A novel bacterial community index to assess ... - Wiley Online Library

0 downloads 0 Views 1MB Size Report
Centre for Microbial Innovation, The University of Auckland, Auckland, New ... Research, Investigations and Monitoring Unit, Auckland Council, Auckland, New ...
Freshwater Biology (2015) 60, 1988–2002

doi:10.1111/fwb.12625

A novel bacterial community index to assess stream ecological health KELVIN E. M. LAU*†, VIDYA J. WASHINGTON*†, VICKY FAN*‡, MARTIN W. NEALE*§, GAVIN LEAR*†, JAMES CURRAN¶ AND GILLIAN D. LEWIS*† *School of Biological Sciences, The University of Auckland, Auckland, New Zealand † Centre for Microbial Innovation, The University of Auckland, Auckland, New Zealand ‡ Bioinformatics Institute, The University of Auckland, Auckland, New Zealand § Research, Investigations and Monitoring Unit, Auckland Council, Auckland, New Zealand ¶ Department of Statistics, The University of Auckland, Auckland, New Zealand

SUMMARY 1. In stream ecosystems, bacterial communities play an important role in nutrient and energy cycling processes as they are among the most numerous and active organisms at the basal trophic level of the stream food web. Bacterial communities in stream biofilms have been shown to correlate well with different catchment land use and therefore provide an opportunity for the development of a novel ecological indicator of stream ecosystem health. 2. In this study, a bacterial community index (BCI) model was developed and validated using a national data set of biofilm bacterial community profiles collected from 223 streams across seven geographical regions in New Zealand. The six-component BCI model was generated using the partial least squares regression method to associate the multivariate bacterial community profile with the macroinvertebrate community index, which is a well-established indicator of stream health. 3. Despite strong regional clustering of the bacterial community profiles, the BCI was indicative of the level of disturbance in the catchment, as shown by significant correlations with a wide range of independent indicators of water quality, macroinvertebrate community data, ecosystem functioning and catchment land-use data. The BCI was able to explain 35% of the variation in a multi-metric index incorporating ten common ecological parameters, suggesting that the stream bacterial communities could provide useful information about the ecosystem integrity. 4. The BCI provides a novel ecosystem assessment tool, which can be used to complement existing stream health measures in the management of anthropogenic impacts on freshwater streams and rivers. Keywords: automated ribosomal intergenic spacer analysis, ecosystem health, indicator, partial least squares regression, stream biofilm

Introduction Biological indicators are often used as a surrogate measure to estimate the ecological status of entire ecosystems (Metcalfe, 1989; Holt & Miller, 2011). Indicators used to assess the biodiversity of freshwater ecosystems have largely been confined to macro-organisms including benthic macroinvertebrates (Boothroyd & Stark, 2000; Klemm et al., 2002), parasites (Shea et al., 2012), fishes (Harris, 1995; Seilheimer & Chow-Fraser, 2006;

Hermoso et al., 2010), birds (Morrison, 1986), amphibians (Micacchion, 2002), periphyton (Hill et al., 2000) and plants (Dennison et al., 1993; Whitton & Kelly, 1995). In particular, macroinvertebrates are the most commonly used indicators of ecological health for streams and rivers (Rosenberg & Resh, 1993), with widespread application for monitoring programmes in New Zealand (Stark & Maxted, 2007), Australia (Keen, 2001), Europe (European Union, 2000) and North America (Barbour et al., 1999). To enable a simple and direct comparison of

Correspondence: Gillian D. Lewis, School of Biological Sciences, The University of Auckland, Auckland Mail Centre, Private Bag 92019, Auckland 1142, New Zealand. E-mail: [email protected]

