Response of an invasive native wetland plant to environmental flows: Implications for managing regulated floodplain ecosystems

Journal of Environmental Management 132 (2014) 268e277

Contents lists available at ScienceDirect

Journal of Environmental Management journal homepage: www.elsevier.com/locate/jenvman

Response of an invasive native wetland plant to environmental flows: Implications for managing regulated floodplain ecosystems Lyndsey M. Vivian*, David J. Marshall, Robert C. Godfree CSIRO Plant Industry, GPO Box 1600, Canberra 2601, Australia

a r t i c l e i n f o

a b s t r a c t

Article history: Received 25 June 2013 Received in revised form 4 November 2013 Accepted 6 November 2013 Available online 8 December 2013

The natural flow regimes of rivers underpin the health and function of floodplain ecosystems. However, infrastructure development and the over-extraction of water has led to the alteration of natural flow regimes, resulting in the degradation of river and floodplain habitats globally. In many catchments, including Australia’s MurrayeDarling Basin, environmental flows are seen as a potentially useful tool to restore natural flow regimes and manage the degradation of rivers and their associated floodplains. In this paper, we investigated whether environmental flows can assist in controlling an invasive native floodplain plant in Barmah Forest, south-eastern Australia. We experimentally quantified the effects of different environmental flow scenarios, including a shallow (20 cm) and deeper (50 cm) flood of different durations (12 and 20 weeks), as well as drought and soil-saturated conditions, on the growth and survival of seedlings of Juncus ingens, a native emergent macrophyte that has become invasive in some areas of Barmah Forest following river regulation and alteration of natural flow regimes. Three height classes of J. ingens (33 cm, 17 cm and 12 cm) were included in the experiment to explicitly test for relationships between treatments, plant survival and growth, and plant height. We found that seedling mortality occurred in the drought treatment and in the 20-week flood treatments of both depths; however, mortality rates in the flood treatments depended on initial plant height, with medium and short plants (initial heights of
17 cm) exhibiting the highest mortality rates. Both the 20 cm and 50 cm flood treatments of only 12 weeks duration were insufficient to cause mortality in any of the height classes; indeed, shoots of plants in the 20 cm flood treatment were able to elongate through the water surface at rapid rates. Our findings have important implications for management of Barmah Forest and floodplain ecosystems elsewhere, as it demonstrates the potential for using environmental flows to limit the spread of invasive plants by targeting a life-stage that is particularly sensitive to prolonged submergence. However, there may be narrow thresholds between the conditions that provide effective control of an invasive species, and those that instead facilitate growth and may promote further invasion. Ó 2013 Elsevier Ltd. All rights reserved.

Keywords: Juncus ingens Submergence Flooding tolerance Invasive species Ecosystem degradation

1. Introduction Freshwater ecosystems are highly productive habitats, rich in biodiversity, and provide a range of ecosystem services (Millennium Ecosystem Assessment, 2005, Dudgeon et al., 2006). Yet despite their ecological, social and economic importance, freshwater habitats e including rivers, floodplains and wetlands e are experiencing accelerating rates of loss and degradation (Tockner and Stanford, 2002; Millennium Ecosystem Assessment, 2005). Among the major threats are infrastructure development and water extraction, resulting in the modification of natural flow

* Corresponding author. Tel.: þ61 2 6246 5137. E-mail addresses: [email protected] (L.M. Vivian), David.Marshall@csiro. au (D.J. Marshall), [email protected] (R.C. Godfree). 0301-4797/$ e see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.jenvman.2013.11.015

regimes, and the encroachment of invasive species (Kercher and Zedler, 2004; Dudgeon et al., 2006). Natural flow regimes underpin the health of many freshwater ecosystems, particularly rivers and floodplains. Consequently, using environmental flows to mimic components of the natural flow regime is a widespread strategy for managing degraded river and floodplain ecosystems (Poff and Zimmerman, 2010). For example, environmental flows have been used to meet the breeding requirements of water birds (Kingsford and Auld, 2005) and native fish (Naiman et al., 2008; King et al., 2010; Cross et al., 2011). Conversely, flow regimes could be manipulated to control populations of invasive plant species by providing unfavourable conditions for their persistence (Catford et al., 2011). This may be particularly effective in habitats that have experienced a loss of their natural wet-dry cycles, which can facilitate the establishment of invasive species that would otherwise not be able to tolerate

L.M. Vivian et al. / Journal of Environmental Management 132 (2014) 268e277

more frequent, prolonged, or deeper floods (Bunn and Arthington, 2002). This study aims to investigate the potential for environmental flows to control seedlings of Juncus ingens (giant rush), an Australian native emergent macrophyte that has become invasive in a south-eastern Australian floodplain, Barmah Forest, due to changes in natural flow regimes. Regulation of the adjacent Murray River in the mid-1930s has reduced the frequency, depth and duration of winterespring floods, which historically occurred in the majority of years (Bren et al., 1988). Conversely, shallow summer floods, which were rare prior to river regulation, have increased in frequency. These changes have enabled J. ingens to expand from wetland margins into areas that historically would have been both too dry in summer and too deeply flooded in winter/spring for its persistence (Mayence et al., 2010). For many plants, prolonged submergence from flooding can hamper growth and eventually cause death (Bailey-Serres and Voesenek, 2008). However, mature J. ingens forms dense, rhizomatous clumps over 3 metres tall. Achieving the complete submergence of these large plants using environmental flows e which tend to be shallow in Barmah Forest due to the risk of flooding adjacent private land (Abel et al., 2006; MDBA, 2012) e is extremely difficult. Instead, environmental flows may be more effectively targeted at submerging shorter seedlings. The effectiveness of using environmental flows to control an invasive plant will depend on the plant’s strategy for responding to submergence stress: some species can ‘escape’ submergence though rapid shoot elongation to the water surface, whereas others exhibit a ‘quiescent’ strategy by tolerating submergence for a certain period (Voesenek et al., 2006; Colmer and Voesenek, 2009). Therefore understanding the submergence response of J. ingens seedlings is critical for managing any environmental flows aimed at reducing growth and causing mortality. In this study we experimentally quantify the effects of different environmental flow scenarios on J. ingens seedling survival and growth: a shallow (20 cm) flood and a deeper (50 cm) flood of two different durations (12 and 20 weeks), and two non-flood treatments: drought and soilsaturated conditions. Since the height of seedlings will determine the depth at which plants are submerged, we also test three height classes of J. ingens seedlings: short (12 cm tall), medium (17 cm) and tall (33 cm). We hypothesise that plants that are completely submerged will have the shortest survival time, with shorter plants succumbing more rapidly than taller plants. We expect that the soil-saturated treatment will provide the most favourable conditions for rapid growth and survival, with the tallest plants exhibiting the fastest growth rates.

