Responses of Aquatic Plant Communities to Stream and Riparian Restoration and Management

Home , Elodea nuttallii

EMILY PEFFER ZEFFERMAN B.S (Florida State University) 2007 DISSERTATION

Submitted in partial satisfaction of the requirements for the degree of DOCTOR OF PHILOSOPHY

Ecology

in the

OFFICE OF GRADUATE STUDIES

of the

UNIVERSITY OF CALIFORNIA DAVIS

Approved:

______Truman P. Young, Chair

______Eliška Rejmánková

______Peter B. Moyle

Committee in Charge

2014

UMI Number: 3685316

INFORMATION TO ALL USERS The quality of this reproduction is dependent upon the quality of the copy submitted.

In the unlikely event that the author did not send a complete manuscript and there are missing pages, these will be noted. Also, if material had to be removed, a note will indicate the deletion.

UMI 3685316 Published by ProQuest LLC (2015). Copyright in the Dissertation held by the Author. Microform Edition © ProQuest LLC. All rights reserved. This work is protected against unauthorized copying under Title 17, United States Code

ProQuest LLC. 789 East Eisenhower Parkway P.O. Box 1346 Ann Arbor, MI 48106 - 1346 Acknowledgements

I owe a debt of gratitude to a large number of people without whom this dissertation would not have been possible. The following words cannot express the full magnitude of all the ways in which I have been helped and humbled by my amazing colleagues, friends, and family throughout this journey. Nor can they include every individual who has helped me along the way. With these words, I merely hope to skim the surface of the deep pool of gratitude I feel for the many wonderful people who have been, and will hopefully continue to be, in my life.

First, I am grateful to my family, especially my parents Claire and Steve Peffer, for never failing to support and believe in my ambitions. I thank my advisor, Truman Young, for his eternal encouragement, constant willingness to provide ideas and advice, and his unabashed use of the word ‘audacious’ when describing my PhD work. I thank my other committee members,

Peter Moyle and Eliška Rejmánková, for their feedback and advice on my dissertation, and for being role models as scientists who are making a difference in conservation and restoration policy and practice.

Thank you to the students of the Young Lab (who over my years in Davis included Kurt

Vaughn, Lauren Porensky, Kevin Welch, Marit Wilkerson, Jen Balachowski, Kelly Gravuer,

Mila Dunbar-Irwin, Steve Fick, Laura Morales, Kristina Wolf, Derek Young, and Grace Charles) for their ideas, feedback, friendship, and logistical help. I would not have been able to complete my dissertation work were it not for their help, and that of many other kind souls who helped me in the field, most notably Elaine Chow and Karen Askeland. At multiple times, Neil Willits provided valuable statistical advice.

I am especially grateful to Andrew Fulks and J.P. Marie of the Putah Creek Riparian

Reserve for creating an artificial channel system for my shade experiments twice, and to the UC

Davis Center for Aquatic Biology and Aquaculture for providing the water for these

ii experiments. I also thank the staff at the Sierra Nevada Aquatic Research Laboratory and the

University of California Natural Reserve System. I thank Rich Marovich, Putah Creek

Streamkeeper, for suggesting that I submit a grant to the Solano County Water Agency for studying their aquatic vegetation management issues. And I thank SCWA, especially Alex

Rabidoux, for funding said study and for being open and enthusiastic about working with me.

I am greatly appreciative to the University entities that provided funding for my graduate education: the Graduate Group in Ecology (for several Block Grants), the Plant Sciences

Department (several scholarships and a multi-year research fellowship), and the Center for

Aquatic Biology and Aquaculture (scholarship). I also thank the benefactors of these funding sources and their awards committees.

I am grateful for the academic peer review process that has led to improvements to my first two dissertation chapters (and other publications), and that will lead to improvements in future manuscripts. I also thank the many wonderful professors and researchers at UC Davis who have taught fantastic classes, and/or opened their doors to answer my questions and provide advice on a variety of topics.

While completing my PhD, my psychological health was preserved by the many amazing friends I have made during my time in Davis, and I cannot thank them enough for all they have taught me about science, life, love, and costumery. I am continually humbled by their greatness, both personally and academically.

Last but not least, I thank my husband Matthew Zefferman for continually agreeing to help me in the field despite enduring many hours of exceedingly boring work, for editing and providing feedback on my papers, for providing emotional support when needed, and for co- creating an awesome last name.

iii

Responses of Aquatic Plant Communities to Stream and Riparian Restoration and Management

Abstract

Submersed macrophytes (plants that grow underwater) are important components of freshwater ecosystems, providing food and habitat for a variety of invertebrates, fish, and other wildlife, and influencing key chemical and physical processes. In moderate abundance, native submersed macrophytes contribute positively to overall stream health, but excessive proliferation of submersed macrophytes, particularly non-native invasive species, can cause ecological harm and create management difficulties. My dissertation examines how management actions can affect submersed macrophyte communities, and thus, the ecological integrity of stream ecosystems. This work incorporates and contributes to ecological theory from the disciplines of restoration ecology, invasion ecology, and plant community ecology, but also addresses sustainable management of streams and riparian areas.

My first chapter asks how riparian restoration, through the alteration of canopy shading, might influence the ability of a native and a non-native invasive submersed macrophyte species—elodea (Elodea nuttallii) and Eurasian watermilfoil (Myriophyllum spicatum), respectively— to establish and grow. I conducted separate experiments in artificial stream channels in two locations in California, USA, using shade cloth as a proxy for four different levels of overstory canopy shading. I found that increasing shade decreased the growth rates of both species but had no effect on establishment: both species had high survival in 0- 90% shade.

For my second chapter, I built on the work from chapter one and conducted a similar experiment in an artificial stream channel system in Davis, CA. I examined how shade levels could affect competitive dynamics between Eurasian watermilfoil and elodea. I also asked

1 whether giving the native elodea a temporal priority (i.e., planting it several weeks earlier than the non-native watermilfoil) could reduce the growth rates or survival of the non-native, and whether this priority effect would interact with shade level. Similar to my earlier findings, the growth rates of both species decreased with greater shade, but I found no significant effect of priority on the growth rates of the non-native species in any shade level.

For my third chapter, I examined the abiotic factors influencing the prolific submersed macrophyte growth in the Interdam Reach of Putah Creek in CA. In this stream reach, huge quantities of aquatic vegetation pose significant management challenges for the water resource agency in charge of regulating water flow and delivery in this area. I surveyed macrophyte cover and a suite of environmental factors throughout the Interdam Reach and, with a collaborator, created boosted regression tree models to identify the most important factors related to macrophyte cover. I found that factors associate with light availability and water velocity had the greatest influence on nuisance submersed macrophyte abundance in the models.

Chapter 1

Increasing canopy shading reduces growth but not establishment of Elodea nuttallii and Myriophyllum spicatum in stream channels

Summary

Submersed macrophytes are often important drivers of instream structure and function, but can be problematic when overabundant. The establishment success, growth rates, and morphology of submersed macrophytes could be affected by alteration of instream light levels during riparian restoration (via removing or planting canopy-forming vegetation), potentially influencing the success of riparian restoration projects aimed at improving aquatic habitats. To examine the effects of canopy shading on two common submersed macrophytes—Elodea nuttallii (native) and Myriophyllum spicatum (non-native)—I conducted experiments in artificial stream channels in two locations in California, USA. Initial establishment of stem fragments of both species was close to 100% in all shade levels, including shade that reduced incident light by

94%. Growth rates of the two species were similar across shade levels, and lowest in the highest shade. Full light appeared to have a photoinhibitory effect on E. nuttallii at the higher elevation site. Higher shade increased the length:biomass ratio and decreased the branching of E. nuttallii.

My findings suggest that altering canopy cover during riparian restoration is unlikely to affect the ability of these species to establish, but higher shade levels should slow their growth and create more favorable conditions for other instream organisms.

Introduction

Whether at the scale of large rivers or small creeks, the goals of riparian restoration generally include improving instream habitat for native flora and fauna (Palmer et al., 2007; Roni

3 et al., 2008). Follow-up monitoring of riparian restoration, though rare, has generally focused on fish, birds, mammals, invertebrates, or native riparian vegetation (Bash & Ryan, 2002; Roni et al., 2008; Feld et al., 2011; Gardali & Holmes, 2011). Submersed macrophytes often play critical roles in the structure and function of stream ecosystems (Rejmánková, 2011), but the effects of riparian restoration on these communities are rarely addressed.

The response of submersed macrophyte communities to riparian restoration, in terms of increased or decreased abundance, invasion of new species, or loss of species, can affect whether or not overall goals of improving instream habitat quality and restoring ecosystem function are achieved. Submersed macrophytes provide food, substrate, and habitat for aquatic and terrestrial organisms (Newman, 1991; Rejmánková, 2011) and drive important physical and chemical processes: they create heterogeneity in water velocity and substrate texture (Sand-Jensen &

Mebus, 1996), trap sediments and particulate matter, oxygenate the water column and rhizosphere, and absorb and transform nutrients (Carpenter & Lodge, 1986). Yet submersed macrophytes can also pose management challenges: excessive growth of weedy macrophytes can lead to reduced water velocity, increased water temperature, lower dissolved oxygen, and degraded fish spawning areas (Unmuth et al., 2000; Anderson, 2011). Invasion of non-natives can competitively reduce desirable native macrophytes (Madsen et al., 1991; Boylen et al., 1999) which may reduce fish and invertebrate diversity and abundance (Krull, 1970; Keast, 1984;

Wilson & Ricciardi, 2009; Schultz & Dibble, 2012). Therefore, practitioners should be concerned with both the positive and negative ways submersed macrophytes can influence the success of riparian restoration projects, and should consider how restoration could affect these important communities.

While restoration of riparian zones may affect submersed macrophyte communities in a number of ways (e.g., changes in water quality and flow rates), here I focus on one important mechanism: alteration of light availability. Light is an important limiting resource for submersed macrophytes (Lacoul & Freedman, 2006; Bornette & Puijalon, 2011), and riparian restoration can either decrease or increase this resource. Riparian restoration practitioners generally aim to revegetate stream banks with native plants, often with an explicit goal of creating more shade over streams (decreasing light) to moderate temperatures and improve instream habitat for fish and other organisms (Opperman & Merenlender, 2004). Conversely, non-native riparian plants are often removed in the initial stages of restoration, and removing canopy-forming species can cause an immediate increase in instream light levels.

In plant communities, increases in resources (such as light) are commonly linked with greater invasion potential (Davis et al., 2000), and competitive dominance of non-native over native species (Daehler, 2003). Decreases in canopy cover might therefore be expected to facilitate establishment and growth of non-native macrophytes at the detriment of native macrophytes, while increasing canopy shading may hinder invasion and benefit natives, but these hypotheses are rarely, if ever, tested in submersed macrophyte communities.

Most submersed macrophytes reproduce primarily asexually through fragmentation of stems and dispersal of stem fragments (Sculthorpe, 1967). Vegetative propagules are created through natural and anthropogenic processes (e.g., mechanical harvesting, boat propellers), and can range widely in size (Northwest Hydraulic Consultants, 2010). The ability for stem fragments of different sizes to initially establish under different levels of canopy cover is particularly important to understand, because non-native invasive macrophytes can be extremely difficult, if not impossible, to eradicate once established (Anderson, 2011). On the other hand,

5 fostering or maintaining the ability of native submersed macrophytes to establish in restoration sites through natural recruitment or active planting may be desirable.

To make predictions and guide management decisions on the effects of altered canopy cover on submersed macrophyte propagule establishment and growth, practitioners would ideally look to studies conducted in flowing conditions, but such studies are usually conducted in tanks with stagnant water (e.g., Sand-Jensen and Madsen (1991), Barrat-Segretain (2004), Mielecki and Pieczyńska (2005), and Angelstein and Schubert (2009)). Results of experiments in non- flowing conditions may not directly apply to macrophytes in flowing conditions, as water velocity can affect macrophyte physiology. Increasing water velocities can enhance photosynthesis and growth by increasing the rate of nutrient and gas exchange, yet when velocities are too high, increased shear stress can cause a decrease in photosynthetic rates and plant growth (Madsen & Sondergaard, 1983; Madsen et al., 1993). As such, controlled experiments concerning the establishment and growth of submersed macrophytes in streams should ideally be conducted in flowing conditions, yet this is rarely done (but see Dawson and

Kern-Hansen (1979)).

Morphological responses of macrophytes to shade levels (e.g., differences in stem branching or density) may also be important, because macrophyte growth forms influence instream habitat structure. Differences in macrophyte density and structural complexity have been shown to affect the growth rates, abundance, and behavior of fish and invertebrates

(Crowder & Cooper, 1982; Warfe & Barmuta, 2006).

To better understand the effects of riparian restoration on submersed macrophyte communities, I studied the establishment and growth of two macrophyte species under different shade levels in artificial stream channels, using Elodea nuttallii (Planch.) H. St. John (western

6 waterweed) and Myriophyllum spicatum L. (Eurasian watermilfoil) as study species. Both are widespread throughout North America—the former a native and the latter a notoriously invasive non-native—and both are perennials that reproduce almost exclusively through vegetative fragmentation (Sculthorpe, 1967). In the summer of 2011, I conducted one experiment with E. nuttallii only, and one with both E. nuttallii and M. spicatum, in artificial stream channels at a higher elevation site, the Sierra Nevada Aquatic Research Laboratory (SNARL), and a lower elevation site, the University of California, Davis (UC Davis), respectively. I conducted experiments in two different locations to gain a sense of the generality of results across systems with different abiotic conditions. Because submersed macrophyte vegetative propagules can vary in size, I tested whether two different sizes of fragments differed in ability to establish across shade levels.

For the experiment conducted with E. nuttallii alone (SNARL), I hypothesized that establishment success (rooting into soil), growth rates, and branching would decrease with increasing shade. For the experiment with both species (UC Davis), I hypothesized that while each species would show reductions in growth rates with greater shade, the non-native species would perform better than the native in the lower shade levels, but the native would do as well or better in the higher shade levels (i.e., an interaction between species and shade). For both experiments, I expected that smaller stem fragments would have lower establishment success than larger fragments in the higher shade treatments but not the lower shade treatments (i.e., an interaction between shade and initial fragment length).

Methods

Sierra Nevada Aquatic Research Laboratory Experiment

Site description

The Sierra Nevada Aquatic Research Laboratory (SNARL), a University of California

Natural Reserve operated by UC Santa Barbara, is located in Mono County, California in the eastern Sierra Nevada mountains (37º 36' 51" N, 118º 49' 47" W, elevation 2160 m). Water from

Convict Creek, an oligotrophic stream that drains Convict Lake, is diverted through a system of nine replicated artificial concrete-lined channels, three of which were used for this experiment.

