Predicting diversity versus community composition of aquatic plants at the river scale

Available online at www.sciencedirect.com

Aquatic Botany 88 (2008) 338–346 www.elsevier.com/locate/aquabot

Predicting diversity versus community composition of aquatic plants at the river scale Kristina Makkay a,1, Frances R. Pick a,*, Lynn Gillespie b a

Ottawa-Carleton Institute of Biology, University of Ottawa, 30 Marie Curie, Ottawa K1N 6N5, Canada Research Division, Canadian Museum of Nature, P.O. Box 3443, Station D, Ottawa K1P 6P4, Canada

b

Received 4 October 2006; received in revised form 21 December 2007; accepted 29 December 2007 Available online 6 January 2008

Abstract We tested the relative importance of physical versus chemical factors in explaining aquatic plant species diversity and community composition within a temperate lowland river. A total of 38 macrophyte species were identified at 33 sites along the 104 km length of the Rideau River, a National Heritage River of Canada. Species richness ranged from 0 to 15 species per site, and Shannon diversity from 0 to 2.98. Macrophyte species richness and Shannon diversity were significantly related to the physical characteristics of sites. For Shannon diversity, 77% of the increase was explained by an increase in sediment organic content and a decrease in water velocity. For species richness, 70% of the increase was explained by the latter factors in addition to an increase in the littoral zone (0–2 m depth contour) width and planktonic chlorophyll concentrations. River water chemistry did not explain any observed variation in either Shannon diversity or species richness in this moderately enriched system. In contrast to species richness, the physical and chemical variables measured failed to explain variation in community composition. Cluster analysis did not reveal any grouping of species into distinct communities. Canonical correlation analysis showed that environmental variables had minimal effect on the distribution of most species, with only floating-leaved species responding to water velocity. We conclude that physical factors can predict species diversity at the within river scale but not the species composition at a given site, underlying the need to preserve the geomorphological diversity of rivers to maintain plant diversity. # 2008 Elsevier B.V. All rights reserved. Keywords: Aquatic plant community; Shannon diversity; Species richness; Lowland river; Stochastic process

1. Introduction Ecologists have long been interested in what controls plant diversity. A number of theories have been put forward to explain what controls the diversity both at a global and local scale. On a local scale, Huston (1979) proposed a twodimensional ‘‘dynamic equilibrium’’ model that predicts species richness will be highest at an intermediate point on the productivity–stress gradient and at intermediate levels of disturbance. This model postulates that extreme stress and disturbance eliminate all but the most tolerant species, whereas minimal stress and high productivity results in species being eliminated through competitive exclusion. The effects of disturbance offset the effects of productivity. While this model * Corresponding author. Tel.: +1 613 562 5800x6364; fax: +1 613 562 5486. E-mail address: [email protected] (F.R. Pick). 1 Present address: Environment Canada, 351 Joseph Blvd., Gatineau, Quebec K1A 0H3, Canada. 0304-3770/$ – see front matter # 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.aquabot.2007.12.011

was developed based on terrestrial systems, it may also apply to aquatic systems as well (Murphy, 2002). Aquatic plant (macrophyte) species richness has been found to be related to factors correlated with productivity and stress such as light attenuation (Willby et al., 2001), standing crop (Dodson et al., 2000), nutrient concentrations and trophic status (Rørslett, 1991; Jeppesen et al., 2000) as well as with disturbance regimes affecting biomass removal such as water velocity (Nilsson, 1987), boat traffic (Willby et al., 2001), water level fluctuations (Rørslett, 1991), and flood frequency (Barrat-Segretain and Amoros, 1996; Riis and Biggs, 2003), although these relationships are not consistently unimodal in shape. Species richness has also been found to increase with habitat area across lakes (e.g. Dodson et al., 2000; Dahlgren and Ehrle´n, 2005). Plant community composition has also long been of interest to ecologists: ecosystems and biomes are described in terms of dominant plant communities. Most community composition models have been developed for animal species (e.g. Diamond, 1975) and later extended to terrestrial and wetland plants. There

