Temporal Dynamics of Benthic Responses to Habitat Disturbance

Louisiana State University LSU Digital Commons

LSU Doctoral Dissertations Graduate School

2016 Temporal Dynamics of Benthic Responses to Habitat Disturbance in Coastal Plain Headwaters of Southwestern Louisiana Catherine Elizabeth Murphy Louisiana State University and Agricultural and Mechanical College

Follow this and additional works at: https://digitalcommons.lsu.edu/gradschool_dissertations Part of the Environmental Sciences Commons

Recommended Citation Murphy, Catherine Elizabeth, "Temporal Dynamics of Benthic Responses to Habitat Disturbance in Coastal Plain Headwaters of Southwestern Louisiana" (2016). LSU Doctoral Dissertations. 4402. https://digitalcommons.lsu.edu/gradschool_dissertations/4402

This Dissertation is brought to you for free and open access by the Graduate School at LSU Digital Commons. It has been accepted for inclusion in LSU Doctoral Dissertations by an authorized graduate school editor of LSU Digital Commons. For more information, please [email protected]. TEMPORAL DYNAMICS OF BENTHIC RESPONSES TO HABITAT DISTURBANCE IN COASTAL PLAIN HEADWATERS OF SOUTHWESTERN LOUISIANA

A Dissertation

Submitted to the Graduate Faculty of the Louisiana State University and Agricultural and Mechanical College in partial fulfillment of the requirements for the degree of Doctor of Philosophy

The School of Renewable Natural Resources

by Catherine Elizabeth Murphy B.S., University of Texas at Dallas, 1998 M.Ap.St., Louisiana State University, 2014 May 2017 DEDICATION

I dedicate this work in loving memory of my dear friend, Ralph Anthony Decuers (1962-2011), whose optimism and humor were surpassed only by his generosity of spirit. His unique perspective and words of encouragement ushered me through so many of life’s biggest challenges, including this one. Among other things, Ralph taught me three very important life lessons: 1) family comprises more than just those to whom you are related; 2) being a New Orleans Saints fan is about way more than football; and 3) modular arithmetic is a greatly underutilized resource for organization and decision-making. Having benefitted greatly from this wisdom, I am obligated to pay it forward.

ACKNOWLEDGEMENTS

First, I would like to thank my adviser, Dr. Mike Kaller, for taking a chance on an unconventional student and allowing me the freedom and time to conduct the research that interested and motivated me. I thank my Committee members, Dr. Ken Brown, Dr. Andy Nyman, Dr. Sammy King, and Dr. Bill Kelso, for their patience, support and encouragement. I am especially grateful to Dr. Kelso for challenging me to widen the lens through which I view ecology and for helping me to focus it when my logic became fuzzy.

Mr. Leonard Butter, Mrs. Hellen Hibbard and Mr. Layne Richard facilitated field collections by providing access to private property and taking time to share historical context about local natural resources. I am grateful for their stewardship and hospitality. A small army of graduate students, student workers and research associates assisted me in the lab and in the field. With professionalism and good humor, they stared for hours into pans of detritus to find bugs or waded through questionable water in less than ideal weather conditions. For their willingness to laugh with me and at me when things didn’t go my way and their ability to turn a disastrous day in the field into a hilarious story, I am deeply grateful to Michael

Baker, Sarah Benson, Dr. Chris Bonvillain, Will Budnick, Brooke Constant, Kelsey Daroca, Anna Evans,

Ali Fitzgerald, Eric Fontenot, Dr. Melissa Fries Gray, Raynie Harlan, Zach Herrington, Leticia

Kaczmarowski, Deb Kelly, Claire Labarbera, Ryan Leeson, Tyler Loeb, Peter Markos, Kaitlyn Matherne,

Kelsey McCray, Brett Miller, Devon Oliver, Tiffany Pasco, Christina Perez, Madeline Richard, Jessica

Sabo, Will Sheftall, Matt Songy, Rachel Tessier, Jose Vasquez, Angela Williamson, Will Young and Sarah

Zaunbrecher. I am humbled by the wisdom imparted to me by each of these researchers.

I thank Tommy Blanchard for his patience, good nature and organizational skills in analyzing my water samples. Cheryl Duplechain always managed to make me smile, no matter the circumstances of my day.

I owe a debt of gratitude to Deb Kelly for sharing her laboratory expertise and for rescuing me from myself on so many occasions. I am obliged to Dr. Ladorian Latin for her friendship, wit and study skills. Dr.

Maureen Corcoran, Audrey Harrison and Amanda Oliver provided invaluable technical and moral support during the writing process over dinners, microscope chats, and GIS maps, respectively. Emily Hahn graciously spent hours at the microscope and always brightened my day.

iii

I am indebted to Dr. Jack Killgore, Dr. Jan Hoover and Dr. Neil Douglas for sparking my interest in stream ecology and encouraging me to pursue graduate education. By enthusiastically sharing their wealth of knowledge and gently pushing me outside my comfort zone, each of them helped to shape my professional career and taught me to love my work. Dr. Chris Bonvillain was instrumental in every stage of this research and patiently helped me navigate through roadblocks along the way. His optimism, work ethic, leadership and generosity have inspired me to keep going on the many occasions when I thought that I could not. I am eternally grateful to my sisters (and technical advisers), Karen Murphy and Anne-

Marie Ireland, my mother, Rosemary H. Murphy, and my father, Bill Murphy, for teaching me to be resourceful and for unconditionally supporting my endeavors. My artistic directors, Mia and Abby Ireland, brightened my office and provided much needed comic relief. I hope they are inspired to dream big and keep moving forward. Finally, I thank my furry alarm clocks, Memphis and Dewey, for waking me every morning to remind me of what is really important.

TABLE OF CONTENTS

DEDICATION...... ii

ACKNOWLEDGEMENTS ...... iii

LIST OF TABLES ...... vii

LIST OF FIGURES ...... viii

ABSTRACT ...... x

CHAPTER 1: SEASONAL VARIABILITY OF IN-STREAM HABITAT CONDITION WITHIN THE PLEISTOCENE TERRACES OF SOUTHWESTERN LOUISIANA, WITH COMMENTS ABOUT DROUGHT ...... 1 Introduction ...... 1 Study Area ...... 2 Study Sites ...... 4 Methods ...... 10 Field ...... 10 Laboratory ...... 11 Results ...... 13 Discussion ...... 20 References ...... 22

CHAPTER 2: TAXONOMIC DISTINCTNESS AND SEASONAL STABILITY OF BENTHIC MACROINVERTEBRATE ASSEMBLAGES AMONG PLEISTOCENE TERRACES OF THE GULF COASTAL PLAIN OF SOUTHWESTERN LOUISIANA ...... 26 Introduction ...... 26 Methods ...... 29 Field Methods...... 30 Laboratory Methods ...... 32 Results ...... 36 Discussion ...... 49 References ...... 52

CHAPTER 3: LARVAL DEVELOPMENT OF CAENIS SP. (EPHEMEROPTERA: CAENIDAE) IN THE COASTAL TERRACES OF SOUTHWESTERN LOUISIANA ...... 57 Introduction ...... 57 Methods ...... 58 Results ...... 61 Discussion ...... 63 References ...... 69

CHAPTER 4: DISCUSSION...... 71

APPENDIX A: OVERVIEW AND INDIVIDUAL SITE MAPS WITH LAND COVER AND STREAM NETWORK WITHIN UPSTREAM PORTION OF SAMPLING WATERSHED ...... 75

VITA ...... 87

LIST OF TABLES

1.1 Abbreviations, names, river basins, ecoregions, size of watershed upstream of site and coordinates for 12 study sites. Ecoregion abbreviations are SCP South Central Plains, WGC Western Gulf Coastal Plain, STU Southern Tertiary Uplands, FLT Flatwoods, LLP Lafayette Loess Plains, NHG Northern Humid Gulf Coastal Prairies ...... 7

1.2 Summary statistics by terrace for physicochemical variables measured during 2011 ...... 15

1.3 Summary statistics by terrace for physicochemical variables measured during 2013 ...... 16

1.4 Variable component loadings for principal component analyses for each study year. |Loading|>0.3 indicated with shading. Variable names beginning with “L” have been log10- transformed to stabilize variance ...... 17

1.5 PERMANOVA results for pairwise test of differences between terraces by sampling event for each study year. Values shown represent average Euclidean distance between/within groups ...... 20

2.1 Abbreviations, names, river basins, ecoregions, size of watershed upstream of site and coordinates for 12 study sites. Ecoregion abbreviations are SCP South Central Plains, WGC Western Gulf Coastal Plain, STU Southern Tertiary Uplands, FLT Flatwoods, LLP Lafayette Loess Plains, NHG Northern Humid Gulf Coastal Prairies ...... 30

2.2 Summary statistics by terrace for instream habitat variables measured during study period ...... 37

2.3 Variable component loadings for principal component analysis of habitat variables. |Loading|>0.3 indicated with shading. Variable names beginning with “L” have been log10- transformed to stabilize variance ...... 38

2.4 List of insect taxa with potential habitat preference for either upland (U) or lowland (L) sites based on collections for this study. Taxa exhibiting no preference (i.e., habitat generalist) not listed...... 46

3.1 Water quality variables included in the regression analysis with summary statistics. *Supersaturated condition caused by dense filamentous algal mats ...... 66

vii

LIST OF FIGURES

1.1 Location of study stream sites within river basins and elevations (LIDAR, LA Atlas 2009). Study sites are designated with light blue dots ...... 4

1.2 Hydrograph (daily discharge m3 s-1) over study period (solid black line) versus mean daily discharge for period of record (dotted red line) at centrally located reference gage (Ouiska Chitto Creek near Oberlin, LA, USGS gage 08014500) ...... 5

1.3 Percent composition land cover (2011 NLCD; NRCS, FSA and RD 2016) within the watershed above each sampling site. Sites are grouped by terrace: West Fork Sixmile (WF6), East Fork Sixmile (EF6), Big Brushy (BB), Barber (BAR) and Hurricane (HUR) are Tertiary Uplands sites; West Fork Caney (WFC), Petite Passe (PP) and Indian (IND) are Flatwoods sites; Serpent (S165 and S99), Lacassine (LAC) and Grand Marais (GM) are Prairie sites ...... 8

1.4 Percent composition soil type (SSURGO; NRCS 2016) within the watershed above each sampling site. Sites are grouped by terrace: West Fork Sixmile (WF6), East Fork Sixmile (EF6), Big Brushy (BB), Barber (BAR) and Hurricane (HUR) are Tertiary Uplands sites; West Fork Caney (WFC), Petite Passe (PP) and Indian (IND) are Flatwoods sites; Serpent (S165 and S99), Lacassine (LAC) and Grand Marais (GM) are Prairie sites ...... 9

1.5 Change in terrain elevation (i.e., minimum versus maximum meters elevation, NAVD88) within the watershed above each sampling site. West Fork Sixmile (WF6), East Fork Sixmile (EF6), Big Brushy (BB), Barber (BAR) and Hurricane (HUR) are Tertiary Uplands sites. West Fork Caney (WFC), Petite Passe (PP) and Indian (IND) are Flatwoods sites. Serpent (S165 and S99), Lacassine (LAC) and Grand Marais (GM) are Prairie sites ...... 10

1.6 Box and whisker plots (median, 25th and 75th quartiles, and range) of overall means by terrace for size of dominant substrate, log10-turbidity, percent overstory cover, and log10- count of woody debris ...... 14

1.7 Principal component bi-plot (PC1 and PC2 together explain 42.3% of the variation) of 2011 physicochemical variables by geologic terrace. Vector direction and length indicate sign and magnitude of contribution to axis, respectively. Variable names preceded by “L” indicate log10-transformation ...... 18

1.8 Principal component bi-plot (PC1 and PC2 together explain 48.2% of the variation) of 2013 physicochemical variables by geologic terrace. Vector direction and length indicate sign and magnitude of contribution to axis, respectively. Variable names preceded by “L” indicate log10-transformation ...... 19

2.1 Map of sampling sites within southwestern Louisiana showing stream network within upstream portion of watershed. West Fork Sixmile (WF6), East Fork Sixmile (EF6), Big Brushy (BB), Barber (BAR) and Hurricane (HUR) are Tertiary Uplands sites. West Fork Caney (WFC), Petite Passe (PP) and Indian (IND) are Flatwoods sites. Serpent (S165 and S99), Lacassine (LAC) and Grand Marais (GM) are Prairie sites ...... 31

2.2 Change in relative abundances of major taxonomic groups across sampling dates by geologic terrace ...... 36

viii

2.3 Principal component bi-plot (i.e., PC1 and PC2 explain 42.3% of variation, combined) of habitat variables by geologic terrace. Vector direction and length indicate sign and magnitude of contribution to axis, respectively. Variable names preceded by “L” indicate log10-transformation ...... 39

2.4 Non-metric multidimensional scaling ordination 2-D bi-plot of Jaccard similarities among samples (presence/absence data) by geologic terrace ...... 40

2.5 Seasonal variability of mean 4th-root transformed individual Diptera taxa within terraces ...... 41

2.6 Seasonal variability of mean 4th-root transformed individual Coleoptera taxa within terraces ...... 42

2.7 Seasonal variability of mean 4th-root transformed individual Odonata taxa within terraces ...... 42

2.8 Seasonal variability of mean 4th-root transformed individual Ephemeroptera taxa within terraces ...... 43

2.9 Seasonal variability of mean 4th-root transformed individual Plecoptera taxa within terraces ...... 44

2.10 Seasonal variability of mean 4th-root transformed individual Trichoptera taxa within terraces ...... 44

2.11 Seriated shade plot (presence/absence) of individual taxa (rows) within macroinvertebrate assemblages across samples (columns). Taxa are ordered according to similarity across samples using Whittaker’s Index of Association. Samples are ordered according to Jaccard similarity across taxa ...... 45

2.12 Proportion of functional feeding group represented by insect taxa (in wood samples) within each geologic terrace. Error bars indicate plus one standard error for the mean proportion. Bars with different letters are significantly different within a trophic group ...... 47

2.13 Proportion of functional feeding group represented by insect taxa (in sediment samples) within each geologic terrace. Error bars indicate plus one standard error for the mean proportion. Bars with different letters are significantly different within a trophic group ...... 48

3.1 Map of study area showing individual watersheds at 12 stream sites ...... 59

3.2 Model-adjusted mean head capsule width within instar development classes for Caenis sp. larvae collected in streams of Prairie and combined Flatwoods/Uplands terraces ...... 62

3.3 Frequency distribution of Caenis sp. head capsule widths by sampling date ...... 64

3.4 Seasonal frequency of larval instar development classes I-V during sampling period ...... 65

ABSTRACT

Weak biotic responses to habitat gradients within Northern Gulf of Mexico streams have been attributed to spatial and temporal variability. Landscape and in-stream habitat descriptions are presented for watersheds within Pleistocene terraces of the Coastal Plains geomorphic province of Louisiana, USA.

Geologic influences on stream habitat were inferred by comparing multivariate ordinations on physicochemical measurements between terraces. Seasonal variability was assessed during a drought year (2011) and a typical water year (2013). Within coastal plains of Louisiana, stream condition was more similar within terraces than within river basins. Permutational MANOVA models indicated significantly different stream habitat between Uplands and Prairie, with intermediate habitat in Flatwoods.

Seasonal differences were detected more frequently during normal flow condition, suggesting that baseflow impacts habitat heterogeneity between adjacent terraces.

Macroinvertebrates were collected throughout a drought year at stream sites stratified among coastal plain terraces to quantify spatial and temporal variability and identify functional habitat gradients.

Macroinvertebrate assemblages differed between Uplands and Prairie terraces, especially regarding insect taxa, which were associated with better water quality and structurally complex habitat. Drought and other disturbances selected against lotic taxa expected in the intermediate Flatwoods terrace.

Widening the lateral scope of the study landscape helped identify habitat thresholds and define regional habitat preference of individual taxa. Aquatic habitat improvement in Prairie terrace bayous should include restoring baseflow, increasing structural complexity and protecting macroinvertebrate source populations in the Uplands.

Aquatic insect larvae are important bio-indicators and flexible life histories of many taxa may reflect regional or seasonal variability in environmental conditions. Larval development and reproductive strategy inferred from seasonal size distributions are presented for specimens of Caenis sp.

(Ephemeroptera: Caenidae) in the coastal plain terraces of Louisiana. Influence of regional drought, landscape features and water quality on growth rate, terminal size and voltinism are examined. Caenis sp. in subtropical Louisiana exhibited bivoltine emergences in November and July. Size at instar development class did not differ by terrace, but was influenced by local water quality (e.g.,

x orthophosphate concentration, specific conductance and biochemical oxygen demand). Maintenance of baseflow during drought enhanced abundance of Caenis larvae in streams with chronic disturbance from agriculture.

CHAPTER 1: SEASONAL VARIABILITY OF IN-STREAM HABITAT CONDITION WITHIN THE PLEISTOCENE TERRACES OF SOUTHWESTERN LOUISIANA, WITH COMMENTS ABOUT DROUGHT

Introduction

A stream is structured by the landscape it carves (Hynes 1970), a relationship that holds true even for intermittent headwaters (Fritz and Feminella 2011). Every facet of the watershed, its topography, geology, soils, hydrology and vegetation, contributes to its value as aquatic habitat (Vondracek et al.

2005). Physical, chemical and biological processes in streams respond to landscape level changes through sediment and nutrient loading, increased temperature, channel incision, and loss of woody debris

(Gardiner et al. 2009). Land use metrics (e.g., percent composition land cover) are often used to quantify impacts to stream biota, especially in areas with extensive agriculture (Allan 2004), but responses are sometimes only weakly correlated (Williams et al. 2005, Riseng 2011, Fitzgerald 2012). Often, these results are confounded by the fact that land use and vegetation (e.g., forest cover or cultivated crops) correlate strongly with geology and soil structure, such that their contributions to ecosystem structure cannot be partitioned (Vondracek et al. 2005). In addition, land use interacts with other anthropogenic drivers that affect stream condition, including climate change (Allan 2004).

Many coastal plains and alluvial valleys have been transformed by agriculture (Benke and Cushing 2004), which is not surprising given their gentle slopes, fertile soils and shallow aquifers. Impacts to surface waters in these landscapes are often pervasive and cumulative, with physical habitat degradation surpassing water quality as the primary stressor to many watersheds (Shields et al. 2006). In the Gulf of

Mexico coastal plain of Louisiana, agricultural impacts are more recent by almost a century compared with other regions, such as the Yazoo Basin in Mississippi, where suspended sediment load in rivers is still being influenced by historical soil erosion that occurred prior to implementation of conservation practices (Vidrine et al. 2004, Cormier 2007, Merten et al. 2016). Yet, in Southwest Louisiana streams, habitat generalists and taxa tolerant of poor water quality tend to dominate both fish and macroinvertebrate assemblages (Kaller and Kelso 2007, Fitzgerald 2012, Justus et al. 2014), indicating either relatively recent and severe impacts from changes in land use or long-term adaptation to low- gradient stream systems that transition routinely between lotic and lentic conditions (or a combination of these). The area originally contained wet prairies (Skrobialowski et al. 2004), many of which were

1 drained to create pasture (Cormier 2007), so surface waters were probably rarely lotic, although interaction with groundwater has been anthropogenically altered since 1900 (Borrok and Broussard

2016).

Evidence from previous biotic investigations across the region suggests that ecoregions, which are primarily delineated by geomorphic characteristics in Louisiana (Daigle et al. 2006), and river basins structure different aspects of the aquatic communities in streams (Kaller et al. 2013). To better understand the interactions of these stream systems with the landscape of the Pleistocene terraces through which they flow, the study objectives were 1) compare in-stream habitat features at spatial (i.e., terrace and basin) and temporal (i.e., seasonally during two distinct water years) scales; 2) discuss drought influences on the strength of the relationship between stream and landscape; and 3) describe watershed characteristics as they relate to underlying geology and hydrology to inform future use of land use metrics in bio-assessment of streams.

Study area

Southwestern Louisiana is located on some of the youngest land in the North American continent. The

Pleistocene terraces consist of relatively flat remnant flood plains forming parallel belts along the Gulf of

Mexico coastline (Louisiana Geological Survey 2008). These areas lie within the Coastal Plains geomorphic province, which averages 1,020 to 1,350 mm of precipitation annually (McNab et al. 2007).

Above the chenier ridges along the coast lies the Prairie Terrace, characterized by deep, poorly drained soils formed by loamy and clayey fluviomarine deposits on gentle slopes of less than 0.5 m/km.

Dominant soils composing these complexes include the Crowley, Vidrine and Mowata series (NRCS Soil

Survey 2016). Originally a rugged prairie with intermittent bayous and wetlands, this area was a hindrance to overland travel by the Spanish and French explorers who divided and developed the surrounding land for their respective crowns. Consequently, settlement of the area by Europeans and the concomitant anthropogenic modification of the landscape were delayed until the early 19th century

(Vidrine et al. 2004, Cormier 2007). At first, grasslands and drained wetlands were used primarily to graze cattle, while cypress, pine and hardwoods from the area were harvested for lumber and shipbuilding.

More recently, the slowly permeable clay soils have facilitated widespread rice cultivation, which in many

2 areas is seasonally rotated with crayfish aquaculture and soybean production (McClain et al. 2007). Rice irrigation constitutes greater than 75% of groundwater use (approximately 140 Mgal/day) within the parishes of the Prairie Terrace (White et al. 2014). Major waterways include the Mermentau River and its tributary bayous, as well as a few bayous of the Vermilion-Teche Basin along the northeastern periphery.

Soils of the Vermilion-Teche are composed mainly of silty loam with well drained loamy alluvium of the

Rexor series to the north and poorly drained loess to the south (NRCS 2016).

