tian möllmann & ole vestergaard emi: jonne kotta fgfri: antti … · 2008. 4. 30. · for...

Title Guidelines for harmonisation of marine data

BALANCE Report No. 32 Date December 2007

Authors Lena Bergström (Ed.), Ulf Bergström, Martin Isæus, Jonne Kotta, Christian Möllmann, Alfred Sandström, Claus R Sparrevohn, Jonna Tomkiewicz & Ole Vester-gaard Method descriptions provided by: DIFRES: Claus R Sparrevohn, Jonna Tomkiewicz, Chris-tian Möllmann & Ole Vestergaard EMI: Jonne Kotta FGFRI: Antti Lappalainen IAE: Juris Aigars IFM-GEOMAR: Gerd Kraus, Hans-Harald Hinrichsen & Rüdiger Voss Metria: Sandra Wennberg Metsä: Minna Tallqvist & Martin Snickars NERI: Karsten Dahl & Jürgen Hansen NIVA: Martin Isæus SBF: Lena Bergström, Ulf Bergström, Alfred Sandström & Göran Sundblad SFI: Wlodimirz Grygiel

Approved by Johnny Reker

Revision Description By Checked Approved Date

Key words Classification

Open

Internal

Proprietary

Distribution No of copies

BALANCE Secretariat BALANCE Partnership

ii

BALANCE Interim Report No.

iii

CONTENTS

1 INTRODUCTION ............................................................................................................ 1 1.1 Purpose of the report ...................................................................................................... 1 1.2 Relationship to current international recommendations.................................................. 1 1.3 Data sets and methods................................................................................................... 1

2 GENERAL GUIDELINES FOR HABITAT MAPPING...................................................... 2 2.1 Analyses of distributions ................................................................................................. 2 2.1.1 Spatial interpolation ........................................................................................................ 2 2.1.2 Criteria analysis .............................................................................................................. 3 2.1.3 Statistical modelling ........................................................................................................ 3 2.2 Variables in habitat mapping .......................................................................................... 5 2.2.1 Response variables ........................................................................................................ 5 2.2.2 Environmental predictor variables .................................................................................. 5

3 GENERAL GUIDELINES FOR DATA INTERCALIBRATION......................................... 7 3.1 Data types....................................................................................................................... 7 3.2 Method efficiency............................................................................................................ 7 3.3 Definitions of terms related to marine habitat mapping .................................................. 8 3.4 Data sets used in BALANCE: Compilation of method descriptions ................................ 9

4 HARMONISATION OF FISH DATA.............................................................................. 10 4.1 Biological aspects ......................................................................................................... 10 4.2 Description of methods for data collection.................................................................... 10 4.3 Data Intercalibration...................................................................................................... 15 4.3.1 Nursery areas ............................................................................................................... 15 4.3.2 Foraging and spawning areas (Pilot area 1, study area 2) ........................................... 17 4.4 Data applicability for habitat mapping........................................................................... 17 4.5 Suggestions for forthcoming data collection ................................................................. 18

5 HARMONISATION OF MACROFAUNA DATA ............................................................ 20 5.1 Biological aspects ......................................................................................................... 20 5.2 Description of methods for data collection.................................................................... 23 5.3 Data intercalibration...................................................................................................... 24 5.4 Data applicability for habitat mapping........................................................................... 25 5.5 Suggestions for forthcoming data collection ................................................................. 27

6 HARMONISATION OF ZOOPLANKTON DATA........................................................... 28 6.1 Biological aspects ......................................................................................................... 28 6.2 Description of methods for data collection.................................................................... 28 6.3 Recommendations for data intercalibration .................................................................. 29 6.4 Data applicability for habitat mapping........................................................................... 29 6.5 Suggestions for forthcoming data collection ................................................................. 29

7 HARMONISATION OF PHYTOBENTHIC DATA.......................................................... 30 7.1 Biological aspects ......................................................................................................... 30 7.2 Description of methods for data collection.................................................................... 30

BALANCE Interim Report No.

iv

7.3 Recommendations for data intercalibration .................................................................. 31 7.4 Data applicability for habitat mapping........................................................................... 31 7.5 Suggestions for forthcoming data collection ................................................................. 32

8 SUMMARY AND CONCLUSIONS ............................................................................... 33 8.1 Fish ............................................................................................................................... 33 8.1.1 Suggestions for future data collection........................................................................... 33 8.2 Macrofauna................................................................................................................... 34 8.2.1 Suggestions for future data collection........................................................................... 34 8.3 Zooplankton .................................................................................................................. 35 8.3.1 Suggestions for future data collection........................................................................... 35 8.4 Benthic vegetation ........................................................................................................ 35 8.4.1 Suggestions for future data collection........................................................................... 36

9 REFERENCES ............................................................................................................. 37

10 APPENDICES............................................................................................................... 43 10.1 Appendix 1. Methods for sampling fish data................................................................. 43 10.2 Appendix 2. Methods for sampling macrofauna data ................................................... 49 10.3 Appendix 3. Methods for sampling zooplankton data ................................................... 51 10.4 Appendix 4. Methods for sampling phytobenthos data................................................. 53 10.5 Appendix 5. Methods for sampling phytoplankton, hydrography & hydrochem. data ... 56

BALANCE Interim Report No. 32

1

1 INTRODUCTION

Comprehensive information on the distribution of marine habitats is currently scattered and limited, particularly in comparison to terrestrial habitats. One main aim of the BALANCE project is to produce transnational maps of marine species, habitats and landscapes in the Baltic Sea Region (BSR) by collating existing biological and geologi-cal data sets and by spatial modelling. As data sets are commonly produced by differ-ent standards in different nations, the harmonisation of data sets from different sources is an essential part of the progress.

1.1 Purpose of the report

This report provides general guidelines for biological data collection and intercalibration of data sets from the perspective of their usefulness in habitat mapping and modelling. Specific recommendations are given for the harmonisation of biological data sets used within BALANCE. Data sets that are openly available may be retrieved upon request to each data owner. At the end of the project GIS shapefiles of key products will also be accessed through the HELCOM website (www.helcom.fi). Hydrographic data are not specifically included in this report, but as they are often an essential part of marine habitat mapping and spatial modelling, and are routinely sam-pled together with biological data in many cases, some methods used for hydrographic data sampling are also described. This also applies for phytoplankton data, which is a potentially useful environmental variable in modelling, although not explicitly mapped within BALANCE.

1.2 Relationship to current international recommendations

International recommendations for data collection and quantification developed within HELCOM, ICES and EU are of high significance for research cooperation within the BSR. As discussed and decided at a workshop held in Uppsala, November 8-10, 2005 and at a workshop arranged by the Swedish EPA in Stockholm February 3, 2006, the approach of BALANCE in this context should be to encompass available international recommendations to the extent that they fulfil requirements for large-scale habitat mapping, and the demands of the stakeholders. Thus, the focus of this report will be to describe and discuss the available systems for data collection and view their useful-ness from a habitat mapping and modelling perspective.

1.3 Data sets and methods

Method descriptions in the report refer to data sets in use or planned for use within BALANCE by June 30, 2006, as defined by a questionnaire that was sent out to and responded on by all involved partners (See section 4). The data sets confine to either of the following two categories:

1. Spatial raw data: Data sets are used in mapping or modelling and are openly available. The project partners or data owners enter Metadata for the biological data into the BALANCE data portal, and data may be retrieved upon request. This is in order to ensure their standard and that relevant updates are included.

2. Spatial modelling data: The data sets are used in modelling but may not be openly available due to restrictions of the data owners. However, the maps of species and habitat distributions produced by BALANCE partners are available.

http://www.helcom.fi/


2

2 GENERAL GUIDELINES FOR HABITAT MAPPING

All data sets that are georeferenced can be plotted spatially, and the maximum resolu-tion of the obtained map is identical to the sample size. However, the accuracy and coverage of the obtained map will depend on which methods are used for data sam-pling and mapping.

2.1 Analyses of distributions

Maps based on direct field sampling typically have high biological accuracy but provide data sets with only partial coverage. Direct mapping with total coverage is economically unfeasible at larger scales, as an extensive amount of data collection and analysis is required. Instead, different modelling approaches are used to predict the distribution of species beyond that of the observed data. Three basic approaches for increasing cov-erage are defined as:

1. spatial interpolation 2. criteria analysis 3. statistical modelling

Their basic assumptions and applicability are reviewed in the sections below. Additionally, distributional maps may be provided by the analysis of remotely sensed data, based on satellite images, aerial photography or aerial laser scanning. Satellite images may be used to obtain biological maps over large areas of shallow or surface waters, and importantly, also maps of some relevant abiotic factors such as water tur-bidity, water depth and temperature. The reliability of the results is highly dependent on access to ground-truth data on which the image reclassification is based, and the po-tential output is limited in terms of depth range, particularly in the BSR where waters are commonly turbid. However, the maps that can be provided have a potentially good resolution, with a grain size down to 10m or less. Preliminary results (presented in BALANCE draft report “Evaluation of remote sensing methods as a tool to characterize shallow marine habitats” (BALANCE WP2 MS1) indicate that remote sensing and as-sociated analyses are a promising tool for mapping the distribution of shallow marine habitat types. Further evaluation of the usage of remote sensing is currently in process within WP2.

Table 1. Analytical approaches applied in habitat mapping. Coverage Env.data

required Data format

Direct mapping Same as input data No All Interpolation Higher than input

data No (op-tional)

Continuous, Ordinal

Criteria analysis Higher than input data

Yes Ordinal, Nominal

Statistical model-ling

Higher than input data

Yes All

2.1.1 Spatial interpolation The term interpolation refers here to spatial predictions that are based only on informa-tion available in the response variable itself at its spatial positions. Thus, predictions


3

are made in geographic space, rather than considering properties of the ambient envi-ronment. Spatial interpolation techniques are based on the assumption that adjacent sites are spatially autocorrelated, which means that a variable at one site is more likely to have the value of an adjacent site than that of a more distant site. Methods for interpolation are based on mathematical functions that may be defined as either exact or approximate. Exact methods preserve the values of the input data, while approximate methods produce smoother surfaces. For biological data, approximate methods are often more appropriate, because they may account for local uncertainties in the input data. Common approximate methods for interpolating point data are inverse distance weighted averages, trend surface analyses and kriging, all of which are usually implemented in common GIS software. Interpolation may be useful for producing layers of environmental data, such as esti-mating the probable depth or salinity in unsampled areas. For biological data, interpola-tion may be used for generalising the distribution of species at large, regional scales. At small scales, interpolation may be motivated when the environment is non-patchy with respect to the habitat demands of the species.

2.1.2 Criteria analysis In criteria analysis, spatial predictions of the occurrence of a response variable are made in relation to a set of categorical (nominal or ordinal) layers of environmental variables. The environmental layers are arranged on top of each other, and combined to one layer where all possible combinations of the input layers are represented. The occurrence of the response variable within each combination is then defined from its observed frequency in field samples from corresponding environmental settings. For example, combinations of bottom substrate and wave exposure that correspond to cer-tain abundances of a certain species may be identified. The environmental layers should have full coverage and data for the response variable should be collected so that the relevant environmental gradients are covered. Although the procedure is basically non-statistical, the relevance of the output increases if the layers and categories applied are properly statistically defined and known to be rele-vant for the distribution of the response variable. Criteria analyses are easy to apply and to communicate. Robust results can potentially be achieved, but the accuracy of the prediction may be low unless a high number of classes are defined within each layer. Thus, the procedure may require high amounts of data processing, especially when predicting the distribution of individual species and when a large part of the environmental data is originally in continuous data format.

2.1.3 Statistical modelling Statistical modelling is technically more demanding, but may provide a way of utilising more (or all) of the information available on the relationship between predictor and re-sponse variables. The output is numerical, for example the abundance of a species, or the probability of finding a certain species or life-stage at a site may be estimated. Predictions are based on a mathematical function describing the relationship between the response variable and relevant environmental variables. This function is then used to estimate the value of the same response variable at other sites based on existing in-formation in environmental layers. The response variable should be represented by point data, and the environmental predictor variables must have full coverage over the


4

area to be predicted. The environmental predictor variables should preferably be casu-ally related to response variable (see section 2.2.2) and cover a gradient from the minimum to maximum for the response variable along each of the predictor variable gradients. The quality of the output depends on the level of compatibility between the properties of the applied data sets and the regression method. Commonly applied methods vary in their response curve and assumed error functions (Table 2). The simplest way of de-scribing the relationship between predictors and the response variable is multiple linear regression, while generalized linear models (GLM) and generalized additive models (GAM) are more technically complex, they are more flexible and better suited for de-scribing non-linear relationships (Guisan & Zimmerman 2000, Guisan et al. 2002, Leh-mann et al. 2002, Garza-Pérez et al. 2004, Francis et al. 2005). GAM has the advan-tage of high flexibility, as the response curves are not pre-defined but may be explored and fitted to observed response distribution along environmental gradients. This makes GAM potentially better suited for modelling the spatial distribution of species with asymmetrical or polynomial distributions, which are often observed in real data (Leh-mann et al. 2002). GRASP (Generalized Regression Analysis and Spatial Prediction) is a statistical software for spatial prediction based on regression analyses (GAM) that is implemented as an interface and a collection of functions in the statistical software S-plus and R. It also has the advantage of being compatible with the GIS-software Arc-View 3.x. In some cases other methods may be preferred. For example classification and regres-sion trees (CART) are more suitable for modelling interactions, although it does not al-low observation of response shapes and is restricted to assuming Gaussian distribu-tions (Lehmann et al. 2002). Another flexible modelling tool is artificial neural networks (ANN), which has the advantage of providing a way of modelling assemblages of spe-cies, as several response variables can be modelled simultaneously (Brosse et al. 1999, Lek & Guégan 1999, Joy & Death 2004). In datasets with many zero values in the response variable multivariate analytical models such as Canonical correspon-dence analyses (CCA), may be more appropriate than models of individual species (Austin 2002). As an alternative, the included data sets can be cut down to only repre-sent the realized distribution of the response variable (Lehmann et al. 2002), or a “presence only” approach can be applied, using only the sites where the species is present (Zaniewski et al. 2002).

