Modelling spatial distribution of hard bottom benthic communities and their functional response to environmental parameters
Bandelj, V.; Curiel, D.; Lek, S.; Rismondo, A.; Solidoro, C. (2009). Modelling spatial distribution of hard bottom benthic communities and their functional response to environmental parameters. Ecol. Model. 220(21): 2838-2850. https://dx.doi.org/10.1016/j.ecolmodel.2009.04.024
In: Ecological Modelling. Elsevier: Amsterdam; Lausanne; New York; Oxford; Shannon; Tokyo. ISSN 0304-3800; e-ISSN 1872-7026, more
|  |
| Keywords |
Aquatic communities > Benthos
Benthic communities
Clustering
Environmental conditions
Environmental factors > Abiotic factors
Macrobenthos
Modelling
Neural networks
Spatial distribution
Zonation (ecological)
MED, Italy, Veneto, Venice Lagoon [Marine Regions]
Marine/Coastal |
| Author keywords |
Macrobenthos; Artificial neural networks; Fuzzy k-means; Zonation;Lagoon of Venice |
| Authors | | Top |
- Bandelj, V.
- Curiel, D.
- Lek, S.
|
- Rismondo, A.
- Solidoro, C.
|
|
| Abstract |
In this study we analyzed and modelled spatial distribution of hard bottom benthic communities in the Lagoon of Venice, and used the model to derive functional response of these communities to changing environmental conditions. Three communities were identified by using a fuzzy clustering algorithm to experimental observations (macrophytobenthos and macrozoobenthos data), and interpreted from a biological perspective by considering their indicator and dominant taxa. The results showed that the lagoon can be divided in areas dominated by a ‘marine’, an ‘intermediate’ and a ‘confined’ benthic community. Some differences in the extension of the three communities between the two years of monitoring were observed and related to significant differences in environmental parameters. The relationships between benthic communities’ membership grades and a set of water quality, hydrodynamics and sediment composition parameters were then modelled by means of an artificial neural network model, using a back-propagation algorithm. Furthermore, the model was used to assess the relative contribution of predictors on membership grades of different communities, and to understand the relationships between cluster membership grades and predictors with highest relative contribution. Several responses of benthic communities to environmental parameters were non-linear and complex, which confirmed the advantage of using neural networks instead of other, more traditional statistical methods. The benthic community showed mainly ‘marine’ characteristics with low temperature, silt percentage in sediments and POC concentration in water. The ‘confined’ characteristics were pronounced in areas of low PO4 and high DOP concentrations in water, low water energy and on wooden substrates. The ‘intermediate’ community was that predicted worst and was related mainly to high water salinity and temperatures around 16 °C. Finally, the model was used as a tool for the detection of anomalies, i.e., cases in which the predicted community was different from that observed. The anomalies were inspected and related to possible causes, such as a not optimal substrate, very low-bottom depth, complex local morphology, human disturbances, and slow adaptation of communities to changing abiotic characteristic. |
|