North and west of the Prairie Terrace are the more intermediate Flatwoods, which range in elevation from

4 to 66 m above mean sea level with poorly to moderately drained loamy soil complexes of the

Beauregard, Caddo, Kinder, Messer and Guyton series (Daigle et al. 2006, NRCS 2016). Land use in this southern part of the South Central Plains ecoregion is a mix of silviculture, pasture and cultivated crops (McNab et al. 2007). Groundwater withdrawals (26-30 Mgal/day) are primarily used for rice irrigation and wood production (Prakken et al. 2012a,b). As further demonstration of the transitional nature of this area, surface waters in the Flatwoods are split between the Calcasieu and Mermentau River

Basins.

The Southern Tertiary Uplands to the north of the Flatwoods contain hilly areas formed by extensive dissection of bedrock strata (Daigle et al. 2006). Soils are primarily fine sands and loams, including

Ruston, Briley, Malbis, Betis, Smithdale and others, intermixed with complexes of the poorly drained

Guyton series (NRCS 2016). Elevation ranges from 15 to 140 m above mean sea level and land uses include forest, pine plantations, forested wetlands and some pasture. Most notably, this region contains several of Louisiana’s scenic rivers (e.g., Spring Creek in Rapides Parish, Whiskey Chitto Creek in

Vernon Parish), as well as tracts of Kisatchie National Forest on which longleaf pine (Pinus palustris,

Miller) and the endangered red-cockaded woodpecker (Picoides borealis, Vieillot) are managed through the use of prescribed burns and other conservation efforts (Daigle et al. 2006). Groundwater withdrawals

(6.5-34 Mgal/day) are mainly for drinking water (Prakken et al. 2012c). Major waterways include headwaters of the Calcasieu River to the west and upland tributaries of the Vermilion-Teche system to the east.

Study sites

Channel features and water quality parameters were measured at each of 12 study streams on 16 different occasions (i.e., 8 dates during Aug 2010 – Nov 2011 and 8 dates during Aug 2012 – Aug 2013).

Streams were stratified among river basins and geologic terraces (based on ecoregion delineations of

Daigle et al. 2006), and I selected specific sampling reaches within these streams that could be waded during all seasons (Figure 1.1). Year 1 was characterized by historic drought conditions in southwest

Figure 1.1. Locations of study stream sites within river basins and elevations (LIDAR, LA Atlas 2009). Study sites are designated with light blue dots.

Louisiana with the third driest hydrologic year (Oct 2010 - Sep 2011) in the 1895-2011 record (NOAA

2012), so time was allotted between sampling years to allow surface waters to return to more typical flow conditions (Figure 1.2). All 12 sites were visited within a two-week period during each of the 16 sampling events, as hydrographic conditions permitted. The mean (± SD) number of days between sampling events at a site was 65 (± 10) for Year 1 and 53 (± 19) for Year 2. Generally, three day trips were required to complete a sampling event with four sites visited per trip. When feasible, care was taken to visit sites in the same order on each trip, minimizing the introduction of variability from diel fluctuations in water quality parameters.

Figure 1.2. Hydrograph (daily discharge m3 s-1) over study period (solid black line) versus mean daily discharge for period of record (dotted red line) at centrally located reference gage (Ouiska Chitto Creek near Oberlin, LA, USGS gage 08014500).

Watershed boundaries were created with the DEM to Raster, Mosaic Raster, Fill, Flow Direction, Flow

Accumulation, Snap Pour Point, Watershed, and Raster to Polygon tools in Arc GIS (ESRI 10.2,

Redlands, CA). Bare earth Digital Elevation Models (DEM) downloaded from LA Atlas (2009) were used as the basis for watershed generation. DEM files were created at a 5-m resolution from 2003 - 2005 Light

Detection and Ranging (LiDAR) data as part of a joint project between the state of Louisiana and Federal

Emergency Management Agency (Watershed Concepts 2005). National hydrography data stream lines were created from USGS hydrologic digital line graph files and updated with local data, where available

(USGS 2013). Soil data were compiled from the National Cooperative Soil Survey (Soil Survey Staff,

NRCS 2016). The 2011 National Landcover dataset classified the U. S. land cover, as depicted in 2011

Landsat satellite images, into 16 classes at a spatial resolution of 30x30m (Homer et al. 2015). An overview map of the study area and individual watershed maps with stream networks and land cover are provided in Appendix A.

Site names, site abbreviations used throughout this document, river basin, ecoregion classifications

(Levels III and IV), watershed area upstream of study site, latitude and longitude are provided in Table

1.1. Note that Bayou Serpent was sampled in two different locations, such that the watershed for Upper

Bayou Serpent at the headwaters was entirely contained within the watershed for Middle Serpent Bayou.

These two sites demonstrated very different habitat characteristics and afforded a unique opportunity for at least a qualitative comparison, acknowledging their spatial dependence as well as the disparity in watershed size.

Percent composition of land cover within each upstream watershed is shown in Figure 1.3. Dominant land covers within watersheds located in the Southern Tertiary Uplands (WF6, EF6, BB, BAR, HUR) were evergreen forest, shrub and mixed forest, while Northern Humid Gulf Coastal Prairies sites (LAC, GM,

S99 and S165) drained lands dominated by cultivated crops and pasture. Watersheds within the

Flatwoods (WFC and IND) and Lafayette Loess Plains (PP), which share a border near this site, comprised land covers similar to either upland or prairie patterns. Percent composition of soil type with study watersheds demonstrated a clearer delineation of the geologic terraces, with upland watersheds containing mostly fine sands and loamy soils, and both flatwoods and prairie watersheds dominated by

6 clay complexes and silt loam (Figure 1.4). Apparent differences in the terrain among the study watersheds can be seen by comparing the maximum and minimum elevations within each (Figure 1.5).

Table 1.1. Abbreviations, names, river basins, ecoregions, size of watershed upstream of site and coordinates for 12 study sites. Ecoregion abbreviations are SCP South Central Plains, WGC Western Gulf Coastal Plain, STU Southern Tertiary Uplands, FLT Flatwoods, LLP Lafayette Loess Plains, NHG Northern Humid Gulf Coastal Prairies.

Ecoregion Watershed Site Stream Name Basin Latitude Longitude III / IV Area km2 West Fork WF6 Calcasieu SCP/STU 47.01 31° 2'21.76"N 93° 0'15.28"W Sixmile Creek East Fork EF6 Calcasieu SCP/STU 50.15 31° 2'6.04"N 92°58'0.25"W Sixmile Creek Big Brushy BB Calcasieu SCP/STU 28.18 31° 1'10.09"N 92°53'59.42"W Creek Vermilion- BAR Barber Creek SCP/STU 23.99 31° 0'7.08"N 92°33'49.26"W Teche Hurricane Vermilion- HUR SCP/STU 17.56 31° 0'47.07"N 92°30'59.32"W Creek Teche West Fork WFC Mermentau SCP/FLT 15.07 30°44'42.22"N 92°37'9.49"W Caney Vermilion- PP Petite Passe WGC/LLP 37.65 30°42'17.54"N 92°11'32.52"W Teche IND Indian Bayou Calcasieu SCP/FLT 12.91 30°26'30.33"N 93°13'55.49"W Middle Bayou S165 Calcasieu WGC/NHG 279.62 30°23'19.06"N 92°54'23.42"W Serpent Upper Bayou S99 Calcasieu WGC/NHG 44.77 30°28'1.06"N 92°47'25.00"W Serpent East Bayou LAC Mermentau WGC/NHG 57.80 30°15'21.21"N 92°49'17.09"W Lacassine Bayou Grand GM Mermentau WGC/NHG 19.26 30°21'53.23"N 92°41'36.50"W Marais

Figure 1.3. Percent composition land cover (2011 NLCD; NRCS, FSA and RD 2016) within the watershed above each sampling site. Sites are grouped by terrace: West Fork Sixmile (WF6), East Fork Sixmile (EF6), Big Brushy (BB), Barber (BAR) and Hurricane (HUR) are Tertiary Uplands sites; West Fork Caney (WFC), Petite Passe (PP) and Indian (IND) are Flatwoods sites; Serpent (S165 and S99), Lacassine (LAC) and Grand Marais (GM) are Prairie sites.

Figure 1.4. Percent composition soil type (SSURGO; NRCS 2016) within the watershed above each sampling site. Sites are grouped by terrace: West Fork Sixmile (WF6), East Fork Sixmile (EF6), Big Brushy (BB), Barber (BAR) and Hurricane (HUR) are Tertiary Uplands sites; West Fork Caney (WFC), Petite Passe (PP) and Indian (IND) are Flatwoods sites; Serpent (S165 and S99), Lacassine (LAC) and Grand Marais (GM) are Prairie sites.

Figure 1.5. Change in terrain elevation (i.e., minimum versus maximum meters elevation, NAVD88) within the watershed above each sampling site. West Fork Sixmile (WF6), East Fork Sixmile (EF6), Big Brushy (BB), Barber (BAR) and Hurricane (HUR) are Tertiary Uplands sites. West Fork Caney (WFC), Petite Passe (PP) and Indian (IND) are Flatwoods sites. Serpent (S165 and S99), Lacassine (LAC) and Grand Marais (GM) are Prairie sites.

Methods

Field

I established 50-m study reaches at each stream site, and before sediment was disturbed by wading, I collected a 1-l water sample from the middle of the water column in a clean, dark bottle, capped underwater and packed in ice for transport to the laboratory. In situ measurements of water temperature, dissolved oxygen, pH, conductivity, and turbidity were also recorded with a handheld multi-parameter water quality sonde (YSI 650MDS unit with 6820 V2-1 sonde, Yellow Springs, Ohio, USA). I placed 5

10 cross-channel transects equidistantly across each reach, as measured with a meter tape, and marked each transect with flags. At each transect, wetted width was recorded and divided into three equal segments, oriented from left bank to right bank looking upstream. At the midpoint of each segment, water velocity and depth were measured with a FlowTracker Handheld 2D acoustic Doppler velocimeter

(SonTek/Xylem Inc., San Diego, California, USA) mounted on a 1.2 m top-setting wading rod (Marsh-

McBirney/Hach Co., Loveland, Colorado, USA). Dominant substrate type was noted (i.e., ordinal index based on particle size) within each segment and pieces of woody debris were counted within a 0.5 m radius about the wading rod. Additionally, percent overstory cover (measured with a Model-C concave spherical densiometer; Forestry Suppliers, Inc., Jackson, Mississippi, USA), degree of channel incision

(i.e., ordinal index based on floodplain width), approximate bank height, and dominant vegetation type

(i.e., ordinal index based on complexity/height) were recorded at transects 1, 3 and 5,. Any anomalies with potential to affect the physical, chemical or biological measurements at a site were noted and/or photographed (e.g., channel obstructions, atypical water odors, or rotting animal carcasses).

Laboratory

Stream water samples were allowed to warm to room temperature (i.e., 20-22°C) in the laboratory before analyses were performed. Carbonaceous biochemical oxygen demand (i.e., with nitrification inhibitor added) was measured in duplicate samples of stream water incubated at 20°C for at least 20 days with a

Thermo Orion model 850A+ dissolved oxygen meter (Thermo Fisher Scientific, Waltham, Massachusetts,

USA) per American Public Health Association standard method 5210B (APHA 2005). Fecal coliforms were measured in each sample following APHA standard method 9222A. Diluted stream water was vacuum-filtered onto sterile 47-mm, gridded 0.45-µ nitrocellulose filters that were placed in padded sterile petri plates to which 2 mL of m-ColiBlue24TM broth (EMD Millipore, Billerica, Massachusetts, USA) was added. Plates were inverted, sealed in water tight bags and incubated in a 35°C water bath for 24 hours.

Blue and red colony forming units (CFU) corresponding to Escherichia coli and total coliforms, respectively, were enumerated on duplicate plates of two different dilutions for each sample. Because coliform counts varied by site and season, dilutions were adjusted to ensure that one would contain at least 50 CFU. To quantify algal abundance in water samples, 100 mL of stream water was filtered onto

47-mm 0.7-µ retention GF/F glass microfiber filters, which were then folded into squares of aluminum foil, labeled and frozen for chlorophyll analysis. Chlorophyll a concentration was measured at Louisiana State

University’s Department of Oceanography and Coastal Sciences Wetland Biogeochemistry Analytical

Services with fluorescence detection (TD-700 Fluorometer, Turner Designs, Sunnyvale, California, USA) following EPA Method 445.0 (Arar and Collins 1997). Finally, concentrations of nitrate (Cadmium

Reduction Method 8192), nitrite (Diazotization Method 8507), N-ammonia (Salicylate Method 8155) and orthophosphate (Ascorbic Acid Method 8048) were measured on a Hach DR/2500 spectrophotometer

(Hach Co., Loveland, Colorado, USA). These parameters (i.e., cBOD, coliforms, chlorophyll a, and nutrients) were measured for every site on all sampling dates during both years. N-ammonia was removed from the final variable set, however, because of missing observations resulting from over- and under-range values.

Statistical analyses were performed separately for 2011 and 2013 to qualitatively assess the influence of drought between the data sets. Habitat variables that were strongly right-skewed were log10-transformed to stabilize variance. Variables were also normalized because measurement units were unique to each variable. Ordination via principal component analysis (PCA) was performed on normalized habitat variables to determine data dimensionality, reduce contributions to variability from redundant variables, and identify habitat features that explained the most variability in the dataset. Ordinations via principal coordinates analysis and non-metric multidimensional scaling were also examined, but all produced similar patterns and PCA eigenvectors provided direct relationships with original variables. All multivariate analyses were performed with PRIMER v.7 and PERMANOVA+ software (PRIMER-E Ltd.,

Plymouth, UK). Habitat measurements were taken at the same 12 locations at regular intervals throughout the study period. Therefore, landscape level differences in stream habitat were tested with permutational MANOVA (henceforth, PERMANOVA), with fixed effects of terrace and time (i.e., sampling event 1-8 for each study year), and site within terrace as a nested random effect. Significance of the pseudo-F statistic for Type III sums of squares was interpreted as permutation p-values <0.05. Type III sums of squares were used despite the unbalanced design in order to produce the most conservative results. For pairwise tests of between terrace differences at various time points, Monte Carlo tests were used to generate p-values because the number of unique permutations was generally less than 100,

12 making permutation p-values less reliable (Anderson et al. 2008). All models were executed with 9999 permutations and were calculated based on the matrix of Euclidean distances between samples for each study year.

Results

Qualitative assessment of watershed characteristics demonstrated that land cover was primarily structured by soil type and terrain, with the Flatwoods terrace (i.e., sites WFC, PP, IND) representing a transitional zone with some features of both the Uplands and Prairies (Figures 1.3-1.5). For reference, maps illustrating stream networks and land cover within individual watersheds are provided in Appendix

A. Summary statistics by terrace and study year are listed for each habitat variable in Tables 1.2 and 1.3.

Alignment of in-stream habitat and terrace geomorphology was evident from the means of individual habitat variables such as turbidity, woody debris count, size of dominant substrate and percent overstory cover (Figure 1.6).

Numerous examples of landscape influence were evident based on habitat variable means (Tables 1.2 and 1.3) viewed in the context of watershed features (Figures 1.3-1.5). Clay soils are not easily erodible and bayous in watersheds of the Prairie terrace had high, steep banks (>2 m) and wide, stable channels

(mean±SD 7.0±0.86m). Extensive development of cropland has left water bodies with little or no riparian vegetation, affecting overstory cover (41.8±23.9%), woody debris count (1.7±1.4 m-2) and water temperature (22.9±6.1°C). Stream networks within the Prairie appeared web-like due to extensive anthropogenic modification for drainage of rice ponds (see watershed map for Bayou Serpent, Appendix

A). The impact of this repurposing of surface waters throughout the region was reflected in observed water quality. Compared to streams in the other terraces, Prairie bayous had more autochthonous production (i.e., inferred from higher chlorophyll a concentrations), and higher specific conductance, turbidity, and concentrations of all nutrients measured (e.g., largest values in 2013 when precipitation was more typical). In addition, because of their larger channel dimensions, Prairie bayous routinely carried higher discharge, despite having generally low flow velocities from a lack of stream gradient. Comparison of sites on Bayou Serpent revealed that, as expected, water velocity (mean±SD, m s-1) was greater in the

Middle site (with the largest watershed), during 2011 (0.224 ±0.08 for Middle; 0.019 ±0.022 for Upper)

Figure 1.6. Box and whisker plots (median, 25th and 75th quartiles, and range) of overall means by terrace for size of dominant substrate, log10-turbidity, percent overstory cover, and log10- count of woody debris.

Table 1.2. Summary statistics by terrace for physicochemical variables measured during 2011.

Uplands Flatwoods Prairie 2011 Mean ±SD Mean ±SD Mean ±SD Temperature °C 18.24 ±7.30 20.00 ±8.57 20.46 ±8.66

Dissolved O2 mg/L 6.12 ±2.56 5.12 ±4.55 5.54 ±2.33 Specific Cond. mS/cm 0.057 ±0.028 0.259 ±0.172 0.321 ±0.177 pH 7.27 ±0.73 7.56 ±0.64 7.60 ±0.50 Turbidity NTU 19.6 ±27.3 91.5 ±76.7 220.2 ±356.1 Wood count per m2 6.15 ±4.82 9.39 ±8.60 2.51 ±3.38 Dom. Substrate Index 2.23 ±0.41 0.59 ±0.47 1.08 ±0.52 Bank Height m 1.68 ±0.47 1.77 ±0.36 2.74 ±0.61 Dom. Bank Veg. Index 2.78 ±0.37 2.48 ±0.44 2.41 ±0.47 Channel Incision Index 1.93 ±0.97 1.00 ±0.52 1.59 ±0.50 % Overstory cover 80.64 ±15.25 67.23 ±17.81 41.71 ±22.71 Wetted width m 4.90 ±1.27 4.71 ±1.20 6.88 ±0.83 Water depth m 0.319 ±0.168 0.442 ±0.114 0.441 ±0.163 Velocity m/s 0.103 ±0.102 0.007 ±0.016 0.072 ±0.101 Discharge m3/s 0.131 ±0.174 0.016 ±0.039 0.177 ±0.274

cBOD20 mg/L 3.556 ±2.514 7.906 ±3.269 7.888 ±4.971 Chlorophyll a ug/L 2.341 ±3.622 6.322 ±10.096 9.720 ±11.101 Nitrite mg/L 0.003 ±0.004 0.007 ±0.013 0.017 ±0.020 Nitrate mg/L 0.015 ±0.023 0.005 ±0.012 0.020 ±0.024 Orthophosphate mg/L 0.191 ±0.243 0.579 ±0.422 0.614 ±0.466 Total coliform CFU/100mL 1955 ±2042 3620 ±5114 7488 ±6865 Fecal coliform CFU/100mL 133 ±215 183 ±367 353 ±483 N Ammonia mg/L 0.035 ±0.062 0.119 ±0.114 0.221 ±0.149 and 2013 (0.308 ±0.080 for Middle; 0.038 ±0.067 for Upper). Size aside, the most conspicuous difference between the Middle Serpent Bayou site and all of the others was the presence of numerous Unionid mussel species forming what Vidrine et al. (2004) described as “an elaborate reef community” that

“cobbled the bottom” of this shallow, perennial stream and supported a diverse benthic community.

Water availability may also be a determining factor in cultivated crop land cover in the Prairie terrace.

The Chicot aquifer is shallow in many locations (61-213 m below ground surface, Borrok and Broussard

2016), which facilitates pumping of groundwater for irrigation. Saltwater intrusion is a concern in these coastal watersheds, as this shallow aquifer is currently being overdrafted by 1.3 million m3 per day, primarily to flood rice fields. Increased groundwater use (e.g., in response to drought) may accelerate interaction with the deeper, saltier Evangeline aquifer underlying the Chicot aquifer (Borrok and

Broussard 2016). Skrobialowski et al. (2004) observed increased groundwater demand during the

Table 1.3. Summary statistics by terrace for physicochemical variables measured during 2013.

Uplands Flatwoods Prairie 2013 Mean ±SD Mean ±SD Mean ±SD Temperature °C 19.70 ±5.08 20.88 ±6.45 22.86 ±6.13

Dissolved O2 mg/L 7.50 ±2.13 3.10 ±2.26 5.01 ±2.20 Specific Cond. mS/cm 0.065 ±0.042 0.239 ±0.126 0.251 ±0.095 pH 6.95 ±0.61 7.23 ±0.39 7.26 ±0.41 Turbidity NTU 14.0 ±14.1 64.2 ±51.5 266.0 ±356.2 Wood count per m2 4.86 ±4.82 6.20 ±8.63 1.69 ±1.42 Dom. Substrate Index 2.25 ±0.39 0.36 ±0.41 0.65 ±0.67 Bank Height m 1.55 ±0.38 1.49 ±0.39 2.24 ±0.46 Dom. Bank Veg. Index 2.68 ±0.38 2.42 ±0.42 2.42 ±0.44 Channel Incision Index 0.91 ±0.67 0.58 ±0.64 0.91 ±0.37 % Overstory cover 84.05 ±11.83 73.45 ±12.70 41.76 ±23.91 Wetted width m 5.40 ±1.38 5.02 ±0.97 7.02 ±0.86 Water depth m 0.362 ±0.157 0.559 ±0.152 0.507 ±0.182 Velocity m/s 0.117 ±0.097 0.005 ±0.006 0.098 ±0.128 Discharge m3/s 0.187 ±0.257 0.015 ±0.017 0.271 ±0.396

cBOD20 mg/L 3.777 ±4.186 10.190 ±6.103 10.979 ±4.322 Chlorophyll a ug/L 0.713 ±0.913 7.175 ±9.749 10.143 ±12.983 Nitrite mg/L 0.001 ±0.002 0.006 ±0.010 0.054 ±0.065 Nitrate mg/L 0.012 ±0.019 0.010 ±0.014 0.066 ±0.079 Orthophosphate mg/L 0.155 ±0.209 0.600 ±0.401 0.793 ±0.334 Total coliform CFU/100mL 2358 ±1800 9441 ±9025 15719 ±13237 Fecal coliform CFU/100mL 108 ±94 1294 ±5764 237 ±257 N Ammonia mg/L 0.020 ±0.025 0.159 ±0.133 0.312 ±0.159 drought of 1998-2000 and reported elevated sodium and chloride concentrations in surface waters of the

Mermentau River Basin. During the 2011 drought, elevated salinities (0.5 – 2.0 ppt) were also detected in wells within the Mermentau Basin by Borrok and Broussard (2016). During this study, specific conductance measured at only one study site within the Mermentau Basin, Bayou Grand Marais, in

February of 2011 was close to 1.0 mS/cm (approximate conductance for 0.5 ppt salinity).