2. Materials and methods 2.1. Study area and study species Barmah Forest is a 28,500 ha floodplain located in Australia’s MurrayeDarling Basin, dominated by Eucalyptus camaldulensis (river red gum) forest and woodland, interspersed with treeless wetland areas dominated by grasses, rushes and sedges (Chesterfield, 1986) (Fig. 1). Topography is uniformly flat. Mean annual precipitation is 440 mm (Mathoura State Forest weather station, 35.81 S 144.90 E). Historically, deep floods of up to 6 months occurred in most years, commencing in winter and receding in early summer (Dexter et al., 1986; Bren et al., 1988). The construction of the Hume Reservoir in the upper Murray River in the mid-1930s has reduced the extent, duration and frequency of winterespring floods (Bren et al., 1988). In addition, high channel volumes are maintained in summer to provide water for irrigation of surrounding agricultural land, resulting in frequent overbank

269

flows and shallow flooding of the forest in a period that was historically usually dry (Chong and Ladson, 2003). J. ingens (giant rush) is a native emergent macrophyte that can grow to over 3 m, occurring in south-eastern mainland Australia (The Council of Heads of Australasian Herbaria, 2013). Historically, J. ingens was a minor component of the vegetation at Barmah Forest; however, since river regulation it has encroached into many treeless wetlands and lakes, including areas dominated by the native grass Pseudoraphis spinescens (Moira grass) (Fig. 1a), which is of particular concern due to the value of P. spinescens grasslands as breeding and feeding habitat for many wetland species (Chesterfield, 1986). Recent mapping of J. ingens spread shows that the rate of invasion increased during the 1970s, and it is now wellestablished as dense monospecific stands in lakes and treeless wetland areas throughout the forest (DPI, 2009). 2.2. Experimental treatments To address our aim of investigating the potential for environmental flows to control seedlings of J. ingens, we used a controlled experiment to test the effects of different flood depths and durations on J. ingens seedling growth and survival, conducted at CSIRO Black Mountain Laboratories in Canberra. Flood treatments were designed to reflect realistic scenarios of environmental flows used in Barmah Forest, which have been implemented since 1998 (Ward, 2005; Abel et al., 2006; GB CMA 2013). These involve the release of a specific volume of water from Hume Reservoir, often to reduce the rate of natural flood recession until colonially-nesting water birds successfully complete breeding, but because of restrictions in the volume of water that can be used for any single environmental flow, the depths of the resulting floods tend to be shallow (Ward, 2005; King et al., 2010; MDBA, 2010). Other environmental objectives have included enhancing breeding opportunities for native fish and frogs, and improving vegetation health, particularly P. spinescens, which is thought to require at least 50 cm depth flooding for 5e9 months of the year (Ward, 1991). We therefore selected a 50 cm flood, and a shallower 20 cm flood, to reflect typical depths of environmental flows in treeless plain areas dominated by J. ingens. For each of these two flood depths, we selected two submergence durations: (a) 20 weeks, and (b) 12 weeks followed by an 8 week ‘recovery phase’ in non-flooded conditions of soil saturation. This latter scenario was selected to examine whether seedlings can recover after being exposed to a 12 week flood, a typical maximumlength duration environmental flow in Barmah Forest (K. Ward pers. comm.), followed by flood recession and well-watered (soilsaturated) conditions (e.g. van Eck et al., 2004). The 20-week scenario was chosen to ensure a sufficiently long period to estimate the maximum duration of submergence tolerance for J. ingens seedlings. We also included two additional treatments: drought, and soil saturation, implemented for the full 20 weeks of the experiment, to determine the resilience of seedlings to prolonged drought, as populations of adult plants were observed to die back during a recent regional drought (Mayence et al., 2010), and to examine growth rates under favourable wet conditions, (i.e. saturated soil), as previous experimental research suggests that this can result in vigorous growth for adult plants (Mayence et al., 2010). Our experiment therefore contained six experimental treatments, including two non-flood treatments and four flood treatments (Fig. 2):  Treatment A: soil saturation;  Treatment B: drought (no watering, with soil in pots allowed to dry out);  Treatment C: 20 cm flood for 20 weeks;

270

L.M. Vivian et al. / Journal of Environmental Management 132 (2014) 268e277

Fig. 1. Invasion of Juncus ingens seedlings at (a) Steamer Plain (10/12/2012), establishing in a Pseudoraphis spinescens patch, and (b) Barmah Lake (26/10/2007). (c) Established mature stands of J. ingens surrounding a remnant patch of P. spinescens. Photo in (b) courtesy of Keith Ward.

 Treatment D: 20 cm flood for 12 weeks, followed by 8 weeks of ‘recovery’ well-soil-saturated conditions;  Treatment E: 50 cm flood for 20 weeks; and  Treatment F: 50 cm flood for 12 weeks, followed by 8 weeks of ‘recovery’ in soil-saturated conditions. For the flood treatments, the first 12 weeks of the experiment is referred to as Phase 1, with the subsequent 8 weeks referred to as Phase 2 (Fig. 2). 2.3. Seed and soil collection Experimental plants were grown from seed collected in Barmah Forest during February 2011 from mature J. ingens plants growing along a creek in Steamer Plain (lat: 35.944, long: 144.981), a ca.

