewed the averages (T-Below=3.11 + 3.40 g, P=9.97 + 14.47 g, T-Above=1.77 + 2.58 g). Due to that variability in T streams be low the canal, leaf litter collected in drift was neither significantly different between stream types nor between T stream reaches above and below the canal, corresponding to canopy cover resu lts of lack of significant difference between stream types. This suggests that instream orga nic matter availability was similar between stream types. However, storage time of litter within the stream was not investigated. Aquatic Macroinvertebrates Results: Dry Months A total of 3570 aquatic m acroinvertebrates we re collected from pools (N=11) in T streams below the canal, which had an average surface area of 8.9 m^2 + 7.8 m^2 with a range from 0.4 m^2 to 27.3 m^2. Diptera was the dominant orde r (76%), followed by Ephemeroptera (14%), Coleoptera (2%), Oligochaeta (4%), Hemiptera (2 %), and Gastropoda (1%) (Fig. 2-5A). Before the onset of the wet season in early May, the only pool remaining had primarily Hemiptera, Diptera, and some Coleoptera (F ig. 2-5B). The dominant ta xa overall were Dipterans: Chironomini (46%), Tanypodinae (15%), Cu licidae (8%), Tanyt arsini (4%), and
37 Ceratopogoninae (2%) (Fig. 2-5C ). An Ephemeropteran, Caenis (14%), and Oligochaeta (4%) were also representated. Figure 2-5D shows the dominant taxa over time. Drying can create isolated pools, which can exhibit major changes in community composition and density and can influence recolo nization of the stream upon rewetting (Lake 2000). Permanent and temporary pools in intermittent streams can provide refugia for macroinvertebrates, influence assemblage st ructure via physical-che mical conditions and biological interactions, and aid recovery from drought by providing a source of macroinvertebrates for recolonizat ion (Lake 2003) through aerial colonization by adults or drift upon rewetting (Paltridge et al. 1997). The pres ence of aquatic macroinvertebrate larvae in isolated pools in the intermittent streams in this study demonstrated that these pools did provide refugia and could become a source for macroinver tebrates in drift upon an initiation of flow, thereby influencing colonization and successional patterns. The macroinvertebrate community of tempor ary pools in this study was dominated by tolerant taxa such as Oligoch aeta, Diptera, Hemiptera, and Caenis Formation of pools from cessation of streamflow can result in deoxygenation, higher wa ter temperatures, detritus accumulation, and increases in conductivity and nutrients (Lake 2000), requiring greater tolerance by macroinvertebrates. Chironomids, a gastropod, and dytiscid beetles were found in harsh physical-chemical conditions of drying pools in Australia (Closs and Lake 1994). Macroinvertebrates characteristic of temporary pond habitats incl ude Hydracarina, Collembola, Odonata, Diptera, Hemiptera, Coleoptera, Ol igochaeta, Trichoptera, and Ephemeroptera (Williams 1996, 1997). Common Diptera in such ponds include Tipulidae, Culicidae, Ceratopogonidae, and Chironomidae (Williams 1996). The Ephemeropteran, Caenis has been found in high densities in temporary pools in interm ittent streams during the dry season (Nolte et
38 al. 1997), supporting the findings of this study that it has high tolerance for intermittent conditions. Most taxa found in temporary pools we re also found in flowing intermittent streams and perennial streams. Taxa inhabiting temporary pools, such as Chironomidae, Hemiptera, Coleoptera and Gastropoda, have evolved physiologi cal tolerances and modified th eir life history to survive (Williams 1996). Oligochaeta are also common in isolated pools due to their high tolerance levels and capability of living in the hyporheic zone during drie r conditions (Paltridge et al. 1997). Isolated pools can attract le ntic taxa, such as Veliidae, which has been found in other temporary streams (Boulton and Lake 1992, Acuna et al. 2005). Even the Ephemeropteran Caenis has a high tolerance for low dissolved oxygen conditions typical of residual pools (Nolte et al. 1997, Taylor and Kennedy 2006), explaining the large propor tion of Ephemeroptera found in isolated pools in this study. However, Ephemeropterans were most notable at the transition into the dry season when large pools still existed and were only beginning to dry, rather than at the beginning of the wet season (Fig. 2-5B), which may indica te some limit of tolerance. The high abundance of macroinvertebrates in the temporary pools in this study indicates that pools act as refugia for many ta xa in drier times. The great va riety of taxa also suggests that the pools enhance biodiversity. Sm all water bodies such as ponds, rivers, streams, and ditches can contribute to regional biodiversity (Williams et al. 2003). Similar macroinvertebrate orders were domin ant in the benthos and drift in P streams during the dry and wet seasons (Fig. 2-6). At th e end of the dry season, during El Veranillo de San Juan, and at the beginning of the dry season, Diptera and Ephemeroptera were the dominant orders in the benthos, followed by Gastropoda, Bivalvia, and Oligochaeta to a much lesser degree. The similarity in macroinvertebra te assemblages throughout this study indicates
39 relatively stable conditions in P streams as a result of continuous flow. Total macroinvertebrate abundances during dry periods were also comparable to abundanc es during wet periods (Fig. 27A and 2-7B). Aquatic Macroinvertebrates Results: Wet Months Benthic macroinvertebrates P stream s (N=35) had 9125 individuals, nearly three times as many as T streams (N=27), which had 3448 individuals (Fig. 2-7C). Total ab undance increased in both stream types as the wet season progressed from June to October, esp ecially in October (Fig. 2-7A and 2-7B). Diptera was the dominant order, comprising 56% of T stream and 47% of P stream benthic macroinvertebrates, and Ephemeroptera was next in T and P streams, representing 26% and 29%, respectively. Other taxa with notable represen tation were Oligoch aeta (9% and 4%) and Gastropoda (3% and 9%) in both stream t ypes and Sphaeridae (8%) in P streams. Order dominance differed between stream types over time (Fig. 2-8A and 2-8B). In T streams, Oligochaeta dominanted in May-J une (57%), but by Oct ober comprised only one percent of benthic macroinvertebrates. In c ontrast, Ephemeroptera increased in relative abundance as the wet season progressed, shifting from 1% to 36% representation. Diptera initially comprised 18% in May-June then was dominant in September and October at 60% in each period. In P streams, Diptera increas ed in relative abundance over time while Ephemeroptera increased into September then leveled off. Oligochaeta was not well-represented after May-June and the Sphaeridae relative proportion decreased cont inuously over time. Gastropoda had distinctly lower relative a bundance in October than previous months. Plecopterans were abse nt throughout the study. Overall, Caenis was both the most abundant Ephemer opteran and the most abundant taxon in T and P streams (21% and 26% of all taxa, resp ectively), but its abundance patterns differed
40 between the two stream types over time (F ig. 2-8C and 2-8D). In T Streams, Caenis was rare in May-June when Oligochaeta was the dominant taxon (57%), but it became the dominant taxon (29%) by October when Oligochaeta was rare (1%). In the case of P streams, Caenis dominated every month (May-June: 20%, Sept: 29%, Oct: 28%). The Chironomidae sub-families Tanytarsini, Tanypodinae, and Chironomini were the next most abundant taxa overall for both T stream s (18%, 15%, and 15%) and P streams (16%, 15%, and 11%) (Fig. 2-8C and 2-8D). In T streams, Orthocladiinae was the dominant Dipteran in May-June (9%), but had no notable representation thereafter. In September, Tanypodinae (23%), Tanytarsini (13%), and Chironomini (17%) were th e dominant taxa in T streams. This strong representation continued into October, when they attained the highest re lative abundances of all taxa after Caenis (14%, 23%, and 16%). In P streams, Tanypodinae representation remained relatively level, while Tanytarsini and Ch ironomini representation increased over time. Overall proportions of functional feeding gr oups were similar between stream types. collector-gatherers-scrapers-shredders were dominant in T streams (31%) and P streams (34%), followed by exclusive collector-gat herers (26% and 15%) then colle ctor-gatherers-filterers (19% and 17%) or predators (18% and 16%). The ma jor difference between stream types was the notable presence of filterers (8%) in P Stream s, which were likely more abundant due to persistent flows carrying particles downstream (Wantzen et al. 2008). Functional feeding group proportions changed ov er time for both stre am types (Fig. 2-9A and 2-9B). In T streams, collector-gatherers dominated (70%) in MayJune, but were only 18% of all taxa by October, reflecting decreased Oligochaeta dominance over time. Collectorgatherers-scrapers-shred ders increased in dominance over time from 12% in May-June and to 37% by October. Collector-gathe rers-filterers increased in do minance over time, representing
41 23% of all taxa by October. The increase in coll ector-gatherers-filterers and collector-gatherersscrapers-shredders over time likely reflects the ability of these macroinvertebrates to persist, to develop, and to utilize resources as a result of mo re consistent flows. In contrast, P streams had several functional feeding groups with similar proportions in May-June, which shifted in September to dominance by collector-gatherers-scrapers-shredders (31%) and collectorgatherers-filterers (20%). In October, relative abundances were similar to September, except for a slight increase in predators (13% to 19%). The increase in collector-g atherers-filterers and collector-gatherers-scrapers-shredders later in the wet season could reflect the influence of seasonal precipitation on aquatic macroinvertebrate s life cycles (Jacobsen et al. 2008) or greater resistance of these feeding groups to spates occurring in the wet s eason compared to other feeding groups (Ramirez and Pri ngle 1998b, Melo and Froehlich 2001). Direct calculations indicated that T streams had slightly less taxa richness than P streams. For all sampling periods combined by stream type, T streams had 47 versus 56 taxa in P streams. Taxa richness in T streams was always less than P streams (May-June: 27 vs. 40; Sept: 30 vs. 44; Oct: 32 vs. 42), and taxa richness values did not fluctuate over time within each stream type. Such factors as loss of aquatic habitat, dete riorated water quality, a nd alteration of food resources stemming from intermittent flows can have negative effects on the aquatic community composition and species richness (Lake 2003), which likely explains the lower taxa richness in T streams. NMDS ordinations were used to compare stream types across all stre amflow periods as a whole and within each sampling period, as well as change over time. The ordinations comparing stream types and sampling periods had a 2-dimens ional solution with a final stress of 20.022, and the two axes accounted for 87% of variation (axis 1=27%, axis 2=60% ). The ordination
42 comparing stream type in May-June had a 2dimensional solution with a final stress of 13.512, and the two axes accounted for 87% of variation (axis 1=70%, axis 2=17%). The ordination comparing stream types in September had a 2-di mensional solution with a final stress of 12.223, and the two axes accounted for 91% of variation (axis 1=37%, axis 2=53%). The ordination comparing stream types in October had a 3-dime nsional solution with a final stress of 7.899, and axis 1 and 2 accounted for 75% of variation (axis 1=52%, axis 2=23%). Comparison across all streamflow periods indi cated that taxa with lower water quality demands, such as Oligochaeta, Nematoda, Othoc ladiinae, associated with T streams (Fig. 210A). Discharge also associated with T stream s due to the large flow event captured in MayJune in one stream. Meanwhile, Gastropods a nd conductivity associated with P streams. Caenis chironimid pupae, and other chironomids (Tanytarsini, Tanypodinae, and Chironimini) associated with both stream types. The ordina tion examining change over time showed taxa that have high tolerances and/or are semi-terrestrial associated more with May-June, as well as flow, likely due to the outlier T stream (Fig. 2-10B). September and October samples largely grouped together, and Caenis and most chironomids associated w ith those months. Gastropods had no clear association. Comparison of stream types within each sa mpling period revealed striking differences between stream types in the ea rly wet season that faded as th e wet season progressed. In MayJune, stream types distinctly sepa rated (Fig. 2-11A). Semi-terrest rial and tolerant taxa strongly associated with T streams, along with flow and dissolved oxygen due to the outlier T stream. Orthocladiinae also strongly associated with T streams. The other chironomids and dipterans associated with P streams, perhaps due to perennial flows having already permitted colonization and larval development. Diverse taxa typified P streams, with Gastropoda, Sphaeridae, Caenis
43 Phylloicus, Elmidae, and Coengrionidae associating with P streams, as well as temperature. The higher diversity shown in the P streams compar ed to the T streams likely stems from the perennial versus nascent streamfl ow at this time period. In Se ptember, the two stream types were still separated, but less so than in May-June (Fig. 2-11B). A variety of di pterans strongly associated with T streams along with very tole rant taxa, like Oligochaeta and Nematoda. One Trichopteran also associated w ith the T streams. Concurren tly, Sphaeridae, Gastropods and Ephemeropterans associated with P streams, wh ich include taxa with greater water quality demands. Temperature and conductivity also associ ated with P streams. In October, the two stream types grouped very closely, but certain taxa still associated with on e stream type or another (Fig. 2-11C). S phaeridae, Gastropods and Caenis continued to associate more strongly with P streams. A wide array of taxa did not clearly associate w ith either stream type, such as chironomids and several Ephemeroptera. These fi ndings suggest that the longer periods of flow in T streams allowed larval development of more taxa, though perhaps th ere was still not enough time for sufficient growth of algae and bacteria on substrate for scraping Gastropods to increase in T streams. The number of indicator taxa decreased as the wet season progressed, indicating greater similarity between stream types over time (Table 2-1). In May-June, se veral Gastropoda, a few dipterans, Sphaeridae, Caenis and Coengrionidae were indicators of P streams, while Orthocladiinae and Collembola were indicators of T streams. In September, only one Gastropod (Physidae), Sphaeridae, and an Ephemeropteran, Farrodes were indicators of P streams, while Tanypodinae, Oligochaeta, and Nematoda were indicators of T str eams. By October, the only indicator taxon was Sphaeridae, for P streams. Decreasing numbers of i ndicator taxa over the course of the wet season reflect increasing similarity between th e two stream types, with taxa
44 that initially only associated with P streams later colonizing T streams. This colonization corresponds with the presence and persistence of streamflow in intermittent streams. Drift macroinvertebrates The total abundance of m acroinvertebrates in drift was very similar between the two stream types despite T streams having intermittent flow and P streams having perennial flow (Fig. 2-12A). T streams had a total of 2006 indi viduals (N=11) and showed a pattern in which the number of individuals collecte d declined in September, likely due in part to one less sample taken compared to other months, then reached its highest level in Oc tober (Fig. 2-12C). P streams had a total of 1903 individuals (N=12) an d showed an increasing trend over time in the number of individua ls collected. Overall representation of orders was similar between stream types, but individual orders showed trends over time. As a whole, T stream s had proportionally more Te rrestrial taxa, insects that fell into the stream, and Coleoptera, while P streams had more Epheme roptera, Diptera, and Bivalvia represented in drift (Fig. 2-12A). Ov er time in T streams, Terrestrial taxa and Collembola decreased proportionally as the wet season progressed (Fig. 2-13A). Ephemeroptera was not notable in May-June, but became the dom inant order by October, and Diptera remained one of the most dominant orders throughout the wet season. Over time in P streams, Ephemeroptera increased in propor tion from May-June to September, but decreased dramatically in October, when Diptera substantially increas ed in proportion (Fig. 2-13B). Hydracarina had a proportionally large presence in both stream types that decreased over time in drift. As a whole, the dominant taxa in both stream types were similar (Fig. 2-12B). The major differences were higher represen tation of Baetid ae, excluding Baetodes, and Terrestrials in the intermittent streams whereas Caenis and Tanytarsini had higher re presentation in the perennial streams. In T streams, Terrestrial taxa and Or thocladiinae were only dominant in May-June (Fig.
45 2-13C). Caenis became particularly domi nant in September, continuing into October when Baetidae, excluding Baetodes became the most dominant taxon. Dytiscidae, Coengrionidae, and Tanypodinae also were better repres ented in September and October. Planorbidae had a strong presence in September only, and Noteridae was well-represented in T streams only. Caenis was the dominant Ephemeroptera in P stream drift, followed by Haplohyphes (Fig. 2-13D). Caenis remained the most dominant taxon in each samp ling period, but had similar representation as chironomids in October. Haplohyphes was well-represented in May-June and September. Total taxa richness for all sampling periods wi th streamflow using direct calculation was 56 taxa in P streams and 53 in T streams. Overall, Haplohyphes (IV=70, p=0.011) was the only indicator taxon of P streams, while Hydraenidae, Dytiscidae, and Culicid ae (IV=50, p=0.012; IV=79.4, p=0.001; IV V=70.6, p=0.003) were indicators for T streams. The two Coleoptera taxa could be a remnant of isolated pools flushed out by streamflow since Coleopter a are characteristic of lentic conditions (Williams 1996). Both Hydraenidae and Dytiscidae taxa inhabit waters low in dissolved oxygen and increase in abundance as flow decreases (Acuna et al. 2005). Culicidae can survive in a wide variety of low oxygen habi tats, such as swamps, permanen t or temporary pools, ephemeral puddles, and artificial containe rs (Merritt and Cummins 1996). Structure of benthic versus drift communities The eight m ost abundant taxa in both the bent hos and drift (16 total taxa) in both stream types were categorized by the habitat they occupied and compared (Table 2-2). A drop in abundance values after the top eigh t taxa created an appropriate cu t-off point for analysis. Of these most abundant taxa in T streams, all f our chironomid sub-families, Baetidae (excluding Baetodes), and Caenis were in both the benthos and drif t. Oligochaeta and Ceratopogoninae were only in the benthos, while Hydracarina and No teridae were only in drift. Of these most
46 abundant taxa in the P streams, Caenis Chironomini, Tanytarsini, Tanypodinae, and Ceratopogoninae were among the most abundant in both the benthos and drift. Sphaeridae, Physidae, and Oligochaeta were dominant on ly in the benthos while Hydracarina, Haplohyphes and Orthocladiinae were only dominant in drift. For both intermittent and perennial streams, drift only macroinvertebrates were primarily Coleoptera and Hemiptera taxa wh ile the benthos only macroinverteb rates were a mix of orders (Table 2-3). When combining da ta from both stream types, the same associations of taxa with habitat existed, and intermittent only and perennial only taxa were a mix of orders (Table 2-4). These taxa found exclusively in a particular habita t or stream type were comparatively rare and lacked a dominant role in th e macroinvertebrate community. Aquatic Macroinvertebrates Discussion Spates and drying events create disturban ces that significan tly influence the aquatic macroinvertebrate community (Closs and Lake 1994, Acuna et al. 2005) due to streamflow patterns affecting habitat size, dissolved oxyge n, primary productivity, and water temperature (Williams 1996). In this study, the effects of st reamflow patterns were apparent for physicalchemical (temperature and conductivity) paramete rs and macroinvertebrates assemblages. For physical-chemical differences, P streams had si gnificantly higher temperature and conductivity compared to flowing T streams. However, dry peri ods resulted in cessation of flow in T streams, as well, producing considerable variability in physical-chemical conditions over time. These findings agree with other studies that revealed greater variability in physical-chemical conditions in intermittent than perenni al streams (Boulton and Lake 1990, Closs and Lake 1994). P streams had an established aquatic macroinvertebrate community, whereas T streams did not, but aquatic macroinvertebrates were able to colonize and repr oduce in streams with intermittent flows quickly. Within days of stre amflow initiation in intermittent streams in May-
47 June, chironomid larvae were collected. Alt hough the streams dried again in July, by midSeptember the chironomids and Ephemeropterans became more relatively abundant than MayJune, indicating that rainfall sufficient in fre quency and magnitude had occurred since the July dry interval to allow a comparatively robust eg g hatching then adequate development time for larvae. This growth in the chironomid and E phemeropteran assemblage size continued into October (Fig. 2-8A). Both the resistance and resilience of taxa ar e influenced by methods of colonization, rates of development, resource util ization, and life history characteristics (Lake 2003). In order to survive dry periods, aquatic insects of temporar y waters tend to have tremendous dispersal abilities, rapid growth, short and flexible life cycles, small sizes, generalist feeding, temperaturelinked development, and in some cases, diap ause (Williams 1996). Resilience rather than resistance appears to be the main survival strate gy of aquatic macroinverteb rates in these tropical dry forest streams, based on evidence of the rapi d colonization, development, and succession of the macroinvertebrate community in this study, wh ich appears to be the case for intermittent streams in the Sonoran desert (Stanley et al. 1994), Kansas (Fritz and Dodds 2004), and Spain (Acuna et al. 2005). Colonization A literatu re review by Paltridge et al. ( 1997) noted that local conditions (i.e. air temperature, streamflow, and the amount of stre ambed desiccation), substratum (i.e. sediments stability and hyporheic moisture), and proximity to permanent water sources have significant influences on recolonization by macroinvertebrates. Aerial colonization and drift from upstream perennial water sources, such as streams or isolated pools, are dominant modes of colonization under severely dry conditions (Paltridge et al. 1997, Fritz and Dodds 2004). Studies of intermittent streams in tropical or arid zones, in particular, have identified aerial recolonization
48 by ovipositing adults as the dominant mechan isms for streams recolonization, such as intermittent streams of both the Sonoran desert (Fisher et al. 1982) and Australia (Boulton and Lake 1992). Drift can be the dominant form of recolonization upon rewetting for intermittent streams with upstream perennial water sources in Australia (Paltridge et al. 1997) and the Great Plains (Dodds et al. 2004). The proximity of intermittent streams to a pere nnial water source or refugium influences the ability of larvae to drift downstream to colonize the rewetted stream and the ability of adults to aerially migrate to the rewe tted stream for oviposition (Fritz and Dodds 2004). No permanent upstream water sources in the intermittent streams were present that could have provided colonizers via drift in this study. The few small pools that did exist during the dry periods, could have aided colonization some what upon rewetting, though. Another route of recolonization can be via diapause of taxa wi thin the substratum (Williams 1996), but diapause has not been reported among aquatic macroinvertebrates collected from the Rio Tempisquito in Guanacaste (J ackson and Sweeney 1995). Low densities of diapausing taxa were reported for intermittent streams in Kansas, indicating low resistance of macroinvertebrates to dry conditions (Fritz and Dodds 2004). Kansas streams with particularly long dry periods were primarily co lonized through drift from upstream perennial refugia or aerial colonization (Fritz and Dodds 2002). Furthermore, the hyporheos probably contributed little to survival due to the duration of the dry season an d the generally small particle and interstitial space sizes of the substrate (Moellend, a combinati on of conditions that resulted in few species colonizing an Australian stream from the hyporheos once flow resumed (Paltridge et al. 1997). Thus, most colonization of the intermittent streams probably resulted from adults ovipositing upon rewetting. Consequen tly, resilience rather than resistance appears the primary
49 mode of recolonization, which wa s also the case for intermittent streams of the Great Plains (Fritz and Dodds 2004), Spain (Acuna et al. 2005), and the Sonoran desert (Stanley et al. 1994). The ability of extant aquatic macroinvertebrates to recolonize after disturbance varies according to larval size and species. While sm all body size enhances resistance to droughts by increasing the ability to use refugia, such as the hyporheos, it also may improve resilience by shortening the life cycle (Beche et al. 2006), which likely occurred for macroinvertebrates in the intermittent streams of this study. Larval size in fluenced the ability of Ephemeropteran species to respond to seasonal changes in disturbances, with Baetidae and Caenis early instar populations colonizing disturbed patches more quickly than la ter instars. However, a succession of distinct instar size classes did not occur after disturbance, which was attr ibuted to rapid development, continuous emergence, oviposition, interstitial refugia, and downstream drift (Mathooko 2002). The rapid population increase of Hydracarina, Oligochaeta, Nematoda, and many Diptera, particularly chironomids, upon rewetting of the in termittent streams in this study was similar to previous investigations. Some macroinvertebrate taxa appeared within days after rewetting in Sonoran desert streams (Fisher et al. 1982). Recolonization of str eams upon rewetting in Australian intermittent streams have include d macroinvertebrates such as water mites (Hydracarina), Ceratopogonids, tipul ids, oligochaetes, and nematods (Paltridge et al. 1997). Dipteran larvae, particularly Chironomidae and Ceratopogonidae, are typical pioneer taxa after rewetting of streams (Stanley et al. 1994, Acuna et al. 2005). Chir onomids are especially adept at inhabiting stream systems with extreme hydrological variability and habita t disturbance (Aguiar et al. 2002, Wood et al. 2005). The hi gher resiliency of chironomids has been attributed to their rapid life cycles (Dodds et al. 2004), and their ability to colo nize rapidly, which makes them particularly abundant at intermittent sites (Wood et al. 2005). High resilience to these
50 disturbances that permits rapid colonization and populat ion growth also has been attributed to traits such as smaller body size, short and as ynchronous life cycles, aerial reproductive adults, high dispersal abilities, and proximity to ups tream refugia (Dodds et al. 2004, Fritz and Dodds 2004). Taxa with slower life cycles and those with larger adult si zes arrived much later (Dodds et al. 2004). Of the chironomids in this study, the subfamily Orthocladiinae appears more particular about its habitat condi tions since it only was dominant in intermittent streams early in the wet season. The presence of Orthocladiinae only in the initial phase of succession in temporary streams was also found in tem porary streams in Spain (Casas 2008). The ability to recolonize quickly extends to other taxa, as well. Within four days of rewetting, 21 taxa had already r ecolonized intermittent streams in southern Oklahoma, especially chironomids, Oligochaeta, and Caenis In this study, oligochaetes and chironomids were dominant early in the wet season in both stream types followed by Caenis and chironomids later in the wet season. Caenis, Leptoplebia and Baetis mayflies were more resistant to spates than other taxa, and Caenis recolonized intermittent streams more quickly than other taxa after dewatering than other taxa (M iller and Golladay 1996). Alth ough Dipterans, particularly Chironomidae and Tipulidae, tend to be the dominant taxa in temporary streams, taxa such as Ephemeroptera, Hemiptera, Coleoptera, and Tric hoptera are considered well-suited to temporary streams (Williams 1996). For Epheme roptera in Brazilian streams, Farrodes dominated where there was higher streamflow, while Caenis dominated in more lentic conditions with lower streamflow velocities or absence of flows (Nolte et al. 1997), which also seemed to be the case in this study as Caenis dominated both stream types in all periods of streamflow and Farrodes was only a very minor element later in the wet season. In this study, the gradual shift from tolerant taxa, such as oligochaetes and chironomids, to other less tolerant taxa in T streams over the
51 course of the wet season refl ects the influence of more cons istent flows, which allowed colonization of taxa requiring higher wate r quality and longer development times. Development times The presenc e and growth of aquatic larvae in intermittent streams after relatively brief periods of flow suggest that aquatic macroinvertebrates of tr opical dry forest streams of northwestern Costa Rica, particularly chironom ids, have rapid development times. Rapid development has been reported for aquatic macroi nvertebrates in the Guanacaste region (Jackson and Sweeney 1995). Chironomids had especially fast hatching times, with the first hatching occurring in four days for Tanypodinae, 3-9 days for Orthocladiinae, and 2-9 days for Chironominae. Larval development time ranged from 26-86 days, 17-73 days, and 17-105 days, respectively. For Ephemeroptera, one species of Baetidae hatched after only 17 days, and the quickest larval development time was 26 days. Leptohyphes and Tricorythodes hatched after as little as 19 days and larval de velopment lasted 76-86 days, an d a Leptophlebiidae species first hatched after 18 days and developed in 131165 days (Jackson and Sweeney 1995). Other studies of intermittent streams documented relatively rapid colonization with development times of 3-5 weeks for Ephemeroptera in intermittent streams within the Sonoran Desert and draining into the Pantanal in Brazil (F isher et al. 1982, Nolte et al. 1997 ). A baetid mayfly in an intermittent stream in Kansas had a life cycle of 18 days, and a chironomid species had a life cycle as short as 6 days (Fritz and Dodds 2002). In Guanacaste, the trichopterans Leptonema Phylloicus, Polycentropodidae, and Helicopsychidae showed quick egg hatching times ranging from 9 to 24 days and larval development from 55-178 days, though Wormaldia developed in 45 days (Jackson and Sweeney 1995). These quick development times for chironomids co mpared to other taxa suggest that they should dominate stream communities in early stages of recovery from drying or flooding. This
52 would explain the presence of chironomids in T streams in May-June after only a brief period of streamflow. Overall, development times in northwestern Co sta Rica streams were fa ster than those of the temperate zones, with a few exceptions, in cluding Sonoran desert streams (Jackson and Sweeney 1995). However, the development times of tropical dry forest stream macroinvertebrates could potentially be even faster than those reported in the Jackson and Sweeney study (1995). Eggs of ch ironomid species reared at warm er water temperatures of 2022C versus 15C developed several days fast er (Jackson and Sweeney 1995). Warmer water temperature facilitates rapid development and ex tensive periods of reproduction and emergence of the Ephemeropteran, Caenis luctuosa, resulting in increased resilience to variations in water level and habitat availab ility (Peran et al. 1999). The wa ter temperature of the intermittent streams in this study averaged close to 26C, much higher than the 22C average rearing temperature of the Jackson study. Smaller adult size facilitates rapid developmen t (Beche et al. 2006). Larval and adult taxa from intermittent and perennial streams in this study were extremely small, which is consistent with the observations of Jackson and Sweeney (1995) for this region of Costa Rica. Rapid development times and lack of egg or la rval diapause suggest that many species are multivoltine, which would directly influence structure a nd function of these tropical dry forest streams. Multivoltinism has been found in semi-a rid or seasonally variable environments, and enhances resilience for populations facing unpredic table habitat availabil ity by permitting rapid exploitation of temporally variab le habitats and resources (Mill er and Golladay 1996, Taylor and Kennedy 2006). Multivoltinism and regional larval development could explain Chironomid assemblage growth observed in this study, as we ll as that of the Ephemeroptera assemblage,
53 which was slower due to their comparatively sl ower egg and larval development times. Rapid, asynchronous life cycles likely improves resiliency after spates or droughts (Fritz and Dodds 2004), such as findings suggest for these lo w-order tropical dr y forest streams. Succession Interm ittent streams in this study show ed much greater changes in aquatic macroinvertebrate abundances and composition over time than perennial streams, suggesting that streamflow has a strong influence on the co lonization and successi on of these aquatic macroinvertebrates. This corresponds with another intermittent stream study in Spain that associated changes in the macroinvertebrate co mmunity over the course of the wet season with flow (Acuna et al. 2005). That change seen ove r time in the intermittent streams in this study likely resulted from seasonal rainfall patterns in Guanacaste affecting streamflow, thereby influencing organic matter inputs, algal bioma ss, and aquatic macroinvertebrate growth and survival in dry forest streams. Clearer successional patterns are also expected for intermittent streams, particularly those without pools, than perennial streams due to the lack of viable eggs in the former once flow returns (Jackson and Sweeney 1995), whereas the perennial streams provide continuous habitat for egg ha tching and larval development. In addition, the aquatic macroinvertebrate comm unity in intermittent streams became more similar to perennial streams with increasing dur ation of streamflow, as shown by the increasing similarities in ordinations, decreasing numbers of indicator taxa, and increasing commonalities in community compositions over time. This trend of increasing similarity between stream types over time has also been observed in streams in Kansas (Fritz and Dodds 2004). Some variation was seen in composition and abundance over time in the perennial stream assemblage during the wet season in this study, perhaps due to spates, cha nges in food availability, or life history traits, like growth rates and multivoltinism (Jacobsen et al. 2008).