In contrast to the Prairie terrace, streams in the Tertiary Uplands terrace flowed through highly erodible sands, some with fairly large patches of gravel (substrate index >2) but exhibited very low turbidity (<20

NTU). Some watersheds in this terrace had close to 50% forested land cover, which extended to the stream bank as overstory (>80%). Combined with eroding, shallow banks (1.6±0.4 m), this riparian cover contributed larger amounts of woody debris (4.9±4.8 m-2). Differences in water quality of upland streams were most conspicuous with respect to their lower nutrient loads (e.g., orthophosphate and N-ammonia),

16 specific conductance (0.06±0.04 mS/cm), and autochthonous production (chlorophyll a 2.3±3.6 ug/L).

Despite lower than normal flows during the drought, these streams experienced slightly higher dissolved oxygen levels (6.1±2.6 mg/L), lower water temperature (18.2±7.3), and lower biochemical oxygen demand (3.8±4.2 mg/L).

Vector patterns on principal components 1 and 2 were fairly well-conserved between the two water years, but variability explained by first two components was slightly higher and separation of terrace groupings was better for 2013 data (Table 1.4, Figures 1.7 and 1.8) The first axis was structured along a gradient between the Uplands and Prairie terraces, characterized primarily by water quality (e.g., specific conductance, turbidity and orthophosphate), but with contributions from habitat complexity (e.g., woody debris, substrate size and overstory cover) (Figures 1.7 and 1.8). Water velocity and channel dimensions

(e.g., wetted width, bank height, depth) contributed significantly to the second axis, which appeared to represent the gradient between the Flatwoods and the other terraces.

Table 1.4. Variable component loadings for principal component analyses for each study year. |Loading|>0.3 indicated with shading. Variable names beginning with “L” have been log10-transformed to stabilize variance.

Eigenvectors 2011 Eigenvectors 2013 Variable PC1 PC2 PC3 PC4 PC5 Variable PC1 PC2 PC3 PC4 PC5 Temp_C 0.130 -0.007 0.457 -0.108 0.273 Temp_C 0.118 -0.027 0.573 -0.100 0.240 DO_mg -0.069 0.188 -0.398 -0.080 0.211 DO_mg -0.187 0.321 -0.320 0.016 -0.053 LSpCond 0.339 -0.061 -0.056 -0.057 0.032 LSpCond 0.308 -0.037 -0.211 0.096 0.168 pH 0.066 0.141 -0.414 -0.013 0.006 pH 0.088 -0.003 -0.311 0.289 0.275 LTurbid 0.324 -0.054 -0.063 -0.097 -0.109 LTurbid 0.304 -0.006 -0.224 -0.100 -0.113 wWidth 0.250 0.394 -0.103 0.017 -0.070 wWidth 0.199 0.353 -0.189 0.047 -0.048 Depth 0.190 -0.262 -0.143 0.081 -0.470 Depth 0.123 -0.381 -0.158 -0.107 -0.326 LVelocity -0.087 0.521 -0.068 -0.059 -0.091 LVelocity -0.102 0.470 -0.039 -0.204 0.021 LWoodCt -0.199 -0.180 0.018 -0.364 -0.290 LWoodCt -0.166 -0.258 -0.141 -0.452 0.031 Substrate -0.261 0.203 0.177 0.253 -0.055 Substrate -0.274 0.243 0.178 -0.016 -0.144 BankHt 0.201 0.341 0.002 -0.165 0.330 BankHt 0.174 0.379 0.064 0.096 0.234 BankVeg -0.155 0.235 0.043 -0.418 -0.245 BankVeg -0.171 0.135 -0.160 -0.420 0.306 Incision -0.006 0.071 0.060 0.604 -0.232 Incision 0.015 0.096 0.373 0.298 -0.324 %Cover -0.271 -0.002 0.262 -0.354 -0.136 %Cover -0.280 -0.131 0.127 -0.240 0.197 cBOD20 0.263 -0.213 0.114 -0.117 0.173 cBOD20 0.268 -0.024 0.061 0.063 0.183 LNitrite 0.193 0.284 0.272 0.014 -0.138 LNitrite 0.236 0.162 0.124 -0.320 -0.132 LNitrate 0.107 0.209 0.419 0.141 -0.199 LNitrate 0.227 0.207 0.168 -0.329 -0.150 LOrthoP 0.303 0.092 0.044 -0.119 -0.213 LOrthoP 0.309 -0.006 -0.022 -0.151 0.160 LChla 0.289 -0.086 0.074 0.012 0.193 LChla 0.283 -0.097 0.138 -0.068 0.137 LTCOLI 0.265 -0.050 0.104 -0.120 -0.131 LTCOLI 0.300 -0.045 -0.018 -0.189 -0.069 LFCOLI 0.196 0.068 -0.178 -0.089 -0.343 LFCOLI 0.079 0.094 -0.105 -0.131 -0.527 %Variation 30.3 12 10.4 8.7 7.2 %Variation 35.9 12.3 9.4 7.8 7.5

2011 6 Terrace Uplands Flatwoods Prairie 4 LVelocity wWidth BankHt LNitrite 2 SubstrateDO_mg LNitrate pH

Incision LOrthoP

2 C

P %Cover Temp_C LChlaLSpCond 0 LWoodCt cBOD20 Depth

-2

-4 -6 -4 -2 0 2 4 6 PC1

Figure 1.7. Principal component bi-plot (PC1 and PC2 together explain 42.3% of the variation) of 2011 physicochemical variables by geologic terrace. Vector direction and length indicate sign and magnitude of contribution to axis, respectively. Variable names preceded by “L” indicate log10-transformation.

PERMANOVA models for 2011 data indicated a significant effect of terrace on habitat features (pseudo-

F=4.29, p=0.0006), but not for basin (pseudo-F= 1.15, p=0.34). Not enough study sites were sampled to construct a test of interaction between terrace with basin. Pairwise comparisons based on Monte Carlo tests demonstrated significant differences in 2011 between Uplands and Prairie (p=0.0012) and Uplands and Flatwoods (p=0.011). Similar results were found for 2013 with significant effect for terrace (pseudo-

F= 6.03, p=0.0001), but not for basin (pseudo-F= 1.50, p=0.18). Pairwise comparisons for 2013 data demonstrated significant differences among all terraces [Uplands/Prairie (p=0.0013), Uplands/Flatwoods

(p=0.0035), Flatwoods/Prairie (p=0.033)].

2013 6 Terrace Uplands Flatwoods Prairie 4

LVelocity BankHt DO_mg wWidth Substrate LNitrate 2 LNitrite

BankVeg Incision

2 C

P pH LTurbidLSpCond %Cover LChla 0 LWoodCt Depth

-2

-4 -6 -4 -2 0 2 4 6 PC1

Figure 1.8. Principal component bi-plot (PC1 and PC2 together explain 48.2% of the variation) of 2013 physicochemical variables by geologic terrace. Vector direction and length indicate sign and magnitude of contribution to axis, respectively. Variable names preceded by “L” indicate log10-transformation.

The main effect of time (i.e., sampling event) was significant in the PERMANOVA models for both 2011

(pseudo-F=4.70, p=0.0001) and 2013 (pseudo-F=6.83, p=0.0001). Therefore, seasonal habitat differences by terrace were examined through pairwise tests (with Monte Carlo generated p-values) between levels of terrace for the terrace with time interaction. Some tests could not be constructed because of insufficient data due to missing observations. Differences in stream habitat between Uplands and Prairie, based on average distance (i.e., Euclidean distance) between/within groups, which was the multivariate surrogate for difference in means, were fairly consistent throughout both years (Table 1.5).

Seasonal differences between Uplands and Flatwoods were not found (i.e., significant only for 1 of 8 tests) in 2011, but were demonstrated in 2013 for all but the Spring samples. Stream habitat in the

Flatwoods did not differ seasonally from that of the Prairie (except for May 2013).

Table 1.5. PERMANOVA results for pairwise tests of differences between terraces by sampling event for each study year. Values shown represent average Euclidean distance between/within groups.

Sampling Event Uplands, Prairie Uplands, Flatwoods Prairie, Flatwoods Aug 2010 -- 5.9607 -- Oct 2010 6.8494 * 6.2386 5.4683 Dec 2010 -- 6.0559 * -- Feb 2011 6.202 * 5.9624 5.3248 Apr 2011 8.8097 * 6.9603 7.6078 Jun 2011 7.4127 * 6.3324 6.7606 Aug 2011 6.2687 * 5.8937 5.6161 Nov 2011 7.1295 * 6.1712 5.0934 Aug 2012 7.0092 * 7.0779 * 6.3417 Oct 2012 6.2254 * 5.5442 * 4.9611 Nov 2012 6.2962 * 6.618 * 5.755 Feb 2013 6.8951 * 6.094 5.1827 Apr 2013 6.9175 7.0483 -- May 2013 9.4343 * 6.0444 * 7.907 * Jul 2013 7.6526 * 5.0875 * 5.6934 Aug 2013 6.6455 * 5.9571 * 5.5261 * Pairwise P(Monte Carlo) <0.05 -- Insufficient data

Discussion

Within the lowland coastal plain streams of Southwest Louisiana, geologic terrace has a direct and significant influence on stream habitat in general, as well as seasonally. A transition occurs within the

Flatwoods, as soil type shifts from impermeable clay in the Prairie to loamy sand of the Uplands. The change in soil type also appears to govern land cover in this rural region, with compact clays facilitating the flooding of rice fields in the flat Prairie and a slight step up in elevation with more permeable soils allowing better drainage for timber production in the Flatwoods and Uplands. The close linkage between soil and land cover has the potential to confound results of studies of land use on streams if care is not taken to choose sites that represent the environmental gradient in question and not some underlying regional effect (Vondracek et al. 2005). Additionally, stream habitat condition was more similar within terraces than within river basins. This probably results from a combination of the strong influence of terrace geology and the weak influence of coastal plain topography, evidence of which can be found in the delineation of ecoregions for this area (Omernik 1987, Daigle et al. 2006). Regardless, knowledge about the primary sources of variability within and between study sites facilitates design of landscape scale experiments and enables researchers to extend inference beyond the geographical area of study.

Other descriptions of coastal plain water bodies have tended to generalize these habitats as low-gradient, warmwater streams with shifting sand or silt substrates, low dissolved oxygen and moderate discharge

(Felley 1992, Benke and Wallace 1998, Smock 1998, Ice and Sugden 2003). Although these descriptors are fitting for many “blackwater” coastal plain streams, such as those in the eastern Gulf and southern

Atlantic coastal plains and even some within the Tertiary Uplands terrace sampled in this study, they exclude many of the features with significant contributions to habitat gradients that were found during this study. Kaller (2005) suggested that influences of a warmer climate, periods of glaciation and sea level rise, and Mississippi River avulsion to the east may have uniquely defined this area of Louisiana among other coastal plains (with regard to stream habitat for macroinvertebrates). The fact that the Pleistocene terraces demonstrated a gradient of stream habitat condition throughout the region may provide evidence for that assertion.

Physical and chemical stream features measured throughout the study period illustrated the influence of terrace on aquatic habitat stability and complexity. Multivariate ordination on these features demonstrated that overall structure of stream habitat was conserved between a severe drought year and a typical water year, although the ability to detect the effect of terrace improved with return to normal stream flows. Permutational MANOVA models indicated significant and consistent differences in stream habitat between the Uplands and the Prairie, with intermediate habitat in the flatwoods. Stream habitats of the Flatwoods terrace were also conducive to natural disturbance. Within at least two sites, beaver

(Castor canadensis) activity disrupted flow and pooled water, which prevented sites from dewatering during drought, but shifted the entire reach to lentic condition for an extended period of time. Stable soils

(i.e., transitional clay complexes and loams) and an abundance of woody debris from forested riparian zones, combined with low flows during 2011, may have provided highly favorable conditions for dam construction by beavers in this terrace.

Ice and Sugden (2003) and Justus et al. (2014) found that, in lowland streams with intermittent flow, low dissolved oxygen levels often result naturally from seasonal input of allochthonous material. In this study, seasonal differences were detected less frequently during extreme drought condition, as, in the absence of regular flushing spates, water quality measures demonstrated more extreme values at all sites,

21 increasing within-terrace variability. These results suggest that baseflow is not only beneficial to water quality in coastal plain streams, but also an important determinant of habitat heterogeneity between adjacent terraces. Groundwater surveys in the study area have shown that the Chicot Aquifer, the primary shallow aquifer used for both public water supply and irrigation of crops, has lost 30 m of potentiometric surface in some locations (Borrok and Broussard 2016). Continued drop in the hydraulic head in this aquifer could affect recharge into river systems resulting in loss of baseflow, especially with increased demand for groundwater during drought. Conversion from groundwater discharge (once forming artesian wells in the area) to widespread recharge has had a significant impact on native mussel communities in the area (Vidrine et al. 2004). Only one field measurement during this study indicated abnormal ion concentrations (i.e., specific conductance near 1.0 mS/cm), so saltwater intrusion most likely did not affect surface waters sampled, at least not on days visited.

In summary, as rivers and streams drain across geologic terraces toward the Gulf of Mexico, physical and chemical characteristics are more similar across a common terrace than within a common river basin. The physical, chemical, and biological features that are expected for low-order streams in upland systems have been attributed to biological and geomorphological processes, as well as stream aging (Leopold et al. 1964; Hynes 1970; Vannote et al. 1980). However, Justus et al. (2014) found that certain habitat criteria used for upland streams are not applicable to lowland streams, such as the ones measured in this study. In this landscape, it appears that geology is a greater driving factor, potentially because of the relative youth of the coastal terraces, and that local climatic patterns decrease differences in drought years and enhance differences in wetter years. Kaller et al. (2013) did not consider geologic terrace in their comparison of river basin or ecoregional stratification. This study suggests that geologic terrace could be a more important consideration, if habitat is being considered. Studies that compare habitat or compare species-habitat associations in this landscape should consider stratification among terraces along with ecoregion or basin.

References

Allan, J. D. 2004. Landscapes and Riverscapes: The Influence of Land Use on Stream Ecosystems. Annual Review of Ecological and Evolutionary Systems 35: 257–84.

Anderson, M. J., R. N. Gorley and K. R. Clarke. 2008. Permanova+ for Primer: Guide to Software and Statistical Methods. 214 p.

APHA, AWWA, and WEF. 2005. Standard Methods for the Examination of Water and Wastewater, 21st ed. American Public Health Association, Washington, D.C.

Arar, E., and G. Collins. 1997. Method 445.0 In Vitro Determination of Chlorophyll a and Pheophytin a in Marine and Freshwater Algae by Fluorescence. In Office of Research and Development National Exposure Research Laboratory, editor. Environmental Protection Agency.

Benke, A. C., and J. B. Wallace. 1998. Wood dynamics in Coastal Plain blackwater streams. Canadian Journal of Fisheries and Aquatic Science 47: 92-99.

Benke A.C. and C.E. Cushing (Editors), 2004. Rivers of North America. Academic Press/Elsevier, San Diego, California. 1168 p.

Borrok, D. M. and W. P. Broussard III. 2016. Long-term geochemical evaluation of the coastal Chicot aquifer system, Louisiana, USA. Journal of Hydrology 533: 320–331.

Cormier, A. 2007. A Timeline History of Lake Charles and Southwest Louisiana. Prepared for the Southwest Louisiana Genealogical Society. Calcasieu Historical Preservation Society. Lake Charles, Louisiana.

Daigle, J.J., G.E. Griffith, J.M. Omernik, P.L. Faulkner, R.P. McCulloh, L.R. Handley, L.M. Smith, and S.S. Chapman. 2006. Ecoregions of Louisiana (color poster with map, descriptive text, summary tables, and photographs): Reston, Virginia, U.S. Geological Survey (map scale 1:1,000,000).

Felley, J.D. 1992. Medium-low-gradient streams of the Gulf coastal plain. Pages 233-270 in C.T. Hackney, S.M. Adams, and W.H. Martin, editors. Biodiversity of the Southeastern United States. Aquatic Communities. John Wiley & Sons, Inc., New York, New York.

Fitzgerald, A. M. 2012. Effects of Varying Land Use on Headwater Stream Fish Assemblages and In- Stream Habitats in Southwestern Louisiana. M.S. Thesis, Louisiana State University, Baton Rouge, Louisiana. 137 p.

Fritz, K. M. and J. W. Feminella. 2011. Invertebrate colonization of leaves and roots within sediments of intermittent Coastal Plain streams across hydrologic phases. Aquatic Science 73:459–469.

Gardiner, E. P., A. B. Sutherland, R. J. Bixby, M. C. Scott, J. L. Meyer, G. S. Helfman, E. F. Benfield, C. M. Pringle, P. V. Bolstad, and D. N. Wear. 2009. Linking stream and landscape trajectories in the southern Appalachians. Environmental Monitoring and Assessment 156:17–36.

Homer, C.G., J.A. Dewitz, L. Yang, S. Jin, P. Danielson, G. Xian, J. Coulston, N.D. Herold, J.D. Wickham, and K. Megown. 2015. Completion of the 2011 National Land Cover Database for the conterminous United States-Representing a decade of land cover change information. Photogrammetric Engineering and Remote Sensing, v. 81, no. 5, 345-354 p.

Hynes, H. B. N. 1970. The ecology of running waters. University of Toronto Press. 555 p.

Ice, G. and B. Sugden. 2003. Summer dissolved oxygen concentrations in forested streams of northern Louisiana. Southern Journal of Applied Forestry 27:92-99.

Justus, B. G., S. V. Mize, J. Wallace and D. Kroes. 2014. Invertebrate and fish assemblage relations to dissolved oxygen minima in lowland streams of Southwestern Louisiana. River Research and Applications 30: 11–28

Kaller M. D. 2005. Macroinvertebrate community ecology of lowland, subtropical streams in Louisiana. Ph.D. Dissertation, Louisiana State University, Baton Rouge, Louisiana. 152 p.

Kaller, M. D. and W. E. Kelso. 2007. Association of macroinvertebrate assemblages with dissolved oxygen concentration and wood surface area in selected subtropical streams of the southeastern USA. Aquatic Ecology 41: 95–110.

Kaller M. D., C. E. Murphy, W. E. Kelso and M. R. Stead. 2013. Basins for Fish and Ecoregions for Macroinvertebrates: Different Spatial Scales Are Needed to Assess Louisiana Wadeable Streams. Transactions of the American Fisheries Society 142: 767-782.

LA Atlas (Atlas: The Louisiana Statewide GIS). 2009. LSU CADGIS Research Laboratory, Baton Rouge, LA, http://atlas.lsu.edu.

Leopold, L. B., M. G. Wolman, and J. P. Miller. 1964. Fluvial processes in geomorphology. San Francisco, CA: Freeman: 79-80.

Louisiana Geological Survey. 2008. Generalized geology of Louisiana. Louisiana State University. Baton Rouge, Louisiana. www.lgs.lsu.edu/deploy/uploads/gengeotext.pdf.

McClain, W. R., R. P. Romaire, C. G. Lutz, M. G. Shirley. 2007. Louisiana Crawfish Production Manual. Louisiana State University Agricultural Center Publication 2637. Baton Rouge, Louisiana. 58 p.

McNab, W.H., D.T. Cleland, J.A. Freeouf, J.E. Keys, G.J. Nowacki, and C.A Carpenter. 2007. Description of ecological subregions: sections of the conterminous United States. General Technical Report WO-76B.

Merten, G.H., Welch, H.L., and Tomer, M.D. 2016. Effects of hydrology, watershed size, and agricultural practices on sediment yields in two river basins in Iowa and Mississippi. Journal of Soil and Water Conservation. 71(3):267-278.

NOAA National Centers for Environmental Information.” State of the Climate: Drought for Annual 2011,” published online January 2012, retrieved on October 9, 2016 from http://www.ncdc.noaa.gov/sotc/drought/201113.

NRCS (Natural Resources Conservation Service USDA), FSA (Farm Service Agency USDA), and RD (Rural Development USDA). 2016. National Land Cover Database 2011. Geospatial Data Gateway. USDA, Natural Resources Conservation Service. http://dx.doi.org/10.15482/USDA.ADC/1241880. Downloaded June 2014.

NRCS (Soil Survey Staff, Natural Resources Conservation Service, United States Department of Agriculture). 2016. SSURGO (Soil Survey Geographic Database). Available online at http://websoilsurvey.nrcs.usda.gov/. Downloaded June 2016.

Omernik, J. M. 1987. Ecoregions of the conterminous United States. Annals of the Association of American Geographers. 77: 118-125.

Prakken, L.B., J.M. Griffith, and R.B. Fendick, Jr. 2012a. Water resources of Allen Parish: U.S. Geological Survey Fact Sheet 2012–3064. 6 p.

Prakken, L.B., J.M. Griffith, and R.B. Fendick, Jr. 2012b. Water resources of Beauregard Parish: U.S. Geological Survey Fact Sheet 2012-3065. 6 p.

Prakken, L.B., J.M. Griffith, and R.B. Fendick, Jr. 2012c. Water resources of Vernon Parish: U.S. Geological Survey Fact Sheet 2012–3063. 6 p.

Riseng, C. M., M. J. Wiley, R. W. Black and M. D. Munn. 2011. Impacts of agricultural land use on biological integrity: a causal analysis. Ecological Applications 21(8):3128-3146.

Rosenberg, D. M. and V. H. Resh (Eds). 1993. Freshwater biomonitoring and benthic macroinvertebrates. New York: Chapman and Hall. 488 p.