K. Makkay et al. / Aquatic Botany 88 (2008) 338–346

are a growing number of studies pertaining to aquatic plants in lakes, although a consensus has yet to emerge on which variables are most important in affecting composition. Some studies on lakes showed trophic state and nutrient concentration to have a strong effect on species composition (Srivasta et al., 1995; Vestergaard and Sand-Jensen, 2000) while others showed no relationship (Pip, 1979; Jackson and Charles, 1988). A similar debate surrounds pH (Pip, 1979; Jackson and Charles, 1988; Srivasta et al., 1995). Other influences include conductivity (Toivonen and Huttunen, 1995) and alkalinity (Srivasta et al., 1995; Vestergaard and Sand-Jensen, 2000). In streams and rivers, community composition was affected by alkalinity (Riis et al., 2000), conductivity, and nutrient concentrations (Murphy et al., 2003) as in lakes, but other factors include the downstream distance from source and spatial connectivity (Demars and Harper, 2005), stream size (Riis et al., 2000), velocity (Riis and Biggs, 2003), calcium and magnesium ion concentrations (Ali et al., 1995; Murphy et al., 2003), substrate characteristics (French and Chambers, 1996), and flood frequency (Barrat-Segretain and Amoros, 1996; Riis and Biggs, 2003). Most ecological studies of aquatic systems have been regional scale comparisons between lakes or small streams (Kalff, 2002). There is a need for studies on single medium to large rivers, rivers in populated areas, and rivers that are modified or dammed. These rivers, while not in a ‘‘natural state’’, nonetheless make up a significant part of the landscape, and are usually managed based on a series of local scale decisions. If management decisions are to be made on scientific principles, some understanding of local scale patterns and processes is necessary. The purpose of this study was to test the relative importance of physical and chemical habitat characteristics on aquatic plant (macrophyte) diversity and community composition at the within—river (local) scale. At this scale, characteristics that were expected to have the greatest influence on diversity were nutrient concentrations, water velocity, and littoral area. We also hypothesized that community composition could be predicted using the same variables that could predict diversity. 2. Materials and methods 2.1. Study area The Rideau River is a lake-fed medium size river (about equivalent to fifth order) located in eastern Ontario, Canada. It is 104 km in length, has an average width of 250 m, and drains an area of 3830 km2. Once a shipping route, the Rideau is now a National Heritage River used as a recreational waterway. Water levels are regulated for optimal navigation through a series of locks and channels and fluctuations rarely exceed 10 cm during the navigation (and growing) season (Parks Canada, unpublished data). The maximum depth of the river is 10 m, with the deeper portions limited to the navigational channel, but much of the river is less than 2 m in depth. The river varies from mesotrophic to eutrophic, with total phosphorus (TP) levels ranging between 7 and 39 mg L1 with

339

the higher concentrations observed in the downstream half of the river (Basu and Pick, 1997). Nitrogen to phosphorus ratios are between 19 and 28, suggesting that the system is primarily phosphorus limited. Water clarity allows for macrophyte development throughout the river, but particularly within the littoral zone (0–2 m depth contour). Habitat types vary along the length of the river. Areas of high water velocity are found in the main river channel adjacent to locks or downstream of dams built beside the locks. Adjacent land use varies from source to mouth. The most undisturbed portions of the river are the first 30 km, which are predominantly forested with some cottage development. Water velocity is slow and the littoral shelf wide, with shorelines characterized by cattail (Typha spp.) marshes or forested swamps. From 30 to 75 km downstream the land use is mostly agricultural, with a strip of riparian vegetation being usually present between fields and the river. The last 30 km is mostly privately owned residential land characterized by reinforced banks and boat docks. The urban portion of the river within the capital city of Ottawa is mostly surrounded by public parks on both sides of the river. This portion of the river is lacking in littoral variety and tends to have higher water velocity. There are no major industries in the watershed. 2.2. Sampling methods Thirty-three sites were sampled along the Rideau River during July and August 2000 and July 2001. Sites were chosen to represent the range of conditions on the Rideau River in terms of channel morphology, distance along the river, water velocity, river width, and adjacent land use. Sites were first selected from bathymetric maps, and adjusted depending on field conditions. For the purposes of this study, aquatic macrophytes included vascular plants, and colonial algae in the family Characeae. Filamentous algae (e.g. Cladophora, Spirogyra), which were often associated with macrophyte beds, were not included. At each site, sampling and measurements were made in six 1 m2 quadrats aligned in a belt transect perpendicular to the shore, at approximate depths of 0.5 m, 0.8 m, 1.0 m, 1.2 m, 1.5 m, and 2.0 m. If the depth did not reach 2.0 m, quadrat 6 was established at the deepest point in the channel, and the other quadrats were distributed evenly between that point and the shore and at the predetermined depths where possible. The exact location of the transect was randomly selected, as was the placement of the quadrats. The distance to shore from furthest quadrat (2.0 m depth) was determined to be an adequate reflection of the littoral width. Preliminary sampling from the Rideau River in 1999 showed six quadrats were adequate in order to sample most of the species at a site, and that beyond 2 m depth, species richness declined to one or two species (usually Myriophyllum spp.). At each quadrat, species were listed and cover estimates made for each species using a Braun-Blanquet scale from one to five (Braun-Blanquet, 1932) with cover values as follows—1: up to 5%, 2: 5–25%, 3: 25–50%, 4: 50–75%, 5: 75–100%. Species were observed either from a boat using a viewing tube,

340

K. Makkay et al. / Aquatic Botany 88 (2008) 338–346

or from the water using a mask and snorkel. Samples were taken of unidentified macrophyte species for later identification. Water velocity was calculated at each quadrat using the time it took an orange to travel 1 m. If after 2.0 min the orange had not yet moved 1.0 m, the velocity was considered below detection and was entered as 0.005 m s1. The distance from each quadrat to shore was measured directly. Bank height and the water depth at shore were also measured. A sediment sample was collected from the top 5 cm of the surface at each quadrat. Sediments were analyzed for total organic content by measuring loss on ignition at 500 8C for 2 h. Underwater slope was calculated from depth and distance measurements using the following formula solving for u: tan u ¼