54 Functional feeding groups The dom inant functional feeding group of th e aquatic macroinvertebrate community for both stream types over time was collector-gatherer-scraper-shredde r, and other well-represented groups included collector-gathere rs, collector-gatherers-filterers and predators. Intermittent rivers during the spring in Portugal were domin ated by collector-gathe rers, then collectorfilterers and shredders to a lesser extent, which was attributed to an abundance of organic matter, particularly from agricultural sources (Aguiar et al. 2002). Organic matter persists on the streambed in low flow conditions due to lack of sufficient velocity to flush it downstream (Gasith and Resh 1999, Aguiar et al. 2002). Due to extensive canopy cover over these first-order streams and dominance of deciduous trees in tropical dry forests, as well as the generally dry or low flow conditions, detritus was present in the streambed and could pers ist there for relatively lengthy periods of time. Availabi lity of fine particulate organi c matter to macroinvertebrates is facilitated by fast leaf decomposition in wa rm tropical streams (Covi ch 1988). This could explain the abundance of various types of colle ctor-gatherers and the lack of exclusive Plecopteran shredders found in this study. Shredders are also lacki ng in the tropics in general, a phenomena attributed to the fast decomposition of tropical leaves limiting the need for shredders for their breakdown and the flushing of accumulate d dry season leaves during wet seasons spates before the leaf matter can be processe d by shredders (Wantzen et al. 2008). Functional feeding group designa tions used in this study were based on temperate species (Merritt and Cummins 1996, Brown 2001) due to lack of information about feeding habits of tropical stream taxa and may not fully reflect the flexibility in feeding types of the macroinvertebrate taxa of tropical dry forest streams. Many neotropi cal stream taxa are collector-gatherers of fine detritus, which tend s to be present and abunda nt throughout the year. Feeding strategies that involve high mobility or visiting unstable substrates, such as those used
55 by scrapers, predators, and shredders, are also co nsidered less suitable for macroinvertebrates in streams with frequent and unpredictable disturbances (Tomanova et al. 2006). This flexibility in feeding was found in northwestern Costa Rica, where Leptohyphes and chironomids unexpectedly consumed leaves while Trichoptera species considered primarily detritivorous shredders consumed algae when available (Jackson and Sweeney 1995). Tomavina et al. (2006) developed fuzzy codes that better reflected the functional feeding profiles of taxa in neotropical streams and these fuzzy codes showed similariti es to the Merritt & Cummins (1996) functional feeding classifications for the le vel of taxa identification empl oyed in this study. Thus, this study still provides insight into feeding habits, bu t revisions may be required in the future. Taxa richness and rankings The inte rmittent streams in this study had sim ilar or slightly less taxa richness than the perennial streams, corresponding with several st udies that have shown similar taxa richness between intermittent and perennial streams (F eminella 1996, Casas 2008) and less taxa richness in temporary streams than perennial streams (S mith et al. 2003, Wood et al. 2005). This study also found that benthic taxa richness decrea sed over time in intermittent streams while abundances increased. In contrast, intermittent streams in Victoria, Australia increased in taxonomic richness with duration of streamflow (Boulton and Lake 1992). This study did not find a similarity between benthic and drift taxa richness patterns. In spite of differences in taxa richness and abundance, this study found substantial taxonomic overlap, as shown by the many shared do minant taxa between stream types (Table 22) and only a small amount of rela tively rare taxa found exclusivel y in a particular stream type (Table 2-4). A high degree of similarity in taxa between the two stream types was found in Australia (Fritz and Dodds 2004) and in England (Wood et al. 2005). The ability of intermittent streams, particularly headwaters, to support ta xonomically rich macroinvertebrate communities
56 and the considerable taxonomic similarity between the intermittent and perennial streams have led to calls for the inclusion of intermittent streams in biodiversity protection efforts and conservation plans (Meyer et al. 2007, Casas 2008). Fish Results & Discussion Dry months Fish were found in the perennial stream s in early May, July, November, and January (N=778) and residual pools of the intermittent st reams (N=100) during El Veranillo de San Juan in July, and in the early dry season in November and January. Brachyrhaphis olomina (Meek 1914) (BROL) was the dominant species, compri sing between 95% of the fish in early May, 79% in July, 88% in late N ovember, and 80% in January. Poecilia gillii (Kner & Steindachner 1863) (POGI) was the second most dominant specie s with 2%, 17%, 8%, and 9% representation, respectively. Although intermittent streams did not have flow, fish were collected from residual, temporary pools within the stream channel th at had an average pool size of 11.7 m^2 + 8.4 m^2. Of the four species collected, two species were poecilids ( B. olomina and P. gillii ), the characin Astyanax aeneus (Gunter 1860) (ASAE), and Archocentrus nigrofasciatus (Gunter 1869) (ARNI), the convict cichlid. Overall, B. olomina comprised 44% of fish collected, followed by P. gillii (36%), A. aeneus (13%), and A. nigrofasciatus (7%). Poecilia gillii accounted for 64% of fish collected in July, followed by A. aeneus (23%). The only fish collected in November was A. aeneus. Brachyrhaphis olomina accounted for 53% of fish co llected in January, followed by P. gillii (29%). Archocentrus nigrofasciatus was only collected in January. This finding that some fish species are capable of inhabiting isolated pools within tropical dry forest streams is supported by previous studies of P. gillii where pools served as refugia and species survived and reproduced under low disso lved oxygen conditions (Chapman et al. 1991).
57 The fish in this study moved upstream from pere nnial streams during peri ods of streamflow and were trapped in the pools as flow disappeare d. Chapman and Kramer (1991a) also found that P. gillii moved upstream, but to a lesser extent than dow nstream due to more instream obstacles in upstream reaches. Decreasing pool volumes in dr y season refugia increases phys ical-chemical extremes and fish concentrations. Habitat use as refugia depe nds on the species, size, and life history traits of the fish, and the fish community that develops in pool refugia can be shaped by predation and competition (Magoulick and Kobza 2003). The lack of fish collected in November could be attributed to large pool sizes in this sampling period providing greater food availability and dispersal ability within the pool, instead of a re flection of fish presence/absence or abundances. The relatively high number of fish collected in January supports this assertion, since no other rain events occurred between N ovember and January that would have allowed fish migration into the pools. However, isolated pools may not serve as adequate refugia for fish to persist through the entire dry season, as shown by the absence of fi sh in isolated pools in early May found in this study. Isolated pools can dry, become uninhabitable, or diminish enough in size to make fish easy prey, and thus absent, by the end of the dry season (Mag oulick and Kobza 2003). Most mortalities of P. gillii in the SRNP stream study were attr ibuted to becoming trapped in pools that eventually dried completely (Chapman a nd Kramer 1991a). The most influential water chemistry parameters on fish survival in temporary pools are oxygen, temperature, pH, and nutrients. Low dissolved oxygen and high temperat ures are typical conditions of these refugia (Magoulick and Kobza 2003). Characteristics th at aid survival in temporary pools for the species found in this stu dy are discussed later.
58 Wet months Total abundance of fish collected in m innow traps for May-June, September, and October was much greater in P streams than T str eams (585, N=24 and 196, N=18, respectively). Six species were collected between both st ream types. Relative abundance of B. olomina (70%) was greatest followed by P. gillii (24%) in P streams, whereas P. gillii (64%) was greatest followed by B. olomina (22%) in T streams (Fig. 2-14A). These tr ends were consistent over the course of the wet season in P streams for B. olomina (May-June: 63%, September 68%, and October 75%) and P. gillii (May-June: 36%, September: 12%, and October 20%) (Fig. 2-14B). Poecilia gillii dominated T streams over time (May-June: 86%, September: 59%, and October 68%) followed by B. olomina except in May-June (May-June: 0%, Se ptember 16%, and October 28%) (Fig. 214B). Beyond those dominant species, A. aeneus and A. nigrofasciatus had a small yet notable presence as a whole (P streams: 4% and 2%; T streams: 8% and 5%). Rhamdia nicaraguensis a catfish, and Synbranchus marmoratus, called the swamp eel though not a true eel, represented 1% or less of all fish collected. No fish were seen or collected in T streams above the canal. Only four freshwater shrimp were collected during the course of the entire study and all from the same P stream with sampling points in close proxim ity to a large perennial river, Rio Piedras, a potential source. Consequently, it appears that shrimp did not commonly use these first-order streams, intermittent or perennial. Environmental conditions and traits of fish species influence colonization and survival. Stream hydrology is considered the ultimate driver of variation in habi tat quality and quantity. Increased water level in the wet season in low-gradient tropical streams creates additional aquatic habitat, increases ecosy stem productivity, and stimulates fish reproduction (Winemiller et al. 2008). Additional aquati c habitat provides additional food sources, such as organisms in drift. Terrestrial insects that have found their way into stream drift increase food availability for
59 fishes and can remove some pred ation pressure for aquatic insect s (Wantzen et al. 2008). Drift samples in this study did contain a considerable amount of terrestrial fauna that had fallen into the stream, which could have added to the diet of fishes in these streams beyond the aquatic macroinvertebrates in the benthos and drift. Conversely, the lack of precipitation in the dry season decreases aquatic habitat availability, decreases productivit y, and diminishes water quality, which lead to greater fish mortality (Winemiller et al. 2008). Poor water qua lity in the low-flow streams or residual pool refugia usually consists of low dissolved oxyge n and high temperatures (Magoulick and Kobza 2003). Effects from drought can also include changes in community composition, altered movement within the stream, and crowding of fi sh from habitat reduction (Matthews and MarshMatthews 2003). The intermittent streams in th is study experienced these conditions whereas perennial streams generally did not. Factors attributed to fish survival in dr ought conditions include disturbance intensity associated with streamflow, distance to refuge and refuge size, and species size, type, and mobility (Magoulick and Kobza 2003). While some fish can move into residual pools for refuge during dry periods, as already show n in this study, they can also move into perennial streams for refuge before smaller tributaries dry completel y. Perennial streams then become a source for colonists to dry tributaries upon rewetting (Mag oulick and Kobza 2003). This was the case for the intermittent tropical dry forest streams. Wa ter depth, current velocity, and predation risk could also inhibit movement of fish species or individuals of particular sizes (Magoulick and Kobza 2003). Meanwhile, perennial tropical dr y forest streams maintained a continuous connection to a source of colonists. The first-order intermittent and perennial streams were
60 tributaries to either the downstream perennial Rio Piedras near the lined canal portion or the downstream perennial reach of Quebrada Barbudal near the unlined canal portion. The resilience of fish to drought is demonstrated by their ability to recolonize streams quickly when flow returns and their resistan ce to a drought disturbance depends upon their adaptations and the quality of refugia available (Magoulick a nd Kobza 2003). Species in the intermittent streams in this study showed re silience by quickly migrating upstream from perennial water sources or disp ersing out of residual pools to colonize new stream habitat upon rewetting. Others showed resistance by surv iving in residual pools during dry periods. The two most dominant species, B. olomina and P. gillii, share traits that allow them to tolerate a wider range of habitats and conditions, but P. gillii appears the more tolerant overall. Small fishes with high reproductive efforts are ty pical inhabitants of trop ical headwater streams with large variation in flow (Winemiller et al. 2008). Brachyrhaphis olomina is an endemic species found in waters with lo w to moderate velocities, such as streams and sometimes shallow waters of rivers. It can tolerate temperatures of 19 to 34C, is insectivorous, and can reach 60 mm in length. Poecilia gillii inhabits streams with various velocities, but is particularly abundant in slower moving waters, such as swamps small streams, and shallow waters of large rivers. It can also tolerate a wide range of temperatures, 19 to 37C (Bussing 2002), and hypoxic conditions that occur in isolated pools (Chapman and Kramer 1991b). The diet of P. gillii consists of detritus, ooze and filamentous al gae from the substratum. This livebearing fish reproduces throughout the year and can reach 7 mm in length, with some reaching 105 mm (Bussing 2002). The wider tolerance of st ream velocities and temperatures of P. gillii as well as its broader diet, could account for its greater presence in the more dynamic intermittent streams than the perennial streams.
61 Chapman and Kramer (1991a) showed that P. gillii can persist in intermittent streams despite stressful conditions within isolated, re sidual pools and high mortality in flood events. Population densities were greatest in the dry season in isolated pools. Rainfall-induced flows permitted P. gillii to move both upstream and downstream from these isolated pools, though mortality reached 75% after a heavy rainfall due to fish becoming trapped in desiccating pools. However, the prevalence of P. gillii in pools during the dry season demonstrated that pools in intermittent streams served as refugia, which provided a more stable environment in intermittent streams over time and permitted greater resilience in post-flood recovery (Chapman and Kramer 1991a). Although, no similar study has occurred for B. olomina to compare the two species directly, these studies of P. gillii indicate that it can handle conditions of intermittent streams, such as taking advantage of flow events to disp erse and inhabiting residu al pools with poor water quality. The less dynamic environment of perennial streams could provide better conditions for the seemingly less tolerant B. olomina. The other species found also have traits that would permit surv ival in both stream types. Astyanax is one of the dominant gene ra in Central America. Of the three species in Costa Rica, A. aeneus has the widest distribution, living in all types of fres hwaters from high velocity streams to swamps and stagnant pools. However, Bussing (2002) reporte d it as absent from temporary streams, which this study disproves. Astyanax aeneus was found in the temporary streams during flow periods and in temporary pools, at least for a short time. Massive migrations upstream to avoid reduction of habitat in th e drying lowlands have been observed for A. aeneus (Lopez 1978 in Bussing 2002). In this case, it mi grated upstream to exploit new habitat then became trapped in residual pools in dry periods. Astyanax aeneus tolerates stream temperatures
62 from 20 to 37C, typically reaches lengths of 70 to 80 mm, and reproduces year-round or seasonally. Its diet includes algae, seeds, leaves, insects (aquatic and terres trial), and fish fry. Archocentrus nigrofasciatus is similarly omnivorous, eating aquatic insects, seeds, leaves, algae, and detritus, and has a nearly identical str eam temperature tolerance range. It also inhabits streams with a wide range of cu rrent velocities and is capable of growing to 100 mm in length, but usually is much smaller. Rhamdia nicaraguensis inhabits streams of lo w to moderate velocity with temperatures between 20 and 28C and is found in streams with stone, sand, or mud substrate, feeding mainly on aquatic insects, seeds, and small fish. It can grow to 260 mm, though individuals found in this study were much smaller. Synbranchus marmoratus inhabits rivers with low to moderate current velocities with stream temperatures between 22 and 34C. It usua lly buries itself in the substratum and is capable of aestivating in burrows covered with mud during the dry season. Its diet consists of small fish and river prawns, and individuals can range from 70 cm to 1.5 m, though those collected in this study were at the low end of the spectrum (Bussing 2002). Similar to the fishes in the tropical dry forest streams in this study, those in the Great Plains can migrate into intermittent streams a nd reproduce quickly. Fishes rapidly recolonized intermittent streams in Kansas and Oklahoma once rewetting allowed fish dispersal from perennial reaches into those previously dry streams (Ma tthews and Marsh-Matthews 2003, Franssen et al. 2006). In addition to environmental cond itions affecting fish, fish in turn can influence the stream ecosystem by altering primary productivity, nutrient cycling, and macroinvertebrate community composition. Droughts can make these effects more localized or patchy (Matthews and Marsh-
63 Matthews 2003). However, the influence of fish on intermittent or perenni al tropical dry forest stream ecosystems was not directly investigated in this study. Conclusion Tropical stream s face many threat s, particularly deforestati on, pollution, water abstraction for agricultural use or municipa l use, and alterations in stre am connectivity. Compounding those threats is the lack of scientific information in the tropics needed to make management decisions (Ramirez et al. 2008). This study begins to fill the gap in knowledge of tropical dry forest stream ecosystems, illuminating not only their structure a nd function but also the influence of regional water management practices on them. Stream flow has a major impact on colonization, development, and succession of biota as shown by differences between intermittent and perennial streams. However, de spite differences in streamflow between the two streams, many similarities existed in the macroinvertebrate a nd fish assemblages. The major impacts of the canal on these tropical dry fo rest streams were increased habitat during the dry season, fragmentation of habitat in the wet season when fish cannot pass through the canal culverts to access habitat above the canal, scouring of stream bed below culverts that enhances the pooling of water and its persistence, and higher water temperatur e and conductivity in the canalinfluenced perennial streams. In addition to the importance of understa nding how the irrigation infrastructure influences stream systems during this period of agricultural expansion, understanding the dynamic structure and function of tropical dry fore st streams is particularly pressing due to potentia l effects of global climate change on precipitation and streamflow patterns in the region, which could affect bi odiversity (Matthews a nd Marsh-Matthews 2003). Consequently, further research on tropical dry fore st streams is essential for proper management of freshwater resources and bi odiversity in this un ique, already water sc arce environment.
64 A B Figure 2-1. Ecohydrology Study si te maps. A) Lined canal section, and B) Unlined canal section. P=Perennial, T=Intermittent.
65 A B Figure 2-2. Precipitation data from the Palo Ve rde Meterological Station (OTS): A) Monthly rainfall, (*) indicates months sampled, and B) Daily precipitation 2007-2008, arrows indicate sampling dates.
66 A B Figure 2-3. Discharge. A) Aver age discharge over time by stre am type, without outlier in intermittent stream in May-June, and B) Water levels below maximum storage capacity of the canal ov er 2007-2008 study period.
67 A B Figure 2-4. Physical-chemical average values, by stream type for periods of streamflow (MayJune, September, and October) and pools (Early May, July, Late-Nov, Jan). A) Water temperature, B) Conductivity, C) Dissol ved oxygen, and D) Stream discharge.
68 C D Figure 2-4. Continued
69 A B Figure 2-5. Isolated pool benthic macroinvertebrates having 5% or greater relative abundances. A) Intermittent stream, by order, all dry pe riods, B) Intermittent stream, by order, by sampling round, C) Intermittent stream, by taxa, all dry periods, and D) Intermittent stream, by taxa, by sampling round.
70 C D Figure 2-5. Continued
71 A B Figure 2-6. Benthic & drift macroinvertebrate orders having 5% or greater relative abundances, by dry & wet periods. A) Benthos, perennial stream, by order, dry periods, B) Benthos, perennial stream, by or der, entire study, C) Drift, perennial stream, by order, dry period, and D) Drift, perennial stream, by order, entire study.
72 C D Figure 2-6. Continued
73 A B Figure 2-7. Benthic macroinverteb rate total abundances. A) Over time, by stream type, B) By stream type for wet periods (May-June, Se ptember, and October), and C) By stream type and order.
74 C Figure 2-7. Continued
75 A B Figure 2-8. Benthic macroinvertebra tes having 5% or greater relativ e abundance, by stream type. A) Intermittent stream, by order, B) Perennial stream, by order, C) Intermittent stream, by taxa, and D) Perennial stream, by taxa.
76 C D Figure 2-8. Continued
77 A B Figure 2-9. Benthic macroinverteb rate functional feeding groups (FFG) having 5% or greater relative abundances, over time. A) Interm ittent stream, and B) Perennial stream. Coding: PRED=Predators; FILT=Filterers; SC=Scrapers; CG=Collector-gatherers; CG-SH, CG-SC=Collector-gatherers-scrapers-shredders; CG-CF=Collectorgatherers-filterers; ALL=All feeding m odes; and OTHER=Shre dders and FFG with less than 5% representation.
78 A Figure 2-10. Relativized ordinati ons, all streamflow periods comb ined. A) By stream type, Diamond=Perennial and Triangle=Intermittent, and B) By sampling round, Diamond=May-June, Triangle=Septemb er, Square=October. Coding: ACARINA=Hydracarina, ANCYL ID=Ancylidae, CAENIS= Caenis CHIR_P=Chironomidae pupae, CHRMINI=Chironomini, COLLEMB=Collembola, HYDROBI=Hydrobiidae, LperS=Discha rge (l/s), NEMATODA=Nematoda, OLIGO=Oligochaeta, ORTHOCL=Or thocladiinae, PHYSID=Physidae, SPHAERID=Sphaeridae, TANYPOD=Tanypodi nae, TANYTAR=Tanytarsini, and uS_cm=Conductivity (uS/cm).
79 B Figure 2-10. Continued
80 A Figure 2-11. Relativized Ordinati ons, by sampling month. A) MayJune, B) September, and C) October. Diamond=Perennial, Tria ngle=Intermittent. Coding: ACARINA=Hydracarina, ANCYL ID=Ancylidae, ASIOPLX= Asioplax, BAET=Baetidae (NonBaetodes ), BAETOD= Baetodes CAENIS= Caenis CERATO=Ceratopogoniinae, CHRMINI=Ch ironomini, COEN=Coengrionidae, COLLEMB=Collembola, DO_p=Dissolved oxygen (%), DO_mgL=Dissolved oxygen (mg/l), ELMID=Elmidae, FARRODES= Farrodes FORCIPO=Forcipomiinae, GOMPH=Gomphidae, HAPLO= Haplohyphes HYDROBI=Hydrobiidae, LEPTOHY= Leptohyphes LperS=Discharge (l/s), NAUCOR=Naucoridae, NEMATODA=Nematoda, NL=Hydroptilidae sp., OLIGO=Oligochaeta, ORTHOCL=Or thocladiinae, PHORID=Phoridae, PHYLLO= Phylloicus, PHYSID=Physidae, PLANORB=Planorbidae, SIMULID=Simuliidae, SPHAERID=Sph aeridae, T=Temperature (C), TANYPOD=Tanypodinae, TANYTAR=Tanytar sini, TDS=Total Dissolved Solids (TDS), THIARID=Thiaridae, and uS_cm=Conductivity (uS/cm).
81 B Figure 2-11. Continued
82 C Figure 2-11. Continued
83 Table 2-1. Indicator species anal ysis, by stream type over time.
84 A B Figure 2-12. Drift macroinvertebrates, by stream type. A) Total abundance, by stream types, B) Relative abundance by taxa having 5% or greater representation (May-June, September, and October), by stream t ype, and C) Total abundance, by sampling month.
85 C Figure 2-12. Continued
86 A B Figure 2-13: Drift macroinvertebrate relative abundance, over time. A) Intermittent streams, by order, (5%+ representation), B) Perennial streams, by orde r (5%+ representation), C) Intermittent streams, by taxa (10%+ representation), and D) Perennial streams, by taxa (10%+ representation).
87 C D Figure 2-13. Continued
88 Table 2-2. Eight most abundant ta xa in the benthos and drift, by stream type (16 taxa total per stream type), categorized by habitat occupied.
89 Table 2-3. Taxon habitat exclusivity by stream type.
90 Table 2-4. Taxon exclusivity, by stream type and habitat.
91 A B Figure 2-14. Fish relative abunda nce by stream type. A) Wet pe riods combined, and B) Over time.
92 CHAPTER 3 RIPARIAN PATCH STUDY Introduction Stream riparian zones are interaction zones be tween terrestrial and a quatic ecosystems that extend parallel to the stream course, laterally to the extent of stream flooding, and vertically into the forest canopy. They can profoundly affect the light and temperature regimes of streams as well as nutrient, sediment, and organic matter lo adings (Gregory et al. 1991). Allochthonous organic matter is a major contributor to the detritus pool of the str eam (Allan 1995) and an important energy source for the food web (Gregor y et al. 1991). The positive contribution of riparian zones to the stream food base is often counterbala nced by reduced production of autochthonous energy sources (macrophytes and algae) from low light availability resulting from extensive canopy cover (Bunn et al. 1999). Breaks (patches) in the longitudinal extent of riparian forest along a stream can create a mosaic of habitat types displaying distinctly di fferent abiotic and biotic structure and function, including light and water temperature regimes, nutrient availability and spiraling, decomposition rates, colonizing species, species interactions, and community re sistance and resilience (Pringle et al. 1988, Townsend 1989, Gomi et al. 2002). Such habitat pa tchiness can affect biotic communities both up and downstream, especially fish (Pringle et al. 1988), with impacts often extending over hundreds of meters of st ream length (Storey and Cowley 1997). Most patches in the riparian zone result fr om forest clearance for urban expansion and agriculture. While the bulk of research has focused on both mechanisms responsible for differences in patch structure and function relative to intact riparian zones and interpatch heterogeneity (Pringle et al. 1988), the literature base is sparse for both tropical streams in general, especially those of dry fo rests, and small patch dynamics.