260 ha treeless wetland with a large area of J. ingens infestation (DPI, 2009). Soil was collected from the same location from the upper 20 cm profile and was air dried, blended and crushed to pass through a 2 mm sieve. 2.4. Seedling growth To ensure that plants of similar heights were available in three different height classes for the experiment, we germinated and grew fifty J. ingens plants in soil-saturated conditions to determine average growth rates and the corresponding time required to establish the experimental plants. Seeds were germinated in a moist seed-raising mix, and after 18 days were transplanted into larger pots (8.5 cm diameter by 15 cm height) containing air dry soil, placed in trays containing ca. 2 cm of water in an unheated

L.M. Vivian et al. / Journal of Environmental Management 132 (2014) 268e277

2.5. Experimental design

Timeline of experiment (Weeks) Treatments

Non-flood

*1 2* 3 4 5* 6 7 8 9 10 11 12*

13 14 15 16 17 18 19 20^

A

Soil-saturation

B

Drought (unwatered)

treatments Flood

Phase 1

Phase 2

treatments C

20 cm flood 20 cm flood

D

Soil-saturation

E

50 cm flood 50 cm flood

F

Soil-saturation

Fig. 2. Experimental treatments and timeline of experiment, including timing of measurements. For the flood treatments (Treatments CeF), the experiment was split into two phases (Phase 1: Weeks 1e12, Phase 2: Weeks 13e20). In Phase 2, a subset of five plants from each height class in Treatments C and E were removed from flooding and placed into soil-saturated conditions in order to examine recovery following flooding (shaded cells), indicated by the dotted arrows. The remainder of the plants was left in the flood for the remaining eight weeks, indicated by the solid arrows. * ¼ At these points during the experiment (Day 1, Day 12, Day 34 and Day 84), a subset of plants were harvested from Treatments C and E for biomass measurements;^¼ At the end of the 20 week experiment, all surviving plants from all treatments were harvested for biomass and shoot morphology measurements. In addition, heights of the tallest shoot were measured on a subset of plants from each treatment at semiregular intervals throughout the experiment.

glasshouse, and watered daily. Plants took approximately 55 days to reach 10 cm (short height class), 62 days to reach 20 cm (medium) and 96 days to reach 40 cm (tall). These durations were used for germinating and establishing the three height classes of plants to be used in the experiment, with the aim of each height class reaching the appropriate height on September 7th 2012. Plants were germinated and established in the same conditions as described above. Growth rates were slower, perhaps due to the reduced light regime at a different time of year. Final mean heights ( standard deviation) of each height class by the start of the experiment were: tall ¼ 33.3 (8.3) cm; medium ¼ 16.9 (5.0) cm; short ¼ 11.7 (3.2) cm.

8 cm

3 cm

271

In total, 120 plants in each of the three height classes (n ¼ 360) were established and randomly assigned to one of the six treatments (Fig. 2): Treatment A: 10 plants per height class (n ¼ 30); Treatment B: 10 plants per height class (n ¼ 30); Treatments C and D combined: 50 plants per height class (n ¼ 150); and Treatments E and F combined: 50 plants per height class (n ¼ 150). At the end of Phase 1, a subset of five plants from each height class in the flood treatments were removed from flooding and placed in soilsaturated conditions for Phase 2 (Fig. 2). A greater number of plants were allocated to the flood treatments than the non-flood treatments to provide sufficient numbers for biomass sampling (see 2.8 below). Plants in the flood treatments were placed in a circular aboveground swimming pool (diameter ¼ 365 cm; height ¼ 76 cm). The pool was filled to 65 cm deep with water collected from the Murray River approximately 12 km downstream of Steamer Plain (lat: 36.019, long: 144.956), and transported to CSIRO Black Mountain Laboratories by truck in 1000 L containers. The pool was topped up to maintain constant depth. Plants in the 50 cm flood treatments (Treatments E and F) were placed on the pool floor, leaving 50 cm water above the top of the pot (considered to be the ‘soil surface’), while plants in the 20 cm flood treatments (Treatments C and D) were placed on 30 cm tall stands, leaving a depth of 20 cm above the top of the pot (Fig. 3). In this way, in Treatments E and F, short (12 cm), medium (17 cm) and tall (33 cm) seedling height classes were submerged by 38 cm, 33 cm and 17 cm respectively, while in Treatments C and D they were submerged by 8 cm, 3 cm and 13 cm respectively (i.e. the tall height class emerged 13 cm above the water level; Fig. 3). Plants were randomly arranged at least 60 cm distance from the pool edge to reduce shading. Water temperatures ranged between 9  C and 34  C, with average temperatures increasing into summer. Water quality remained good with pH of 6.9e7.1, conductivity ca. 70 mS/cm, and low turbidity. Plants in the non-flood treatments were randomly placed in trays on benches alongside the pool. Trays in Treatment A contained 2 cm of water for continuous soil saturation. Trays in

-13 cm

Water surface 38 cm

33 cm

17 cm

20 cm flood (Treatments C and D)

15 cm pot

15 cm pot

15 cm pot

30 cm stand

30 cm stand

30 cm stand

Medium cohort (17 cm)

Tall cohort (33 cm)

Short cohort (12 cm)

50 cm flood (Treatments E and F)

15 cm pot

15 cm pot

Short cohort (12 cm)

Medium cohort (17 cm)

Total water depth = 65 cm

15 cm pot

Tall cohort (33 cm)

Fig. 3. Experimental design for the flood treatments. Numbers in boxes indicate the average depth to which plants were submerged by at the commencement of the experiment, with tall plants in Treatments C and D actually emergent (see 2.5).

272

L.M. Vivian et al. / Journal of Environmental Management 132 (2014) 268e277

Treatment B were kept dry, with soil allowed to dry. All plants received natural precipitation (183 mm total) during the experiment; comparable to the total (187.5 mm) for the same period at Barmah Forest. 2.6. Plant survival rates and time to death All plants were monitored for time to senescence, defined as the number of days from the start of the experiment on which green material was no longer visible. After senescence, plants from Treatments BeF were removed and placed into soil-saturated conditions to allow any recovery, as indicated by any re-greening or production of new shoots. However, no senesced plants showed any signs of recovery after monitoring for at least three weeks; as such, time to senescence will henceforth be referred to as time to plant death. Survival rates were calculated as the percentage of plants remaining alive in each treatment at the end of the experiment (20 weeks). 2.7. Shoot heights, elongation rates and morphology The height of the tallest shoot was measured at semi-regular intervals during the experiment on all plants in Treatments A and B, as well as on ten randomly selected plants from each height class in Treatments CeF. Height data was used to estimate shoot elongation rates during the experiment, with average shoot elongation rates calculated for Phase 1 and Phase 2 separately for plants in the flood treatments (Treatments CeF). At the end of the experiment, the height of the tallest shoot, the number of shoots and diameter of the tallest shoot (measured at 10 cm from the base) were recorded for all remaining live plants (Fig. 2). 2.8. Biomass partitioning During Phase 1, biomass was measured at four time intervals on plants in Treatments C and E to compare the effects of different flood depths (50 cm and 20 cm) on plant growth over time. Twenty plants from each height class (excluding those used for measuring shoot elongation) were selected at random for measurement on Days 1 (pre-treatment), 12, 34 and 84 (Fig. 2). At each time interval, five plants were removed, root and shoot material was separated and washed of soil, dried at 60  C for 72 h and weighed. At the end of the experiment, 5 individuals from the remaining live plants in each height class/treatment combination were randomly selected and biomass harvested in the same manner (Fig. 2). 2.9. Statistical analyses Measurements taken at the conclusion of the experiment (after 20 weeks) on surviving plants (height and basal diameter of the tallest shoot, number of shoots, shoot biomass, root biomass and root to shoot ratio) were analysed using a two-factor analysis of variance (ANOVA), fitting height class with three levels (short, medium, tall) and treatment with five levels (Treatments A, C, D, E, F) as the two main factors, along with all interactions. Treatment B (drought) was not included as a factor level because no plants survived in this treatment. Because there were uneven numbers of plants left in each height class/treatment combination, the ANOVA was an unbalanced design; hence, Type 3 sums of squares were used in the estimation of each factor (Logan, 2010). The change in biomass allocation measured on plants in Treatments C and E at four time intervals during Phase 1 were analysed using three-factor ANOVA, fitting time with four levels (Day 1, 12,