Shields, Jr., F. D., E. J. Langendoen, and M. W. Doyle. 2006. Adapting Existing Models to Examine Effects of Agricultural Conservation Programs on Stream Habitat Quality. Journal of the American Water Resources Association (JAWRA) 42(1):25-33.

Skrobialowski, S. C., S. V. Mize and D. K., Demcheck. 2004. Environmental Setting, Water Quality, and Ecological Indicators of Surface-Water Quality in the Mermentau River Basin, Southwestern Louisiana, 1998-2001. Reston, Virginia. U. S. Geological Survey Water-Resources Investigations Report 03-4185. 73 p.

Smock, L. A.1998. Life histories, abundance and distribution of some macroinvertebrates from a South Carolina, USA coastal plain stream. Hydrobiologia 157: 193 -208.

Soil Survey Staff, Natural Resources Conservation Service, United States Department of Agriculture. Official Soil Series Descriptions. Available online. Accessed Sep 2016.

U.S. Geological Survey. 2013. National Hydrography Geodatabase: The National Map viewer available on the World Wide Web (http://viewer.nationalmap.gov/viewer/nhd.html?p=nhd). Accessed Jun 2016.

Vannote, R. L., G. W. Minshall, K. W. Cummins, J. R. Sedell and C. E. Cushing. 1980. The River Continuum Concept. Canadian Journal of Fisheries and Aquatic Sciences, 37:130-137.

Vidrine, M. F.; G. J. Quillman-Vidrine, M. F. Vidrine, II; D. J. Vidrine, and C. E. Vidrine. 2004. Freshwater Mussels (Bivalvia: Unionidae) in the Cajun Prairie Ecosystem in Southwestern Louisiana. Proceedings of the North American Prairie Conferences. Paper 77.

Vondracek, B., K. Blann, C. Cox, K. Mumford, B. Nerbonne, and J. Nerbonne. 2005. Land use, spatial scale, and stream systems: Lessons from an agricultural region. Environmental Management 36: 775– 791.

Watershed Concepts. 2003 - 2005. Digital Elevation Model (USGS DEM), Task Area 12, 13, 16, 21, 22, 23. Louisiana / Federal Emergency Management Agency (FEMA) Project - Phase 2 - 4 of Louisiana LIDAR Data Development. Watershed Concepts contract EMT-2002-CO-0048.

White, V.E., and L.B. Prakken. 2014. Water resources of Jefferson Davis Parish, Louisiana: U.S. Geological Survey Fact Sheet 2014–3074. 6 p.

Williams, L.R., T.H. Bonner, J.D. Hudson III, M.G. Williams, T.R. Leavy, and C.S. Williams. 2005. Interactive effects of environmental variability and military training on stream biota of three headwater drainages in Western Louisiana. Transactions of the American Fisheries Society, 134:192–206.

CHAPTER 2: TAXONOMIC DISTINCTNESS AND SEASONAL STABILITY OF BENTHIC MACROINVERTEBRATE ASSEMBLAGES AMONG PLEISTOCENE TERRACES OF THE GULF COASTAL PLAIN OF SOUTHWESTERN LOUISIANA

Introduction

Assessment of ecological condition in water bodies has employed biological endpoints for more than 100 years (Rosenberg and Resh 1993, Buss et al. 2015). In the shadow of industrialization, however, chemical contaminants became the focus of freshwater evaluations in the first half of the twentieth century (Karr and Chu 2000). Rosenberg (1998) likened measuring chemical pollutants to “taking snapshots of the ecosystem, whereas biological measurements are like making a videotape.” Indeed, the abundance and diversity of biota present in a habitat embody cumulative effects of multiple environmental stressors, key to characterizing ecological integrity (Karr 1993). Effects on biota are cumulative because immediate stresses on individuals compile over time to eventually shape whole assemblages as more tolerant organisms usurp niches once held by more sensitive taxa (Colas et al. 2014). Angermeier and

Karr (1994) defined ecological integrity as a state in which all appropriate biotic elements are present in the ecosystem and processes occur at appropriate rates. Which elements and rates are appropriate depends on the type of ecosystem and the regional forces that structure it (Buffagni and Furse 2006,

Lake et al. 2007). Natural resource agencies use assessments of ecological integrity to prioritize development of pollution budgets and implement strategies to reduce pollution loads [Clean Water Act

1972 40 CFR §305(b)]. Over the past four decades, significant improvements to bio-assessment protocols have been achieved by integrating sound ecological principles with regional approaches and moving away from short-sited, binary classifications of impairment (Karr 1991, Karr 1993, Davy-Bowker et al. 2006, Bailey et al. 2007, Heino et al. 2007, Daniel et al. 2014). Biological endpoints are most useful when they are sensitive to stressors of interest, exhibit a response that can be distinguished from natural variability, and can be monitored in a cost-effective manner (Karr and Chu 1999, Verdenschot and Moog

2006, Southerland et al. 2007, Beghelli et al. 2012).

Evaluation of habitat quality is essential to management, conservation and restoration of freshwater ecosystems. In lotic systems, evaluation of biological responses to anthropogenic habitat disturbance

(e.g., turbidity from agricultural runoff) is often confounded by the influence of natural disturbance (e.g.,

26 seasonal extremes in discharge; Clarke and Hering 2006). Novel approaches to modeling biotic-abiotic interactions have been developed that, in combination, may reduce noise in data from disturbed ecosystems with a well-defined reference condition (Kennard et al. 2006, Sandin and Verdonschot 2006).

In highly-disturbed, low-gradient streams, such as those in the Gulf coastal plain, however, conventional biotic metrics may not characterize obvious disturbances, such as heavy sedimentation or hypoxia, because pervasive impacts may represent only one side of the disturbance gradient or the extant biota may have adapted strategies for tolerating extreme conditions that occur naturally for brief periods of time

(Kaller and Kelso 2007, Justus et al. 2014). Biological responses modeled along these truncated gradients may confound management decisions when the goals of determining habitat impairment or designating aquatic life use require a reference condition as a basis of comparison (40 CFR §131.10).

Capturing the functional gradient of condition, therefore, is essential to describing a response across the spectrum of environmental stressors, especially when the goal is identification and mitigation of anthropogenic disturbance (Hering et al. 2006). This functional portion of the stressor gradient comprises the ranges of the various habitat features that affect diversity and abundance of biota, beyond the influence of natural habitat fluctuations or trophic interactions (e.g., predation, competition) (Greathouse and Pringle 2006, Kennard et al. 2006).

Aquatic macroinvertebrates have been used as indicators of habitat impairment in streams for over a century (Chen et al. 2014). Their distinctive life histories, many with sedentary and protracted aquatic larval stages followed by brief emergence as highly mobile adults, uniquely qualify them as indicators of chronic environmental stress (Beghelli et al. 2012). Reproductive strategies, especially frequency and timing of emergence, may also be indicative of response to environmental cues and/or competition

(Smock 1988). In addition, macroinvertebrates inhabit all available niches from the water surface to the benthos and have evolved a wide variety of physiological and morphological features to subsist in an unstable and often inhospitable aquatic environment (Merritt et al. 2008). For example, some taxa prefer lentic habitats, whereas others have respiratory requirements that preclude their survival in non-flowing waterbodies. The breadth of habitats exploited by various groups of aquatic insects, in particular, can complicate use of biotic metrics that tend to oversimplify the response to a stressor, in that a shift in habitat (e.g., from lotic to lentic) may select for a different assemblage, but show no change in the

27 response of a generalized metric (e.g., taxa richness or diversity; Heino et al. 2007). Examination of taxonomic distinctness has, therefore, been identified as a critical first step in describing macroinvertebrate relationships with habitat features, especially in landscape-scale studies for biotic index development (Heino et al. 2008, Colas et al. 2014).

Extensive research on freshwater abiotic-biotic interactions has been conducted in the highlands of Gulf coastal states and the Atlantic coastal plain (Jacobi and Benke 1991, Carlisle et al. 2008, Benke and

Huryn 2010), but few studies have examined the unique challenges to streams in the lowlands of the Gulf of Mexico coastal plain (Skrobialowski et al. 2004, Kaller and Kelso 2007, Justus et al. 2014). In-stream habitat in low-gradient, warm-water streams, such as those found in the Gulf coastal plain of southwestern Louisiana, varies from shifting sand substrate with abundant overstory cover to open- canopy bayous with hardpan clay substrate (Chapter 1). Seasonal variability in discharge in these streams is an important natural disturbance resulting from a combination of flat topography, extensive watersheds and a subtropical climate (Felley 1992). Consequently, the natural disturbance regime plays an important role in structuring faunal integrity in coastal plain streams and bayous (Williams et al. 2005), and designing meaningful and effective bioassessment protocols. Anthropogenic disturbances, such as deforestation, military training, agriculture, and animal husbandry, also exert profound influences on habitat quality in these streams by altering sediment load, allochthonous input, and physicochemical condition (Skrobialowski et al. 2004, Williams et al. 2005, Kaller and Kelso 2006a, Kaller and Kelso

2006b, Mize et al. 2008). Attempts to differentiate effects of anthropogenic disturbance gradients from those of natural origin have been further confounded by the “generalist” faunal communities that characterize many low-gradient streams (Kaller and Kelso 2007, Fitzgerald 2012). Specifically, organisms tolerant of turbidity, organic enrichment, and low dissolved-oxygen concentrations are common and often dominate the macroinvertebrate assemblages in these systems (Skrobialowski et al.

2004). Therefore, abiotic factors that yield a significant response in upland fauna may be inappropriate for investigations in lowland coastal plain streams, where organisms are more tolerant of some environmental extremes (Justus et al. 2014). Parreira de Castro et al. (2016) suggested a change in land use from natural vegetation to agriculture led to shifts in the input of allochthonous nutrients and autochthonous production, disrupting the balance between functional feeding groups and widening their

28 trophic niches. Essentially, overlap of trophic niches resulting from simplification of food resources in disturbed habitats might select for a generalist fauna.

Substantial differences in underlying geology among coastal plain and coastal upland stream habitats suggest that macroinvertebrate assemblages may differ between Pleistocene terraces, especially the

Tertiary Upland and Prairie terraces. Previous studies predicted that relationships between invertebrate taxa and specific habitat variables, however, would be weak. Kaller et al. (2013) suggested that macroinvertebrate communities within streams of Louisiana are organized primarily by ecoregions, which closely follow geologic boundaries. In this study, macroinvertebrates sampled throughout the year at stream sites stratified between geologic terraces (e.g., Pleistocene terraces within the coastal plain) and river basins were used to examine temporal stability of assemblages within different geologic landscapes.

Research objectives were 1) identify abiotic influences on aquatic biota, relating similarities among macroinvertebrate assemblages to in-stream habitat features and primary environmental gradients; 2) determine the utility of taxonomic distinctness, individual taxa, and functional feeding groups as indicators of environmental differences among streams; and 3) examine the effects of historic drought conditions observed throughout the study period on these findings and any implications on macroinvertebrate communities.

Methods

Twelve stream sites throughout Southwest Louisiana were sampled during even numbered months beginning in August 2010 and ending in August 2011. Sites were stratified between three adjacent river basins (i.e., Calcasieu, Mermentau and Vermilion-Teche) and three adjacent geologic terraces (i.e.,

Tertiary Uplands, Flatwoods and Prairie terraces). West Fork Sixmile Creek (WF6), East Fork Sixmile

Creek (EF6), Big Brushy Creek (BB), Barber Creek (BAR) and Hurricane Creek (HUR), located in the

Tertiary Uplands terrace, were characterized by shifting sand substrates and closed canopies (Table 2.1;

Figure 2.1). West Fork Caney Bayou (WFC), Bayou Petite Passe (PP) and Indian Bayou (IND), located in the Flatwoods terrace, had intermediate to full canopy cover and hardpan clay substrate. Additionally, all three of these stream reaches were impacted by beaver activity at some point during the study period.

Two locations on Bayou Serpent (S165 and S99), a site on East Bayou Lacassine (LAC) and a site on

Table 2.1. Abbreviations, names, river basins, ecoregions, size of watershed upstream of site and coordinates for 12 study sites. Ecoregion abbreviations are SCP South Central Plains, WGC Western Gulf Coastal Plain, STU Southern Tertiary Uplands, FLT Flatwoods, LLP Lafayette Loess Plains, NHG Northern Humid Gulf Coastal Prairies.

Bayou Grand Marais (GM) were located within the Prairie terrace and demonstrated intermediate to open canopies with silt on hardpan clay substrates. Historic drought conditions were observed for eastern

Texas and western Louisiana throughout 2011, resulting in extremely low flows in all study reaches

(Chapter 1).

Field Methods

Macroinvertebrates and in-stream habitat features were collected in 50-m study reaches in each stream.

During each site visit, I collected a water sample from the middle of the water column in a clean, dark bottle, capped underwater and packed in ice for transport to the laboratory. In situ measurements of water temperature, dissolved oxygen, pH, conductivity, and turbidity were also recorded using a handheld multi-parameter water quality sonde (YSI 650MDS unit with 6820 V2-1 sonde, Yellow Springs, Ohio,

W E

Figure 2.1. Map of sampling sites within southwestern Louisiana showing stream network within upstream portion of watershed. West Fork Sixmile (WF6), East Fork Sixmile (EF6), Big Brushy (BB), Barber (BAR) and Hurricane (HUR) are Tertiary Uplands sites. West Fork Caney (WFC), Petite Passe (PP) and Indian (IND) are Flatwoods sites. Serpent (S165 and S99), Lacassine (LAC) and Grand Marais (GM) are Prairie sites.

USA). Five equidistantly-spaced cross-channel transects measured with a meter tape were marked with flags, each transect was measured for wetted width and divided into three equal segments, oriented from left bank to right bank looking upstream. At the midpoint of each segment, water velocity and depth were measured with a FlowTracker Handheld 2D acoustic Doppler velocimeter (SonTek/Xylem Inc., San

Diego, California, USA) mounted on a 1.2 m top-setting wading rod (Marsh-McBirney/Hach Co.,

Loveland, Colorado, USA). Dominant type of substrate was noted (i.e., ordinal index based on particle size) within each segment and pieces of woody debris were counted within a 0.5 m radius about the wading rod. Additionally, percent overstory cover (measured with a Model-C concave spherical densiometer; Forestry Suppliers, Inc., Jackson, Mississippi, USA), degree of channel incision (i.e., ordinal index based on floodplain width), approximate bank height, and dominant vegetation type (i.e., ordinal index based on complexity/height) were recorded at transects 1, 3, and 5. Any anomalies with potential to affect the physical, chemical or biological measurements at a site were noted and/or photographed

(e.g., channel obstructions, atypical water odors, or rotting animal carcasses).

Laboratory Methods

Stream water samples were removed from the laboratory refrigerator and allowed to warm to room temperature (i.e., 20-22°C) before analyses were performed. Carbonaceous biochemical oxygen demand (i.e., with nitrification inhibitor added) was measured in duplicate samples of stream water incubated at 20°C for at least 20 days with a Thermo Orion model 850A+ dissolved oxygen meter

(Thermo Fisher Scientific, Waltham, Massachusetts, USA) according to American Public Health

Association standard method 5210B (APHA 2005). Fecal coliform counts were determined following

APHA (2005) standard method 9222A. Diluted stream water was vacuum-filtered onto sterile 47-mm, gridded 0.45-µ nitrocellulose filters that were then placed in padded sterile petri plates to which 2 mL of m-ColiBlue24TM broth (EMD Millipore, Billerica, Massachusetts, USA) was added. Plates were inverted, sealed in water tight bags and incubated in a 35°C water bath for 24 hours. Blue and red colony forming units (CFU) corresponding to Escherichia coli and total coliforms, respectively, were enumerated on duplicate plates of two different dilutions for each sample. Because coliform counts varied by site and season, dilutions were adjusted to ensure that one would contain at least 50 CFU. To quantify algal

32 abundance in water samples, 100 mL of stream water was filtered onto 47-mm 0.7-µ GF/F glass microfiber filters that were then folded into squares of aluminum foil, labeled and frozen for chlorophyll analysis. Chlorophyll a concentration was measured at Louisiana State University’s Department of

Oceanography and Coastal Sciences Wetland Biogeochemistry Analytical Services with fluorescence detection (TD-700 Fluorometer, Turner Designs, Sunnyvale, California, USA) following EPA Method

445.0 (Arar and Collins 1997). Finally, concentrations of nitrate (Cadmium Reduction Method 8192), nitrite (Diazotization Method 8507), N-ammonia (Salicylate Method 8155) and orthophosphate (Ascorbic

Acid Method 8048) were measured on a Hach DR/2500 spectrophotometer (Hach Co., Loveland,

Colorado, USA). These parameters (i.e., cBOD, coliforms, chlorophyll a, and nutrients) were measured for every site on all sampling dates.

Macroinvertebrates were collected in sediments with a 0.086- m2 Hess stream bottom sampler constructed with 500-micron mesh (Wildlife Supply Co., Yulee, Florida). In addition to sediment, approximately 5 moderately decomposed pieces of submerged wood were collected and all samples were placed carefully into pre-labeled gallon zipper bags, preserved with ethanol, double bagged, and placed on ice. One sample each of wood and sediment was collected at transects 1, 3 and 5 in conjunction with habitat measurements. Preserved samples were stored at -20°C until processing, at which time all collected invertebrates were preserved in 95% ethanol prior to identification. Some members of Diptera and Annelida were mounted on glass microscope slides with CMC-10 (Master’s

Chemical Company, Elk Grove, Illinois, USA) following the protocol by Epler (2001) and identified to lowest practical taxon, usually sub-family, under high magnification. Other macroinvertebrate taxa, such as insects, were identified to lowest practical taxon, usually genus. Functional feeding group (FFG) and pollution tolerance score (usually modified Hilsenhoff tolerance) were assigned to taxa based on a combination of references, with a default family-level value (or next higher level for non-insect taxa) used for genera with no published information (Barbour et al. 1999, Mandaville 2002, Merritt et al. 2008). If no tolerance score could be found, a neutral score of 5 was assigned. Additionally, mean taxa richness and

Shannon diversity (H’) were estimated for sediment and wood at each site.

Multivariate statistical analyses were performed with PRIMER v.7 and PERMANOVA+ software

(PRIMER-E Ltd., Plymouth, UK). Habitat variables that were strongly right-skewed were log10- transformed to stabilize variance. Variables were also normalized because measurement units were unique to each variable. Ordination via principal component analysis (PCA) was performed on normalized habitat variables to determine data dimensionality, reduce contributions to variability from redundant variables, and identify habitat features that explained the most variability in the dataset.

Ordinations via principal coordinates analysis and non-metric multidimensional scaling were also examined, but all produced similar patterns and PCA eigenvectors provided direct relationships with original variables.

Abundances for macroinvertebrate taxa collected in sediment were standardized by surface area (cm2) and abundances from wood samples were standardized by volume (i.e., wood volume in mL measured by water displacement). Taxonomic similarity among samples was examined via non-metric multidimensional scaling (NMDS) ordination on a matrix of Jaccard similarities calculated on combined

(i.e., sediment and wood samples) presence/absence data. Macroinvertebrate samples were collected at the same 12 locations at regular intervals throughout the study period. Therefore, landscape level differences in assemblage structure were tested with permutational MANOVA (hereafter, PERMANOVA) with fixed effects of terrace and time (i.e., sampling events 1-7 from 2010-2011) and site within terrace as a nested random effect. The effect of river basin was also tested in the same manner, but not enough sites were sampled to test a terrace by river basin interaction. Significance of the pseudo-F statistic for

Type III sums of squares was interpreted as permutation p-values <0.05. Type III sums of squares were used despite the unbalanced design in order to produce the most conservative results. For pairwise tests of between-terrace differences at various time points, Monte Carlo tests were used to generate p-values because the number of unique permutations was generally less than 100, making permutation p-values less reliable (Anderson et al. 2008). All models were executed with 9999 permutations and were calculated on the matrix of Jaccard similarities between samples.

Agreement between the multivariate ordination pattern based on macroinvertebrate assemblages (NMDS on Jaccard similiarity) and the ordination pattern based on in-stream habitat features (PCA on Euclidean

34 distance) was measured with rank correlation (rho) between the resemblance matrices. Distribution of the test statistic for the null hypothesis rho=0 (i.e., no agreement or observed by chance) was generated by randomly permuting one set of samples relative to the other (Clarke and Gorley 2015). This procedure is a non-parametric form of the Mantel test and was performed with 9999 permutations. Next, a subset of habitat variables that optimized the correlation between the assemblage matrix and the habitat matrix was chosen via stepwise selection based on the BEST procedure in PRIMER7. This allowed for inference about the primary environmental gradients or stressors that “explained” the pattern in the macroinvertebrate assemblages. .

To examine stability of macroinvertebrate assemblage structure through time, relative abundances of various taxonomic groups (e.g., insect orders) were qualitatively compared at time points between terraces. Additionally, seasonality of individual taxa within the most taxa-rich insect orders (Diptera,

Coleoptera, Odonata, Ephemeroptera, Plecoptera and Trichoptera) was examined with simple line plots of mean density (4th-root transformed to facilitate direct comparison on one scale) versus time within terrace. Habitat preference within the macroinvertebrate community was qualitatively assessed by seriating (re-ordering by similarity measure) a shade plot matrix of taxa (rows) by samples (columns) and comparing taxa that were unique to a terrace versus generalist taxa that occurred throughout the study area. Samples were re-ordered based on their Jaccard similarities (across taxa) and taxa were re- ordered based on Whittaker’s index of association (across samples) (Clarke and Gorley 2015).