93 The literature that does exist has documented impacts on biotic communities from riparian patches. In the humid lowlands of south eastern Costa Rica, deforestation altered the macroinvertebrate taxonomic composition and de creased diversity (Lorion and Kennedy 2009). The macroinvertebrate community in deforested agricultural streams in Madagascar shifted basal resource use, increased taxa capabl e of capitalizing on hi gher primary production compared to forested streams, and declined in species richness (Benst ead et al. 2003, Benstead and Pringle 2004). In Indonesia, ch annelization and convers ion of forest to agriculture resulted in fewer benthic morphospecies, the presence of gastropods, increased relative abundance of chironomids and baetids, decreased relative a bundance of hydropsychids, finer sediments, and higher nutrient concentrations (Dudgeon 2006). Although deforestation in the Ecuadorian Amazon led to decreased detritus, macroinvertebrate diversity, and relative density of collectors, and increased periphyton biomass, total macroi nvertebrate density, and relative density of predators (Bojsen and Jacobsen 2003), riparian canopy cover had a negative impact on night and day drift densities as a result of increased riparian cover that led to de creased fish and benthic macroinvertebrate abundance (Jacobs en and Bojsen 2002). Finally, d eforestation of riparian zones can result in reduced fish diversity a nd survival due to loss of shade, increased temperature, loss of allochthonous food resour ces, increased sedimentation, and increased nutrient loading (Pusey and Arthington 2003, Winemiller et al. 2008). Given the paucity of information about effects of deforestation on tropical streams, as well as lack of information on struct ure and function of streams in tropical dry forests, Quebrada Barbudal in Costa Rica provides a unique opportun ity to investigate bot h. A large irrigation canal was recently constructed that passes comple tely under the streambed of this second-order seasonal stream via an inverted siphon. The stream course remains uninterrupted, but the
94 riparian zone corresponding to the canal width plus an adjacent road (approximately 34 m in stream length) is continually cleared (San tana 2007 Personal Comm unication). Above and below the deforested zone, riparian vegetation is intact. However, the stream section below the deforested zone could have significantly different habitat quality than the stream section above the zone due to potential dow nstream export of primary production, sediment, and physicalchemical conditions from the deforested zone. Consequently, three distinct habitats exist within a stream length measuring only a few hundred meters, allowing fish and macroinvertebrat es to select among them. In addition to the canal resulting in a cleared riparian patch, an ou tlet near the siphon releas es canal water directly into Quebrada Barbudal approximately 10 m downst ream of the three distinct habitats, which makes Quebradal Barbudal a perennial stream beyond that confluence. This creates a fourth distinct habitat, a forested reach of perennial fl ow that could be compared to the seasonal reaches upstream. Thus, this study addresses two questions: 1) Do canals influence stream structure and function, and 2) Do patches in riparian vegetatio n affect macroinvertebra te and fish community structure within tropical dry fo rest streams? These questions are addressed by studying abiotic characteristics and biotic communities of four zones within Quebrada Barbudal: a forested zone above the deforested patch, the deforested patch where the canal passes, a forested zone below the deforested patch, and a forested zone fa rther downstream that r eceives both seasonal streamflow and direct cana l water discharge year-round. Methods Study Area A second-order stream Quebrada Barbudal, that runs within a tropical dry forest wildlife corridor between Palo Verde National Park a nd the Lomas Barbudal Reserve was the study site
95 (Fig. 3-1). The principal irriga tion canal in the region, Canal Oe ste, crosses Quebrada Barbudal where the wildlife corridor begins at the transi tion from the foothills within the Lomas Barbudal Reserve to the lowland plains. Canal Oeste provides water for agricultural fields in the lowland plains surrounding the wild life corridor. It passes undernea th Quebrada Barbudal through an inverted siphon (Fig. 3-2A and 3-2B), such that the only disturbances to the stream are the maintenance of a cleared riparian zone for canal passage under the stream and a road adjacent to the canal that crosses the stream via a low water crossing, which impounds water within the cleared riparian zone during low flows and enha nces pooling when streamflow ceases. Quebrada Barbudal is a seasonal stream until 41 m belo w the Canal Oeste crossing, when it becomes perennial due to constant water releases from a canal outlet up gradient from the siphon. The canal outlet is managed both to provide stream habitat year-r ound and to release excess canal water that would otherwise overtop the canal during the rare occasions when run-off from storm events combine with already high volumes of can al water or when the siphon becomes suddenly and unexpectedly clogged with debris (S antana 2007 Personal Communication). The tropical dry forest surrounding Quebrada Ba rbudal is primarily deciduous, with leaf out at the beginning of the wet season and leaf fall mostly in the dry season. The foothills are of volcanic origin, while the lowland plains consist of primarily allu vial deposits. More information on the phenology and geology of th e region, as well as details on Canal Oeste, are provided in Chapter 2. Site Selection Four distinct, sequential instream habitats we re sampled within Quebrada Barbudal. From upstream to downstream, the sites were defined as : 1) Above: Immediately upstream of the canal crossing with a forested riparian zone, 2) Open: Stream section with a deforested riparian zone and full sun exposure where the can al goes under the stream and the stream crosses the adjacent
96 canal road, 3) Below: Immediat ely downstream of the canal cro ssing with a forested riparian zone and a potential recipient of exported materi al from the Open zone, and 4) Canal-Influenced zone (downstream of the Below zone) with a fo rested riparian zone and a combination of seasonal streamflow from Quebrada Barbudal and direct, perennial flow from Canal Oeste. Data were collected at three points 10 m apar t in all zones, except for the Open zone, which had three points that were 8 m apart (Fig. 3-3). The entire Above zone could be considered everything upstream of the Open z one, but only 20 m in stream length was sampled in the Above zone, and the most downstream sa mpling point was 10 m upstream of the Open zone. The entire Open zone was a total of 25 m in stream length, but only 16 m in stream length was sampled within the zone. The canal road that the stream crossed was 8.8 m wide, and the most downstream sampling point in the Open zone was 4 m upstream of the canal road. The entire stream length of the Below zone was 41 m, but only a 20 m section within that was sampled. The most upstream sampling point in the Below zone was 11 m downstream of the road. The Canal-Influenced zone could be cons idered everything downstream of the confluence of Quebrada Barbudal and canal water released from the Canal Oeste outlet, but only 20 m in stream length was sampled, with the first point approximately 100 m downstream of the end of the Below zone. Sampling Schedule Sa mpling occurred from early May 2007 to January 2008 at 4 to 6 week intervals, with the exception of one eight week interval between the July and September samplings. The sampling spanned the end of the dry season in early May before heavy rainfall began, the wet season, and the early months of the following dr y season in November and January.
97 Field Sampling Physical-chemical Bankfull width and height, channel width and height, and slope were recorded once f or each zone at the beginning of this study in early May. Slope was determined using a laser level. Canopy cover was determined using a convex spherica l densiometer, and data were collected in early May, May-June, October, La te November, and January. Phot osynthetically active radiation (PAR) was measured using a LICOR-190 (terrestrial PAR meter) and LICOR-192S (underwater PAR meter) (Li-COR, Inc., Lincoln, NE, U.S.A.). Physical-chemical and streamflow data we re collected each sa mpling period in the daytime. Dissolved oxygen (% and mg/L), salinity, to tal dissolved solids (g/L), temperature (C), and conductivity (mS/cm-adjusted for 25C) were measured using a YSI 556 Handheld Multiparameter Instrument (YSI, Inc. Yellow Springs, OH, U.S.A.). Stream velocity was measured using a Marsh-McBirney flowmeter at three equidistant points across the width of the stream channel. In stream depths less than 20 cm, velocity was measured at 60% of the depth from the stream surface. When depth was 20 cm or greater, velocity was measured at 20% and 80% of the depth from the stream surface and then averaged. Discharge was calculated using the velocity, depth, and channel width measurements. Rainfall data were collected from a rain gauge operated by the Organization for Tropica l Studies at Palo Verde National Park. Biological Benthic m acroinvertebrates were collected us ing a 243 um Surber sampler. Sediment, small stones, and debris within the 0.093 m2 area of the sampler were disturbed by hand, and organic debris and macroinvertebrates were collecte d in the attached net. Macroinvertebrates in drift were collected using a 363 um drift net pl aced at the most downstream sampling point of each zone for 20 hours, spanning dusk, night, and dawn. Drift samples from zones in the
98 exclusively seasonal stream section were always taken on the same day. Macroinvertebrates on stones were sampled by removing a stone with one hand from the stream, following that removed stone with a Surber sampler net held in the other hand to catch any dislodged aquatic macroinvertebrates. After collecting macroinvert ebrates that fell into the net, the stone was scrubbed with a toothbrush within a plastic tray to remove macroinverte brates, using distilled water to aid the process. Contents of the plastic tray were emptie d into a plastic bag. Stone size was determined by measuring the B-axis (Gordon et al. 1992). All macroinvertebrate samples were preserved with ethanol in the field. Fish were sampled using bread-baited minnow tr aps, placed in the deepest part of the stream and left for 20 hours, spanning dusk, night, and dawn. Captured fish were measured for standard and total length, identified, and released back into the stream. A 20-hour sampling period for the macroinvertebrate drift and minnow trap samples was chosen due to logistical constraints in the field. Algal biomass was estimated as both peri phyton chlorophyll-a and Ash Free Dry Mass (AFDM), but periphyton chlorophylla analysis only estimates algal biomass whereas AFDM includes both algal biomass and a mixture of bi omasses from bacteria, fungi, small fauna and detritus. Periphyton was collected for chlorophy ll-a analysis by removing a stone from the stream using a Surber sampler net, placing a hollow rubber template with an internal area of 24.6 cm^2 on the stone to sample a consistent ar ea and prevent loss, then scraping the periphyton from stones using a nylon bristled brush. The periphyton collected on the brush was cleaned in 50 ml of water and stored with plastic ice in a cooler. This same procedure employed to collect periphyton was repeated on different stones for AFDM analysis (Steinman and Lamberti 1996).
99 Laboratory Methods All aquatic m acroinvertebrate samples were put through 1 mm and 250 um sieves in the laboratory. Both size classes were exam ined under a dissecting microscope, and macroinvertebrates were removed and placed in to dated and labeled vials containing 70% ethanol. Samples were then shipped to the United States for identification. Macroinvertebrates were generally identified to Family or Genus, except the families Chironomidae and Ceratopogonidae, which were identified to Sub-Fam ily. Non-insect macroinvertebrates, such as Oligochaeta, Nematoda, Hydracarina, and Hirudinea, were identified to Order. Identification of tropical aquatic macroinvertebrates to the specie s level is difficult due to lack of taxonomic tropical literature (Jacobsen et al. 2008). Identi fication utilized Merritt & Cummins (Merritt and Cummins 1996), assistance from M. Springer and P. Hanson at the University of Costa Rica, and Roldn (Roldn 1996). The periphyton chlorophyll-a and AFDM sa mples were filtered separately through Whatman pre-ashed and weighed filter-disks on the same day that they were collected. Samples were then frozen for preservation and shipped to the University of Florida for analysis. Periphyton chlorophyll-a samples were correc ted for pheophytin and measured with a spectrophotometer using standard procedures (A PHA 1998). AFDM filter samples were dried in an oven at 60C for 48 hours, weighed, then as hed in a muffle furnace for five hours at 550C and weighed again. Data Analysis To test for differences in variables betw een stream types and over tim e, Repeated Measures ANOVAS using PROC MIXED were run in SAS (SAS Institute, Inc 2006). To test for correlations between different variables, Pearsons Correlati ons using PROC CORR were run in SAS. Variables were log-normalized before running statistics when appropriate. When
100 differences were significant, pos t-hoc analysis was conducted usi ng Tukeys test and Bonferroni corrections. Taxa richness was used rather than species richness because not all identifications were made to the same taxonomic level. Conseq uently, taxa richness was calculated by summing the number of different taxa found to the level of identification chosen in this study. To investigate relationships of abio tic and biotic stream components with stream type and time, nonmetric multidimensional scaling ordinations (NMDS) using Sorensons metrics (i.e. Bray-Curtis) and Monte Carlo tests were created in PC-Ord (McCune and Grace 2002). Taxa data were logged and relativized. Bi plots with a cut-off of P=0.2 were used to define associations between the sample and taxa distributions. Indicator species analysis using Monte Carlo tests was conducted in PC-Ord. Results Physical-Chemical Parameters Canopy cover was significantly different only between the deforested Open zone and the Above (t =7.02, p= 0.0005), Below (t=-8.11, p= 0.0002), and Canal-influenced (CI) (t=-8.06, p=0.0002) zones, all of which had forested ripari an zones. Average canopy cover for the Open zone during the study period was 20 + 16%. The Above, Below, and Canal-influenced zones had average canopy cover of 70 + 11%, 78 + 11%, and 78 + 7%, respectively (Fig. 3-4A). No significant difference in canopy cover was found over time. Photosynthetically active radiation (PAR) valu es fluctuated considerably during sampling, but averaging these values over th e entire study still rev ealed that the terrest rial and underwater PAR were much greater in the Open than other zones (Fig. 3-4B). In the Open zone, mean terrestrial PAR was 807 + 769 umol and underwater PAR was 717 + 637 umol. Among the forested zones, average terrestrial PAR did not exceed 186 + 336 umol and the average underwater PAR did not exceed 244 + 345 umol. This difference between the Open zone and the
101 other zones was significant for terrestrial PAR (Above: t=-3.52, p=0.008; Below: t=4.95, p=0.0011; CI: t=4.98, p=0.0011) and underwater PAR (Above: t=-3.32, p=0.0106; Below: t=5.32, p=0.0007; CI: t=5.35, p=0.0007). The significantl y greater PAR levels in the Open zone correspond with significantly less canopy cover in the Open zone compared to the forested zones. Average chlorophyll-a levels were much greater in the Open than other zones over the entire study, except for one inst ance when the Canal-influenced zone was greater in LateNovember (Fig. 3-4C). However, chlorophyll-a wa s not significantly diffe rent among the various zones, except between the Above and Canal-influenced zones in the Late-November (t=-3.68, p=0.026). Ash-free Dry Mass (AFDM) was significantly different between the Open zone and other zones in May-June (Above: t=-4.96, p=0.005; Below: 3.67, p=0.026), September (Above: t=4.68, p=0.007; Below: t=3.7, p=0.025; CI: t=4.31, p=0.011), and January (Above: t=-5.17, p=0.004; Below: t=3.71, p=0.025; CI: t=3.92, p=0.019). There were no significant differences among Above, Below, and Canal-influenced zones, which were all forested. There was a significant difference over time (F=20.18, p<0.001), specifically between Ma y-June and all other months except January (Sept: t=7.02, p<0.0001; Oct: t=7.95, p<0.0001; Mid-Nov: t=7.29, p<0.0001; Late Nov: t=5.19, p<0.0001), and between January and September (t=-4.71, p=0.0004), October (t=-5.44, p<0.0001), and Mid-N ovember (t=-4.8, p=0.0003). Throughout the entire study, the average levels of AFDM in the Open zone were greater than all other zones, particularly in May-June and January (Fig. 3-4D). Stream discharge changed dramatically over ti me, with streamflow continuously increasing over time until peaking in October then droppi ng to low levels by November and remaining
102 constantly low thereafter (Fig. 3-5A). There was no significant difference in streamflow between the Above, Open, and Below zones, with the ex ception of one case in September between the Above and Below zones (t=-3.36, p=0.04), when th e Below zone discharge was significantly greater for unknown reasons. The Canal-influenc ed zone had significantly different discharge from the other zones: lower in July (Open: t=3.6, p=0.029; Below: t=4.17, p=0.013), greater in September (Above: t=-5.24, p=0.004; Below: t=3.92, p=0.019), and lower in October (Below: t=4.37, p=0.01). In July and October, the Canalinfluenced zone had much smaller discharges than the other zones, while in September it was much greater, which could reflect some interaction with the hyporheos or management of the canal outlet opening that permits water to enter Quebrada Barbudal. Discha rge from the canal out let changed over the course of this study, but unfortunately, this was not recorded during the study. Stream temperatures in the Open zone were significantly greater th an the Above zone in May-June (t=-3.42, p=0.037) and Late November (t=-3.81, p=0.022), and si gnificantly less than the Above zone in July (t=34.54, p=<0.0001) September (t=5.55, p=0.002), and January (t=11.57, p=<0.0001). Open zone stream temperatur es were significantly greater than the Below zone in May-June (t=3.88, p=0.02), Late-Nove mber (t=11.57, p=<0.0001), and January (t=6.24, p=0.001), and significantly less th an the Below zone in Sept ember (t=-7.86, p=0.0002). These significant differences between the Open and Belo w zones indicate that the Open zone stream temperatures did not have a downstream influe nce on the Below zone. The Above and Below zone were both significantly greater or less than the Open zone during most of the same sampling rounds, demonstrating similarities betw een the two the forested zones, as well as differences between the forested and deforested z ones. However, the reasons for differences in stream temperatures among zones during these pa rticular sampling rounds are unclear. As a
103 whole, the Above and Open zones fluctuated between having higher average temperature over time, while the Below zone usually had lower temp eratures on average than the Open zone. No significant difference was found between the Open zone and the Above and Below zones in late October and mid-November (Fig. 3-5B). Late October had the highest recorded stream discharge, which was likely great enough to hom ogenize stream temperatures in the adjacent Above, Open, and Below zones, and this homoge nization probably carried over into the next sampling in mid-November. In addition, the Above and Below zones were significantly different in temperature in late November (t=7.77, p=0.0002) and January (t=20.82, p=<0.0001), with the Below zone having lower average temp eratures than the Above zone. The Canalinfluenced zone was significantly less than th e Open zone in May-June (t=4.44, p=0.009). The Canal-influenced zone was also significantly di fferent from all other zones in July (Above: t=16.87, p=<0.0001; Open: t=-17.67, p=<0.0001; Below t=-17.47, p=<0.0001), September (Above: t=30.12, p=<0.0001; Open: t=24.57, p=< 0.0001; Below t=32.43, p=<0.0001), and LateNovember (Above: t=4.09, p=0.015; Open: t=7.8 9, p=0.0002; Below t=-3.68, p=0.026), varying in the type of significance among the zones. In January, the Canal-influenced zone was only significantly greater than the Open and Below z ones (t=-11.45, p=<0.0001; t=-17.69, p=<0.0001). No clear stream temperature trends over time were ev ident in the Canal-influenced zone. Significant differences in dissolved oxygen am ong zones in the seasonal stream section only occurred in May-June and January, the periods of lowest stream discharge. In May-June, the Above zone had significan tly lower dissolved oxygen levels than the Open (t=-4.4, p=0.01) and Below (t=-4.07, p=0.015) zones. In Januar y, the Open zone had significantly greater dissolved oxygen levels than the Above (t =-8.9, p<0.0001) and Below (t=7.37, p=0.0004) zones.
104 In both periods, average dissolved oxygen was hi ghest in the Open zone, followed by the Below zone, which suggests that the Open zone expor ted oxygen downstream (Fig. 3-5C). The Canalinfluenced zone also had significantly greater dissolved oxygen levels compared to other zones in May-June (Above: t=-6.92, p=0.0006), Septem ber (Above: t=-6.04, p=0.001; Open: t=-3.64, p=0.027; Below: t=-6.14, p=0.001), and January (Above: t=-5.66, p=0.0002; Below: t=-4.13, p=0.014), but the Canal-influenced zone had signif icantly lower levels than the Open zone in January (t=3.24, p=0.047), when the Open zone had higher average dissolved oxygen and significantly greater AFDM. The Canal-influenced zone had a gr eater average stream discharge associated with each case of significantly great er dissolved oxygen levels, such that higher stream discharge could have in creased dissolved oxygen levels. Conductivity was only significantly different on one occasion among zones in the strictly seasonal stream section. In September, the Open zone had significantly lower conductivity than the Below zone (t=-4.32, p=0.011) (Fig. 3-5D). However, the Canal-influenced zone had significantly lower c onductivity than all the other zones in September (Above: t=9.16, p=<0.0001; Open: t=7.08, p=0.0005; Below: t=11.4, p=<0.0001), late-November (Above: t=80.54, p=<0.0001; Open: t=80.54, p=<0.0001; Be low: t=82.56, p=<0.0001), and January (Above: t=19.45, p=<0.0001; Open: t=18.46, p=<0.0001; Below: t=19.94, p=<0.0001). The Canal-influenced zone had significantly greater conductivity than all the other zones in October (Above: t=-8.14, p=0.0002; Open: t=-8.93, p=<0 .0001; Below: t=-7.21, p=0.0004) and midNovember (Above: t=-7.56, p=0.0003; Open: t= -7.56, p=0.0003; Below: t=-6.8, p=0.0006). Lower average conductivity in the Canal-influen ced zone in September, Late-November, and January corresponds with higher average stream di scharge in the Canal-influenced zone than other zones during these periods. When average conductivity was higher than the other zones in
105 October, stream discharge was lowest in the Canal-influenced zone. The si gnificant difference in mid-November is unclear since the Canal-influe nced zone fell in the middle of the average conductivity range of the zones. Overall, for each zone, stream discharge positively correlated with dissolved oxygen (Above: F=0.737, p=<0.0001; Open: F=0.669, p= 0.001; Below: F=0.74, p=0.0001; CI: F=0.843, p=<0.0001) and was negatively correlated with conductivity (A bove: F=-0.794, p=<0.0001; Open: F=-0.845, p=0.001; Below: F=-0.847, p=0.0001; CI: F=-0.848, p=<0.0001). Increased stream discharge would permit greater oxygenati on of the water while diluting the amount of solutes. In the Above and Canal-influenced zones, conductivity negatively correlated with dissolved oxygen (Above: F=-0.455, p=0.038; Canal-influenced: F=-0.593, p=0.005), which could reflect a similar influence of stream flow on dissolved oxygen and conductivity as previously explained. In the Above and Open zones, temperature negatively correlated with dissolved oxygen (Above: F=-0.733, p=0.002; Open: F=-0.471, p=0.035), so stream temperatures decreased as oxygenati on increased. This correlation could demonstrate an indirect influence of stream discharge on stream temperatures since dissolved oxygen and discharge were found to be correlated. In the Below zone, temper ature negatively correl ated with conductivity (F=-0.652, p=0.001), which was unexpected since incr eased temperatures would indicate less stream discharge that would decrease solute dilution, so they should be positively correlated. However, warming of the water in the Open zone and exporting those higher temperatures into the Below zone could have altere d that relationship. In the Cana l-influenced zone, temperature negatively correlated with stream discha rge (F=-0.448, p= 0.042), which likely reflected increased flows cooling the water.
106 Biotic Parameters Benthic macroinvertebrates The greatest num ber of individuals was collected from the Open zone (8688), followed by Above zone (7263), Below zone (4289), and Canalinfluenced zone (3489). Total abundance of benthic macroinvertebrates in each zone remained relatively stable until after the wet season, when it increased markedly in November and Janua ry (Fig. 3-6A). The two dominant orders in all zones were Diptera (Above: 42%, Open: 20% Below: 50%, CI: 51%) and Ephemeroptera (Above: 53%, Open: 75%, Below: 42%, CI: 42%), with Ephemeroptera much more dominant in the Open zone than the other zo nes (Fig. 3-7A). Each zone s howed changes over time in the relative abundance of the aquatic macroinvertebrate s orders (Fig. 3-8). In each case, Diptera and Oligochaeta dominated in May-June, at the onset of the we t season following a period of no streamflow. The proportion of Ephemeroptera incr eased over the course of the wet season to become the dominant order in September a nd October. Trichoptera also increased in representation as the wet season progressed. Afte r the wet season, in November and January, the proportion of Ephemeroptera a nd Trichoptera decreased while Diptera became much more dominant in the Above and Below zones. In th e Open zone, Ephemeroptera remained the most dominant order, although Trichopt era no longer had strong represen tation. The Canal-influenced zone followed similar trends in macroinverteb rate order dominance as the Above and Below zones. Overall, the Ephemeropteran Caenis was the dominant taxon in each zone (Above: 21%, Open: 41%, Below: 26%, and CI: 24%). Haplohyphes was the second most dominant taxa in the Above (22%) and Open zones (24%), while chiron omids were the second most dominant taxa in the Below (Tanypodinae: 25%) and Canal-influe nced zones (Chironomini: 23%). Tanypodinae was also well-represented in the Above (20%) an d Canal-influenced (13%) zones. The Open
107 zone had similarly low abundances among the chironomids Chironomini (5%), Tanypodinae (5%), and Tanytarsini (6%). Other Ephemeropt era in all zones includ ed Baetidae taxa (nonBaetodes), Farrodes, and Leptohyphes. The dominant functional feeding groups in th e benthos shared similarities among zones (Fig. 3-10A). In the Above zone, the order of decreasing abundance wa s collector-gatherersfilterers (35%), collector-gatherers-scrapers-shredders (25%), and predators (22%). The Open zone was dominated by two groups, collector-gatherers-scrapers-shredders (45%) and collectorfilterers (37%), which likely reflects the greater dominance of Ephemeropter a in this zone than other zones. In the Below zone, the order of abundance was collect or-gatherer-scrapersshredders (30%), predators (27%), and then collector-gatherers -filterers (21%). The Canalinfluenced zone was dominated by collector-gat herers (25%), collector-gatherers-scrapersshredders (30%), predators (15%), and collector-filterers (18%). Both total and average taxa ri chness of benthic macroinverteb rates were similarly slightly greater in the Above zone (55 and 26 taxa) and Open zone (5 5 and 26.3 taxa) than the Below zone (53 and 23.2 taxa) and Canal-influenced z one (48 and 24.5 taxa) (Tab le 3-1). Table 3-4 shows the twelve most abundant benthic taxa by zone. The NMDS ordination of associations over time had a 3-dimensional solution with a final stress of 13.231, and axis 1 and 2 accounted fo r 71% of variation (axis 1=0.440, axis 2=0.265). It revealed that May-June separated from the other sampling periods, likely due to its recent rewetting and low streamflows (Fig. 3-13A). Oli gochaeta and Diptera taxa associated with this early wet season sampling. The remaining peri ods largely grouped together, but several distinctions are evident. Str eam discharge and dissolved oxygen associated with October, along with taxa that require higher water quality, Farrodes and Leptohyphes. September, November,
108 and January grouped particularly tightly, with Caenis and Haplohyphes associating with these periods. Indicator ta xa were only found for the Open zone (Corixidae: IV=100, p=0.017; Tipulidae: IV=64.4, p=0.036) and Canal-influenced zone ( Asioplax : IV=72.8, p=0.017; Chironomini: IV=36.5, p=0.044). Stone macroinvertebrates The average B-axis length of stones sam pled for each zone over the study ranged between 13.53 + 2.31 cm and 14.35 + 2.57 cm. No significant differen ce between zones was found, but a slightly significant difference was found among stone sizes over time (F=2.56, p=0.042). The minimum B-axis length of sampled stones was 9 cm, which was greater than the maximum Baxis length of any stone encountered during benthic sampling, such that no overlap occurred in habitat sampling. For the entire study period, the Open zone stones had a much greater abundance of macroinvertebrates (2793) than other zones and the Canal-influenced zone had a substantially lower abundance (883) than other zones. Above and Below zones were relatively similar (1707 and 1346, respectively). Over time, abundances fluctuat ed for all zones, with distinct peaks in the Open zone in July and September (Fig. 3-6B). The dominant order of macroinvertebrates on stones was Diptera, particularly in the Open and Below zones (75% and 69%) compared to Above and Canal-influenced zones (43% and 37 %) (Fig. 3-7B). Ephemeroptera (Above: 34%, Open: 16%, Below: 17%, CI: 25%) followed by Trichoptera (Above: 12%, Open: 6%, Below: 10%, CI: 15%) were the next most common orders for each zone, with Ephemeroptera representing a much greater share of the macroi nvertebrates in the Above and Canal-influenced zones. Over time, Ephemeroptera and Trichopter a generally increased in representation as the wet season progressed then decreased with the onse t of the dry season in November in each zone (Fig. 3-9). However, the Above zone had some variability in proportions of aquatic
109 macroinvertebrate orders compared to other zo nes, and the Canal-influenced zone had much greater diversity of orders represented in Ma y-June, the beginning of the wet season, compared to the other zones in the seasonal stream sec tion, which were almost completely dominated by Diptera, Hydracarina and Oligochaeta. Chironomids were the dominant taxon in all zo nes as a whole. In the Above zone, the dominant taxa were Chironomini (14%) and Ort hocladiinae (14%), while Baetidae (13%) and Haplohyphes (12%) were the dominant Ephemeroptera. The Open zone was similar to the Above zone, but Chironomini ( 41%) and Orthocladiinae (28%) were particularly dominant, followed by Baetidae (5%) and Haplohyphes (7%) to a lesser extent In the Below zone, Chironomini (39%) and Orthocla diinae (22%) dominated, and Haplohyphes (7%) had the next greatest representation. In the Canal-influenced zone, Chiron omini (10%) and Orthocladiinae (18%) were less dominant, while various other ta xa, such as Hydracarina (13%), Baetidae (6%), Haplohyphes (7%), Farrodes (6%), and an early instar Trichopt era, Hydroptilidae sp. (4%), were notable. Functional feeding group dist ributions varied among zones (Fig. 3-10B). In the Above zone, the dominant group was collector-gathe rers-scrapers-shredders (34%), followed by collector-gatherers (17%), collect or-gatherers-filterers (17%), and predators (12%). In the Open and Below zones, the major functional feeding groups were collector-gat herers (44%, 40%) and collector-gatherers-scrapers-shredders (35%, 27%). Collector-gatherers-filterers was the next most important group (9%, 12%). In the Canalinfluenced zone, collec tor-gatherers-scrapersshredders dominated (31%), followed by a more ev en distribution of pred ators (16%), collectorgatherers (11%), collector-fil terers (12%), and collector -filterers-predators (14%).