34, 84), flood depth with two levels (50 cm or 20 cm), and height class with three levels (short, medium, tall) as the main factors, along with all interactions. The response variables were root biomass, shoot biomass, and root to shoot ratio. Post-hoc Tukey’s honestly significant differenced tests were used to test for differences among group means when significant differences were found. For all analyses, response variables were checked for homogeneity of variance and normality and transformed where necessary to meet the assumptions of ANOVA (Quinn and Keough, 2002). 3. Results 3.1. Plant survival after 20 weeks All J. ingens seedlings in Treatment B had a zero percent survival rate, with a time to death of approximately 60 days (Table 1). In contrast, all plants in Treatments A, D and F survived for 20 weeks. Tall plants in Treatment C also had a 100% survival rate; however, short and medium plants had survival rates of 60% and 20% respectively, with a time to death of over 100 days (Table 1). In Treatment E, survival rates were 76%, 8% and 4% for tall, medium and short plants respectively, with time to death of over 100 days (Table 1). 3.2. Shoot heights and elongation rates In Treatment A, the heights of the tallest shoots of seedlings in each height class converged after 20 weeks, with heights tripling in short plants, doubling in medium plants, and remaining relatively stable in tall plants (Fig. 4a). Heights of the tallest shoots of all plants in Treatment B remained relatively stable until plant death after approximately 60 days (Fig. 4a). For short and medium plants in the 20 cm flood treatments (Treatments C and D), which commenced the experiment with their shoots submerged below the water surface by 8 cm and 3 cm respectively, the tallest shoots emerged above the water after ca. 40 and 20 days respectively (Fig. 4b). After approximately 70 days, many shoots of medium plants snapped, resulting in a decline in the height of the tallest shoot. During Phase 2, short and medium plants in Treatments C and D continued to increase in height, whereas tall plants remained relatively stable (Fig. 4b). The tallest shoots of plants in the 50 cm flood in Treatments E and F remained suppressed below the water level for the duration of the experiment, despite an initial height increase (Fig. 4c). During Phase 2, any short and medium plants that still survived in Treatment E declined in height and died, whereas tall plants eventually recommenced shoot growth. In plants removed from the

Table 1 Plant survival rates and number of days until death for each treatment. Treatment A ¼ soil-saturation; Treatment B ¼ drought; Treatment C ¼ 20 cm flood for 20 weeks; Treatment D: 20 cm flood for 12 weeks followed by 8-week recovery; Treatment E ¼ 50 cm flood for 20 weeks; Treatment F ¼ 50 cm for 12 weeks followed by 8-week recovery. Dashes indicate that no plants died in the treatment; hence no time until death was calculated. Numbers in brackets are standard errors. Treatment

A B C D E F

Survival rate

Average no. days until death (1 s.e)

Tall

Medium

Short

Tall

Medium

Short

100% 0% 100% 100% 76% 100%

100% 0% 60% 100% 8% 100%

100% 0% 20% 100% 4% 100%

e 61 (1.1) e e 137 (0.9) e

e 64 (1.4) 117 (4.5) e 114 (1.9) e

e 63 (1.2) 104 (2.5) e 104 (1.1) e

L.M. Vivian et al. / Journal of Environmental Management 132 (2014) 268e277

273

Table 2 Average shoot elongation rates of plants (measured by the height of the tallest shoot) in each treatment, expressed as mm per week. Treatment A ¼ soil-saturation; Treatment B ¼ drought; Treatment C ¼ 20 cm flood for 20 weeks; Treatment D: 20 cm flood for 12 weeks followed by 8-week recovery; Treatment E ¼ 50 cm flood for 20 weeks; Treatment F ¼ 50 cm for 12 weeks followed by 8-week recovery. For the flood treatments, growth rates are given for Phase 1 and Phase 2 separately to compare between those plants remaining in the flood conditions (Treatments C and E) and those placed into soil-saturated conditions (Treatments D and F). Treatment

Tall

Medium

Short

Non-flood treatments (Weeks 1e20)

A B

3.6 Dead

10.1 Dead

12.8 Dead

Flood treatments: Phase 1 (Weeks 1e12)

C/D E/F

22.6 7.6

10.4 7.4

13.9 5.7

Flood treatments: Phase 2 (Weeks 13e20)

C D E F

2.2 6.4 Dead 3.4

6.6 42.5 Dead 2.6

2.2 75.3a Dead 2.5

a

Only one plant remained in this treatment after 96 days.

plants, the fastest elongation rates occurred during Phase 2 in Treatment D, after removal from the 20 cm flood and placement into wet conditions; however, these very rapid rates were recorded on just one and two plants respectively, with the remaining plants monitored for shoot elongation having previously died. The second fastest shoot elongation rates for short and medium plants were found in Treatments A and in Treatments C/D during Phase 1 (Table 2). In Treatment F during Phase 2, when plants were removed from the 50 cm flood and allowed to recover in soil-saturated conditions, shoot elongation rates were slow, with shoot height declining due to breakages, although plants remained alive. Shoot elongation rates were also relatively slow during Phase 2 in Treatment C (Table 2). 3.3. Biomass allocation during Phase 1 in the flood treatments