Each channel is 1 m wide and 50 m long. The channels have concrete walls and a rock and gravel substrate. Sandy soil from the surrounding area was added to the centers of plots to provide a more uniform surface for plant establishment, and any existing aquatic plants were removed before the experiment began. Wooden weirs were placed in the middle and downstream ends of each channel to create a more uniform depth throughout the channel. Depth was on average 19 cm, and ranged from 10 to 30 cm.

Temperature, pH, dissolved oxygen, specific conductance, and turbidity were measured with a YSI multiprobe sonde (YSI Inc., Yellow Springs, Ohio, USA) at upstream and downstream points in each channel in the afternoon of August 29, 2011 (Table 1.1). Flow into each channel was controlled by sluice gates, and water velocity was kept uniform among the channels. Due to weather and stream flow variation, water velocity fluctuated throughout the duration of the experiment. During sampling on August 29, velocity averaged 0.1 m/s, but was probably as high as 0.2 m/s (estimated) near the start of the experiment. Water nutrients were not measured for this study, but historical data indicate that very low nitrogen and phosphorus levels are typical for Convict Creek (SNARL, personal communication).

Experimental Design

To test the effects of shade on the establishment and growth of two different fragment sizes of E. nuttallii, shade treatments were randomly assigned to plots, with fragment sizes nested within plot in a split-plot design.

Each of the three channels was divided into 20 1.5 m-long plots (Figure 1.1A). Into the center of each plot, four 5 cm (“short”) stem fragments and four 10 cm (“long”) stem fragments of E. nuttallii were placed 5 cm apart from each other on the substrate surface in a 4x2 arrangement alternating long and short fragments. The fragment lengths chosen are within the typical range of naturally dispersing propagule sizes observed for a similar Elodea species in a

California stream (Northwest Hydraulic Consultants, 2010). Source material for the fragments was collected on site, and any roots, branches, periphyton, or invertebrates were removed from the initial fragments before planting. Each fragment included an apical tip. Stem fragments were secured in the plots with plastic-coated metal ground staples. All fragments were planted on 9

July 2011. Additional 5 and 10 cm fragments (ten each) were dried overnight in an electric drying cabinet (Fisher Hamilton Scientific Inc., model # 218S632) at 60˚C to determine average initial weights for calculating dry weight biomass gained (final – initial weight, hereafter,

“biomass gained”), and relative growth rate (RGR).

Each of the 60 plots was randomly assigned one of four levels of shade using 30%, 60%, or 90% black polyethylene shade cloth or no shade cloth. Subsequent analysis of the actual amount of light reduction produced by the shade cloth, as measured in full sun with a LI-COR

LI-193 spherical quantum sensor (LI-COR, Lincoln Nebraska, USA), revealed that the 30%,

60%, and 90% shade cloth reduced incident photosynthetically active radiation (PAR) by an average of 40%, 72%, and 94%, respectively. Therefore, shade levels will henceforth be referred

to as “zero” (no shade cloth), “low”, “medium”, and “high”. Shade cloth was placed over plots

approximately 0.2 m above the water surface.

On 29 August 2011 at mid-day, PAR was measured with a LI-COR LI-193 spherical

quantum sensor below the water in the four shade treatments to assess the amount of light

reaching the planted macrophytes. Measurements were 1578, 990, 419, and 100 μmol photons m-

2 s-1 for the zero, low, medium, and high shade levels, respectively.

Data collection

During the experiment, high winds detached some of the 60% and 90% shade cloth,

reducing the number of replicates for the medium and high treatments from 15 to 14 and 6,

respectively. In the remaining plots, individual stem fragments were harvested from 30 August -

1 September 2011, after growing for an average of 53 days. This length of time was expected to

allow plants time to establish and experience the effects of shade without becoming limited by

other factors (e.g., substrate nutrients, space). Each plant was measured for number of branches

and total length (sum of lengths of all branches). Presence of roots was noted for each fragment

as an indication of successful establishment, but roots were removed along with any attached

algae and invertebrates to obtain above-substrate biomass. For each plot, the (initial) short

fragments were combined separately from the long fragments, and dried in an electric drying

cabinet at 60˚C for 48 hours. Using these plot-level total biomass values, average per-plant

values were calculated for biomass gained, RGR, and length:biomass ratio.

Statistical analysis

Mixed effects ANOVA models were used to evaluate the effects of shade level, initial fragment length, and their interaction on biomass gained, RGR, length:biomass ratio, and number of branches per plant. “Plot” was included as a random effect with initial fragment length nested within plot. Relative growth rate was calculated as ([ln(final weight)-ln(initial weight)]/days). For length:biomass ratio and RGR, weighted least squares approaches were used due to unequal variances among treatment groups. Number of branches was log transformed

(base 10) to comply with normality assumptions. Differences among shade treatments within each response variable were analyzed using least squares means with Tukey adjustments in p- values. All analyses were conducted in SAS software version 9.2 (SAS Institute Inc., Cary,

North Carolina, USA).

University of California, Davis Experiment

Site description

A series of connected stream channels was created at the University of California, Davis

(UC Davis) Putah Creek Riparian Reserve in the Central Valley of California (38˚ 31’ 42” N,

121˚ 47’ 7’’ W, elevation 21 m). The soil at this site is classified as Yolo Silt Loam, and the terrestrial plant community was dominated by non-native grasses and forbs. Seven channels were excavated, each ~ 24 m long, 1 m wide, and 0.6 m deep. The channels were connected into a single system so that water diverted from an aquaculture facility flowed through all of the channels, with water inputs at three of the sections (Figure 1.1B). Water velocity averaged 0.05 m/s throughout the experiment.

Temperature, pH, dissolved oxygen, specific conductance, and turbidity were measured

11 on 15 September 2011 with a YSI multi-probe sonde in the center of the three middle channels

(Table 1.1).

Experimental Design

Elodea nuttallii and M. spicatum source material was collected locally from Putah Creek

(38˚ 31’ 36” N, 121˚ 48’ 13” W), cleaned of attached algae and invertebrates, and cut into 4 cm

(“short”) and 8 cm (“long”) apical fragments. These fragment sizes are in the modal range for vegetative propagules of E. nuttallii and M. spicatum found locally in Putah Creek (Northwest

Hydraulic Consultants, 2010). Any roots or branches were removed from fragments. Ten additional fragments of 4 and 8 cm each were dried overnight in an electric drying cabinet at

60˚C to determine average initial weights for calculating biomass gained and RGR.

Each channel was divided into 1.5 m-long plots, and each plot was randomly assigned one of eight treatment combinations comprised of two species—E. nuttallii and M. spicatum— crossed with four shade levels—zero, low, medium, and high (see actual percent light reduction of shade cloth levels above). In each plot, four long and four short fragments of either M. spicatum or E. nuttallii were staked to the soil surface underwater using plastic-coated ground staples in a 2x4 arrangement alternating short and long fragments. Fragments were placed 10 cm apart and were planted on 25 and 26 July 2011. Shade cloth was placed over plots approximately

0.2 m above the water surface.

To assess the amount of light reaching the planted macrophytes, PAR levels were measured underwater near the sediment surface at mid-day with a LI-COR LI-193 spherical quantum sensor on 15 September 2011. Measurements were 1721, 1062, 419, and 168 μmol photons m-2 s-1 for the zero, low, medium, and high shade levels, respectively.

Data collection

At the time of harvesting (50-51 days after planting), many plants had become very large and intertwined and were fragmenting at the slightest touch, making it impossible to collect individual plant-level data. Therefore, analyses were based on plot-level biomass only. From 14-

16 September, total plot biomass was collected, dried in a 60˚C oven for 48 hours, and weighed.

Twelve replicates of each treatment were obtained.

Statistical analysis

Differences in plot-level biomass gained and RGR across species and shade levels were analyzed using two-way weighted least squares ANOVA. Biomass gained was log transformed

(base 10) to comply with normality assumptions. A shade*species interaction term was included in the models to determine whether the two species responded differently to the four shade levels. To analyze differences among shade levels within each species, ANOVA was followed by means comparisons using Tukey’s Honest Significant Difference method. One outlier, as identified by a Grubb’s outlier test (Grubbs, 1950), was excluded from all analyses. (This outlier was the first plot in the first channel, and may have grown larger due to higher water velocity in that location.) Analyses were completed using the “stats” package in R, version 2.15.2 (R

Development Core Team, Vienna, Austria).

Results

SNARL Experiment- effects of initial fragment length and shade on E. nuttallii

A total of 348 individual plants (87% of originally planted) in 50 plots were harvested and measured. The remaining 52 plants were missing entirely, and had apparently washed away.

No significant effects of shade level or original length on the number of remaining plants in each plot were detected (p = 0.19, p = 0.46, respectively), all remaining plants had rooted and grown, and I did not see any evidence of plant death within any plots at any point during the experiment; therefore, I concluded that survival of E. nuttallii was close to 100% in all shade levels for both initial lengths.

ANOVA statistics for the effects of initial fragment size and shade level on E. nuttallii growth metrics are shown in Table 1.2. Initial fragment length had significant effects on RGR, with short fragments having 28% higher growth rates than long fragments across shade levels.

However, initial fragment length had no significant effects on biomass gained, length:biomass ratio, or number of branches. The interaction between shade level and initial fragment length was not significant for any of the response variables.

Shade level, in contrast, had highly significant effects on biomass gained, RGR, length:biomass ratio, and number of branches, but the pattern of response to shade level differed between response variables. These differences were explored using Tukey means comparisons on the main effect of shade (Figure 1.2).

Biomass gained and RGR were highest in intermediate shade levels. (Figure 1.2, A and

B). The plants in low and medium shade gained on average 75% and 64% more biomass than the high shade level, and 33% and 25% more biomass than the zero shade level, respectively.

Length:biomass ratio (Figure 1.2C) increased significantly with higher shade: the medium shade level had a 24% larger ratio and the high shade level a 83% larger ratio than the zero and low treatments combined (which were not significantly different from each other).

Though not measured, I observed that the internode lengths of E. nuttallii were consistently longer in the higher shade levels. These findings suggest a plastic response of stem elongation in lower light environments.

The number of branches produced by E. nuttallii plants was similar in the zero, low, and medium shade levels, but significantly lower in the high shade level (Figure 1.2D). Plants in the high shade level produced 55% fewer branches compared to the three lower shade levels combined.

UC Davis Experiment- effects of shade on E. nuttallii and M. spicatum

No propagule mortality was observed in any of the plots for either species, though exact numbers were impossible to discern; individuals were intertwined and rooted in multiple locations in most zero, low, and medium shade plots, and attempts at separating individuals resulted in stem fragmentation. Based on observations made during harvesting, it appeared that close to 100% of fragments of both species and initial lengths established in all shade levels. All collected plants had rooted into the substrate.

ANOVA tables for the effects of shade and species on biomass gained and RGR are shown in Table 1.3. Both shade level and species had significant effects on plot-level biomass gained and RGR, but there was no significant interaction, indicating similar responses to shading between species. Results of Tukey means comparisons within species across shade levels are shown in Figure 1.3.

Elodea nuttallii gained 85% less biomass in the high shade level compared to the zero shade level. Despite an apparent linear trend in decreasing biomass with greater shade (Figure

1.3A), the zero, low, and medium treatments were not significantly different from each other.

Myriophyllum spicatum biomass gained was not significantly different in the zero and low shade levels, suggesting that light may be saturating at these levels. Biomass gained was significantly reduced in the higher shade levels, with 42% less biomass gained in the medium shade level and 87% less in the high shade level compared to the low and zero shade levels combined.

Biomass gained and RGR were significantly different between the two macrophyte species across shade levels, but conclusions on the relative performance of the species depend on which metric is used for evaluation (Figure 1.3). Myriophyllum spicatum gained more biomass than E. nuttallii in all four shade treatments. However, RGR was higher in E. nuttallii than M. spicatum in all treatments. This disparity can be accounted for by the fact that while the lengths of the fragments in the plots of both species were the same initially, M. spicatum fragments had

3.8 times more dry weight biomass per unit length, on average. Therefore, E. nuttallii had higher growth than M. spicatum relative to the initial fragment weights.

Discussion

The ability of both native and non-native submersed macrophytes to establish and grow in flowing conditions under different levels of shade has important implications for restoration projects that alter canopy cover over streams, especially if establishing native plants and preventing the establishment of non-natives is a goal. Because macrophytes often play important ecological roles in streams, practitioners may want to foster native macrophyte establishment and

(moderate) growth while reducing the likelihood of invasion or proliferation of non-native submersed macrophytes. These topics are also relevant to managers of canals, irrigation ditches, and other waterways where both native and non-native submersed macrophytes can be a nuisance.

My results suggest that the alteration of canopy cover is unlikely to affect the ability of E. nuttallii or M. spicatum to establish. Contrary to my hypothesis, both species were able to establish with high success in all shade levels, even under shade cloth that reduced PAR by 94%.

Despite this large reduction in incident light, PAR levels in the highest shade treatment were measured at 168 and 100 μmol photons m-2 s-1 underwater around peak daylight for UC Davis and SNARL, respectively. These light levels, which were possibly elevated because of light scattering underwater, are high enough to support growth of both macrophyte species:

Angelstein and Schubert (2009) found under experimental conditions that 8 cm stem fragments of E. nuttallii could grow in as low as 10 μmol photons m-2 s-1, and Van et al. (1976) found the light compensation point of M. spicatum to be 35 μmol photons m-2 s-1. Barrat-Segretain (2004) found that 5 cm stems of E. nuttallii had lower survival in 28 than in 48 μmol photons m-2 s-1, but these light levels are quite low. My results show that in flowing conditions with a natural photoperiod over the summer, light levels under 94% shade were still high enough to support establishment and growth of E. nuttallii and M. spicatum.

I hypothesized that shorter stem fragments would be less successful than longer fragments in establishing, particularly in the higher shade treatments, but found that both fragment sizes appeared to have close to 100% establishment success for both species. Riis et al.