1988

© 2015 John Wiley & Sons Ltd

Bacterial community index ecosystem health between streams, multivariate macroinvertebrate community composition data are commonly summarised into an index, such as the Macroinvertebrate Community Index (MCI) in New Zealand (Stark, 1985), the Biological Monitoring Working Party (BMWP) score in the United Kingdom (Hawkes, 1998), or the Stream Invertebrate Grade Number Average Level (SIGNAL) index in Australia (Chessman, 1995). Bacteria have short life cycles and are among the most numerous and active organisms in the basal trophic level of the stream food web (Pernthaler, 2013). Consequently, bacterial communities have the potential to offer an early indication of shifts in the ecosystem before organisms at higher trophic levels (i.e. macro-organisms) respond. As a result, several studies have highlighted the need to include microbial variables in predictive models of ecosystem change (Sarmento et al., 2010; Treseder et al., 2012). Microbial indicators based on algal communities (Taylor et al., 2007; Lavoie et al., 2014) and specific faecal bacteria (Ashbolt, Grabow & Snozzi, 2001; Nnane, Ebdon & Taylor, 2011) are already used to measure water quality and ecosystem health, but an index of ecological integrity based on stream bacterial communities has yet to be developed for use in biomonitoring programmes. Most bacteria occurring in freshwater ecosystems are attached to surfaces as biofilms (Gessey, Mutch & Costerton, 1978; Edwards, Meyer & Findlay, 1990; Hall-Stoodley, Costerton & Stoodley, 2004). Benthic biofilms provide several advantages in the context of systematic environmental monitoring since they can be sampled easily with minimal disturbance to the sampling site (Lear et al., 2012). Biofilm communities are relatively sessile and therefore are more likely to be indicative of local conditions when compared to more mobile and transient organisms in the water column. They are also amendable to analysis by relatively inexpensive and sensitive molecular fingerprinting techniques (Fisher & Triplett, 1999; Danovaro et al., 2006), which improves the feasibility of collecting replicate samples that are essential for robust biological studies (Prosser, 2010). Compared to molecular methods, the routine morphological identification of a large number of physical specimens, such as macroinvertebrates and diatoms, is often relatively more labour intensive and potentially prone to error (Haase et al., 2010). The diversity of bacterial communities within stream biofilms could provide a more sensitive indicator of ecological health in highly impacted streams when compared to current methods that rely on macro-organisms, such as fish and macroinvertebrates (Lear et al., 2009). The abundance and diversity of macro-organisms in © 2015 John Wiley & Sons Ltd, Freshwater Biology, 60, 1988–2002

1989

impacted sites generally decreases as sensitive species are lost (Violin et al., 2011; Scott, Steward & Stober, 1986). This change in biodiversity is the basis for several ecological health models including the Index of Biotic Integrity (Karr, 1981) and EPT Index (Lenat, 1988), which associates a decrease in the abundance or number of sensitive taxa observed with an increase in ecosystem deterioration. However, low diversity and the absence of sensitive indicator species in highly impacted sites reduce the functionality and accuracy of such indices (Ostermiller & Hawkins, 2004). This creates a paradox, whereby the locations where management interventions are undertaken in an effort to improve ecological health are the locations where detecting any improvement is challenging because of the limitations of existing biological indicators. In contrast, bacterial community diversity in impacted sites has been found to remain relatively high, with anthropogenic impact resulting in changes in bacterial community composition, but not a significant loss of diversity (Lear et al., 2011; Reis et al., 2013; Washington et al., 2013). The high bacterial diversity that is maintained even in highly degraded streams provides the potential for a sensitive biological indicator that can operate across the full range of impacted sites. Evidence is accumulating that bacterial communities embedded in stream biofilms are suitable as biological indicators (Burns & Ryder, 2001; Lyautey et al., 2005; Lear & Lewis, 2009; Ancion, Lear & Lewis, 2010; Wang et al., 2011). A recent conceptual model describing the structure of stream biofilm bacterial communities proposed that there are specific impact-related and naturalstate bacteria, which could be used as biological indicators of stream ecosystem health (Washington et al., 2013). However, the lack of models or algorithms that can translate complex multivariate bacterial community descriptions into a standardised quantitative index has hampered the application of stream biofilm bacterial communities as an ecological indicator. Here, we build on the current descriptions of correlations between microbial communities and levels of anthropogenic impacts in an attempt to generate a biological index as proof of concept that bacterial communities can be applied to the assessment and management of streams. The objective of this work is to develop and test a model to summarise multivariate bacterial community data obtained using molecular methods into a quantitative bacterial community index (BCI) that can be applied consistently throughout New Zealand. We report on the use of mathematical modelling to account for both natural and anthropogenic factors influencing the biofilm bacterial communities into the BCI.

