Optimal flow for brown trout: Habitat - prey optimization.

TL;DR: This work developed density-environment relationships for three different life stages of brown trout that show the limiting effects of hydromorphological variables at habitat scale and proposed a new approach for the identification of optimal flows using the limiting factor approach and the evaluation of basic ecological relationships.

About: This article is published in Science of The Total Environment.The article was published on 2016-10-01 and is currently open access. It has received 17 citations till now.

Summary

1. Introduction

• Freshwater is a fundamental and limiting resource, both for the development of human society and for the maintenance of biodiversity and functionality of aquatic ecosystems.
• Human demand for water is constantly increasing, chiefly for hydroelectric power production and for agricultural purposes (de Fraiture andWichelns, 2010).
• Habitat-based models have been widely used to define a relationship between in-stream flow and habitat availability for various species offish and, thus, to define the optimal orminimumflow rate (i.a.
• The use of density-environment relationships that show the limiting effects of the habitat characteristics and not the average effects of the same variables seems more adequate.
• Quantile regression allows the association of the different rates of change to the different parts of the response distribution, being a method for estimating relationships among variables for all the portions of a probability distribution (Koenker and Bassett, 1978).

2.1. Study area

• The study area is the Alpine Valley of the Serio River in northern Italy.
• It was chosen because it provides a large variety of environmental conditions.
• The river substratum varies from sand to bedrock including all the intermediate substrate classeswith a heterogeneous distribution.
• This stretch of the river can be considered to be near-pristine due to the absence of other anthropogenic impacts (Canobbio et al., 2010).

2.2. Fish sampling

• In order to produce habitat suitability curves, fish samplingwas performed according to Mäki-Petäys et al. (1997) and Van Liefferinge et al. (2005).
• To minimize the flight bias, which may cause the displacement of individuals from their original position, a modified point electrofishing procedure was used.
• The activated anode was submerged for several seconds every 0.5–1.0 m (measured between the anode centers of two consecutive ‘dippings’).
• The point of the first sighting of fish was noted with a different reconnaissance symbol (colored stake) in order to know the placement of the different individuals after electrofishing.
• Some individuals were also stomach-flushed (Bridcut and Giller, 1995), and the ingested prey were collected and identified in laboratory.

2.3. Characterization of sampling sites, habitat use and availability

• After sampling, different habitat descriptors weremeasured for each individual in order to define the habitat use.
• Water velocity was measured at 40% of the depth in order to obtain the mean velocity of the water column in the sampling point.
• Using these three descriptors the authors derived the maximum substrate size (MSS) and the percentage of fine substrates.
• The availability of refugia was also evaluated.
• S, B)Map of water depth, C)Map of water velocity, D)Map of themaximum substrate size, he location of refugia, G) Reclassified habitat map evaluated using the water velocity and indicate riffles and number 4 indicate runs.

3.1. Development of quantitative habitat suitability models

• To avoid multicollinearity a selection of variables using Variance Inflation Factor (VIF) were performed before quantile regression analyses (Neter et al., 2004).
• Since the wide range of values of raw data all the independent variables were unit-based standardized [X′ = (X − Xmin) / (Xmax − Xmin)].
• Hence, quantile regression has been used instead of traditional central response models in order to examine with more ease the boundaries of density–habitat relationships, i.e. the upper limits imposed by the limiting factors.
• The authors also created a mesohabitat-level dummy variable for evaluate fixed-effects for different intercepts among habitat categories.
• For each model a τ-specific version of Akaike Information Criterion, corrected for small sample size (AICc(τ)), was calculated for every studied quantile.

3.2. Habitat simulations

• The authors selected 8 sites, different from those used for the suitability evaluation, in order tomodel the changes induced by flowmodifications on physical habitat.
• The collected data were used as input to simulate the habitat features and availability by PHABSIM (Physical Habitat SimulationModel) (Waddle, 2001)which has been used for decades in ecological flow studies.
• The sections were placed at a distance comparable to the streamwidth and in order to represent properly the morphological variability, so the distance between sections was not the same in all sampling sites.
• Water velocity and depth measurements were repeated 3 or 4 times with different flows in order to produce a rout densities.
• All variables are grouped by habitat type.