(2009) showed that smaller fragments of Elodea canadensis (a closely related species to E. nuttallii, often morphologically indistinguishable) had lower establishment success than longer

17 fragments, but the fragment sizes used in their experiment were smaller than in mine—1 cm and

0.5 cm. They also found 100% establishment of M. spicatum fragments of 2-5 cm. The plants in their experiments were grown in non-flowing conditions and higher light levels (225 μmol photons m-2 s-1 in a 16/8 h light/dark cycle) than my highest shade treatment, and therefore may have been expected to have different outcomes. Taken together, the results of these studies suggest that only very small stem fragments in very low light may be unable to establish successfully, given adequate conditions for other growth factors. Shading from riparian vegetation alone may not reduce light to low enough levels to reduce establishment success in shallow, clear-water streams in the summer, but establishment may be reduced if canopy shade is combined with greater water depth, turbidity, and/or color.

Shade level had significant effects on all response variables in both locations. At

SNARL, E. nuttallii gained the most biomass and had the highest RGRs in the low and medium shade levels, and had significantly lower biomass in the zero and high shade levels. Finding that high shade reduced growth was not surprising, but the demonstration of a possible photoinhibitory effect of full light was unexpected. Elodea nuttallii has been called a “sun- adapted” plant (Jahnke et al., 1991), and often forms a canopy at the water’s surface (Barrat-

Segretain, 2004). Photoinhibition in submersed macrophytes is rarely studied or documented (but see Hussner et al. (2010)), possibly because most studies of light’s effect on macrophyte growth use artificial light. Neither species showed reduced growth in the zero shade treatment at UC

Davis, but plants in the UC Davis experiment probably experienced lower levels of irradiance compared to SNARL due to greater water depth, lower site elevation, and later experimental initiation. It is also important to note that results from both experiments suggest that light saturation may have been reached at the low (~40%) shade level, which means that alterations to

18 canopy cover that increase or reduce shade within the 0-40% range in shallow, clear water systems may not have a significant effect on growth of these species.

Plant morphology was affected by shade level as well. At SNARL, E. nuttallii stems in the highest shade level were elongated (greater length:DW biomass ratio) and had fewer branches, likely an adaptation for reaching toward light at the water’s surface. Plants in the three lower shade treatments had similar amounts of branching and generally had more prostrate growth forms with many adventitious roots. Though not directly measured, I observed the same phenomenon for E. nuttallii in the UC Davis experiment. Similar trends in morphological variation under different shade levels have been demonstrated in E. canadensis in experiments by Barko et al. (1982) and Sand-Jensen and Madsen (1991) in non-flowing conditions, suggesting that these morphological responses to shade are consistent in a variety of environmental conditions. Because E. nuttallii reproduces primarily through fragmentation of stems, reduced branching in high shade levels could also mean lower potential for populations to spread.

Shade-driven morphological differences could also have important implications for other instream organisms. For example, compared to plants with sparser, elongated stems, mat-like plants with many rooting branches may provide more effective hiding locations for invertebrates and small fish, but may exclude larger fish. Intermediate stem densities are probably best for fish populations (Crowder & Cooper, 1982), which my research suggests may occur at medium to high shade levels.

Contrary to the hypothesis that natives perform better than non-natives in lower resource conditions and vice-versa, the native E. nuttallii and the non-native M. spicatum had very similar performance in terms of establishment success and growth across shade levels: relative growth

19 rate was higher in all shade levels for E. nuttallii than M. spicatum, but the opposite was true for biomass gained. Classifying the two species as “native” and “non-native” is perhaps less ecologically meaningful than for other species pairs, because E. nuttallii is an aggressive invader and competitor in its non-native range, including regions where M. spicatum is native

(Angelstein & Schubert, 2009). Results of my experiment suggest that canopy cover may not strongly influence the dominance of one species over the other, but competition experiments with both species planted together in one plot under different shade levels would better explore these dynamics. Interestingly, variance within treatments was consistently greater for E. nuttallii, suggesting that E. nuttallii’s growth rate may be more influenced by other factors (e.g., soil nutrients, water velocity) that may have varied randomly in this experiment, while growth of M. spicatum may be more highly influenced by light level.

Conclusions

Increasing riparian shading has been proposed as a management tool to reduce problematic growth of macrophytes (Dawson & Kern-Hansen, 1979; Anderson, 2011), and the potential efficacy of this idea has been demonstrated by multiple studies. For example, field surveys by Canfield and Hoyer (1988), Madsen and Adams (1989), Julian et al. (2011), Köhler et al. (2010), Ali et al. (2011), and Wood et al. (2012) all found that higher levels of riparian shade were associated with lower submersed macrophyte abundance. Experiments have also shown reduced submersed macrophyte growth in higher shade, but these studies are almost always conducted in non-flowing conditions (e.g., Barko and Smart (1981)). My studies provide experimental evidence under stream-like conditions that increasing riparian shading could be effective in reducing growth rates and biomass of two common submersed macrophytes, but only

20 at relatively high shade levels, and even then it may not reduce their establishment rates.

Conversely, my results suggest that well-intentioned removal of invasive canopy-forming riparian vegetation could have an unintended consequence of increasing the density of submersed macrophytes.

It should be noted that because the effects of riparian shading on macrophyte growth can vary depending on season and plant growth phase (Wood et al., 2012), the timing of restoration in relation to macrophyte lifecycles may influence the outcomes of management actions. Further research into the important connections between riparian restoration and macrophyte communities (for example, similar studies on different species or in different seasons) could help restoration practitioners anticipate aquatic responses to riparian restoration, leading to more targeted and effective management actions.

Tables and Figures

Table 1.1 Water quality data for SNARL and UC Davis

SNARL UC Davis Sampling Date August 29, 2011 September 15, 2011

Water Quality Parameter Mean Range Mean Range Temperature (˚C) 17.75 (17.6-17.9) 19.44 (19.01-19.82) pH (SU) 8.21 (8.16-8.24) 8.11 (8.03-8.18) Specific Conductance (μS/cm) 112 (112-112) 716 (716-717) Turbidity (NTU) 7.5 (7.5-7.6) 8.0 (7.8-8.2) Dissolved Oxygen (% saturation) 81.2 (80.0-82.3) 111.5 (103.6-117.9)

Table 1.2 ANOVA Table for SNARL experiment

Factor F Value P Dry Weight Biomass Added Shade level 8.31 <0.001 Fragment size 1.43 0.235 Shade*Fragment size 0.32 0.809

Relative Growth Rate Shade level 9.06 <0.001 Fragment size 41.58 <0.001 Shade*Fragment size 0.33 0.802 Length:DW Biomass Ratio Shade level 21.18 <0.001 Fragment size 1.16 0.285 Shade*Fragment size 0.34 0.794 Number of Branches Shade level 7.62 <0.001 Fragment size 0.06 0.801 Shade*Fragment size 0.44 0.727

Boldfaced values are significant at the α= 0.0125 level. For all response variables, DF=3 for shade level and shade*fragment size, and DF=1 for fragment size.

Table 1.3 ANOVA table for UC Davis experiments

Factor F Value P Biomass Gained (Log transformed) Shade 76.78 <0.001 Species 52.72 <0.001 Shade*Species 0.59 0.622 Relative Growth Rate Shade 72.94 <0.001 Species 11.52 0.001 Shade*Species 0.40 0.753

Boldfaced values are significant at the α = 0.025 level. For both response variables, DF=3 for shade level and shade*species, and DF=1 for species.

Figure 1.1 Diagram of channel layouts for (A) SNARL and (B) UC Davis experiments (not to scale). Plots are shown as rectangles shaded according to assigned shade level- zero, low, medium, or high. In all plots, four short and four long stem fragments were planted. For the UC Davis layout, the species planted in each plot is indicated with a letter (M= M. spicatum; E= E. nuttallii). At SNARL, only E. nuttallii was planted. Water inputs are indicated with arrows. At SNARL, each treatment had 15 replicates initially, but high winds caused the loss of nine high shade plots and one medium shade plot. At UC Davis, the most downstream replicate of each of the eight treatments was not collected due to time constraints, so final N=12.

Figure 1.2 Response of E. nuttallii plants to shade level in SNARL experiments. Short and long fragments were averaged to show main effects of shade. Error bars show standard errors. Letters above columns indicate results of Tukey means comparisons (treatments with the same letter are not significantly different at the α=0.05 level). Number of branches is shown on a log axis.

Figure 1.3 Response of E. nuttallii and M. spicatum to shade level in UC Davis experiments. Results are based on plot-level biomass. Error bars show standard errors. Lowercase letters show results of Tukey means comparisons within species across shade level (treatments with the same letter are not significantly different at the α=0.05 level). Note the log scale for biomass gained.

Chapter 1 Literature Cited

Ali, M. M., S. A. Hassan, & A. S. M. Shaheen, 2011. Impact of riparian trees shade on aquatic plant abundance in conservation islands. Acta Botanica Croatica 70: 245–258.

Anderson, L., 2011. Freshwater Plants and Seaweeds In Simberloff, D., & M. Rejmanek (eds), Encyclopedia of Biological Invasions. University of California Press, Berkeley and Los Angeles: 248-258.

Angelstein, S., & H. Schubert, 2009. Light acclimatisation of Elodea nuttallii grown under ambient DIC conditions. Plant Ecology 202: 91–101.

Barko, J. W., D. G. Hardin, & M. S. Matthews, 1982. Growth and morphology of submersed freshwater macrophytes in relation to light and temperature. Canadian Journal of Botany 60: 877–887.

Barko, J. W., & R. M. Smart, 1981. Comparative influences of light and temperature on the growth and metabolism of selected submersed fresh-water macrophytes . Ecological Monographs 51: 219–235.

Barrat-Segretain, M.-H., 2004. Growth of Elodea canadensis and Elodea nuttallii in monocultures and mixture under different light and nutrient conditions. Archiv für Hydrobiologie 161: 133–144.

Bash, J. S., & C. M. Ryan, 2002. Stream restoration and enhancement projects: is anyone monitoring? Environmental management 29: 877–885.

Bornette, G., & S. Puijalon, 2011. Response of aquatic plants to abiotic factors: a review. Aquatic Sciences 73: 1–14.

Boylen, C. W., L. W. Eichler, & J. D. Madsen, 1999. Loss of native aquatic plant species in a community dominated by Eurasian watermilfoil. Hydrobiologia 415: 207–211.

Canfield, D. E., & M. V Hoyer, 1988. Influence of nutrient enrichment and light availability on the abundance of aquatic macrophytes in Florida streams. Canadian Journal of Fisheries and Aquatic Sciences 45: 1467–1472.

Carpenter, S. R., & D. M. Lodge, 1986. Effects of submersed macrophytes on ecosystem processes. Aquatic Botany 26: 341–370.

Crowder, L. B., & W. E. Cooper, 1982. Habitat structural complexity and the interaction between bluegills and their prey. Ecology 63: 1802–1813.

Daehler, C. C., 2003. Performance comparisons of co-occurring native and alien invasive plants: Implications for conservation and restoration In Futuyma, D. J. (ed), Annual Review of Ecology Evolution and Systematics. Volume 34. Annual Reviews: 183–211.

Davis, M. A., J. P. Grime, & K. Thompson, 2000. Fluctuating resources in plant communities: A general theory of invasibility. Journal of Ecology 88: 528–534.

Dawson, F. H., & U. Kern-Hansen, 1979. The effect of natural and artificial shade on the macrophytes of lowland streams and the use of shade as a management technique. Internationale Revue der gesamten Hydrobiologie 64: 437–455.

Feld, C. K., S. Birk, D. C. Bradley, D. Hering, J. Kail, A. Marzin, A. Melcher, D. Nemitz, M. L. Pedersen, F. Pletterbauer, D. Pont, P. F. M. Verdonschot, & N. Friberg, 2011. From natural to degraded rivers and back again: a test of restoration ecology theory and practice In Woodward, G. (ed), Advances in Ecological Research, Vol 44. Academic Press: 119–209.

Gardali, T., & A. L. Holmes, 2011. Maximizing benefits from riparian revegetation efforts: local- and landscape-level determinants of avian response. Environmental Management 48: 28–37.

Grubbs, F. E., 1950. Sample criteria for testing outlying observations. Annals of Mathematical Statistics 21: 27–58.

Hussner, A., H. P. Hoelken, & P. Jahns, 2010. Low light acclimated submerged freshwater plants show a pronounced sensitivity to increasing irradiances. Aquatic Botany 93: 17–24.

Jahnke, L. S., T. T. Eighmy, & W. R. F. Departments, 1991. Studies of Elodea nuttallii grown under photorespiratory conditions. I. Photosynthetic characteristics. Plant, Cell and Environment 14: 147–156.

Julian, J. P., S. Z. Seegert, S. M. Powers, E. H. Stanley, & M. W. Doyle, 2011. Light as a first- order control on ecosystem structure in a temperate stream. Ecohydrology 4: 422–432.

Keast, A., 1984. The introduced aquatic macrophyte, Myriophyllum spicatum, as habitat for fish and their invertebrate prey. Canadian Journal of Zoology 62: 1289–1303.

Köhler, J., J. Hachol, & S. Hilt, 2010. Regulation of submersed macrophyte biomass in a temperate lowland river: interactions between shading by bank vegetation, epiphyton and water turbidity. Aquatic Botany 92: 129-136.

Krull, J. N., 1970. Aquatic plant macroinvertebrate associations and waterfowl. Journal of Wildlife Management 34: 707–718.

Lacoul, P., & B. Freedman, 2006. Environmental influences on aquatic plants in freshwater ecosystems. Environmental Reviews 14: 89–136.

Madsen, J. D., & M. S. Adams, 1989. The distribution of submerged aquatic macrophyte biomass in a eutrophic stream, Badfish Creek: the effect of environment. Hydrobiologia 171: 111–119.

Madsen, J. D., J. W. Sutherland, J. A. Bloomfield, L. W. Eichler, & C. W. Boylen, 1991. The decline of native vegetation under dense Eurasian watermilfoil canopies. Journal of Aquatic Plant Management 29: 94–99.

Madsen, T. V., & M. Sondergaard, 1983. The effects of current velocity on the photosynthesis of Callitriche stagnalis Scop. Aquatic Botany 15: 187–193.

Madsen, T. V., H. O. Enevoldsen, & T. B. Jorgensen, 1993. Effects of water velocity on photosynthesis and dark respiration in submerged stream macrophytes. Plant, Cell and Environment 16: 317–322.

Mielecki, M., & E. Pieczyńska, 2005. The influence of fragmentation on the growth of Elodea canadensis Michx . in different light conditions. Polish Journal of Ecology 53: 155–164.

Newman, R. M., 1991. Herbivory and detritivory on freshwater macrophytes by invertebrates: a review. Journal of the North American Benthological Society 10: 89–114.