1990

K. E. M. Lau et al.

Methods Study sites The bacterial community index model was developed using a New Zealand national data set (Washington et al., 2013) collected from streams located in seven geographical regions in New Zealand, with a latitude range of 36°200 40.24″S to 44°540 37.81 S and a longitude range of 169°350 03.62″E to 177°350 0362″E. The geographical regions included Auckland, Waikato, Hawkes Bay, Manawatu-Wanganui and Wellington regions in the North Island, as well as Tasman and Canterbury regions in the South Island of New Zealand. This study utilised bacterial and macroinvertebrate community data from 223 sites across New Zealand, collected during the austral summer of 2010. The study sites form part of the New Zealand routine state of the environment monitoring programme (Bunnik et al., 2007) and encompass sites with a range of natural environmental conditions and anthropogenic impacts. Landcover information obtained from the New Zealand Land Cover Database (LCDB3) was derived from satellite imagery (Landcare Research, 2012) and was used to categorise the catchment land use as urban, rural, exotic forest or native forest. For many streams in New Zealand, there is often a lack of information on the reference condition due to the widespread development of landscapes for human use (Clapcott et al., 2011). Therefore, pristine or near pristine sites with more than 90% native forest land cover in the catchment were accepted as reference sites (Death & Collier, 2010), while sites with urban cover were assumed to be highly impacted. A quantitative indication of the anthropogenic impact on the catchment was provided by the land-use stress (LUS) scores, which were calculated from the weighted sum of upstream land cover as described by Collier (2008). LUS scores range from 0 for sites with only native forest in the catchment to 300 for sites in entirely urban catchments. A range of water quality, macroinvertebrate, fish and ecosystem function metrics, as well as a multi-metric index (MMI) of ecological integrity for the sites used in this study (Table 1), were obtained from the New Zealand Department of Conservation and Cawthron Institute (Clapcott & Goodwin, 2010; Clapcott et al., 2011). The MMI was derived from a combination of ten metrics of ecological integrity, weighted according to the strength and precision of the responses of each metric category to four land-use stressors (native vegetation loss in catchment, stream nitrogen load, impervious

Table 1 Means (and ranges) for water quality, macroinvertebrate, fish, ecosystem function and land-use metrics for study sites used to validate the BCI model. Ecological Indicator Water quality Clarity DRP* NOx* Macroinvertebrates Cycle*

MCI* ITR Fish F.Introduced*

F.IBI*

Mean (Range)

Black disc clarity Dissolved reactive phosphorous (ppm) Nitrates–nitrites (ppm) Invertebrate taxa reproducing only once (%) Macroinvertebrate community index Invertebrate taxonomic richness Percentage of introduced fish taxa (%) Fish index of biological integrity

Ecosystem processes ER* Ecosystem respiration (gO2 m2 d1) del15N* Delta 15N of primary consumers (&) Cotton* Cellulose decomposition potential (k dd1) GPP* Gross primary production (gO2 m2 d1) Land use LUSS Land-use stress score Multi-metric index MMI Multi-metric index predicted from the category weighted response of 10 metrics of ecological integrity to four land-use stressors FENZ river Multi-metric index pressures incorporating a series of pressures in the catchment

1.64 (0.50–6.79) 0.043 (0.010–0.124) 0.600 (0.047–4.268) 73.0 (33.6–91.3)

103.1 (42.6–156) 16.9 (9.7–38.1)

0.375 (0.006–2.222)

40.0 (15.0–56.5)

7.50 (3.31–14.38) 6.33 (1.08–11.88) 0.0027 (0.0003–0.0076)

2.91 (0.42–18.31)

129 (0–298) 0.74 (0.43–1.21)

0.28 (0.00–0.88)

*Ten metrics included in the calculation of the MMI.

cover and surface water allocation) as detailed in Clapcott & Goodwin (2010). In addition, the Freshwater Ecosystems of New Zealand (FENZ) river pressure index (Leathwick et al., 2010) was also included in the comparison because it did not include macroinvertebrate community data, thereby avoiding a cyclical validation of the BCI which was modelled on the MCI. The FENZ river pressure index was generated by incorporating a © 2015 John Wiley & Sons Ltd, Freshwater Biology, 60, 1988–2002

Bacterial community index series of anthropogenic pressures in the catchment, including impervious cover, nitrogen concentration, reduction in indigenous vegetation cover, barriers to water flows, point discharges and introduced fish species, as described by Leathwick & Julian (2007).

Biofilm bacterial communities Stream biofilm samples were collected according to a standardised protocol (Lear & Lewis, 2009) from the upper surface of submerged rocks using dehydrated sterile Speci-Sponges (VWR International Ltd., Arlington Heights, IL, USA). For each stream, replicate biofilm samples were collected from up to five rocks selected randomly within a 10 m reach. Streams with 119)

Good (MCI 100–119)

Fair (MCI 80–99)

Poor (MCI 119) Good 826 953 560 169 (BCI 100–119) Fair 65 486 628 536 (BCI 80–99) Poor 0 0 60 94 (BCI