Fig. 4. Height of the tallest shoot in (a) Treatment A ¼ soil-saturation and Treatment B ¼ drought; (b) Treatment C ¼ 20 cm flood for 20 weeks and Treatment D: 20 cm flood for 12 weeks followed by 8-week recovery; and (c) Treatment E ¼ 50 cm flood for 20 weeks and Treatment F ¼ 50 cm for 12 weeks followed by 8-week recovery. Dotted horizontal lines in (b) and (c) indicate flood depths, and dotted vertical lines indicate the transition between Phase 1 and 2. A: Only one plant remained in this treatment after 96 days.

flood (Treatment F), the tallest shoots of plants in all height classes began to increase in height after being placed in soil-saturated conditions (Fig. 4c). Shoot elongation in tall plants was fastest during Phase 1 in the 20 cm flood (Treatments C/D) (Table 2). For short and medium

Biomass allocation across four time periods (during Phase 1) for plants in the two flood depths of 50 cm and 20 cm was strongly related to height class, with significant differences between height classes identified for shoot biomass (log-transformed; F2,96 ¼ 90.3; P < 0.001), root biomass (log-transformed; F2,96 ¼ 87.2; P < 0.001) and root to shoot ratio (log-transformed; F2,96 ¼ 41.0; P < 0.001) (Fig. 5). Post-hoc tests indicated that tall plants had the greatest shoot and root biomass, and the highest root to shoot ratio; whereas short plants exhibited the lowest root and shoot biomass and root to shoot ratio. There was little difference in biomass allocation between the two flood depths, with a higher allocation to shoots in the 20 cm flood evident only in tall plants (height class  treatment interaction: F2,96 ¼ 3.97; P ¼ 0.022; Fig. 5a). Allocation to root biomass increased after 84 days for tall plants only, in both the 20 cm and 50 cm floods (height class  day interaction: F6,96 ¼ 5.14; P < 0.001; Fig. 5b). Overall, plants allocated relatively more biomass to shoots over roots, with only tall and medium plants able to increase root to shoot ratio after 84 days (height class  day interaction: F6,96 ¼ 4.47; P < 0.001; Fig. 5c,f). 3.4. Shoot morphology and biomass allocation after 20 weeks At the end of the experiment, after 20 weeks, shoot height was significantly different between treatments (F4,112 ¼ 10.69; P < 0.001) and height class (F2,112 ¼ 13.86; P < 0.001) (Fig. 6a). Posthoc tests indicated that the tallest shoots were in Treatment C, with Treatment E resulting in the shortest plants, and that plants in the

274

L.M. Vivian et al. / Journal of Environmental Management 132 (2014) 268e277 Treatment C (20 cm flood) Treatment E (50 cm flood)

Shoot biomass (grams)

2.5

2 1.5 1 0.5 0

Root biomass (grams)

Day 1

Day 14

Day 34

Day 84

2.5

2

2

1.5

1.5

1

1

0.5

0.5

0

0

Day 1

Day 14

Day 34

Day 84

2.5

2.5

2.5

2

2

2

1.5

1.5

1.5

1

1

1

0.5

0.5

0.5

0

0 Day 1

Root:Shoot

2.5

Day 14

Day 34

Day 84

Day 14

Day 34

Day 84

2.5

2.5

2

2

2

1.5

1.5

1.5

1

1

1

0.5

0.5

0.5

0

Day 1

Day 14

Day 34

Day 84

Day 14

Day 34

Day 84

Day 1

Day 14

Day 34

Day 84

Day 1

Day 14

Day 34

Day 84

0 Day 1

2.5

0

Day 1

0

Day 1

Day 14

Day 34

Day 84

Fig. 5. Shoot and root biomass and root to shoot ratio in plants harvested at four time intervals during the experiment. Dotted lines in (c), (f) and (i) represents an equal ratio between shoot and root biomass, with any values above this line indicating relatively greater investment in root biomass.

short height class remained significantly shorter than the medium and tall plants. For total shoot numbers, a significant interaction term (F8,112 ¼ 3.15; P ¼ 0.003) indicated that Treatment E resulted in the lowest number of shoots, but only in medium and short plants, with tall plants having no significant difference in shoot number between treatments (Fig. 6b). Plants in treatment C had the thickest shoot diameters; however, a significant interaction term showed that again this was evident only in short and medium plants (F8,98 ¼ 2.05; P ¼ 0.05; Fig. 6c). After 20 weeks, shoot biomass was greater in Treatments A, C, and D than Treatments E and F (square-root transformed; F4,52 ¼ 6.31; P < 0.001; Fig. 6d), and was greater in the tall height class compared to the medium and short height class (F2,52 ¼ 7.31; P < 0.002). Greater root biomass was found in Treatments A, C, D, and F compared to Treatment E (log-transformed; F4,52 ¼ 13.18; P ¼ 0.001; Fig. 6e), and was also greater in the tall height class compared to the medium and short height class (F2,52 ¼ 9.33; P < 0.001). A significant interaction term indicated that root to shoot ratio was higher in Treatments A, C, D, F compared to Treatment E, but only for medium plants (square-root transformed; F8,52 ¼ 2.31; P ¼ 0.034; Fig. 6f).

ecosystems (Zedler, 2000). Removal of many invasive plants can be costly and resource-intensive, particularly for well-established populations (Hershner and Havens, 2008). Therefore, an effective approach could be to manipulate flow regimes to target invasive species while simultaneously providing the water requirements of other biota. Our experimental results suggest that at Barmah Forest, shallow environmental flows, which are typically used to extend flood durations for fish, birds and the native grass P. spinescens (Ward, 2005; MDBA, 2010), could potentially cause the mortality of invasive J. ingens seedlings. However, substantial mortality only occurred in medium and short seedlings in the two flood depths of 50 cm and 20 cm, and for taller seedlings in the 50 cm flood. The most conducive conditions for rapid shoot elongation was the 20 cm flood, with shoots of many submerged plants able to elongate through the water surface. These findings have important implications for management in Barmah Forest and other floodplains as it suggests that although some environmental flow scenarios may result in substantial plant mortality, there may be a narrow threshold between the conditions that result in the control of an invasive species, and conditions that result in providing favourable conditions for its growth. 4.1. Submergence strategy of J. ingens

4. Discussion The encroachment of invasive plant species is a widespread consequence of flow regime change in riverine and floodplain