Northwest Hydraulic Consultants, 2010. Species identification and seasonal biomass flux monitoring in Putah South Canal, September 2008 through September 2009.

Opperman, J. J., & A. M. Merenlender, 2004. The effectiveness of riparian restoration for improving instream fish habitat in four hardwood-dominated California streams. North American Journal of Fisheries Management 24: 822–834.

Palmer, M., J. D. Allan, J. Meyer, & E. S. Bernhardt, 2007. River restoration in the twenty-first century: data and experiential knowledge to inform future efforts. Restoration Ecology 15: 472–481.

Rejmánková, E., 2011. The role of macrophytes in wetland ecosystems. Journal of Ecology and Field Biology 34: 333–345.

Riis, T., T. V Madsen, & R. S. H. Sennels, 2009. Regeneration, colonisation and growth rates of allofragments in four common stream plants. Aquatic Botany 90: 209–212.

Roni, P., K. Hanson, & T. Beechie, 2008. Global review of the physical and biological effectiveness of stream habitat rehabilitation techniques. North American Journal of Fisheries Management 28: 856–890.

Sand-Jensen, K., & T. V. Madsen, 1991. Minimum light requirements of submerged freshwater macrophytes in laboratory growth experiments. Journal of Ecology 79: 749–764.

Sand-Jensen, K., & J. R. Mebus, 1996. Fine-scale patterns of water velocity within macrophyte patches in streams. Oikos 76: 169–180.

Schultz, R., & E. Dibble, 2012. Effects of invasive macrophytes on freshwater fish and macroinvertebrate communities: the role of invasive plant traits. Hydrobiologia 684: 1–14.

Sculthorpe, C. D., 1967. The Biology of Aquatic Vascular Plants. Edward Arnold Publishers, London.

Unmuth, J. M. L., R. A. Lillie, D. S. Dreikosen, & D. W. Marshall, 2000. Influence of dense growth of Eurasian watermilfoil on lake water temperature and dissolved oxygen. Journal of Freshwater Ecology 15: 497–503.

Van, T. K., W. T. Haller, & G. Bowes, 1976. Comparison of the photosynthetic characteristics of three submersed aquatic plants. Plant Physiology 58: 761–768.

Warfe, D. M., & L. a Barmuta, 2006. Habitat structural complexity mediates food web dynamics in a freshwater macrophyte community. Oecologia 150: 141–154.

Wilson, S. J., & A. Ricciardi, 2009. Epiphytic macroinvertebrate communities on Eurasian watermilfoil (Myriophyllum spicatum) and native milfoils Myriophyllum sibericum and Myriophyllum alterniflorum in eastern North America. Canadian Journal of Fisheries and Aquatic Sciences 66: 18–30.

Wood, K., R. Stillman, R. Clarke, F. Daunt, & M. O’Hare, 2012. Understanding plant community responses to combinations of biotic and abiotic factors in different phases of the plant growth cycle. PLOS ONE 7: e49824.

Chapter 2

Experimental tests of priority effects and light availability on relative performance of Myriophyllum spicatum and Elodea nuttallii propagules in artificial stream channels

Summary

Submersed macrophytes have important ecological functions in many streams, but fostering growth of beneficial native species while suppressing weedy invasives may be challenging. Two approaches commonly used in management of terrestrial plant communities may be useful in this context: (1) altering resource availability and (2) establishing desirable species before weeds can invade (priority effects). However, these approaches are rarely used in aquatic systems, despite widespread need for sustainable solutions to aquatic weed problems. In artificial stream channels in California, USA, I conducted experiments with asexual propagules of non-native invasive Myriophyllum spicatum (Eurasian watermilfoil) and native Elodea nuttallii (western waterweed) to address the questions: (1) How does light availability affect relative performance of the two species?; (2) Does planting the native earlier than the invasive decrease survival or growth rate of the invasive?; and (3) Do light level and priority effects interact? The relative performance between E. nuttallii and M. spicatum had an interesting and unexpected pattern: M. spicatum had higher growth rates than E. nuttallii in the zero and medium shade levels, but had similar performance in the low and high shade levels. This pattern is most likely the result of E. nutallii’s sensitivity to both very low and very high light, and M. spicatum’s sensitivity to very low light only. Native priority did not significantly affect growth rate or survival of M. spicatum, possibly because of poor growth of the E. nuttallii planted early.

This study suggests that altering light levels could be effective in reducing growth of an invasive macrophyte, and for changing the competitive balance between a native and a non-native species

32 in the establishment phase. Further investigations into the use of priority effects and resource alteration for submersed macrophyte management are warranted, given their mixed results in other (limited) studies.

Introduction The role of resource availability in determining relative performance of plant species is often examined in the context of competition between native and non-native invasive species

(hereafter, ‘invasives’). Higher resource levels are commonly thought to favor invasive over native species, as plant traits that confer invasiveness (e.g., high growth rates and fecundity) are often associated with higher resource use (Dukes & Mooney, 1999; Alpert et al., 2000). Many studies do show a competitive advantage of invasive species over natives in high resource conditions and vice-versa (Daehler, 2003). However, invasives in certain ecosystems can be more efficient at using scarce resources than natives, thus outperforming natives in lower resource conditions (Reinhart et al., 2006; Funk & Vitousek, 2007; Funk, 2013). Research into the competitive outcomes between natives and invasives across resource levels has focused largely on terrestrial plant communities, with relatively little attention to aquatic plant

(submersed macrophyte) communities, particularly in flowing systems.

Light is an important limiting resource in freshwater systems (Lacoul & Freedman, 2006) and can be manipulated through management of canopy-forming riparian vegetation. Reducing light by increasing canopy shading has been recommended for controlling growth of submersed macrophytes generally (Dawson & Kern-Hansen, 1979; Canfield & Hoyer, 1988; Anderson,

2011). However, few, if any, studies specifically address how reductions in light may affect relative performance of invasive over native macrophytes when grown in a competitive environment.

Native submersed macrophytes play an important and beneficial role in many streams, providing food and habitat for aquatic organisms and modifying the physical and chemical environment (Carpenter & Lodge, 1986; Rejmánková, 2011). Invasive macrophytes, however, can have large negative economic and ecological impacts in natural and human-made waterways

(Pimentel et al., 2005; Anderson, 2011), and have been linked with reduced abundance and/or diversity of native macrophytes (Aiken et al., 1979; Mjelde et al., 2012), invertebrates (Keast,

1984; Cheruvelil et al., 2001) and fish (Keast, 1984; Schultz & Dibble, 2012). Controlling invasive macrophytes through traditional physical or chemical means can pose logistical difficulties in flowing systems due to the potential for dispersal of unintentionally-created vegetative propagules and reduced contact time with herbicides (Anderson, 2011). In addition, these methods often cause collateral damage to other aquatic organisms and generally have only short-term impacts (Anderson, 2011). Therefore, understanding how manipulating resource availability affects the competitive balance between native and invasive macrophytes should be useful to managers of flowing waters who are interested in preserving or restoring native diversity while reducing the impact of invasives.

In addition to resource availability, priority effects—the impacts of species arrival and establishment order on community structure—are known to affect the relative performance of species. The idea that early-arriving species can preempt resources, reducing the success of later arrivers, is a basic tenet of community assembly theory (Young et al., 2001). Priority effects have been studied extensively in terrestrial plant communities, often in the context of restoration and with a goal of giving native species a competitive edge over invasive species (Vaughn &

Young, in press). Yet, the role of priority in determining macrophyte community composition and the potential for using priority to achieve management outcomes have only just begun to be

34 addressed in freshwater systems (Larned et al., 2006; Chadwell & Engelhardt, 2008).

Priority effects may be particularly relevant after stream restoration, as these projects may involve large disturbances (e.g., temporary drying of stream channels or creation of new channels) that ‘reset’ macrophyte communities by killing off resident species or creating new substrate. While active planting of riparian vegetation often takes place after stream restoration, submersed macrophytes are typically expected to recolonize on their own (Larned et al., 2006).

If actively planting native macrophytes early could effectively reduce the impact of invasive macrophytes, it would be a useful management tool.

Perhaps as interesting as the individual effects of either altering resource availability or giving native macrophytes a temporal priority advantage is how these two potential management techniques could interact. For example, priority effects may be stronger in higher light environments if early-arriving species are able to preempt other resources (e.g., soil nutrients) more quickly. Interactions between priority effects and resource manipulation are rarely studied

(but see Moore & Franklin 2012; Kardol, Souza & Classen 2013) and to my knowledge, have never been studied in aquatic systems.

To investigate the effects of light level (resource availability), temporal advantage

(priority effects), and their potential interaction on the relative performance of native and non- native submersed macrophytes during initial establishment, I conducted a mesocosm experiment in central California, USA, using native Elodea nuttallii (western waterweed), and invasive

Myriophyllum spicatum (Eurasian watermilfoil) as study species. In North America, M. spicatum is considered one of the most pernicious non-native invaders of fresh waters (Smith & Barko,

1990). Elodea nuttallii is a common native in California, and co-occurs with M. spicatum in local waterways (Peffer, 2013). Both are perennials that reproduce primarily through dispersal of

35 vegetative fragments (Sculthorpe, 1967). Because M. spicatum has been associated with altered fish and invertebrate assemblages (Keast, 1984; Weaver et al., 1997; Wilson & Ricciardi, 2009) and declines in native macrophyte richness (Madsen et al., 1991; Boylen et al., 1999), the competitive outcomes between these species could have important effects on aquatic communities. To date, most experiments on submersed macrophytes, including these two species, have been conducted in laboratory conditions with little water flow and under artificial light. Because I wanted my experimental results to be applicable to flowing systems, I created artificial stream channels for these experiments under natural light conditions.

I predicted that (1) decreasing light availability (by increasing shading) would improve performance of the native E. nuttallii relative to the invasive M. spicatum; (2) planting E. nuttallii earlier than M. spicatum (priority advantage) would reduce survival and growth rates of

M. spicatum; (3) giving E. nuttallii an establishment priority would have a greater suppressive effect on M. spicatum in high light treatments than in the high shade treatments (a priority x light availability interaction).

Methods

Site description A system of connected artificial stream channels was created for this experiment at the

University of California, Davis (UC Davis) Putah Creek Riparian Reserve in the Central Valley of California (38.52833˚N, 121.78528˚W, elevation 21 m). All of the methods described herein took place within, and were approved by, the Putah Creek Riparian Reserve. Three channels were excavated in a field dominated by grasses and forbs with soil classified as Yolo Silt Loam.

Each channel was approximately 60 m long, 1 m wide, 0.6 m deep (water depth range = 0.32-

0.48 m). Water from an aquaculture facility was diverted to flow through all three channels, with

36 water inputs at the upstream end of each channel (Figure 2.1). The channels were flooded for 17 days prior to initial planting. Water velocity averaged 0.05 m/s throughout the experiment and was slightly higher near the three water inputs and slightly lower at the downstream ends of the channels.

To characterize the nutrient levels in the system, I collected soil samples from five different locations within the channels on 7 September 2012, and took water samples from upstream, middle, and downstream locations in the channel system on 18 September 2012 and again on 23 September 2012. Samples were analyzed by the UC Davis Analytical Laboratory

(http://anlab.ucdavis.edu) for ammonium and nitrate (SOP 312 for soil, SOP 847 for water), and phosphorus (SOP 340 for soil, SOP 865 for water). Sediment and water nutrient levels were similar throughout the channel system (Table 2.1). Typical values for specific conductance, hardness, and pH in the source water for this system are, respectively, 750 μS/cm, 350 mg/L

(CaCO3), and 8.1 SU (personal communication, Paul Lutes, UC Davis Center for Aquatic

Biology and Aquaculture).

I deployed HOBO U22 Water Temp Pro v2 temperature loggers (Onset Computer

Corporation, Cape Cod, MA) in the middle of each channel from 18-29 September 2012, and recorded temperatures every two hours. Temperatures were similar among channels (within 0.25

˚C difference on average), and ranged from daily lows around 18.0 ˚C to highs around 20.6 ˚C.

Experimental Design Using a completely randomized design, I assigned one of eight treatment combinations— two competition treatments crossed with four shade treatments—to 1.5 m-long plots in the channel system. In both competition treatments, three 5 cm stem fragments of E. nuttallii and three 5 cm stem fragments of M. spicatum were planted into each plot by staking each fragment

37 to the sediment surface, but in the ‘priority’ competition treatment E. nuttallii was planted 35 days before M. spicatum, while in the ‘concurrent’ competition treatment both species were planted at the same time. In each plot, stem fragments were placed 5 cm apart alternating species in a 2x3 arrangement. The timing, number, and spacing of plants was informed by observed growth of the same two species in a study conducted the previous year (Zefferman, 2014).

I collected macrophytes locally from Putah Creek (38.52667˚N, 121.80361˚W) and removed all attached algae, invertebrates, and/or roots before cutting stems into 5 cm fragments.

All fragments were apical. Five cm is in the modal range for naturally dispersing vegetative propagules of E. nuttallii and M. spicatum found locally in Putah Creek (Northwest Hydraulic

Consultants, 2010). Ten additional 5 cm fragments of each species were dried at 60˚C to determine average initial weights for calculating relative growth rates.

Shade treatments were implemented using black polyethylene shade cloth of three different weights—30% (‘low’), 60% (‘medium’), or 90% (‘high’)—or no shade cloth (‘zero’).

Previous work (Zefferman, 2014) determined that the actual amount of photosynthetically active radiation (PAR) reduced by the 30%, 60%, and 90% shade cloth was on average 40%, 72%, and

94%, respectively. Shade cloth was secured over the plots approximately 0.2 m above the surface of the water after planting. To determine the amount of light reaching the planted macrophytes under each shade treatment, I measured PAR below the water on 14 and 18 August

2014 between 12:00pm and 12:30pm with a LI-COR LI-193 spherical quantum sensor (LI-COR,

Lincoln Nebraska, USA).

Elodea nuttallii was planted into the priority plots on 13 and 14 July 2012. On 17 and 18

August 2013, M. spicatum was planted into the priority plots, and all of the concurrent plots were planted with both species. On 29 and 30 September 2012 (43 days after planting M. spicatum in

38 both treatments), biomass was harvested at the plot level. All above-soil biomass was collected in each plot; the species were separated, placed into paper bags, dried for 48 hrs at 60˚C in a drying oven, and weighed.