ðdepth2  depth1 Þ ðdistance2  distance1 Þ

where depth2 and distance2 were depth and distance at the deeper adjacent quadrat and depth1 and distance1 were the depth and distance at the shallower adjacent quadrat. At quadrats 3 and 6 (1.0 m and 2.0 m) pH was measured using a Hydrolab MiniSonde1 4a. Water samples were taken at approximately 0.5 m depth and were analyzed by the City of Ottawa Surface Water Quality Laboratories using standard methods for total phosphorus (TP) and reactive phosphate (RP), total Kjeldahl nitrogen (TKN) and ammonium (NH4+). Chlorophyll a (Chl a) was extracted using DMSO and 90% acetone (Burnison, 1980) and concentrations calculated from absorbance (Jeffery and Humphrey, 1975). Chl a concentration is an estimate of phytoplankton biomass, as well as an indicator of light attenuation. Shoreline features approximately 300 m upstream and downstream from the site and adjacent land use within 1– 2 km from the site were noted and then assigned to one of four categories. For adjacent land use the categories were— agricultural, forested, residential and urban. For shoreline features the categories were lawn (mowed to the water’s edge), forest, wetland, and undisturbed riparian strip (usually a strip of 2–5 m back from the river’s edge that is let to grow undisturbed). 2.3. Statistical analyses For site level analyses, data were summarized for each site as follows: each species was tallied according to the number of quadrats they were found in, giving a frequency measure between 1 and 6; species richness was the total number of species found at the site; velocity and sediment organic content were averaged over the six quadrats; width of littoral zone was simply the distance to shore at quadrat 6 (usually at 2 m depth); water quality measurements were averaged over the two samples. Vascular plant species were identified using Crow and Hellquist (2000), and charophytes using Wood (1967). Plant species that were primarily wetland species were noted but excluded from species richness counts when they occasionally appeared in shallow quadrats. These species were: Decodon

verticillatus, Scirpus pungens, Phalaris arundinacea, and Typha angustifolia. Linear regression analysis was used to analyze the relationship between species richness and each of the environmental variables (water velocity, organic content, slope, littoral width, Chl a, pH, NH4+, TKN, RP, and TP), and between Shannon diversity and the aforementioned variables. Log transformation was done on all independent variables except total TKN and pH in order to obtain a normal distribution of the data and a linear relationship between the dependent and independent variables. Variation in species richness according to land use and adjacent shoreline features was analyzed using analysis of variance (ANOVA). All continuous independent variables were combined in a multiple regression in order to build predictive models for species richness and Shannon diversity. Backwards and forwards stepwise selection using alpha to enter/leave at 0.15 were compared to determine the optimal model for species richness and Shannon diversity. The relationships between continuous independent variables (log-transformed) were examined using a correlation matrix in order to eliminate problems with multicolinearity. A cluster analysis connecting sites using Euclidean distances and average linkages was used to examine the patterns in species assemblages. Canonical correspondence analysis (CCA) was used to investigate the relationship between environmental variables and plant community composition, and a correspondence analysis (CA) was used to examine the patterns in species composition between quadrats independent of the environmental variables. Like other multivariate techniques, these tests require a sample to variable ratio of at least 20 in order for canonical variates and correlates to be reliable (McGarigal et al., 2000). Because of this limitation, detailed species and species–environment relationships could only be investigated at the 1 m2 quadrat scale, that is to say, using each quadrat as a separate data point. The environmental variables that had been measured at each quadrat were sediment organic content, velocity, slope, and distance from shore. In addition, depth, location downstream (in km) and sampling date (based on the number of days from 29 June) were also added into the analysis to determine their influence on community composition. All variables except depth, location downstream, and date were log-transformed to approximate normal distributions. Only species found in more than 5% and less than 90% of the quadrats were included in the analyses. The statistical significance of the relationship between the species data and both the first canonical axis and the entire set of axes were tested using 199 random Monte Carlo permutations. It was recognised that the data were not entirely independent since the quadrats were distributed among 33 sites, rather than being randomly placed throughout the river. One effect of this would be spatial autocorrelation which results in the real degrees of freedom being somewhat less than the apparent degrees of freedom. The second effect was that there would be equal weight put on within-site relationships as between-site relationships. As a result, the effects of environmental variables

K. Makkay et al. / Aquatic Botany 88 (2008) 338–346

would be under-emphasized and the analysis may be somewhat conservative. Analysis of variance showed that the F ratios of among-site to within-site variation were highest for sediment organic content (F = 23.36) and lowest for depth (F = 0.36). The other variables were between 2.97 and 7.23. Thus, the effects identified were more likely to be a problem for detecting effects of sediment organic content and less for the other variables. Before conducting the CCA, principal components analysis (PCA) was undertaken to determine the extent to which environmental variables were correlated. Since the pair-wise correlations among environmental variables were relatively low, the PCA did not result in much dimensionality reduction. Depth and distance showed similar loadings (as expected) but not to the extent to justify excluding one of these variables from the analysis. All the original variables were therefore retained. CANOCO for Windows 4.0 (ter Braak and Sˇmilauer, 1998) was used for the CCA and PCA. All other statistical procedures were done with Systat 10 (SSPS Inc., 2000).