Table 2. Examples of regression analyses used to make spatial predictions Method Statistic distribution Response curve Least Square Regression (LSR) Gaussian Parametric, quadratic Logistic regression

Binomial Parametric, quadratic

Generalized linear models (GLM) Gaussian, binomial, Pois-son…

Parametric, quadratic

Generalized additive models (GAM) Gaussian, binomial, Pois-son…

Non-parametric, smoothed, any shape

Classification and regression trees (CART)

Gaussian Non-linear, non-additive


5

2.2 Variables in habitat mapping

There is a wide range of variables applicable for mapping marine habitats and many different approaches depending on sector, country etc. The range of variables will be presented below.

2.2.1 Response variables

The fundamental ecological response variable is the species. Modelling of individual species is conceptually appropriate, because the direct relationship between response and predictor variables is the potentially least complex. The habitat of a certain species is then defined by the most relevant environmental predictors for that species. Focus-ing on the species level agrees with ecological awareness that the distributional dy-namics of species are typically independent of each other, and that communities or as-sociations are unlikely to move as an entity along temporal and spatial gradients (Guisan & Zimmermann 2000). However, mapping of pre-defined habitat types has a high applied relevance and is the most common basis for management decisions. Using the species approach, other re-sponse variables, such as species richness and certain community or assemblage types, may be derived by compiling maps of individual species distributions (Lehmann et al 2003, Austin 2002). The species approach is also motivated by the fact that many habitat types are defined by the presence of major structuring species. Thus, maps with full coverage of structuring species are an important input in for example the Fin-nish national habitat classification system BalMar, and the EUNIS classifications. For species with discrete life-stages, which are ecologically as well as spatially distinct, a combination of several stage-specific models may be required for complete habitat definition. For example, several fish species have a number of well-defined life-stages, which are separated by physiological or ontogenetic shifts leading to changes in habitat preference and, thus, spatial distribution. In other cases, sufficient data on species dis-tributions may not be available, and the distribution of functional groups, habitats or other ecological units of organisation is used as the primary focus for mapping.

2.2.2 Environmental predictor variables

Statistical models and criteria analyses are based on observations on correlations be-tween a response variable and a set of environmental predictor variables. Thus, spatial predictions may potentially be achieved at high precision also in cases where the model has no direct ecological process basis. However, the spatial and temporal ro-bustness of purely correlative predictions are potentially weaker than predictions based on relevant underlying ecological knowledge. Environmental predictors may be defined as being either proximal or distal, based on the position of the predictor in the chain of processes that links it to the species. Using a similar concept, environmental gradients may be defined into three idealized and not exclusive categories (Guisan and Zimmerman, 2000, Austin 2002):

1. Indirect gradients, based on variables that have no physiological effect on

growth or competition (for example depth, latitude). These are typically de-scribed by distal variables.


6

2. Direct gradients, based on variables that have a direct physiological influence (for example temperature, salinity and oxygen). These predictor variables can be either proximal or distal.

3. Resource gradients, based on variables that are consumed (ex light, prey, nu-trients). These are typically described by proximal variables.

Models based on proximal variables are potentially more robust and general, whereas species distribution models using distal variables will often primarily be of local value (Austin 2002). However, the relative importance of distal and proximal variables varies with ecological context. Proximal variables are usually more successful predictors where the environment changes slowly and a species occupies its optimal realized niche. Where gradients are steep and environments are extreme, simple distal vari-ables will be as successful as environmental process based models (Austin 2002). Sometimes, GIS coverage for proximal variables are not easily obtained, which in-creases the relative importance of distal predictor variables in habitat modelling. In these cases, distal variables may be used to replace a combination of different proxi-mal variables in a simple way (Guisan and Zimmermann, 2000). For example, wave exposure can be used as a distal variable assumed to at least partially replace proxi-mal variables such as water temperature and bottom substrate, but wave exposure may also be a relevant proximal variable for some species. The shape of the response function will vary with the nature of the environmental pre-dictor variables. Responses to a predictor variable describing an indirect gradient could take any shape. For example, two predictor variables with unimodal shape may cause a bimodal shape in the response variable. In biological variables, skewed distributions are commonly seen, as different predictor variables may affect the abundance of the response variable at different ends of its range of distribution. Additionally, environ-mental predictor variables may interact which can also influence the response curve.

Table 3 Main environmental gradients structuring the distribution of marine species in the BSR. Many variables may be considered either proximal or distal, depending on context, as exemplified in footnotes for the variables nutrients and wave exposure. Scale of prediction Predictor type Larger Smaller Proximal Distal Climate X X Depth X X Latitude X X Light X X Nutrients X X1 X2

Oxygen X X Salinity X X X Substrate type X X Temperature X X Wave exposure X X3 X4

1for primary producers 2for secondary producers 3for species/habitats affected by physical disturbance 4when used as a proxy for, e.g. temperature (see text, section 2.2.2)


7

3 GENERAL GUIDELINES FOR DATA INTERCALIBRATION

When large-scale maps are based on data originating from different surveys, potential differences in the approaches used for measuring and estimating data variables should be considered before pooling data sets. Initial evaluations should verify that data sets are compatible regarding sampling strategy, method efficiency and data format. Data sets may then be either pooled directly, after data processing, or after data reduction. As an alternative, datasets collected by different methods may be used for a model building dataset and a validation dataset, respectively. For validation of models focus-ing on relative abundances, intercalibrated data sets may not be required. However, this is only advisable for methodologically similar datasets that do not differ in catch ef-ficiency (see section 3.3).

3.1 Data types

The following terms are used for data types: 1) numerical, 2) ordinal, 3) nominal includ-ing presence/absence, and 4) presence only. Data types should be compatible among data sets and, in case of modelling, with the particular assumptions in the applied analytical approach. In the case of ordinal and nominal data, common intervals and definitions should be applied or developed. Data sets using different data types should be intercalibrated by data reduction.

3.2 Method efficiency

Differences among data sets regarding survey design and sampling methodology may affect the potential accuracy of quantitative estimates, and which data formats to apply. Methods should be compatible regarding gear efficiency, and size of area sampled. This may be verified by referring to common standards (e.g. HELCOM or ICES rec-ommendations), or alternatively, a field intercalibration of gears may be used or re-ferred to in order to ensure compatibility. Field intercalibration may be used to estimate quantitative differences among methods, and subsequently to recalculate estimates and combine datasets. For ordinal and nominal data formats, and in cases where different methodological ap-proaches clearly vary in their level of efficiency, data reduction may be a means for en-suring a common level of accuracy. Special attention should be paid to the efficiency of different methods in estimating absences, since this is highly dependent on the area sampled, sample efficiency and species in target. In such cases, data reduction to presence only should be considered (Zaniewski et al. 2002). A definition of terms is presented in table 4.


8

3.3 Definitions of terms related to marine habitat mapping

Table 4. Definitions of terms related to marine habitat mapping Term Definition Accuracy The degree to which an estimated mean differs from the true mean Assemblage Coexisting populations within the same trophic level

Classification The grouping of a data set or a set of variables into predefined cate-

gories Community A group of populations that coexist and interact

Criteria analysis A non-statistical overlay analysis based on criteria developed within

the current and/or earlier studies Data reduction Removing spatial or numerical details in a data set

Data types The classification of data according to their characteristics

Environmental envelope The realized ecological niche of a species

Georeferenced data Data with spatial attributes, such as position coordinates

Habitat Ecological niche referring to a certain species and life stage, e g habi-

tat for perch, spawning habitat of cod or sprat larvae Habitat types Environment with similar structural characteristics and composition of

associated species. Intercalibration The process of correcting data collected by different means so that

they may be comparable with each other. Interpolation The estimation of values of a variable at non-sampled sites from

measurements made at surrounding sites Layer A logical separation of mapped information according to themes, can

be in the form of vector or raster data. (syn. GIS coverage) Marine landscape Geological and biological large-scale spatial characteristics of the ma-

rine environment Nominal data Data type where the variable assumes one of several alternative

states, e g presence/absence or turbid/clear Numerical data Data type where the variable assumes continuous values or values

with equal intervals from each other Ordinal data Data type where the variable assumes values with unknown or un-

equal intervals, e.g. temperature. Overlay analyses Spatial analyses where data is interpreted by superimposing several

thematic layers on each other Population Interbreeding individuals of the same species in a given area

Resolution The smallest size that can be mapped or sampled as one unit. Also

applicable on GIS raster layers when referring to pixel/grain size. Spatial interpolation Spatial predictions based on information present only in the variable

itself and its position Spatial modelling Spatial predictions based on statistical models


9

3.4 Data sets used in BALANCE: Compilation of method descrip-tions

A questionnaire on sampling methodology was sent out to all involved BALANCE part-ners, and was responded on by May 2006. In order to ensure compatibility among method descriptions, a common template for biological and geological data collection was designed at the workshop in Uppsala 8-10 November, 2005. The focal points of the questionnaire were to address: 1. Technical descriptions of gears and instruments 2. Sampling designs & comparability of methods from a habitat mapping perspective 3. Suitability for habitat modelling Details of the obtained method descriptions are presented in appendices 1-5. The main part of the descriptions was allocated to either of four main target biota; fish, macro-fauna, zooplankton and phytobenthos (Table 5). These were submitted for further evaluation to experts of each field, according to common guidelines, as described in table 6. Method descriptions for phytoplankton and hydrographic data were additionally provided. These are described in the appendix.

Table 5. Grouping of biological data collected Target biota Methodological sub-groups Number of

methods de-scribed

De-scribed within pilot area

Evaluator

Fish Spawning habitats 2 1, 2, 3 DIFRES, Egg habitats 2 2, 3 SBF Larvae habitats 2 2, 3 Nursery areas 6 1, 3 Foraging habitats 3 2, 3 Macrofauna Soft bottom macrofauna 5 1, 4 EMI Zooplankton Zooplankton 2 2, 4 IFM-GEOMAR Phytobenthos Macrovegetation and sessile

animals on hard substrates 7 2, 3, 4 NIVA

Additional method descriptions provided Phytoplankton Species composition 1 4 - Hydrography Temperature, oxygen, salinity 3 2 -

Table 6. Points addressed in the evaluation and harmonisation of methods for data col-lection. Principles for data intercalibration (focal point number 3 in Table 6) concern cases where more than one method was described 1. Biological aspects General biological characteristics

General method requirements 2. Methods for data collection Description of gear, methodology, special adaptations and

details. 3. Data intercalibration Identification of data sets to intercalibrate; Evaluation of how

to intercalibrate data sets; Recommended data format, accu-racy, and spatial resolution

4. Data applicability for habi-tat mapping

May the data be spatially generalized in order to be represen-tative of larger areas, and if so, how? Do the data sets con-form to the modelling approaches preferred? Main environ-mental data required and its availability

5. Suggestions for forthcom-ing data collection

How should forthcoming data collection be modified/ de-signed in order to facilitate mapping and modelling?


10

4 HARMONISATION OF FISH DATA

In order to harmonise fish information for a marine region such as the Baltic Sea Re-gion, it is important to take biological aspects, sampling methodologies and how raw data are handled and interpreted into account.

4.1 Biological aspects

The horizontal and vertical distribution of different fish species depend on whether they are pelagic or demersal, occupy coastal or open water, or if they prefer fresh or saline water. In addition, different life stages, such as the eggs, larvae, juveniles and adults of various fish species may be associated with different ecological niches and habitat characteristics (hydrograhy, bathymetry, substrate etc.). For example for cod, the eggs occupy the mid water layer in the deep basins, the larvae live in the open sea surface layers, juveniles inhabit the layer near bottom in shallow water so-called nursery areas, while immature and adult cod live near the bottom at deeper water depths. In addition, the foraging and spawning habitats of the adult fishes often differ causing seasonal mi-grations between different areas. One of the scopes of BALANCE is to identify spawning and nursery areas, which are important for recruitment of fish species and important aspects when designating MPAs (Marine Protected Areas). The location, size and seasonality of the spawning and nursery areas will be defined by mapping distributions of eggs, larvae and juvenile fish stages. Also, the preferred foraging areas of adult fish in comparison to spawning ar-eas will be mapped. In all cases, the spatial and temporal distributions of different life forms depend on environmental variability influencing the size and location of habitats over time. The choice of survey design, methods and gears, timing and frequency consequently differ among species and life stages (Table 7 & 8). For example, the abundance of ju-venile fish in nursery areas is both temporally and spatially highly variable, which af-fects sampling efficiency at the scale of hours to seasons.

4.2 Description of methods for data collection

Diving surveys Used by SBF. Surveys of perch (Perca fluviatilis L) egg strand abundance in bays and lagoons are conducted in spawning time during April-June. Mapping of egg strands is performed by snorkelling along parallel transect lines drawn perpendicular from the shore to the opposite shore. The first line is placed 5 m from and parallel with the inner-most shore, or as near as possible of it if the depth conditions allow it. The second line is placed at 50 m distance from the first one, the third at 100 m distance from the sec-ond one, and the following ones at 100 m distance from the previous one until the entire area is surveyed. Additional lines are placed 5 m from the shoreline between the starting and end points of the first and second line, respectively. All egg strands within one me-tre on both sides of the transect lines are registered. The substrate on which the strand is attached, depth and distance from nearest shore is noted. Temperature is measured at start, mid-point and end of each transect line 20 cm above the bottom.