Statistical analysis Although initial fragment lengths were the same for both species in all treatments, the E. nuttallii fragments were lower in biomass. To account for this initial difference between species,

I used relative growth rate (RGR) as the response variable in all of the analyses, calculated as

[ln(final dry weight) - ln(initial dry weight)]/#days. In cases where final biomass was zero or smaller than initial biomass, I assigned a RGR of ‘0’. All analyses were completed using the

‘stats’ package in R (R Development Core Team, Vienna, Austria).

To determine whether shade changed the relative performance of E. nuttallii and M. spicatum when grown concurrently, I first used analysis of variance (ANOVA), including shade, species, and their interaction as predictors. I used a weighted least squares approach to accommodate unequal variances between the two species. ANOVA was followed by Tukey means comparisons of RGR across shade levels within each species, and pairwise t-tests comparing E. nuttallii to M. spicatum RGR in each shade level. I also tested the correlation between M. spicatum and E. nuttallii RGR within each shade level, using Pearson’s product moment correlation coefficient.

To determine whether M. spicatum growth rates were affected by the timing of E. nuttallii planting, and whether the magnitude of these effects was influenced by shade level, I used ANOVA with shade, competition treatment, and their interaction as predictors and M. spicatum RGR as the response variable. To comply with ANOVA assumptions of normality of residuals, I Winsorized the dataset by adjusting the RGR values with the two lowest and two

39 highest residuals to the 2nd and 98th percentiles, respectively (Tukey, 1962). (Winsorizing did not ultimately change the conclusions of the analysis.) ANOVA was followed with means comparisons by Tukey’s Honest Significant Different method.

Results Mean PAR levels measured underwater below the zero, low, medium, and high shade treatments at mid-day in mid-August were 1922, 1218, 643, and 233 μmol photons m-2 s-1, respectively. Final biomass and RGR means and standard errors, and number of replications per treatment are shown in Table 2.

Effect of shade level on competition between M. spicatum and E. nuttallii

Plot level survival was 100% for both species in the concurrent treatment. Overall, M. spicatum had higher RGR than E. nuttallii (significant species effect). Shade had a significant effect on RGR, but these differences across shade levels did not differ significantly between the two species (no significant species x shade interaction) (Table 2.3A, Figure 2.2). The mean RGR of E. nuttallii in the concurrent treatment was highest in the low shade level, intermediate in the zero and medium shade levels, and lowest in the high shade level—41.1% lower than the low shade level (Figure 2.2). However, Tukey means comparisons for E. nuttallii in the concurrent treatment only show a significant difference in RGR in the high shade level compared to the zero

(p = 0.022) and low shade levels (p = 0.0004).

Pairwise comparisons of the two species in each shade level found that M. spicatum RGR was 19% higher than E. nuttallii RGR in the zero shade treatment (p = 0.029) and 20% higher in the medium shade treatments (p = 0.038). In the low and high shade levels, performance between species was similar (mean difference of M. spicatum – E. nuttallii RGR = 0.003 g/g/d and -0.004

40 g/g/d, respectively) and was not significantly different (p = 0.58 and p = 0.61, respectively) (Fig.

2).

Myriophyllum spicatum and E. nuttallii RGR were positively correlated in all shade levels within the concurrent treatment, although the correlations were only significant in the medium and low shade levels (Figure 2.3).

Effect of native priority and shade on M. spicatum

Plot level survival for M. spicatum was 100% in both competition treatments.

Competition treatment did not significantly affect M. spicatum growth, and RGRs were similar between treatments in each shade level. Shade level did have a significant effect on M. spicatum

RGR, but there was no significant interaction between shade and competition treatments (Table

2.3B, Figure 2.4). Therefore, Tukey means comparisons were performed on the main effect of shade. Mean RGR of M. spicatum was highest in the zero and low shade treatments, which were not significantly different from each other (p = 0.47). Mean RGR in the medium shade treatment was significantly lower than in the zero (p <0.0001), and low (p = 0.009) shade treatments, and the high shade treatment was significantly lower than the other three (p < 0.0001). Compared to the average of the zero and low shade treatments, mean RGR was 11% lower in medium shade and 49% lower in high shade.

Discussion

Light level affects relative performance of M. spicatum and E. nuttallii

Competitive suppression of native submersed macrophytes by non-native invasives such as M. spicatum is widely regarded as problematic (Anderson, 2011), but there is little information on how factors such as resource availability or priority effects enhance or reduce

41 these competitive effects. While some experiments have tested macrophyte competition in different nutrient levels (e.g., Van et al., 1999; Mony et al., 2007; Angelstein et al., 2009), experiments on the effects of light availability on submersed macrophyte growth are generally conducted on plants grown individually or with conspecifics (e.g., Barko et al., 1982; Sand-

Jensen & Madsen, 1991; Angelstein & Schubert, 2009; Zefferman, 2014). My experiments tested the effect of light availability on relative performance of a native and an invasive submersed macrophyte when planted in a competitive environment, which is arguably more relevant to natural systems.

I expected performance of M. spicatum to decrease relative to that of E. nuttallii with increasing shade, consistent with the commonly-observed pattern that invasive species have a performance advantage in high resource conditions while natives have an advantage in low resource conditions (Daehler, 2003). However, relative performance of the two species varied with shade level in a different and unexpected way: the two species had similar growth rates in the low and high shade levels, but M. spicatum had significantly higher growth rates than E. nuttallii in the zero and medium shade levels (Figure 2.2).

This pattern could be attributed to greater competitive abilities of M. spicatum compared to E. nuttallii at the zero and medium shade levels and similar competitive abilities in the low and high shade levels. Alternatively (or in part), it could be due to differences in the shape of the growth response to shade for each species independently. If competition was limiting growth, negative correlations between the two species’ growth rates within each shade level might be expected. However, the positive correlations between E. nuttallii and M. spicatum growth rates

(Figure 2.3), suggests both species were responding similarly to conditions in individual plots that may have varied spatially (e.g., moderate differences in soil nutrients in different plots), and

42 that competition was not limiting growth. This lack of competitive suppression may have been due to the abilities of both species to expand laterally within a plot, and implies that competition may not be important between these two species in the early establishment phase, given adequate soil nutrients.

It is likely that the different patterns in growth rate between the two species across shade levels were caused by individual species-specific growth responses to light. Myriophyllum spicatum had highest growth rates in the zero and low shade levels, moderately lower growth rates in the medium shade level, and substantially lower growth rates in the high shade level.

This pattern is similar to that found by Zefferman (2014) when M. spicatum was planted under the same four shade levels with conspecifics. The similar growth rates in the zero and low shade levels are probably due to light saturation: Su et al. (2004) identified the saturation point of M. spicatum as 1000 μmol photons m-2 s-1, which is lower than the light levels I measured below water in the zero and low shade treatments at mid-day. Elodea nuttallii also had the lowest growth rate in the highest shade level, but in contrast to M. spicatum, had decreased growth in full light. Although this decrease in the zero shade treatment was not statistically significant, it suggests a photoinhibitory response caused by clear, shallow water and intense mid-summer radiation. Though photoinhibition is rarely documented in freshwater macrophytes, photoinhibitory effects have been found in E. nuttallii by Hussner et al. (2010) and Zefferman

(2014). The demonstrated potential for photoinhibition in macrophytes illustrates the importance of conducting experiments in natural light conditions. Even in a glasshouse, UV from the sun may be diminished to an extent that affects macrophyte performance.

Elodea nuttallii showed much greater within-treatment variability in growth rates than M. spicatum (Figure 2.2). This was also found by Zefferman (2014) and suggests a difference in

43 sensitivity to factors such as water velocity or spatial variation in soil nutrients. Observationally, it appeared that E. nuttallii grew larger in plots with moderately higher flow, as occurred near the water inputs.

High resources are generally thought to favor invasive over native species (reviewed in

Daehler (Daehler, 2003)), as invasives often have traits that are associated with high resource acquisition. When the opposite is found, it may be attributed to superior resource conservation traits in the invasive species (Funk & Vitousek, 2007). My results show that relative performance between natives and invasives across resource levels can have a complex pattern when individual species response curves differ. In this case, E. nuttallii appears to be sensitive to light at both high and low levels during establishment, while M. spicatum is sensitive only at low levels. The lack of a linear relationship between light availability and relative performance of these two macrophytes has potentially complicated implications for managers who may wish to use canopy shading to control relative abundance of these species or species with similar growth response curves.

It should be noted that in contrast to many invasive-native species pairs, this particular pair shows an interesting quality: each species is invasive in the other’s home range. In fact, many aquatic plants are invasive outside their home ranges (Sculthorpe, 1967). In this way, using

‘native’ and ‘invasive’ as guild designations for purposes of comparing plant traits may be less relevant in aquatic systems than in terrestrial systems.

No effect of native priority on M. spicatum

Priority planting of desirable (often native) species is another method commonly used to reduce invasive weed growth in terrestrial systems (reviewed by Vaughn and Young (in press)), but remains under-explored in freshwater systems. In addition, interactions between resource

44 level and priority effects in any plant community are even more rarely addressed. Priority effects may be particularly important in streams after restoration activities like channel modification, during which existing submersed macrophyte communities may be wiped out. I expected growth rate and survival of M. spicatum to be reduced where E. nuttallii was planted five weeks before

M. spicatum, and to show the greatest difference in growth rates in higher light treatments.

However, planting E. nuttallii earlier appeared to have no measurable effect on growth rates of

M. spicatum (Figure 2.4) and had no effect on plot level survival, which was 100% in both competition treatments. This result was likely due to the failure of E. nuttallii to grow as quickly as needed to preempt resources during the initial phase of the experiment, when E. nuttallii was planted alone in the priority treatment plots. The final biomass of E. nuttallii was similar for both competition treatments, despite having a five week head start in the priority treatment (Table

2.2).

The poor performance of E. nuttallii planted earlier appeared to be caused by insect herbivory. After initial flooding of the channels, I observed a flush of aquatic insect colonization.

The number of invertebrate larvae in the channels decreased over time, presumably as insects emerged as adults and predators consumed insect larvae. Much of the E. nuttallii planted earlier appeared to have herbivory damage, but I did not observe damage to E. nuttallii planted later.

While herbivory by insects on living macrophytes is often thought to be relatively unimportant in freshwater systems, it is not uncommon and has been documented in many cases (Newman,

1991).

This unexpected failure to demonstrate a priority effect illustrates an important consideration in the study of priority effects in general: that relative time is often conflated with real time. In other words, it is difficult to disentangle the effects of an establishment advantage

45 with differences in environmental conditions at the time of plantings, such as differences in temperature, photoperiod, presence and abundance of other species, etc. In addition, there appears to have been a strong year effect (sensu Vaughn and Young (in press)) on the performance of E. nuttallii. This experiment was an extension of a study I conducted one year earlier (Zefferman (2014) using the same experimental location at the same time of year, with the same species collected from the same area, yet I observed rapid initial growth of E. nuttallii in the previous year’s experiment and no obvious herbivory.

In general, priority effects in aquatic systems may pose unique challenges compared to terrestrial systems, and the limited number of studies that address the usefulness of priority in flowing systems have yielded mixed results. Native macrophytes can be successfully established in degraded or newly-created stream reaches (Riis et al., 2009; Zefferman, 2014), but this process can be difficult and expensive, and new plantings are subject to being washed away

(Suren, 2009). Establishing natives has been shown to limit the spread of invasive macrophytes in restored stream reaches with little weed pressure (Larned et al., 2006), but not in streams where invasive macrophytes have been weeded but not removed entirely (Suren, 2009).

Furthermore, establishing native submersed macrophytes early may have both inhibitory and facilitative effects in flowing systems. For example, invasion success of Hydrilla verticillata was hindered by the presence of established native Vallisneria americana in closed mesocosms due to nutrient preemption; however, in the field this inhibitory effect was offset by the constant influx of nutrients and the facilitative effect of established macrophytes catching propagules of the invasive Hydrilla (Chadwell & Engelhardt, 2008).

The results of this experiment and others suggest that simply planting native macrophytes is not enough to reduce invasion and spread by invasive submersed macrophyte propagules. To

46 sufficiently preempt space and nutrients, native macrophytes may need considerable time to establish and grow before invasive propagules reach the area, which may or may not be feasible in actual streams. In addition, the risk of facilitation through propagule catching in streams is worth considering, and would be dependent on site-specific conditions. Of course, priority plantings of macrophytes under different physical conditions, native stem densities, and species may have different outcomes than those of my experiment.

Tables and Figures

Table 1 Mean values and ranges of key nutrient levels of water and soil samples in the experimental channels.

Parameter Mean Range Water Ammonium-N (mg/L) <0.05* NA Nitrate-N (mg/L) 6.24 (6.19-6.33) Orthophosphate-P (mg/L) 0.07** (<0.05-0.10) Soil Ammonium-N (ppm) 2.95 (2.09-3.78) Nitrate-N (ppm) 2.18 (1.55-2.61) Phosphorus (Olsen-P) (ppm) 6.79 (5.05-8.20)

*Ammonium was not detected (mdl=0.05 mg/L) in any samples. **Three samples had orthophosphate levels below detection limits (0.05 mg/L)

Table 2 Means, standard errors, and number of replicates for each species in each treatment combination

Competition Shade level Species RGR (g/g/d) Biomass (mg) N treatment Elodea Concurrent 0.0667 ± 0.0055 882.9 ± 374.4 12 nuttallii Priority 0.0330 ± 0.0049 920.9 ± 354.6 14 Zero Myriophyllum Concurrent 0.0795 ± 0.0021 1945.6 ± 175.5 12 spicatum Priority 0.0831 ± 0.0019 2268.3 ± 200.3 14

Elodea Concurrent 0.0767 ± 0.0067 1404.6 ± 462.4 12 nuttallii Priority 0.0410 ± 0.0051 1602.3 ± 659.7 12 Low Myriophyllum Concurrent 0.0799 ± 0.0025 2011.5 ± 201.1 12 spicatum Priority 0.0764 ± 0.0026 1749.3 ± 199.1 12

Elodea Concurrent 0.0599 ± 0.0044 496.0 ± 82.0 12 nuttallii Priority 0.0450 ± 0.0065 5493.6 ± 3108.2 14 Medium Myriophyllum Concurrent 0.0721 ± 0.0020 1404.2 ± 117.1 12 spicatum Priority 0.0696 ± 0.0018 1267.7 ± 95.6 14

Elodea Concurrent 0.0452 ± 0.0028 240.1 ± 32.7 12 nuttallii Priority 0.0267 ± 0.0036 410.1 ± 122.9 13 High Myriophyllum Concurrent 0.0422 ± 0.0041 416.8 ± 44.4 12 spicatum Priority 0.0399 ± 0.0040 390.1 ± 47.1 13

Table 2.3 ANOVA statistics

Degrees of F-statistic p-value Freedom A. Effect of shade level and species on RGR in concurrent treatment Shade level 3 45.7 <0.0001 Species 1 4.8 0.03 Shade*Species 3 1.7 0.17 B. Effect of shade level and competition treatment (priority) on M. spicatum RGR Shade level 3 118.8 <0.001 Competition treatment 1 0.6 0.45 Shade*Competition 3 1.0 0.38 treatment

Figure 2.1 Diagram of experimental layout in artificial channels. Water flowed through the entire channel system with inputs at the upstream end of each of the three channels.