3.1. Site characteristics The physical characteristics of the sites varied over several orders of magnitude with respect to littoral slope and width, average water velocity and sediment organic content (Table 1). The chemical characteristics of the sites varied as well but generally over less than one order of magnitude. RP ranged from the analytical detection limit (0.5 mg L1) to as high as 46 mg L1 and TP from 13 mg L1 to 51 mg L1. Nutrient concentrations were within the range reported in previous studies of the Rideau River (Basu and Pick, 1997). A total of 38 species of aquatic plants were observed in the quadrats (Table 2). The most common species found were tapegrass (Vallisneria americana, 29 of 33 sites), Canada waterweed (Elodea canadensis, 25 sites), star duckweed (Lemna trisulca, 24 sites), and coontail (Ceratophyllum demersum, 23 sites). An additional 13 species were observed in the river, but were not in the quadrats themselves. Species richness per site ranged from 0 to 15, and Shannon diversity from 0 to 2.98. Up to 11 species were found in a 1 m2 Table 1 Physical and chemical variables of the Rideau River measured at 33 sites

Water velocity (m s ) Sediment organic content (%) Littoral width (m) Littoral slope (degrees) pH Chlorophyll a (mg L1) Reactive phosphate (mg L1) Total phosphorus (mg L1) Ammonium (NH4+) (mg L1) Total Kjeldahl nitrogen (mg L1)

quadrat, with the maximum richness found at depths from 0.8 m to 1.5 m. 3.2. Univariate analyses of species diversity and environmental variables Linear regression showed that both Shannon diversity and species richness increased significantly with an increase in littoral width and organic content, and decreased with an increase in water velocity and cross-sectional slope (Fig. 1, Table 3). Littoral width and cross-sectional slope were highly correlated (r = 0.83; p < 0.001), which was expected. Only littoral width was then used in multiple regression analyses as it was the better predictor. The model best describing Shannon diversity (H) using backwards and forwards selection included (log-transformed) organic content and water velocity (r2 = 0.77; p < 0.001) (Table 4a) and was as follows: H ¼ 1:65 þ 0:28log OC  0:29log WV For species richness (S), the best model using backwards and forwards selection included (log-transformed) per cent sediment organic content (OC), water velocity (WV), littoral width (LW), and Chl a (CA), (r2 = 0.70; p < 0.001) (Table 4b) and was as follows:

3. Results

1

341

Mean  S.D.

Range

0.062  0.084 14.1  13.5 57.7  63.2 3.56  3.56 8.21  0.45 5.06  3.54 14  13 34  19 33  18 664  87

0.005–0.385 0.5–44.1 7.0–272.0 0.29–15.95 7.13–8.89 1.22–20.40 0.5–46 13–51 9.0–76 492–825

S ¼ 1:04 þ 1:16 log OC  1:05 log WV þ 1:35 log LW þ 1:34 log CA Manual selection showed that water velocity and organic content alone accounted for most of the variation (r2 = 0.65; p < 0.001). No significant relationships were found between Shannon diversity or species richness and pH, ammonium, TKN, reactive phosphorus, and total phosphorus (Table 3). Dissolved nutrient levels increased with distance downstream with the highest levels found in the urban and suburban areas in the last 30 km of the river, but this did not appear to have a direct impact on species diversity. 3.3. Multivariate analysis of community composition The cluster analysis showed a high degree of chaining. The two main groups that emerged were one group that was comprised of six of the seven most common species, and another group containing all other species (Fig. 2). In Canonical Correlation Analysis (Fig. 3), the first two axes explained 59% of the species–environment relations with eigenvalues of 0.179 and 0.161, respectively. The environmental variables that accounted for most of these species– environment relationships, as indicated by their correlation to the canonical axes, were sediment organic content (r = 0.839 with axis 1), and depth (r = 0.718 with axis 2). However, these relationships accounted for a very small proportion of the overall variation in species composition. The first four canonical axes only explained 10.2% of the variation in species composition, with the first axis explaining only 3.8%.

342

K. Makkay et al. / Aquatic Botany 88 (2008) 338–346

Table 2 List of macrophyte species within the Rideau River, and the number of sites and quadrats in which each species were found Species

Family

Sites found

Quadrats found

Vouchers

Vallisneria americana L. Elodea canadensis Michx. Lemna trisulca L. Ceratophyllum demersum L. Myriophyllum sibiricum Komarov Myriophyllum spicatum L. Potamogeton zosteriformis Fern. Zosterella dubia (Jacq.) Small Nymphaea odorata Aiton Najas flexilis (Willd.) Rostkov and Schmidt Ranunculus aquatilis L. Potamogeton pusillus L. Butomus umbellatus L. Potamogeton crispus L. Potamogeton richardsonii (A. Benn.) Rydb. Spirodela polyrhiza (L.) Schleiden Stuckenia pectinata (L.) Bo¨rnor Wolffia borealis (Engelm.) Landolt Potamogeton friesii Rupr. Hydrocharis morsus-ranae L. Nuphar variegata Durand. Potamogeton robbinsii Oakes Megalodonta beckii (Torr.) Greene Alisma gramineum C.C. Gmel. Chara globularis Thuill. Nitella flexilis (L.) C. Agardh Lemna minor L. Chara vulgaris L. Sagittaria latifolia Pursh. Sparganium eurycarpum Engelm. Typha angustifolia L. Chara braunii C.C. Gmel. Potamogeton foliosus Raf. Scirpus pungens Vahl Decodon verticillatus (L.) Elliott Phalaris arundinacea L. Potamogeton amplifolius Tuckerman Utricularia vulgaris L.

