Fish Habitat Associations

"Machine learning (ML) techniques have become important to support decision making in management and conservation of freshwater aquatic ecosystems. Given the large number of ML techniques and to improve the understanding of ML utility in ecology, it is necessary to perform comparative studies of these techniques as a preparatory analysis for future model applications. The objectives of this study were (i) to compare the reliability and ecological relevance of two predictive models for fish richness, based on the techniques of artificial neural networks (ANN) and random forests (RF) and (ii) to evaluate the conformity in terms of selected important variables between the two modelling approaches. The effectiveness of the models were evaluated using three performance metrics: the determination coefficient (R2), the mean squared error (MSE) and the adjusted determination coefficient (R2adj) and both models were developed using a k-fold crossvalidation procedure. According to the results, both techniques had similar validation performance (R2 = 68% for RF and R2 = 66% for ANN).
Although the two methods selected different subsets of input variables, both models demonstrated high ecological relevance for the conservation of native fish in the Mediterranean region. Moreover, this work shows how the use of different modelling methods can assist the critical analysis of predictions at a catchment scale. Copyright ONEMA, 2013. "

Seagrass beds and mangroves are distributed in tropical regions all over the world and considered important nurseries for coral reef fish species. Studies in the Caribbean have shown that seagrass beds and mangroves are the most important nurseries for juvenile coral reef fish, and when juveniles become adults they move to the coral reefs. Coral reef fish in the Indo-Pacific region do not show such a consistent pattern. In this region it is not clear to what extent coral reef fish use available habitats in a non-reef – coral reef gradient and what the influence is of the presence of non-reef habitats as a nursery on coral reef fish populations on coral reefs. The present study was carried out on four islands along the coast of Tanzania and on one island in the Comoros archipelago in the western Indian Ocean. Using a point count visual census method during the day, the length, abundance and distribution of 76 commercially important coral reef fish species were studied. In total, 2279 quadrats were surveyed at six different non-reef habitats, on coral reefs adjacent to seagrass beds and mangroves (reef sg-mg), on coral reefs at a greater distance to seagrass beds and mangroves (reef far) and on coral reefs in the Comoros archipelago completely lacking seagrass beds and mangroves (reef Comoros). Results showed that shallow seagrass beds did harbour much higher juvenile fish densities than mangroves, deep seagrass beds, intertidal flats, algal beds and coral reefs. This suggests that shallow seagrass beds are the most important nurseries for juvenile coral reef fish. Furthermore, all 76 observed species were categorised in six species groups: sg residents, sg-reef transients, generalists, reef generalists, reef residents and rare species, based on their juvenile and adult distribution patterns between non-reef and reef habitats. Two species groups, sg-reef transients and reef generalists, showed possible migrations between non-reef and reef habitats, but only sg-reef transients, consisted of two species, showed a possible ontogenetic migration from seagrass beds towards the coral reef adjacent to non-reef habitats. This suggests that the nursery hypothesis affected only two species and all other species can use alternative habitats. On reef Comoros the six species groups were also present, but the number of species found on reef Comoros was much lower than on reef far. On reef far three species groups and on reef Comoros four species groups showed significant higher adult densities on reef sg-mg than on reef far or reef Comoros. This suggests that on the scale of an entire island, non-reef habitats have an impact on species groups. The presence of non-reef habitats influences coral reef fish densities on reef sg-mg. At species level, 22 species of the 50 observed species on reef far and 5 species of the 19 observed species on reef Comoros showed significantly higher adult densities on reef sg-mg than on reef far or reef Comoros. This indicates that these species have an advantage of nearby non-reef habitats as a juvenile habitat, but they can sustain on reef far or reef Comoros. On reef far 28 species and on reef Comoros 14 species showed no effect of the absence of non-reef habitats on adult fish densities on the coral reef. It is suggested that these species are able to use the coral reef as an alternative nursery and do not depend on non-reef habitats. They can possibly spend their entire life cycle on the reef and are to some extent self-sustaining.

As changes in landuse and the demand for water accelerate, regulators and resource managers are increasingly asked to evaluate water allocation against the need for protection of in-stream habitat. In the United States, only a small number of river basins have the longterm hydrograph data needed to make these assessments. This paper presents an example of how to bridge the conceptual and physical divide between GIS-based watershed modelling of basin-discharge and in-stream hydraulic habitat models. Specifically, we used a Soil andWater Assessment Tool (SWAT) model for the Hillsborough River to produce data for use in a Physical HABitat SIMulation (PHABSIM) model of the same river. This coupling of models allowed us to develop long-term discharge data in ungauged river systems based on watershed characteristics and precipitation records. However this approach is not without important limitations. Results confirm that accuracy of the SWAT-predicted hydrograph declines significantly when either the DEM resolution becomes too coarse or if DEM data are resampled to a coarser or finer resolution. This is due to both changes in the size and shape of the river basin
with the varying DEMs and subsequent shifts in the proportions of land use, soils and elevation. Results show the use of 30m DEMs produced hydrographic patterns amenable for using in-stream habitat protocols like PHABSIM model, especially where little or no hydrographic and land use information exists.