Mean macroinvertebrate densities (number cm-2 for sediment, number mL-1 for wood) were compared among terraces for all taxa, as well as for insect taxa with general linear mixed models (PROC GLIMMIX,

SAS version 9.4, Cary, N.C.). Autocorrelation between observations, resulting from the fact that sites were visited multiple times, was addressed with a first order autoregressive covariance structure, although other structures (i.e., compound symmetry, Toeplitz, and heterogenous first order autoregressive) were tested with model fit (i.e., corrected Akaike information criterion) (Gutzwiller and

Riffell 2007). Assumptions of the general linear model were assessed by examination of model residuals and predicted values. In addition, differences among terraces of the proportions of various functional

35 feeding groups represented by insect taxa were also compared with general linear mixed models. Type I error rate for pairwise t-tests of differences was controlled with the Tukey-Kramer adjustment.

Results

Macroinvertebrate collections yielded 44,466 individuals from 167 insect taxa and 40 non-insect taxa (11 crustacean taxa, 15 mollusc taxa, 14 others including worms, mites, etc.). Change in relative abundances of major taxonomic groups within assemblages over time for each geologic terrace revealed seasonal fluctuations in coleopterans and dipterans within Upland streams and gradual increases in oligochaetes, gastropods and amphipods in Flatwoods and Prairie streams (Figure 2.2). Dipterans

100%

90% Oligochaeta Hirudinea 80% Platyhelminthes Veneroida 70% Gastropoda Trichoptera 60% Plecoptera Odonata 50% Neuroptera Megaloptera 40% Hemiptera Ephemeroptera 30% Diptera Collembola 20% Coleoptera Malacostraca 10% Isopoda Decapoda 0% Amphipoda

Arachnida

4_2011 8_2010 2_2011 6_2011 8_2011 8_2010 2_2011 4_2011 6_2011 8_2011 8_2010 2_2011 4_2011 6_2011 8_2011

10_2010 10_2010 12_2010 10_2010 12_2010 12_2010 Uplands Flatwoods Prairie

Figure 2.2. Change in relative abundances of major taxonomic groups across sampling dates by geologic terrace.

(mostly midges from family Chironomidae) accounted for 47.4% of overall macroinvertebrate density.

Oligochaetes constituted 15.3% of total, but densities were higher in lowlands than in upland sites and appeared to increase over time throughout the drought. The opposite was true for coleopterans, which accounted for 14.9%, but with higher densities in upland sites. The next most abundant taxonomic groups overall were gastropods (5.6%), freshwater clams (3.5%), trichopterans (3.1%), ephemeropterans

(2.7%), amphipods (2.4%), flatworms (1.7%), leeches (1.7%), odonates (0.6%) and plecopterans (0.2%).

Means and standard deviations for all in-stream habitat variables measured during the study period indicated differences in both physical habitat and water quality parameters among terraces (Table 2.2).

Table 2.2. Summary statistics by terrace for instream habitat variables measured during study period.

Uplands Flatwoods Prairie Mean ± SD Mean ± SD Mean ± SD Temperature °C 18.24 ± 7.30 20.00 ± 8.57 20.46 ± 8.66

Dissolved O2 mg/L 6.12 ± 2.56 5.12 ± 4.55 5.54 ± 2.33 Specific Cond. mS/cm 0.057 ± 0.028 0.259 ± 0.172 0.321 ± 0.177 pH 7.27 ± 0.73 7.56 ± 0.64 7.60 ± 0.50 Turbidity NTU 19.6 ± 27.3 91.5 ± 76.7 220.2 ± 356.1 Wood count per m2 6.15 ± 4.82 9.39 ± 8.60 2.51 ± 3.38 Dom. Substrate Index 2.23 ± 0.41 0.59 ± 0.47 1.08 ± 0.52 Bank Height m 1.68 ± 0.47 1.77 ± 0.36 2.74 ± 0.61 Dom. Bank Veg. Index 2.78 ± 0.37 2.48 ± 0.44 2.41 ± 0.47 Channel Incision Index 1.93 ± 0.97 1.00 ± 0.52 1.59 ± 0.50 % Overstory cover 80.64 ± 15.25 67.23 ± 17.81 41.71 ± 22.71 Wetted width m 4.90 ± 1.27 4.71 ± 1.20 6.88 ± 0.83 Water depth m 0.319 ± 0.168 0.442 ± 0.114 0.441 ± 0.163 Velocity m/s 0.103 ± 0.102 0.007 ± 0.016 0.072 ± 0.101 Discharge m3/s 0.131 ± 0.174 0.016 ± 0.039 0.177 ± 0.274

Principal component analysis of in-stream habitat variables revealed clear differences among geologic terraces, as evidenced by the variable loadings on each of 5 principal components by study year (Table

2.3). The first axis was structured along a gradient between the Uplands and Prairie terraces,

Table 2.3. Variable component loadings for principal component analysis of habitat variables. |Loading|>0.3 indicated with shading. Variable names beginning with “L” have been log10-transformed to stabilize variance.

Eigenvectors Variable PC1 PC2 PC3 PC4 PC5 Temp_C 0.130 -0.007 0.457 -0.108 0.273 DO_mg -0.069 0.188 -0.398 -0.080 0.211 LSpCond 0.339 -0.061 -0.056 -0.057 0.032 pH 0.066 0.141 -0.414 -0.013 0.006 LTurbid 0.324 -0.054 -0.063 -0.097 -0.109 wWidth 0.250 0.394 -0.103 0.017 -0.070 Depth 0.190 -0.262 -0.143 0.081 -0.470 LVelocity -0.087 0.521 -0.068 -0.059 -0.091 LWoodCt -0.199 -0.180 0.018 -0.364 -0.290 Substrate -0.261 0.203 0.177 0.253 -0.055 BankHt 0.201 0.341 0.002 -0.165 0.330 BankVeg -0.155 0.235 0.043 -0.418 -0.245 Incision -0.006 0.071 0.060 0.604 -0.232 %Cover -0.271 -0.002 0.262 -0.354 -0.136 cBOD20 0.263 -0.213 0.114 -0.117 0.173 LNitrite 0.193 0.284 0.272 0.014 -0.138 LNitrate 0.107 0.209 0.419 0.141 -0.199 LOrthoP 0.303 0.092 0.044 -0.119 -0.213 LChla 0.289 -0.086 0.074 0.012 0.193 LTCOLI 0.265 -0.050 0.104 -0.120 -0.131 LFCOLI 0.196 0.068 -0.178 -0.089 -0.343 %Variation 30.3 12 10.4 8.7 7.2 characterized primarily by water quality parameters such as specific conductance, turbidity and orthophosphate, but with contributions from several habitat characteristics such as woody debris, substrate size and overstory cover (Figure 2.3). Water velocity and channel dimensions (e.g., wetted width, bank height, depth) contributed significantly to the second axis, which appeared to represent the gradient between the Flatwoods and the other terraces better than the first principal component.

NMDS ordination of Jaccard similarities among macroinvertebrate samples revealed strong taxonomic distinctness between the Tertiary Uplands terrace and the Prairie and Flatwoods terraces, which overlapped significantly (Figure 2.4). Stress, a measure of the “fit” of the NMDS ordination of multidimensional data in a lower dimension, was fairly high (0.26) for the 2-dimensional solution (Clarke and Gorley 2015). In fact, stress was reduced to 0.2 for the 3-dimensional solution, which revealed that the overlap between the Prairie and Flatwoods terraces was not as complete as shown in the 2-D biplot.

6 Terrace Uplands Flatwoods Prairie 4 LVelocity wWidth BankHt LNitrite 2 SubstrateDO_mg LNitrate pH

Incision LOrthoP

2 C

P %Cover Temp_C LChlaLSpCond 0 LWoodCt cBOD20 Depth

-2

-4 -6 -4 -2 0 2 4 6 PC1

Figure 2.3. Principal component bi-plot (i.e., PC1 and PC2 explain 42.3% of variation, combined) of habitat variables by geologic terrace. Vector direction and length indicate sign and magnitude of contribution to axis, respectively. Variable names preceded by “L” indicate log10-transformation.

Although the PERMANOVA test of differences between river basins was not significant (pseudo-F=1.23, p=0.2116), differences among terraces was (pseudo-F=3.38, p=0.0016), and pairwise comparisons based on Monte Carlo tests indicated assemblages in the Uplands sites were significantly different from sites in both the Prairie (p=0.0078) and Flatwoods (p=0.0145) terraces, which did not differ from each other (p=0.1664). Differences in assemblage structure between sampling dates was also significant

(pseudo-F=4.9028, p=0.0001), with significant pairwise differences mostly occurring between winter and spring samples. Differences in assemblage structure between terraces were consistent between the two time periods.

NMDS of Macroinvertebrate Assemblages Transform: Presence/absence Resemblance: S7 Jaccard 2D Stress: 0.26 Terrace Uplands Flatwoods Prairie

Figure 2.4. Non-metric multidimensional scaling ordination 2-D bi-plot of Jaccard similarities among samples (presence/absence data) by geologic terrace.

Agreement between multivariate ordination patterns for habitat variables and macroinvertebrate assemblages was significant, but not very strong (rho=0.301, p=0.0001). Stepwise selection of specific variables to improve the rank correlation between the Jaccard similarity matrix of macroinvertebrate assemblages and the Euclidean distance matrix of habitat features yielded a suite of 7 variables (log10- specific conductance, log10-turbidity, water depth, log10-velocity, dominant substrate size, percent canopy cover, and cBOD20) with a final rank correlation of rho=0.505. Selected variables matched many of the variables with strong correlations to the first two principal components (Table 2.3).

Line plots of mean 4th-root transformed densities by terrace and sampling date further demonstrated taxonomic distinctness between terraces, as well as seasonal patterns for certain taxa (Figures 2.5-2.10).

Dipterans appeared to be fairly ubiquitous and seasonally stable, although identification of chironomids to subfamily most likely masked some taxonomic distinctness between terraces (Figure 2.5). Coleopterans

Diptera taxa by Terrace and Date 0.5 Atherix Allognosta Atrichopogon Odonto_Hedrio Bezz alp Odontomyia Ceratopogonidae Chlorotabanus Culicoides Chrysops 0.4 Forcipomyia Tabanidae

Chaoborus Tabanus y

t Chironomidae Ellipteroides i

s Chironomini Erioptera n

e 0.3 Cryptochironomus Hexatoma

t Orthocladiinae Limnophila

o Tanypodinae Limonia

o r

- Tanytarsini Megistocera h

t Xestochironomus Molophilus 4 0.2

n Dolichopodidae Ormosia a

e Hemerodromia Paradelphomyia

M Pericoma Pilaria Psychodidae Pseudolimnophila 0.1 Sciomyzidae Tipula Simuliidae Tipulidae Simulium Diptera

P F F

Figure 2.5. Seasonal variability of mean 4th-root transformed individual Diptera taxa within terraces.

were very diverse and habitat preferences of individual taxa, especially among riffle beetles, appeared to be related to stream flow and wood availability (Figure 2.6). Stenelmis sp. (Family: Elmidae) was ubiquitous throughout terraces and seasons, but densities were lower in the Flatwoods sites, where beaver activity reduced flow entirely during the study period. Macronychus sp. and Ancyronyx sp.

(Elmidae) appeared to prefer upland sites, whereas Dubiraphia sp. (Elmidae) densities were higher within the Flatwoods sites, possibly indicating a preference for a more lentic environment with abundant woody debris. Odonate richness was fairly consistent across terraces, but habitat affinities were apparent within certain families (Figure 2.7). Generic richness and abundances within the Family Libellulidae were higher for the Prairie and Flatwoods terraces, whereas richness within the family Gomphidae was higher for

Uplands sites. Taxa such as Boyeria sp. (Aeshnidae) and Calopteryx sp. (Calopterygidae) were exclusive to the Uplands. Argia sp. (Coenagrionidae) was ubiquitous among terraces, whereas Ischnura sp. and Nehalennia sp. were much more associated with Prairie terrace sites.

Coleoptera taxa by Terrace and Date 0.20 Chrysomelidae Hydraena Curculionidae Berosus Cyrtobagous Cymbiodyta Sphenophorus Helochares Tournotaris Helocombus Dryopidae Helophorus 0.15 Helichus Hydrophilidae

Dytiscidae Tropisternus

y t

i Neoporus Hydrocanthus s

n Pachydrus Noteridae

e Ancyronyx Notomicrus

t Dubiraphia Suphisellus o

o Elmidae Ectopria

r 0.10

- Macronychus Ptiliidae h

t Microcylloepus Prionocyphon

n Stenelmis Scirtes

a Dineutus Scirtidae e

M Gyretes Staphylinidae 0.05 Peltodytes

P F F

Figure 2.6. Seasonal variability of mean 4th-root transformed individual Coleoptera taxa within terraces.

Odonata taxa by Terrace and Date 0.08 Aeshnidae Gomphus Boyeria Hagenius Nasiaeshna Progomphus Calopteryx Stylurus Hetaerina Erythemis Amphiagrion Libellula

0.06 Argia Libellulidae y

t Chromagrion Macrothemis i

s Coenagrionidae Miathyria n

e Enallagma Nannothemis

t Ischnura Orthemis

o Nehalennia Pachydiplax o

r 0.04

- Telebasis Perithemis h

t Epitheca Sympetrum

n Neurocordulia Didymops a

e Aphylla Macromia

M Ariogomphus Neoneura 0.02 Dromogomphus Protoneuridae Gomphidae Unknown

P F F

Figure 2.7. Seasonal variability of mean 4th-root transformed individual Odonata taxa within terraces.

Among Ephemeroptera, Plecoptera and Trichoptera taxa, habitat selection evidence was much stronger

(Figures 2.8-2.10). Sites within the Uplands terrace demonstrated the greatest ephemeropteran diversity

(19 genera; Figure 2.8). Caenis sp. [likely Caenis hilaris (Say), family Caenidae] was ubiquitous across all terraces, dominating mayfly abundances within the Flatwoods and Prairie sites, but with much lower densities in the Uplands terrace. Plecoptera taxa (9 genera) were restricted to uplands sites (Figure 2.9) and were generally rare during the study year. Caddisflies demonstrated substantial differences in both taxonomic distinctness and abundance between terraces (Figure 2.10). After the first sampling event,

Trichoptera taxa were not collected at any of the Flatwoods sites. By contrast, 18 genera were found in the adjacent Uplands terrace, a few of which (e.g., Hydropsyche sp., Oecetis sp.) also occurred in the

Prairie terrace immediately to the east. The Prairie terrace caddisfly assemblage was dominated by

Hydropsychidae, but was also fairly diverse (10 genera). Nyctiophylax sp. (Polycentropodidae) demonstrated the highest densities throughout the year in the Uplands terrace.

Ephemeroptera taxa by Terrace and Date 0.25 Acerpenna Anafroptilum Baetidae Callibaetis Centroptilum 0.20 Heterocloeon Paracloeodes

Procloeon

y t

i Pseudocentroptiloides s

n Caenis

e 0.15 Eurylophella

t Hexagenia o

o Heptageniidae r

- Maccaffertium h

t Stenacron 4 0.10 n Stenonema

a Asioplax e

M Leptohyphidae Trichorythodes 0.05 Choroterpes Leptophlebia Leptophlebiidae Paraleptophlebia Anthopotamus

0 0 0 1 1 1 1 0 0 0 1 1 1 1 0 0 0 1 1 1 1

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2

______

8 0 2 2 4 6 8 8 0 2 2 4 6 8 8 0 2 2 4 6 8

_ 1 1 _ _ _ _ _ 1 1 _ _ _ _ _ 1 1 _ _ _ _

______

P P P P P

R R R R R

W W W W W

P P

R R

U U U U U P P P P P

W W

F F F F F

U U P P F F

Figure 2.8. Seasonal variability of mean 4th-root transformed individual Ephemeroptera taxa within terraces.

Plecoptera taxa by Terrace and Date 0.05 Acroneuria Agnetina Attaneuria Beloneuria 0.04 Neoperla

Paragnetina

y t

i Perlesta

s n

e 0.03 Perlidae

t Taeniopteryx

t 4

0.02

e M

0.01

P F F

Figure 2.9. Seasonal variability of mean 4th-root transformed individual Plecoptera taxa within terraces.

Trichoptera taxa by Terrace and Date 0.15 Brachycentrus Dipseudopsidae Phylocentropus Cheumatopsyche Hydropsyche Hydropsychidae

Potamyia y

t Hydroptila i 0.10

s Hydroptilidae n

e Neotrichia

t Ochrotricia

o Oxyethira

o r

- Leptoceridae h

t Oecetis

n Setodes a

e 0.05 Molanna

M Psilotreta Chimarra Philopotamidae Neureclipsis Nyctiophylax Polycen_Cyrn Polycentropodidae 0 Lype

Trichoptera

0 0 0 1 1 1 1 0 0 0 1 1 1 1 0 0 0 1 1 1 1

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2

______

8 0 2 2 4 6 8 8 0 2 2 4 6 8 8 0 2 2 4 6 8

_ 1 1 _ _ _ _ _ 1 1 _ _ _ _ _ 1 1 _ _ _ _

______

P P P P P

R R R R R

W W W W W

P P

R R

U U U U U P P P P P

W W

F F F F F

U U P P F F

Figure 2.10. Seasonal variability of mean 4th-root transformed individual Trichoptera taxa within terraces.

A seriated (i.e., ordered) shade plot based on presence/absence data of individual taxa (as rows) within

assemblages across samples (as columns) illustrates the subtle gradient of habitat affiliations within the

entire macroinvertebrate community (Figure 2.11). Taxa are ordered according to similarity across

samples based on Whittaker’s Index of Association (Clarke and Gorley 2015). Samples are ordered

according to Jaccard similarity across taxa, as calculated previously for NMDS ordination and

PERMANOVA tests. Taken as a whole, the macroinvertebrate community contained a large number of

generalist taxa. The Uplands terrace sites, however, demonstrated higher insect diversity than the Prairie

or Flatwoods. Furthermore, the seriated shade plot suggested several insect taxa exhibited positive

associations for either upland (i.e., Uplands terrace) or lowland (i.e., Prairie or Flatwoods terrace) habitats

(Table 2.4).

Sisyridae For cipom yia Trichogr am m at idae Valvat idae Pleur oblem a Terrace Sympet r um Limonia O dont om yia Prot oneur idae Helophor us M egistocer a Asioplax Cymbiodyt a Sisyridae Figitidae For cipom yia Trichogr am m at idae Ptiliidae Valvat idae Cyrt obagous Uplands Pleur oblem a Helichus Terrace Sympet r um Taeniopt er yx Limonia G om phidae O dont om yia Nem at om or pha Prot oneur idae Trepobat es Helophor us M egistocer a Simulium Asioplax Smint hur idae Cymbiodyt a Daphnia Figitidae Libellula Ptiliidae P s eudolimnophila

Cyrt obagous Uplands Flatwoods Att aneur ia Helichus Ellipt er oides Taeniopt er yx Pilar ia

G om phidae

Nem at om or pha M olophilus Trepobat es Eriopt er a Simulium Brachycent r us Smint hur idae H ydr ophilidae Daphnia Ant hopot am us Libellula Par acloeodes P s eudolimnophila Trichor yt hodes Flatwoods Att aneur ia Simuliidae Prairie Ellipt er oides Pilar ia Par agnet ina M olophilus O r t hem is Eriopt er a Beloneur ia Brachycent r us Set odes H ydr ophilidae Staphylinidae Ant hopot am us Boyer ia Par acloeodes Trichor yt hodes Lept ohyphidae Aeshnidae Simuliidae Prairie Par agnet ina Veliidae O r t hem is Pachydr us Beloneur ia D ipseudopsidae Set odes Neoper la Staphylinidae Acroneur ia Boyer ia O r m osia Lept ohyphidae Aeshnidae Tabanus Veliidae Hagenius Pachydr us Calopt er yx D ipseudopsidae P seudocent r opt iloides Neoper la Stylur us Acroneur ia Stenonem a O r m osia Agnet ina Tabanus Hagenius Upland O xyet hira Calopt er yx Per lest a P seudocent r opt iloides Het aer ina Stylur us O chr ot r icia Stenonem a Synur ella Agnet ina Neot r ichia O xyet hira Lept ophlebia Per lest a Philopot am idae Het aer ina O chr ot r icia M olanna Synur ella Hydr om et r a Neot r ichia Hexat om a Lept ophlebia P ar alept ophlebia Philopot am idae Hem er odr om ia M olanna Ather ix Hydr om et r a Per lidae Hexat om a Procloeon P ar alept ophlebia Hem er odr om ia Progom phus Ather ix M accaf f er t ium Per lidae Hexagenia Procloeon Chimar r a Progom phus Lept ocer idae M accaf f er t ium Hexagenia Acer penna Chimar r a Hept ageniidae Nyctiophylax Lept ocer idae Acer penna L ept ophlebiidae Hept ageniidae Anaf r opt ilum Nyctiophylax Eur ylophella L ept ophlebiidae Ectopr ia Anaf r opt ilum G om phus Eur ylophella Chor ot er pes Ectopr ia G om phus Tipula Chor ot er pes M icrocylloepus Tipula Brachonidae M icrocylloepus M acr onychus Brachonidae Hydr opt ila M acr onychus Lepidopt er a Hydr opt ila Xest ochironom us Lepidopt er a Xest ochironom us G yr et es G yr et es Lype Hydr opt ilidae Lype Hydr opt ilidae Het er ocloeon Het er ocloeon Cheum at opsyche Cheum at opsyche Ancyr onyx Ancyr onyx O ecet is O ecet is Sialis Sialis Chr ysops Chr ysops Polycen_Cyrn Polycen_Cyrn Hydr opsyche Hydr opsyche Cam bar idae Cam bar idae Drom ogom phus Drom ogom phus D olichopodidae D olichopodidae Neur ocor dulia Neur ocor dulia Stenacr on Stenacr on Argia Neur eclipsis Argia Dipt er a Neur eclipsis Culicoides Dipt er a Stenelmis Culicoides Hirudinea Stenelmis Per icom a Sphaer iidae Hirudinea Chironom idae Per icom a Sphaer iidae Baet idae Trichopt er a Chironom idae O r t hocladiinae Baet idae Cor ixidae Trichopt er a Lirceus O r t hocladiinae Sem icer ur a Cor ixidae Chironom ini Lirceus Chr ysom elidae Lum br iculidae Sem icer ur a Tanyt ar sini Chironom ini Cor ydalus Chr ysom elidae Dubiraphia Lum br iculidae Psychodidae Tanyt ar sini Tipulidae Cor ydalus Tanypodinae Dubiraphia Neopor us Acar i Psychodidae Tipulidae L ibellulidae Tanypodinae Tubificidae Tur bellar ia Neopor us Crangonyx Acar i Nem at oda L ibellulidae Chauliodes Tubificidae Dineut us Tur bellar ia Enallagm a Scirt idae Crangonyx Prionocyphon Nem at oda M ysidae Chauliodes Copepoda Dineut us Laevapex Enallagm a Cladocer a Scirt idae Palaem onet es Prionocyphon E r pobdellidae M ysidae Naididae Atr ichopogon Copepoda G lossiphoniidae Laevapex Chaobor us Cladocer a O st r acoda Palaem onet es Physidae E r pobdellidae C ur culionidae Naididae Haplot axidae Generalist Atr ichopogon Cor bicula G lossiphoniidae Peltodyt es Collem bola Chaobor us Chr om agr ion O st r acoda H elobdella Physidae Hyalella C ur culionidae Planor bidae Haplot axidae Caenis Cor bicula P lacobdella Lym naeidae Peltodyt es Per ithem is Collem bola Ber osus Chr om agr ion M usculium H elobdella Scirt es Hyalella Hydr obiidae Planor bidae Polychaet a Caenis Bat r acobdella Pachydiplax P lacobdella Lym naeidae Let hocer us Per ithem is Pisidium I schnur a Ber osus Crypt ochironom us M usculium Unionidae Scirt es Euper a Hydr obiidae Nehalennia Polychaet a Cer at opogonidae Toxolasm a Bat r acobdella Chlor ot abanus Pachydiplax Hydr a Let hocer us Tritogonia Pisidium Trichocor ixa I schnur a Aphylla Crypt ochironom us Ligum ia Unionidae Hydr aena Euper a M acr ot hem is Neoneur a Nehalennia Dryopidae Cer at opogonidae Sciom yzidae Toxolasm a Callibaet is Chlor ot abanus Seira Hydr a Ariogom phus Phylocent r opus Tritogonia Trichocor ixa Limnophila Aphylla Rheum at obat es Limnocor is Ligum ia Tabanidae Hydr aena Uniom er us M acr ot hem is C oenagr ionidae Neoneur a