Rapid shoot elongation is a common strategy for many emergent macrophytes to escape from complete submergence by reestablishing contact with air (Mauchamp et al., 2001; Voesenek

L.M. Vivian et al. / Journal of Environmental Management 132 (2014) 268e277

Treatment A

Treatment D

Treatment C

Treatment F

275

Treatment E

800

(a)

(b)

600 No. green shoots

Shoot height (mm)

700 500 400

300 200 100 0 Tall

Medium

Short

20 18 16 14 12 10 8 6 4 2 0 Tall

(c)

3.5

3

(d)

2 1

2.5 2 1.5 1 0.5

0 Tall

4

Medium

0

Short

Tall

(e)

2.5

3.5 3

Medium

Short

Medium

Short

(f)

2

2.5

Root : Shoot

Root biomass (g)

Short

3

4 Shoot biomass (g)

Shoot basal diameter (mm)

5

Medium

2 1.5

1.5

1

1 0.5

0.5 0 Tall

Medium

Short

0 Tall

Fig. 6. Shoot morphology and biomass allocation after 20 weeks: (a) shoot height; (b) number of green shoots; (c) shoot basal diameter; (d) root biomass; (e) root biomass; and (f) root:shoot ratio. No living plants were remaining in the drought treatment. Dotted line in (f) represents an equal ratio between shoot and root biomass, with any values above this line indicating relatively greater investment in root biomass. Treatment A ¼ soil-saturation; Treatment B ¼ drought; Treatment C ¼ 20 cm flood for 20 weeks; Treatment D: 20 cm flood for 12 weeks followed by 8-week recovery; Treatment E ¼ 50 cm flood for 20 weeks; Treatment F ¼ 50 cm for 12 weeks followed by 8-week recovery.

et al., 2006; Smith and Brock, 2007). Similarly, the rapid shoot elongation observed in J. ingens seedlings e particularly through a shallower flood of 20 cm e suggests an escape strategy, rather than a quiescent strategy of tolerating complete submergence. This investment into shoot elongation was reflected in the high shoot to root ratios in all plants during the earlier stages of the experiment. However, response strategies can depend on water depth, which determines the proportion of plant tissue submerged and the distance to the water surface (Mauchamp et al., 2001; Edwards et al., 2003). J. ingens seedlings in the 50 cm flood treatment were initially submerged by an average of 17 cm (tall plants), 33 cm (medium

plants) and 38 cm (short plants), and were unable to maintain positive growth after ca. 80 days, after which shoots began to break and overall plant height declined. In contrast, shoots of short and medium J. ingens seedlings in the 20 cm flood treatment, which were submerged by an average of just 3 cm and 8 cm respectively, were able to elongate to the water surface. Hence, it is not simply submergence per se that results in mortality of J. ingens seedlings, but submergence that overtops plant tissue by a depth sufficient to prevent shoots from reaching the surface, possibly by exhausting the energy and carbohydrate reserves needed to maintain shoot elongation (Voesenek et al., 2006). Plant height in relation to the

276

L.M. Vivian et al. / Journal of Environmental Management 132 (2014) 268e277

depth of flooding is therefore an important consideration, as it determines whether shoots are emergent or submerged (Edwards et al., 2003), and the depth to which shoots may be emerged. The height cohorts in our experiment were different ages; hence, their variable responses could also be a result of increased resilience with age, such as through the development of carbohydrate reserves and rhizomes (Mauchamp et al., 2001; Mauchamp and Methy, 2004). For example, we found that tall plants increased root to shoot ratios earlier in the experiment compared to medium and short plants, as allocation to shoots became less critical for survival, allowing plants to instead reallocate resources to below-ground reserves. Wetland plants removed from prolonged submergence can suffer from shoot mortality and breakage, particularly for shoots that have elongated rapidly and lack structural support (Mauchamp et al., 2001; Zhu et al., 2012). However, all J. ingens seedlings in the 50 cm flood were still alive after 12 weeks, and were able to continue shoot elongation when removed from flooding and placed in soilsaturated conditions, although some shoot breakages did occur, particularly in the taller plants. For many of the plants monitored for shoot elongation, height gains were achieved through the elongation of newer shoots overtaking the height of the older shoots, which had elongated underwater. Similarly, Edwards et al. (2003) found that shoots of the emergent wetland plant Eleocharis cellulosa (spikerush) that matured in deep water tended to die when plants were subsequently placed in shallower water, with a new set of shoots produced. This suggests that a switch from deep flooding to nonflooded soil-saturated conditions may be costly for plants, as they invest in the elongation of new shoots rather than the pre-existing older shoots. Nevertheless, in our experiment, plants continued to recover after removal from flooding, demonstrating that 12 weeks of a 50 cm flood, despite completely submerging the experimental plants, is inadequate to cause death. In contrast, after 20 weeks in the 50 cm flood treatment, survival rates of medium and short plants were reduced to less than 10%, and a quarter of tall plants died. The strategy of ‘escape’ from submergence via rapid shoot elongation is likely to be of less importance in prolonged deeper floods, where plants take a longer time to reach the water surface, rendering the costs of elongation too high to sustain (Voesenek et al., 2006; Colmer and Voesenek, 2009). For J. ingens, the decreased survival e particularly of shorter plants e in prolonged deeper flooding suggests that it is better able to survive in conditions of shallow floods, or deeper floods of short durations. 4.2. Potential for environmental flows to control J. ingens seedlings The spread of J. ingens in Barmah Forest is a challenging management issue because it is a locally-occurring native species. The objective therefore is not to completely remove J. ingens, but rather to control its spread, particularly in locations that were once dominated by P. spinescens (Mayence et al., 2010). The encroachment of J. ingens is thought to be caused by the reduction of deep flooding, allowing seedlings to establish (Chesterfield, 1986), and an increase in shallow summer flooding, effectively watering plants during a dry time of year (Mayence et al., 2010). Can environmental flows therefore be used to return deeper floods to prevent successful seedling establishment of J. ingens? Our results suggest that a flood of 50 cm for at least 20 weeks could cause >90% mortality in J. ingens seedlings shorter than 17 cm. Hence, an environmental flow used to maintain a 50 cm flood e providing the flood requirements for the ecologically important P. spinescens grasslands, for example e could potentially achieve an additional objective of killing shorter J. ingens seedlings. However, even after 20 weeks, 76% of tall plants (33 cm) persisted in the 50 cm flood, indicating that any population targeted for