Figure 2.2 Boxplots comparing RGR of M. spicatum and E. nuttallii planted concurrently across shade levels. Box edges mark the 1st and 3rd quartile, and the median is shown with a dark line. Whiskers extend to a maximum of 1.5 x interquartile range outward, and values beyond this are indicated by circles. Diamonds show the mean of each shade level*species combination. Asterisks indicate significant differences (p < 0.05) in pairwise t-tests between species, and “NS” indicates non-significant differences.

Figure 2.3 Correlation of M. spicatum and E. nuttallii RGR in the concurrent treatment. Thin, solid line is 1:1 line. Most points fall to the upper left of this line, indicating a higher RGR for M. spicatum over E. nuttallii. The Pearson correlation coefficient (r) and p-values for each shade level are as follows (DF = 10 for all tests): zero: r = 0.54, p = 0.07; low: r = 0.80, p = 0.002; medium: r = 0.61, p = 0.04; high: r = 0.14, p = 0.68.

Figure 2.4 Boxplots comparing M. spicatum RGR in the concurrent and priority treatments across shade levels. Box edges mark the 1st and 3rd quartile, and the median is shown with a dark line. Whiskers extend to a maximum of 1.5 x interquartile range outward, and values beyond this are indicated by circles. Diamonds show means of each competition treatment*shade level combination.

Chapter 2 Literature Cited

Aiken, S. G., R. Newroth, & I. Wiles, 1979. The biology of Canadian weeds. 34. Myriophyllum spicatum L. Canadian Journal of Plant Science 59: 201–215.

Alpert, P., E. Bone, & C. Holzapfel, 2000. Invasiveness, invasibility and the role of environmental stress in the spread of non-native plants. Perspectives in Plant Ecology Evolution and Systematics 3: 52–66.

Anderson, L., 2011. Freshwater Plants and Seaweeds In Simberloff, D., & M. Rejmanek (eds), Encyclopedia of Biological Invasions. University of California Press, Berkeley and Los Angeles: 248–258.

Angelstein, S., & H. Schubert, 2009. Light acclimatisation of Elodea nuttallii grown under ambient DIC conditions. Plant Ecology 202: 91–101.

Angelstein, S., C. Wolfram, K. Rahn, U. Kiwel, S. Frimel, I. Merbach, & H. Schubert, 2009. The influence of different sediment nutrient contents on growth and competition of Elodea nuttallii and Myriophyllum spicatum in nutrient-poor waters. Fundamental and Applied Limnology / Archiv für Hydrobiologie 175: 49–57.

Barko, J. W., D. G. Hardin, & M. S. Matthews, 1982. Growth and morphology of submersed freshwater macrophytes in relation to light and temperature. Canadian Journal of Botany 60: 877–887.

Boylen, C. W., L. W. Eichler, & J. D. Madsen, 1999. Loss of native aquatic plant species in a community dominated by Eurasian watermilfoil. Hydrobiologia Springer Netherlands 415: 207–211

Carpenter, S. R., & D. M. Lodge, 1986. Effects of submersed macrophytes on ecosystem processes. Aquatic Botany 26: 341–370.

Chadwell, T. B., & K. A. M. Engelhardt, 2008. Effects of pre-existing submersed vegetation and propagule pressure on the invasion success of Hydrilla verticillata. Journal of Applied Ecology 45: 515–523.

Cheruvelil, K. S., P. A. Soranno, & J. D. Madsen, 2001. Epiphytic macroinvertebrates along a gradient of Eurasian watermilfoil cover. Journal of Aquatic Plant Management 39: 67– 72.

Dukes, J. S., & H. A. Mooney, 1999. Does global change increase the success of biological invaders? Trends in Ecology & Evolution 14: 135–139.

Funk, J. L., 2013. The physiology of invasive plants in low-resource environments. Conservation Physiology 1: 1–17.

Funk, J. L., & P. M. Vitousek, 2007. Resource-use efficiency and plant invasion in low-resource systems. Nature (London) 446: 1079–1081.

Hussner, A., H. P. Hoelken, & P. Jahns, 2010. Low light acclimated submerged freshwater plants show a pronounced sensitivity to increasing irradiances. Aquatic Botany 93: 17–24.

Kardol, P., L. Souza, & A. T. Classen, 2013. Resource availability mediates the importance of priority effects in plant community assembly and ecosystem function. Oikos 122: 84–94.

Keast, A., 1984. The introduced aquatic macrophyte, Myriophyllum spicatum, as habitat for fish and their invertebrate prey. Canadian Journal of Zoology 62: 1289–1303

Lacoul, P., & B. Freedman, 2006. Environmental influences on aquatic plants in freshwater ecosystems. Environmental Reviews 14: 89–136.

Larned, S. T., A. M. Suren, M. Flanagan, B. J. F. Biggs, & T. Riis, 2006. Macrophytes in urban stream rehabilitation: Establishment, ecological effects, and public perception. Restoration Ecology 14: 429–440.

Mjelde, M., P. Lombardo, D. Berge, & S. W. Johansen, 2012. Mass invasion of non-native Elodea canadensis Michx. in a large, clear-water, species-rich Norwegian lake – impact on macrophyte biodiversity. Annales de Limnologie - International Journal of Limnology 48: 225–240.

Mony, C., T. J. Koschnick, W. T. Haller, & S. Muller, 2007. Competition between two invasive Hydrocharitaceae (Hydrilla verticillata (L.f.) (Royle) and Egeria densa (Planch)) as influenced by sediment fertility and season. Aquatic Botany 86: 236–242.

Moore, J. E., & S. B. Franklin, 2012. Water stress interacts with early arrival to influence interspecific and intraspecific priority competition: a test using a greenhouse study. Journal of Vegetation Science 23: 647–656.

Newman, R. M., 1991. Herbivory and detritivory on freshwater macrophytes by invertebrates: a review. Journal of the North American Benthological Society 10: 89–114.

Northwest Hydraulic Consultants, 2010. Species identification and seasonal biomass flux monitoring in Putah South Canal, September 2008 through September 2009.

Peffer, E., 2013. Aquatic Vegetation Assessment of Putah Creek, Lake Solano, the Putah South Canal, and the Terminal Reservoir. Vacaville, CA: 1–123.

Pimentel, D., R. Zuniga, & D. Morrison, 2005. Update on the environmental and economic costs associated with alien-invasive species in the United States. Ecological Economics 52: 273–288.

Reinhart, K. O., J. Gurnee, R. Tirado, & R. M. Callaway, 2006. Invasion through quantitative effects: Intense shade drives native decline and invasive success. Ecological Applications 16: 1821–1831.

Rejmánková, E., 2011. The role of macrophytes in wetland ecosystems. Journal of Ecology and Field Biology 34: 333–345.

Riis, T., R. Schultz, H. M. Olsen, & C. K. Katborg, 2009. Transplanting macrophytes to rehabilitate streams: experience and recommendations. Aquatic Ecology 43: 935–942.

Sand-Jensen, K., & T. V. Madsen, 1991. Minimum light requirements of submerged freshwater macrophytes in laboratory growth experiments. Journal of Ecology 79: 749–764.

Schultz, R., & E. Dibble, 2012. Effects of invasive macrophytes on freshwater fish and macroinvertebrate communities: the role of invasive plant traits. Hydrobiologia 684: 1– 14.

Sculthorpe, C. D., 1967. The Biology of Aquatic Vascular Plants. Edward Arnold Publishers, London.

Smith, C. S., & J. W. Barko, 1990. Ecology of Eurasian watermilfoil. Journal of Aquatic Plant Management 28: 55–64.

Su, W., G. Zhang, Y. Zhang, H. Xiao, & F. Xia, 2004. The photosynthetic characteristics of five submerged aquatic plants. Acta Hydrobiologia Sinica 28: 391–395.

Suren, A. M., 2009. Using macrophytes in urban stream rehabilitation: a cautionary tale. Restoration Ecology 17: 873–883.

Tukey, J. W., 1962. The future of data analysis. The Annals of Mathematical Statistics 33: 17– 21.

Van, T. K., G. S. Wheeler, & T. D. Center, 1999. Competition between Hydrilla verticillata and Vallisneria americana as influenced by soil fertility. Aquatic Botany 62: 225–233.

Vaughn, K. J., & T. P. Young, (in press). Short-term priority over exotic annuals increases the initial density and longer-term cover of native perennial grasses. Ecological Applications.

Weaver, M. J., J. J. Magnuson, & M. K. Clayton, 1997. Distribution of littoral fishes in structurally complex macrophytes. Canadian Journal of Fisheries and Aquatic Sciences.

Young, T. P., J. M. Chase, & R. T. Huddleston, 2001. Community Succession and Assembly. Ecological Restoration 19: 5-18.

Zefferman, E., 2014. Increasing canopy shading reduces growth but not establishment of Elodea nuttallii and Myriophyllum spicatum in stream channels. Hydrobiologia 734: 159–170.

Chapter 3

Predicting importance of abiotic correlates to nuisance macrophyte cover in a regulated California stream using boosted regression tree models

Co-authored with David J. Harris

Summary

Proliferation of submersed macrophytes in regulated waters often requires expensive and logistically difficult interventions, driving water resource managers to look for sustainable, cost- effective strategies to reduce problematic macrophyte growth. Our study investigated the extent and causes of excessive submersed macrophyte growth in the Interdam Reach of Putah Creek in central California, USA, where biomass and propagules clog canal infrastructure downstream. In summer-fall 2011, we surveyed submersed macrophyte cover and environmental conditions, including canopy cover, water velocity, depth, sediment nutrients, and substrate texture. Eurasian watermilfoil (Myriophyllum spicatum) and western waterweed (Elodea nuttallii) were the most abundant species, and along with six additional species comprised our response variable of

‘nuisance macrophyte cover’. Using boosted regression tree models, we identified the abiotic factors most important in predicting nuisance macrophyte cover to be those associated with light availability (sun hours and water depth) and flow (water velocity and substrate texture). This machine learning-based modeling approach enabled us to find biologically-relevant thresholds in predicted macrophyte cover that can be used to guide management decisions. Overall, increasing canopy shading or water depth, and increasing water velocity to flush out fine sediments (e.g., channel narrowing) are likely to be most effective in reducing nuisance macrophyte abundance in the Interdam Reach in the long term. Due to the cosmopolitan distributions of the most

59 abundant species found in this study, our findings have broader relevance to water managers dealing with problem aquatic vegetation in many other regions.

Introduction

Submersed macrophytes (plants that grow underwater) are natural components of many lakes and streams, and contribute to the ecological health and productivity of aquatic ecosystems.

Submersed macrophytes provide food and habitat for invertebrates, fish, and other wildlife, and influence water and sediment chemistry and nutrient cycling (Carpenter & Lodge, 1986;

Rejmánková, 2011). However, overabundance of submersed macrophytes impairs ecosystem functioning and commercial and recreational activities, creating problems for managers of aquatic systems (Anderson, 2011). In flowing waters, excessive submersed macrophyte growth can decrease water velocity, impairing water delivery and increasing flood risk (Bal & Meire,

2009). Dislodged macrophyte biomass and asexual propagules (stem fragments) can clog intake pumps and screens, hindering municipal, industrial, or agricultural water withdrawals (Anderson,

2011). Submersed macrophyte proliferation can also cause extreme daily swings in dissolved oxygen and can reduce the diversity and abundance of fishes and other aquatic species (Killgore

& Hoover, 2001; Schultz & Dibble, 2012).

When submersed macrophytes over-proliferate in a regulated waterway, water resource managers face the often difficult challenge of reducing macrophyte biomass. Traditional methods of submersed macrophyte control include mechanical removal and herbicide use, both of which cause collateral damage to non-target organisms (Nichols, 1991). In addition, these methods generally have only short-term effects, thus requiring repeated and costly intervention

(Anderson, 2011). To control submersed macrophyte growth in a more sustainable and cost- effective way, it is important to better understand how the manipulation of physical or chemical

60 conditions of aquatic systems could reduce proliferation of submersed macrophytes in the long term.

The goal of this study was to inform sustainable management of nuisance aquatic vegetation in a regulated section of Putah Creek called the Interdam Reach in California, USA. Our objectives were to (1) document the extent of vegetation growth in the Interdam Reach, (2) identify the most important environmental factors driving it, and (3) model how macrophyte cover is predicted to vary over a range of these factors. To identify important patterns and thresholds, we modeled our data using boosted regression trees, a relatively new modeling technique that combines methods from machine learning and traditional statistical approaches to achieve high predictive accuracy (Elith et al., 2008).

Materials and Methods

Site Description

Putah Creek flows from the California Coast Range to the Yolo Bypass (Sacramento River floodplain) near Sacramento, CA. In the 1950s, the United States Bureau of Reclamation began the Solano Project to store and deliver water from Putah Creek to municipal, industrial, and agricultural users in the Sacramento Valley. Construction of Monticello Dam in 1957 created the

Lake Berryessa reservoir, which provides up to 1.93 billion cubic meters of water storage

(Harrison et al., 2001). Eleven km downstream, construction of the Putah Creek Diversion Dam formed the Lake Solano reservoir, which stores 888,000 cubic meters of water (Harrison et al.,

2001). From the Diversion Dam, most water is diverted south through the Putah South Canal.

The subject of this study, the Interdam Reach (IDR), consists of the 6.6 km of stream habitat

(Putah Creek) and 4 km of slow-moving lacustrine habitat (Lake Solano) between the Monticello and Putah Diversion Dams (Figure 3.1).

The two dams have altered the flow regime in the IDR: instead of seasonal flooding in the winter months, flows are typically highest in the summer when water is released through the IDR to the Putah South Canal for irrigation. Decreased flooding frequency and intensity in the IDR have caused sediment to accumulate, particularly in the wide, shallow section of Lake Solano just upstream of the Putah Diversion Dam. Submersed macrophytes are abundant in this reach.