110 Both total and average taxa ri chness of stone macroinvertebra tes were similarly greater in the Above zone (48 and 20.8 taxa) and Open zone (46 and 21 taxa) than the Below zone (37 and 16.2 taxa) and Canal-influenced zone (38 and 18 .2 taxa) (Table 3-2). Table 3-5 shows the twelve most abundant stone taxa by zone. The NMDS ordination of associations over time had a 3-dimensional solution with a final stress of 15.138, and the axis 1 and 2 accounted for 72% of variation (axis 1=0.308, axis 2=0.405). It revealed that stone macroinvertebra tes grouped together by time, similar to benthic macroinvertebrates (Fig. 3-13B). Dipterans, particularly Chironomids, and Oligochaeta associated with low flow periods of Ma y-June and January, as well as conductivity. Ephemeropterans, Trichopterans, stream discha rge, and dissolved oxygen associated with the high flow periods of September and October. Th e indicator taxa for each zone in the seasonal portion of Quebrada Barbudal were relati vely rare taxa. The Above zone had Alisotrichia (IV: 35.1, p=0.009), and the Open zone had Hydroptila (IV: 20, p=0.012) and Psephenidae (IV: 26.6, p=0.01), as indicator taxa. The Canal-influenced zone had two indicator taxa: Hydroscaphidae (IV: 30.2, p=0.003) and Neotrichia (IV: 35.1, p=0.005). Drift macroinvertebrates Drift sam ples from Above and Below zones were the focus of analysis. Drift samples taken from the Open zone were not analyzed due to the complete submergence of the nets throughout the study, whereas all other drift nets were at or above the water level. The complete submergence would have excluded macroinvertebrates drifting on the stream surface that nets in the other zones would have collected, potentially causing a significantly lo wer collection in the Open zone. Thus, the Above zone captured drift from above the deforested stream section and the Below zone drift net potential ly contained macroinv ertebrates in drift from the Above, Open, and Below zones. Any difference between Above and Below zone drift would be related to
111 habitat in the Open and Below zone. The Ca nal-influenced zone drift could include macroinvertebrates from the seasonal and perennial stream sections. Total abundance of macroinvertebrates in drif t was much greater in the Above (10,316) and Below zones (8123) than the Canal-influenced zone (5166), despite the latter having yearround flow. Macroinvertebrate assemblages in dr ift were very similar between the Above and the Below zones (Fig. 3-11). Ephemeropterans were dominant in the Above zone (39%) and Below zones (35%). In the Above zone, Trichopte ra was the next most abundant order (28%), followed by Diptera (20%). In the Below zone, Trichoptera and Diptera had similar representation (26% and 28%). Terrestrials comp rised only a small propor tion of organisms in drift in Above and Below zones (6% and 5%). In the Canal-influenced zone, the dominant orders were Diptera (37%), Ephemeroptera (25%), Trichoptera (15%), and Hydracarina (13%). Above and Below zones showed similar tre nds in macroinvertebrates assemblages over time (Fig. 3-12). Each was dominated by Diptera at the beginning of the wet season in MayJune (Above: 60%, Below: 45%). Hydracarina and Coleoptera had notable representation in May-June and after the wet season in November and January, when streamflow was comparable to low levels recorded in Ma y-June. Both Ephemeroptera and Trichoptera showed an increasing trend as the wet season progressed, with Trichopt era peaking in September (Above: 38%, Below: 48%) and Ephemeroptera peaking in mid-Novemb er (Above: 51%, Below: 64%). Drift samples could not be taken in October due to prohibitively high streamflow. However, these two orders remained dominant into the dry season in the Above zone, while they decreased dramatically in dominance in the Below zone. In the Canal-infl uenced zone, Ephemeroptera was the dominant order as a whole throughout the st udy and Trichoptera remained re latively stable, except for an increase in September. Diptera also spiked in dominance in Late-November.
112 The Above and Below zones had similar taxa representation. The dominant taxon in both zones was a Hydropsyc hidae grouping of Smicridea and/or Leptonema (21% and 19%). The Ephemeropterans Farrodes, Haplohyphes and Baetidae were other dominant taxa (Above: 12%, 12%, 7%; Below: 11%,11%, 6%). Chironomini was the dominant Dipteran in each zone (Above: 8% and Below: 13%). Representation of ta xa was more evenly distributed in the Canalinfluenced zone. Ephemeropterans were the dominant taxa with Haplohyphes (15%), Baetidae (11%), and Farrodes (8%), as well as the Hydropsychidae grouping of Smicridea and/or Leptonema (11%). Several functional feeding groups had high representation, an d their proportions were similar between the Above zone and Below z one. The dominant group was collector-filterers (25% and 23%), which corresponds with the hi gh relative abundance of Trichoptera. Other groups with considerable repres entation were collector-gatherers -scrapers-shredders (17% and 18%), collector-gatherers-filterers (17% and 16%), collector-gatherers (10% and 15%), and taxa within all functional feeding groups but predator s (14% and 14%). Predators and scrapers were represented to a lesser extent (8% and 7%, 7% and 4%). In th e Canal-influenced zone, the dominant groups were collector-g atherers-scrapers-shredders and collector-gatherers-filterers, both comprising 22% of total taxa. Predators a nd collector-filterers ha d the next greatest representation (16% and 14%). Scrapers and ta xa falling within all functional feeding groups except predators also had notable representation (11% and 9%). Both total and average taxa ri chness in drift were greatest in the Above zone (69 and 37.7 taxa), followed closely by the Canal-influenced zone (67 and 35.3 taxa) and the Below zone (63 and 29.8 taxa) (Table 3-3). Table 3-6 shows th e twelve most abundant drift taxa by zone.
113 NMDS ordination showed no clear distinction between zones or within zones over time. Only the Above and Canal-influenced zones had indicator taxa. Indicators in the Above zone were Alisotrichia (IV: 39.4, p=0.05) and Planorbidae (IV: 52, p=0.036). Indicators in the Canalinfluenced zone were the Trichopterans Macronema (IV: 80, p=0.005), Mayatrichia (IV: 51.9, p=0.028), Leucotrichia (IV: 54.2, p=0.006), and the Hemipteran family Veliidae (IV: 54.2, p=0.022). Taxa exclusivity Select taxa were found either exclusively in one habitat type or only in two of the three habitats (Table 3-7). Drift had the highest num ber of exclusive taxa, com prised primarily of Hemipterans, Philopotomidae, and rare taxa. The benthos had a divers e mix of orders, and stones had no exclusive taxa. Taxa found only in the benthos and stones were a diverse mix of orders. Only two Diptera pupae were found on st ones and in drift, but many taxa were only found in the benthos and drift, including severa l Hemiptera, Coleoptera, and Trichoptera taxa. Fish Seven species of fish were collected over the study: Astyanax aeneus (G unter 1860) Brachyrhaphis olomina (Meek 1914), Archocentrus nigrofasciatus (Gunter 1869), Poecilia gillii (Kner & Steindachner 1863), Synbranchus marmoratus (Bloch 1795), Rhamdia nicaraguensis (Gunther 1864), and Parachromis dovii (Gunther 1864). Total fish abundance was greatest in the Above zone (450, N=13) and decreased by zone downstream: Open (340, N=15), Below (135, N=15), then Canal-influenced (107, N=12) z ones (Fig. 3-14A). Differences in the number of samples (N) are the result of some minnow traps containing no fish. No minnow traps were placed instream in September and October due to prohibitively high stream discharge, which the minnow traps could not withstand. Fish abunda nce generally increase d as the wet season progressed then decreased again well into the dr y season in January (Fig. 3-14B). For all zones,
114 the greatest number of indivi duals was collected in late N ovember, following the month with highest streamflow, which likely allowed upstream migr ation of fish species. Overall, Astyanax aeneus was the most abundant fish in each zone (Above: 55%, Open: 60%, Below: 68%, CI: 75%) (Fig. 3-14B). The second most abundant fish was Brachyrhaphis olomina in the Above zone (29%), Poecilia gillii in the Open zone (30%), Archocentrus nigrofasciatus in the Below zone (21%), and A. nigrofasciatus in the Canal-influenced zone (18%). A shift in the dominant fish species occurred over time in all zones. In the Above, Open, and Below zones, P. gillii dominated in the early part of the wet season in May-June (75%, 68%, 56%) and July (52%, 82%, 60%) then A. aeneus dominated in Mid-November (53%, 86%, 76%), Late November (81%, 89%, 87%), and January (62%, 64%, 42%), after the highest periods of streamflow at the end of the wet se ason and the onset of the dry season. Brachyrhaphis olomina maintained a steady relative abundance in th e Above zone (June: 25%, July: 36%) until MidNovember, when it overtook P. gillii in dominance ( B. olomina: Mid-Nov: 38%, Late-Nov: 18%, Jan: 30%; P. gillii : Mid-Nov: 8%, Late-Nov: 1%, Jan: 6%), but had comparatively little representation in the Open or Below zones. Archocentrus nigrofasciatus only had notable representation in the Below and Canal-influenced zones, and it dominated the Canal-influenced zone (June: 67%, July: 40%) along with P. gillii (June: 33%, July: 40%) until Mid-November when A. aeneus became dominant (Mid-November: 91%, Late-November: 81%, January: 88%). Discussion Stream s are a mosaic of patches based on th e distribution of habitats, nutrients, and organisms (Pringle et al. 1988) This study centered on a patc h of stream with open canopy within an otherwise continuously forested ripari an zone, a patch created by the construction and maintenance of a large irrigation canal. Th is open canopy patch, or Open zone, created a different stream environment and biotic assemb lage than the forested stream reaches both
115 upstream (Above zone) and downstream (Below zone ). However, even these forested reaches differed based on their upstream and downstream re lationship to the Open zone, indicating that instream conditions and the biotic community in these zones were linked not only to riparian cover, but also to the environm ent within that zone and in the neighboring zone. In addition, abiotic and biotic characteristic s varied over time in all zones. At the same time, zones still shared many similarities. These findings exemplify the multi-dimensional linkages of patches, which include longitudinal linkages between ups tream and downstream, lateral linkages between the stream and riparian zone, and vertical linkages between the stream and hyporheic zone, as well as the influence of time on these linka ges (Ward 1989, Lake 2000). The multi-dimensional interactions among patches, or zones in this st udy, paint a complex yet re vealing picture of the dynamics of a tropical dry forest stream influenced by localized ri parian canopy removal. Riparian Canopy, Stream Lighting, and Primary Production Canopy cover had a clear influence on stream lighting and prim ary production. The Open zone had the lowest canopy cover and highest photosynthetically ac tive radiation values throughout the study, as well as greater levels of periphyton chlorophylla and AFDM, reflecting algal growth permitted by increased light availabi lity. Greater light penetration from less riparian canopy cover increases periphyt on chlorophyll-a (Larned and Santos 2000, Lorion and Kennedy 2009) and algal biomass (AFDM) (Hill and Knig ht 1988, Hetrick and Brusven 1998, Ambrose et al. 2004, Boothroyd et al. 2004, Quinn et al. 2004). Hi gher light irradiance in laboratory streams also resulted in grea ter algal biomass (DeNic ola and McIntire 1991). Periphyton chlorophyll-a and AFDM levels we re strikingly higher in May-June and January, both periods of lowest stream discharge. The lower stream discharge may allow more light to penetrate the water column and reach st one surfaces to enhance algal growth, and lower flows could also decrease disturbance of stone surfaces, allowing algae to remain attached and
116 grow (Allan 1995). Deposition of remains from observed floating algal mats during these periods also could have contributed to peak s in algal biomass on stones. Higher stream discharge during September and October would have decreased algal biomass on stones due to increased flows and less light penetration from greater water depths, and a lag time would likely exist for growth and accumulation of algae, expl aining the lack of signi ficant differences during these months. However, differences among zone s in periphyton chlorophyll-a and AFDM were usually not significant, likely due to variability existing amo ng the substrate samples. High variability was also found in other Costa Rican streams reaches with open canopies (Lorion and Kennedy 2009). Riparian Canopy, Physical-Chemical Conditions, and Streamflow Changes over tim e in the effects of canopy cover on physical-chemical characteristics in Quebrada Barbudal were largely mediated by seasonal streamflow patterns. Despite the difference in the light availability and primar y production between the Open zone and the other zones, physical-chemical parameters were only signi ficantly affected at the streamflow extremes. Stream water temperature was significantly greate r in the Open zone except during periods of high stream discharge, in dicating an effect of lack of canopy cover on the thermal regime of the stream throughout most of its period of flow. Riparian canopy cover, resulting from the width and density of streamside vegetati on, is one of the fundamental dete rminants of solar heat inputs into streams (Gregory et al. 1991), such that incr eased stream lighting a nd solar radiation from riparian canopy removal increase stream temp eratures (Hetrick and Brusven 1998, Johnson and Jones 2000, Quinn et al. 2004), affecting diurnal and maximum temp eratures (Boothroyd et al. 2004). Deforested stream reaches in southeastern Costa Rica also had significantly higher stream temperatures than forested sections, and stream temperatures varied over time, with the lowest values during periods of highest rainfall (Lorion and Kennedy 2009), patterns mirrored in
117 this study. Dissolved oxygen was si gnificantly greater in the Open zone during periods of lowest streamflow, suggesting a larger influence of algal growth on dissolved oxygen levels due to lower flows allowing greater accumulation. Mean while, conductivity was largely unaffected in this study and within the range of values in other streams in Costa Rica (Lorion and Kennedy 2009). These suggest that effect s of riparian canopy removal are primarily seen under low streamflow conditions. Streamflow is considered an environmental parameter capable of shaping and linking the biotic community structure of i ndividual patches within stream s (Malmqvist 2002). Patches in streams are strongly influenced by floods and dr oughts, with the latter potentially affecting longitudinal ecological processes, such as move ment of nutrients and biota, as well as stimulating algal blooms and increased competit ion and predation in more isolated, stressed conditions (Lake 2000). Deforestation of tropical wa tersheds with erosion prone soils increases stream sedimentation, which may also affect biota (Ramir ez et al. 2008). Sedimentation may increase in streams due to lack of riparian vegetation in creasing streambank erosi on (Quinn et al. 2004), as well as road run-off (Bilby et al. 1989). Although sedimentation was not measured in this study, the road crossing at the Open zone likely cont ributed to sedimentation in the Open zone and immediately downstream in the Below zone, wh ich had developed sandbars near the road. Aquatic Macroinvertebrates This study investigated aquatic m acroinvertebrates in the be nthos, on stones, and in drift both to gain a broader perspectiv e of assemblages occupying differe nt habitats in tropical dry forest streams and to determine how riparian removal at the local s cale may affect them. Information on multiple habitats in neotropical streams is considered vital to understanding macroinvertebrate community dynamics (Buss et al. 2004), and there are few studies on
118 macroinvertebrates in drift (Pringle and Ra mirez 1998, Ramirez and Pringle 1998a, Boyero and Bosch 2002, Callisto and Goulart 2005) and on stones (Collier 1995, Boyero and DeLope 2002, Boyero and Bosch 2004) in the tropics. The studies that exist revealed that different habitats within the same stream reach can contain different macroinvertebrate community co mpositions. Substrate type primarily structured the macroinvertebrates assemblage in Brazilian streams rather than environmental degradation, water quality, or season (Buss et al. 2004). Vegeta tion and cobble habitats in Zimbabwe rivers had distinct macroinvertebrate communities, and these two habitats supported a greater abundance and taxa richness than gravel and sand habitats, which was attributed to vegetation and cobbles offering more complex habitat for food and refuge (Chakona et al. 2008). Sampling drift and benthos provided more information on the macroinvertebrates community in an Indonesian stream than either alone (Dudgeon 2006). Drift samp ling revealed that shrimp constituted a major component of the stream community in the Caribbean slope lowland tropical streams in Costa Rica, which would have been missed by simply sampling benthos (Pringle and Ramirez 1998). Drift also added information on ta xa richness for those Costa Rican streams. Pringle and Ramirez (1998) reco mmended that drift sampling be included with benthic sampling in standard biological assessments of tropical streams. Similarly, this study found differences in macroinvertebrates assemblages based upon the habitat sampled. Benthic macroinvertebrates Benthic aquatic m acroinvertebrates in this study did show a response in abundance and community structure to the lack of canopy cover. Macroinvertebrate abunda nces were highest in the Open zone, consistent with other studies sh owing higher abundances or densities in stream reaches with decreased forest canopy cover (Haw kins et al. 1982, Paaby et al. 1998, Bojsen and Jacobsen 2003). In a light manipulation experi ment that held water temperature constant,
119 primary consumer abundance positively correlated with light and periphyton (Kiffney et al. 2004). Another light manipulation experiment compared periphyton and chironomids in New Zealand streams with 0, 60, 90, and 98% photos ynthetically active ra diation. Periphyton productivity progressively decr eased with increasing shade, and total invertebrate and chironomid densities decreased si gnificantly as shade increased from 60 to 90% (Quinn et al. 1997). The Open zone also differed from the other zones by greater Ephemeropteran dominance. Ephemeroptera were primarily Caenis and the Leptohyphidae genus Haplohyphes followed by Baetidae, Farrodes and Leptohyphes to a much lesser extent. Diptera was the next most abundant order, but considerably less represen ted. Tolerant and gene ralist Ephemeroptera families, such as Caenidae and Trichorythidae, were bioindicators of agricultural sites in Uganda (Kasangaki et al. 2008), along w ith Diptera, Hemiptera, and Co leoptera. Compared to other orders, Hemiptera and Coleoptera were not well-repr esented in this study, but despite their lack of representation, the Coleoptera family Corixidae was an indicator taxon in the Open zone. By affecting food resources, riparian vegetation also influences representation of functional feeding groups within the community (Gregory et al. 1991), which was seen in this study. The dominant functional feed ing groups in the Open zone were collector-gatherersscrapers-shredders and collector-filterers, followed by predators. Other zones had a lower representation of collector-gathere rs-scrapers-shredders but greater representation of predators. All zones lacked Plecopteran shredders and lack of shredders, such as these, is common in the tropics (Wantzen et al. 2008). Similar to this study, streams in Oregon with open canopies had higher abundances of collector-gat herers, filter feeders, and pr edators, as well as no notable difference in scrapers relative to canopy cover (Hawkins et al 1982). Generalist collector-
120 gatherers, mostly Ephemeroptera, also domina ted sites with less canopy cover in Madagascar (Benstead et al. 2003, Benstead and Pringle 2004), wh ich was attributed to greater algal growth present. Low levels of riparian deforestation in headwater streams in Georgia reduced consumer dependence on allochthonous resources, implicat ing canopy cover as a primary driver in food web changes through its influence on heterotrophy and autotrophy (England and Rosemond 2004). This helps explain the greater representati on of collector-gatherers -scrapers-shredders in the Open zone that are capable of consum ing both allochthonous and autochthonous food resources. However, changes in ripari an conditions and the food resource base only weakly affected the macroinvertebrate community and functional feeding guild organization in New Guinea streams (Dudgeon 1994). Community composition was similar across those New Guinea streams, irrespective of shading, and wa s dominated by Baetidae, Leptophlebiidae, Orthocladiinae, Elmidae, and Hydropsychidae. The findings from the Open zone in this study were not fully consiste nt with findings from other studies of streams lacking riparian ca nopy. Reduced canopy cover ove r coastal streams in British Columbia resulted in greater algal biomass and decreased total invertebrate and Ephemeroptera biomass, as well as decreased co mmunity diversity, compared to heavily shaded reaches (Kelly et al. 2003). Diptera, rather than Ephemeroptera, had highest abundances in pasture reaches in southeastern Costa Rica streams (Lorion and Kennedy 2009), and chironomids dominated New Zealand pasture streams after a spate (Collier and Quinn 2003). Non-forested sites had lower Ephemeroptera, Plecoptera, and Trichoptera (EPT) taxa richness in Brazil (Nessimian and Venticinque 2008) and showed d eclines in sensitive EPT taxa in Washington state, Madagascar, and New Zealand (Benstead et al. 2003, Qui nn et al. 2004, Jackson et al. 2007). However, this study reve aled no striking difference in taxon richness between zones,
121 similar to a study of Oregon streams that show ed no influence of canopy cover or substrate on taxa richness (Hawkins et al. 1982). Both macroinvertebrate diversity and collector density decreased with diminishing canopy cover in Ecuadorian Amazon streams, while macroinvertebrate density, periphyton biomass, and predator density increased (Bojsen and Jacobsen 2003), demonstrating some differences and similarities with this study. Tanypodinae positively related to canopy cover or litter detritus (Bojsen and Jacobsen 2003), which was also found in this study, with Tanypodina e representing only 5% of taxa in the Open zone versus between 13%-25% in the fo rested zones (Above: 20%, Be low: 25%, CI: 13%). Macroinvertebrates character istic of non-forested str eam reaches have included Chironomidae, Baetidae, Gastropods, and Oli gochaeta (Collier a nd Quinn 2003, Dudgeon 2006, Jackson et al. 2007, Lorion and Kennedy 2009), and deforested streams in Madagascar had a higher abundance and biomass of Ephemeroptera pr imarily due to the generalist Ephemeroptera families Baetidae and Leptohyphidae (Benstead et al. 2003). This study also found greater dominance of Baetidae a nd Leptohyphidae taxa ( Leptohyphes and Haplohyphes ) in the Open zone than other zones, but this zone did not have notable differences in Gastropods or Oligochaeta. The Open zone actually shared similari ties with macroinvertebrate communities characteristically found in forested streams, ofte n to a greater extent than the forested Above zone. Ephemeroptera dominate d forested streams in New Z ealand (Collier and Quinn 2003), which this study found for the Open zone, even more so than the forested zones. The Ephemeropteran families Leptophlebiidae and Baetid ae dominated forested sites, while several Trichopteran famlies (Leptoceridae, Hydropsychi dae, and Calamoceratidae), and Tipulidae were considered good bioindicators of forested st reams in Uganda (Kasangaki et al. 2008). Farrodes
122 (Family Leptophlebiidae) and Tricorythodes (Leptohyphidae) had higher abundances in forested streams than pasture streams in southeaste rn Costa Rica (Lorion and Kennedy 2009). While Farrodes was better represented in Above zone than the Open zone, the zones were not strikingly different. No clear dist inctions in Trichoptera represen tation emerged for the benthos. Tipulidae was actually an indicator group for the Open zone, suggesting good water quality despite lack of canopy cover. High species rich ness of EPT taxa and Diptera, consisting of shredders, collector-gatherers, a nd collector-filterers characterized Madagascar forested streams (Benstead et al. 2003). Collect or-filterers and th e generalist collector -gatherers-scrapersshredders dominated the Open z one, along with the other zones. Riparian vegetation and canopy cover influe nce benthic macroinvertebrate community composition by affecting instream conditions (Nessimian and Venticinque 2008) and allochthonous and autochthonous food resource s (Aguiar et al. 2002, Benstead et al. 2003, Benstead and Pringle 2004). Macroinvertebrate assemblage and abundance responses to riparian canopy removal have been attributed to subseque nt increases in periphyton biomass and water temperature resulting from great er stream lighting (Benstead and Pringle 2004, Quinn et al. 2004, Kasangaki et al. 2008). Di fferences in the macroinvertebrate community between the Open zone and the Above and Below zones most likely reflect exploitati on of more abundant autochthonous food resources in the Open zone, as shown by the generally greater levels of periphyton chlorophyll-a and AFDM, and possibly the greater temperatures in the Open zone. Although the Canal-influenced zone had pere nnial flow, macroinvertebrate relative abundance was not markedly different from other forested zones, indicating that the seasonal stream can develop a community similar to perennial streams, and a perennial stream can maintain a similar community to that found in s easonal streams. This was found also to be the
123 case for first-order tropical dry fo rest streams (Chapter 2). Over all, taxa richness was slightly less, which suggests that perennial streamflow may inhibit the pres ence of certain benthic taxa. However, the Canal-influenced zone did have tw o benthic indicator taxa whereas the forested zones did not; so the influence of perennial streamflow on taxa richness is unclear. Stone macroinvertebrates The Open zone had the greatest abu ndance of macroinvertebrates on stones. Diptera dominated in all zones, particularly in th e Open and Below zones, and Chironomini and Orthocladiinae were the domina nt Dipterans in all zones. Ephemeroptera had greater representation in the Above zone and the dominant Ephemeropter a in all zones were Baetidae and Haplohyphes Overall, relative abunda nce of the Canal-influenced zone most closely resembled the Above zone despite different str eamflow regimes and much lower total abundance in the Canal-influenced zone, and the Canalinfluenced zone had s lightly more diverse representation. The dominant f unctional feeding groups in the Above, Open, and Below zones were collector-gatherers and collector-gatherers-scrapers-shredders followed by collectorfilterers. The Canal-influenced zone differed in having a more even spread of functional feeding groups, reflecting the slightly more even distributi on of taxa. Despite this, overall taxa richness was lower in the Canal-influenced zone, along with the Below zone, indicating that a factor other than streamflow may influence taxa richness. Similar communities have been found in other systems where stones were sampled. Diptera and Ephemeroptera were the dominant orders of macroi nvertebrates that recolonized stones in Costa Rican tropical rainforest streams (Boyero and Bosch 2004). Simuliidae and Baetidae were the most abundant families that recolonized stones, followed by Chironomidae and Leptohyphidae. The primary stone recolonizers in tropical island streams in Panama were Ephemeroptera, Diptera, Trichoptera, Coleopter a, Lepidoptera, and Hydracarina. The dominant
124 Ephemeroptera families were Baetidae, Leptophl ebiidae, and Leptohyphidae, while the dominant Diptera families were Chironomidae, Simuliidae and Ceratopogonidae. Odonata, Oligochaeta, and Mollusca were not found on stones (Boyero and DeLope 2002). Aquatic insects in Caribbean slope Costa Rican streams effectively re duced algal biomass, even when grazing fish were present in moderate abundances. Grazing Ephemeroptera, particular ly Tricorythidae and Baetidae, and Chironomidae significantly redu ced periphyton abundance on stones in riffle habitats in Costa Rica (Barbee 2005 ). Other physical char acteristics of stones that influence the macroinvertebrate community density and richne ss included water depth (s trong negative effect), current velocity (positive effect ), horizontal position, embeddedne ss, and surface area (Jacobsen 2005). Canopy cover and water temperature we re among the factors that influenced Ephemeroptera, Plecoptera, and Trichoptera taxa richness on New Zealand stones, which led to the suggestion of retaining and planting trees in riparian areas to decrease water temperatures and light exposure to enhance macroinvertebrate biodiversity (Collier 1995). It should be noted that severa l characteristics of stones may affect their taxa richness and abundance. Increased surface area of stones has been related to increased taxa richness, with stones having larger surface areas supporting more large predators, likel y due to higher prey densities. This relationship has been attributed to enhanced ha bitat diversity on larger stones (Douglas and Lake 1994). However, stones were not significantly different in size among zones, so surface area likely did not influence differences in richness or abundance between zones. Stones with rougher texture also support higher species richness and abundance (Downes et al. 1995, Downes et al. 2000), and gr eater stone stability has been related to greater total macroinvertebrate density, taxa richness, and chir onomid densities after spate disturbances than unstable stones (Matthaei et al. 2000). Howe ver, densities of hydropsychids on fixed and
125 unfixed bricks were similar to each other and to natural stones, suggesti ng that stone stability does not always influence macroinvertebrate co mmunities (Bond et al. 2000). Stone texture and stability were not directly measured in this study. Drift macroinvertebrates The drif t community was very similar upstr eam and downstream of the Open zone, and was dominated by Ephemeroptera, Trichoptera, and Diptera. The Ephemeroptera consisted mainly of Haplohyphes Farrodes and Baetids. The Trichoptera were primarily Hydropsychidae taxa ( Smicridea or Leptonema ) and Neotrichia (Hydroptilidae). Chironomini and Orthocladiinae constituted most of the Diptera. Above and Be low zones had similar trends over time, as well, until the beginning of the dry season when Ephemeroptera remained dominant in the Above zone, while Diptera became dominant in the Below zone. Divergence in community assemblage for the Above and Below zones in late-November and January likely reflects conditions within the Below zone rather than taxa drifting downstream and conditions in the Open zone, since Diptera was more dominant in th e benthos in the Below zone th an in the zones upstream and Diptera representation on stones was similar am ong the Above, Open, and Below zones. In contrast, the Below zone had much greater Dipt era dominance for both substrates than the Above and Open zones, which likely explains the greater representation by Diptera in drift in this zone. The Canal-influenced zone also had greater dominance of Diptera, as well as larger representation of Coleoptera and Hydracarina. Drift tota l abundance was lower, yet taxa richness similar, compared to other zones, suggesti ng that seasonal streams are able to develop macroinvertebrate communities similar to perenni al streams, with only small differences in dominant taxa. Other drift studies have revealed sim ilarities and differences in dominant macroinvertebrate orders and taxa compared to this study. As in this study, dominant taxa in
126 Caribbean slope tropical streams in Costa Rica have included Ephemeroptera of the families Baetidae and Leptohyphidae, Dipt era in the family Chironomidae, and Trichoptera in the families Hydropsychidae and Hydroptilidae (P ringle and Ramirez 1998, Ramirez and Pringle 1998a, 2001, Boyero and Bosch 2002). Larval shrimp, Simuliidae, and Elmidae were also major components of those streams, but absent or not well represented in drift in this study. Larval shrimp have been found in low-order mountain streams and have dominanted drift more than aquatic insects in lowland forest streams draining the Caribbean slope of Costa Rica (Pringle and Ramirez 1998, Ramirez and Pringle 1998a), which starkly contrasts with this study, since no larval shrimp were collected in any zone. Research on the drift community in southeastern Brazilian streams revealed that Chironomidae and Ephemeroptera represented 80% of total density of macroinvertebrates in drift. Drift sampling in these Brazilian streams also allowed collection of taxa not normally sampled in the benthos, such as Veliidae and Gerridae taxa, and added new or rare taxa to the macroinvertebra te collection, as well as additional knowledge of Chironomidae pupae (Callisto and Goulart 2005). This study also had high representation of Chironomidae and Ephemeroptera. Furthermore, the only two macroinvertebrates ex clusively found in drift in this study were Chironomidae pupae, and several Hemiptera famili es, among other taxa, were exclusively found only in the benthos and drift (not on stones), s upporting the benefit of drift collection noted by Callisto (2005). Canopy cover had an indirect ne gative effect on macroinvertebrate drift density in Ecuadorian Amazon streams due to increased ripa rian cover associated w ith a decline in fish and benthic macroinvertebrate abundance (Jacobsen and Bojsen 2002). Drift in these streams was dominated by Leptophlebiidae, Baetid ae, Tricorythidae, Hydropsychidae, and Chironomidae.