Hydrocharitaceae Hydrocharitaceae Lemnaceae Ceratophyllaceae Haloragaceae Haloragaceae Potamogetonaceae Pontederiaceae Nymphaeaceae Najadaceae Ranunculaceae Potamogetonaceae Butomaceae Potamogetonaceae Potamogetonaceae Lemnaceae Potamogetonaceae Lemnaceae Potamogetonaceae Hydrocharitaceae Nymphaeaceae Potamogetonaceae Asteraceae Alismataceae Characeae Characeae Lemnaceae Characeae Alismataceae Sparganiaceae Typhaceae Characeae Potamogetonaceae Cyperaceae Lythraceae Poaceae Potamogetonaceae Lentibulariaceae

29 25 24 23 17 17 17 17 13 11 11 10 9 9 8 8 7 7 7 5 5 5 4 3 3 2 2 2 2 2 2 1 1 1 1 1 1 1

91 74 91 85 47 47 42 39 17 27 27 23 14 11 17 12 12 12 9 8 8 8 7 5 3 6 3 2 2 2 2 2 2 2 1 1 1 1

00–036 02–004 00–043 00–042 00–031 00–050 00–035 00–013 00–007 00–018 01–002 00–015 00–016 00–023 00–019

00–017 00–052 00–056 00–012 00–013 00–025 00–008 00–048 00–027 00–045 00–051

00–041

A total of 198 quadrats in 33 sites were sampled. Voucher specimens (numbers as indicated) have been deposited in the Canadian National Herbarium (CAN), Aylmer, Que´bec.

For this reason, the majority of the species were located close to the origin of the biplot, which indicates that very little of the variation in species composition is accounted for by the environmental variables measured. These results were also corroborated using a series of Mantel tests that determined there was little correlation between sites with similar species composition and sites with similar environmental characteristics. There was some influence of the environmental variables on the floating species Wolffia borealis, and Spirodela polyrhyza, and to a lesser extent on rooted floating-leafed species, Nymphaea odorata, and Nuphar variegatum. The floating species clustered together somewhat opposite the water velocity vector, indicating an association with slow-moving water. The one emergent species that appears in the analysis, Butomus umbellatus, as well as Stuckenia pectinata is shown to be characteristic of shallow water close to shore in the downstream portion of the river. The Monte Carlo test showed

the relationship between the species data and the canonical axes was statistically significant ( p = 0.005). 4. Discussion 4.1. Macrophyte species diversity The dynamic equilibrium concept of Huston (1979) is partially supported by the results seen in the Rideau River. Variation in macrophyte diversity in the Rideau River appears primarily associated with sediment organic content and water velocity, with some influence of littoral width and planktonic Chl a. The increase in diversity with organic content is likely a reflection of the relationship between productivity and diversity. Fine organic particulate matter tends to accumulate in areas of low current and slope (Sand-Jensen, 1998), and be high in nutrients. As nutrient uptake in rooted macrophytes comes mostly, if not entirely, from the sediments rather than

K. Makkay et al. / Aquatic Botany 88 (2008) 338–346

343

Fig. 1. Regression of macrophyte Shannon diversity against (a) percent sediment organic content (b) water velocity (c) littoral width (distance to shore from a depth of 2 m), and (d) littoral slope in the Rideau River.

from the water column (Chambers et al., 1989), productivity would be expected to increase with sediment organic content. The relationship in the Rideau River showed a logarithmic increase in species richness, rather than the unimodal relationship predicted by the dynamic equilibrium concept, however, as maximum sediment organic content was about 44% (Table 2), this suggests that levels of productivity may not Table 3 Results of regression of species richness and Shannon diversity on independent variables Variable

Species richness p r2

Water velocity Littoral width Organic content Littoral slope pH Chl a Reactive phosphate Total phosphorus Total nitrogen (TKN) NH4+

0.518 0.436 0.616 0.256

<0.001 <0.001 <0.001 0.003 0.278 0.228 0.125 0.309 0.644 0.864

Shannon diversity r2 p 0.617 0.338 0.708 0.137

All independent variables except pH and TKN were log-transformed.