11

Table 7. Potential biological layers in the mapping of fish, biological characteristics that relate to differences in sampling strategy, and methods used for data collection Potential habitat layer

Potential species layers

Temporal varia-tion within a year

Main habitat Method

Cod, subadult and adult

Constant/seasonal Demersal, depth > 30 m

Demersal and pe-lagic trawl surveys

Sprat, herring

Constant/seasonal

Pelagic, basin slopes

Hydroacoustics, pelagic trawl sur-veys

Foraging habi-tats

Perch, pike, whitefish, sander, roach, bream (>8cm)

Sample scale (hours-days), seasonal

Coastal areas 0-20 m depth

Nordic nets of coastal survey type

Cod

Seasonal

Pelagic/demersal

Demersal and pe-lagic trawl surveys

Spawning habi-tats

Sprat Seasonal Pelagic, open sea Hydroacoustics Pelagic trawl sur-veys

Perch egg strands

Seasonal

Sheltered, partly vegetated, 0-6 m depth

Diving survey Egg habitats

Cod eggs Seasonal Pelagic open sea Bongo survey

Larvae of pike, roach, burbot

Sample scale (hours-days), Seasonal


White plate and scoop

Larval habitats

Cod larval stages

Seasonal

Pelagic, open sea

Bongo survey

Juvenile plaice, sole, flounder, turbot


Sandy/muddy, 0-6 m depth

Beamtrawl, Drop trap, Pushnet, Trawl sampling

Nursery areas (juveniles, age 0+ and 1+)

Juvenile perch, pike, roach, sander, bream



Beach seine, LIPS

Juvenile white-fish


Moderately ex-posed, sand, boul-ders, 0-6 m depth

Beach seine, pot. LIPS

White plate and scoop Used by FGFRI. This method is used to sample and observe pike, burbot and cyprinid larvae among and in shallow vegetated shores. The plate is a 20*30 cm white plastic plate fixed to a 1 m long arm. The plate is slowly moved at a depth of 10-40 cm and the typically 13-25 mm long larvae are easily detected against the white background. The scoop is an ordinary white 2-3 l water scoop, which is used for sampling and counting larvae against the white background. Surveys are performed during the latter half of


12

May and June. Sampling sites are randomly selected 100 m long shorelines, and the method produces data on presence/absence of larvae on the sites.

Drop-trap This gear can be designed in several ways (see Pihl & Rosenberg 1982, Wennhage et al. 1997 for different designs) and is used not only for flatfish sampling but also for various benthic organisms. At DIFRES the size of the trap is 1 m2 whereas in other studies a size of 0.5 m2 has been used. It is generally assumed that the efficiency of the gear is 100 % and that it is non-selective for different fish sizes. It is possible to ac-complish 5-10 samples per hour (Elliot & Hemingway 2002). The sampling procedure is as follows. The drop-trap is carried out to the desired locations and quickly dropped at the bottom. Then the area within the drop-trap is fished through several times with a small net in order to catch all flatfish enclosed by the drop-trap. The use of drop-traps is restricted to smooth bottom substrates.

Beam-trawl There are several different designs of beam-trawls. The size and weight of the beam has been altered as well as the numbers and sizes of chains in front of the beam. (see Riley & Corlett 1966, Kuipers 1975, Rogers & Lockwood 1989, Kuipers et al. 1992, Kaiser et al. 1994, Wennhage et al. 1997 for different designs). Besides different de-signs the beam trawl can be used from a boat or it can be pulled by hand in shallow ar-eas. The beam-trawl used by DIFRES is 2-m wide and has a 7.4 m rope at each end of the beam for dragging of the gear. In front of the gear an iron-chain is mounted in order to chase the flatfish up from the sediment. This chain also has the purpose of smooth-ing the bottom and thereby ensuring the best contact between the gear and the bottom. The use of beam trawls is restricted to relatively smooth bottom substrates.

Push-net (Standard and Improved) Even though the push-net was originally designed to catch shrimps it has also been used for sampling juvenile flatfish in the coastal zone (Elliot & Hemingway 2002, figure 1). It has the advantage of being very easy to use and that the sampling can be carried out by a single person. A commonly used push-net is the Riley push-net (Elliot & He-mingway 2002), which is 1.5 m wide. This gear resembles a beam-trawl but is pushed instead of being dragged. Push-nets are divided into two different types, one is a stan-dard push-net designed for catching shrimps whereas the other one is a push-net modified by Else Nielsen from DIFRES in order to improve the efficiency for catching flatfish. The two different push-nets are both 63 cm wide, the only difference being that the standard push-net has a sharp edge, and the improved push net has a round edge, which makes it easier to keep a steady fast speed. The use of push-nets is restricted to smooth bottom substrates.

Beach seine Used by SBF and FGFRI. A beach seine is an active gear designed to catch small fishes and shrimps. It is made up by two long leading net-arms joining in a cod end with a slightly lower mesh size. The beach seine is pulled towards the shoreline on relatively smooth substrates. It is possible to use at a low or moderately low coverage of vegeta-tion, although it has the disadvantage of damaging some species of vegetation to a smaller extent. The beach seines used by SBF and FGFRI are 2 m deep, have 10 m long arms and 5 mm mesh size in the arms and 2 mm in the cod end.

Juvenile trawl In fishing for juvenile flatfish in the coastal zone with boat a standard juvenile trawl is towed for 10 min at 1 knot (40 m min-1). Total length of the net is 10.8 m and the width


13

is 4.5 m. The trawl is attached directly to 14 kg otter-boards measuring 85x50 cm sepa-rated by a 5.5 m long chain. The height of the opening is estimated at 36 cm. The body is divided into three sections with decreasing stretched mesh size of 10, 8 and 6 mm. The stretched mesh size of the cod end is 5 mm. The use of the juvenile trawl is re-stricted to relatively smooth substrates.

A) B)

C) D)

Fig. 2. Man-powered gears used for catching flatfish in the coastal zone. A) drop trap, B) Beam trawl, C) Standard push-net and D) Improved push-net

Low impact pressure wave sampling (LIPS) Used by SBF and FGFRI. Data on abundance of juvenile fish is collected by point sampling using small detonation-capsules (0.94g have been used at FGFRI, 0,94 plus 10g at SBF) that stun small fish within an area of c. 15-100 m2. This method allows sampling of fish sized 15-150 mm with well-developed swim bladders. A capsule is detonated at a depth of 0.5 m from a boat using a long fishing rod. The capsule is ig-nited through an ignition cord (figure 2a). After detonation, all stunned fish are netted from the surface and collected from the bottom by snorkelling/diving. Fish sampling is preferably randomly stratified by depth, wave exposure and vegetation composition. A maximum daily effort is c. 10-30 detonations depending on travelling distance between sites. To assure that sampling stations do not interfere with each other, they should be at least 30 m apart. Sampling is conducted in late July-August during daytime.


14

Fig. 2a. Low impact pressure wave sampling Fig. 2b. Bongo with Babybongo

Nordic nets of coastal survey type Used for national coastal fish monitoring by SBF. Coastal surveys of adult fish are per-formed using Nordic nets of coastal survey type. The nets are 1.8 m deep and 45 m long, and are composed of nine sections of five meters length, with mesh sizes of 10, 12, 15, 20, 24, 30, 38, 47, and 60 mm. The sampling design is random stratified within depth intervals of 0–3 m, 3–6 m, 6–10 m, and 10-20 m. The main part of the surveys is performed within two weeks in August each year. Ten-fifteen stations are fished at each depth interval, except for the deepest stratum where 5 stations are fished,. Nets are laid at 14-17 pm and taken up at 7-10 am the following day.

Bongo with Babybongo Gear used for sampling of ichthyoplankton (and zooplankton) used by IFM-GEOMAR and DIFRES (figure 2b). It consists of 4 net frames (2x0.6m diameter plus 2x0.2m di-ameter), with the large nets normally equipped with 300 and 500µm mesh size for sampling fish eggs and larvae, while the small nets with 150 and 50µm mesh sizes are used to sample different life stages of zooplankton organisms. The Bongo is towed ap-plying double-oblique tows within five meters of bottom or to a maximum depth of 200 meters with a ship speed of 3 knots. Filtered volumes are measured either by me-chanic or electronic flow-meters or estimated via deployment time. Sampling in the Bornholm Basin is performed on a regular station grid covering the deep basin limited by the 60m depth line at a distance of approximately 10nm. Between 3 and 6 surveys per year are conducted, encompassing the main period of productivity (cod & sprat, sprat spawning time) from April to August/September.

Demersal trawl Bottom trawling is conducted with different trawl gears applying bobbins, enforced bot-tom etc. usually towed for 30-60 min at 3.5 knots. The trawls have different proportions depending on the size of the research vessel and surveys type. The International Baltic trawl survey (BITS) coordinated by ICES use a standard gear (TV3 trawl) since 1999. The surveys are intercalibrated and the present gear has been calibrated against the formerly used gear types that varied among countries. Commonly used to fish fishes inhabiting the zone close to the bottom, e.g. cod, flatfishes, etc. Area and volume sam-pled by trawling depends on the gear type and trawl time.

Pelagic trawl Pelagic trawling is conducted with different trawl gears usually towed for 30-60 min at 3.5 knots. The trawls have different proportions depending on the size of the research vessel and surveys. The International Baltic Acoustic Survey coordinated by ICES use


15

a standard gear to catch fish species inhabiting the upper and midwater layers (e.g. sprat and herring) obtain biological data to validate the hydroacoustic measurements. Area and volume sampled by trawling depends on the gear type and trawl time.

Hydroacoustics Hydroacoustic survey methodologies are used to provide rapid estimates of fish stock distribution and biomass independent of fisheries data. The method surveys pelagic fish species that swim in large schools utilising most of the water column, in contrast to demersal species. The method is based on instruments, which emit pressure waves in the water and detect part of this energy reflected by fish targets. Hydroacoustic surveys are routinely conducted as part of the Baltic International Acoustic Surveys (BIAS) to estimate the abundance of sprat and herring. The survey methods are described in the Manual for the Baltic International Acoustic Surveys, or BIAS (ICES 2003). Table 8. Sampling methods used and details of data obtained Method Obtained

variables Obtained unit Area covered by

sample Pilot area involved

Diving survey Presence of egg strands

Presence/absence 0.5x0.5m 3


Catch rates Presence/absence ~100 m2 3

Drop trap Abundance n/m2 1m2 1 Beam trawl Catch rates n/effort 50m2 1 Push net Catch rates n/effort 20m2 1 Beach seine Catch rates n/effort 100-200m2 3 Juvenile trawl Catch rates n/effort 1800m2 1 LIPS Catch rates n/m2 16 and 100m2 3 Coastal survey nets Catch rates n/effort - 3 Bongo Abundance n/l Volume water fil-

tered is measured 2

Demersal trawl types Catch rates n/effort Varies 2 Pelagic trawl types Catch rates n/effort Varies 2 Hydroacoustics Biomass t/volume Varies 2

4.3 Data Intercalibration

Intercalibration of fish data needs to take into account all the habitats where a species can be found in various life stages.

4.3.1 Nursery areas Estimating fish abundance in nursery areas from their catch-rates is complex due to high temporal and spatial variability of the target species, and due to high variability in the efficiency of different gears. Although the abundance of the target species may be estimated as numbers per area or per known effort (Table 9), the main target species may differ and the relationship between observed catch-rates and the actual abun-dance of fish in an area is unknown in most cases. One way of dealing with this problem is to estimate the relationship between the catch-rates of different gears by intercalibration experiments. Calibration experiments are quite straightforward as long as the gears calibrated are in use at the same location. In an intercalibration study carried out by DIFRES, the efficiency of the most commonly used gears was estimated to vary between 10% and 100%, depending on gear type, but also on the physical characteristics of the environment, such as depth, sediment, and in some cases wave-height. As a consequence, potential differences in method ef-


16

ficiency should be considered not only when comparing different gears, but also when collating data collected by the same gear but in different areas. Method efficiency varies among target species and may also vary with the size of the target species. In a study with releases of a known number of turbot, Sparrevohn and Støttrup (in press.) estimated that the efficiency of a boat-driven young-fish trawl was approximately 50 % for sizes 7-12 cm and decreasing to around 10 % for 17 cm large turbot. Similar results have been found in various other studies with flounder and plaice. In some cases, intercalibration of data sets is not at all possible, because several gears and methods cannot be used on all types of substrates. As an example, trawling may not be used in areas where sea-grass is abundant. Intercalibration experiments are po-tentially realistic among gears that target the same species and habitat types, provided that local environmental differences may be accounted for in the intercalibration design. An estimation of the catch efficiency of different methods in relation to target species, size of target species, bottom substrate, season and vegetation is presented in 9. When catch efficiency is strongly affected, data reduction to presence only is recom-mended. In the case of similar target species and low effects on catch efficiency, data sets may potentially be pooled quantitatively however this should be done with caution and only following situation-specific intercalibration.

Examples of method efficiency evaluations Within DIFRES, intercalibration experiments have been conducted regarding gears used for collecting individuals and sampling abundance of juvenile flatfish in the coastal zone. The methods compared were man-powered (drop-trap, beam-trawl and push-net) and vessel-powered (juvenile trawl) methods. These gears are typically used at different depths. Manpowered method can only be applied in the very shallow areas of less than 1-1.5 m depth, whereas vessel powered methods may reach a sample depth of 1 meter at the shallow end, depending on the boat used. However, in general, man-powered methods are recommended for depths less than 1.5 meters. In the intercali-bration experiment, it was assumed that the drop-trap catch 100 % of all fish. The effi-ciency of the other gears in relation to this was estimated to approximately 9% (SD 9.4) for beamtrawl, 31% (SD 18.9) for standard push-net, 28 (SD 26) for improved push-net and 29% (SD 21) for beam-trawl.

The efficiency of low impact pressure wave sampling (LIPS) in sampling perch, pike and roach was evaluated for vegetated and non-vegetated areas by Snickars et al. (submitted manuscript). Method efficiency was measured as the effective area in which >95% of the individuals present were affected, and varied from 7 to 28 m2 using a 0.94g detonator. The study indicates that calibrations among areas should be done with reference to sampled depth, presence or absence of vegetation and detonation size.