Nannot hem is Dryopidae Telebasis Sciom yzidae Not er idae

P iscicolidae Callibaet is

Neot r idact ylus Seira A l boglossiphonia Ariogom phus Sisyra Phylocent r opus Psilot r et a Limnophila Buenoa Rheum at obat es M acr om ia Limnocor is Pot am yia Epitheca Tabanidae Uniom er us Par aplea Dytiscidae C oenagr ionidae Not om icrus Nannot hem is Helocom bus Telebasis Tropister nus Not er idae Cor ydalidae P iscicolidae Eryt hem is Neot r idact ylus Hydr ocant hus Nasiaeshna A l boglossiphonia Cram bidae Sisyra Lophognat hella Psilot r et a Abedus Buenoa M iat hyr ia M acr om ia P olycent r opodidae Pot am yia M ar vinm eyer ia Epitheca Par adelphom yia O ligochaet a Par aplea Hesper ocor ixa Dytiscidae Sphenophor us Not om icrus Elmidae Helocom bus Hem ipt er a Tropister nus Helochar es Cor ydalidae Taphr om ysis Not onect a Eryt hem is Hydr ocant hus Allognost a Suphisellus Nasiaeshna O donat a Cram bidae Hydr opsychidae Lophognat hella I sot om idae Abedus Limnopor us M iat hyr ia Amphiagr ion P olycent r opodidae Ancylidae Lowland M ar vinm eyer ia M ackenziella Par adelphom yia Tour not ar is O dont o_Hedr io O ligochaet a Cent r opt ilum Hesper ocor ixa Didym ops Sphenophor us Hirudinidae Elmidae Ent om obr yidae Hem ipt er a Helochar es Taphr om ysis Not onect a Allognost a Suphisellus O donat a Hydr opsychidae I sot om idae Limnopor us Amphiagr ion Ancylidae M ackenziella Tour not ar is O dont o_Hedr io Cent r opt ilum Didym ops Hirudinidae Ent om obr yidae

Figure 2.11. Seriated shade plot (presence/absence) of individual taxa (rows) within macroinvertebrate assemblages across samples (columns). Taxa are ordered according to similarity across samples using Whittaker’s Index of Association. Samples are ordered according to Jaccard similarity across taxa.

Table 2.4. List of insect taxa with potential habitat preference for either upland (U) or lowland (L) sites based on collections for this study. Taxa exhibiting no preference (i.e., habitat generalist) not listed.

Order Family Genus Order Family Genus Collembola Entomobryidae ( L ) Seira Coleoptera Dytiscidae ( U ) Pachydrus Collembola Mackenziellidae ( L ) Mackenziella Coleoptera Hydraenidae ( L ) Hydraena Collembola Onychiuridae ( L ) Lophognathella Coleoptera Hydrophilidae ( U ) Cymbiodyta Collembola Sminthuridae ( U ) Sminthuridae Coleoptera Hydrophilidae ( L ) Helochares Diptera Athericidae ( U ) Atherix Coleoptera Hydrophilidae ( L ) Helocombus Diptera Empididae ( U ) Hemerodromia Coleoptera Hydrophilidae ( L ) Tropisternus Diptera Simuliidae ( U ) Simulium Coleoptera Noteridae ( L ) Hydrocanthus Diptera Tabanidae ( L ) Chlorotabanus Coleoptera Noteridae ( L ) Notomicrus Diptera Tipulidae ( U ) Ellipteroides Coleoptera Psephenidae ( U ) Ectopria Diptera Tipulidae ( U ) Erioptera Ephemeroptera Baetidae ( U ) Acerpenna Diptera Tipulidae ( U ) Hexatoma Ephemeroptera Baetidae ( U ) Anafroptilum Diptera Tipulidae ( U ) Molophilus Ephemeroptera Baetidae ( L ) Callibaetis Diptera Tipulidae ( U ) Ormosia Ephemeroptera Baetidae ( U ) Paracloeodes Diptera Tipulidae ( L ) Paradelphomyia Ephemeroptera Baetidae ( U ) Procloeon Diptera Tipulidae ( U ) Pilaria Ephemeroptera Baetidae ( U ) Pseudocentroptiloides Diptera Tipulidae ( U ) Pseudolimnophila Ephemeroptera Ephemeridae ( U ) Hexagenia Hemiptera Belostomatidae ( L ) Abedus Ephemeroptera Heptageniidae ( U ) Maccaffertium Hemiptera Corixidae ( L ) Hesperocorixa Ephemeroptera Heptageniidae ( U ) Stenonema Hemiptera Corixidae ( L ) Trichocorixa Ephemeroptera Leptohyphidae ( U ) Tricorythodes Hemiptera Gerridae ( L ) Rheumatobates Ephemeroptera Leptophlebiidae ( U ) Choroterpes Hemiptera Gerridae ( U ) Trepobates Ephemeroptera Leptophlebiidae ( U ) Leptophlebia Hemiptera Hydrometridae ( U ) Hydrometra Ephemeroptera Leptophlebiidae ( U ) Paraleptophlebia Hemiptera Naucoridae ( L ) Limnocoris Ephemeroptera Potamanthidae ( U ) Anthopotamus Hemiptera Notonectidae ( L ) Buenoa Plecoptera Perlidae ( U ) Acroneuria Hemiptera Notonectidae ( L ) Notonecta Plecoptera Perlidae ( U ) Agnetina Hemiptera Pleidae ( L ) Paraplea Plecoptera Perlidae ( U ) Attaneuria Odonata Aeshnidae ( U ) Boyeria Plecoptera Perlidae ( U ) Beloneuria Odonata Aeshnidae ( L ) Nasiaeshna Plecoptera Perlidae ( U ) Neoperla Odonata Calopterygidae ( U ) Hetaerina Plecoptera Perlidae ( U ) Paragnetina Odonata Coenagrionidae ( L ) Ischnura Plecoptera Perlidae ( U ) Perlesta Odonata Coenagrionidae ( L ) Telebasis Plecoptera Taeniopterygidae ( U ) Taeniopteryx Odonata Corduliidae ( L ) Epitheca Trichoptera Brachycentridae ( U ) Brachycentrus Odonata Gomphidae ( L ) Aphylla Trichoptera Dipseudopsidae ( U ) Phylocentropus Odonata Gomphidae ( L ) Ariogomphus Trichoptera Hydropsychidae ( L ) Potamyia Odonata Gomphidae ( U ) Hagenius Trichoptera Hydroptilidae ( U ) Neotrichia Odonata Gomphidae ( U ) Stylurus Trichoptera Hydroptilidae ( U ) Ochrotricia Odonata Libellulidae ( L ) Erythemis Trichoptera Hydroptilidae ( U ) Oxyethira Odonata Libellulidae ( U ) Libellula Trichoptera Leptoceridae ( U ) Setodes Odonata Libellulidae ( L ) Macrothemis Trichoptera Molannidae ( U ) Molanna Odonata Libellulidae ( L ) Miathyria Trichoptera Odontoceridae ( L ) Psilotreta Odonata Libellulidae ( L ) Nannothemis Trichoptera Philopotamidae ( U ) Chimarra Odonata Libellulidae ( U ) Orthemis Trichoptera Polycentropodidae ( U ) Nyctiophylax Odonata Libellulidae ( L ) Pachydiplax Neuroptera Sisyridae ( L ) Sisyra Odonata Macromiidae ( L ) Macromia Orthoptera Tridactylidae ( L ) Neotridactylus Odonata Protoneuridae ( L ) Neoneura

-1 Mean macroinvertebrate density in wood samples (number mL ; F2,13.38=1.89, p=0.1899) and total mean density of all insect taxa (F2,13.73=1.62, p=0.2334) did not differ between geologic terraces (. The proportion of the macroinvertebrate assemblage represented by insect taxa, however, did demonstrate a

46 terrace effect (F2,10.11=19.91, p=0.0003), with the largest proportion in the Uplands terrace sites

(mean±SE 0.92±0.04), the lowest proportion in the Flatwoods (0.48±0.06) and intermediate proportion in the Prairie terrace (0.68±0.05). Proportion of insect taxa was not significantly different between the

Prairie and Flatwoods, but was higher in the Uplands than both Prairie (p=0.0096) and Flatwoods

(p=0.0003). Further comparisons based on functional feeding groups revealed a significant terrace effect for predators (F2,10.49=6.31, p=0.0159), collector-gatherers (F2,12.82=19.24, p=0.0001), collector-filterers

(F2,6.27=19.63, p=0.002) and scrapers (F2,9.02=8.09, p=0097) with similar pairwise patterns, although not all differed significantly (Figure 2.12). Low numbers of shredders and piercers precluded model estimation for those groups.

Proportion of functional feeding groups as insect taxa by terrace 1.2 b b b 1 c ab

0.8 b a a ab Uplands 0.6 a Flatwoods a a Prairie

0.4

0.2

0 Predators Collector-Gatherers Collector-Filterers Scrapers

Figure 2.12. Proportion of functional feeding group represented by insect taxa (in wood samples) within each geologic terrace. Error bars indicate plus one standard error for the mean proportion. Bars with different letters are significantly different within a trophic group.

Mean macroinvertebrate density in sediment samples (number cm-2) was significantly higher

(F2,14.12=10.93, p=0.0014) in the Prairie terrace (mean±SE 0.054±0.005) than in the Flatwoods

(0.031±0.006) or the Uplands (0.022±0.005), which did not differ from each other. In contrast, mean density of insect taxa among terraces was not significantly different (F2,12.1=1.59, p=0.2446). As with wood samples, the proportion of the macroinvertebrate assemblage represented by insect taxa in sediment did demonstrate a terrace effect (F2,15.04=24.94, p<0.0001), with a significantly larger proportion of insect taxa found in the Uplands terrace (mean±SE 0.81±0.04), and smaller proportions in the

Flatwoods (0.34±0.06) and Prairie terraces (0.40±0.05). Similar patterns were demonstrated within each functional feeding group, especially collector-filterers and scrapers (Figure 2.13).

Proportion of functional feeding groups as insect taxa by terrace 1.2

b 1 b

a a a b 0.8

Uplands 0.6 a Flatwoods Prairie a a 0.4 a a a 0.2

0 Predators Collector-Gatherers Collector-Filterers Scrapers

Figure 2.13. Proportion of functional feeding group represented by insect taxa (in sediment samples) within each geologic terrace. Error bars indicate plus one standard error for the mean proportion. Bars with different letters are significantly different within a trophic group.

Discussion

Habitat features within streams of the Gulf coastal plain of southwestern Louisiana were distinct for the

Tertiary Uplands and Prairie Pleistocene terraces. The primary habitat gradient was structured on water quality (specific conductance, turbidity) and habitat complexity (amount of woody debris, percent overstory cover, substrate size), while a secondary gradient demonstrated differences in stream flow and channel dimensions (width, bank height). These features had a strong influence on macroinvertebrate assemblages, which also were diagnostic for the Uplands and Prairie terraces, despite containing a number of generalist taxa. Agreement (i.e., rank correlation) between the multivariate patterns for abiotic and biotic data was significant overall, but was improved by focusing on habitat variables that described the strong differences in water quality and habitat complexity between the terraces. As indicated by other studies conducted in this region, benthic macroinvertebrate taxa generally were not responsive to the dissolved oxygen gradient (Skrobialowski et al. 2004, Kaller and Kelso 2007, Justus et al. 2014). Kaller

(2005), Daniel (2012) and Budnick (2015) found habitat gradients in streams of the Louisiana coastal plain that were strongly influenced by specific conductance, which can be correlated with system productivity, i.e. high nutrient nitrogen and phosphorus concentrations typically occur in waters with high specific conductance (Dodds and Whiles 2010). Elevated specific conductance can indicate higher concentrations of non-nutrient salts, and although the range of measurements during this study (0.022 -

0.968 mS/cm) was relatively large, readings at the upper end of the range were rare, and there was no indication of salt water intrusion at the study sites.

Historic drought condition throughout the study year undoubtedly affected the strength of these relationships. Seasonal fluctuations in density were shown in a few taxa, but others were consistently present throughout the study year. Stronger temporal fluctuations in density were expected for these sites, as regular spates characteristic of these flashy streams have more than likely shaped the macroinvertebrate assemblages (Felley 1992, Miller and Golladay 1996, Reznickova et al. 2007).

Drought reduced flow at all sites such that baseflow and occasional surface runoff from draining rice ponds constituted the only water inputs during the study period. Effects were most pronounced in the

Flatwoods sites where flows ceased altogether due to obstruction by beaver dams. Low water level,

49 stable substrate and abundant woody debris probably increased the likelihood of disturbance by beavers within the Flatwoods terrace, as compared to the Uplands sites which had shifting sand substrate and the

Prairie sites which contained less wood. Macroinvertebrates, especially beetles, within these sites tended to be more lentic or slow moving water associates (e.g., Dubiraphia sp., Hydrophilidae, Scirtidae; Kaller and Kelso 2007, Merritt et al. 2008), with certain groups (i.e., EPT taxa) severely under-represented, given their abundances in adjacent terraces.

Although the macroinvertebrate fauna of the Gulf coastal plain of Louisiana comprises many generalist taxa, which have adapted to tolerate seasonal shifts between a lotic and lentic environment in these low- gradient streams, the presence of several taxa with strong habitat preferences may improve the potential for diagnostic indicators in the region. Non-specific biotic metrics such as taxa richness or diversity should be restricted to Ephemeroptera, Plecoptera and Trichoptera, which demonstrated strong relationships with water velocity, habitat complexity and water quality. Members of other diverse insect orders, such as the Coleoptera, exhibited a wide range of habitat and water quality tolerances, with some taxa exhibiting associations with particular terraces, as with certain Elmidae, Scirtidae and Hydrophilidae.

Presence/absence data were sufficient to distinguish between terraces, indicating that taxonomic distinctness rather than differences in abundances discriminate between these terrace assemblages.

This provides further evidence that the gradient of habitat conditions experienced by macroinvertebrate taxa in these low-gradient streams may be so extreme that it not only selects for habitat-generalist taxa, but also has the potential to select between lentic and lotic assemblages.

In sediment samples, overall macroinvertebrate density was greater in the Prairie terrace, but was dominated by non-insect taxa. In wood samples, neither overall density of macroinvertebrates nor density of insect taxa differed significantly among terraces, but the proportion of the assemblage represented by insect taxa was much higher in the Uplands. The same pattern was demonstrated for each functional feeding group, and was most pronounced for scrapers and collectors (Figures 2.12 and

2.13). This indicated that, within the Prairie and Flatwoods terraces, insect taxa within each trophic niche have been replaced by more tolerant non-insect taxa such as gastropods (scrapers), bivalves (filterers) and oligochaetes or crustaceans (collectors). In the Flatwoods sites, which had abundant woody debris

50 inputs, the loss of insect specialists was probably the result of the cessation of stream flow due to beaver activities during the drought. In the Prairie terrace, however, simplification of food resources resulting from agricultural impacts, as suggested by Parreira de Castro (2016), may be an important factor structuring macroinvertebrate assemblages. Consequently, the proportion of insect taxa in the community may represent a useful bio-assessment metric across these terraces.

In the context of a potential climate shift in this region, with alternating extreme flood and drought conditions, aquatic fauna already pressed by chronic environmental stressors may be pushed beyond their ability to tolerate these extremes. Borrok and Broussard (2016) provided evidence of significant drawdown of the Chicot aquifer since 1900, citing rice irrigation as a culprit, and showed that demand increased during the 2011 drought. The documented loss of potentiometric surface (30 m in some locations) suggests potential disconnection of some prairie water bodies from this aquifer, which provides baseflow during low-precipitation periods maintaining or restoring baseflow, especially in the most vulnerable bayous of the Prairie terrace, may reduce the impacts of extreme fluctuations in habitat condition. Vidrine et al. (2004) described historic condition in these slow-moving bayous as perennially fed by fresh groundwater discharge and “cobbled with mussels.” Very similar conditions were observed throughout the drought period at one site, Middle Bayou Serpent, which drained the largest watershed sampled within the Prairie terrace. Macroinvertebrate diversity (mean±SD, Shannon H’=2.39±0.53) and richness (S=13±5.5) at this site were consistently higher than other Prairie sites (H’=2.14±0.47,

S=9.9±4.2) and, although mussels were not enumerated as part of this study, high diversity and abundance of Unionid mussels were observed. In southeastern Louisiana streams, higher mussel abundance in silt versus sand has been attributed to the stability of the former substrate over time (Brown and Banks 2001, Daniel and Brown 2013). During this study, mussels were observed in greatest abundance at sites with a small amount of silt over hardpan clay. Large mussel beds may support abundant and diverse invertebrate and haptobenthic assemblages by increasing complexity in substrates that lack woody structure, providing surface area for growth of aufwuchs and creating refuges for predator avoidance (Vidrine et al. 2004). These unique ecosystems clearly represent reference condition for the

Prairie terrace, and a shift in conservation focus from non-point source agricultural impacts to restoration

51 of baseflow and high mussel abundance could improve habitat conditions and outcomes for all aquatic biota.

Daniel et al. (2014) proposed a model linking habitat features, including hydrologic disturbance, to freshwater mussel assemblage structure in southeastern Louisiana. A version of this model, adjusted to accommodate the hydrologic conditions in bayous such as those sampled for this study, could be applied to the Prairie terrace to identify least impaired sites with the greatest restoration potential. These data could be related to data from nearby wells to evaluate the effects, if any, of water withdrawals on baseflow and habitat suitability (Borrok and Broussard 2016). I believe managing water resources in this area to maximize mussel diversity will result in benefits to fish and macroinvertebrate assemblages as well, because of their interrelated life histories and similarities in habitat associations (Vidrine et al. 2004,

Kaller 2005, Daniel et al. 2014). In addition, macroinvertebrate assemblages inhabiting streams of the

Tertiary Uplands terrace represent the best potential source population for recolonization of improved habitats of the Flatwoods and Prairie terraces and should, therefore, be protected to the greatest extent possible.

References

Adams, S. M., M. G. Ryon, and J. G. Smith. 2005. Recovery in diversity of fish and invertebrate communities following remediation of a polluted stream: investigating causal relationships. Hydrobiologia 542: 77-93.

Anderson, M. J., R. N. Gorley and K. R. Clarke. 2008. Permanova+ for Primer: Guide to Software and Statistical Methods. 214 p.

Angermeier, P. L., and J. R. Karr. 1994. Biological integrity versus biological diversity as policy directives: Protecting biotic resources. BioScience 44:690-697.

Bailey, R. C., T. B. Reynoldson, A. G. Yates, J. Bailey and S. Linke. 2007. Integrating stream bioassessment and landscape ecology as a tool for land use planning. Freshwater Biology 52(5):908-917.

Barbour, M.T., J. Gerritsen, B.D. Snyder, and J.B. Stribling. 1999. Rapid Bioassessment Protocols for Use in Streams and Wadeable Rivers: Periphyton, Benthic Macroinvertebrates and Fish, Second Edition. EPA 841-B-99-002. U.S. Environmental Protection Agency; Office of Water; Washington, D.C.

Beghelli, F.G.S., A.C.A., Dos Santos, M.V. Urso-Guimaraes, and M.C. Calijuri. 2012. Relationship between space distribution of the benthic macroinvertebrates community and trophic state in a Neotropical reservoir (Itupararanga, Brazil). Biota Neotropica 12(4): http://www.biotaneotropica.org.br/v12n4/en/abstract?article+bn02812042012.

Benke, A. C. and A. D. Huryn. 2010. Benthic invertebrate production–facilitating answers to ecological riddles in freshwater ecosystems. Journal of the North American Benthological Society 29: 264-285.