3.3. Habitat availability evaluations

• The authors have defined as microhabitats a portion of the river that has dimensions of one square meter and may represent fairly homogeneous habitat for macroinvertebrates, conversely they defined four kinds of mesohabitats (shallow pool, deep pool, riffle, run) that may represent fairly homogeneous habitat for trout.
• The authors evaluated the potential density (ind/m2) in each cell of each site for each discharge and, multiplying this value for the cell area and summing the obtained results, they were able to provide a potential number of individuals in each site for each discharge.
• The authors used the density models developed in this work to evaluate the potential density of each class of age of brown trout in each habitat per site.
• Habitats were classified and described according to the same criteria used during the analyses of field data for the development of density models.
• Þ½ where Qopt are the optimal flows for fish, Q are vectors of modeled discharge, HA are vectors of available habitat.

4.1. Habitat suitability models

• In the 13 sampled sites, a total of 73 different habitat units were characterized.
• Also the effect of MSS was best described by the quadratic functionwhich predicts higher densities for intermediate size (Fig. 2B).
• The models selected to explain the densities of all macroinvertebrate families consider velocity as amain driving force (both in univariate or bivariatemodels).

4.2. Optimal flows

• Habitat availability patterns were generated by the different effects of flow-dependent depth and velocity variations, but also by the substrate characteristics of the gained or lost river bed and later a range of flows that maximize the habitat availability for the different life stages of brown trout were obtained for each modeled site.
• The area of available habitat in each site, for the different life stages, can increase up to three times (Table 4) in the consider range of flows.
• Habitats for adult brown trout are the most influenced by flows (Figs. 4 and 5) showing a within-site percentage increase of 156 ± 77 (mean ± sd).
• Across the modeled sites, between theminimum and themaximum optimal flows selected for trout, the dry biomass of macroinvertebrate increases by 141±183% (Table 5).
• Thismeans that the available energy can increase N2 times inside the range offlows that preservemost of the habitat for fish.

5. Discussion

• Among the modeled sites, higher invertebrate production occurred in mesohabitats with greater water velocities, such as riffles and runs, while fish density increased in pools and near the refugia where the trout could find better cover.
• In some sites most of the habitats for fish are already available for really low flows, while, on the contrary, the macroinvertebrate biomass always increases with increasing flow, as shown in Fig.
• For all the life stages, the selectedmodels considered simultaneously more than one variables and implied that the suitability changes among mesohabitat.
• The availability of refugia, evaluated as the fractions of available habitat that were characterized by discontinuity in the riverbed profile, where the depth of the water was over 0.30 m and the MSS was N0.5 m, had a significant effect only on adult trout densities.
• On the other hand the density of juvenile brown troutwere higher in habitats characterized byMSS between 0.4 and 0.7m, and as also for fry, shallow pools habitats could provide a better environment than others.

6. Conclusions

• Habitat models that predict flow-related changes in productivity are usually used for the definition of environmental flows, and thus the information they provide must be as accurate as possible and useful for the water managers.
• Above such threshold, mesohabitats switch from pools to riffles and runs that are less suitable for all the life stages.
• These findings should be taken in account as they have implications in flowmanagement.
• Looking at the limiting response of densities to flow-related variables seems a promising approach.

Figures

Fig. 3. Multivariate quantile regression models for macroinvertebrate densities: (A) Leuctrida Independent variables are water velocity and mean substrate diameter (S). Lines and surface white points outside (reproduced from Fornaroli et al. 2015).

Fig. 1. Examples of physical habitat maps: A) Points of measurement of environmental variable E) Map of the fine substrate percentage, F) Map of refugia availability, black shading indicate t water depth maps, number 1 indicate shallow pools, number 2 indicate deep pools, number 3

Table 4 Habitat availability variation for adult, juvenile and fry brown trout in the 8 modeled sites. X = [(Xmax − Xmin) / Xmin] ∗ 100.