Throughout the summer, tens of thousands of macrophyte fragments flow into the Putah South

Canal Headworks each hour, clogging intake screens and infrastructure within the canal

(Northwest Hydraulic Consultants, 2010). Seeds and vegetative propagules establish and grow inside the canal itself, and their eventual decay impairs drinking water quality. Managers of the

Solano Project must conduct costly and logistically-difficult canal clean-outs every year. This study seeks to inform long-term solutions for reducing macrophyte biomass in the IDR, to reduce vegetation management costs and improve ecosystem function.

Sampling Methods

The two sections of the IDR were sampled in two stages: Lake Solano, the deeper 4.0 km stretch of the IDR directly upstream of the Putah Diversion Dam, was surveyed by canoe from

23-26 August 2011; the 6.6 km stretch of Putah Creek between Lake Solano and the Monticello

Dam, was surveyed by foot on 12-21 November 2011.

In Lake Solano, 24 transects were established across the channel at 175 m intervals. In Putah

Creek, 13 transects across the channel were established, with locations based on accessibility and the intention to sample physically variable habitats. Three points were sampled (when possible) on each transect, located one meter inward from the each bank, and in the middle of the channel

(Figure 3.1). In total, we sampled 72 points from Lake Solano and 31 points from Putah Creek.

At each sampling point, we measured the following: percent cover of all macrophyte species found in a 0.25 m2 quadrat; substrate size class (qualitative estimate, Table 3.1); water velocity at

10 cm depth (MJP Geopacks Student Stream Flowmeter); and water depth. In addition, at each sampling point we measured canopy shading with a Solar PathfinderTM and calculated sun hours, a measure of average daily solar radiation reaching a location over one year. In Lake Solano only, we measured dissolved oxygen, specific conductance, pH, temperature, and turbidity using a YSI multi-probe sonde at a depth of 20-30 cm.

We took sediment cores (when possible) measuring 5 cm in diameter and ~15 cm deep at one randomly selected point on each transect using an AMS Multi-stage sludge and sediment sampler. A total of 21 samples from Lake Solano and 5 samples from Putah Creek were obtained. We analyzed sediment samples for nitrate and ammonium (KCl extraction) and plant- available inorganic phosphorus (Olsen-P method; Murphy and Riley 1962). The UC Davis

Analytical Laboratory in Davis, CA (http://anlab.ucdavis.edu) analyzed sediment total nitrogen and carbon (SOP 320.03) and particle size distribution (SOP 470.03), and analyzed the water samples for nitrate and ammonium (SOP 847.03) and orthophosphate (SOP 865.03).

To characterize general water nutrient levels and clarity during the study (not for use in the modeling exercise), we took water samples near the downstream, middle, and upstream portions of Lake Solano and in two locations in Putah Creek. We also measured photosynthetically active radiation (PAR) at several depths in the middle point of each transect with a LI-COR LI-193 spherical quantum sensor to calculate vertical extinction coefficients, kd ( = ; Io = 𝑙𝑙𝑙𝑙𝑙𝑙𝑜𝑜−𝑙𝑙𝑙𝑙𝑙𝑙𝑧𝑧 𝑘𝑘𝑑𝑑 𝑧𝑧 light intensity just below the water’s surface, Iz = light intensity at depth z).

Modeling approach

To explore the relationships between measured environmental factors and macrophyte abundance in the IDR, we created boosted regression tree models using R software (R version

2.15.2, R Development Core Team, Vienna, AT) with the gbm package version 1.6 (Ridgeway,

2013). Boosted regression trees (BRT) use machine learning to combine multiple decision trees, also called classification or regression trees, into models with high predictive ability. Boosting refers to the process of creating many simple trees in succession, with each tree built on the residual error of the previous tree. The trees are then combined into a single predictive model.

Boosted regression trees have several advantages over linear models, such as being better able to identify nonlinearities like threshold effects and interactions. For overviews of this technique, see

Elith et al. (2008) and De’ath (2007).

We created 150 models from bootstrapped samples of the full dataset. We resampled by transect rather than by individual points to reduce fitting based on spatial autocorrelation. Each model was based on 4100 trees, as decided by the .632 bootstrap procedure (Efron, 1983). Each tree had two splits (three branches), and the shrinkage, or learning rate, was 0.001.

Predictor variables used in the models are presented in Table 3.2 with summary statistics. All but three predictors were continuous variables. “System” (Lake Solano or Putah Creek) and

“Position” (left, middle, or right sampling location on a transect when facing upstream) were categorical. Qualitatively-assessed substrate was also treated as a categorical variable. Because some sampling points had more than one substrate class present, this variable was input as the proportion of a given substrate class at each sampling point, divided evenly between the number of substrate types present. For example, if a point had both classes 1 and 4, the substrate for that point was assigned as 50% substrate 1 and 50% substrate 4.

For our response variable of total nuisance macrophyte abundance, we combined percent cover of the following submersed macrophyte species: Myriophyllum spicatum (Eurasian watermilfoil), Potamogeton foliosus (leafy pondweed), Stuckenia pectinata (sago pondweed),

Zannichellia palustris (horned pondweed, z-grass), Elodea nuttallii (western waterweed),

Potamogeton crispus (curly leaf pondweed), and Ceratophyllum demersum (coontail). These species made up the majority (~95%) of submersed macrophyte cover in the IDR and are the most problematic, either because of their extensive growth within the Putah South Canal (first four), and/or because they produce a significant amount of floating plant material that can clog screens at the canal Headworks (all seven), as determined by a prior vegetation monitoring study and personal observations (Northwest Hydraulic Consultants, 2010; Peffer, 2013). Mosses

(bryophytes) were excluded from the calculation of total nuisance macrophyte cover because even though moss cover was often high on boulders, mosses do not produce large quantities of biomass. The response variable of percent cover was converted to a discrete variable composed of 20 ‘spaces’ per sampling point and modeled as a binomial process whereby each space could be either occupied or not occupied by nuisance macrophytes.

Using the BRT model results, we calculated relative importance of each variable in predicting nuisance macrophyte cover with the ‘summary’ function in the gbm package. Relative importance of a variable describes the percent by which a model improves over a baseline model by adding that variable. For the six top predictors, we modeled predicted macrophyte abundance across the range of values found in the dataset. By averaging over all models, we obtained mean predictions, 50% confidence intervals (25% and 75% quantiles), and 95% confidence intervals

(2.5% and 97.5% quantiles). Overall model performance was assessed using the .632 bootstrap method (Efron, 1983).

Results

Background conditions

Percent cover of nuisance submersed macrophytes was high throughout the IDR, with 68 out of 103 sampled points having 80% or greater cover (Figure 3.2). In both the Lake Solano and

Putah Creek sections of the IDR, cover ranged from 0-100%.

The submersed macrophytes found in the IDR, in order of greatest to least average percent cover, were M. spicatum, E. nuttallii, P. crispus, C. demersum, S. pectinata, Z. palustris, bryophytes (mosses), and P. foliosus (Figure 3.3). All of these species are native to California except M. spicatum and P. crispus. The percent of sampling points in which each taxon was present followed a similar pattern (Figure 3.3), but E. nuttallii was found more frequently than

M. spicatum.

All water samples came back below detection limits (0.05 mg/L) for ammonium and soluble phosphorus. Nitrate values were 0.12 mg/L and 0.09 mg/L at upstream and downstream transects, respectively, in Putah Creek. The samples from the upstream and middle points in

Lake Solano both had nitrate concentrations of 0.06 mg/L, while the downstream sample was below detection limits.

Vertical extinction coefficients, which measure light reduction per unit depth, averaged 0.46 m-1 and were similar throughout the system. This value is in the low-middle range for freshwater lakes (Kirk, 1994). Lower values indicate more light penetration through the water column.

Modeling results

Relative performance of all predictor variables used in the BRT models is presented in Figure

3.4, and predictions of total cover of nuisance macrophytes along the ranges of each of the six

66 most important predictor variables are shown in Figure 3.5. Yearly mean sun hours (Figure

3.5A) was the most important variable in the BRT models, with 24.7% of the explanatory power.

At sampling points with sun hours of approximately 2.4 kWh/m2/d and higher, macrophyte cover was predicted to be around 80%. Below this, macrophyte cover was predicted to be around 65% or less. Seventy-seven percent of sampled points were above this 2.4 kWh/m2/d threshold.

The proportion of soft substrate and boulders were the second and third most important variables, with 20.5% and 19.3% of the importance, respectively. Soft substrate was positively correlated with nuisance macrophyte cover (Figure 3.5B), while boulders were negatively correlated (Figure 3.5C). The majority of sampling points (62%) contained soft substrate, while only 17% had boulders. The interaction of the two most important variables—percent soft substrate and sun hours—is shown in Figure 3.6.

Depth had a relative importance of 15.6%. Depths between 0.3 and 2.4 m showed relatively little difference in predicted macrophyte cover (Figure 3.5D). At greater than 2.4 m, macrophyte cover was predicted to decline with greater depth. There was also a predicted decline in macrophyte cover at the shallowest depths.

The relative importance of sediment total N was 7.0%. The response of nuisance macrophyte cover to sediment total N levels was relatively uniform, with a reduction in cover only predicted to occur below 0.05% (Figure 3.5E). However, only three sampling points out of 26 were at or below 0.05% N.

Water velocity had a relative importance of 4%, and BRT model predictions show a slight negative effect on macrophyte cover (Figure 3.5F). At the time of sampling, 80% of points had water velocities of 0.25 m/s or lower, with 41% being below detection limits, indicating relatively low water velocities throughout the IDR. Relative importances of the remaining

67 predictors were all less than 2%. Notably, ‘System’ was among these less important predictors, indicating that sampling differences between the Lake Solano and Putah Creek sections of the

IDR did not strongly affect model predictions.

Evaluation of model performance using the .632 bootstrap method yielded an R2 of 0.60. The models tended to underpredict high values and overpredict low values (Figure 3.7) for two main reasons: first, with binomial data, predicting extremes often results in overfitting, and it’s ‘safer’ to predict intermediate values; second, averaging over 150 predictions smooths out more extreme predictions that might arise in only a subset of models.

Discussion

For many years, prolific macrophyte growth in the Putah Creek IDR has created management problems by producing huge quantities of propagules and plant material that flow into the Putah

South Canal Headworks, clogging screens and intakes, and ultimately causing vegetation growth in the canal itself. However, the extent of submersed macrophyte coverage in the IDR, and the physical factors driving it, have been largely unknown. The main objective of our study was to determine the primary factors influencing growth of nuisance macrophytes within the IDR, so that management could target these factors to sustainably reduce problematic macrophyte growth.

Nuisance submersed macrophyte abundance was high throughout the IDR, with the majority of surveyed points having over 80% cover of macrophytes (Figure 3.2). Non-native

Myriophyllum spicatum and native Elodea nuttallii dominated the communities in both the Putah

Creek and Lake Solano sections of the IDR (Figure 3.3). These two species have often been a problem for managers throughout North America and Europe: both have relatively high photosynthetic efficiency and growth rates, and can produce large quantities of vegetative

68 propagules (stem fragments) during the summer growing season (Nichols & Shaw, 1986;

DiTomaso & Healy, 2003).

The most important physical factors associated with nuisance submersed macrophyte growth in the BRT models were those associated with light availability and water velocity (Figure 3.4).

Yearly average sun hours, which reflect the quantity and position of overstory shade at a specific location, comprised almost a quarter of the relative importance. Negative correlations between riparian shading and submersed macrophyte abundance have been identified by numerous studies (e.g., Madsen and Adams 1989; Köhler et al. 2010; Julian et al. 2011), and this study joins others that have found light availability to be one of the most important predictors of macrophyte abundance in streams (e.g., Canfield and Hoyer 1988; Ali et al. 2011; Wood et al.

2012). Increasing riparian shading has been recommended as a management strategy for reducing submersed macrophyte growth (Dawson & Kern-Hansen, 1979; Anderson, 2011), and the importance of sun hours in our study suggests reducing incident light levels in the IDR should be prioritized. However, sun hours were positively associated with greater macrophyte cover only up to approximately 2.4 kWh/m2/d, above which greater solar radiation was predicted to have little effect on macrophyte cover (Figure 3.5A). Although orientation of canopy cover influences the calculation of sun hours, a value of 2.4 kWh/m2/d corresponds to approximately

50% canopy cover. Thus, increasing riparian shading in this system could be effective in reducing macrophyte abundance, but only in areas where achieving high levels of canopy cover are possible.

In this relatively shallow, clear system, the influence of water depth—the fourth most important factor in the BRT models—probably also reflects of the importance of light availability. A small reduction in macrophyte cover was predicted at the shallowest depths

(Figure 3.5D), which may be due to yearly water level fluctuations that cause shallow depths to dry out at certain times of the year. However, in general, nuisance submersed macrophyte cover was predicted to decrease with greater depth, but only at depths greater than around 2.4 m. The similar predicted macrophyte cover over a range of shallow depths suggests that, as with sun hours, light is probably only limiting macrophyte growth below a threshold level. Given a typical amount of light at the surface of unshaded water in the summer in central California (~1900

μmol/m2/s PAR), and the average vertical extinction coefficient of 0.46 m-1 for the IDR, a depth of 2.4 m corresponds to around 625 μmol/m2/s PAR. This value is within the range of typical light saturation points for submersed macrophytes (Van et al., 1976; Kirk, 1994). Therefore, plants growing at depths shallower than 2.4 m may be receiving non-limiting levels of PAR. As such, increasing the depth of certain sections of the IDR through dredging may reduce the growth of problem macrophytes, but probably only at depths greater than 2.4 m.

Factors directly or indirectly related to water velocity also emerged as key predictors of nuisance submersed macrophyte cover. Previous studies have found complicated, and sometimes contradictory effects of water velocity on submersed macrophyte growth (reviewed in Madsen et al. (2001)). At lower ranges, increasing water velocity may enhance growth rates by increasing gas and nutrient exchange (Westlake, 1967; Madsen & Sondergaard, 1983); however, high water velocity can remove fine sediments that are favorable for macrophyte growth, and/or physically remove, damage, or stress macrophytes and their propagules (Madsen et al., 1993; Riis & Biggs,

2003).

In our study, water velocity appeared to have a small negative effect on macrophyte abundance (Figure 3.5F). The water velocity data used in the BRT models were collected at single time points, and do not reflect the full range of velocities that occur intra- and inter-

70 annually. However, they do represent an approximation of relative water velocities across sampling points within the IDR.