<0.001 <0.001 <0.001 0.037 0.439 0.473 0.210 0.502 0.584 0.997

reach the point where competitive exclusion would start to reduce diversity. Field observations also did not reveal any areas of dense monoculture that are typical where there is strong competitive exclusion. Another explanation of the diversity—organic content relationship is that areas accumulating fine sediment might also accumulate a greater diversity of seeds and propagules. Water velocity also had a strong negative influence on species diversity. The relationship was also log-linear, rather than unimodal. While there is no doubt that sites of high water velocity present a stress that makes it unlikely for a wide variety of species to become established, extremely low levels of water velocity simply create lake-like conditions, in which case other factors would be more responsible for determining diversity such as water chemistry, light penetration, exposure, and interactions with other biota. It has been speculated that the inhospitable conditions at higher water velocity are due to organic material and fine sediment having been washed away (Nilsson, 1987). Others have observed the negative correlation of water velocity with organic material and fine sediment (Sand-Jensen, 1998; Chambers et al., 1991), but in this study organic content and water velocity both made a significant contribution to model fit.

344

K. Makkay et al. / Aquatic Botany 88 (2008) 338–346

Table 4 Multiple regression models for (a) Shannon diversity and (b) species richness, using backwards stepwise selection with p to enter/leave = 0.10 Effect (a)Shannon diversity Constant Velocity Organic content (b)Species richness Constant Velocity Littoral width Organic content Chl a

Co-efficient

S.E.

Standard co-efficient

Tolerance

T

p (2 tail)

1.652 0.286 0.283

0.514 0.088 0.059

0.000 0.386 0.572

N/A 0.513 0.513

3.213 3.262 4.829

0.003 0.003 0.001

1.037 1.149 1.354 1.156 1.339

4.221 0.591 0.622 0.441 0.775

0.000 0.255 0.283 0.421 0.179

N/A 0.465 0.566 0.372 0.893

0.246 1.776 2.176 2.623 1.728

0.808 0.087 0.038 0.014 0.095

Similar results were found using forward selection.

This suggests there are other influences of water velocity beyond its effect on sediment organic content, an observation that has been confirmed through experimentation (Chambers et al., 1991). Likely causes might be damage under high flows from breakage, uprooting or scouring which would prevent species with vulnerable growth forms from becoming established. The species found at high velocity sites of the Rideau River tended to be flexible species such as V. americana, and Myriophyllum spp., which have been shown to be wellsuited to such conditions (Sand-Jensen, 2003). The third best predictor of macrophyte diversity was littoral width, which was measured as the distance from shore to a depth of 2 m. A larger littoral area means more habitat area suitable for aquatic macrophyte colonization is available for a given length of river (also more microhabitat types). A similar

Fig. 2. Cluster analysis of species connecting the sites surveyed on the Rideau River using Euclidean distances and median linkages. BUM, Butomus umbellatus; CDE, Ceratophyllum demersum; ELA, Elodea canadensis; HMO, Hydrocharis morsus-ranae; LTR, Lemna trisulca; MSI, Myriophyllum sibiricum; MSP, Myriophyllum spicatum; NFL, Najas flexilis; NOD, Nymphaea odorata; NVA, Nuphar variegatum; PCR, Potamogeton crispus; SPE, Stuckenia pectinata; PPU, Potamogeton pusillus; PRI, Potamogeton richardsonii; PZO, Potamogeton zosteriformus; RAQ, Ranunculus aquatilis; SPO, Spirodela polyrhiza; VAM, Vallisneria americana; WBO, Wolffia borealis; ZDU, Zosterella dubia.

effect was observed by Vestergaard and Sand-Jensen (2000) in lakes, while in contrast, Dahlgren and Ehrle´n (2005) found lake area to be a better predictor of species richness than littoral width. In the Rideau River, large littoral zones also tended to be adjacent to wetlands and areas with few shoreline modifications, so there could be an additional effect of the adjacent land area. Overall, the Rideau River had a higher total macrophyte species richness than that which would be predicted from lakes of comparable surface area when applying the predictive model of Dodson et al. (2000). This is probably the result of its greater relative littoral area and may be the case for rivers in general. Interestingly, the Rio Parana in Brazil appears to contain a similar total macrophyte species richness (50 species, Murphy et al., 2003) as the Rideau River, despite having a much larger surface area. While planktonic Chl a did not show a significant relationship with macrophyte diversity when compared

Fig. 3. Canonical correlation analysis biplot illustrating the main pattern of variation in the species assemblage in relation to environmental variables. See Fig. 2 for species codes. (*) submerged species; (& ) floating and floatingleaved species; (^) emergent species.