There is a long-standing debate whether mangrove and seagrass habitats in the Indo-Pacific region function as nurseries for coral reef fishes. We studied the use of all major shallow-water habitat types by juvenile coral reef fish using visual census surveys at 4 islands along the Tanzanian
coast (East Africa) and at the island of Grande Comoros (Comoros archipelago). We investigated the value of mangroves, seagrass beds, coral reefs, macroalgae and intertidal flats as a juvenile habitat for fish by studying density distribution patterns of juveniles and adults of 76 reef fish species in these habitats. We assessed (1) which part of the reef fish community used mangrove–seagrass habitats as
juvenile or adult habitats, (2) whether adult fish densities and diversity on adjacent reefs were related to the presence of these shallow habitats, and (3) whether adults of species that use these habitats when juvenile were less abundant on coral reefs situated far away from these juvenile habitats. Seagrass beds and coral reefs were the most important juvenile fish habitats. Ontogenetic migrations between seagrass beds and reef habitats possibly occur, since several species showed their highest juvenile densities on seagrass beds, whereas adults showed their highest densities on reefs adjacent
to these seagrass beds. The presence of areas with seagrass beds positively influenced adult densities of many reef fish species on adjacent coral reefs. Of the 36 fish species whose juveniles were observed in seagrass beds along the Tanzanian coast, 32 species were absent from or showed low densities on coral reefs of the island of Grande Comoros (lacking seagrass beds or mangroves). On reefs far from seagrass beds and mangroves along the Tanzanian coast, 25 of these 36 species were absent or showed low densities in comparison with reefs adjacent to these habitats.

Coral reefs and seagrass beds are often located adjacent to one another, but little is known about the degree to which their fish communities are interlinked. To determine whether coral reef fishes on the coral reef are interlinked with or segregated from fishes on adjacent seagrass beds, a 60 m coral reef–seagrass gradient was studied on the island of Zanzibar in the western Indian Ocean. Using underwater visual census, coral reef fishes were surveyed in 4 habitat zones: (1) a coral patch reef, (2) seagrass beds bordering the coral reef, (3) seagrass beds at a 30 m distance from the coral reef edge and (4) seagrass beds at a 60 m distance from the coral reef edge. Based on the densities of juveniles and adults in the 4 zones, the 48 species that were observed were classified into reef-associated species, seagrass-acsociated species, nursery species, generalists and rare species. Reef-associated species occurred almost exclusively on the coral reef and at the reef–seagrass edge, while seagrass-associated species occurred almost exclusively on the seagrass beds. Generalists and seagrass-associated species occurred in all 3 seagrass zones, but densities of generalists on seagrass beds decreased with increasing distance from the reef, whereas that of seagrass-associated species increased. Reef-associated and generalist species showed an edge effect, where densities on the seagrass beds near the reef edge were higher than on the seagrass beds further away. Juvenile densities of nursery species on seagrass beds also increased with the distance from the reef, whereas their adults showed the highest densities on the coral reef, suggesting a possible ontogenetic shift from the seagrass beds to the reef. The results of the present study show that this seagrass–coral reef landscape features habitat segregation between species and life stages and shows an edge effect, possibly driven by competition mechanisms between species or life stages.

Soil erosion due to accelerating runoff in various land cover types pose a serious threat to the long term sustainability of the fragile Himalayan landscape characterized by subsistence farming. Delimitation of high runoff zones, fostering soil erosion in the agricultural dominated watersheds is thus a necessity for watershed managers, NGO’s, urban planners, policy makers, and municipal administrations. The approach is practical, SWAT is a straightforward modeling system using GIS information. The BMP is also a very practical approach to mitigation of runoff accumulation on sub basin. A set of interviews with people familiar to the issues locally strengthened the confidence in the results. Results from this research show that terraces can reduce storm water runoff very effectively. There is a huge difference in amount of runoff after terraces are added to the simulation. Results showed a reduction of 57% in runoff volume during August 2004. If the area near the river is preserved as...