control must be considerably shorter in height than the depth of the planned flood. Indeed, it may be extremely difficult to identify new invasions of J. ingens while plants remain as short as 17 cm, requiring regular on-ground monitoring for seedling establishment. For example, a report of seedling invasions in October 2007 was not made until plants were 60 cm tall (Ward, 2007) (Fig. 1b). The control of short J. ingens seedlings through submergence is likely to therefore require environmental flows that result in flood depths of at least 50 cm for 20 weeks. The total volume of water required to achieve such a scenario will depend on the location of the J. ingens seedling invasion, as wetlands in Barmah Forest commence to flood at different flow rates of the Murray River (Stokes et al., 2010). Submergence may be more feasible in lowlying wetlands which flood at a lower flow rate. As an example, a 50 cm flood on parts of Steamer Plain (Fig. 1a) equates to a flow of approximately 15,000 megalitres (ML) day1 at Yarrawonga Weir, upstream of Barmah Forest (authors’ unpublished data). The volume of environmental water needed to maintain this rate depends on the natural flow of the Murray River. For example, at a natural flow rate of 10,000 ML day1 (i.e. high channel volume, but below the threshold for flooding to occur in the forest), an extra ca. 5,000 ML day1 for 140 days is needed to maintain 50 cm of flooding on Steamer Plain for 20 weeks. This equates to a total allocation of 700 GL; more than the largest previously implemented environmental flows (512 GL during 2005e2006; and 428 GL during 2010e11 and 2011e12) (King et al., 2010; GB CMA, 2013), but within the scope of current watering scenarios (GB CMA, 2013). However, natural flows are variable and any drop below 10,000 ML day1 would require additional environmental water to maintain the required flood depth. In particular, channel volumes can be very low during dry periods, potentially requiring a much larger volume of environmental water than that allocated to Barmah Forest for environmental needs. We also emphasise that our example of a 50 cm flood would be effective only for J. ingens seedlings that are considerably shorter than the proposed flood depth. Hence, as seedlings grow taller, opportunities for submerging them for a sufficiently long period will decrease, and will require increasingly more environmental water. Furthermore, with mature plants developing rhizomes (Mayence et al., 2010), older plants may be more resistant to submergence. Chesterfield (1986) hypothesised that soil saturation provided favourable conditions for J. ingens growth. Indeed, our study showed that soil saturation provided good growth conditions, particularly for shorter plants. However, the 20 cm flood resulted in the fastest growth, thickest shoots, and tallest plants. Possibly the 20 cm flood provided warmer conditions compared to the air temperatures experienced by plants growing in conditions of soil saturation. Hence, any environmental flow that partially submerges plants e or only overtops the tallest shoots by a few centimetres e may actually increase growth rates and result in plants with taller and healthier shoots. For the purposes of controlling J. ingens seedlings, a conservative management strategy may be to prevent any watering of seedlings, rather than to risk a partial submergence event; indeed, drought conditions in our experiment resulted in the fastest plant mortality. Unfortunately, with environmental flows used to extend shallow floods in Barmah Forest, this could present a conflict between delivering the water requirements for desired floodplain species and worsening the spread of an undesired species in some areas. 4.3. Conclusions Using an experimental approach, our study suggests that environmental flows could potentially be used to reduce the spread of an invasive plant by targeting a life-stage (short seedlings) that is

L.M. Vivian et al. / Journal of Environmental Management 132 (2014) 268e277

particularly sensitive to prolonged submergence. However, the amount of environmental water available for maintaining a flood of sufficient depth and duration is likely to only be available if the flow rate of water through the Murray River is already relatively high. An important conclusion is that there may be a narrow threshold between the conditions that result in effective control of a species, and those that may worsen the invasion. For J. ingens seedlings, these thresholds occurred in plant height, flood depth and flood duration: a flood that is not quite deep enough to submerge plants, or a flood that is not quite sufficiently long to cause mortality, can become a flood that instead provides favourable growth conditions. Understanding whether these thresholds exist requires knowledge of species’ hydrological responses; our study demonstrates that a controlled experiment can be a useful approach to quantifying such floodeplant relationships. Finally, our case study highlights potential conflicts between environmental objectives when applying environmental flows to manage floodplain ecosystems. Whilst some species require the restoration of particular flow components that have been eliminated or reduced by river regulation, the provision of environmental flows to meet these needs may have adverse consequences for other parts of the ecosystem. For example, Howell and Benson (2000) suggested that environmental flows could disadvantage riparian vegetation by encouraging the spread of invasive plant species downstream. Unfortunately, invasive plants are wellestablished in many modified ecosystems, particularly floodplain habitats. As such, management strategies should be assessed for their potential to worsen the impact of any undesired species present. Acknowledgements We thank the Victorian Environmental Water Holder for permission to collect Murray River water to use in this experiment, and Parks Victoria for permission to work in the Barmah Forest. We also thank Matthew Lynch and Tom McLucas for their help with the water collection and transport. Daryl Nielsen and Carmel Pollino and two anonymous reviewers provided constructive feedback on the manuscript. We also thank Keith Ward for his advice in planning the experiment and for his feedback on the manuscript. References Abel, N., Roberts, J., Reid, J., Overton, I., O’Connell, D., Harvey, J., Bickford, S., 2006. Barmah Forest: A Review of its Values, Management Objectives, and Knowledge Base. Report to the Goulburn Broken Catchment Management Authority. CSIRO, Canberra. Bailey-Serres, J., Voesenek, L.A.C.J., 2008. Flooding stress: acclimations and genetic diversity. Annu. Rev. Plant Biol. 59, 313e339. Bren, L., O’Neill, I.C., Gibbs, N.L., 1988. Use of map analysis to elucidate flooding in an Australian riparian river red gum forest. Water Resour. Res. 24, 1152e1162. Bunn, S.E., Arthington, A.H., 2002. Basic principles and ecological consequences of altered flow regimes for aquatic biodiversity. Environ. Manage. 30, 492e507. Catford, J.A., Downes, B.J., Gippel, C.J., Vesk, P.A., 2011. Flow regulation reduces native plant cover and facilitates exotic invasion in riparian wetlands. J. Appl. Ecol. 48, 432e442. Chesterfield, E.A., 1986. Changes in the vegetation of the river red gum forests at Barmah, Victoria. Aust. For. 49, 4e15. Chong, J., Ladson, A.R., 2003. Analysis and management of unseasonal flooding in the BarmaheMillewa Forest, Australia. River Res. Appl. 19, 161e180. Colmer, T.D., Voesenek, L.A.C.J., 2009. Flooding tolerance: suites of plant traits in variable environments. Funct. Plant Biol. 36, 665e681. Cross, W.F., Baxter, C.V., Donner, K.C., Rosi-Marshall, E.J., Kennedy, T.A., Hall, R.O., Kelly, H.A.W., Rogers, R.S., 2011. Ecosystem ecology meets adaptive