Fig. 4. Trends in mesohabitat availability and total wet area in the 8 modeled sites.

Table 1 Model equations used to describe the effect of a single gradient.

Table 2 Mean values and standard deviations of the characteristics of the sampled habitats and mean t

Fig. 5. Trends in available habitat for adult, juvenile, fry brown trout and potential macroinvertebrate biomass in the 8 modeled sites.

Table 5 Optimal flows calculated for the 8 modeled sites and relative biological metrics. Macroinvertebrate dry weight was compared to the worst situation inside the acceptable range of flows identified for fish.

Table 3 Variable selection results for each brown trout life stage (adult, juvenile, fry). Variable selectionwas performed using Akaikeweights (wi) averaged from small-sample Akaike information criterion AICc(τ) selection of quantile regression models. For each selected variable are also listed its respective shape and the mesohabitat effect (“+”: different intercepts among mesohabitat, “−”: same intercept for all the mesohabitat). Only the models with the 5 highest average Akaike weights (wi) are shown.

Frequently asked questions

Q1. What are the contributions in "Optimal flow for brown trout: habitat – prey optimization" ?

Forna et al. this paper defined optimal flows using both physical habitat and prey availability, and found that most of the habitat for juvenile and fry brown trout is available at low flows. 

Q2. What functions were the performing for the effect of MSS?

The functions that best described the effects offinepercentage and water depth were respectively the exponential and the logarithmic one. 

Q3. Why has quantile regression been used instead of traditional central response models?

quantile regression has been used instead of traditional central response models in order to examine with more ease the boundaries of density–habitat relationships, i.e. the upper limits imposed by the limiting factors. 

Q4. What is the main reason why the results of habitat based models are important?

As physical habitat is a necessary, but not sufficient condition for the development and survival of fish, the results of habitat based models may best be viewed as indicators of population potential, in systems where the habitat conditions described by the models are the major population constraints. 

Q5. Why are they rarely used in habitat-based methods?

Macroinvertebrates are rarely used in habitat-basedmethods because of the high heterogeneity of the density response along environmental gradients. 

Q6. What factors limiting the densities of brown trout were?

In their analyses, the authors found that the factors limiting the densities of trout were water depth, substrate characteristics and refugia availability. 

Q7. What were the three types of substrates used to determine the maximum size of the fish?

The substrates were classified as dominant, sub-dominant and matrix as the more abundant, the second more abundant or the finer class which occupies the interstices between the larger sized elements. 

Q8. Why is the Serio River considered near-pristine?

Despite the presence of dams, this stretch of the river can be considered to be near-pristine due to the absence of other anthropogenic impacts (Canobbio et al., 2010). 

Q9. How was the habitat availability – discharge relationship evaluated?

The habitat availability – discharge relationships were evaluated by fitting a spline function for each species in each site to the output of the model and using it for making prediction for every 0.050 m3/s in the considered range. 

Q10. How many times did the study section be sampled?

To avoid faulty observations of habitat utilization caused by the displacement of individuals due to flight from the electric current, each study section was sampled only once with this technique. 

Q11. What is the model for describing brown trout densities?

themodel consideringwater depth andMSS as the independent variables and accounting for themesohabitat effect was selected as the best (averagedwi=0.339) for describing juvenile brown trout densities. 

Q12. What was the calibration of the stage discharge model?

The stage-discharge model (STGQ, Waddle 2001) was calibrated using the measured water surface levels recorded during the hydraulic surveys. 

Q13. How many times can the area of available habitat in each site increase?

The area of available habitat in each site, for the different life stages, can increase up to three times (Table 4) in the consider range of flows. 

Q14. What is the way to estimate the biomass of macroinvertebrate?

Across the modeled sites, between theminimum and themaximum optimal flows selected for trout, the dry biomass of macroinvertebrate increases by 141±183% (Table 5).