Table 9. Target species of the different gears used in surveys of fish nursery areas, together with indication on whether size of the fish caught, bottom-substrate, and vegetation are likely to affect gear catch efficiency (+) or not (0). Method Species Size Substrate Vegetation


17

Diving survey Perch and other species with easily recognisable egg strands

Ns + 0


Larvae Ns 0 0

Drop trap Mainly flatfish

0 + +

Beam trawl Mainly flatfish

+ + +

Push net Demersal species

+ + +

Beach seine Demersal species

+ + +

Juvenile trawl Demersal species

+ + +

LIPS Fish with swim bladders

+ 0 0

Coastal survey nets

The majority of occurring species

0 0 0

4.3.2 Foraging and spawning areas (Pilot area 1, study area 2) Data intercalibration is needed in order to estimate population size as absolute density or abundance indices based on catch rates from fish surveys conducted by different vessels and using different gear types, sampling strategies etc. The Baltic International Trawl Surveys (BITS) and Baltic International Acoustic Surveys (BIAS) are conducted in spring and autumn by different research vessels from the various countries sur-rounding the Baltic Sea and Kattegat to obtain fisheries independent data for stock as-sessment of cod, herring and sprat. The surveys, which originally used different gears and methodology, are since 1999 coordinated by ICES and use similar gear and stan-dardised methods. (ICES, 2003). The vessels, gear and coverage differ calibration ex-periments have been conducted and calibration factors are used to calibrate catch rates per species, size etc. These data can also be used on a temporal scale to com-pare distribution of cod, sprat and herring over a range of years in relation to environ-mental changes determining their habitat volume and quality. Data from Danish and German fish surveys not related to BITS and BIAS are addition-ally used to map cod spawning areas and spawning time. Comparison of relative com-position catch rates from different surveys can be made without calibration of gear catchability and efficiency as long as the sampling strategy used is similar, e.g. per cent female cod in spawning condition per haul can be plotted for an area and be com-pared to similar data from other surveys, in order to identify spawning locations and timing of spawning independent of trawl gear used. The sub-sampling of fish from the catches may differ among surveys, e.g. representative sampling of the catch versus systematic sampling by length groups) but this can be corrected by up-weighting the samples to total catch. Sampling of ichthyoplankton using Bongo and Babybongo is made using standard gear in a standardised way and data are independent of vessel.

4.4 Data applicability for habitat mapping

Data on fish abundance collected using different methods and gears should generally be viewed as not directly quantitatively comparable. This is mainly due to differences in


18

the target species and overall catching efficiency, as explained in the sections above (figure 3a-d). Because of high spatial and temporal variability of target species, method efficiency varies with environmental setting, and quantitative comparisons between studies using the same gear should also be done with caution. Also, data on fish abun-dance are often highly skewed with non-normal frequency distributions, because sev-eral species and/or life stages have a strong schooling behaviour that results in strongly zero-inflated catch data. The schooling behaviour may also be dependent on the environmental conditions. As an example, juvenile fishes have been found to alter their tendency to school as a response to changes in habitat complexity (Rangeley & Kramer, 1998). Thus, the following recommendations may be made:

1. Generally, data reduction to “presence only” is recommended for large-scale

mapping. 2. In all instances where the predictor variable influences the schooling ten-

dency of the target species, presence/absence models may be seriously bi-ased and should be used with care.

3. Presence/absence data may be applicable for methods with similar target species and catch efficiency.

4. Quantitative pooling is only recommended on smaller scales and after situa-tion-specific intercalibration exercises.

5. In statistical spatial modelling, presence/absence data or presence only data of good quality is preferred over more detailed data of uncertain relevance, as the estimated output may still be expressed as the numerical probability for occurrence of the predictor variable.

4.5 Suggestions for forthcoming data collection

Standards for fish sampling methods are under continuous development within ICES and EU, to ensure use of data beyond the national and regional level. New standards are in progress for monitoring of fish communities in lakes and streams and standards for trawling of pelagic and benthic species in off-shore areas with larger vessels are developed and set within ICES. However, similar standards for monitoring and mapping of young fishes in shallow coastal areas are not available. Since the sampling techniques are highly specialised for certain species or habitats, such standards are harder to develop for shallow coastal areas. A first step towards harmonisation would be to conduct transnational in-tercalibration efforts, comparing and evaluating fish sampling surveys in shallow coastal areas within the BSR. Although such an attempt is relevant and highly recom-mended, it would require a large effort and should be conducted over a longer time pe-riod and is thus not possible to include in BALANCE. Currently, for shallow nursery areas, the sampling methods drop-traps, LIPS and coastal survey nets (gill nets) appear the most general by the aspects analysed in 9. However, these are limited in their potential areas of use, and it may not be possible to agree on one single gear type or method to be used throughout the Baltic Sea. Drop-traps can only be used in very shallow waters, and LIPS will only catch fish with swim bladders, which means that flatfish will not be caught. Gill-nets, again, do not easily al-low quantitative abundance estimates as the catch method is passive, which means that the catch rate of different species will depend on their level of activity in the survey area. Also, gillnetting can be very time consuming during the summer season in the more saline areas of the Baltic Sea due to large catches of crabs. In most cases, small trawls or push-nets are preferred for sampling at sandy beaches, as these gears are


19

generally quite efficient and economically feasible. However, they cannot be used at the whole range of different bottom types and vegetation coverage. From the perspective of habitat mapping, fish surveys that are a priori designed for sampling of the most likely habitats of the target species are relevant for mapping, but have limited value for modelling unless the survey also covers habitats were the target species is less likely to be found. This is because the predictive strength of the statisti-cal models relies on full inclusion of the whole potential distributional range of predicted species in order to properly define the environmental envelope, and not only the opti-mal habitats. Thus, developing methods that enable sampling of fish in the majority of potential habitats in the BSR is thus an important challenge for future fisheries related research in coastal areas.

Fig. 3a. Gadus morhua, an adult cod in a pe-lagic habitat. Photo from www.fishbase.de.

Fig. 3b. Esox lucius, a juvenile pike caught in a near-shore habitat. Photo: The Swedish Board of Fisheries.

Fig. 3c. Trigla lucerna, a red garnet in a deep sea mud habitat. Photo: Orbicon.

Fig. 3d. Pleuronectes platessa, a flatfish in a shallow water macroalgae habitat. Photo: Or-bicon.

http://www.fishbase.de/


20

5 HARMONISATION OF MACROFAUNA DATA

Macrozoobenthos, or benthic macrofauna, is defined as benthic invertebrates that are retained on 1 mm sieve. Benthic macrofauna represent important ecological functions and contribute significantly major ecological processes in the Baltic Sea (figure 4). Ad-ditionally, benthic fauna species have a value as indicators in water quality assessment and in modelling of habitat types.


Three main components may be considered as potentially useful parameters in map-ping and modelling, as described in table 10 and below.

1. Key macrozoobenthic species and/or functions that can be used to pre-dict various habitat types in the Baltic Sea area As a group, benthic macrofauna is functionally very diverse and includes practically all major animal phyla. However, as the Baltic Sea is isolated, has short developing time, low salinity and temperature, only a limited number of macrozoobenthic species have been able to adapt to the local conditions. A mixture of marine and lacustrine organ-isms characterizes the communities. Specific brackish-water or endemic forms are rare. In its northern and north-eastern ends the number of benthic invertebrate species is particularly low and often each ecosystem function is represented by a single spe-cies. Thus, the loss of a species may correspond to the loss of ecosystem function (Segerstråle 1957, Järvekülg 1979, Hällfors et al. 1981). As the number of benthic in-vertebrate species is low, each ecosystem function is often represented by a single species. Thus, modelling of the distribution of the key species gives good representa-tion of the distribution of the habitats of interest. For example certain herbivorous inver-tebrate species are only found within Zostera marina meadows and the invasive am-phipod species Gammarus tigrinus indicates the presence of the endangered charophyte communities.

2. Species that may be used as biological indicators It is essential to develop biodiversity and eutrophication relevant indicators for the BSR. This can be done by statistical analyses assessing responsiveness and robustness. Links between ecosystem components can be analyzed using multivariate statistics and/or modelling. Multivariate statistics can also provide insight into the performance and statistical power of the data available and show which biodiversity components are essential for inclusion in monitoring activities. The ecosystems of the Baltic Sea are very dynamic and characterized by high physical and biological disturbances. Wave induced currents and ice scraping are the prevailing physical disturbance in the shallower areas and semi-natural periods of hypoxia/anoxia in the deeper areas (Kotta et al. 1999, Laine et al. 1997). Eutrophication induced blooms of phytoplankton and macroalgae and their decomposition are ranked among the most severe biological disturbances affecting local invertebrate distribution (Norkko & Bonsdorff 1996ab, Paalme et al. 2002). Benthic invertebrate communities represent an intermediate trophic level and nutrient additions affect them in many ways. According to the Pearson-Rosenberg model (Pearson & Rosenberg, 1978) the bio-mass of benthic invertebrates increases gradually to a maximum as the load of organic matter increases. After this, the biomass falls and often shows a secondary peak but lower than the first maximum. Increasing nutrient loads enhance the production of ben-


21

thic and/or pelagic microalgae (Granéli & Sundbäck 1985, Howarth, 1988) and thereby increase available food for benthic grazers, suspension feeders or deposit feeders. As a consequence, abundance and growth responses of invertebrates are observed from an initial increase of nutrients (Posey et al., 1999). Further eutrophication leads to hy-poxia and the disappearance of benthic invertebrates. Thus, benthic communities are highly sensitive to eutrophication, which makes them a good indicator of water quality (Pearson & Rosenberg, 1978, Grall & Chauvaud, 2002, Gray et al., 2002). The effects of eutrophication are more pronounced in coastal areas where they are ex-pressed as an excessive growth of filamentous algae (Rosenberg 1985, Hull 1987, Gray 1992, Kolbe et al. 1995), changes in herbivore assemblages (Kotta et al., 2000), and the development of dense populations of filter-feeding mussels (Barnes & Hughes 1988, Kautsky et al. 1992, Kautsky 1995). Owing to their large filtration capacity, popu-lations of filter-feeding mussels are able to filter major parts of the water column each day (Riisgård & Møhlenberg 1979, Kautsky & Evans 1987), and thereby directly control the standing stock of pelagic primary producers. Consequently, filter-feeders are con-sidered to play a key role in the stability of coastal ecosystems (Herman & Scholten 1990).

3. Invasive species Biological invasions are considered another key factor affecting the dynamics of ben-thic fauna (Jansson 1994, Olenin & Leppäkoski 1999). Due to environmental instability, a low number of species and an increasing intensity of freight transportation, the eco-system of the Baltic Sea is very exposed to invasions, and biological invasions have resulted in relatively large-scale ecological changes. Examples from invasions in the 1980s and 1990s have shown that successful invasive species may render previously stable systems unbalanced and unpredictable (Leppäkoski 1991, Carlton & Geller 1993, Mills et al. 1993, Carlton 1996, Ruiz et al. 1999) and may severely affect biologi-cal diversity in an area (Baker & Stebbins 1965, Gollasch & Leppäkoski 1999, Gollasch et al. 1999). A number of benthic animals that presently live in the BSR have only re-cently invaded the area, some of them only in the last decades (Kotta 2000, Lep-päkoski and Olenin 2001). A few of the non-native animals add unique ecological func-tions for the species-poor Baltic Sea ecosystem (Leppäkoski et al. 2002) whereas others share the same food resources with the local species and, thus, may reduce the native biological diversity (Kotta et al. 2001, Kotta and Ólafsson, 2003). Among the most recent newcomers is the amphipod Gammarus tigrinus which has caused strong impacts on the Baltic Sea ecosystem including the disappearance of native amphipods and potentially a change in fish diet (Kotta, unpublished data).


22

Table 10. Potential biological layers in the mapping of macrofauna, biological character-istics that relate to differences in sampling strategy, and methods used for data collec-tion. Examples are given for planned continuous benthic invertebrate layers in pilot area 4. Potential habitat layer

Potential species layers

Temporal variation

Main habitat Method

Key invertebrate species repre-sentative of pilot area

e.g. Mytilus tros-sulus

seasonal and/or an-nual

e.g. M. trossulus, exposed, hard bot-tom substrate, sa-linity above 5 psu, mainly photic zone

Core sampling by diving on mixed bot-toms, van Veen grab sampling on soft bottoms

Biological indi-ces (biodiversity, eutrophication)

e.g. biomass of deposit feeders indicating state of eutrophication

annual soft accumulation bottoms, depth 10-40 m

een and/or Ekman type grabs

Key invasive species

e.g. Gammarus tigrinus

seasonal Sheltered partly vegetated, 0-6 m depth

Core sampling by divers, Van Veen and/or Ekman type grabs


23


This section describes the usage of grabs (Van Veen, Ekman type grabs) and core samplers (remote or diver operated) in macrozoobenthos sampling (Table 11). The 0.1 m2 Van Veen grab is the standard gear for benthic macrofauna sampling in the Baltic Sea, because of its very good reliability and manoeuvrability at sea. In some cases, however, the use of devices with smaller sampling area may be appropriate, e.g. if the fauna is very dense and uniform. In areas where the burrowing depth of the fauna is beyond the penetration depth of the grabs, or in sites where that type of gear cannot be used, remote core samplers may be advisable. Alternatively, the core sampler can be operated by diver and is usually used at depths less than 20 m.

Grabs The empty van Veen grab sampler should weigh about 30 kg when used for fine grain sizes and up to 80 kg in sandy bottoms. The settling down and the closing of the grab must be done as gently as possible. Winch operation should be standardized, in order to reduce the shock wave and the risk of sediment loss as a result of lifting the grab be-fore completed closure. The wire angle must be kept as small as possible to ensure that the grab is set down and lifted up vertically. If, as often happens on sandy bottom or erosion sediments, less than 5L of sediment is collected by a van Veen grab (the critical volume is smaller for other grab types), the sample should be regarded as not quantitative, and a new sample should be taken after loading the grab with an extra weight. This may as much as double the effective sampling depth of the grab. The evi-dence of this problem may be different in different parts of the Baltic Sea, depending on, e.g., how deep in the sediment the species live. In the northern parts of the Baltic Sea, benthic macrofauna rarely penetrate sediment deeper than 10 cm except for the nonindigenous polychaete Marenzelleria neglecta. Criteria for rejection of samples col-lected by grabs are given by Rees et al. (1991) and in the HELCOM guidelines (http://sea.helcom.fi/Monas/ CombineManual2/PartC/ CFrame.htm).