Borrok, D. M. and W. P. Broussard III. 2016. Long-term geochemical evaluation of the coastal Chicot aquifer system, Louisiana, USA. Journal of Hydrology 533: 320–331.

Brown, J. H. 1995. Macroecology. The University of Chicago Press, Ltd. London, UK. Pp 269.

Brown, K.M., and P.D. Banks. 2001. The conservation of unionid mussels in Louisiana Rivers: diversity, assemblage composition and substrate use. Aquatic conservation: Marine and Freshwater Ecosystems, 11: 189-198.

Budnick, W. R. 2015. Metacommunity dynamics and the biogeography of Central Louisiana crayfishes. M.S. Thesis, Louisiana State University, Baton Rouge, Louisiana. 97 p.

Buffagni, A. and M. Furse. 2006. Intercalibration and comparison – major results and conclusions from the STAR project. Hydrobiologia 566: 357-364.

Buss, D. F., D. M. Carlisle, T. S. Chon, J. Culp, J. S. Harding, H. E. Keizer-Vlek, W. A. Robinson, S. Strachan, C. Thirion and R. M. Hughes. 2014. Stream biomonitoring using macroinvertebrates around the globe: a comparison of large-scale programs. Environmental Monitoring and Assessment, 187(1): 4132.

Carlisle, D. M., C. P. Hawkins, M. R. Meador, M. Potapova, and J. Falcone. 2008. Biological assessments of Appalachian streams based on predictive models for fish, macroinvertebrate, and diatom assemblages. Journal of the North American Benthological Society 27(1): 16-37.

Chen, K., R. M. Hughes, S. Xu, J. Zhang, D. Cai, and B. Wang. 2014. Evaluating performance of macroinvertebrate-based adjusted and unadjusted multi-metric indices (MMI) using multi-season and multi-year samples. Ecological Indicators 36: 142-151.

Clarke, R. T. and D. Hering. 2006. Errors and uncertainty in bioassessment methods – major results and conclusions from the STAR project and their application using STARBUGS. Hydrobiologia 566: 433-439.

Clarke, K.R. and R. N. Gorley. 2015. PRIMER v7: User Manual/Tutorial. PRIMER-E, Plymouth, 296pp.

Colas, F., A. Vigneron, V. Felten and S. Devin. 2014. The contribution of a niche-based approach to ecological risk assessment: Using macroinvertebrate species under multiple stressors. Environmental Pollution 185: 24-34.

Daniel, W. M. 2012. Modeling Effects of Instream Variables, Land Use, and Life History Attributes on Community Structure of Freshwater Mussels in Louisiana Streams. Department of Biological Sciences. Baton Rouge, LA, Louisiana State University. Doctor of Philosophy dissertation. Pp. 97.

Daniel, W. M. and K. M. Brown. 2013. Multifactorial model of habitat, host fish, and landscape effects on Louisiana freshwater mussels. Freshwater Science 32(1):193-203.

Daniel, W., K.M Brown, and M.D. Kaller. 2014. A hybrid Tiered Aquatic Life Unit (Hybrid-TALU) bioassessment model for Gulf of Mexico coastal streams. Fisheries Management and Ecology 21: 491- 502.

Davy-Bowker, J., R. T. Clarke, R. K. Johnson, J. Kokes, J. F. Murphy and S. Zahradkova. 2006. A comparison of the European Water Framework Directive physical typology and RIVPACS-type models as alternative methods of establishing reference condition for benthic macroinvertebrates. Hydrobiologia 566: 91-105.

Dodds, W. K., and M. R. Whiles. 2010. Freshwater ecology: concepts and environmental applications of limnology. 2nd edition. Elsevier, Dordrecht, The Netherlands. Pp 829.

Felley, J. D. 1992. Medium-low-gradient streams of the Gulf coast plain. Pp. 233-269, In: C. T. Hackney, S. M. Adams, and W. H. Martin, eds. Biodiversity of the southeastern United States. John Wiley & Sons, Inc., New York.

Greathouse, E.A. and C.M. Pringle. 2006. Does the river continuum concept apply on a tropical island? Longitudinal variation in a Puerto Rican stream. Canadian Journal of Fisheries and Aquatic Science 63: 134–152.

Gutzwiller, K. J., and S. K. Riffell. 2007. Using statistical models to study temporal dynamics of animal- landscape relations. Pages 93-118 in J.A. Bissonette and I. Storch (eds), Temporal Dimensions of Landscape Ecology: Wildlife Responses to Variable Resources. Springer-Verlag, Inc., New York. 284 pp.

Heino J, H. Mykra, H. Hamalainen, J. Aroviita and T. Muotka. 2007. Responses of taxonomic distinctness and species diversity indices to anthropogenic impacts and natural environmental gradients in stream macroinvertebrates. Freshwater Biology 52:1846–1861.

Heino, J., H. Mykra and J. Kotanen. 2008. Weak relationships between landscape characteristics and multiple facets of stream macroinvertebrate biodiversity in a boreal drainage basin. Landscape Ecology 23:417–426.

Hering, D., C. K. Feld, O. Moog, and T. Ofenbock. 2006. Cook book for the development of a Multimetric Index for biological condition of aquatic ecosystems: experiences from the European AQEM and STAR projects and related initiatives. Hydrobiologia 566: 311-324.

Jacobi, D. I, and A. C. Benke. 1991. Life histories and abundance patterns of snag-dwelling mayflies in a blackwater Coastal Plain river. Journal of the North American Benthological Society 10:372- 387.

Kaller, M. D. 2005. Macroinvertebrate community ecology of lowland, subtropical streams in Louisiana. The School of Renewable Natural Resources. Baton Rouge, LA, Louisiana State University. Doctor of Philosophy dissertation. Pp. 152.

Kaller, M. D., and W. E. Kelso. 2006a. Swine activity alters invertebrate and microbial communities in a coastal plain watershed. American Midland Naturalist 156: 163-177.

Kaller, M. D., and W. E. Kelso. 2006b. Effects of a small-scale clearing on habitat and macroinvertebrates of a costal bottomland stream in Louisiana. The Southwestern Naturalist 51(2): 143-151.

Karr, J. R. 1991. Biological integrity: A long-neglected aspect of water resource management. Ecological Applications 1(1): 66-84.

Karr, J. R. 1993. Defining and assessing ecological integrity: beyond water quality. Environmental Toxicology and Chemistry 12: 1521-1531.

Karr, J. R. & E. W. Chu, 1999. Restoring Life in Running Waters: Better Biological Monitoring. Island Press, Washington, DC. pp 220.

Karr, J. R. and E. W. Chu. 2000. Sustaining living rivers. Hydrobiologia 422: 1-14.

Karr, J. R. and C. O. Yoder. 2004. Biological Assessment and Criteria Improve Total Maximum Daily Load Decision Making. Journal of Environmental Engineering 130(6): 594-604.

Kennard, M. J., B. D. Harch, B. J. Pusey and A. H. Arthington. 2006. Accurately defining the reference condition for summary biotic metrics: A comparison of four approaches. Hydrobiologia 572:151-170.

Kinsolving, A. D., and M. B. Bain. 1993. Fish assemblage recovery along a riverine disturbance gradient. Ecological Applications 3(3): 531-544.

Kentucky Department for Environmental Protection. 2002. Methods for Assessing Biological Integrity of Surface Waters in Kentucky. Revision 1. KDOW Biological Assessment Methods. 182 p.

Lake, P. S., N. Bond, and P. Reich. 2007. Linking ecological theory with stream restoration. Freshwater Biology 52: 597-615.

Mandaville, S.M. 2002. Benthic Macroinvertebrates in Freshwaters-Taxa Tolerance Values, Metrics, and Protocols. Project H-1. Soil &Water Conservation Society of Metro Halifax, Nova Scotia, A58.

Merritt, R. W., K. W. Cummins and M. B. Berg. 2008. An introduction to the aquatic insects of North America, 4th ed. Kendall Hunt, Dubuque, Iowa. 1158 p.

Miller, A. M. and S. W. Golladay. 1996. Effects of spates and drying on macroinvertebrate assemblages of an intermittent and perennial prairie stream. Journal of the North American Benthological Society 15: 670–689.

Mize, S. V., S. D. Porter, and D. K. Demcheck. 2008. Influence of fipronil compounds and rice-cultivation land-use intensity on macroinvertebrate communities in streams of southwestern Louisiana, USA. Environmental Pollution 152: 491-503.

Parreira de Castro, D. M., Reis de Carvalho, D., Pompeu, P. dos S., Moreira, M. Z., Nardoto, G. B. and Callisto, M. 2016. Land Use Influences Niche Size and the Assimilation of Resources by Benthic Macroinvertebrates in Tropical Headwater Streams. PLoS ONE 11(3) e0150527.

Reznickova, P., P. Paril and S. Zahradkova. 2007. The Ecological Effect of Drought on the Macroinvertebrate Fauna of a Small Intermittent Stream – An Example from the Czech Republic. International Review of Hydrobiology 92 : 514–526.

Rosenberg, D. M. and V. H. Resh (Eds). 1993. Freshwater biomonitoring and benthic macroinvertebrates. New York: Chapman and Hall. 488 p.

Rosenberg, D.M. 1998. A National Aquatic Ecosystem Health Program for Canada: We should go against the flow. Bulletin of the Entomological Society of Canada. 30(4):144-152.

Sandin, L. and P. F. M. Verdonschot. 2006. Stream and river typologies – major results and conclusions from the STAR project. Hydrobiologia 566:33-77.

Scott, M. C. 2006. Winners and losers among stream fishes in relation to land use legacies and urban development in the southeastern U.S. Biological Conservation 127: 301-309.

Sloman, K. A., A. C. Taylor, N. B. Metcalfe, and K. M. Gilmour. 2001. Effects of an environmental perturbation on the social behaviour and physiological function of brown trout. Animal Behaviour 61: 325- 333.

Smith, D. G. 2001. Pennak’s Freshwater Invertebrates of the United States: Porifera to Crustacea. Douglas G. Smith 4th ed. John Wiley and Sons, Inc.: New York, NY. 638 p.

Smock, L.A. 1988. Life histories, abundance and distribution of some macroinvertebrates from a South Carolina, USA coastal plain stream. Hydrobiologia 157: 193-208.

Southerland, M. T., G. M. Rogers, M. J. Kline, R. P. Morgan, D. M. Boward, P. F. Kazyak, R. J. Klauda, and S. A. Stranko. 2007. Improving biological indicators to better assess the condition of streams. Ecological Indicators 7: 751-767.

Verdenschot, P. F. M. and O. Moog. 2006. Tools for assessing European streams with macroinvertebrates: major results and conclusions from the STAR project. Hydrobiologia 566: 299-309.

Williams, L. R., T. H. Bonner, J. D. Hudson III, M. G. Williams, T. R. Leavy, and C. S. Williams. 2005. Interactive effects of environmental variability and military training on stream biota of three headwater drainages in western Louisiana. Transaction of the American Fisheries Society 134: 192-206.

CHAPTER 3: LARVAL DEVELOPMENT OF CAENIS SP. (EPHEMEROPTERA: CAENIDAE) IN THE COASTAL TERRACES OF SOUTHWESTERN LOUISIANA

Introduction

Secondary production is a fundamental driver of aquatic ecosystem function and is an increasingly important tool for the study of benthic ecology (Benke and Huryn 2010). Mayfly larvae (Ephemeroptera) are important not only to biomonitoring and assessment of freshwater ecosystems, but also as major constituents of macroinvertebrate biomass and secondary production (Rodgers 1982, Cayrou and

Cereghino 2003). Many taxa exhibit great flexibility in life history, including regional and/or seasonal variability, so preconceived notions about cohort production intervals and production/biomass ratios rarely hold true (Gonzalez et al. 2001). Multivoltinism is common in Ephemeroptera, including members of the family Caenidae, especially in warm water streams (Clifford 1982, Brittain 1990, Taylor and Kennedy

2006). The non-seasonal multivoltine strategy, with four to six generations during the course of a year, is typical of mayfly populations from tropical and subtropical regions, where temperature does not limit hatching and larval development (Peran et al. 1999). In temperate zones, however, temperature is considered to be the major ecological factor structuring mayfly development (Vannote and Sweeney

1980), and growth in most taxa has been shown to be correlated with temperature (Cayrou and

Cereghino 2003). Either univoltinism, with overwintering nymphs, or bivoltinism, with winter and summer generations, is expected in these temperate regions (Corkum 1985).

Caenid mayflies are cosmopolitan, occurring in a wide variety of lotic and lentic habitats throughout the world, with the exception of oceanic islands (Corkum 1985, Provonsha 1990). Although considered tolerant of habitat impairment relative to other Ephemeroptera taxa (Peran et al. 1999), Caenis have demonstrated sensitivity to pollutants, such as pesticides (Demcheck et al. 2004). Life histories of many species of Caenis have been described and both larval size and voltinism appear to be influenced by latitude (i.e., temperature) and local conditions (Clifford 1982). Some species, such as C. hilaris, may exhibit more predictable emergence patterns than other taxa in this group (Provonsha 1990). Life cycle characteristics inferred entirely from emergence data may underestimate the number of annual generations (Lancaster and Downes 2013), underscoring the importance of examining larval development stages and seasonality of their frequency distributions (Peran et al. 1999, Cayrou and Cereghino 2003).

Targeted studies of Caenis life cycle and/or secondary production have been conducted regionally in streams of Oklahoma (Taylor and Kennedy 2006), Alabama (Rodgers 1982), and Georgia (Jacobi and

Benke 1991), but life history characteristics for Caenis populations of Louisiana have not been reported.

Descriptions of macroinvertebrate assemblages and impacts from agriculture have been reported for southwest Louisiana (Kaller and Kelso 2007, Mize et al. 2008). Louisiana has a humid subtropical climate, but has recently experienced historic drought conditions (Borrok and Broussard 2016). Due to their importance as biotic indicators and their ability to persist in intermittent pools (Taylor and Kennedy

2006), larvae of Caenis sp. collected throughout this drought were compiled to address the following objectives: 1) describe life history characteristics of Caenid mayflies in southwest Louisiana; 2) assess environmental cues that influence those characteristics; and 3) investigate water quality parameters with potential to affect growth rate or reproductive strategy. Because of apparent differences in soil, elevation, land cover, stream habitat and water quality, the geologic terrace of origin (Louisiana Geological Survey,

2008) was predicted to significantly influence these metrics.

Methods

Larvae of Caenis sp. (likely C. hilaris, Say 1839) were collected in 10 of 12 warm water streams sampled across Southwest Louisiana from August 2010 to August 2011, as part of a larger study of spatial and temporal variability within macroinvertebrate assemblages. Watersheds ranged in size from 13 to 280 km2 (Figure 3.1) and were located within three adjacent river basins (i.e., Calcasieu, Mermentau and

Vermilion-Teche) across three Pleistocene terraces, the Northern Humid Gulf Coastal Prairies of the

Western Gulf Coastal Plain and the Flatwoods and Southern Tertiary Uplands of the South Central Plains ecoregions (Daigle et al. 2006). The sampling period was characterized by historic drought condition in western Louisiana and eastern Texas (Borrok and Broussard 2016) and, as a result, stream flow was low or zero in many streams for extended periods. Land use in approximately half of the sampled watersheds was dominated by agriculture, with pasture and timber production predominating in the remainder.

Stream sites were visited on 8 occasions during the study period with a mean (± SD) number of days between sampling events at a site of 65 (± 10).

W E

Figure 3.1. Map of study area showing individual watersheds at 12 stream sites.

Macroinvertebrates were collected in sediments with a Hess sampler with 500 micron mesh and substrate surface area of 0.086 m2 (Wildlife Supply Co., Yulee, Florida). In addition to sediment, approximately 5

59 moderately decomposed pieces of submerged wood were collected, with all samples placed into pre- labeled gallon zipper bags, preserved with ethanol, double bagged, and placed on ice. One sample each of wood and sediment was collected at three locations within a 50-m reach in conjunction with measurements of wetted width, depth-velocity, substrate type, woody debris count, channel incision, overstory cover, bank height, and dominant bank vegetation collected along 5 transects placed equidistantly along the reach. Water temperature, specific conductance, pH, dissolved oxygen and turbidity were measured with a handheld multi-parameter water quality sonde (YSI 650MDS unit with

6820 V2-1 sonde, Yellow Springs, Ohio, USA). Before sediment was disturbed by wading, water was collected from the middle of the water column in a clean, dark bottle, capped underwater and packed in ice for transport to the laboratory. In the laboratory, water samples were used to measure carbonaceous biochemical oxygen demand, chlorophyll a concentration and coliform counts, as well as nitrate, nitrite, orthophosphate and N-ammonia concentrations following standard protocols APHA 5210B, EPA 445.0,

APHA 9222A, and Hach 8192, 8507, 8048 and 8155, respectively.

Preserved macroinvertebrate samples were stored at -20°C until they could be picked, re-picked and preserved in 95% ethanol prior to identification. Macroinvertebrate taxa were identified to lowest practical taxon, usually genus, and head capsule measurements of Caenis sp. were made with an ocular micrometer. In addition, Caenis larvae were classified by instar development class according to descriptions by Taylor (2001) for C. latipennis in Oklahoma. Caenids were chosen for further investigation because they were collected in large enough numbers to compare size distributions both spatially and seasonally. Adult male specimens are generally required to identify Caenis to species, and larval keys provided by Provonsha (1990) could not be applied to early instars, which constituted a large portion of the specimens in this study. Based on a majority of collections at the Louisiana State

Arthropod Museum, however, Caenis specimens taken from the study region more than likely belong to

C. hilaris.

To investigate landscape scale effects on development of Caenis sp. larvae, maximum width of head capsule within instar development classes was compared between the different geologic terraces with general linear mixed models (PROC GLIMMIX, SAS version 9.4, Cary, N.C.). Autocorrelation between

60 observations, resulting from the fact that sites were visited multiple times, was addressed with a first order autoregressive covariance structure, although other structures (i.e., compound symmetry, Toeplitz, and heterogenous first order autoregressive) were tested based on model fit (i.e., corrected Akaike information criterion; Gutzwiller and Riffell 2007). Because data were unbalanced (i.e., Caenis observations were not equally distributed among sites or sampling dates) and a covariance structure was specified, the degrees of freedom used to construct the F statistic were adjusted based on the Kenward-

Roger approach (Gbur et al. 2012). Assumptions of the general linear model were assessed by examination of model residuals and predicted values.

Seasonality of larval stages and timing of emergences were examined by comparing instar development class frequencies and size frequency distributions by sampling date. Patterns were assessed qualitatively, as further division of data by sampling date resulted in sparse tables and reduced statistical power (e.g., numerous sampling dates had less than 5 representative instars). Variability in habitat condition among streams was well-explained by landscape features, including geologic terrace, but temperature and nutrient levels have been suggested as major influences of growth in Caenid mayflies

(Cayrou and Cereghino 2003, Back et al. 2008). Therefore, the degree to which water quality parameters measured at the time of collection predicted larval size was investigated by pooling all collections and modeling head capsule width as the response in a multiple regression with these continuous variables and instar class (hereafter regressors). Continuous variables (other than temperature) were log10 transformed to stabilize variance. Regressors were chosen through stepwise model selection, followed by collinearity and variance inflation diagnostics, inspection of model residuals, and testing for autocorrelation via the Durbin-Watson statistic (PROC REG, SAS 9.4). Significance of selected regressors was then evaluated with the mixed model structure, defined previously, to account for unbalanced design, covariance structure and random effects.

Results

Head capsule width was measured on a total of 744 Caenis sp. specimens from 133 individual macroinvertebrate collections of wood and sediment in 10 streams. Greater than half of the specimens

(52%) came from one site, Middle Bayou Serpent, which had the largest upstream watershed and

61 maintained flow, albeit substantially less than normal, during the 2011 drought. Caenis collections were sparse in streams of the Tertiary Uplands terrace (i.e., only 24 individuals), so those specimens were pooled with collections from the Flatwoods for comparison with Prairie terrace streams. Tertiary Uplands and Flatwoods belong the same level III ecoregion (South Central Plains) and are more similar in terms of soils, land cover and elevation compared to the Prairie terrace (Daigle et al. 2006). The main effect for terrace in the general linear model of head capsule width within development class was not significant

(F=0.16, p=0.6930) (Figure 3.2). Pairwise comparisons within instar class were therefore not attempted.

1100

Prairie 1000 Up_Flatwoods

900

800

700

600

500 Mean Mean Head WidthCapsule (microns)

400

300 1 2 3 4 5 Instar Class

Figure 3.2. Model-adjusted mean head capsule width within instar development classes for Caenis sp. larvae collected in streams of Prairie and combined Flatwoods/Uplands terraces.

Larval size frequency distributions differed somewhat by sampling date in terms of central tendency, indicative of cohort development, but variability was fairly consistent throughout the year (coefficient of variation range 21.1 – 26.1%; Figure 3.3). Highest occurrence of small size classes occurred in

December of 2010 and in August of 2011, which were also the only months in which the largest

62 specimens (HCW>1200 microns) were not collected. Similarly, instar development classes II-V were collected in every sampling month, and very small individuals of class I were represented in all except two, albeit generally in small numbers (Figure 3.4). Distribution of earliest instar classes appeared bimodal with peaks in December of 2010 and August of 2011, whereas mature nymphs of class V peaked in June 2011.

Multiple regression selected instar class (+), water temperature (-), orthophosphate concentration (-), specific conductance (+), chlorophyll a concentration (-), turbidity (+), and carbonaceous biochemical oxygen demand (+) as significant predictors of head capsule width. The regression model was significant

(F=79.12, p<.0001) with an adjusted coefficient of determination of 0.5570 and normally distributed residuals. Variance inflation factors for all selected regressors were less than 2.1, and most variables did not exhibit collinearity, although chlorophyll a, turbidity and cBOD were slightly redundant. Positive autocorrelation was detected (D= 1.121, p<.0001), so significance of predictors was re-tested with a general linear mixed model structured for repeated measures. Type III tests of fixed effects were significant for log10-orthophosphate (parameter estimate= -210.25, SE=93.07, p=.025), log10-specific conductance (est.=706.46, SE=182.95, p=.0002), and cBOD (est.=144.74, SE=71.04, p=.0425), although log10-chlorophyll a was close (est.= -64.62, SE=33.00, p=.0509). Water temperature was not a significant effect in the final model.