Substrate class is arguably a better long-term indicator of water velocity, and the finest and coarsest substrates (soft substrate and boulders) were found to be the second and third most important factors in the BRT models, respectively. Very few sampling points had rock or boulders as the main substrate class in the IDR, and fine sediments were widespread. Of course, in addition to being an indicator of water velocity, substrate class itself is important to macrophyte abundance, as larger classes (rocks and boulders) provide a less hospitable rooting medium (Sculthorpe, 1967). Interestingly, although sediment texture has been shown to influence growth rates of submersed macrophytes (Barko & Smart, 1986), the quantitative measures of particle size distribution in the sediment (% sand, silt, and clay) were not important predictors in the BRT models.

The accumulation of fine sediments in the IDR may be caused, at least in part, by the moderation of flows between the Monticello and Putah Diversion Dams. Periodic high flow events are known to reduce macrophyte proliferation (Lacoul & Freedman, 2006), and planned

“flushing flows” to remove sediments and vegetation have been used successfully to control macrophytes in flow-regulated rivers (Rorslett & Johansen, 1996; Merz & Setka, 2004; Batalla

& Vericat, 2009). Such planned dam releases may be a viable solution for reducing nuisance macrophyte growth in the IDR as well, though downstream impacts are important to consider.

Whether or not sediment nutrients play an important role in controlling submersed macrophyte abundance is controversial. Sediment fertilization in experimental conditions often results in increased growth of submersed macrophytes (e.g., Best et al. 1996; Carr and Chambers

1998). However, when background nutrient levels are high in situ, other factors such as light and

71 carbon dioxide availability often trump the importance of nutrients in limiting macrophyte growth, and it may only be in oligotrophic systems that sediment nutrients play an influential role in macrophyte abundance (Barko et al., 1991; Carr et al., 1997). Of the measured sediment nutrients in this study—nitrate, ammonium, total N, total C, and soluble P—only sediment total

N was an important predictor of nuisance submersed macrophyte growth, and the effect was present only at the lowest N levels. Because our models are based on correlations, we cannot distinguish whether low sediment N causes lower macrophyte growth, or whether decreased macrophyte growth causes less accumulation of sediment N. However, one might expect that if the latter were true, a linear relationship would be detected instead of the more asymptotic relationship that is predicted in the BRT models (Figure 3.5E).

Overall, we found high abundance of nuisance submersed macrophytes in the IDR, which can be attributed largely to the high percentage of shallow, unshaded aquatic habitat, with low to moderate water velocity and fine sediments. Perhaps the best solution for addressing these factors would be narrowing and deepening the main channel of the IDR, particularly in Lake

Solano, and planting canopy-forming vegetation along the banks. This would decrease light availability, increase water velocity (thus decreasing fine sediments), and also decrease the total amount of available habitat for submersed macrophytes. The interaction plot of sun hours and percent soft substrate (Figure 3.6) shows that reducing both light and soft substrate is predicted to be more effective in reducing total macrophyte cover than reducing either independently.

Increased shading and faster transport of water through the IDR, which comes from hypolimnetic discharge from the Monticello Dam, would also decrease water temperatures.

Though not modeled in this study, lower water temperatures are associated with reduced macrophyte growth rates and reproduction (Barko & Smart, 1981; Barko et al., 1982; Lacoul &

Freedman, 2006). Due to the ability for managers to store water in the Lake Berryessa reservoir and regulate water releases at Monticello Dam, storing large amounts of water in the IDR may not be necessary. Channel narrowing, though resource-intensive in the short term, may also be the most sustainable approach to reducing problematic macrophyte growth in the IDR.

Conclusions

Identifying and understanding the primary drivers of nuisance macrophyte abundance is important for prioritizing management actions in regulated streams and human-made waterways.

Using BRT as a modeling approach resulted in a high level of in-sample predictive accuracy (R2

= 0.60), and enabled us to find important biologically-meaningful thresholds in our predictor variables that can aid management decisions. Because the most abundant submersed macrophyte species identified in this study are cosmopolitan in distribution, our findings have relevance to managers grappling with macrophyte overabundance worldwide. To our knowledge, our study represents the first use of BRT in identifying and modeling the most important factors predicting submersed macrophyte cover. A similar approach could be useful to managers of many types of wetland and aquatic systems looking to understand and apply limited resources to a variety of nuisance taxa.

Tables and Figures

Table 3.1 Size classes of substrates

Substrate Qualitative Assessment Scale

Soft Substrate (silt/clay) (1) < 1/16 mm Sand (2) 1mm to 1/16 mm Gravel (3) 2.5 inches to 1 mm Rock (4) 10 inches to 2.5 inches (64mm) Boulders (5) >10 inches (256 mm)

Table 3.2 Summary statistics for predictor variables

Variable Standard Predictor Variable Average* Range N type Deviation Water velocity (m/s) Continuous 0.17 0.22 0.05 - 1.81 103 Depth (m) Continuous 1.04 0.74 0.15 - 3.80 103 Sun Hours-Yearly (kWh/m2/d) Continuous 3.58 1.27 0.95 - 4.87 102 Sand (%) Continuous 58 23 15 - 88 24 Silt (%) Continuous 28 18 5 - 64 24 Clay (%) Continuous 15 6 6 - 30 24 Sediment N (Total %) Continuous 0.097 0.064 0.023 - 0.347 26 Sediment C (Total %) Continuous 1.13 0.64 0.21 - 3.50 26 Sediment Nitrate (μg/g) Continuous 0.15 0.15 0.07 - 0.80 26 Sediment Ammonium (μg/g) Continuous 37.4 42.9 0.9 - 191 26 Sediment Soluble P (μg/g) Continuous 6.8 3.12 1.4 - 15.5 23 Substrate (qualitative scale) Ordinal 1 - 1 - 5 87 Right, middle, Position Categorical - - 103 left Lake Solano, System Categorical - - 103 Putah Creek * Averages are arithmetic means, except for “substrate”, which is the modal value.

Figure 3.1 Map of the Putah Creek Interdam Reach. Sampling points are marked with black diamonds. A total of 103 points were sampled on 37 transects in summer 2011. Inset map shows location of study system within California, USA

Figure 3.2 Histogram of percent nuisance submersed macrophyte cover in all 103 sampled plots in the Interdam Reach

Figure 3.3 Occurrence of submersed macrophyte taxa throughout the Interdam Reach. Dark bars show the percentage of sampling points in which a taxon was present, and light bars show average percent cover across sampling points

Figure 3.4 Relative importance of 18 predictor variables (see Table 3.2) used in the boosted regression tree models. Relative importance of a variable describes the proportion of variation in the data explained by that variable relative to all other variables in the model

Figure 3.5 Panels A-F show predicted nuisance macrophyte percent cover over the range of each of the top six predictor variables. Values were generated from 150 boosted regression tree models fitted to bootstrapped samples from the original data. Means and confidence intervals came from averaging over all models; 50% and 95% confidence intervals were based on 25% and 75% quantiles and the 2.5% and 97.5% quantiles, respectively. Tick marks at the top of plots A, D, E, and F show the distribution of observed points

Figure 3.6 Interaction plot of the two most important predictor variables: sun hours and percent soft substrate

Figure 3.7 Plot of average model predictions of nuisance macrophyte percent cover at a sampling point vs. observed values from field-collected data. Line shows an ideal 1:1 relationship between predicted and observed values. The model tends to predict values that are less extreme than the observed values

Chapter 3 Literature Cited

Ali, M. M., S. A. Hassan, & A. S. M. Shaheen, 2011. Impact of riparian trees shade on aquatic plant abundance in conservation islands. Acta Botanica Croatica 70: 245–258.

Anderson, L., 2011. Freshwater Plants and Seaweeds In Simberloff, D., & M. Rejmanek (eds), Encyclopedia of Biological Invasions. University of California Press, Berkeley and Los Angeles: 248–258.

Bal, K. D., & P. Meire, 2009. The influence of macrophyte cutting on the hydraulic resistance of lowland rivers. Journal of Aquatic Plant Management 47: 65–68.

Barko, J. W., D. Gunnison, & S. R. Carpenter, 1991. Sediment interactions with submersed macrophyte growth and community dynamics. Aquatic Botany 41: 41–65.

Barko, J. W., D. G. Hardin, & M. S. Matthews, 1982. Growth and morphology of submersed freshwater macrophytes in relation to light and temperature. Canadian Journal of Botany 60: 877–887.

Barko, J. W., & R. M. Smart, 1981. Comparative influences of light and temperature on the growth and metabolism of selected submersed fresh-water macrophytes . Ecological Monographs 51: 219–235.

Barko, J. W., & R. M. Smart, 1986. Sediment-related mechanisms of growth limitation in submersed macrophytes. Ecology 67: 1328–1340.

Batalla, R. J., & D. Vericat, 2009. Hydrological and sediment transport dynamics of flushing flows: implications for management in large Mediterranean rivers. River Research and Applications 25: 297–314.

Best, E. P. H., H. Woltman, & F. H. H. Jacobs, 1996. Sediment-related growth limitation of Elodea nuttallii as indicated by a fertilization experiment. Freshwater Biology 36: 33–44.

Carpenter, S. R., & D. M. Lodge, 1986. Effects of submersed macrophytes on ecosystem processes. Aquatic Botany 26: 341–370.

Carr, G. M., & P. A. Chambers, 1998. Macrophyte growth and sediment phosphorus and nitrogen in a Canadian prairie river. Freshwater Biology 39: 525–536.

Carr, G. M., H. C. Duthie, & W. D. Taylor, 1997. Models of aquatic plant productivity: a review of the factors that influence growth. Aquatic Botany 59: 195–215.

DiTomaso, J. M., & E. H. Healy, 2003. Aquatic and Riparian Weeds of the West. University of California, Agriculture and Natural Resources Publication 3421.

Efron, B., 1983. Estimating the error rate of a prediction rule: improvement on cross-validation. Journal of the American Statistical Association 78: 316–331.

Elith, J., J. R. Leathwick, & T. Hastie, 2008. A working guide to boosted regression trees. The Journal of animal ecology 77: 802–813.

Glenn De’ath, 2007. Boosted trees for ecological modeling and prediction. Ecology 88: 243–251.

Harrison, L. L., R. C. MacArthur, & R. A. Sanford, 2001. Lake Solano sediment management study. Watershed Management and Operations Management 2000 1–10.

Julian, J. P., S. Z. Seegert, S. M. Powers, E. H. Stanley, & M. W. Doyle, 2011. Light as a first- order control on ecosystem structure in a temperate stream. Ecohydrology 4: 422–432.

Killgore, K. J., & J. A. N. J. Hoover, 2001. Effects of hypoxia on fish assemblages in a vegetated waterbody. Journal of Aquatic Plant Management 39: 40–44.

Kirk, J. T. O., 1994. Light and Photosynthesis in Aquatic Ecosystems, 2nd Edition. Cambridge University Press, Great Britain.

Köhler, J., J. Hachoł, & S. Hilt, 2010. Regulation of submersed macrophyte biomass in a temperate lowland river: Interactions between shading by bank vegetation, epiphyton and water turbidity. Aquatic Botany 92: 129–136.

Lacoul, P., & B. Freedman, 2006. Environmental influences on aquatic plants in freshwater ecosystems. Environmental Reviews 14: 89–136.

Madsen, J. D., & M. S. Adams, 1989. The distribution of submerged aquatic macrophyte biomass in a eutrophic stream, Badfish Creek: the effect of environment. Hydrobiologia 171: 111–119.

Madsen, J. D., P. A. Chambers, W. F. James, E. W. Koch, & D. F. Westlake, 2001. The interaction between water movement, sediment dynamics and submersed macrophytes. 71–84.

Madsen, T. V., & M. Sondergaard, 1983. The effects of current velocity on the photosynthesis of Callitriche stagnalis Scop. Aquatic Botany 15: 187–193.

Madsen, T. V., H. O. Enevoldsen, & T. B. Jorgensen, 1993. Effects of water velocity on photosynthesis and dark respiration in submerged stream macrophytes. Plant, Cell and Environment 16: 317–322.

Merz, J. E., & J. D. Setka, 2004. Evaluation of a spawning habitat enhancement site for Chinook salmon in a regulated California River. North American Journal of Fisheries Management 24: 397–407.

Murphy, J., & J. P. Riley, 1962. A modified single-solution method for the determination of phosphate in natural waters. Analytica Chemica Acta 27: 31–36.

Nichols, S. a, 1991. The interaction between biology and the management of aquatic macrophytes. Aquatic Botany 41: 225–252.

Nichols, S. A., & B. H. Shaw, 1986. Ecological life histories of the 3 aquatic nuisance plants, Myriophyllum spicatum, Potamogeton crispus and Elodea canadensis. Hydrobiologia 131: 3–21.

Northwest Hydraulic Consultants, 2010. Species identification and seasonal biomass flux monitoring in Putah South Canal.

Peffer, E., 2013. Aquatic Vegetation Assessment of Putah Creek, Lake Solano, the Putah South Canal, and the Terminal Reservoir. Vacaville, CA: 1–123.

Rejmánková, E., 2011. The role of macrophytes in wetland ecosystems. Journal of Ecology and Field Biology 34: 333–345.

Ridgeway, G., 2013. gbm: Generalized Boosted Regression Models. http://cran.r- project.org/package=gbm.

Riis, T., & B. J. F. Biggs, 2003. Hydrologic and hydraulic control of macrophyte establishment and performance in streams. Limnology and Oceanography 48: 1488–1497.

Rorslett, B., & S. W. Johansen, 1996. Remedial measures connected with aquatic macrophytes in Norwegian regulated rivers and reservoirs. Regulated Rivers Research and Management 12: 509–522.

Schultz, R., & E. Dibble, 2012. Effects of invasive macrophytes on freshwater fish and macroinvertebrate communities: the role of invasive plant traits. Hydrobiologia 684: 1– 14.

Sculthorpe, C. D., 1967. The Biology of Aquatic Vascular Plants. Edward Arnold Publishers, London.

Van, T. K., W. T. Haller, & G. Bowes, 1976. Comparison of the photosynthetic characteristics of three submersed aquatic plants. Plant Physiology 58: 761–768.

Westlake, D. F., 1967. Some effects of low-velocity currents on the metabolism of aquatic macrophytes. Journal of Experimantal Botany 13: 187–205.

Wood, K. a, R. a Stillman, R. T. Clarke, F. Daunt, & M. T. O’Hare, 2012. Understanding plant community responses to combinations of biotic and abiotic factors in different phases of the plant growth cycle. PLOS ONE 7.