Core samplers The design of the corer and its usage is described by Kangas (1972). The corer has potential to be used as standard gear in vegetated or unvegetated soft or mixed sedi-ments because of its very good reliability and because it is easy to handle underwater.

General methodological guidelines The standard sieve for the Baltic Sea area has a mesh size of 1.0 x 1.0 mm. In order to collect quantitatively developmental stages of the macrofauna and abundant smaller species, additional smaller sieve with mesh size of 0.25 x 0.25, 0.4 x 0.4 or 0.5 x 0.5 mm are recommended. On the representative stations, at least 3 to 5 samples should be taken, depending on area and species composition, to enable the investigator to reach a certain level of precision by sorting as many samples as necessary. The same procedure is strongly recommended for all other benthos stations unless another sam-pling strategy (area sampling) is employed in national/coastal monitoring programs The choice of sample size and number of samples is always a compromise between the need for statistical accuracy and the effort, which can be put into the study. One way to do this is to calculate an index of precision. The ratio of standard error to arith-metic mean may be used (Elliott, 1983). A suggested reasonable error is 0.2 ± 20%. Biomass determination should be carried out for each taxon separately. All polychaetes should be removed from the tubes, other methods have to be explicitly stated (e.g. for large numbers of polychaetes). The dry weight should be estimated after drying the


24

formalin material at 60°C to constant weight (for 12-24 hours, or an even longer time, depending on the thickness of the material). The total wet and/or dry weight of the ani-mals in each sample is usually calculated for an area of 1 m2. Sampling on shallow stations is recommended to be conducted during daytime, since some benthic species have semipelagic activity during the night. Exact positioning and correct depths when sampling should be noted in the protocols to avoid comparisons between samples taken at different localities (although noted as the same station in the protocols). Experienced and well-trained personnel are a prime basis for maintaining quality stan-dards on a high level. Allocation of resources for proper training and education of field and laboratory personnel is important. Ring tests and intercalibration exercises at least on a regional basis should be undertaken regularly basis and be obligatory. They should be open to all institutions including private industry. Technicians who carry out the actual procedures rather than managing scientists should take part in the exer-cises. Regional taxonomical workshops should be held on a regular basis and be attended by every laboratory. A checklist of species in the area should be developed, distributed to the participating laboratories and updated regularly. It is advisable, even with routine samplings, to place some specimens of each taxon under museum curatorship to make later taxonomic checks possible.

Table 11. Methods applied and details of data obtained Method Obtained variables Obtained unit Area cov-

ered by sample

Pilot area involved

Van Veen grab sampling

Abundance and bio-mass per species

n/m2 and ww/m2 or dw/m2

0.1m2 4

Ekman grab sam-pling



0.02m2 4

Core sampling Abundance and bio-mass per species


0.0143m2 1

Core sampling diver operated



0.03 m2 4

5.3 Data intercalibration

The Van Veen grab sampler combined with 1mm mesh sieving, conform to an estab-lished international standard in the Baltic Sea area. However, all methods described are basically similar and result in the same data format and units. The basic differences between gears are in the size of the sampled area and in depth penetration, which should be accounted for by intercalibration before pooling data sets. Intercalibration of gears should be done within and between regions and on the specific sediment types concerned. Alternatively, literature data on the intercalibration of methods conducted in different basins and on different sediments can be used, when appropriate. The gears described have a relatively similar catch efficiency of invertebrate species, especially for sessile groups (HELCOM, 1982; http://sea.helcom.fi/Monas/ Combine-Manual2/PartC/ CFrame.htm). Differences in grab-specific catch efficiency among spe-cies should, however be noticed in particular for species with high mobility. When com-paring different grabs and corers used in the northern Baltic Sea, the major difference is seen in that the diver operated corer has a better catch efficiency of mobile necto-


25

benthic invertebrates than the remote methods. Also, deep digging invertebrates such as Marenzelleria neglecta can basically only be effectively sampled by core sampler or Van Veen grab type sampler. Also, if datasets are pooled, the size of sieve used should be similar. However, inverte-brate biomasses may be compared even if sieves of smaller size are used, as the ma-jority of biomass is derived from larger clams, polychaetes and crustaceans.


The current report compiles the methodology of datasets used for different monitoring programs in the Baltic Sea area. The purpose of monitoring is mainly to assess the quality of water rather than to perform habitat mapping, and does not include straight-forward suggestions for the spatial resolution. Thus, the datasets obtained by monitor-ing activities need to be combined with other research surveys before any valid interpo-lation based habitat prediction can be done. Alternatively, monitoring data can be used in models, and the model based relationships between environment and invertebrate communities can be used to predict the benthic fauna in areas outside of sampling programs. Especially at shallow sites, the level of information on geomorphologic, hydrographic and biological variables is insufficient for producing reliable habitat maps. Other Euro-pean (e.g. LIFE project: 05 NAT/LV/000100) or national monitoring projects related to Water Framework Directive may partly reduce this limitation in the near future. Predicted variables As the Baltic Sea is isolated, has short developing time, low salinity and temperature the natural diversity is low in the region. Each habitat is often represented by one (or a few) benthic invertebrate species. Thus, modelling of species distribution has high po-tential to predict the distribution of habitats. After this, the distributions of species that are indicative of certain habitat types should be selected and combined in order to pre-dict the distribution of habitat types of interest. As the number of samples within the regions is often low, interpolation techniques (such as kriging) typically result in poor habitat maps. However, using existing point data is advisable in order to initially establish the functional relationships between physical and biological environment and benthic invertebrate communities. This tech-nique together with GIS modelling can be regarded as the most rewarding method for the prediction of species and/or habitats. In case there is no sufficient data to perform interpolation or other modelling the point data may be represented as such. The spatial and temporal resolution of data is essential, as different processes operate at different spatial scales, thus, modelling needs to be done at a predefined spatial scale in order to make the correct comparisons between regions. The sampling resolu-tion is far too low to cover small-scale habitat heterogeneity in most studied areas. Bio-logical processes operating at smaller scale impact the formation of habitats at larger scales. Thus, there is a need for establishing process pattern relationships at various spatial scales in order to achieve accurate modelling of habitats Environmental variables Wave exposure, fronts, salinity, hypoxia, nutrients, depth, sediment type, bottom slope are likely key environmental variables for predicting the presence and/or densities of the key macrozoobenthos species. However, knowledge on the relationships between


26

distribution patterns and ecosystem processes is currently more qualitative than quanti-tative (Hector et al. 2001). Experimental studies manipulating functional diversity and abiotic environment are needed to quantify how environmental variability modifies the relationships between distribution of species and ecosystem processes (Loerau et al. 2001). Strong fluctua-tion in abiotic factors such as oxygen concentration or wave induced disturbance is ex-pected to reduce the importance of biotic interactions within communities (Laine et al. 1997, Kotta et al. 1999, Worm et al. 2002) and physical control of ecosystem proc-esses is more commonplace (Flöder & Sommer 1999, Buckling et al. 2000). There is also some experimental evidence that physical disturbance override the effect of biotic interactions in terms of community development in the northern Baltic Sea (Kotta, unpublished data). The number of macrofauna significantly decreases with in-creasing exposure. Exposed areas with high sediment mobility are known as poor habi-tats for biota as compared to sheltered areas, which generally host diverse benthic in-vertebrate communities (Gray 2002). Thus, it is likely that the distribution of the key invertebrate species can be predicted using primarily abiotic environmental factors. Consequently, data describing the physical environment is needed to predict large scale distribution of macrofauna (fig. 5).

Fig. 5. A satellite image illustrating very high small-scale variability of geomorphologic and hy-drological conditions (represented as different colours in the figure) within the eastern part of the pilot area 4 coupled with insufficient availability of macrozoobenthos data (stars). The type of environmental data required depends on the scale of the modelling exer-cises. As an example, salinity is essential at larger scales, whereas bottom slope and exposure might be the most interesting variable at smaller scales. Also, specific needs may depend on the location of modelled area, for example ice scraping may be the prime structuring parameter in the northern Baltic Sea whereas accumulation rate of organic matter is probable in the southern Baltic Sea. On the smallest scales, diversity and structure of macrophyte communities may be the best predictor of macrofauna, especially in coastal areas. Both attached and drifting macroalgae and sedimentation of macroalgal debris are likely to modify coastal benthic invertebrate communities.


27


In general many shallow sites lack the information on geomorphologic, hydrographic and biological variables that is needed to produce reliable habitat maps. In these un-derrepresented sites, photographic and video records are recommended as a comple-ment to traditional sampling methods. Sediment profile imaging (see e.g. Rhoads and Germano (1982), Rumohr (1995)) may provide a useful means for rapid surveys and classification of sediment structure and bioturbation depth. Side-scan sonar images will provide information on bottom topography and substrate type, which can be useful in the planning of benthos monitoring programs or in the interpretation of data. Images should be verified by ground-truthing by underwater video recording and/or grab sam-pling of sediments. Descriptive and experimental studies should focus on the effects of habitat fragmenta-tion on habitat functionality and seek the relationships between habitat units, number of habitats, natural biodiversity and functional diversity. This information is needed for the investigation of potential blue corridors and for evaluation of the NATURA 2000 net-work. Macrofauna can be large and the complexity of the habitat will be reflected in the sampling methods (figure 6a-d).

Fig. 6a. Crangon crangon, a large individual hiding in a crevice in a submarine structure made by leaking gas. Photo: Orbicon.

Fig. 6b. Metridium senile, dead man’s fingers attached to the rock on a deep-water boulder field. Photo: Orbicon.

Fig. 6c. Nephros norwegicus, a Norwegian lobster in a deep-water mud habitat. Photo: Orbicon.

Fig. 6d. A log worm burrow in a sandy/soft bottom habitat. Photo: Orbicon.


28

6 HARMONISATION OF ZOOPLANKTON DATA

Mesozooplankton is the only size group of zooplankton for which data availability al-lows the mapping of their distribution. Usually, the mesozooplankton community is dominated by copepods, mainly Pseudocalanus acuspes, Temora longicornis and the group of Acartia spp., and cladocerans.


The different copepod species are characterized by differences in their life-cycle and vertical distribution (Dippner et al. 2000, Möllmann et al. 2000). Pseudocalanus acus-pes is a marine copepod, or perhaps a glacial relict, which displays on average one generation per year with peak reproduction in early spring and an overwintering stock of mainly C4-5 copepodites (Hansen et al. 2006, Renz and Hirche 2006). Further, an ontogenetic vertical distribution has been shown with the later copepodites and adults dwelling in the deep water of the Baltic basins, while younger stages occur in the ther-mocline region (Hansen et al. 2006, Renz and Hirche 2006). Long-term changes of the copepod are related to variable salinity and/or oxygen levels in the deep water, which are dependent on major Baltic inflow events (Möllmann et al. 2003, Renz and Hirche 2006). In contrast to P. acuspes, the taxa T. longicornis and Acartia spp. have multiple generations per year and produce resting eggs to overcome adverse winter conditions. These copepods are mainly distributed in the vicinity of the thermocline and their long-term development is steered by climate-induced tempera-ture changes (Dippner et al. 2000, Möllmann et al. 2000).


Mesozooplankton data are generally collected using different types of net samplers. Abundance and biomass values integrated over the water column are determined us-ing either gears towed by a research vessel in double-oblique hauls from the surface to close to the seafloor, e.g. Bongo, or in vertical hauls on a fixed position, e.g. Judai, WP-2. Vertically resolved data may be achieved using opening/closing nets (e.g. Multi-net, BIOMOC, BIONESS), but also Judai and WP-2, either towed by a ship or vertically operating on a fixed position. Values per square or cubic meter are derived by applying the filtered volume from flowmeters and the sampling depth. Different mesh sizes be-tween 50 and 300 µm are used which is contingent on targeted species or stages. In the future either modern underwater videotechnology (Video-Plankton Recorder, VPR) or sampling by ships of opportunity (Continuous Plankton Recorder, CPR) will be in-creasingly used for monitoring zooplankton. For the central Baltic Sea, different kinds of zooplankton datasets are available.

1. A long-term dataset >1960 is held by the Latvian Fish Resources Agency (LATFRA). Samples are collected seasonally (usually February, May, August and November) by a Judai-Net (net opening 0.36m; mesh-size 160µm) at varying depth intervals and on a varying number of stations (Möllmann et al. 2000).

2. Horizontally resolved samples are available from IFM-GEOMAR, based on

Bongo sampler (net opening 0.30m; mesh-size 150µm) on station grids in the Bornholm Basin with a varying number of stations per year and month since the late 1980s.


29

3. Monthly to bimonthly surveys in the Bornholm Basin have been conducted be-

tween May 2002 and 2003 using the Bongo-sampler and a 0.5 m2 Multinet (mesh-size 50µm), which provided horizontally and vertically resolved data.

6.3 Recommendations for data intercalibration

The catchability of the different zooplankton species and their stages varies strongly according to gear and mesh-size used. Consequently, no single gear/mesh-size can be recommended for mapping as a combination of mesh-sizes would be necessary to capture all species/stages representatively. A wide range of gears is thus used in moni-toring and process investigation programmes depending on institute, country or target species.