127 Only a few studies have compared the drift macroinvertebrate communites to those of the benthos. In this study, drift showed little difference in composition to the benthos, which was also the case for streams in New Zealand (C ollier and Quinn 2003) and Indonesia, but agricultural streams showed greater differe nces (Dudgeon 2006). Simuliidae and Baetidae, followed by Leptohyphidae and Chironomidae, were the most abundant families that recolonized stones in Costa Rican streams on the Caribbean sl ope, and these same families dominated drift except that Chironomidae was less re presented likely due to the larg e mesh size of the drift nets (Boyero and Bosch 2004). Benthic and drift taxa from forested sites in New Zealand were dominated by Ephemeroptera af ter floods, while pasture sites were dominated by Chironomids, terrestrials, Hydroptilidae, and Molluscs (Colli er and Quinn 2003). Although this study may not have always had the same taxa found in other st udies, it did show similarities between the drift and benthic macroinvertebrate communities like other studies. Overall lack of differences between the drif t community in the Above zone and the Below zone suggest that characteristic ta xa from the Open zone generally either did not enter drift or did not drift far. Mean drift distan ces are generally short, 10-20 m at moderate current speeds (Allan 1995), but distances traveled can vary based on length of larval life span considered (Malmqvist 2002) and seasonal streamflow (Jackson et al. 19 99). Jackson, McElravy & Resh (1999) found that most larvae drifted less than 4 m in the dry season and no more than 222 m in the wet season, although 18 m was the median dispersal distance during th e wet season. Drift distance was limited by reduced flows in streams in New Zeal and, and few larvae drifted more than 1 to 2 m under reduced flows. (James et al. 2009). The amount and arrangement of obstacles and dead water zones (backwaters and pools) also decrease drift distance due to forced lateral movement and settling (Bond et al. 2000).
128 Overall As a whole, the zones had very few indicator taxa, and ordinations did not reveal distinct groupings by zone, demonstrating the large degree of taxonom ic overlap among zones. This does not seem related to any shortfa ll in the power of the analysis due to the level of taxonomic identification, given that ordination of family-lev el data was considered a promising method for rapid bioassessment of stream conditions in S outheast Asia and had a dvantages when dealing with a fauna that has received limited taxonom ic study (Dudgeon 2006). Instead, the similarity in taxa among zones despite differences in physi cal-chemical parameters likely reflects a linkage among macroinvertebrate communities through upstream movement by flying adults or downstream drift. These linkages would aid ta xa resiliency after less optimum conditions or disturbances occur within a given zone and they demonstrate aspects of patch dynamics (Townsend 1989). Floods may enhance downs tream transport links, while droughts may fragment stream continuity, but both have the ability to destro y and create abiotic and biotic patches in streams (Lake 2000). Over time, benthos, stones, and drift showed similar succession trends among all zones, which mirrored streamflow patterns, as shown by taxa total abundances relative abundances, and ordinations. The increase in total abundance in each zone in late-November and January was likely related to a decrease in storm related sp ates and high flows from the wet season, with the large increase in January likely reflecting a lag time in recruitm ent, hatching, and development since the end of the wet season. As far as general trends in relative abundan ces for all the zones, Diptera and Oligochaeta dominated in the early wet season, Ephemeroptera and Trichoptera grew in dominance as the wet season progressed then decreased in the dry s eason, when Diptera regained dominance. The increasing trend in Ephemeroptera and Trichoptera over the course of the wet season in the three
129 adjacent zones of the seasonal stream section (Above, Open, & Below zones) of Quebrada Barbudal corresponds with increases in stream di scharge and flow persistence, as well as increases in dissolved oxygen, over the course of the wet season. The decrease in dominance of these orders in the Above and Below zones in No vember and January reflect decreases in stream discharge and dissolved oxygen. The maintenanc e of Ephemeroptera dominance in the Open zone could be attributed to the higher average di ssolved oxygen levels in the Open zone resulting from a burst of algal primary production related to a decrease in streamflow and water depth. Greater availability of algal biomass as a f ood source in the Open zone may also have contributed to the presence of more Ephemeropt era. The similar trend found between the zones in the seasonal stream section and the Canal-in fluenced zone in the perennial stream section demonstrated that the perennial stream was still in fluenced by the same aquatic insect cycles of recolonization, colonization, and development as the Above, Open, and Below zones in the seasonal section of Quebrada Barbudal. However, the trend in the Canal-influenced zone was not as pronounced as the other zones, likely due to its perennial flow regime. The trend of colonization and succession in these zones also mirro rs patterns observed in first-order tropical dry forest intermittent streams (Chapter 2). Dominant taxa differed based on the habita ts sampled. Ephemeroptera, particularly Caenis and Diptera dominated the benthos Drift was dominated by these, as well as Trichoptera. Stones were dominated by Diptera, followed by Ephemeroptera. Interestingly, Caenis was not notably represented in drift or on stones compared to the benthos. These di fferences likely relate to life history characteristics and feeding guilds, and support the value of sampling multiple habitats. Streamflow influences macroinvertebrate co mmunities in other studies, as well. Along with canopy cover, the average number of flowle ss summer months was a significant predictor
130 of variations in macroinveterbrate assemblages in streams in Portugal (Aguiar et al. 2002). Intermittent streams had greater taxa richness th an ephemeral streams with shorter flow periods, indicating the effects of flow permanence on the macroinvertebrate community in Zimbabwe streams (Chakona et al. 2008). Substrate disturbance from str eamflows and periphyton abundance influenced macroinverte brate diversity in New Zealand forest streams, with species richness decreasing as disturba nces increased and increasing as periphyton abundance increased when left undisturbed by stream flow (Death 2002). Macroinverte brate community structure and functional feeding groups in Ecua dorian highland streams differed in wet and dry seasons, with the dry season having higher numbers of individual s and species due to the unstable environment in the wet season related to stream discha rge (Jacobsen and Encal ada 1998). During both seasons, Baetidae, Simuliidae, Chironomidae, and Elmidae were most common. Collectorgatherers were most common with filterers, predators, and shredders of lesser importance. Macroinvertebrate consumers in Oregon stream s were influenced primarily by autotrophic production, even when allochthonou s detritus was abundant, and st ream current had a greater impact on macroinvertebrates abundance than the quality of detr itus, with the abundances of scrapers, shredders, and filter feeders increasing as current velocity increased (Hawkins et al. 1982). Collectors and filte r-feeders appear to play a larger role in th e processing of organic matter than shredders in aseasonal, flashy tropic al streams due to spat es flushing out intact leaves, leaving only finer particles instream (Wantzen et al 2008). However, recent studies on Hong Kong streams suggest that algae could be either a more important food resource than detritus, or nearly as important, in small tropical streams (Salas and Dudgeon 2001, 2003, Lau et al. 2009). Growth rates of Ephemeroptera were enhanced by algal food, and the importance of
131 algal food grew in the dry season when spates stopped depleting alga l standing stocks. The trophic base of mayflies was auto chthonous, but detritus was more important in the wet season. Thus, both canopy cover and streamflow influence patchiness in streams. Each stretch of stream is patchy on some scale due to particular disturbances, colonizers, biotic interactions, and physical characteristics of that patch (Townsend 1989). Disturbances can be characterized as pulses, presses, and ramps. Pulses are considered short-term and concentr ated events, like floods. Presses are typically characterized by quickly aris ing disturbances that re ach a constant level, often as a result of human ac tivities (Lake 2000), such as ch annelization and deforestation. Ramps are disturbances that incr ease in intensity over time, such as drought or the spread of invasive species. The type of disturbance can depend on the biota in question (Lake 2000). At the same time, these linkages between pa tches due to streamflow did not seem to supercede the influence of the open canopy on the macroinvertebrate assemblage during most of this study, as shown by the Above and Below z ones having more similar macroinvertebrate relative abundances than with the Open zone. The Below zone showed no clear intermediate community relative to Above and Open zones that would indicate export of conditions or taxa. In the absence of substrate distur bances, resource levels most likely determine species richness once they are established by immigration (Death 2002). Fish Astyanax aeneus was the m ost abundant species in each zone. Species had access to each zone, and the zones were immediately adjacent to one another, except for the Canal-influenced zone, which was approximately 100 m downstream of the Below zone. Astyanax aeneus took advantage of high stream discharge in Sept ember and October to migrate upstream then dominate the fish community in each zone, demons trating their tolerance of high streamflow and of varied habitat conditions therea fter. However, the other species seemed to select for habitats
132 since the second most relatively abunda nt fish differed among zones: Brachyrhaphis olomina in the Above zone, Poecilia gillii in the Open zone, and Archocentrus nigrofasciatus in the Below and the Canal-influenced zones. The Above zone had a forested riparian area, the shallowest average depth (16 cm), second smallest average wi dth (350 cm), and primarily stony substrate. The Open had a deforested riparian zone, the second shallowest average depth (24 cm), second greatest average width (411 cm), a nd substrate of stones and fine sediment. The Below zone had a forested riparian area, the greatest average de pth (30 cm), greatest average width (460 cm), and more fine sediment deposits than other zones. Th e Canal-influenced zone had a forested riparian area, the second greatest average de pth (28 cm), smallest average width (219 cm), and mixture of stone and fine sediment substrate. Habitat unique to each zone may have infl uenced the distribution of species within Quebrada Barbudal, but exactly what fact ors are most important is unclear. Habitat can have the greatest influence on fish distribution and community structure, with mesohabitat characteristics, such as water depths and velocity, having the next greatest influence (Arrington and Winemiller 2006). Water depth influenced presence and distributi on of fish species in New York (Singkran and Meixler 2008), West Africa (Kouame et al. 2008), Texas (Ostrand and Wilde 2002), and Mediterranean streams (Moran-Lopez et al. 2006), where habitat size (d epth) had the greatest influence on species distribution in assembla ge structure during stressful summer months. However, differences in stream depth among zones were not particularly great. Stream width does not appear to be a major factor since A. nigrofasciatus was the second most abundant fish in both the narrowest and widest zones. No clear preference for a particular substrate was evident either, since substrate type did not differ greatly between the Above, Open, and Canal-influenced zone, such that no striking difference in fish relative abundances would be
133 expected among these zones. The Below zone cl early had more fine sediment depositions, but that did not appear to influence A. nigrofasciatus. Changes in instream thermal regime can influence fish distribution and survival (Pus ey and Arthington 2003), but given the wide tolerance range of the species pr esent, temperature was unlikely the primary influence. While streamflow permitted colonization of the seasonal stretches of Quebrada Barbudal, it did not seem to play the greatest role in fish dist ribution and abundances, since the Below and Canalinfluenced zones had nearly identica l total and relative fish abundances. Canopy cover could have influenced fish di stribution. Lack of canopy cover may have caused P. gillii to prefer the Open zone. Light, peri phyton chlorophyll-a, and AFDM were much greater in the Open zone than other zones, providing P. gillii with a greater quantity of key elements of its diet: filamentous algae and ooze. The lack of P. gillii in the Below compared to the Open zone, suggests that the Open zone did not export conditions or foods favored by P. gillii. Open canopy stream reaches enhance fish densities and biomass (Murphy et al. 1981, Wilzbach et al. 2005), as well as enhance growth rates of most age classes (Wilzbach et al. 2005) and prey capture for some species (Wilzbach et al. 1986). These may have been factors that contributed to dominance of P. gillii in the Open zone. However, fish diet and the influence of light on prey capture were not directly investig ated in this study. The effects of canopy cover were species and age-class depe ndent in Alaskan streams (Keith and Bjornn 1998). A study of Southern Appalachian stream fish communities sugge sted that streams with reduced forest cover have lower fish assemblage structural and func tional diversity and/or ri chness (Burcher et al. 2008). Food availability likely influenced fish distribution, as well. Total abundance of macroinvertebrates in the benthos and on stone s combined was much greater in the Above
134 (8,970) and Open (11,481) zones compared to th e Below (5635) and Canal-influenced (4372) zones. When adding drift to the total, the A bove zone had a substantia lly higher abundance of macroinvertebrates (19,286) than Below (13,758) and Canal-influenced zones (9538). Thus, Above and Open zones would provide more food for the insectivorous B. olomina, which may have caused the species to have greater relative abundances in these two zones. In terms of total fish abundances, the size of deforested patches is influential. As the length of the riparian removal upstream increased in Appalachian streams, fish abundance immediately downstream decreased, but assemblage diversity did not change significantly. This suggested that riparian deforest ation less than 1 km in length ma y only create minor disturbances in fish communities if the watershed upstream is largely forested (Jones III et al. 1999). In another study, riparian canopy removal along a 100 m stretch of stream increased salmonid biomass, density, and growth, except for young-of-the-year fish that likely had trouble competing with older fish for resources (Wilzbach et al. 2005). In Washington, increased primary production and slightly elevated temper atures from open canopy, created by a clear-cut, were suggested as causes of increased fish production (Bilby and Bisson 1992). Since the Open zone in this study had intermediate fish abunda nces between the forested Above zone and the Below and Canal-influenced zones downstream, th e size of the Open zone was unlikely a major factor influencing fish species abundances and distribution. Although minnow traps are capable of providing information about fish populations, it should be noted that they also have two notable li mitations. The first limitation is the size of fish that can enter the trap, but visual observation re vealed that very few fish would be excluded based on size. The second limitation is the behavior of fish, particularly convict cichlids, which are territorial (Wisenden 1995), and preferred defe nding their territory rath er than explore and
135 enter minnow traps. However, this behavior woul d be the same across zones, so lower collection would have occurred in every zone. Conclusion Stream communities can be affected by human disturbances, such as impacts from climate change, forestry practices, and construction (Ma lmqvist 2002). Effects from canal management were seen in this study, as shown by changes in abiotic characteristics and biotic community structure of stream zones due to changes in ri parian cover and streamflow. Although the effects may be detrimental to some biota, differences be tween zones also created a greater diversity of habitats that benefited certain macroinvertebrate taxa and fish species, as demonstrated by the presence of indicator taxa in particular z ones and differentiation of fish species relative abundances among zones. In spite of effects from canal management, zones still showed similarities in macroinvertebrate taxa richness and relative abundances, as well as fish species richness, suggesting that the effects may be limited. Riparian restoration has been s uggested to decrease the negati ve effects of deforestation on endemic stream taxa (Benstead et al. 2003, Benstead and Pringle 2004, Urban et al. 2006, Ramirez et al. 2008). While that may be an appropr iate action for larger scale deforestation, this study suggested that loss of riparian canopy cover of less th an 35m may not have severe detrimental effects on the stream community. The small deforested patch in this tropical dry forest stream may actually provide greater diversity of niches, resour ces, and habitats for biota to exploit, influencing their distri butions, abundances, and presence in ways that maintain, if not enhance, aquatic macroinvertebrates taxa richness and fish species richness. The perennial discharge into Quebradal Bar budal from the canal may also support a source of colonizers for the seasonal stream reaches an d may provide refuge for macroinvertebrates and fish in the dry season. The primary concerns as sociated with direct canal discharge are likely
136 occasional floods from excess canal flows and potential introduction of species from other regions of Costa Rica into these streams.
137 Figure 3-1. Riparian Patch Study site map.
138 A B Figure 3-2. Photos of Open zone. A) Invert ed siphon at Canal Oeste and Quebrada Barbudal crossing, and B) Deforested Open zone.
139 Figure 3-3. Aerial view of sampling site showing sampli ng lengths and site dimensions.
140 A B Figure 3-4. Stream light and primary production averages. A) Canopy cover, B) Photosynthetically active radiation (PAR), C) Periphyton chlorophyll-a, and D) AshFree Dry Mass (AFDM). CI=Canal-influenced.
141 C D Figure 3-4. Continued
142 A B Figure 3-5. Physical-chemical ch aracteristics over time, by zone. A) Stream discharge, B) Water temperature, C) Dissolved oxygen, and D) Conductivity. CI=CanalInfluenced.
143 C D Figure 3-5. Continued
144 A B Figure 3-6. Total macroinverteb rate abundance over time. A) Benthos, and B) Stones.
145 A B Figure 3-7. Macroinvertebrate relative abundance, entire study, by zone. A) Benthos, and B) Stones.
146 A B Figure 3-8. Benthic macroinverteb rate relative abundance, by sampling round. A) Above zone, B) Open zone, C) Below zone, and D) Canal-Influenced zone.
147 C D Figure 3-8. Continued
148 A B Figure 3-9. Stone macroinvertebra te relative abundance, by samp ling round. A) Above zone, B) Open zone, C) Below zone, a nd D) Canal-Influenced zone.
149 C D Figure 3-9. Continued
150 A B Figure 3-10. Functional feeding group relative abundance, by zone. A) Benthos, and B) Stones. Coding: PRED=Predators; SC=Scrapers; CG=Collector-gatherers; CG-SH, CGSC=Collector-gatherers-scrapers-shredde rs; CG-CF=Collector-gatherers-filterers; CG,CF,SC,SH=Collector-gatherers-filterers -scrapers-shredders; ALL=All feeding modes; and OTHER=FFG with le ss than 5% representation.
151 Figure 3-11. Drift macroinvertebrate re lative abundance, entire study, by zone.
152 A B Figure 3-12. Drift relative abundance, by order. A) Above zone, B) Be low zone, and C) CanalInfluenced zone.
153 C Figure 3-12. Continued
154 Table 3-1. Taxa richness, by zone : Benthos. CI=Canal-Influenced.
155 Table 3-2. Taxa richness, by zone : Stones. CI=Canal-Influenced.
156 Table 3-3. Taxa richness, by zone : Drift. CI=Canal-Influenced.
157 Table 3-4. Top twelve most a bundant taxa, by zone: Benthos.
158 Table 3-5. Top twelve most a bundant taxa, by zone: Stones.
159 Table 3-6. Top twelve most a bundant taxa, by zone: Drift.
160 Table 3-7. Taxa exclusivity, by habitat.
161 A Figure 3-13. Relatived ordinations, by sampling round. A) Benthos, and B) Stones. Coding: ALISOTR=Alisotrichia, BAET=Baetidae (NonBaetodes), BAETOD= Baetodes, CAENIS=Caenis, CERATO=Ceratopogon iinae, CHRMINI=Chironomini, DO_mgL=Dissolved oxygen (mg/l), FARRODES= Farrodes HAPLO= Haplohyphes LEPTOHY= Leptohyphes LEUCOTR=Leucotrichia, LperS=Discharge (l/s), OLIGO=Oligochaeta, ORTHOCL=Or thocladiinae, PHYSID=Physidae, SMI_LPT= Smicridea/Leptonema TANYPOD=Tanypodinae, TDS=Total Dissolved Solids (TDS), and uS_cm=Conductivity (uS/cm).
162 B Figure 3-13. Continued
163 A B Figure 3-14. Fish abundance. A) By zone and B) Over time. CI=Canal-Influenced.
164 CHAPTER 4 CANAL CONTINUUM STUDY Introduction W ater resources development is common in semi-arid regions and such projects may include reservoir storage, small farm dams, groundwater pumping, and inter-basin water transfers (Hughes 2005). One such semi-arid region is the province of Guanacaste in northwestern Costa Rica, which has fertile soils, but experiences distinct wet and dry seasons that limit agricultural production. Irrigation water initially came from river withdrawals, but supplies were limited and variable (Pineda 2007) To expand agriculture and improve water supply dependability, the government began an irrigation canal system called the TempisqueArenal Irrigation Project (PRAT) in 1980, whic h conveys water from Lake Arenal in the highlands to agricultural lowlands of Guanacas te (Pacific 2002). This water transfer and subsequent conversion of wetlands and tropical dry forest enab led agricultural expansion, such that approximately half of all rice and sugar cane produced in Costa Rica comes from the region (Jimenez et al. 2001). The PRAT irrigation can als now crisscross much of Guanacaste, but neither the structure and function of the freshwat er ecosystem potentially created by these canals nor how the canals may contribute to regional freshwater divers ity have been investigated. Many agricultural water dist ribution and drainage systems can support fish and macroinvertebrates (Katano et al. 2003, Rizo-Patrn 2003, Hicks et al. 2006, Herzon and Helenius 2008), or convey water to fields that support fish (K atano et al. 2003) and macroinvertebrates (Leitao et al 2007, Wilson et al. 2008). Ditche s provide valuable habitat for aquatic and terrestrial taxa, incl uding rare species or species not typically found in agricultural habitats. They also provide resources in pl aces that otherwise would be dry, and enhance connectivity within the lands cape (Herzon and Helenius 2008). Ditches in England support
165 aquatic biodiversity, including uni que taxa, due to such factors as a unique physical-chemical environment, slow flow, connection to a major river, and structural complexity of riparian vegetation (Armitage et al. 2003). Although d itches had the least specie s-rich habitat compared to rivers, streams, and ponds in Britain, they still constituted valuable habitat that supported uncommon species and contributed to overall macroinvertebrate diversity (Williams et al. 2003). Williams (2003) defined ditches as man-made ch annels created for agricultural purposes that typically have a linear planform, follow field boundari es that involve right angle turns, and show little relationship to the contours of the natural landscape, which is a description that corresponds with the canals in this study. Aquatic insects supported by ditches, particularly Odonata and Chironomidae, constituted a major component of bi rd diets upon reaching the adult stage in the United Kingdom, further increasi ng the habitat and biodiversity va lue of ditches (Bradbury and Kirby 2006). In addition to providing important habitat, canals and ditches are considered good model systems for research and ecology as they are ge nerally abundant, contain a gradient of clearly delineated and accessible habitats, support relatively simple commun ity structures compared to natural systems, and can be easily altered by hum an actions, which make it easier to evaluate how environmental factors affect the aquati c community (Chlyeh et al. 2006, Herzon and Helenius 2008). Adjacent to Palo Verde National Park are ri ce and sugar cane fields with an irrigation water distribution system fed by Canal Oeste, a ma jor canal within PRAT. This system provides habitat and has very clear changes in physical features along its gradie nt, making it a useful subject for testing both predic tions of structural and functio nal succession of the River Continuum Concept (RCC) and for exploring biodive rsity within irrigation canals in the tropics.
166 The RCC states that physical parameters displa y a predictable gradient of conditions from a river headwater to its mouth, which include stream width, de pth, velocity, discharge, and temperature. The conditions along that gradient in turn influen ce the biotic community. Stream sizes, or orders, each have thei r own predictable set of conditions and biotic assemblages. Loworder streams are considered allochthonous driven with relative ly low biological diversity and are dominated by shredders and collectors that gath er or filter particulate organic matter. Midorder streams are considered au tochthonous driven with high di versity and dominated by grazers and collectors. High-order streams (rivers) ar e considered allochthonous based with lower diversity than mid-order streams and dominated by collectors. Predators change little with stream order (Vannote et al. 1980). The RCC has been refined further to address human regulation of rivers via the Serial Discontinuity Concept (Stanford and Ward 2001) multiple spatial dimensions and a temporal dimension (Ward 1989), and regional and local e nvironmental influences, such as climate, geology, riparian conditions, tr ibutaries, and geomorphology (Minshall et al. 1983). Few investigations into applic ability of the RCC have occurred in the tropics. In Puerto Rico, the RCC prediction that changes in stream size determine changes in basal resources and consumers generally held (Greathouse and Prin gle 2006), and algal resources were found to be very important to stream consumers along the l ongitudinal stream gradie nt, even in forested headwaters, which was attributed to high algal availability and al gal assimilation efficiency of shrimp (March and Pringle 2003). In Bolivia, altitude in combination with longitudinal stream gradient position determined func tional feeding group structure in streams (Tornanoval et al. 2007). In Brazil, macroinvertebrate richness on ly weakly followed a hump-shaped curve along the stream size gradient, because smaller streams (1st and 2nd-order) had greater taxa richness
167 than mid-sized streams (3rd-4th order) (Melo and Froehlich 2001). However, species richness and composition strongly correlated with stream size, indicating that both were related to characteristics of the physical habitat and presence of fine sediments on rocks. In canal systems, predictions of the RCC woul d be expected to be reversed in accordance with channel size. Like stream systems, irriga tion water distribution chan nels change physically in width and flow from source to outlet. Unlik e natural watersheds, irrigation channels start large and become increasingly smaller heading down stream to the fields, resulting in a reversed channel size configuration (Fig. 4-1). This ra ises the question: Does the irrigation water distribution system near Palo Verde National Pa rk follow the tenets of the River Continuum Concept, but in reverse, starting large and becoming smaller? Specifically, does macroinvertebrate community structure change acco rding to such physical variables as flow and channel size, and if so, how does it change? Answering these questions contributes to the understanding of the effects of physical variables on biotic components in streams, as well as awareness and management of biodiversity in irrigation canals in the tropics. Methods Study Area Sa mpling occurred in an irrigation canal system located in agricultural lowlands between Palo Verde National Park and a range of footh ills within the Lomas Barbudal Reserve in the Guanacaste province of Costa Rica (Fig. 4-2). Wate r for the irrigation canal system comes from the principle irrigation canal for the region, Cana l Oeste, which runs along the base of these foothills, and irrigation water is distributed through a series of laterals th at support rice and sugar cane production.