Estuaries are extremely productive ecosystems, providing habitat for numerous species and considered essential nurseries for many marine species. However, environmental changes caused by anthropogenic threats affect the quantity and quality of estuarine habitats. Habitat loss and degradation, conspicuous in estuaries worldwide, is considered a major threat to the function of estuaries. Information on threatening processes leading to habitat loss and degradation and on relative vulnerability of different habitats is imperative for conservation planning. The vulnerability of estuarine habitats (saltmarsh, mudflat and subtidal) from five Portuguese estuaries (Ria de Aveiro, Tejo, Sado, Mira and Guadiana) to the major anthropogenic threats affecting them was assessed through a threat ranking approach developed with basis on expert’s judgment. Simultaneously, fish communities in these habitats (in two sites per estuary) were sampled seasonally in 2009, in order to relate habitat vulnerability and habitat use by fish communities. Experts ranked threats such as nutrient input, changes in river flow and land reclamation as having stronger impacts on estuarine habitats, making them more vulnerable to destruction or degradation. Averaging all threats, vulnerability scores were higher in mudflats, followed by subtidal and saltmarsh habitats. Overall, habitats in the Tejo and Sado estuaries were ranked as the most vulnerable and in the Mira estuary as the least vulnerable. Relationships between vulnerability values and fish community indicators in sampled habitat types and areas evidenced similarities between intertidal habitats (saltmarsh and intertidal mudflats), highlighting their priority in conservation and management contexts. The presented approach provides a useful tool as it allows prioritizing threat mitigation and identifies the most vulnerable habitats to potential degradation and destruction.

Mesohabitat heterogeneity in four mediterranean streams of the Jucar river basin (Eastern Spain) The hydromorphological units or mesohabitats of the headwaters of four Mediterranean streams were studied by classifying them according to their length, width, average and maximum depth, percentage of substrate and water volume. These characteristics of the streams were assessed over four consecutive years (2003-2006). The main objective of the study was to analyse the spatial and temporal pattern of the physical attributes of the mesohabitats. Classification strength and dendrograms were examined for each stream, mesohabitat type, and combination of stream and mesohabitat type and for stream mesohabitat type and year, whereas Nonmetric Multidimensional Scaling ordination (NMDS) was used to analyse classification patterns. The Villahermosa and Ebrón streams were more homogeneous than the Palancia and Vallanca streams, whereas riffles and glides were more homogeneous than pools and rapid...

The effects of tourist visitation and food provisioning on fish assemblages were assessed by visual censuses (stationary technique) carried out in a tropical reef in Northeastern Brazil. Comparisons of species abundance, richness, equitability, and trophic structure in the presence (PT) and absence (AT) of tourists suggest that tourist visitation and supplementary food influenced the structure of the fish assemblage, as follows: (a) diversity, equitability and species richness were significantly higher on the AT period, while the abundance of a particular species was significantly higher during PT; (b) trophic structure differed between the AT and PT periods, omnivores being more abundant during the latter period, while mobile invertivores, piscivores, roving herbivores and territorial herbivores were significantly more abundant on AT. Reef tourism is increasingly being regarded as an alternative to generate income for human coastal communities in the tropics. Therefore, closer exam...

Soil erosion due to accelerating runoff in various land cover types pose a serious threat to the long term sustainability of the fragile Himalayan landscape characterized by subsistence farming. Delimitation of high runoff zones, fostering soil erosion in the agricultural dominated watersheds is thus a necessity for watershed managers, NGO’s, urban planners, policy makers, and municipal administrations. The approach is practical, SWAT is a straightforward modeling system using GIS information. The BMP is also a very practical approach to mitigation of runoff accumulation on sub basin. A set of interviews with people familiar to the issues locally strengthened the confidence in the results. Results from this research show that terraces can reduce storm water runoff very effectively. There is a huge difference in amount of runoff after terraces are added to the simulation. Results showed a reduction of 57% in runoff volume during August 2004. If the area near the river is preserved as a buffer strip and a tree cover is maintained around it then it minimizes the effect surface runoff. The effect of riparian buffer is modeled in SWAT and the results show about 6% decrease in surface runoff when a buffer of 500 m is applied around the main river only. This study has drawn a location map of BMP implementations of most meaningful impact for the rehabilitation and safeguard of rivers, streams, lakes and wetlands around the river Siran in the Mansehra Tehsil, Pakistan.