K. Makkay et al. / Aquatic Botany 88 (2008) 338–346

directly, it did show a significant positive relationship when water velocity, sediment organic content, and littoral width were held constant. Usually macrophyte abundance decreases with Chl a due to phytoplankton inhibiting macrophyte growth by competing for nutrients and light (Scheffer et al., 1993). There have, however, been other studies showing Chl a to increase in the presence of macrophytes (Basu et al., 2000). In this study, the positive correlation between macrophyte diversity and Chl a may simply be an indirect effect of the overall productivity of the site. The relationships between diversity, adjacent land use, and shoreline features are complicated by the fact that there were also relationships between adjacent land use and riparian features with some of the other independent variables, namely sediment organic content and littoral width. While there is little doubt that adjacent land use has a strong effect on aquatic ecosystems (Moss, 1998), it appears from this study that macrophyte diversity responds more closely to river morphology than what occurs on the shoreline or the larger catchment. It is possible that a stronger correlation might have been found had a more systematic approach been used to categorise land use and shoreline features (e.g. GIS information) instead of a qualitative survey. Water nutrient levels showed no significant relationships with species diversity. Riis et al. (2000) also found that nutrients were not useful in predicting macrophytes diversity across and within Danish streams. This is in contrast to the results of Murphy et al. (2003) on the Rio Parana who showed that water column TP had a significant negative effect on macrophytes diversity, along with sediment concentrations of iron and calcium. However, in both these studies water column nutrient levels were one to two orders of magnitude higher than those found in the Rideau. Another reason for the lack of any relationship with nutrients may have been the relatively small variation in water chemistry between river sites. 4.2. Macrophyte species composition While biodiversity could be predicted from simple physical features, these same features were not useful in predicting community composition, even taking into consideration that the CCA test was likely to be conservative for reasons explained previously. Floating and floating-leafed species were the only ones to demonstrate any environmental preference, which was for areas with low water velocity, consistent with existing research (Crow and Hellquist, 2000). Aside from this, there was little evidence of defined communities. There was also no apparent effect of downstream position on community composition, likely due to the Rideau’s relative consistency in width and depth. Even independently of the environmental variables there was little evidence of distinct communities. The groups defined in the cluster analysis predominantly reflected diversity—one cluster was composed of species that were only found at the more diverse sites, whereas the other comprised species that were nearly ubiquitous. Apparent random patterns of colonization have also been observed in other recent macrophyte studies

345

(Rolan and Maltchik, 2006; Evardsen and Okland, 2006), supporting neutral models of community that suggest species are functionally similar (Bell, 2001; Hubbell, 2001). Overall, the results indicate the Rideau River supports essentially one macrophyte community, which will appear wherever habitat conditions suit macrophyte growth and will become more diverse if habitat conditions improve. This suggests that stochastic processes likely determine plant community composition, whereas physical factors ultimately constrain the number of species at a given site. This underlines the need to preserve the geomorphological diversity of rivers as much as water quality in order to maintain plant diversity, particularly littoral marshes vulnerable to shoreline modifications. In river systems, the neutral model of community assemblages may be sufficient to explain community structure. Acknowledgements We thank Jessica Strike, Marc Demers, and Marsha Ostrovsky, from the Department of Biology, University of Ottawa, and Micheline Beaulieu-Bouchard from the Canadian Museum of Nature for assistance in sampling the Rideau River, and C.S. Findlay (University of Ottawa) for statistical advice. Funding from the EJLB Foundation to the Rideau River Biodiversity Project (L. Gillespie) and NSERC Discovery grants (F. Pick) is gratefully acknowledged. References Ali, M.M., Hamad, A.M., Springer, I.V., Murphy, K.J., 1995. Environmental factors affecting submerged macrophyte communities in regulated water bodies in Egypt. Arch. Hyrobiol. 133, 107–128. Barrat-Segretain, M.H., Amoros, C., 1996. Recovery of riverine vegetation after experimental disturbance: a field test of the patch dynamics concept. Hydrobiologia 321, 53–68. Basu, B.K., Pick, F.R., 1997. Phytoplankton and zooplankton development in a lowland, temperate river. J. Plankton Res. 19, 237–253. Basu, B., Kalff, J., Pinel-Alloul, B., 2000. The influence of macrophyte beds on plankton communities and their export from fluvial lakes in the St.Lawrence River. Freshwater Biol. 45, 373–382. Bell, G., 2001. Neutral macroecology. Science 293, 2413–2418. Braun-Blanquet, J. In: Fuller, G.D., Conrad, H.S., (Trans.) 1932. Plant Sociology: The Study of Plant Communities, McGraw-Hill Book Company, New York Burnison, B.K., 1980. Modified dimethyl sulfoxide (DMSO) extraction for chlorophyll analysis of phytoplankton. Can. J. Fish. Aquat. Sci. 52, 804–815. Chambers, P.A., Prepas, E.E., Bothwell, M.L., Hamilton, H.R., 1989. Roots versus shoots in nutrient uptake by aquatic macrophytes in flowing waters. Can. J. Fish. Aquat. Sci. 46, 435–439. Chambers, P.A., Prepas, E.E., Hamilton, H.R., Bothwell, M.L., 1991. Current velocity and its effect on aquatic macrophytes in flowing waters. Ecol. Appl. 1, 249–257. Crow, G.E., Hellquist, C.B., 2000. Aquatic and Wetland Plants of Northeastern North America. University of Wisconsin Press. Madison, WI, USA. Dahlgren, J.P., Ehrle´n, J., 2005. Distribution patterns of vascular plants in lakes—the role of metapopulation dynamics. Ecography 28, 49–58. Demars, B.O.L., Harper, D.M., 2005. Distribution of aquatic vascular plants in lowland rivers: separating the effects of local environmental conditions, longitudinal connectivity and river basin isolation. Freshwater Biol. 50, 418–437.