168 Sampling Strategy Irrigation water d istribution laterals coming o ff of Canal Oeste were sampled according to their stream order, which was determined by their size and position within the water distribution system. The first la teral off Canal Oeste was termed the Secondary canal (2), and it ran perpendicular to the agricultura l fields (Fig. 4-2). The next smallest lateral, which came off the Secondary canal, was termed the Tertiary canal (3), and it ran parall el to the agricultural fields. The smallest lateral, which came off th e Tertiary canal, was term ed the Quarternary canal (4), and it ran between fields. Sampling occurred in July 2007. Three sample s were taken at 5 m intervals in three different sections of the only Secondary canal, which had no repl icates of similar size and position in the water distribution system. Three samples were taken at 5 m intervals in four separate Tertiary canals, each coming off the Sec ondary canal at different points. Three samples were taken at 5 m intervals from four separate Quarternary canals, ea ch coming off a sampled Tertiary canal. Field Sampling Physical-chemical For channel dim ensions, channel width was collected during during streamflow measurements and considered eq uivalent to bankfull width, and channel depth was measured at bankfull. Canal slope was determined using a laser level. Dissolved oxygen (% and mg/l), water temperature (C), and conductivity (uS/cm-adjus ted for 25C) were measured using a YSI 556 Handheld Multiparameter Instrument (YSI, Inc. Yellow Springs, OH, U.S.A.). Photosynthetically Active Radiation (PAR) was measured usi ng a LICOR-190 (terrestrial PAR meter) and LICOR-192S (underwater PAR meter) (L i-COR, Inc., Lincoln, NE, U.S.A.). Canal water velocity was measured using a Marsh-Mc Birney flowmeter at three equidistant points
169 across the width of the stream channel. In str eam depths less than 20 cm, velocity was measured at 60% of the depth from the stream surface. When depth was 20 cm or greater, velocity was measured at 20% and 80% of the depth from the stream surface and then averaged. Discharge was calculated using the veloci ty, depth, and channel width m easurements. All data were collected in the daytime. Biological Benthic m acroinvertebrates were semi-quantitatively collected using a 243 um dip net to make a 20-second sweep of the canal bottom. All macroinvertebrate samples were transferred to a labeled plastic bag and preserve d with ethanol in the field. Algal biomass was estimated as both peri phyton chlorophyll-a and Ash Free Dry Mass (AFDM), but periphyton chlorophylla analysis only estimates algal biomass whereas AFDM includes both algal biomass and a mixture of bi omasses from bacteria, fungi, small fauna and detritus. Periphyton chlorophyll-a was collected fr om the surface of three tiles placed in each canal type for 29 days. Tiles were placed on the bottom of Quarternary canals and on a submerged portion of the canal wall in Tertiary and Secondary cana ls. Tiles were removed from the stream while using a Surber sampler net to collect any di slodged macroinvertebrates. A hollow rubber disc with internal area of 24.6 cm^2 was then placed on each tile to isolate the sampling area and to prevent loss. Periphyton within the rubber di sc area was scraped from the tiles using a nylon bristled brush. Next, the pe riphyton collected on the brush was cleaned in 50 ml of water, creating a concen trated periphyton solution, which was then stored in the dark inside a cooler containing cold packs. The procedure employed to collect periphyton chlorophyll-a was repeated on a different part of the same tile for AFDM analysis (Steinman and Lamberti 1996).
170 Laboratory Methods Benthic m acroinvertebrate samples were put through 1 mm and 250 um sieves in the laboratory. Both size classes were exam ined under a dissecting microscope, and macroinvertebrates were removed and placed in to dated and labeled vials containing 70% ethanol. Samples were then shipped to the United States for identification. Macroinvertebrates were generally identified to Family or Genus, except for the families Chironomidae and Ceratopogonidae, which were identified to Sub-Fam ily. Non-insect macroinvertebrates, such as Oligochaeta, Nematoda, Hydracarina, and Hirudinea, were identified to Order. Identification of tropical aquatic macroinvertebrates to the specie s level is difficult due to lack of taxonomic tropical literature (Jacobsen et al. 2008). Identi fication utilized Merritt & Cummins (Merritt and Cummins 1996), assistance from M. Springer and P. Hanson at the University of Costa Rica, and Roldan (Roldn 1996). Periphyton chlorophyll-a and AFDM samples we re filtered separately through Whatman pre-ashed and weighed filter-disks on the day of collection. These samples were then frozen for preservation and shipped to the University of Florida for analysis. Periphyton chlorophyll-a samples were corrected for pheophytin and measur ed with a spectrophotometer using standard procedures (APHA 1998). AFDM filter samples were dried in an oven at 60C for 48 hours, weighed, then ashed in a muffle furnace for five hours at 550C and weighed again. Data Analysis To test for differences in variables be tween canal sizes, ANOVAS using PROC MIXED were run in the program SAS (SAS Institute, Inc 2006). To test for correlations between different variables, Pearsons Correlations using PROC CORR were run in the program SAS. Water temperature, dissolved oxygen, canal disc harge, PAR, periphyton chlorophyll-a, and AFDM were log-normalized. When differences were significant, pos t-hoc analysis was
171 conducted using Tukeys test and B onferroni corrections. Taxa ri chness was used rather than species richness since no t all identifications were made to the same taxonomic level. Taxa richness was calculated by summing the number of different taxa found to the level of identification chosen in this study. To investigat e relationships of abiotic and biotic components to canal type, non-metric multidimensional scalin g ordinations using Sorensons metrics (i.e. Bray-Curtis) and Monte Carlo te sts were created in PC-Ord (McCune and Grace 2002). Taxa data were logged and relativize d. Biplots with a cut-off of P=0.2 were used to define associations between the sample and taxa distri butions. Indicator specie s analysis using Monte Carlo tests was conducted in PC-Ord. Results Physical-Chemical Parameters Channel width and depth (height) were dete rm ined to characterize canal dimensions. Average channel width and height was greatest in the Secondary canal, intermediate in the Tertiary canals, and smallest in the Quarternar y canals (Fig. 4-3). The difference in average channel width between the Secondary and Tertiary canals was nearly 1.5 times greater than the difference in average width between the Tertiary and Quarternary canals. The difference in average channel height was similar between Sec ondary and Tertiary canals (21 cm) and between Tertiary and Quarternary canals (20 cm). Average discharge was greatest in the Secondary canal (518.39 + 94.59 l/s) and decreased with decreasing canal size (Tertiary canals: 132.74 + 83.41 l/s, Quarternar y canals: 36.56 + 22.22 l/s) (Fig. 4-4A). Canal discharge in the Secondary canal was significantly greater than both the Tertiary (t=4.24, p=0.0006) and Quaternary canals (t=-6.18, p=<0.001), but there was no significant difference between the Tertiary and Quaternary canals.
172 Average canal water temperature was lo west in the Seconda ry canal (27.02 + 0.07C), intermediate in the Qu aternary canal (28.13 + 1.05C), and highest in the Tertiary canals (28.38 + 0.61C). Canal water temperature in the Second ary canal was significantly different from both the Tertiary (t=-4.24, p=0.0006) and Qu aternary canals (t=3.45, p=0.005). Average dissolved oxygen in canal water in creased as canal size decreased (Secondary: 8.06 + 0 .23 mg/l, Tertiary: 8.91 + 1.83 mg/l, Quaternary: 9.50 + 1.86 mg/l), but there were no significant differences between canal sizes. Average conductivity was similar between th e Secondary (106 uS/cm) and Tertiary (106.33 + 4.34 uS/cm) canals, while conductivity in the Quaternary canals was much higher and varied widely (190.83 + 153.51 uS/cm). However, there wa s no significant difference in conductivity relative to canal sizes. Several correlations were found within the canal network. In the Secondary canal, water temperature negatively correlated to discharge (F=-0.833, p=0.005). In the Tertiary canal, the water temperature positively correlated to dissolved oxygen (F=0.627, p=0.029). In the Quaternary canal, the water temperature nega tively correlated to conductivity (F=-0.894, p=<0.0001) and positively correlated to dissolved oxygen (F=0.987, p=<0.0001). Averages of both terrestrial and underwater pho tosynthetically active radiation (PAR) were greatest in Tertiary canals (terrestrial: 667 + 490 umol, underwater: 618 + 478 umol) and similar between the Secondary (terrestrial: 261 + 100 umol, underwater: 263 + 103 umol) and Quaternary (terrestrial: 369 + 243 umol, underwater: 255 + 214 umol) canals, but the Tertiary and Quaternary canals had high variability (Fi g. 4-4B). No significant difference was found for terrestrial PAR, but Tertiary and Quaternary canals had significantly different underwater PAR (t=-2.92, p=0.018).
173 Biotic Parameters Algal biomass Average periphyton chlorophyll-a levels were m u ch greater in the Secondary canal (4.54 + 3.77 g/m^3) than Tertiary (0.65 + 0 .85 g/m^3) and Quaternary canals (1.13 + 1.08 g/m^3), but values varied widely for all stream sizes. Despit e this variability, periphyt on chlorophyll-a in the Secondary canal was significantly greater than both Tertiary (t=4.27, p=0.0005) and Quaternary canals (t=-2.57, p=0.04). Average Ash-free Dry Mass (AFDM) levels were also much greater in the Secondary canal (16.784 + 13.719 g/cm^3) than the Tertiary (1.399 + 2.293 g/cm^3) and Quaternary canals (2.182 + 3.596 g/cm^3). Although values varied widely for all stream sizes, the Secondary canal had significantly greater algal biomass than the Tertiary (t=4.98, p =<0.0001) and Quarternary canals (t=-3.71, p= 0.002). Aquatic macroinvertebrates Total m acroinvertebrate abundance was much great er in Tertiary (10,391), than Quaternary canals (3203) and the Secondary can al (1179) (Fig. 4-5A). In th e latter, Diptera (50%) was the most common order, followed by Gastropoda (30%) (Fig. 4-5B). Oligochaeta (8%) and Trichoptera (6%) were the next most represente d orders. Gastropoda was by far the dominant order in the Tertiary and Quaternary canals, re presenting 62% and 55% of all macroinvertebrates collected, respectively. The next dominant order in Tertiary canals was Oligochaeta (14%), followed closely by Diptera (10%) and Ephemeropter a (9%). In the Quaternary canals, Diptera (16%), Oligochaeta (11%), and Ephemer optera (8%) were other notable orders. The dominant taxa in the Secondary canal were the gastropod Thiari dae (30%) and several chironomids (Chironomini: 20%, Tanytarsini: 13%, Orthocladiinae: 9%, and Tanypodinae: 3%), which comprised nearly all of the Diptera collected (Fig. 4-5C). Oligoc haeta (8%) and
174 Neotrichia (5%) also had notable representation. Thiaridae was the dominant taxon in Tertiary (45%) and Quarternary (50%) canals, as well as the most dominant Gastropoda. The other notable Gastropoda was Hydrobiidae, but much more so in Tertiary canals (15%) than Quaternary canals (2%). Other notable taxa in the Tertiary canal were Oligochaeta (14%), Baetidae (non-Baetodes ) (8%), and Tanypodinae (6%). In th e Quaternary canal, Oligochaeta (11%), Chironomini (6%), Ta nypodinae (5%), Baetidae (nonBaetodes ) (5%), and Orthocladiinae (3%) were well-represented. Of Ephemeroptera, the dominant taxa in each canal type were Baetidae (Secondary: 65%, Tertiary: 95%, Quarternary: 63%), followed by Farrodes (Secondary: 0%, Tertiary: 2%, Quarternary: 24%), Haplohyphes (Secondary: 22%, Tertiary: 2%, Quarternary: 6%), and Caenis (Secondary: 9%, Tertiary: 1%, Quarte rnary: 4%). Of the Trichopt era, the dominant taxa were Neotrichia (Secondary: 81%, Tertiary: 29%, Quarternary: 42%), Hydroptilidae sp. (Secondary: 16%, Tertiary: 20%, Quarternary: 4%), Oxyethira (Secondary: 0%, Tertia ry: 47%, Quarternary: 7%), Smicridea-Leptonema (Secondary: 0%, Tertia ry: 1%, Quarternary: 35%), and Leptoceridae (Secondary: 1%, Tertiary: 3%, Quarternary: 13%). Of the Diptera, the dominant taxa in each canal type were Chironomidae (Secondary: 89%, Te rtiary: 88%, Quarternary: 95%), followed by Chironomidae pupae (Secondary: 3%, Tertiary: 7% Quarternary: 2%), and Ceratopogoninae (Secondary: 7%, Tertiary: 5%, Quar ternary: 2%). The Chironomidae sub-family distribution in the Secondary canal was Chironomini (45%), Tany tarsini (28%), Orthoc ladiinae (19%), and Tanypodinae (9%) (Fig. 4-5D). Tertiary canal distribution was Tanypodinae (64%), Chironomini (17%), Orthocladiinae (12%), and Tanytarsini (6%). Quarternary canal distribution was Chironomini (41%), Tanypodinae (33%), Ort hocladiinae (21%), and Tanytarsini (4%).
175 Dominant functional feeding groups in the Secondary canal were collector-gatherersscrapers-shredders (41%) and collector-gatherers (31%), as well as collector-gatherers-filterers (13%), scrapers (5%), and predators (4%) to a much lesser extent (Fig. 4-6). Tertiary and Quaternary canals had similar representations of functional feeding groups, with collectorgatherers-scrapers-shredders dominating (57% 62%), followed by collector-gatherers (16%, 19%) and scrapers (16%, 5%). Predators comprise d only 4% of all taxa in the Secondary canal and 7% of all taxa in Ter tiary and Quaternary canals. Taxa richness was nearly identical in Tertiary and Quaternary canals, with 38 taxa and 37 taxa, respectively. Taxa ric hness in the Secondary canal was less, with only 28 taxa. The NMDS ordination provided a two-dimensi onal solution with a stress value of 13.963, and these axes accounted for 87% of the variation (axis 1=58%, axis 2=29%). It revealed that Secondary canal grouped apart from Quaternary canals, and Tertiary canals grouped primarily with the Quaternary canals, but did have some si milarities to the Seconda ry canal (Fig. 4-7). Canal discharge associated with the Secondary canal, along with Dipter a taxa (Ceratopogininae, Orthocladiinae, and Chironomini) and the Trichopteran Neotrichia Canal water temperature, dissolved oxygen, and both terrestrial and underw ater PAR associated with Tertiary and Quaternary canals. Several families of Gast ropoda (Hydrobiidae, Thiaridae, and Physidae), Oligochaeta, Coengrionidae, Chironomidae pupae, and Ephemeropteran taxa Farrodes and Haplohyphes all associated with these two canal sizes, as well. Only the Secondary canal and Tertiary canal s had indicator taxa. Two Chironomids were indicators in the Secondary canal: Chironomini (IV: 54.3, p= 0.001) and Tanytarsini (IV: 62, p= 0.005). The indicator taxa in the Tertiary canal were Hydrobiidae (IV: 53.2, p= 0.006),
176 Oxyethira (IV: 45.2, p= 0.006), Coengrionidae (IV: 49.3, p= 0.009), and Libellulidae (IV: 33.3, p= 0.025). Discussion The reverse trend of the River Continuum Concept was evident for certain physicalchemical canal water parameters. Canal size had a similar influence as stream size on discharge and dissolved oxygen. Average canal discharge decr eased as canal size decr eased, just as stream discharge decreases as stream size decreases. Average dissolved oxygen increased as canal size decreased, which resembles higher oxygenation of h eadwater streams compared to larger rivers. However, only the Secondary canal was significantly different from the other canal sizes in discharge and dissolved oxygen, which likely refl ects the greater similarity in channel size between the Tertiary and Quaternary canals than th e Secondary canal with th e Tertiary canals. Meanwhile, canal water temperature did not fo llow a gradient according to canal size, but was actually greatest in Tertiary canals. This likely corresponds to the much higher light penetration in Tertiary canals compared to ot her canal sizes, as shown by the highest average PAR values in Tertiary canals. Stream lighti ng regimes can influence stream temperatures (Hetrick and Brusven 1998, Johnson and Jones 2000) The RCC predicts that greatest species richness would be at the sites experiencing the widest diel temp erature changes, which provides more taxa an opportunity to encounter their optimum temperature (V annote et al. 1980). However, Melo and Froehlich (2001) found the greatest species richness in second order streams that had the smallest temperature variations, indi cating that temperature ma y not necessarily be a factor that determines richness in the tropics (Melo and Froehlich 2001). This study found that the Secondary canal had low water temperature variability and distinct ly low taxa richness compared to Tertiary and Quaternary canals, which had similarly high vari ability and higher taxa
177 richness. Thus, this study agrees with the RCC prediction that greater temperature variability supports greater taxa richness. Despite higher light penetration in Tertiary canals than other canals, the Secondary canal had the highest values of peri phyton chlorophyll-a and AFDM wh ereas Tertiary canals had the lowest. While these findings were surprising, they could be explaine d by total abundance of macroinvertebrate communities in each canal size. Tertiary canals had a substantially higher abundance of macroinvertebrates than other canal si zes, such that more in dividuals were present to consume algae. In additi on, 62% of those macroinverteb rates were Gastropods, which consume algae exclusively or as part of thei r diet (Brown 2001), and Gastropoda dominated the macroinvertebrate community to a greater extent in the Tertiary canal than the others canal sizes. Although high autochthonous produc tion typical of mid-sized streams described by the RCC could not be shown directly for Tertiary canals, high macroinvertebrate abundances and dominance by grazing gastropod su ggest high primary productivity. At the same time, the finding that collector-gatherer-sc rapers-shredders (57%) and scrapers (16%) combined were proportionally greatest in Tertia ry canals follows the expectation of the RCC that grazers (scrapers in some form) characterize mid-sized st reams. Consequently, this study contradicts a New Zealand study that found that epilithic pe riphyton pigment concentr ation best predicted invertebrate density (Dea th and Winterbourn 1995). An alternative explanation for high levels of both periphyton chlorophyll-a and AFDM in Tertiary canals could be that another subs trate besides sediment was supporting the macroinvertebrate community, such as thick gr owth of macrophytes, whic h was observed on the bottom of Tertiary canals. Macroinvertebrat e abundance and richness positively related to aquatic plant diversity in ditche s in England (Armitage et al. 2003), and macroinvertebrate taxa
178 richness was related to the dive rsity and quantity of submerged plants in Japanese wetlands (Takamura et al. 2009). Macrophytes provide cover and habita t structure for algal diatoms on which macroinvertebrates feed (Shannon et al. 1994). A New Zealand study demonstrated that macroinvertebrates were associated with aquati c macrophytes in canals (Hicks et al. 2006). Grass carp additions to a canal resulted in an 80% reduction in aquatic macrophytes, which caused a large reduction in the macroinvertebra tes they supported, pa rticularly gastropods. Besides gastropods, macroinvertebrates found on macrophytes included back swimmers, chironomids, Odonata larvae, leeches, tubifici ds, and water boatman, while the sediments contained chironomids, leeches, and Oligochaeta. The gastropod Bulinus truncatus inhabiting a Moroccan concrete canal was positively associated with macrophytes (Chlyeh et al. 2006), and filamentous green algae, Cladophora glomerata, covering a canal in Poland likely increased collector-gatherer densities by reducing current ve locity, providing substr ate, and allowing fine detritus to accumulate in algal mats (Fleituch 2003). Not only does the presence of submerged ve getation influence the macroinvertebrate community, the structure of that vegetation does, as well. Vegetation with more complex architecture (complex filamentous mats and subm erged plants versus simple emergent stems) harbored higher biomass of epiphyton and macroinvertebrates in a fluvial lake in the St. Lawrence River, Quebec (Tessier et al. 2008) Periphyton biomass and macroinvertebrate community abundance and diversity were grea ter on more structurally complex aquatic vegetation in a Tasmanian river, suggesting that the surface area of macrophytes provided the macroinvertebrate community with more habita t and food resources (Warfe and Barmuta 2006). Emergent vegetation removal by grazing cattle ca used a decrease in Odonata abundance and reproductive effort in Canadian prairie pothole wetlands, and vegetation structure (height and
179 density) was more important than species co mposition of the plant community (Foote and Hornung 2005). Chironomidae and Oligochaeta were dominant macroinvertebrate taxa inhabiting aquatic macrophytes in a slow-flowing channel in Croatia particularly in macrophytes with dissected leaves (Bogut et al. 2007). Consequently, submerged macrophytes in Tertiary canals in this study likely had a great influence on macroinvertebrates abundance. The intermediate channel size with high light penetration and intermed iate canal water discharge and depth would have contributed to the ability of Tertiary canals to support these macrophytes, following RCC predictions. Complimenting the RCC emphasis on the importan ce of streamflow, a literature review of ditches found that aspects of streamflow had th e greatest effect on species composition (Herzon and Helenius 2008). These aspects included dept h, discharge, and water chemistry, and they determined ditch type and size. Factors that co ntributed to aquatic biodi versity in ditches in southern England included distinct water chem istry, slow flow, connection to rivers, and a vegetation management regime that left ditche s mostly open and capable of supporting a variety of ecological niches (Armitage et al. 2003). St ream stability was the be st predictor of the number of species, with reduced stability resulting in a declin e in the number of species, and important factors influencing st ream stability included depth, cu rrent velocity, substrate and channel stability, and water te mperature (Death and Winterbo urn 1995). In South Carolina, stream width (1st-4th order) positively related to the total number of taxa, number of Ephemeroptera-Plecoptera-Trichoptera (EPT) taxa, and total number of organisms per sample, and stream width was inversely re lated to biotic index values, such that organism tolerance decreased as stream size increase d (Paller et al. 2006). Aspects of streamflow also seemed to have major effects on the macroinvertebrate comm unities within each canal type in this study.
180 Macroinvertebrate taxa richness and abundance both appeared to be influenced by discharge, channel depth, and channel width, all of whic h affect light penetr ation to the benthos, macrophyte growth, dissolved oxygen, and stream temperature variability, as previously mentioned. Beyond comparisons to the RCC, the macroinvert ebrates community in this study shared similarities and differences with other canal and ditch studies. A literature review of ditches found that common aquatic macroinvertebrates include Gastropods, Diptera larvae, water beetles, and crayfish, and that larger ditches may support a variet y of fish species (Herzon and Helenius 2008). The canals in this study were dominated by ga stropods and Diptera larvae followed by Oligochaeta, and fish and shrimp also were collected in minnow traps from the canals. A stream canal in Poland had the highest re presentation of collect or-gatherers, ranging from 87 to 96% of the taxa, along with so me scrapers and a few predators. C luster analysis based on taxa abundance showed that the stream canal was dominated by Baetidae, Rhyacophilidae (Trichoptera), and Chironomidae (Fleituch 2003). After collector-gatherersscrapers-shredders, collector-gatherers had the highest representation in each canal size, and predators constituted a relatively small percentage in this study. As in the Polish stream canal, this study found that Baetidae was the dominant Ephemeroptera and that Chironomidae was the dominant Diptera. A study of an adjacent irrigation project to th e one in this study found that an irrigation canal had primarily Trichoptera and Ephemeroptera and that most of the Odonata and Coleoptera collected, whereas a moderately stable and mo re contaminated drainage ditch had mostly Trichoptera and Gastropods, and th e most contaminated drainage ditch had mostly Gastropods and Veneroida bivalves (Rizo-Patrn 2003). Overall, most Ephemeroptera were Fallceon
181 (Baetidae), Tricorythodes (Leptohyphidae), and Traverella (Leptophlebiidae), which was similar to the finding in this study that E phermeroptera were primarily Baetidae, Haplohyphes (Leptohyphidae), and Farrodes (Leptophlebiidae). Most Trichoptera were Neotrichia with a few other genera in the families Hydropsychida e and Hydroptilidae, as was found in this study. Most Odonata were in the families Protoneuridae, Libellulidae, Gomphidae, and Coenagrionidae, all of which were found in this study, except Pr otoneuridae, and Libellu lidae and Coengrionidae were considered indicator taxa of Tertiary canals. Of the Diptera, dominant taxa were Chironomidae (68%) and Simuliidae (31%), and the families (Ceratopogonidae, Empididae, Stratiomyidae, and Tipulidae) representated less than 1% of the remaining Diptera. Chironomids were found in almost the same quantity in spite of differences in water quality. This study also found that Chironomidae larvae (90%) and Chir onomidae pupae (5%) dominated the Diptera, but that Ceratopogoninae, Empididae, Stratiomy idae, Psychodidae, Simuliidae, and Tipulidae comprised less than 5% of the remaining Dipt era. Chironomidae dominance was high in each canal size, though slightly greater in the Quarternary canal. As in this study, gastropods were dominated by Hydrobiidae, a family particularly abundant in places with considerable aquatic vegetation (Roldn 1996), and Thiaridae, an introduced taxon resistant to contamination (RizoPatrn 2003). The combination of these two studies provides a cl earer picture of the macroinvertebrate community that can be expected in different types of irrigation canals and drainage ditches in the dry tropics. Streams associated with sugar cane cultivation showed similar findings to those of canals and ditches. Primary productivity supported the macroinvertebrate community in sugar cane streams in Australia, and the microalgae living on macrophytes most likely served as a more important component of the diet of grazers and collector-gatherers than the macrophytes
182 themselves (Bunn et al. 1997). Brazilian sugar ca ne streams had fewer taxa and were dominated to a greater extent by tolerant Ch ironomidae than streams in areas with riparian vegetation (Corbi and Trivinho-Strixino 2008). Ot her dominant macroinvertebrat es in the sugar cane streams included Libellulidae, Dytiscidae, Ceratopogonidae, Polycentropodidae, Anne lida, Oligochaeta, and Hirudinea. Some aspects of the macroinvertebrate communities in the canals did not follow tenants of the RCC. Macroinvertebrate diversity is expected to be greatest for mid-sized streams under the RCC, but taxa richness was almost identical in Tert iary and Quaternary canal s. This lack of a hump-shaped curve in taxa rich ness could reflect too much sim ilarity in canal size for clear differentiation. A study in South Carolina found that taxa richness and biotic assessment metrics were significantly affected by stre am size, but results showed a lin ear rather than a hump-shaped curve in species richness exp ected under the RCC (Paller et al. 2006). This linear trend was attributed to sampling only a small to medium-si zed streams. Consequently, Paller et al. (2006) concluded that their findings were compatible with the RCC predictions for small to mid-sized streams, recognizing that relationships between ta xa richness and stream size can be obscured by the range in stream sizes sampled. The similarity in taxa richness between Tertia ry and Quaternary canal s could also reflect the influence of or interaction with the macroinve rtebrates community in the rice fields, to which Quaternary canals directly connect Macroinvertebrates moved fr om rice fields to adjacent irrigation canals in Japan, but whether a canal was st rongly or weakly connected to rice fields did not significantly affect the num ber of macroinvertebrates (Kata no et al. 2003), suggesting that proximity of Quaternary canals to the rice fields would not necessa rily explain th e structure of the macroinvertebrate community.