Mapping of zooplankton species is difficult due to their patchy small-scale distribution which is very unstable due to advective transport. Consequently, a spatially and tempo-rarily very highly resolved sampling would be necessary to reliably model their horizon-tal distribution. This is probably not possible with the data available from net sampling, and requires VPR or CPR sampling (see above). However, recent studies from the Bal-tic have shown that general features of the horizontal distribution of the copepod spe-cies and their relationship to the physical environment can be derived from spatial sampling (Hansen et al. 2004, Hansen et al. 2006, Renz and Hirche 2006). The datasets available have a limited suitability for spatial modelling. The data set by LATFRA cannot be used for distribution modelling for single points in time (seasons) due to the low spatial resolution of the sampling. However, by merging data sets over different periods, for example decades, general Baltic wide patterns of distribution for the main mesozooplankton species in relation to hydrography can be derived. IFM-GEOMAR datasets are more suitable for spatial modelling as they frequently represent a station grid within the Bornholm Basin. From these data species- and stage-specific species-environment relationships are in principal achievable.


Future data collection of zooplankton for reliable spatial modelling can only be con-ducted by highly resolved net sampling with different mesh-sizes and an agreed stan-dard gear, which is very costly. Future alternatives will be VPR- and CPR-sampling, depending on the progress of VPR technology and installation of CPR-lines in the Bal-tic.


30

7 HARMONISATION OF PHYTOBENTHIC DATA

Phytobenthic species include macroalgae and submersed phanerogams. Since the group consists to a large part of photosynthetic organisms, depth distribution is ulti-mately limited by light availability. However, the limiting factor of individual species may as well be limited by lack of appropriate substrate. Phanerogams and charophytes are found on muddy to sandy sediments, while macroalgae are mainly found on stable substrates such as rock and boulders. Mobile substrates are generally less favourable for phytobenthic species. Since the vegetation changes over the year, repeated studies and monitoring are generally conducted during the same time of the year in order to be comparable.


Mapping of phytobenthic species is interesting from a biodiversity perspective, but also because some species may be useful indicators of water quality. For example, some species are only found in clear waters, whereas high abundance of other species may indicate eutrophication. Also, importantly for habitat mapping, many abundant phyto-benthic species are important habitat formers in themselves, e g Fucus belts (wracks), or blue mussel beds. The distribution of large structure-forming species can generally be surveyed by coarse methods, such as UW video techniques, whereas closer obser-vations through diving or sampling are required for estimations of biodiversity and for determining the distribution of more cryptic species. A trade-off between quality and quantity may be seen among these methods, and the final choice of method will de-pend on the target taxa and the aim of the study (Table 12). Table 12. Methods applied for data collection in the phytobenthos and their po-tential biological layers in mapping Data collection method

Potential layer in map

Depth limi-tations

Ranked accuracy (1=highest)

Ranked effi-ciency (1=fastest)

Scuba diving Any species 30 m 1 4 Free diving Any species 4 m 2 3 UW-video and ROV Habitat forming

species Unlimited 3 2

Remote sensing by satellite

Habitat forming species

6 m 4 1


Scuba diving Diving surveys are used as method for registration of species on hard bottoms in sev-eral coastal monitoring programs since they are accurate and repeatable. One method is to dive along transects from deep water to the surface, observing the species vertical distributions and abundance (e.g. Norwegian program and the Swedish East coast monitoring programs). Another approach is to collect point-data at a number of posi-tions (e.g. the Danish coastal monitoring program). The inventory units are percentage cover or abundance classes (Appendix 2). Each monitoring program results in high quality data, and there are benefits and disadvantages of each approach. Underwater video and ROV


31

Underwater video equipments have developed strongly during the recent decade and are now a relevant alternative to diving for some studies. Compared to diving, under-water video is not as accurate since species can not be studied closely and verification samples can not be collected. However, much larger areas can be covered since the video may be dragged after a boat for many hours, or lowered at a large number of sites. Remotely operated vehicles (ROV) are capable of collecting data of higher qual-ity than UW-video due to their manoeuvrability, but the obtained data is still of poorer quality than data collected by diving. On the other hand, a ROV can not cover as large areas as the UW-video. Free diving Free diving may be as accurate as SCUBA diving, but are only reliable at shallow sites, or down to 3-4 meters depth. Compared to SCUBA diving, free diving requires less equipment and personnel and is thereby more efficient, and is also more practical at very shallow depths. National recommendations for surveying vegetation in shallow inlets have been developed in Sweden in connection to the Natura2000 baseline inven-tories (Persson and Johansson 2005). Remote sensing Satellite images may be used to obtain habitat maps over large areas, and a grain size down to 10m or less can be achieved. One main objective within Balance WP2 is to evaluate the possibility of using remote sensing by satellite imagery for characterisation and identification of shallow marine coastal habitats in the Baltic Sea, and to evaluate if satellite based information can complement existing information and hence be incorpo-rated in management strategies. Currently, methodological procedures have been de-veloped (and are under evaluation, and are to be reported by the end of 2006.

7.3 Recommendations for data intercalibration

Unfortunately, there are no international standards for surveying benthic vegetation. The national monitoring programs of each country are not synchronized which make their data less comparable, especially for annual species. The applied methodology also differs, which has to be dealt with when combining data sets from different sur-veys. These differences relate to both sampling design and accuracy. When combining data of different quality and units for a modelling task the least com-mon denominator should therefore be used, which is presence only or pres-ence/absence in the case of data presented in appendix 2. Unfortunately a lot of infor-mation is lost when this is done. An alternative approach is to make multiple models taking the full advantage of each dataset, which will result in a set of models and pre-dictions that vary in quality. A European standard for marine biological surveys on littoral and sublittoral hard bot-tom is currently under development. However, national monitoring programs are still likely to continue without major changes in order not to loose comparability with earlier data series.


Several species should be able to model since they are included the national datasets. These include the brown alga Fucus vesiculosus, the red alga Furcellaria lumbricalis, the phanerogam Zostera marina and charophytes, which are also important habitat building species that equally could be modelled as habitats. The general distribution of phytobenthic species could also be modelled.


32


The monitoring programs are unfortunately not designed for spatial modelling activities since they do not cover full environmental gradients. Surveys directed only at the opti-mal habitat of the target species do not provided an adequate basis for spatial ecologi-cal models, but suboptimal environments should also be registered. This is also an es-sential aspect in marine spatial planning, where potential effects of different actions are to be analysed. There are research projects in which diving inventories have been de-signed for modelling studies (e.g. Isæus 2004). Irrespective of which method is used, the surface substrate, level of sedimentation and water depth should be registered together with the species data. This gives the possi-bility to validate the indata layers used for modelling and to analyse the errors of the predictions. The Secchi-depth should also be measured at each station as a proxy for light availability. There are many types of phytobenthic organisms in the Baltic Sea and Kattegat all characterising different habitats and requiring different sampling method-ologies (figure 7a-d).

Fig. 7a. Fucus vesiculosus, the dominating submerged brown seaweed in the Baltic Sea. Photo: The Natural Heritage Service, Finland.

Fig. 7b. Dilsea carnosa, a red seaweed in central Kattegat. It distribution is limited by the low salinity in the Baltic Sea. Photo: Orbicon.

Fig. 7c. Chara sp., a green algae living in shallow sheltered soft sediment habitats. Photo: The Natural Heritage Service, Finland.

Fig. 7d. Zostera marina, the dominating sea-grass in the Baltic Sea. Photo: The National Environmental Research Institute.


33

8 SUMMARY AND CONCLUSIONS

In general, there are many challenges to be met if ecological relevant maps are to be produced for larger marine areas, such as the Baltic Sea Marine Region. Some of the challenges related to various types of biological information available are discussed be-low. These are challenges that should be met, as ecological maps are essential for in-formed marine spatial planning and thus for the long-term sustainable development within the Baltic Sea region.

8.1 Fish

Data on fish abundance collected using different methods and gears should generally be viewed as not directly quantitatively comparable. This is mainly due to differences in the target species and overall catchability. Because of high spatial and temporal vari-ability of target species, method efficiency varies with environmental setting, and quan-titative comparisons between studies using the same gear should also be done with caution. Also, data on fish abundance are often highly skewed with non-normal fre-quency distributions, because several species and/or life stages have a strong school-ing behaviour that results in strongly zero-inflated catch data. The schooling behaviour may also be dependent on the environmental conditions. As an example, juvenile fishes have been found to alter their tendency to school as a response to changes in habitat complexity. Thus, the following recommendations may be made:

• Generally, data reduction to “presence only” is recommended for large-scale

mapping. • In all instances where the predictor variable influences the schooling tendency

of the target species, presence/absence models may be biased and should be used with care.

• Presence/absence data may be applicable for methods with similar target species and catch efficiency.

• Quantitative pooling is only recommended on smaller scales and after situa-tion-specific intercalibration exercises.

• In statistical spatial modelling, presence/absence data or presence only data of good quality is preferred over more detailed data of uncertain relevance, as the estimated output may still be expressed as the numerical probability for occurrence of the predictor variable.

8.1.1 Suggestions for future data collection Standards for fish sampling methods are under continuous development within ICES and EU, to ensure use of data beyond the national and regional level. New standards are in progress for monitoring of fish communities in lakes and streams and standards for trawling of pelagic and benthic species in off-shore areas with larger vessels are developed and set within ICES. However, similar standards for monitoring and mapping of juvenile fish in shallow coastal areas are not available. Since the sampling techniques are highly specialised for certain species or habitats, such standards are harder to develop for shallow coastal areas. A first step towards harmonisation would be to conduct transnational in-tercalibration efforts, comparing and evaluating fish sampling surveys in shallow coastal areas within the BSR. Although such an attempt is relevant and highly recom-mended, it would require a large effort and should be conducted over a longer time pe-riod and is thus not possible to include in BALANCE.


34

Currently, for shallow nursery areas, the sampling methods drop-traps, LIPS and coastal survey nets (gill nets) appear the most general by the aspects analysed in table 9. However, these are limited in their potential areas of use, and it may not be possible to agree on one single gear type or method to be used throughout the Baltic Sea. Drop-traps can only be used in very shallow waters, and LIPS will only catch fish with swim bladders, which means that flatfish will not be caught. Gill-nets, again, do not easily al-low quantitative abundance estimates as the catch method is passive, which means that the catch rate of different species will depend on their level of activity in the survey area. Also, gillnetting can be very time consuming during the summer season in the more saline areas of the Baltic Sea due to large catches of crabs. In most cases, small trawls or push-nets are preferred for sampling at sandy beaches, as these gears are generally quite efficient and economically feasible. However, they cannot be used at the whole range of different bottom types and vegetation coverage. From the perspective of habitat mapping, fish surveys that are a priori designed for sampling of the most likely habitats of the target species are relevant for mapping, but have limited value for modelling unless the survey also covers habitats were the target species is less likely to be found. This is because the predictive strength of the statisti-cal models relies on full inclusion of the whole potential distributional range of predicted species in order to properly define the environmental envelope, and not only the opti-mal habitats. Thus, developing methods that enable sampling of fish in the majority of potential habitats in the BSR is thus an important challenge for future fisheries related research in coastal areas.

8.2 Macrofauna

Benthic macrofauna represent important ecological functions and contribute signifi-cantly to major ecological processes in the Baltic Sea. Additionally, benthic fauna spe-cies have a value as indicators in water quality assessment and in modelling of habitat types. Three main components may be considered as potentially useful parameters in mapping and modelling: 1) key macrozoobenthic species and/or functions that can be used to predict various habitat types in the Baltic Sea area, 2) species that may be used as biological indicators, 3) invasive species. Data on benthic macrofauna usually comes from environmental monitoring pro-grammes. The purpose of monitoring is mainly to assess the quality of water rather than to perform habitat mapping, and sampling is therefore often not sampled in a way that is directly applicable for mapping. The datasets obtained by monitoring activities may need to be combined with other research surveys before any valid interpolation based habitat prediction can be done. Alternatively, monitoring data can be used in models, and the model based relationships between environment and invertebrate communities can be used to predict the benthic fauna in areas outside of sampling programs.

8.2.1 Suggestions for future data collection In general many shallow sites lack the information on geomorphologic, hydrographic and biological variables that is needed to produce reliable habitat maps. In these un-derrepresented sites, photographic and video records are recommended as a comple-ment to traditional sampling methods. Sediment profile imaging may provide a useful means for rapid surveys and classification of sediment structure and bioturbation depth. Side-scan sonar images will provide information on bottom topography and sub-strate type, which can be useful in the planning of benthos monitoring programs or in


35

the interpretation of data. Images should be verified by ground-truthing by underwater video recording and/or grab sampling of sediments. Descriptive and experimental studies should focus on the effects of habitat fragmenta-tion on habitat functionality and seek the relationships between habitat units, number of habitats, natural biodiversity and functional diversity. This information is needed for the investigation of potential blue corridors and for evaluation of the NATURA 2000 net-work.

8.3 Zooplankton

The catchability of the different zooplankton species and their stages varies strongly according to gear and mesh-size used. Consequently, no single gear/mesh-size can be recommended for mapping as a combination of mesh-sizes would be necessary to capture all species/stages representatively. A wide range of gears is thus used in moni-toring and process investigation programmes depending on institute, country or target species. Mapping of zooplankton species is difficult due to their patchy small-scale distribution, which is very unstable due to advective transport. Consequently, a spatially and tempo-rally very highly resolved sampling would be necessary to reliably model their horizon-tal distribution. This is probably not possible with the data available from net sampling, and requires VPR or CPR sampling. However, recent studies from the Baltic have shown that general features of the horizontal distribution of the copepod species and their relationship to the physical environment can be derived from spatial sampling. The datasets available have a limited suitability for spatial modelling. The data set by LATFRA cannot be used for distribution modelling for single points in time (seasons) due to the low spatial resolution of the sampling. However, by merging data sets over different periods, for example decades, general Baltic wide patterns of distribution for the main mesozooplankton species in relation to hydrography can be derived. Datasets more suitable for spatial modelling would be data collected from a frequently visited station grid. From these data species- and stage-specific species-environment relation-ships are in principal achievable.