Discussion

Despite apparent differences in habitat and water chemistry among geologic terraces, no effect of location in a particular geologic terrace on development of Caenis sp. larvae was detected in this study.

Given the differences in soil type, elevation, land cover, instream habitat features and water chemistry between the geologic terraces (Omernik 1987, Daigle et al. 2006), growth of a benthic consumer such as

Caenis was expected to differ as well (Merritt et al. 2008). Two primary confounding factors likely

40 Aug 2010 30

0 40 Oct 2010 30

0 40 Dec 2010 30

0 40 Feb 2011 30

0 40 Apr 2011 30

0 40 Jun 2011 30

0 40 Aug 2011 30

HCW class (microns)

Figure 3.3. Frequency distribution of Caenis sp. head capsule widths by sampling date.

Instar Development Class Frequencies by Date 80

50 1 2 40 3 4

30 5 Number of individuals of Number 20

0 2010-8 2010-10 2010-12 2011-2 2011-4 2011-6 2011-8 Sampling Year-Month

Figure 3.4. Seasonal frequency of larval instar development classes I-V during sampling period.

resulted in the lack of support for this hypothesis. First was the larger-than-landscape effect of an historic drought throughout much of 2011, and second was the complex interaction between Caenis ontogeny and environmental cues. Both of these factors complicated interpretation of the results, but also provided valuable insights about survival strategies in streams with chronic disturbance.

Historically low flows during the sampling period caused many streams to experience extreme fluctuations in water quality (Table 3.1). For example, mean (± SD) water temperature during the study year across samples was 21.3°C (±8.83) with a range of 4.95-34.19°C. The coefficient of variation for the 2010-2011 water temperatures was 40% compared to 28% during 2012-2013, which was a more typical water year.

Other measured habitat parameters also varied erratically in the absence of regular spates that are typical of the region (Felley 1992) and tend to stabilize water quality by flushing organic material and resetting trophic processes that structure the water column. As a result, biotic responses to environmental gradients may have been atypical and/or representative of extreme stress. Without the

65 stabilizing force of perennial flow within a stream, increased variability across the region caused both biotic and abiotic signatures to look more similar between watersheds than within them.

Table 3.1. Water quality variables included in the regression analysis with summary statistics. *Supersaturated condition caused by dense filamentous algal mats.

Variable Mean Std Dev Minimum Maximum CV Temperature °C 21.3 8.44 4.24 34.19 39.6% Specific Cond. mS/cm 0.291 0.177 0.022 0.968 60.8% DO mg/L 5.82 3.46 0.41 18.77* 59.5% Turbidity NTU 153.6 232.7 0.9 1715 151% Nitrite mg/L 0.0236 0.0309 0 0.2310 131% Nitrate mg/L 0.0241 0.0324 0 0.2410 134% Ortho-P mg/L 0.7033 0.5557 0 2.660 79.0% Chlorophyll a ug/L 6.379 6.937 0.015 55.115 109% cBOD20 mg/L 6.227 3.128 0.025 31.350 50.2%

Additionally, biotic responses to certain environmental gradients, such as nutrients, are often unimodal rather than linear. For example, Back et al. (2008) found that later developmental stages in Caenids have higher phosphorus requirements than earlier classes. Specifically, they showed P uptake increased for classes II through IV (somatic growth), and declined abruptly for class V (nearing eclosion, with a shift to reproductive development. In this study, orthophosphate was negatively related with larval size, but sites that produced the most Caenis individuals were not P-limited. To the contrary, all sites impacted by agricultural runoff maintained baseline concentrations of orthophosphate near 0.5 mg/L, even during dry months. Therefore, holding other factors constant, larvae developing in a nutrient rich environment would have faster growth rates and size at instar class would be expected to be smaller. Similarly, temperature was cited as the most influential factor structuring larval development by several authors, yet temperature was not selected as a significant predictor of larval size by the final model in this study. The extended emergence period afforded by the tropical to subtropical climate in southern Louisiana makes temperature limitation unlikely. In general, however, a faster growth rate and smaller terminal size are expected in temperate streams, as higher temperatures and disturbance regimes select for taxa with short development times (Taylor and Kennedy 2006).

Regression analysis also indicated cBOD and specific conductance were positively associated with larval size. Specific conductance was higher in the prairie terrace (F=14.53, p=0.0009), primarily due to percent cultivated crop land cover within watersheds, but values ranged between 0.1 and 1.0 mS/cm, within the normal range for fresh surface waters, and there was no indication of saltwater intrusion despite evidence of increased salinity in groundwater data during the 2011 drought (Borrok and

Broussard 2016). Reduced base flow and less overall water volume in general throughout the study area, however, may have concentrated dissolved minerals and organics at study sites, especially at sites where flow stopped completely for a protracted period. Both cBOD and specific conductance were negatively correlated with richness and abundance of EPT (Ephemeroptera, Plecoptera, and Trichoptera) taxa in study sites. However, at Middle Bayou Serpent (site “S165” in Figure 3.1), a higher order stream with a much larger watershed than the other sites, base flow was maintained, albeit reduced, throughout the study period. Over half of the Caenis specimens measured for this study were collected at Middle

Bayou Serpent, in which EPT abundance was consistently higher than other sites within the same river basin and geologic terrace. Environmental stressors that were augmented by the 2011 drought condition, including cBOD and specific conductance, possibly selected for more tolerant taxa, such as Caenis, which apparently thrived in Middle Bayou Serpent, perhaps due to maintenance of base flow and reduced competition (Miller and Golladay 1996, Johnson et al. 2013). Further, these stressors may select for asynchronous development, essentially staggered maturation of larvae, resulting in smaller individuals

(Taylor and Kennedy 2006). A simpler explanation, however, is that correlation does not necessarily imply causation in this highly disturbed ecosystem where so many stressors act in concert. Although

Middle Bayou Serpent constituted an outlier compared to the other headwater streams in our study, it provided important insight about the value of even minimal flow during dry periods for macroinvertebrates typically used as indicator taxa in bioassessments.

Another factor masking developmental responses to habitat gradients could be the complex life history characteristics of Caenis sp. within the study area, with multiple cohorts represented across size classes.

As previously mentioned, cohort overlap from asynchronous development is a strategy for mitigating mortality risk in temporally variable environments (Lancaster and Downes 2013). Distributions of both size classes and instar classes were complex and demonstrated a high degree of variability throughout

67 the study year. Early instar frequency appeared bimodal and may provide evidence that Caenis populations in Southwest Louisiana exhibit a bivoltine life cycle, which has been documented for many species of Caenis (Clifford 1982, Corkum 1985, Provonsha 1990, Taylor and Kennedy 2006). Larval development and size distributions examined over one sampling year demonstrated two possible emergence periods, late fall (probably November) and mid to late summer (probably July), but with no consequent absence of larval stages (save the rarer class I individuals). Persistence of multiple instar development classes throughout the year may be evidence of asynchronous development, but data pooled among sites were not robust enough to confirm based on the method outlined by Peran et al.

(1999). Another caveat regarding these data was the use of 500-µ mesh in the Hess sampler, which was larger than that employed in other studies that targeted early life stages (Taylor and Kennedy 2006).

Larger mesh may have resulted in loss of very small specimens, especially first and second developmental classes, but wood samples were taken whole and may have compensated for some of the loss.

Because these data were compiled from a larger study that did not target Caenid mayflies in particular, statistical design elements were unbalanced and conclusions should be considered tentative. The goal of the regression analysis on water quality measurements was only to find parameters that may be structuring growth of Caenis larvae in the study region, not to develop a predictive model. A study designed to address multivoltine strategies, using finer mesh sampling devices and targeting all development classes with more frequent collections throughout the year, such as those reported by

Peran et al. (1999) or Taylor and Kennedy (2006), would be better suited to answer more nuanced research questions. Although production was not calculated in this study due to inadequacy of the data, insights about possible emergence dates and factors influencing growth and abundance of Caenis sp. could help to refine studies of production and biomass in the study area. Additionally, given the drought condition in which these specimens were collected, results could inform further study of the effects of extended dry periods resulting from climate shift or disconnection of streams from underlying aquifers due to overuse of water for agriculture. Recently, water supply and loss of base flow within the coastal plains and alluvial valleys of the United States and abroad have begun demanding the attention of both hydrologists and natural resource managers (Borrok and Broussard 2016). Better information about

68 indicator organisms that interact with these ecosystems in complex ways, such as Caenid mayflies, could refine the tools required to address these and other issues in the near future.

References

Back, J. A., J. M. Taylor, R. S. King, K. L. Fallert and E. H. Hintzen. 2008. Ontogenic differences in mayfly stoichiometry influence growth rates in response to phosphorus enrichment. Fundamental and Applied Limnology 171(3): 233–240.

Benke, A. C. and A. D. Huryn. 2010. Benthic invertebrate production—facilitating answers to ecological riddles in freshwater ecosystems. Journal of the North American Benthological Society, 29(1): 264-285.

Brittain, J. E. 1990. Life history strategies in Ephemeroptera and Plecoptera, pp. 1-12. In I. C. Campbell [ed.], Mayflies and stoneflies: life histories and biology. Kluwer Academic Publishers, Dordrecht, The Netherlands.

Borrok, D. M. and W. P. Broussard III. 2016. Long-term geochemical evaluation of the coastal Chicot aquifer system, Louisiana, USA. Journal of Hydrology 533: 320–331.

Cayrou, J. and R. Cereghino. 2003. Life history, growth and secondary production of Caenis luctuosa and Cloeon simile (Ephemeroptera) in a small pond, S. W. France. Aquatic Insects 25(3): 191-201.

Clifford, H. 1982. Life cycles of mayflies (Ephemeroptera), with special reference to voltinism. Quaestiones Entomologicae 18: 1-4.

Corkum, L. D. 1985. Life cycle patterns of Caenis simulans McDunnough (Caenidae: Ephemeroptera) in an Alberta, Canada, Marsh. Aquatic Insects 7(2): 87-95.

Demcheck, D. K., R. Tollett, S. V. Mize, S. C. Skrobialowski, R. B. Fendick, Jr., C. M. Swarzenski, and S. D. Porter. 2004. Water quality in the Acadian-Pontchartrain Drainages, Louisiana and Mississippi 1999- 2001. Reston, Virginia. U. S. Geological Survey Circular 1232. 47 pp.

Gonzalez, J. M., A. Basaguren and J. Pozo. 2001. Life history and production of Caenis luctuosa (Burmeister) (Ephemeroptera, Caenidae) in two nearby reaches along a small stream. Hydrobiologia 452: 209–215.

Gutzwiller, K. J. and S. K. Riffell. 2007. Using statistical models to study temporal dynamics of animal- landscape relations. Chapter 6 in Temporal Dimensions of Landscape Ecology: Wildlife Responses to Variable Resources, J. A. Bissonette and I. Storch, eds. Springer, New York, NY. 284 pp.

Jacobi, D. I. and A. C. Benke. 1991. Life histories and abundance patterns of snagdwelling mayflies in a blackwater Coastal Plain river. Journal of the North American Benthological Society 10:3 72-387.

Johnson, R. C., H. Jin, M. M. Carreiro and J. D. Jack. 2013. Macroinvertebrate community structure, secondary production and trophic-level dynamics in urban streams affected by non-point-source pollution. Freshwater Freshwater Biology 58: 843–857.

Lancaster, J. and Downes, B.J. 2013. Aquatic Entomology. Oxford University Press. 304 pp.

Louisiana Geological Survey. 2008. Generalized geology of Louisiana. Louisiana State University. Baton Rouge, Louisiana. www.lgs.lsu.edu/deploy/uploads/gengeotext.pdf.

Merritt, R. W., K. W. Cummins and M. B. Berg. 2008. An introduction to the aquatic insects of North America, 4th ed. Kendall Hunt, Dubuque, Iowa. 1158 p.

Miller, A. M. and S. W. Golladay. 1996. Effects of spates and drying on macroinvertebrate assemblages of an intermittent and a perennial prairie stream. Journal of the North American Benthological Society 15: 670-689.

Mize, S. V., S. D. Porter and D. K. Demcheck. 2008. Influence of fipronil compounds and rice-cultivation land-use intensity on macroinvertebrate communities in streams of southwestern Louisiana, USA. Environmental Pollution 152: 491-503.

Omernik, J. M. 1987. Ecoregions of the conterminous United States. Annals of the Association of American Geographers. 77: 118-125.

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: 187-194.

Provonsha, A. V. 1990. A Revision of the Genus Caenis in North America (Ephemeroptera: Caenidae). Transactions of the American Entomological Society. 116(4): 801-884.

Rodgers, E. B. 1982. Production of Caenis (Ephemeroptera: Caenidae) in Elevated Water Temperatures Freshwater Invertebrate Biology 1(2): 2-16.

Sporka, F., Vlek, H.E., Bulankova, E. and Krno, I. (2006). Influence of seasonal variation on bioassessment of streams using macroinvertebrates. Hydrobiologia, 566: 543-555.

Taylor, J. M. 2001. Life History and Secondary Production of Caenis latipennis Banks (Ephemeroptera: Caenidae) in Honey Creek, Oklahoma. Masters of Science, Biology, University of North Texas. 89 pp.

Taylor, J. M. and J. H. Kennedy. 2006. Life History and Secondary Production of Caenis latipennis (Ephemeroptera: Caenidae) in Honey Creek, Oklahoma. Annals Entomological Society America 99(5): 821-830.

Vannote, R. L. and B. W. Sweeney. 1980. Geographic Analysis of Thermal Equilibria: A Conceptual Model for Evaluating the Effect of Natural and Modified Thermal Regimes on Aquatic Insect Communities. The American Naturalist 115(5): 667-695.

CHAPTER 4: DISCUSSION

Headwater streams represent not only the origin of downstream ecosystems, but also an important zone of interaction between our freshwater and the land. Four decades since the adoption of the Clean Water

Act, many of our small streams continue to be used as garbage dumps and drainage ditches. Progress in the development and refinement of bioassessment tools has enabled aquatic ecologists and limnologists to better define ecological integrity and set important benchmarks for impairment, but recovering a stream from decades of environmental degradation is often a slow and costly process, requiring long-term funding and participation of multiple stakeholders. Therefore, when weak evidence of ecological restoration potential is presented by experts, large-scale stream improvement projects tend to lose traction. I examined factors that influence habitat quality in some of the most impaired streams in southwestern Louisiana to address two issues that have historically weakened the argument for stream restoration: 1) quantifying spatial and temporal variability of habitat features and biotic responses, and 2) identifying functional in-stream habitat gradients. To this end, I compared aquatic habitat and benthic macroinvertebrate assemblages across adjacent river basins and geologic terraces in Louisiana’s coastal plain, described seasonal and drought effects, and examined environmental influences on the life history characteristics of a common mayfly. Results from these studies are used to recommend management actions to monitor and restore stream health in the heavily impaired Prairie terrace ecosystem of southwestern Louisiana.

Coastal plains and alluvial valleys contain gentle slopes and fertile soils, making them ideal locations for crop production. Consequently, the low-gradient watersheds found within them frequently contain a large agricultural footprint. In Chapter 1, I explored this linkage between land-use and warm-water stream habitat by examining the geologic influences in three adjacent Pleistocene terraces (i.e., Uplands,

Flatwoods and Prairie) within Louisiana’s Gulf of Mexico coastal plain during a drought year (2011) and a normal water year (2013). Habitat characteristics in the Uplands differed from the other two terraces with regards to both physicochemical parameters and streamflow. Multivariate ordination demonstrated that sites were organized primarily by terrace, driven by water chemistry and in-stream structure, and secondarily by water depth and flow velocity. These patterns were highly conserved between the two

71 study years, but variability among all sites was higher during the drought. Soil type and topography determined the type of resources (e.g., timber, rice, livestock) that the land would support and, therefore, the influences of geology and land-use on stream habitat were difficult to distinguish. In addition, thin riparian buffers (or lack of riparian vegetation, in some cases) precluded the introduction of woody debris into some streams. Over time, this can decrease the structural complexity necessary to support faunal diversity, especially in aquatic systems that lack other hard substrates such as bayous of the Prairie terrace. In Chapter 2, multivariate patterns for macroinvertebrate assemblages demonstrated strongest correlation with the subset of in-stream habitat variables containing specific conductance, turbidity, water depth, velocity, dominant substrate size, percent canopy cover, and biochemical oxygen demand. The influence of terrace on in-stream habitat was also stronger than that of river basin (i.e., Calcasieu,

Mermentau and Vermillion-Teche basins), indicating that local conditions in low-gradient streams determine aquatic habitat characteristics more than longitudinal factors, but this effect may have been inflated in these headwater stream sites, compared to higher-order sites downstream.

Groundwater discharge into a streambed, or baseflow, was also an important determinant of stream health in this region of relatively flat topography, where stream flows slow or cease during dry months.

This was demonstrated in Chapter 1, with more pronounced differences in stream habitat between terraces during the normal water year compared to the drought year. Seasonal differences between terraces were also stronger in 2013, indicating that regular flushing from precipitation helps to moderate variability in water chemistry. Without baseflow, low-gradient water bodies may shift from lotic streams to lentic pools throughout the year, disrupting environmental cues that support the reproductive strategies of many aquatic insect taxa. In Chapter 2, I demonstrated that macroinvertebrate assemblages in the

Uplands terrace could be distinguished from the Flatwoods and Prairie terraces primarily by the large diversity of insect taxa they supported. Aquatic insect taxa have specialized anatomy and generally require flowing water to respire. Other natural disturbance in flow regimes also affected ecological function at my study sites. Beavers capitalized on low water levels during the drought and the abundant woody debris in the Flatwoods terrace, pooling water at each of these study sites and creating an entirely lentic environment. Eventually, these environmental stressors selected against habitat-sensitive organisms, resulting in a generalist fauna dominated by tolerant taxa. Seasonal differences in

72 macroinvertebrate assemblages were consistent across terraces, and primarily reflected periodic emergences.

Investigation of environmental influences on the life history of a common mayfly, Caenis sp.

(Ephemeroptera: Caenidae; most likely C. hilaris) in Chapter 3 included comparisons of size (head capsule width) at larval instar development class and timing of emergence. Environmental factors of interest included geologic terrace, in-stream habitat gradients and regional drought effects. Specimens

(N=744) were compiled from 133 individual macroinvertebrate collections of wood and sediment in 10 streams. Greater than half of the specimens came from one site, Middle Bayou Serpent, which had the largest upstream watershed and maintained flow, albeit minimal, during the drought. No terrace effect was demonstrated, but water chemistry was associated with larval growth. Specific conductance and biochemical oxygen demand, indicative of degraded water quality, were positively correlated with larval size. Orthophosphate was negatively related with larval size, but sites that produced the most Caenis individuals were not P-limited. I concluded that habitat-tolerant larvae, such as Caenids, developing in a nutrient rich environment with large amounts of agricultural runoff, might exhibit faster growth rates and smaller size at instar class in the absence of resource competitors. Size frequency distributions of Caenis sp. in subtropical Louisiana indicated a bivoltine reproductive strategy with emergences in July and

November. As expected for this subtropical region, temperature was not a significant influence on growth or timing of emergence. Maintenance of baseflow during drought appeared to support increased abundance of Caenis larvae in streams with chronic disturbance from agriculture and reduced competition for resources.

In summary, geology is a landscape scale driver of both aquatic habitat gradients and macroinvertebrate assemblages. In an area with extensive habitat impairment, widening the lateral scope of the study landscape helped to identify habitat thresholds and describe regional habitat preference of individual macroinvertebrate taxa (especially insect taxa). Without the stabilizing force of perennial flow within a stream, increased variability across a region can cause both biotic and abiotic signatures to look more similar between watersheds than within them. The stream site with the largest watershed and consistent baseflow throughout the drought, Middle Bayou Serpent, provided important insight about the value of

73 even minimal flow during dry periods for macroinvertebrates typically used as indicator taxa in bioassessments. Management actions to improve aquatic habitat in bayous of the unique Prairie terrace ecosystem should include restoring baseflow, increasing structural complexity and protecting source populations in the Uplands. These actions might include monitoring more carefully the use of groundwater for irrigation to mitigate loss of potentiometric surface in aquifers, improving in-stream habitat condition to support colonization by native mussels, and maintaining or enhancing structural complexity and water quality in streams of the Uplands and Flatwoods terraces. Finally, Chapter 2 contains a list of macroinvertebrate taxa that demonstrated an association with either upland or lowland habitats during the study period. This list could help inform stream monitoring and restoration efforts within the coastal plain region of Louisiana.

APPENDIX A: OVERVIEW AND INDIVIDUAL SITE MAPS WITH LAND COVER AND STREAM NETWORK WITHIN UPSTREAM PORTION OF SAMPLING WATERSHED

W E

VITA

Catherine Elizabeth Murphy was born in July 1976 in Vicksburg, Mississippi. After graduating from

Vicksburg High School in 1994, she attended the University of Texas at Dallas and worked summers as a contract student in fisheries at the Corps of Engineers Waterways Experiment Station (now Engineer

Research and Development Center, ERDC) in Vicksburg. In 1998, Catherine earned a Bachelor of

Science in biology and worked as a laboratory technician in New Orleans, Louisiana and Stoneville,

Mississippi before returning to ERDC in 2001. She left work to pursue graduate education full time at

Louisiana State University in 2008, earning a Master of Applied Statistics in 2014 from the Department of

Experimental Statistics and completing her dissertation research within the School of Renewable Natural

Resources. Catherine currently works as a research ecologist at ERDC and is a candidate to receive the

Doctor of Philosophy in Wildlife and Fisheries Science in May 2017.