183 Although this study did not sample macroinvertebr ates within rice fields fed by the canals, other studies provide an idea of what macroinvertebrates could inhabit them. Portuguese rice fields contained Chironomidae, Gastropoda, Oligoc haeta, and Hirudinea, as well as Hemiptera, Coleoptera, and other Diptera la rvae (Leitao et al. 2007). Macroi nvertebrates larger than 1 mm in length found in water draining from rice fiel ds included Gerridae, Diptera adults, gastropods, Chironomidae larvae, and Odonata (Katano et al. 2003). Australian ri ce fields included Chironomidae, Dytiscidae, Hydrophilidae, Culicidae, Oligochaeta, Planorbidae, Libellulidae, Coengrionidae, and adult Veliidae (Wilson et al. 2008). Thus, the rice fields may have contributed to taxa richness in the Quarternary stream. Additionally, the RCC predicts dominance by shredders in small streams, but no Plecopteran shredders were found in any of the canal sizes, refl ecting the general paucity of shredders in the tropics rather than any stream size effects (Wantzen et al. 2008). Trends in the Chironomidae community were also notable. Representation of sub-families differed between canal sizes. Orthocladiinae a nd Chironomini were dominant in Secondary and Quaternary canals whereas Tanypodinae dominated Te rtiary canals. Averag e water temperatures were lower in Secondary and Quarternary canals th an in Tertiary canals, suggesting that this chironomid distribution may be in fluenced by water temperature. Flow regime, water temperature, and substrate composition were co nsidered main factors influencing Chironomidae distribution in a Canadian river (Ward and Williams 1986). Orthocladiinae dominated in headwaters, while Chironomini dominated farther downstream, which was attributed to stream temperature (Ward and Williams 1986, Rossaro 1991) A study in an Argentine river found that Orthocladiinae generally dominated upland stre am reaches, while Chironominae dominated lower reaches, and aquatic vegetatio n was one of the main factors influencing the distribution of
184 Chironomids in lowland reaches. Substrate characte r and current velocity appeared to influence the Chironomid distribution (Pri ncipe et al. 2008). Orthocladiinae were most common in Secondary and Quarternary canals, and least co mmon in the Tertiary can als, a grouping that corresponds with increasing aver age temperature, but does not follow a longitudinal pattern. Orthocladiinae and Tany tarsini were more common Chironom idae taxa at lower sites along a longitudinal gradient of an alpine glacial stream in Switzerland (Burgherr and Ward 2001), which corresponds with these two groups dominating in the Secondary canal. Tanytarsini generally showed an increasing trend in percent abundance from headwaters downstream in a relatively pristine river in Trinid ad (Helson et al. 2006), which ali gns well with the finding that Tanytarsini were only well-repr esented in the Secondary canal However, a study examining longitudinal zonation patterns of chironomids found that no single factor accounted for species richness patterns of Chironomidae in a physically complex spring system in Kansas (Ferrington et al. 1995). Instead, Ferrington et al. (1995) concluded that an alyzing each habitat as its own system and looking for factors influencing taxa richness within that particular system would provide better understanding of Chironomidae patte rns. This suggested approach may be more appropriate for analyzing the Chironomidae community in this canal system. Another potential influence on the macroinvertebrate commun ity worthy of consideration is the presence of fish in the canals. Ditches ma y provide important habitat for fish (Katano et al. 2003, Hicks et al. 2006). Chironomid larvae were an important food resource for fish in experimental streams that simulated temporary ditches around irrigated rice fields in Japan (Katano et al. 1998). Hicks et al. (2006) noted that the loss of gastropods from canals in New Zealand represented a loss of food resources for fish and wildfowl (Hicks et al. 2006). The majority of fishes in irrigation ditches in Japa n consumed aquatic macroinvertebrate larvae, such
185 as Ephemeroptera, Chironomidae, and Trichoptera, with a few other fish consuming adult insects and benthic algae. Smaller fishes (<4 cm sta ndard length) consumed more Chironomidae, while larger fishes consumed more Trichoptera. Thes e findings demonstrated that temporary waters, such as irrigation ditches in rice paddy fields, can provide habitat impor tant for fish growth, survival, and reproduction due to the presence of abundant food resources, particularly Chironomidae larvae (Katano et al. 2003). Alth ough fish and freshwater shrimp were not sampled in this study, they were present. Conclusion This study dem onstrated that each canal si ze within this Costa Rican irrigation water distribution system contains unique physical-chemical characteristics and macroinvertebrate community structures, some of which correspon d with predictions of the River Continuum Concept, but in reverse order. The largest ca nal size had the highest canal water discharge, lowest dissolved oxygen, lowest taxa richness, lo west total number of individuals, and highest periphyton chlorophyll-a and AFDM levels. The mi d-sized canal had intermediate canal water discharge, intermediate dissolved oxygen, lowe st periphyton chlorophyl l-a and AFDM levels, highest total abundance, and grea test relative abundance of gast ropods. The smallest canal size had the lowest canal water discharge, highest dissolved oxygen, and had similarly low total abundances as the largest canal size, but simila rly high taxa richness as the mid-sized canal. Canal water discharge and dissolv ed oxygen patterns, prevalence of scrapers (exclusive or in part) in mid-sized channels, and higher taxa richness in the smaller to mid-sized channels corresponded most closely with the RCC. Although this irrigation canal network in Costa Rica was compared to stream networks, the two networks are not true equivale nts. Canal Oeste is concrete lined along the majority of its course, unlike high-order rivers that have varied substrate. The Secondary, Tertiary, and
186 Quarternary canals have similar substrates composed primarily of earth and fine sediments. In contrast, natural streams have substrates composed of a combination of st ones, gravel, and fine sediments that vary from headwa ters to high-order reaches. In addition, this canal network lacked extensive canopy cover in the Quarte rnary canal and had no progression in light penetration along its course, while natural stream headwaters are typica lly forested and light penetration progressively increase s as stream order increases. Consequently, limitations exist when comparing the networks. Yet despite diffe rences and limitations, a major characteristic that this canal network and stream networks do share is a size gradie nt along their course, specifically for the first few stream orders (1st-4th). Thus, channel size was used as the basis to compare the ecosystem structure and f unction of these canals and streams. This investigation of the ecosystems within each canal size also provides information to canal managers. As a whole, the largest canal showed strong differences with the other canal sizes, while the smallest and mid-sized canals show ed much weaker differe nces with each other. Consequently, management practices for these ca nal sizes could be grouped into two canal size categories, such that one set of practices is deve loped for the largest canals, while another set of practices is developed for the mid-sized and sma llest canals. However, sp ecial consideration of the value of macrophytes for the macroinvertebrate community in the mid-sized canals should be made. Finally, this study showed that irrigation canals in Costa Rica are not completely different from streams. Like streams, canals do support an aquatic macroinvertebrate community that is influenced by the discharge, depth, channel width, stability, and light pene tration in the canals, which shape the instream habitat. The canals also supported a fish a nd shrimp community, but
187 their diversity, distribution, and influence on the aquatic macroinve rtebrate community remain to be explored.
188 Figure 4-1. River Continuum Con cept versus Canal Continuum.
189 Figure 4-2. Canal Continuum St udy site map. Coding: 2=Secondary canal, 3=Tertiary canal, and 4=Quarternary canal.
190 Figure 4-3. Canal dimensions, by canal size. Coding: 1=Canal Oeste, 2=Secondary canal, 3=Tertiary canal, 4=Quarternary canal.
191 A B Figure 4-4. Physical-chemical pa rameter averages, by canal type. A) Average canal discharge, and B) Average photosynthetic active radiation (PAR).
192 A B Figure 4-5. Macroinvertebrates community stru cture by canal size. A) Total abundance, B) Relative abundance, by order, C) Rela tive abundance, by taxa, and D) Relative abundance, by Chironomidae taxa.
193 C D Figure 4-5. Continued
194 Figure 4-6. Macroinvertebr ate functional feeding groups by canal type. Coding: PRED=Predators; FILT=Filterers; SC=Scrap ers; CG=Collector-gatherers; CG-SH, CG-SC=Collector-gatherers-scrapers-shredders; CF=Collector-filterers; CGCF=Collector-gatherers-filterers; CG-SCSH-CF=Collector-gatherers-filterersscrapers-shredders; and ALL=All feeding modes.
195 Figure 4-7. Relativized Ordination, by canal type. Triangle=2 (Secondary), Square=3 (Tertiary), and Diamond=4 (Quarternary. Codi ng: ANCYLID=Ancylidae, BAET=Baetidae (NonBaetodes ), CHIR_P=Chironomidae Pupae, CHRMINI=Chironomini, COEN=Coengrionidae, FARRODES= Farrodes HAPLO= Haplohyphes, HYDROBI=Hydrobiidae, LperS=Di scharge (l/s), NEOTRICH= Neotrichia NL=Hydroptilidae sp., OLIGO=Oligochaeta, ORTHOCL=Orthocladiinae, PHYSID=Physidae, SPHAERID=Sphaeridae, T=Temperature (C), TANYPOD=Tanypodinae, TANYTAR=Tanytar sini, T_umol=Terrestrial PAR (umol), and THIARID=Thiaridae.
196 CHAPTER 5 STUDY CONCLUSION Ecohydrology The sem i-arid Guanacaste region of Costa Rica c ontains some of the few protected tropical dry forests remaining in the tropics of Centra l America, whose streams have received little scientific study. A recently constructed large irri gation canal intersects some of these streams, adding streamflow year-round to th e typically intermittent headwate r streams. This provided an opportunity to investigate: 1) the ecohydrology of intermittent (T) and perennial (P) tropical dry forest streams, and 2) the impacts of canals on tr opical dry forest stream structure and function. Sampling occurred from May 2007 to January 2008, spanning the wet season and seasonal transitions. Physical-chemical ch aracteristics, aquatic macroinvertebrates, and fish were sampled mostly on a 4-6 week basis. Residual, isolated pools in T streams were found during drier periods, which served as refugia for macroinvertebrates, primarily Dipter a (76%) and the tolerant Ephemeroptera family Caenis (14%), and fish, particularly the poecilids Brachyrhaphis olomina and Poecilia gillii Streamflow was recorded in T streams on three occasions during the wet season, and streamflow was sustained in most P streams for the duration of th e study, with fluctuations in discharge generally reflecting changes in canal wa ter level. P streams had significantly greater water temperature and conductivity. The benthic macroinvertebrate community change d dramatically over time in T streams. At the start of the wet se ason, Oligochaeta dominated, and Diptera only had a small representation, but by the end of the wet season, T and P stream co mmunities were very similar due to increased dominance of Diptera and Ephemeroptera in T streams. Caenis and Chironomidae were the dominant taxa, and collector-gatherers-scrapers-shredders was the
197 dominant functional feeding group in both stream types. Taxa richness was similar between the two stream types, and NMDS ordinations revealed both that taxa with lower water quality demands associated with T streams and that th e two stream types became more similar over the duration of the wet season. The number of indica tor taxa for each stream type decreased over time, indicating increasing similarity in macroinvertebrates communities. Total abundance of macroinvertebrates in drif t was similar between the two stream types despite their distinct streamflow patterns and showed patterns of change similar to benthic macroinvertebrates. T stream drift was character ized by more Terrestrials and Coleoptera, while P streams had greater dominance of Ephe meroptera, Diptera, and Bivalvia. More fish were collected from P streams, but there was no difference in species composition. Poecilia gillii dominated T streams, while B. olomina dominated P streams, likely reflecting their feeding habits and tolerance levels. This study assessed the aquatic community of tropical dry fore st streams and demonstrated the strong influence of streamflow on the colonization, development, and succession of aquatic biota. It also revealed impact s of irrigation canals on tropical dry forest streams, such as increased habitat during the dr y season, habitat fragmentation, channel scouring, and water quality changes. Riparian Patch Study A large irrigation canal crosses a second-or der seasonal stream in Costa Rica via an inverted siphon, and the riparian ve getation within that crossing site is m aintained deforested for canal management purposes. This deforestation created a gap in riparian canopy cover along a stretch of a tropical dry forest st ream, resulting in three distinct environments within the stream: a deforested zone where the canal passes (Open zone), a forested zone upstream of the Open zone (Above zone), and a forested zone downstrea m of the Open zone that could receive inputs
198 from the Open zone (Below zone). In addition, a fourth zone existed further downstream that receives both seasonal streamflow and dire cts canal water discha rge year-round (Canalinfluenced zone). This raised the questions : 1) Do canals influence stream structure and function, and 2) Do patches in riparian vegetatio n affect macroinvertebra te and fish community structure within tropical dry fo rest streams? The physical-chemi cal characteristics and biological communities of these zones were analyzed and compared to answer those questions. Physical-chemical analysis revealed that canopy cover and photosynthetically active radiation were significan tly greater in the Open zone compared to other zones, and average levels of periphyton chlorophyll-a and AFDM were also greater in the Open zone. Streamflow influenced the effects of canopy cover on water temperature and dissolved oxygen. For aquatic macroinvertebrates overall, the zones had few indicat or taxa, and NMDS ordination revealed no clear di stinctions between zones. The benthos, stones, and drift macroinvertebrate communities showed similar tr ends in succession over time among the zones, and these trends followed streamflow patterns. Diptera and Oligochaeta generally dominated in the early wet season, Ephemeroptera and Trichop tera grew more dominant as the wet season progressed then decreased in the dry season, at which point Diptera regained dominance. This succession corresponded with stream discharge and dissolved oxygen patterns. However, the Canal-influenced zone showed less dramatic ch anges. The Above and Below zones showed great similarity, suggesting that conditio ns in the Open zone did not have strong downstream effects. The three habitats sampled also had different community composition with benthos dominated by Ephemeroptera and Diptera, stones domin ated by Diptera, and drift dominated by Ephemeroptera and Trichoptera, but few taxa were exclusive to any one habitat. Benthic and stone macroinvertebrate total abundances were gr eatest in the Open zone, and taxa richness was
199 slightly greater in the Above and Open zones than the Below and Canal-influenced zones. Drift macroinvertebrate total abundance and taxa richness was greatest in the Above zone. For fish, all zones were dominated by Astyanax aeneus, but the second most abundant fish species differed among the zones: Brachyrhaphis olomina in the Above zone, Poecilia gillii in the Open zone, and Archocentrus nigrofasciatus in the Below and Canal-influenced zones. This distribution was likely due to the influence of water depth, stream channel morphology, and canopy cover. Although some differences were found between zones due to canal management, small scale riparian deforestation (< 35 m) may not ha ve had severe detrimental effects on the stream community and may actually have provided a greater diversity of habitats and resources capable of potentially enhancing biotic ri chness. Perennial discharge from the canal into the stream may provide refugia and a source of colonizers for the seasonal stream reach in the dry season. Future research that analyzes macroinvertebrate and fish diet would provi de valuable additional information. Canal Continuum Study Irrigation canals serving suga r cane and rice fields provide f reshwater habitat in the seasonally dry Guanacaste province of Costa Rica. The canals diminish in size over their longitudinal gradient between the primary canal serving the region and the fields the canals locally irrigate, creating three size classes: Secondary (largest ), Tertiary (mid-sized), and Quarternary (smallest) canals. Sampling of the physical-chemical characteristics and aquatic macroinvertebrates occurred in each of these thre e canal sizes both to de termine their structure and function and to examine whether the canals follow the predictions of the River Continuum Concept (RCC), but in reverse order due to the si ze structure of the canals compared to natural streams.
200 This study revealed that canal discharge decreased and disso lved oxygen increased as canal size decreased, and scrapers (exclusive or in pa rt) dominated Tertiary canals more so than the other sizes, all findings in accordance with RCC predictions. However, taxa richness was equally high in the Tertiary and Quarternary cana ls compared to the Sec ondary canal rather than highest in the Tertiary canal, wh ich may reflect greater similarity in size among the Tertiary and Quarternary canals or macroinverteb rate movement from rice fields into the small canals that directly feed them. Periphyton chlorophyll-a and AFDM levels were highest in the Secondary canal rather than Tertiary canal s, and total abundance was far greater in Tertiary canals than other canal sizes, which may reflect the presen ce of thick submerged vegetation in Tertiary canals providing additional habitat and food resources that support a larger population capable of consuming more algae. The dominant macroinve rtebrates in each canal size were Gastropoda, Chironomidae, and Oligochaeta. As a whole, the canals did support an aquati c community whose stru cture and function did change over its longitudinal gradie nt, and the canals did follow some predictions of the RCC, but in reverse order. The greater similarity between Tertiary and Quarternar y canals suggests that two separate canal management practices should be developed, one for the Secondary canal and one for Tertiary and Quarternary canals. Future research should address fish and shrimp use of the canals and their influence on the macroinvertebrate community. Holistic Perspective This entire study dem onstrated that the large irrigation canal, Canal Oeste, did affect the structure and function of tropica l dry forest streams through a ltering the hydrology and riparian vegetation of those streams. Impacts include additional habitat during the dry season and differing biotic assemblage structures, but thes e were not necessarily negative and may have contributed positively to stream diversity. The study also demonstrated that irrigation canals
201 share similarities with streams through providing freshwater habitat and following some tenets of the River Continuum Concept.
202 LIST OF REFERENCES Acuna, V., I. Munoz, A. Giorgi, M. Om ella, F. Sabater, and S. Sabater. 2005. Droughts and post drought recovery cycles in an intermittent Me diterranean stream: structural and functional aspects. Journal of the North American Benthological Society 24:919-933. Aguiar, F. C., M. T. Ferreira, and P. Pinto. 2002. Relative Influence of Environmental Variables on Macroinvertebrates Assemblages from an Ib erian Basin. Journal of the North American Benthological Society 21 :43-53. Allan, J. D. 1995. Stream Ecology: Structur e and Function of Running Waters. Kluwer Academic Publishers, Dordrecht. Ambrose, H. E., M. A. Wilzbach, and K. W. Cummins. 2004. Periphyton response to increased light and salmon carcass introduc tion in northern Calif ornia streams. Journal of the North American Benthological Society 23 701-712. APHA. 1998. Standard Methods for the Examinati on of Water and Wastewater. 20th edition. American Public Health Association, American Water Works Association, Water Pollution Control Federation, Washington, DC. Armitage, P. D., K. Szoszkiewicz, J. H. Bl ackburn, and I. Nesbitt. 2003. Ditch communities: a major contributor to floodplain biodiversity. A quatic Conservation: Marine and Freshwater Ecosystems 13:165-185. Arrington, D. A., and K. O. Winemiller. 2006. Habitat affinity, the seasonal flood pools, and community assembly in the litto ral zone of a Neotropical flo odplain river. Journal of the North American Benthological Society 25:126-141. Avila, S. A., and R. W. Flowers. 2005. Ne w species and records of Ulmeritoides (Ephemeroptera: Leptophlebiidae) from Costa Rica. Zootaxa:1-14. Barbee, N. C. 2005. Grazing insects reduce al gal biomass in the neotropical stream. Hydrobiologia 532:153-165. Beche, L. A., E. P. McElravy, and V. H. Resh. 2006. Long-term seasonal variation in the biological traits of benthicmacroinvertebrates in two Mediterranean-climate streams in California, U.S.A. Freshwater Biology 51:56-75. Benstead, J. P., M. M. Douglas, and C. M. Pri ngle. 2003. Relationships of stream invertebrate communities to deforestation in eastern Madagascar. Ecological Applications 13:14731490. Benstead, J. P., and C. M. Pringle. 2004. Defore station alters the resource base and biomass of endemic stream insects in eastern Madagascar. Freshwater Biology 49:490-501. Biggs, B. J. F., V. I. Nikora, and T. N. Snelde r. 2005. Linking scales of fl ow variability to lotic ecosystems structure and function. Ri ver Research and Applications 21 :283-298.
203 Bilby, R. E., and P. A. Bisson. 1992. Allocht honous versus autochthonous organic matter contributions to the trophic s upport of fish populations and clear -cut and old-growth forest streams. Canadian Journal of Fisheries and Aquatic Sciences 49:540-551. Bilby, R. E., K. Sullivan, and S. H. Duncan. 1989. The generation and fate of road-surface sediment in forested watersheds in S outhwestern Washingt on. Forest Science 35:453-468. Bogut, I., J. Vidakovic, G. Palijan, and D. Cerb a. 2007. Benthic macroinvertebrates associated with four species of macrophytes. Biologia 62 :600-606. Bojsen, B. H., and D. Jacobsen. 2003. Effects of deforestation on macroi nvertebrate diversity and assemblage structure in Ecuadorian Amazon streams. Archiv fur Hydrobiologie 158:317-342. Bond, N. R., G. L. W. Perry, and B. J. Downes 2000. Dispersal of organisms in a patchy stream environment under different settlement s cenarios. Journal of Animal Ecology 69:608-619. Boothroyd, I. K. G., J. M. Quinn, E. R. Langer, K. J. Costley, and G. Steward. 2004. Riparian buffers mitigate effects of pine plantation logging on New Zealand streams 1. Riparian vegetation structure, stream geomorphology and periphyton. Forest Ecology and Management 194 199-213. Boulton, A. J. 2003. Parallels and contrast s in the effects of drought on stream macroinvertebrates assemblages. Freshwater Biology 43:1173-1185. Boulton, A. J., and P. S. Lake. 1990. The ecology of two intermittent streams in Victoria, Australia. I. Multivariate analyses of physicochemical features. Freshwater Biology 24:123-141. Boulton, A. J., and P. S. Lake. 1992. The ecology of two intermittent streams in Victoria, Australia. III. Temporal changes in formal composition. Freshwater Biology 27:123-138. Boyero, L., and J. Bosch. 2002. Spatial and temporal variation of macroinve rtebrate drift into neotropical streams. Biotropica 34:567-574. Boyero, L., and J. Bosch. 2004. Multiscale spa tial variation of stone recolonization by macroinvertebrates in a Costa Rican stream. Journal of Tropical Ecology 20:85-95. Boyero, L., and J. L. DeLope. 2002. Short-term recolonization of stones in a tropical island stream. Marine and Freshwater Research 53:993-998. Bradbury, R. B., and W. B. Kirby. 2006. Farmla nd birds and resource pr otection in the UK: Cross-cutting solutions for multi-functional farming? Biological Conservation 129 :530542. Brown, K. M. 2001. Mollusca: Gastropoda. in J. H. Thorp and A. P. Covich, editors. Ecology and Classification of North American Freshwater Invertebrates. Academ ic Press, San Diego.
204 Bunn, S. E., P. M. Davies, and D. M. Kellaway. 1997. Contributions of su gar cane and invasive pasture grass to the aquatic food web of the tropical lowland stream. Marine and Freshwater Resources 48 :173-179. Bunn, S. E., P. M. Davies, and T. D. Mosisch. 19 99. Ecosystem measures of river health in their response to riparian and catchment degradation. Freshwater Biology 41:333-345. Burcher, C. L., M. E. McTammany, E. F. Benfie ld, and G. S. Helfman. 2008. Fish assemblage responses to forest cover. Environmental Management 41:336-346. Burgherr, P., and J. V. Ward. 2001. Longitudinal in the seasonal distribution patterns of the benthic fauna of an alpine glacial stream (Val Roseg, Swiss Alps). Freshwater Biology 46:1705-1721. Buss, D. F., D. F. Baptista, J. L. Nessimi an, and M. Egler. 2004. Substrate specificity, environmental degradation and disturbance st ructuring macroinvertebrates assemblages in neotropical streams. Hydrobiologia 518:179-188. Bussing, W. A. 2002. Peces de las aguas continentale s de Costa Rica. 2nd edition. Editorial de la Universidad de Costa Rica. Callisto, M., and M. Goulart. 2005. Invertebrate drift along the longit udinal gradient in a neotropical stream in Serra do Cipo National Park, Brazil. Hydrobiologia 539:47-56. Casas, J. J. 2008. Chironomids species richne ss of a permanent and temporary Mediterranean stream: a long-term comparative study. Jour nal of the North American Benthological Society 27 :746-759. Chakona, A., C. Phiri, C. H. D. Magadza, and L. Brendonck. 2008. The influence of habitat structure and flow permanence on macroinvertebrates assembla ges in temporary rivers in northwestern Zimbabwe. Hydrobiologia 607:199-209. Chapman, L. J., and D. L. Kramer. 1991a. The c onsequences of flooding for the dispersal and fate of poeciliid fish in an inte rmittent tropical stream. Oecologia 87:299-306. Chapman, L. J., and D. L. Kramer. 1991b. Limnologi cal observations of an intermittent tropical dry forest stream. Hydrobiologia 226 :153-166. Chapman, L. J., D. L. Kramer, and C. A. Chapman. 1991. Population-dynamics of the fish Poecilia gillii (Poeciliidae) in pools of in intermittent tropical stream. Journal of Animal Ecology 60:441-453. Chlyeh, G., M. Dodet, B. Delay, K. Khallaayoune, and P. Jarne. 2006. Spacial-temporal distribution of freshwater snai l species in relation to migrat ion and environmental factors in an irrigated area from Morocco. Hydrobiologia 553:129-142. Closs, G. P., and P. S. Lake. 1994. Spatial and temporal variation in the structure of an intermittent stream food web. Ecological Monographs 64:1-21.
205 Collier, K. J. 1995. Environmental factors aff ecting the taxonomic composition of aquatic macroinvertebrates communities in lowland waterways of Northl and, New Zealand. New Zealand Journal of Marine and Freshwater Research 29:453-465. Collier, K. J., and J. M. Quinn. 2003. Land-us e influences macroinvertebrate community response following the pools disturbance. Freshwater Biology 48:1462-1481. Corbi, J. J., and S. Trivinho-Strixino. 2008. Re lationship between sugar cane cultivation and stream macroinvertebrate communities. Br azilian Archives of Biology and Technology 51:769-779. Covich, A. P. 1988. Geographical and historical comparisons of Neotropical streams: Biotic diversity in the trial processing in highly variable habitats. Community structure and function in temperate and tropical streams: Proceedings of a symposium. Journal of the North American Benthological Society. Cross, W. F., A. P. Covich, T. A. Crowl, J. P. Benstead, and A. Ramirez. 2008. Secondary production, longevity and resource consumption ra tes of freshwater shrimps in two tropical streams with contrasting geomorphology and food web structure. Freshwater Biology 53:2504-2519. Death, R. G. 2002. Predicting inverteb rate diversity from disturbance regimes in forest streams. Oikos 97:18-30. Death, R. G., and M. J. Winterbourn. 1995. Diversity in patterns in stream benthic invertebrate communities: The influence of habitat stability. Ecology 76:1446-1460. DeNicola, M., and C. D. McIntire. 1991. Effects of hydraulic refuge and irradiance on grazerperiphyton interactions in laboratory streams. Journal of the North American Benthological Society 10 :251-262. Dodds, W. K., K. Gido, M. R. Whiles, K. M. Fr itz, and W. J. Matthews. 2004. Life on the edge: The ecology of Great Plains prairie streams. BioScience 54 :205-216. Douglas, M., and P. S. Lake. 1994. Species richness of stream stones: An investigation of the mechanisms generating the sp ecies-area relationship. Oikos 69 :387-396. Downes, B. J., P. S. Lake, and E. S. G. Schr eiber. 1995. Structure and in vertebrate assemblages on stream stones: A multivariate view from the riffle. Australian Journal of Ecology 20:502-514. Downes, B. J., P. S. Lake, E. S. G. Schreiber, and A. Glaister. 2000. Habitat structure, resources and diversity: the separate effects of surface roughness and macroalgae on stream invertebrates. Oecologia 123:569-581. Dudgeon, D. 1994. The influence of riparian vegetation on macroinvertebrate community structure and functional-or ganization in 6 New Guinea streams. Hydrobiologia 294 :65-85.