8.3.1 Suggestions for future data collection Future data collection of zooplankton for reliable spatial modelling can only be con-ducted by highly resolved net sampling with different mesh-sizes and an agreed stan-dard gear, which is very costly. Future alternatives will be VPR- and CPR-sampling, depending on the progress of VPR technology and installation of CPR-lines in the Bal-tic.

8.4 Benthic vegetation

There are no international standards for surveying benthic vegetation. The national monitoring programs of each country are not synchronized which make their data less comparable, especially for annual species. The applied methodology also differs, which has to be dealt with when combining data sets from different surveys. These differ-ences relate to both sampling design and accuracy. When combining data of different quality and units for a modelling task the least com-mon denominator should therefore be used, which is presence only or pres-ence/absence in the case of data presented in appendix 2. Unfortunately a lot of infor-mation is lost when this is done. An alternative approach is to make multiple models


36

taking the full advantage of each dataset, which will result in a set of models and pre-dictions that vary in quality. A European standard for marine biological surveys on littoral and sublittoral hard bot-tom is currently under development. However, national monitoring programs are still likely to continue without major changes in order not to loose comparability with earlier data series. Several species should be possible to model since they are included the national data-sets. These include the brown alga Fucus vesiculosus, the red alga Furcellaria lumbri-calis, the phanerogam Zostera marina and charophytes, which are also important habi-tat building species that equally could be modelled as habitats. The general distribution of phytobenthic species could also be modelled.

8.4.1 Suggestions for future data collection The monitoring programs are unfortunately not designed for spatial modelling activities since they do not cover full environmental gradients. Surveys directed only at the opti-mal habitat of the target species do not provided an adequate basis for spatial ecologi-cal models, but suboptimal environments should also be registered. This is also an es-sential aspect in marine spatial planning, where potential effects of different actions are to be analysed. There are research projects in which diving inventories have been de-signed for modelling studies. The diversity of phytobenthic organisms and habitats in the Baltic Sea and Kattegat re-quires different methodologies for efficient sampling. Irrespective of which method is used, the surface substrate, level of sedimentation and water depth should be regis-tered together with the species data. The Secchi-depth should also be measured at each station as a proxy for light availability. This gives the possibility to validate the in-data layers used for modelling and to analyse the errors of the predictions.


37

9 REFERENCES

Austin, M. P. 2002. Spatial prediction of species distribution: an interface between eco-logical theory and statistical modelling. Ecological Modelling 157:101-118. Baker, H. & Stebbins, G. 1965. The Genetics of Colonizing Species. Academic Press, New York. Barnes, R. S. K. & Hughes, R. N. 1988. An Introduction to Marine Ecology. Blackwell Scientific Publications, Oxford. Brosse, S., Guegan, J-F., Tourenq, J-N., Lek, S. (1999). The use of artificial neural networks to assess fish abundance and spatial occupancy in the littoral zone of a mesotrophic lake. Ecological Modelling 120: 299-311. Buckling A., Kassen R., Bell G., Rainey P. B., 2000, Disturbance and diversity in ex-perimental microcosms, Nature, 408, 961–964. Carlton J. T. & Geller, J. B. 1993. Ecological roulette: the global transport of nonindi-genous marine organisms. Science (Wash.), 261, 78–82. Carlton, J. T. 1996. Pattern, process, and prediction in marine invasion ecology. Biol. Conserv., 78, 97–106. Dahl et al. 2003- Stenrev, havbundens oaser. (http:www2.dmu.dk/1_viden/2_publikationer/3_miljobib/raporter/MB02.pdf) Dippner, J.W., Kornilovs, G., and Sidrevics, L. 2000. Long-term variability of mesozoo-plankton in the central Baltic Sea. Journal of Marine Systems, 25: 23-32. Dybern I., Ackefors H., Elmgren R. 1976. Recommendation on method for Marine biological studies in the Baltic Sea. BMB, 98p Elliot, M. & Hemmingway, K. L. (2002). Fishes in Estuaries. Blackwell Science. Elliott, J. M., 1983. Some Methods for the Statistical Analysis of Samples of Benthic In-vertebrates. Freshwater Biological Association - Scientific Publication No. 25. 159 pp. Flöder S., Sommer U., 1999, Diversity in planktonic communities: An experimental test of the intermediate disturbance hypothesis, Limnol. Oceanogr., 44, 1114–1119. Francis, M. P., Morrison, M.A., Leathwick, J., Walsh, C., Middleton, C. (2005). Predic-tive models of small fish presence and abundance in northern New Zealand harbours. Estuarine Coastal and Shelf Science 64: 419-435. Garza-Pérez, J. R., Lehmann, A., Arias-Gonzalez, J. E. (2004). Spatial prediction of coral reef habitats: integrating ecology with spatial modelling and remote sensing. Ma-rine Ecology Progress Series 269: 141-152. Gollasch, S. & Leppäkoski, E (Eds.). 1999. Initial Risk Assessment of Alien Species in Nordic Coastal Waters. Nordic Council of Ministers, Copenhagen.


38

Gollasch, S., Minchin, D., Rosenthal, H. & Voigt, M (Eds.). 1999. Exotics across the Ocean Case Histories on Introduced Species. Department of Fishery Biology, Institute for Marine Science, University of Kiel, Germany.

Grall, J. & L. Chauvaud, 2002. Marine eutrophication and benthos: the need for new approaches and concepts. Global Change Biology 8: 813–830. Granéli, E. & K. Sundbäck, 1985. The response of planktonic and microbenthic algal assemblages to nutrient enrichment in shallow coastal waters, southwest Sweden. Journal of Experimental Marine Biology and Ecology 85: 253–268. Gray J. S., 2002, Species richness of marine soft sediments, Mar. Ecol. Prog. Ser., 244, 285–297. Gray, J. S. 1992. Eutrophication in the sea. In Marine Eutrophication and Population Dynamics. Proc. 25th Eur. Mar. Biol. Symp. (Columbo, G. C., Ferrari, I., Ceccherelli, V. U. & Rossi, R., Eds.). Olsen & Olsen, Fredensborg, 3–15. Gray, J. S., R. S. Wu & Y. Y. Or, 2002. Effects of hypoxia and organic enrichment on the coastal marine environment. Marine Ecology Progress Series 238: 249–279. Grip, et al (in prep) Inventering av marina naturtyper på utsjöbankar. Naturvårdsverket rapport. Guisan, A., & Zimmerman, N.E. (2000). Predictive habitat distribution models in ecology. Ecological Modelling 135: 147-186. Guisan, A., Edwards, Jr., T. C., Hastie, T. (2002). Generalized linear and generalized additive models in studies of species distributions: setting the scene. Ecological Model-ling, 157, 89-100. Hansen, F., Möllmann, C., Schütz, U. and Hinrichsen, H.-H. 2004. Spatio-temporal dis-tribution of Oithona similis in the Bornholm Basin (Central Baltic Sea). Journal of Plank-ton Research, 26: 659-668. Hansen, F.C., Möllmann, C., Schütz, U. and Neumann, T. 2006. Spatio-temporal distri-bution and production of calanoid copepods in the central Baltic Sea. Journal of Plank-ton Research, 28: 39-54. Hector A., Joshi J., Lawler S. P., Spehn E. M., 2001, Conservation implications of the link between biodiversity and ecosystem functioning, Oecologia, 129 (4), 624–628. HELCOM, 1998. Manual for Marine Monitoring in the COMBINE Programme of HELCOM HELCOM, 1982. Baltic Sea Environmental Proceedings, Second biological intercalibra-tion workshop, No. 9, Marine Pollution Laboratory and Marine Division of the National Agency of Environmental Protection, Denmark August 17-20, 1982, Rønne, Denmark Herman, P. M. J. & Scholten, H. 1990. Can suspension-feeders stabilize estuarine ecosystems? In Trophic Relationships in the Marine Environment. Proc. 24th Eur. Mar. Biol. Symp. (Barnes, M. & Gibson, R. N., Eds.). Aberdeen University Press, Aberdeen, 104–116. Howarth, R. W. 1988. Nutrient limitation of net primary production in marine ecosys-tems. Annual Review of Ecology and Systematics 19: 89–110.


39

Hull, S. C. 1987. Macroalgal mats and species abundance: a field experiment. Estuar. Coast. Shelf Sci., 25, 519–532. Hällfors, G., Niemi, Å., Ackefors, H., Lassig, J. & Leppäkoski, E. 1981. Biological oceanography. In The Baltic Sea (Voipio, A., Ed.). Elsevier Oceanography Series, Am-sterdam, 30, 219–274. ICES. 2003: Manual for the Baltic International Acoustic Surveys (BIAS) (ICES 2003). In: Report of the Baltic International Fish Survey Working Group. ICES CM 2003/G:05 Ref.: D. H.: 213 – 258. Jansson, K. 1994. Alien Species in the Marine Environment. Introductions to the Baltic Sea and the Swedish West Coast. Report 4357. Swedish Environmental Protection Agency. Joy, M. K, Death, R. G. (2004). Predictive modelling and spatial mapping of freshwater fish and decapod assemblages using GIS and neural networks. Freshwater Biology 49: 1036–1052. Järvekülg, A. 1979. Benthic Fauna of the Eastern Baltic Sea. Valgus, Tallinn (in Rus-sian). Kaiser, M. J., Rogers, S. I. & McCandless, D. T. (1994). Improving quantitative surveys of epibenthic communities using a modified 2 m beam trawl. Marine Ecology Progress Series, 106: 131-138. Kangas, P., 1972. Quantitative sampling equipment for the littoral benthos II. IBP i Nor-den 10: 9–16. Kautsky, H., 1995. Diving inventory of the vegetation of the shallow bottoms in Stock-holm archipelago 1994 (In Swedish). Department of Systems Ecology, Stockholm Uni-versity. Kautsky, H., Kautsky, L., Kautsky, N., Kautsky, U. & Lindblad, C. 1992. Studies on the Fucus vesiculosus community in the Baltic Sea. Acta Phytogeogr. Suec., 78, 33–49. Kautsky, N. & Evans, S. 1987. Role of biodeposition by Mytilus edulis in the circulation of matter and nutrients in a Baltic coastal ecosystem. Mar. Ecol. Prog. Ser., 38, 201–212. Kautsky, U. 1995. Ecosystem Processes in Coastal Areas of the Baltic Sea. Doctoral dissertation, Stockholm University, Sweden. Kolbe, K., Kaminski, E., Michaelis, H., Obert, B. & Rahmel, J. 1995. Macroalgal mass development in the Wadden Sea: first experiences with a monitoring system. Helgol. Meeresunters., 49, 1–4. Kotta J, Ólafsson E (2003) Competition for food between the introduced exotic poly-chaete Marenzelleria viridis and the resident native amphipod Monoporeia affinis in the Baltic Sea. J Sea Res. 342:27–35 Kotta J, Orav H, Sandberg-Kilpi E (2001) Ecological consequence of the introduction of the polychaete Marenzelleria viridis into a shallow water biotope of the northern Baltic Sea. J Sea Res. 46:273–280


40

Kotta, J. 2000. Impact of eutrophication and biological invasions on the structure and functions of benthic macrofauna. Dissertationes Biologicae Universitatis Tartuensis, 63, Tartu University Press, 1–160. Kotta, J., Kotta, I. & Kask, J. 1999. Benthic animal communities of exposed bays in the western Gulf of Finland (Baltic Sea). Proc. Est. Acad. Sci. Biol. Ecol., 48, 107–116. Kotta, J., Kotta, I. Viitasalo, I. 2000. Effect of diffuse and point source nutrient supply on the low diverse macrozoobenthic communities of the northern Baltic Sea. Bor. Env. Res., 5, 235–242. Kuipers, B. (1975). On the efficiency of a two-metre beam trawl for juvenile plaice (Pleuronectes platessa). Netherlands Journal of Sea Research, 9 (1): 69-85. Krause-Jensen, D. et al. Tekninske anvisninger for marin overvågning. NOVANA Laine, A., Sandler, A. O., Andersin, A.-B. & Stigzelius, J. 1997. Long-term changes of macrozoobenthos in the eastern Gotland Basin and the Gulf of Finland (Baltic Sea) in relation to the hydrographical regime. J. Sea Res., 38, 135–159. Lehmann, A., Overton, J. Leathwick, J. R. (2002). GRASP: generalized regression analysis and spatial prediction: Ecological Modelling 157: 189-207. Lek, S., & Guégan, J. F., (1999). Artificial neural networks as a tool in ecological mod-elling, an introduction. Ecological Modelling 120: 65-73. Leppäkoski E, Olenin S, Gollasch S (2002) The Baltic Sea – a Field laboratory for inva-sion biology. In Leppäkoski E, Olenin S, Gollasch S (ends) Invasive Aquatic Species of Europe. Kluver Acad. Publ., Dordrecht, Boston, London, pp 253–259 Leppäkoski E., Olenin S., 2001. The meltdown of biogeographical peculiarities of the Baltic Sea: the interaction of natural and man-made processes, Ambio, 30, 202–209. Leppäkoski, E. 1991. Introduced species – resource or threat in brackish water seas? Examples from the Baltic and the Black Sea. Mar. Pollut. Bull., 23, 219–223. Loreau M., Naeem S., Inchausti P., Bengtsson J., Grime J.P., Hector A., Hooper D. U., Huston M.A., Raffaelli D., Schmid B., Tilman D., Wardle D.A., 2001, Biodiversity and ecosystem functioning: current knowledge and future challenges, Science, 294, 804-808. Mills, E. L., Leach, J. H., Carlton, J. T. & Secor, C. L. 1993. Exotic species in the Great Lakes: a history of biotic crises and anthropogenic introductions. J. Great Lakes Res., 19, 1–54. Möllmann, C., G. Kornilovs, and L. Sidrevics. 2000. Long-term dynamics of main mesozooplankton species in the central Baltic Sea. Journal of Plankton Research, 22: 2015–2038. Möllmann, C., F.W. Köster, G. Kornilovs, and L. Sidrevics. 2003. Interannual variability in population dynamics of calanoid copepods in the central Baltic Sea. ICES Marine Science Symposia, 219: 294-306 Moy, F. Aure, J. et al. 2005. Langtidsovervåkning av miljökvaliteten i kystområdene av Norge. Årsapport for 2004. F Moy. Oslo, SFT, NIVA.