346

K. Makkay et al. / Aquatic Botany 88 (2008) 338–346

Diamond, J.M., 1975. Assembly of Species of Communities. In: Cody, M.L., Diamond, J.M. (Eds.), Ecology and Evolution of Communities. Harvard University Press, Cambridge, MA, pp. 342–444. Dodson, S.I., Arnott, S.E., Cottingham, K.L., 2000. The relationship in lake communities between primary productivity and species richness. Ecology 81, 2662–2679. Evardsen, A., Okland, R., 2006. Variation in plant species composition in and adjacent to 64 ponds in SE Norwegian agricultural landscapes. Aquat. Bot. 85, 92–102. French, T., Chambers, P., 1996. Habitat partitioning in riverine macrophyte communities. Freshwater Biol. 36, 509–520. Hubbell, S.P., 2001. The Unified Neutral Theory of Biodiversity and Biogeography. Princeton University Press, Princeton, NJ, USA. Huston, M., 1979. A general hypothesis of species diversity. Am. Nat. 113, 81–101. Jackson, S.T., Charles, D.F., 1988. Aquatic macrophytes in Adirondack lakes: patterns of species composition in relation to environment. Can. J. Bot. 66, 1449–1460. Jeffery, S.W., Humphrey, G.F., 1975. New spectrophotometric equations for determining chlorophylls a, b, c1 and c2 in higher plants, algae, and natural phytoplankton. Biochem. Physiol. Pfl. 167, 191–194. Jeppesen, E., Jensen, J.P., Sondergaard, M., Lauridsen, T., Landkildehus, F., 2000. Trophic structure, species richness and biodiversity in Danish lakes: changes along a phosphorus gradient. Freshwater Biol. 45, 201–218. Kalff, J., 2002. Limnology: inland water ecosystems. Prentice-Hall Inc., Upper Saddle River, NJ. McGarigal, K., Cushman, S., Stafford, S., 2000. Multivariate Statistics for Wildlife and Ecology Research. Springer, New York. Moss, B., 1998. Ecology of fresh waters: man and medium, past to future. Blackwell Science, Oxford. Murphy, K.J., 2002. Plant communities and plant diversity in softwater lakes of Northern Europe. Aquat. Bot. 73 (4), 287–324. Murphy, K.J., Dickenson, G., Thomaz, S.M., Bini, L.M., Dick, K., Greaves, K., Kennedy, M.P., Livingstone, S., McFerran, H., Milne, J.M., Oldroyd, J., Wingfield, R.A., 2003. Aquatic plant communities and predictors of diversity in a sub-tropical river floodplain: the upper Rio Parana. Braz. Aquat. Bot. 77, 257–276.

Nilsson, C., 1987. Distribution of stream edge variation along a gradient of current velocity. J. Ecol. 57, 513–522. Pip, E., 1979. Survey of the ecology of submerged aquatic macrophytes in central Canada. Aquat. Bot. 7, 339–357. Riis, T., Biggs, B.J.F., 2003. Hydrologic and hydraulic control of macrophyte establishment and performance in streams. Limnol. Oceanogr. 48, 1488–1497. Riis, T., Sand-Jensen, K., Vestergaard, O., 2000. Plant communities in lowland streams: species composition and environmental factors. Aquat. Bot. 66, 255–272. Rolan, A.S., Maltchik, L., 2006. Environmental factors as predictors of aquatic macrophyte richness and composition in wetlands of southern Brazil. Hydrobiologia 556, 221–231. Rørslett, B., 1991. Principal determinants of aquatic macrophyte richness in northern European lakes. Aquat. Bot. 39, 173–193. Sand-Jensen, K., 1998. Influence of submerged macrophytes on sediment composition and near bed flow in lowland streams. Freshwater Biol. 39, 663–679. Sand-Jensen, K., 2003. Drag and reconfiguration of freshwater macrophytes. Freshwater Biol. 48, 271–283. Scheffer, M., Hosper, S.H., Meijer, M.L., Moss, B., Jeppesen, E., 1993. Alternative equilibria in shallow lakes. Trends Ecol. Evol. 8, 275–279. Srivasta, D., Staicer, C., Freedman, B., 1995. Aquatic vegetation of Nova Scotia lakes differing in acidity and trophic status. Aquat. Bot. 51, 181–196. SSPS Inc., 2000. Systat for Windows, Version 10. SPSS Inc., Chicago, IL, USA. ter Braak, C.J.F., Sˇmilauer P., 1998. CANOCO Reference Manual and User’s Guide to CANOCO for Windows: Software for Canonical Community Ordination. Microcomputer Power, Ithaca, New York, U.S.A. Toivonen, H., Huttunen, P., 1995. Aquatic macrophytes and ecological gradients in 57 small lakes in southern Finland. Aquat. Bot. 51, 197–221. Vestergaard, O., Sand-Jensen, K., 2000. Alkalinity and trophic state regulate aquatic plant distribution in Danish lakes. Aquat. Bot. 67, 85–107. Willby, N.J., Pygott, J.R., Eaton, J.W., 2001. Inter-relationships between standing crop, biodiversity and trait attributes of hydrophytic vegetation in artificial waterways. Freshwater Biol. 46, 883–902. Wood, R.D., 1967. Charophytes of North America. University of Rhode Island.