206 Dudgeon, D. 2006. The impacts of human disturbance on stream benthic invertebrates and their drift in North Sulawesi, I ndonesia. Freshwater Biology 51:1710-1729. Dudgeon, D., editor. 2008. Tropi cal Stream Ecology. Academic Press, Amsterdam. England, L. E., and A. D. Rosemond. 2004. Small reductions in forest cover weekend terrestrials-aquatic linkages in head water streams. Freshwater Biology 49:721-734. Feminella, J. W. 1996. Comparison of benthic macr oinvertebrates assemblages in small streams along a gradient of flow permanence. Journal of the North American Benthological Society 15:651-669. Ferrington, L. C., R. G. Kavanaugh, F. J. Schmidt, and J. L. Kavanaugh. 1995. Habitat separation among Chironomidae (Diptera) in Big Springs. Journal of the Kansas Entomological Society 68:152-165. Fisher, S. G., L. J. Gray, N. B. Grimm, and D. E. Busch. 1982. Temporal succession in the desert stream ecosystem following flas h flooding. Ecological Monographs 52:93-110. Fleituch, T. 2003. Structure and f unctional organization of benthic invertebrates in a regulated stream. International Review of Hydrobiology 88:332-344. Foote, A. L., and C. L. R. Hornung. 2005. Odonates as biological indicators of grazing effects on Canadian prairie wetlands. Ecological Entomology 30:273-283. Frankie, G. W., H. G. Baker, and P. A. Opler. 1974. Comparative phenologica l studies of trees in tropical wet and dry forests in the lowl ands of Costa Rica. Journal of Ecology 62:881-919. Franssen, N. R., K. B. Gido, C. S. Guy, J. A. Trip e, S. J. Shrank, T. R. Strakosh, K. N. Bentrand, C. M. Franssen, K. L. Pitts, and C. G. Paukert. 2006. Effects of floods on fish assemblages in the intermittent prairie stream. Freshwater Biology 51:2072-2086. Fritz, K. M., and W. K. Dodds. 2002. Macroinverte brates assemblage structure across a tallgrass prairie stream landscape. Archiv fur Hydrobiologie 154:79-102. Fritz, K. M., and W. K. Dodds. 2004. Resistan ce and resilience of macroinvertebrates assemblages to drying and flood in the tall grass prairie streams system. Hydrobiologia 527:99-112. Gasith, A., and V. H. Resh. 1999. Streams in Med iterranean climate regions: Abiotic influences and biotic responses to predictable seas onal events. Annual Review of Ecology and Systematics 30:51-81. Gomi, T., R. C. Sidle, and J. S. Richards on. 2002. Understanding pr ocesses and downstream linkages of headwater systems. BioScience 52 :905-916. Gordon, N. D., T. A. McMahon, B. L. Finlayson, and R. J. Nathan. 1992. Stream Hydrology: An introduction for ecologists. 2nd edition. John Wiley & Sons, LTD, Chichester.
207 Greathouse, E. A., and C. M. Pringle. 2006. Do es the River continuum concept apply on a tropical island? Longitu dinal variation in a Puerto Rican stream. Canadian Journal of Fisheries and Aquatic Sciences 63:134-152. Gregory, S. V., F. J. Swanson, W. A. McKee, and K. W. Cummins. 1991. An ecosystem perspective of ripari an zones. BioScience 41:540-551. Hawkins, C. P., M. L. Murphy, and N. H. Anderson. 1982. Effects of canopy, substrate composition, and gradient on the structure of macroinvertebrates communities in Cascade Range streams of Oregon. Ecology 63 :1840-1856. Helson, J. E., D. D. Williams, and D. Turner. 2006. Larval chironomid community organization in four tropical rivers: human impact s and longitudinal zonation. Hydrobiologia 559 :413431. Herzon, I., and J. Helenius. 2008. Agricultural drai nage ditches, their biological importance and functioning. Biological Conservation 141:1171-1183. Hetrick, N. J., and M. A. Brusven. 1998. Changes in solar input, water temperature, periphyton accumulation, and allochthonous input and storage after canopy removal along two small salmon streams in southeast Alaska. Transac tions of the American Fisheries Society 127:859-875. Hicks, B. J., H. J. Bannon, and R. D. S. We lls. 2006. Fish and macroinvertebrates and lowland drainage canals with and w ithout Grass Carp. Journal of Aquatic Plant Management 44:8998. Hill, W. R., and A. W. Knight. 1988. Nutrient and light limitation of algae in 2 Northern California streams. Journal of Phycology 24:125-132. Holdridge, L. R. 1967. Life zone ecology. Tropi cal Science Center, San Jose, Costa Rica. Hooke, J. M. 2006. Human impacts on fluvial systems in the Mediterranean region. Geomorphology 79:311-335. Hose, G. C., P. Jones, and R. P. Lim. 2005. Hypor heic macroinvertebrates in riffle and pool areas of temporary streams in south eastern Australia. Hydrobiologia 532:81-90. Hughes, D. A. 2005. Hydrological is sues associated with the de termination of environmental water requirements of ephemeral rivers River Research and Applications 21:899-908. Humphries, P., and D. S. Baldwin. 2003. Drought and aquatic ecosystems: an introduction. Freshwater Biology 48:1141-1146. Jackson, C. R., D. P. Batzer, S. S. Cross, S. M. Haggerty, and C. A. Sturm. 2007. Headwater streams and timber harvests: Channel, macroinvertebrate, and amphibian response and recovery. Forest Science 53:356-370.
208 Jackson, J. K., E. P. McElravy, and V. H. Resh. 1999. Long-term movements of self-marked caddisfly larvae (Trichoptera: Sericostomatid ae) in a California coastal mountain stream. Freshwater Biology 42:525-536. Jackson, J. K., and B. W. Sweeney. 1995. Egg and larval development times for 35 species of tropical stream insects from Costa Rica. J ournal of the North American Benthological Society 14 :115-130. Jacobsen, D. 2005. Temporally variable macroinve rtebrate-stone relatio nships in streams. Hydrobiologia 544:201-214. Jacobsen, D., and B. H. Bojsen. 2002. Macroinverte brate drift in Amazon st reams in relation to riparian forest cover and fish fauna. Archiv fur Hydrobiologie 155:177-197. Jacobsen, D., C. Cressa, J. M. Mathooko, and D. Dudgeon. 2008. Macroinvertebrates: Composition, Life Histories and Production. Pages 66-106 in D. Dudgeon, editor. Tropical Stream Ecology. Academic Press, Amsterdam. Jacobsen, D., and A. Encalada. 1998. The macr oinvertebrate fauna of Ecuadorian highland streams in the wet and dry s eason. Archiv fur Hydrobiologie 142 :53-70. James, A. B. W., Z. S. Dewson, and R. G. Death. 2009. The influence of flow reduction on macroinvertebrate drift density and distance in three New Zealand str eams. Journal of the North American Benthological Society 28:220-232. Janzen, D. H. 1986. Guanacaste National Park: Tropical ecological and cultural restoration. Editorial Universidad Estatal a Distancia, San Jose. Janzen, D. H. 1988. Management of habitat fragment s in a tropical dry forest: growth. Annals of the Missouri Botanical Garden 75:105-116. Jimenez, J., E. Gonzalez, and J. Mateo-Vega, editors. 2001. Perspectives for the integrated management of the Tempisque River Basin, Costa Rica. Organization for Tropical Studies, San Jose. Johnson, S. L., and J. A. Jones. 2000. Stream temp erature responses to forest harvest and debris flows in western Cascades, Or egon. Canadian Journal of Fish eries and Aquatic Sciences 57:30-39. Jones III, E. B. D., G. S. Helfman, J. O. Ha rper, and P. V. Bolstad. 1999. Effects of riparian forest removal on fish assemblages in S outhern Appalachian streams. Conservation Biology 13:1454-1465. Kasangaki, A., L. J. Chapman, and J. Balirwa. 2008. Land-use in the ecology of benthic macroinvertebrates assemblages of high-altitude rainforest streams in Uganda. Freshwater Biology 53:631-697.
209 Katano, O., K. Hosoya, K. Iguchi, M. Yamaguc hi, Y. Aonuma, and S. Kitano. 2003. Species diversity and abundance of freshwater fishes and irrigation ditches around ricefields. Environmental Biology of Fishes 66:107-121. Katano, O., J. Toi, K. Maekawa, and K. Iguchi. 1998. Colonization of an artificial stream by fishes and aquatic macro-invert ebrates. Ecological Research 13:83-96. Keith, R. M., and T. C. Bjornn. 1998. Response of j uvenile salmonids to riparian and instream cover modifications and small streams flowi ng through second-growth forests of Southeast Alaska. Transactions of the American Fisheries Society 127 :889-907. Kelly, D. J., M. L. Bothwell, and D. W. Schindl er. 2003. Effects of solar ultraviolet radiation on stream benthic communities: an intersite comparison. Ecology 84:2724-2740. Kiffney, P. M., J. S. Richardson, and J. P. Bu ll. 2004. Establishing light is a causal mechanism structuring stream communities in response to experimental manipulation of riparian buffer width. Journal of the North American Benthological Society 23 :542-555. Kouame, K. A., S. S. Yao, B. G. Goore, E. P. Kouamelan, V. N'Douba, and N. J. Kouassi. 2008. Influential environmental grad ients and patterns of fish a ssemblages in a West African basin. Hydrobiologia 603 :159-169. Lake, P. S. 2000. Disturbance, patchiness, and diversity in streams. Journal of the North American Benthological Society 19 :573-592. Lake, P. S. 2003. Ecological effects of perturbation by drought in flowing waters. Freshwater Biology 48:1161-1172. Larned, S. T., and S. R. Santos. 2000. A late -and nutrient-limited periphyton and low order streams of Oahu, Hawaii. Hydrobiologia 432:101-111. Lau, D. C. P., K. M. Y. Leung, and D. Dudgeon. 2009. What does stable isotope analysis reveal about trophic relationships and the relative importance of allocht honous and autochthonous resources in tropical streams? A synthe tic study from Hong Kong. Freshwater Biology 54:127-141. Leitao, S., P. Pinto, and M. F. Brito. 2007. Spatial and temporal variabilit y of macroinvertebrate communities in two farmed Mediterra nean rice fields. Aquatic Ecology 41:373-386. Lorion, C. M., and B. P. Kennedy. 2009. Relationshi ps between deforestation, riparian forest buffers and benthic macroinvertebrates in ne otropical headwater streams. Freshwater Biology 54:156-180. Magoulick, D. D., and R. M. Kobza. 2003. The ro le of refugia for fishes during drought: a review and synthesis. Freshwater Biology 48:1186-1198. Malmqvist, B. 2002. Aquatic inve rtebrates in riverine lands capes. Freshwater Biology 47:679694.
210 Maltchik, L., and E. S. F. Medeiros. 2006. Cons ervation importance of semi-arid streams in north-eastern Brazil: implications of hydr ological disturbance and species diversity. Aquatic Conservation: Marine and Freshwater Ecosystems 16:665-677. March, J. G., and C. M. Pringle. 2003. Food web structure and basil reso urce utilization along a tropical island stream conti nuum, Puerto Rico. Biotropica 35 :84-93. Mathooko, J. M. 2002. The sizes, maturity stages a nd biomass of mayfly assemblages colonizing disturbed stream bed patches in cent ral Kenya. African Journal of Ecology 40:84-93. Matthaei, C. D., C. J. Arbuckle, and C. R. To wnsend. 2000. Stable surface stones as refugia for invertebrates during disturbance in the Ne w Zealand stream. Journal of the North American Benthological Society 19 :82-93. Matthews, W. J., and E. Marsh-Matthews. 2003. Effects of drought on fish across axes of space, time and ecological complexity. Freshwater Biology 48:1232-1253. McCune, B., and J. B. Grace. 2002. Analysis of Ecological Communities. MjM Software Design, Gleneden Beach, OR. Medeiros, E. S. F., and L. Maltchik. 2001. Fish assemblage stability in an intermittently flowing stream from the Brazilian semiarid region. Austral Ecology 26 :156-164. Melo, A. S., and C. G. Froehlich. 2001. Macroinv ertebrates in neotropical streams: richness patterns along a catchment in assemblage stru cture between 2 seasons. Journal of the North American Benthological Society 20 :1-16. Merritt, R. W., and K. W. Cummins, editors. 19 96. An Introduction to th e Aquatic Insects of North America. 3rd edition. Kendall/Hunt Publishing Co., Dubuque. Meyer, J. L., D. L. Strayer, J. B. Wallace, S. L. Eggert, G. S. Helfman, and N. E. Leonard. 2007. The contribution of headwater streams to biodi versity in river netw orks. Journal of the American Water Resources Association 43 :86-103. Miller, A. M., and S. W. Go lladay. 1996. Effects of spates and drying on macroinvertebrates assemblages often intermittent and a perennial prairie stream. Journal of the North American Benthological Society 15 :670-689. Minshall, G. W., J. T. Peterson, K. W. Cummins, T. L. Bott, J. R. Sedell, C. E. Cushing, and R. L. Vannote. 1983. Interbiome comparison of stream ecosystem dynamics. Ecological Monographs 53:2-25. Moran-Lopez, R., E. Da Silva, J. L. PerezBote, and C. C. Amado. 2006. Associations between fish assemblages and environmental factors fo r Mediterranean-type rivers during summer. Journal of Fish Biology 69 :1552-1569.
211 Murphy, M. L., C. P. Hawkins, and N. H. Anderson. 1981. Effects of canopy modification and accumulated sediment on stream communities. Transactions of the American Fisheries Society 110 :469-478. Murphy, P. G., and A. E. Lugo. 1986. Ecology of tr opical dry forest. Annua l Review of Ecology and Systematics 17 :67-88. Nessimian, J. L., and E. M. Venticinque. 2008. Landuse, habitat integrity, and aquatic insect assemblages in Central Amazonian streams. Hydrobiologia 614:117-131. Nolte, U., M. J. de Oliviera, and E. Sturs. 1997. Seasonal, discharge-driv en patterns of mayfly assemblages in intermittent Neotro pical stream. Freshwater Biology 37:333-343. Opler, P. A., G. W. Frankie, and H. G. Bake r. 1980. Comparative phenologi cal studies of treelet and shrub species in tropical we t and dry forests in the lowlan ds of Costa Rica. Journal of Ecology 68:167-188. Ostrand, K. G., and G. R. Wilde. 2002. Seasonal a nd spatial variation in a Prairie stream fish assemblage. Ecology of Freshwater Fish 11:137-149. OTS. 2009. Palo Verde Biological Stati on. Organization for a Tropical Studies, http://www.ots.ac.cr/index.php?option=com_ content&task=view&id=238&Item id=415 Viewed April 24, 2009. Paaby, P., A. Ramirez, and C. M. Pringle. 1998. The benthic macroinvertebrate community in Caribbean Costa Rican streams and the effect of two sampling methods. Revista de Biologia 46 :185-199. Pacific, C. I. a. N. E. C., Ltd.. 2002. Estudio del Proyecto de Desarroll o Rural de la Cuenca Media del Rio Tempisque. Volume 1: Informe Principal. Paller, M. H., W. L. Specht, and S. A. Dyer 2006. Effects of stream size on taxa richness and other commonly used benthic bioass essment metrics. Hydrobiologia 568:309-316. Paltridge, R. M., P. L. Dostine, C. L. Hum phrey, and A. J. Boulton. 1997. Macroinvertebrate recolonization after re-wetting of a tropi cal seasonally-flowing stream (Magela Creek, Northern Territory, Australia). Ma rine and Freshwater Research 48:633-645. Peran, A., J. Velasco, and A. Millan. 1999. Life cycle and secondary production of Caenis luctuosa (Ephemeroptera) in a semiarid stream (Southeast Spain). Hydrobiologia 400 :187194. Pineda, N. 2007. SENARA Manager, Personal Communication. Poff, N. L., J. D. Allan, M. B. Bain, J. R. Karr, K. L. Prestegaard, B. D. Richter, R. E. Sparks, and J. C. Stromberg. 1997. The na tural flow regime. BioScience 47:769-784.
212 Principe, R. E., M. F. Boccolini, and M. C. Corigliano. 2008. Structure and spatial-temporal dynamics of Chironomidae fauna (Diptera) in up land and lowland fluvial habitats of the Chocancharava River Basin (Argentina). International Review of Hydrobiology 93:342357. Pringle, C. M., and T. Hamazaki. 1997. Effects of fi shes on algal response to storms in a tropical stream. Ecology 78:2432-2444. Pringle, C. M., R. J. Naiman, G. Bretschko, J. R. Karr, M. W. Oswood, J. R. Webster, R. L. Welcomme, and M. J. Winterbour n. 1988. Patch dynamics in lotic systems: the stream as a mosaic. Journal of the North Am erican Benthological Society 7:503-524. Pringle, C. M., and A. Ramirez. 1998. Use of both benthic and drift sampling techniques to assess tropical stream of inve rtebrate communities along an al titudal gradient, Costa Rica. Freshwater Biology 39:359-373. Pritchard, G. 1996. The life history of a tropical dragonfly: Cora marina (Odonata: Polythoridae) in Guanacaste, Costa Rica. Journal of Tropical Ecology 12:573-581. Pusey, B. J., and A. H. Arthington. 2003. Importance of the riparian zone to the conservation and management of freshwater fish: a revi ew. Marine and Freshwater Research 54:1-16. Quinn, J. M., I. K. G. Boothroyd, and B. J. Sm ith. 2004. Riparian buffers mitigate the effects of pine plantation logging on New Zealand str eams 2. Invertebrate communities. Forest Ecology and Management 191:129-146. Quinn, J. M., A. B. Cooper, M. J. Stroud, and G. P. Burrell. 1997. Shade affects on stream periphyton and invertebrates: an experiment in streamside channels. New Zealand Journal of Marine and Freshwater Research 31:665-683. Ramirez, A., and C. M. Pringle. 1998a. Inverteb rate drift and benthic community dynamics in a lowland neotropical stream Costa Rica. Hydrobiologia 386 :19-26. Ramirez, A., and C. M. Pringle. 1998b. Stru cture and production of the benthic insect assemblage in a neotropical stream. Journal of the North American Benthological Society 17:443-463. Ramirez, A., and C. M. Pringle. 2001. Spatial a nd temporal patterns of invertebrate drift in streams draining a Neotropical landscape. Freshwater Biology 46:47-62. Ramirez, A., C. M. Pringle, and K. M. Wantzen. 2008. Tropical Stre am Conservation. Pages 285-304 in D. Dudgeon, editor. Tropical Stream Ecology. Academic Press, Amsterdam. Rizo-Patrn, F. 2003. Estudio de los arroza les del proyecto Tamarindo: Agroqumicos y macroinvertebrados bentnicos en relacin al Parque Nacional Palo Verde, Guanacaste, Costa Rica. Universidad Nacional, Heredia.
213 Robson, B. J., T. G. Matthews, P. R. Lind, and N. A. Thomas. 2008. Pathways for algal recolonization in seasonally-flowing streams. Freshwater Biology 53:2385-2401. Roldn, G. 1996. Gua para el estudio de los m acroinvertebrados acuticos del Departamento de Antioquia, Colombia. Universi dad de Antioquia, Colombia. Rossaro, B. 1991. Factors that determine Chirono midae species distribution and freshwaters. Bollettino di Zoologia 58 :281-286. Salas, M., and D. Dudgeon. 2001. Stable-isoto pe determination of mayfly (Insecta: Ephemeroptera) food sources in three tropical Asian stream s. Archiv fur Hydrobiologie 151:17-32. Salas, M., and D. Dudgeon. 2003. Life histories, production dynamics and resource utilisation of mayflies (Ephemeroptera) in two tropical Asian forest streams. Freshwater Biology 48:485-499. Santana, E. 2007 Personal Comm unication. COPEVICA Ditchrider. Shannon, J. P., D. W. Blinn, and L. E. Steven s. 1994. Trophic interactions and benthic animal community structure in the Colorado Ri ver, Arizona, USA. Freshwater Biology 31:213220. Singkran, N., and M. S. Meixler. 2008. Influences of habitat and land cover on fish distributions along attributed very to Lake Ontario, New York. Landscape Ecology 23 539-551. Smith, H., P. J. Wood, and J. Gunn. 2003. The in fluence of habitat structure and flow permanence on invertebrate communities in karst spring systems. Hydrobiologia 510 :5366. Stanford, J. A., and J. V. Ward. 2001. Revisiting the serial discontinuity concept. Regulated Rivers: Research & Management 17:303-310. Stanley, E. H., D. L. Buschman, A. J. Bou lton, N. B. Grimm, and S. G. Fisher. 1994. Invertebrate resistance and resilience to in termittency in the desert stream. American Midland Naturalist 131 :288-300. Steinman, A. D., and G. A. Lamberti. 1996. Biom ass and Pigments of Benthic Algae. Pages 295313 in F. R. Hauer and G. A. Lamberti, editors. Methods in Stream Ecology. Academic Press, Inc., San Diego. Storey, R. G., and D. R. Cowley. 1997. Recovery of three New Zealand rural streams as they pass through native forest reminants. Hydrobiologia 353:63-76. Takamura, N., T. Ito, R. Ueno, A. Ohtaka, I. Wa kana, M. Nakagawa, Y. Ueno, and H. Nakajima. 2009. Environmental gradients determining the distribution benthic macroinvertebrates in Lake Takkobu, Kushiro wetland, norther n Japan. Ecological Restoration 24:371-381.
214 Taylor, J. M., and J. H. Kennedy. 2006. Life Hi story and Secondary Production of Caenis latipennis (Ephemeroptera: Caenidae) in Honey Creek, Oklahoma. Annals of the Entomological Society of America 99:821-830. Tessier, C., A. Cattaneo, B. Pinel-Alloul, C. Hudon, and D. Borcard. 2008. Invertebrate communities and epiphythic biomass associ ated with metaphyton and emergent and submerged macrophytes in a large river. Aquatic Sciences 70 :10-20. Thorp, J. H., and A. P. Covich. 2001. Ecology and Classification of North American Freshwater Invertebrates. 2nd edition. A cademic Press, San Diego. Tomanova, S., E. Goitia, and J. Helesic. 2006. Tr ophic levels and functional feeding groups of macroinvertebrates in neotropical streams. Hydrobiologia 556 :251-264. Tornanoval, S., P. A. Tedesco, M. Campero, P. A. Van Damme, N. Moya, and T. Oberdorff. 2007. Longitudinal and altitudinal changes of macroinvertebrates functional feeding groups in neotropical streams: a test of th e River Continuum Concept. Fundamental and Applied Limnology 170:233-241. Townsend, C. R. 1989. The patch dynamics concep t of stream community ecology. Journal of the North American Benthological Society 8 :36-50. Urban, M. C., D. K. Skelly, D. Burchsted, W. Price, and S. Lowry. 2006. Stream communities across a rural-urban landscape grad ient. Diversity and Distribution 12:337-350. Uys, M. C., and J. H. O'Keeffe. 1997. Simple words and fuzzy zones: early directions for temporary river research in South Africa. Environmental Management 21:517-531. Vannote, R. L., G. W. Minshall, K. W. Cummi ns, J. R. Sedell, and C. E. Cushing. 1980. The river continuum concept. Canadian Jour nal of Fisheries and Aquatic Sciences 37:130-137. Wantzen, K. M., C. M. Yule, J. M. Mat hooko, and C. M. Pringle. 2008. Organic Matter Processing in Tropical Streams. Pages 44-65 in D. Dudgeon, editor. Tropical Stream Ecology. Academic Press, Amsterdam. Ward, A. F., and D. D. Williams. 1986. Longit udinal zonation and food of larval chironomids (Insecta: Diptera) along the c ourse of a river in temper ate Canada. Holarctic Ecology 9:4857. Ward, J. V. 1989. The four-dimensional nature of lotic ecosystems. Journal of the North American Benthological Society 8 :2-8. Warfe, D. M., and L. A. Barmuta. 2006. Hab itat structural complexity mediates food web dynamics in a freshwater macrophyte community. Oecologia 150 :141-154. Williams, D. D. 1996. Environmental constraints in temporary freshwaters and their consequences for the insect fauna. Journal of the North American Benthological Society 15:634-650.
215 Williams, D. D. 1997. Temporary ponds and their invertebrate communities. Aquatic Conservation: Marine and Freshwater Ecosystems 7:105-117. Williams, P., M. Whitfield, J. Biggs, S. Br ay, G. Fox, P. Nicolet, and D. Sear. 2003. Comparative biodiversity of rivers, streams, d itches and ponds in an agricultural landscape in Southern England. Biological Conservation 115 :329-341. Wilson, A. L., R. J. Watts, and M. M. Stevens. 2008. Effects of different management regimes on aquatic macroinvertebrate diversity in Aust ralian rice fields. Ecological Restoration 23:565-572. Wilzbach, M. A., K. W. Cummins, and J. D. Hall. 1986. Conference of habitat manipulations on interactions between cutthroat trout and invertebrate drift. Ecology 67:898-911. Wilzbach, M. A., B. C. Harvey, J. L. White, and R. J. Nakamoto. 2005. Effects of riparian canopy opening and salmon carcass addition on th e abundance and growth of resident salmonids. Canadian Journal of Fisheries and Aquatic Sciences 62:58-67. Winemiller, K. O., A. A. Agostinho, and E. P. Caramaschi. 2008. Fish Ecology in Tropical Streams. Pages 107-146 in D. Dudgeon, editor. Tropical Str eam Ecology. Academic Press, Amsterdam. Wisenden, B. D. 1994. Factors a ffecting reproductive success in free-ranging convict cichlids (Cichlasoma nigrofasciatum). Canadian Journal of Zoology 72:2177-2185. Wisenden, B. D. 1995. Reproductive-behavior of free-ranging convict cichlids, Cichlasoma nigrofasciatum Environmental Biology of Fishes 43:121-134. Wood, P. J., J. Gunn, H. Smith, and A. Abas-Kutty. 2005. Flow permanence and macroinvertebrate community diversity with in groundwater dominated headwater streams and springs. Hydrobiologia 545:55-64.
216 BIOGRAPHICAL SKETCH Suzanne M. Moellendorf was born in 1980 in Renton, W ashington, and was the youngest child of Guy and Natalie Moellendorf. She grew up on her grandparents old farmland and enjoyed trips to national parks throughout her childhood, which gave her a great appreciation for agriculture and nature. Suzanne graduated from Kentridge High School in 1999 then attended Wellesley College where she earned a Bachelor of Arts in biol ogy with a minor in economics in 2003. During her years at Wellesley College, she developed a d eep interest in water resources science and management as a result of thought-provoking co urses, diverse summer internships, and a semester abroad in Ecuador. Suzannes internsh ips included aiding a do ctoral student with a tropical watershed project near the Organization for Tropical Studi es La Selva Biological Station in Costa Rica in 2001, working on dam-related projects for American Rivers in Washington, D.C. in 2002, and carrying out water resources re search in the Atacama Desert in Chile through the University of Notre Dame and University of Nevada, Reno. Suzanne also carried out an independent project on natural resource use in an indigenous community in the Ecuadorian Andes in 2002 as part of her study abroad program. After graduation, Suzanne worked for the De schutes River Conservancy (DRC) in the Bend, Oregon where she was able to continue wo rking on water resources management issues. Her two years at the DRC (2003-2005) allowed he r to interact with diverse stakeholders throughout the Deschutes Basin, ta ught her an incredible amount about watershed management strategies and technologies, intr oduced her to irrigation canal proj ects, and inspired her to pursue a graduate school research project that would compliment and extend her work at the DRC. Fortunately, Suzanne found an exciting project th rough the University of Florida (UF) that allowed her to combine her interest in the inter action between irrigation a nd stream systems, as
217 well as her fondness for Latin America and knowle dge of the Spanish language. While in graduate school, Suzanne had the opportunity to help with Organizati on for Tropical Studies (OTS) courses: the 1st and 2nd Annual Wetland Restoration and Management Course for Latin American professionals and the Tropical Agro ecology Course for graduate students from the United States and Costa Rica. She participated in an Environmental Flows workshop in Costa Rica where she met inspiring leaders, as well. Suzannes field resear ch in Costa Rica was remarkable for the interactions that she had with students and staff at the University of Costa Rica (UCR) and OTS Palo Verde Biological St ation, as well as the strong friendship she developed with a Costa Rican family that she met by chance and ended up living with whenever working at UCR. Suzanne also had the unique opportunity to be a visiting researcher associated with the Department of Fisheries & Wildlife at Oregon St ate University (OSU). While at OSU, she interacted with and learned from a cohort of incredible freshwater scientists, participated in a variety of field projects, taught aquatic macroinvertebrate sampling labs and a graduate level seminar, and progressed with her dissertation. Suzanne received her Master of Science in the spring of 2007 and her Doctor of Philosophy in the summer of 2009, both from the Department of Environmental Engineering Sciences from the University of Florida. She plans to pursue work that allows her to address watershed management issues and to improve di ssemination of scientific information to the public.