The cover of seagrasses and macroalgae, landings and fishery-independent measures of spawning stock and recruitment for brown tiger prawns, were monitored immediately following a major cyclone in Exmouth Gulf Western Australia in March 1999. Anecdotal evidence on the extent of seagrass from 1990 to 1998 suggests that the cyclone caused a major, immediate disruption and loss of the  eagrass/macroalgal
beds (to
2% cover), the critical prawn nursery habitat, and mangroves in the shallow inshore waters of the system. Prawn landings and recruitment to the fishery were not affected in the year of the cyclone, but were markedly lower in the two years immediately afterwards and then increased as the cover of macrophytes increased to over 40% in 2003. Tiger prawn landings and catch rates were not affected in Shark Bay, a system 500 km south of Exmouth Gulf that did not experience cyclonic
disturbance. Seagrasses in Exmouth Gulf showed a succession of species from small colonising species (Halophila ovalis and Halodule uninervis) to larger, broad-leaved species (Cymodocea serrulata, Syringodium isoetifolium) only two years after the cyclone. The recruitment and landings of tiger prawns were correlated significantly with the total cover of macroalgae and seagrass. The large loss of seagrass and macroalgae reduced the settling habitat for postlarvae and the nursery habitat for juvenile tiger prawns, probably leading to the lower recruitment to the fishery. These findings suggest that the extent of seagrass and macroalgae are some of the factors defining the productivity of the tiger prawn fishery in Exmouth Gulf.

Indonesia is the world’s richest country regarding reef fish diversity. Nevertheless, the reef ichthyofauna of the Indonesian Archipelago remains poorly known, primarily due to a lack of sampling. Coral reefs in the Kepulauan Seribu Marine National Park close to the Indonesian capital Jakarta are under threat by many destructive activities that trigger a loss of habitat and species diversity. This communication: (1) describes the reef fish community structure from three distinct reef habitats in the Pari Island group dominated by Acropora branching corals (ACB), foliose corals (CF) and massive corals (CM), using a number of community properties such as numerical abundance, species richness, diversity, and multivariate similarity; (2) examines the temporal variation of the fish community from the three habitats; and (3) discusses possible implications for the monitoring of qualitative changes in coral reef systems on small islands. During this study, a total of 13 536 individual fishes were counted, representing 205 species belonging to 36 families. In terms of species richness, Pomacentridae was the dominant fish family in ACB and CF sites (40 % and 48.6 %, respectively), and Labridae (27.4 %) was the dominant family in the CM plots. The most species-rich habitat was ACB with 125 species (with Amblyglyphidodon curacao as the most characteristic species), followed by CM and CF with 117 (Thalassoma lunare) and 79 species (Pomacentrus alexanderae), respectively. Average Shannon-Wiener diversity (ln basis) ranged from 2.0–2.9 (ACB), 2.4–3.1 (CF), and 2.1–3.0 (CM), with no significant difference between growth forms. Abundance, species richness and diversity showed significant seasonal variability, but the effects differed between habitats. Multivariate analysis of the reef fish community was able to detect significant differences between species composition and diversity of the reef fish community between sites with different coral growth forms at Pari Island, both when based on species abundances and when aggregated according to trophic categories. It thus constitutes a useful tool to detect qualitative differences of the species-rich Indonesian coral reef ecosystems.

According to conceptual models, the distribution of resources plays a critical role in determining how organisms distribute themselves near habitat edges. These models are frequently used to achieve a mechanistic understanding of edge effects, but because they are based predominantly on correlative studies, there is need for a demonstration of causality, which is best done through experimentation. Using artificial seagrass habitat as an experimental system, we determined a likely mechanism underpinning edge effects in a seagrass fish. To test for edge effects, we measured fish abundance at edges (0–0.5 m) and interiors (0.5–1 m) of two patch configurations: continuous (single, continuous 9-m2 patches) and patchy (four discrete 1-m2 patches within a 9-m2 area). In continuous configurations, pipefish (Stigmatopora argus) were three times more abundant at edges than interiors (positive edge effect), but in patchy configurations there was no difference. The lack of edge effect in patchy configurations might be because patchy seagrass consisted entirely of edge habitat. We then used two approaches to test whether observed edge effects in continuous configurations were caused by increased availability of food at edges. First, we estimated the abundance of the major prey of pipefish, small crustaceans, across continuous seagrass configurations. Crustacean abundances were highest at seagrass edges, where they were 16% greater than in patch interiors. Second, we supplemented interiors of continuous treatment patches with live crustaceans, while control patches were supplemented with seawater. After five hours of supplementation, numbers of pipefish were similar between edges and interiors of treatment patches, while the strong edge effects were maintained in controls. This indicated that fish were moving from patch edges to interiors in response to food supplementation. These approaches strongly suggest that a numerically dominant fish species is more abundant at seagrass edges due to greater food availability, and provide experimental support for the resource distribution model as an explanation for edge effects.