277

management: food web response to a controlled flood on the Colorado River, Glen Canyon. Ecol. Appl. 21, 2016e2033. Dexter, B.D., Rose, H.J., Davies, N., 1986. River regulation and associated forest management problems in the River Murray red gum forests. Aust. For. 49, 16e27. DPI, 2009. Mapping of Historic Vegetation of the Barmah-Millewa Forest. Department of Primary Industries, Victoria. Dudgeon, D., Arthington, A.H., Gessner, M.O., Kawabata, Z.-I., Knowler, D.J., Leveque, C., Naiman, R.J., Soto, D., Stiassny, M.L.J., Sullivan, C.A., 2006. Freshwater biodiversity: importance, threats, status and conservation challenges. Biol. Rev. 81, 162e182. Edwards, A.L., Lee, D.W., Richards, J.H., 2003. Responses to a fluctuating environment: effects of water depth on growth and biomass allocation in Eleocharis cellulosa Torr. (Cyperaceae). Can. J. Bot. 81, 964e975. GB CMA, 2013. Barmah Forest Seasonal Watering Proposal 2013e2014. GoulburnBroken Catchment Management Authority, Shepparton. Hershner, C., Havens, K.J., 2008. Managing invasive aquatic plants in a changing system: strategic consideration of ecosystem services. Conserv. Biol. 22, 544e550. Howell, J., Benson, D., 2000. Predicting potential impacts of environmental flows on weedy riparian vegetation of the HawkesburyeNepean River, south-eastern Australia. Austral Ecol. 25, 463e475. Kercher, S.M., Zedler, J.B., 2004. Flood tolerance in wetland angiosperms: a comparison of invasive and noninvasive species. Aquat. Bot. 80, 89e102. King, A.J., Ward, K.W., O’Connor, P., Green, D., Tonkin, Z., Mahoney, J., 2010. Adaptive management of an environmental watering event to enhance native fish spawning and recruitment. Freshwater Biol. 55, 17e31. Kingsford, R.T., Auld, K.M., 2005. Waterbird breeding and environmental flow management in the Macquarie Marshes, arid Australia. River Res. Appl. 21, 187e200. Logan, M., 2010. Biostatistical Design and Analysis Using R. Wiley-Blackwell, West Sussex. Mauchamp, A., Blanch, S., Grillas, P., 2001. Effects of submergence on the growth of Phragmites australis seedlings. Aquat. Bot. 69, 147e164. Mauchamp, A., Methy, M., 2004. Submergence-induced damage of photosynthetic apparatus in Phragmites australis. Environ. Exp. Bot. 51, 227e235. Mayence, C.E., Marshall, D.J., Godfree, R., 2010. Hydrological and mechanical control for an invasive wetland plant, Juncus ingens, and implications for rehabilitating and managing Murray River floodplain wetlands, Australia. Wetlands Ecol. Manag. 18, 717e730. MDBA, 2010. Assessing Environmental Water Requirements. Chapter 10: BarmahMillewa Forest. Murray-Darling Basin Authority, Canberra. MDBA, 2012. Assessment of Environmental Water Requirements for the Proposed Basin Plan: BarmaheMillewa Forest. MDBA publication no: 16/12. MurrayDarling Basin Authority, Canberra. Millennium Ecosystem Assessment, 2005. Ecosystems and Human Well-being: Current State and Trends, vol. 1. Island Press, Washington. Naiman, R.J., Latterell, J.J., Pettit, N.E., Olden, J.D., 2008. Flow variability and the biophysical vitality of river ecosystems. Geoscience 340, 629e643. Poff, N.L., Zimmerman, J.K.H., 2010. Ecological responses to altered flow regimes: a literature review to inform the science and management of environmental flows. Freshwater Biol. 55, 194e205. Quinn, G.P., Keough, M.J., 2002. Experimental Design and Data Analysis for Biologists. Cambridge University Press, Cambridge. Smith, R.G.B., Brock, M.A., 2007. The ups and downs of life on the edge: the influence of water level fluctuations on biomass allocation in two contrasting aquatic plants. Plant Ecol. 188, 103e116. Stokes, K., Ward, K., Colloff, M.J., 2010. Alterations in flood frequency increase exotic and native species richness of understorey vegetation in a temperate floodplain eucalypt forest. Plant Ecol. 211, 219e233. The Council of Heads of Australasian Herbaria, 2013. Australia’s Virtual Herbarium. http://avh.chah.org.au. Tockner, K., Stanford, J.A., 2002. Riverine flood plains: present state and future trends. Environ. Conserv. 29, 308e330. van Eck, W.H.J.M., van de Steeg, H.M., Blom, C.W.P.M., de Kroon, H., 2004. Is tolerance to summer flooding correlated with distribution patterns in river floodplains? A comparative study of 20 terrestrial grassland species. Oikos 107, 393e405. Voesenek, L.A.C.J., Colmer, T.D., Pierik, R., Millenaar, F.F., Peeters, A.J.M., 2006. How plants cope with complete submergence. New Phytol. 170, 213e226. Ward, K., 1991. Investigation of the Flood Requirements of the Moira Grass Plains in Barmah Forest, Victoria. Department of Conservation and Environment, Benalla. Ward, K., 2005. Water management in the changing Barmah-Millewa wetlands. Proc. R. Soc. Victoria 117, 77e84. Ward, K., 2007. Proposed Water Management Requirements Downstream of Yarrawonga for Ecological Issues at Barmah-Millewa Forest. Goulburn-Broken Catchment Management Authority. Zedler, J.B., 2000. Progress in wetland restoration ecology. Trends Ecol. Evol. 15, 402e407. Zhu, G., Li, W., Zhang, M., Wang, S., 2012. Adaptation of submerged macrophytes to both water depth and flood intensity as revealed by their mechanical resistance. Hydrobiologia 696, 77e93.