41

Norkko, A. & Bonsdorff, E. 1996a. Population responses of coastal zoobenthos to stress induced by drifting algal mats. Mar. Ecol. Prog. Ser., 140, 141–151.

Norkko, A. & Bonsdorff, E. 1996b. Rapid zoobenthic community responses to accumu-lations of drifting algae. Mar. Ecol. Prog. Ser., 131, 143–157. Olenin, S. & Leppäkoski, E. 1999. Non-native animals in the Baltic Sea: alteration of benthic habitats in coastal inlets and lagoons. Hydrobiologia, 393, 233–243. Paalme, T., Kukk, H., Kotta, J., Orav, H. 2002. “In vitro” and “in situ” decomposition of nuisance macroalgae Cladophora glomerata and Pilayella littoralis. Hydrobiologia, 475/476, 469–476. Persson, J., Johansson, G. 2005. Manual för basinventering av marina habitat (1150, 1160, 1650), Naturvårdsverket. 27pp. (www.naturvardsverket.se) Pearson, T. H. & R. Rosenberg, 1978. Macrobenthic succession in relation to organic enrichment and pollution of the marine environment. Oceanography and Marine Biol-ogy: An Annual Review 16: 229–311. Pihl, L. & Rosenberg, R. (1982) Production, abundance, and biomass of mobile epiben-thic marine fauna in shallow waters, Western Sweden. Journal of Experimental Marine Biology and Ecology, 57: 273-301. Pitkänen, Timo. 2006. The use of GIS methods in planning and carrying out an under-water nature inventory. Univ. Turku, 98 pp Posey, M. H., T. D. Alphin, L. Cahoon, D. Lindquist & M. E. Becker, 1999. Interactive effects of nutrient additions and predation on infaunal communities. Estuaries 22: 785–792. Rees, H. L., C. Heip, M. Vincx and M. M. Parker, 1991. Benthic communities: use in monitoring point-source discharges. ICES Techniques in Marine Environmental Sci-ences No. 16, 70 pp. Renz, J., and H.J. Hirche. 2006. Life cycle of Pseudocalanus acuspes giesbrecht (Co-pepoda, Calanoida) in the central Baltic Sea: I. Seasonal and spatial distribution. Ma-rine Biology, 148: 567. Rhoads, D.C. and Germano, J.D., 1982. Characterisation of organism-sediment rela-tions using sediment profile imaging: an efficient method of remote ecological monitor-ing of the seafloor (REMOTS system). Mar. Ecol. Prog. Ser., 8, pp. 115-128. Riisgård, H. U. & Møhlenberg, F. 1979. An improved automatic recording apparatus for determining the filtration rate of Mytilus edulis as a function of size and algal concentra-tion. Mar. Biol., 52, 61–67. Riley, J. D. & Corlett, J. (1966). The numbers of 0-group plaice in Port Erin Bay, 1964-1966. Marine Biological Station, University of Liverpool, Port Erin, Isle of Man, Annual Report No. 78 for 1965. Rogers, S. I. & Lockwood, S. J. (1989). Observations on the capture efficiency of a two-metre beam trawl for juvenile flatfish. Netherlands Journal of Sea Research 23(3): 347-352


42

Rosenberg, R. 1985. Eutrophication – the future marine coastal nuisance? Mar. Pollut. Bull., 16, 227–231. Ruiz, G. M., Fofonoff, P. & Hines, A. H. 1999. Non-indigenous species as stressors in estuarine and marine communities: assessing invasion impacts and interactions. Lim-nol. Oceanogr., 44, 950–972. Rumohr H., 1995. Monitoring the marine environment with imaging methods. Scientia Marina, 59: 129-138. Segerstråle, S. 1957. Baltic Sea. Mem. Geol. Soc. Am., 67, 757–800. Snickars, M., Sundblad, G., Sandström, A., Ljunggren, L., Johansson, G., Mattila, J. (manuscript). Modelling of spawning selectivity in shallow areas of the Baltic Sea. Sparrevohn, C.R., Støttrup, J. G. (In press.) Post-release survival and feeding in reared turbot. Journal of Sea Research Wennhage, H., Gibson, R. N. & Robb, L. (1997). The use of drop traps to estimate the efficiency of two beam trawls commonly used for sampling juvenile flatfish. Journal of Fish Biology, 51: 441-445. Worm B., Lotze H. K., Hillebrand H., Sommer U., 2002, Consumer versus resource control of species diversity and ecosystem functioning, Nature, 417, 848–851. Zaniewski, A. E., Lehmann, A., Overton, J. McC. (2002). Predicting species distribution using presence-only data: a case study of native New Zealand ferns. Ecol. Model. 157: 259-278.


43

10 APPENDICES

10.1 Appendix 1. Methods for sampling fish data

AP

PE

ND

IX 1

. PA

GE

1a/

3M

ETH

OD

/ EQ

UIP

MEN

TD

rop

trap

Pus

hnet

Man

pow

ered

2m

eter

be

amtra

wl

Traw

l sa

mpl

ing

of

youn

g fis

h

Beac

h S

eine

Div

ing

LIP

SW

hite

pla

tes

and

scoo

psG

ill n

et

surv

eys

Info

rmat

ion

prov

ided

by

CR

S /D

IFR

ES

CR

S

/DIF

RES

CR

S /D

IFR

ESC

RS

/DIF

RE

SAS

/SB

FAS

/SBF

AS

/S

BFA

L

/F

GFR

IAS

/SBF

PILO

T A

REA

11

11

33

33

3TA

RG

ET S

PEC

IES

Spe

cies

/life

sta

geYo

ung

flatfi

shYo

ung

flatfi

shYo

ung

flatfi

shYo

ung

flatfi

shYo

ung

fish,

fre

shw

ater

sp

ecie

s

Youn

g fis

h,

fresh

wat

er

spec

ies

Youn

g fis

h,

fresh

wat

er

spec

ies

Fish

larv

aeFi

sh >

8cm

Mob

ility

type

activ

e (k

m)

activ

e (k

m)

activ

e (k

m)

activ

e (k

m)

activ

e (k

m)

sess

ileac

tive

(km

)ac

tive

(km

)ac

tive

(km

)

Tem

pora

l var

iatio

nse

ason

alse

ason

alse

ason

alse

ason

alse

ason

alse

ason

alse

ason

alse

ason

alse

ason

alO

UTP

UT

VAR

IAB

LES

Nam

eab

unda

nce

catc

h ra

teca

tch

rate

catc

h ra

teca

tch

rate

pres

ence

abun

danc

epr

esen

ceca

tch

rate

Uni

tn/

m2

n/ef

fort

n/ef

fort

n/ef

fort

n/ef

fort

pres

./abs

.n/

m2

pres

./abs

.n/

effo

rtD

eriv

ed v

aria

ble

unit

none

none

none

none

none

none

none

none

none

Pos

ition

ing

accu

racy

3-20

m3-

20m

3-20

m3-

20m

3-65

mM

ETH

OD

DET

AIL

S A

ND

REF

EREN

CES

Orig

inal

pur

pose

of s

tudy

Map

ping

Map

ping

Map

ping

Mon

itorin

gM

appi

ngM

appi

ngM

appi

ngM

appi

ngM

onito

ring

Are

a/vo

lum

e co

vere

d by

eac

h sa

mpl

e1

sqm

20 s

qm50

sqm

1800

sqm

varie

s0.

25 s

qm16

or 1

00

sqm

ca 1

00 s

qmno

t est

imab

le

Est

imat

ed a

ccur

acy

EXAC

TG

OO

DG

OO

DM

OD

ERAT

EG

OO

DG

OO

DVE

RY

GO

OD

VER

Y G

OO

DM

OD

ERA

TE

How

is e

valu

atio

n ac

hiev

ed?

inte

rcal

ibra

tion

with

in

DIF

RE

S

inte

rcal

ibra

tion

with

in

DIF

RES

inte

rcal

ibra

tion

with

in

DIF

RE

S

inte

rcal

ibra

tion

with

in

DIF

RE

S

-ow

n es

timat

eSn

icka

rs e

t al

in p

rep

own

estim

ate

With

in S

BF

Met

hod

refe

renc

ese

e te

xt in

pr

esen

t rep

ortse

e te

xt in

pr

esen

t rep

ortse

e te

xt in

pr

esen

t rep

ortse

e te

xt in

pr

esen

t rep

ortFr

anci

s et

al.

2005

, mod

if ac

c H

udd

(200

0)

Snic

kars

et

al. I

n pr

epSn

icka

rs in

pr

ep,

Sand

strö

m

2005

Kje

llman

20

04S

öder

berg

et

al 2

005


44

AP

PE

ND

IX 1

. PA

GE

1b/

3M

ETH

OD

/ EQ

UIP

MEN

TBo

ngo

Dem

ersa

l tra

wl (

TV3

traw

l)H

ydro

acho

ust

ics

Pela

gic

traw

l

Info

rmat

ion

prov

ided

by

RV

, GK

/IF

M-

GE

OM

AR

JT, H

P /D

IFR

ES

JT, H

P /D

IFR

ESJT

, HP

/DIF

RES

PILO

T A

REA

22

22

TAR

GET

SPE

CIE

SS

peci

es/li

fe s

tage

Ear

ly L

ife

Stag

es o

f cod

an

d sp

rat

Cod

, fla

tfish

Spr

at, h

errin

gSp

rat,

herri

ng

Mob

ility

type

activ

e/pa

ssiv

eac

tive

(km

)ac

tive

(km

)ac

tive

(km

)

Tem

pora

l var

iatio

nse

ason

alan

nual

annu

alan

nual

OU

TPU

T VA

RIA

BLE

SN

ame

abun

danc

e/pr

odu

ctio

nca

tch

rate

biom

ass

catc

h ra

te

Uni

tn/

m²

n/ef

fort

t/vol

ume

n/ef

fort

Der

ived

var

iabl

e un

itno

neno

neno

neno

neP

ositi

onin

g ac

cura

cy10

m3

dec.

min

.3

dec.

min

.3

dec.

min

.M

ETH

OD

DET

AIL

S A

ND

REF

EREN

CES

Orig

inal

pur

pose

of s

tudy

Mon

itorin

gM

appi

ngM

appi

ngM

appi

ngA

rea/

volu

me

cove

red

by e

ach

sam

ple

250

km**

2va

ries

varie

sva

ries

Est

imat

ed a

ccur

acy

GO

OD

GO

OD

GO

OD

How

is e

valu

atio

n ac

hiev

ed?

subj

ectiv

ely

judg

edsu

bjec

tivel

ysu

bjec

tivel

y

Met

hod

refe

renc

eBI

TS (I

CES

)B

IAS

(ICES

)BI

AS

(ICES

)


45 A

PP

EN

DIX

1. P

AG

E 2

a/3

MET

HO

D/ E

QU

IPM

ENT

Dro

p tra

pP

ushn

etM

anpo

wer

ed

2met

er

beam

traw

l

Traw

l sa

mpl

ing

of

youn

g fis

h

Bea

ch S

eine

Div

ing

LIP

SW

hite

pla

tes

and

scoo

psG

ill n

et

surv

eys

ECO

NO

MY

Tim

e in

Fie

ld/s

ampl

eno

t est

imat

edno

t est

imat

edno

t est

imat

edno

t est

imat

ed6-

10

sam

ples

/day

*2

pers

25ha

/day

*2

pers

7-15

sa

mpl

es/d

ay

*2pe

rs

20-3

0 m

in/s

ite6-

12 n

ets/

day

*2pe

rs

Tim

e in

Lab

/sam

ple

(not

incl

udin

g re

gist

ratio

n an

d nu

mer

ical

ana

lyse

s)lit

tle-n

one

little

-non

elit

tle-n

one

none

none

none

none

little

-non

eno

ne

Labo

rato

ry a

naly

sis

taxo

nom

ic

valid

atio

nta

xono

mic

va

lidat

ion

taxo

nom

ic

valid

atio

nta

xono

mic

va

lidat

ion

INTE

ND

ED U

SE IN

BA

LAN

CE

Hab

itat m

apYE

SYE

SYE

SYE

SYE

SYE

SYE

SYE

Spo

ssib

lyR

esp.

Var i

n M

odel

YES

YES

YES

YES

NO

YES

YES

YES

poss

ibly

Env

. Var

in M

odel

NO

NO

NO

NO

NO

NO

NO

NO

NO

Mod

el V

alid

atio

nN

ON

ON

ON

OYE

SN

OYE

SYE

Spo

ssib

lyP

lann

ed m

etho

d fo

r spa

tial

inte

rpol

atio

n -

- -

-so

me:

mov

ing

win

dow

NO

som

e: m

ovin

g w

indo

wN

ON

O

Pla

nned

met

hod

for s

patia

l mod

ellin

gG

RA

SP

GR

AS

PG

RA

SP

GR

AS

PG

RA

SP

GR

AS

PG

RA

SP

GR

AS

PG

RA

SP

Dat

a pr

oces

sing

refe

renc

esLe

hman

n et

al

200

2Le

hman

n et

al

200

2Le

hman

n et

al

200

2Le

hman

n et

al

200

2Le

hman

n et

al

200

2Le

hman

n et

al

200

2Le

hman

n et

al

200

2Le

hman

n et

al

200

2Le

hman

n et

al

200

2ES

TIM

ATI

ON

OF

ENVI

RO

NM

ENTA

L VA

RIA

BLE

S A

FFEC

TIN

G D

ISTR

IBU

TIO

N O

F TA

RG

ET S

PEC

IES

Sal

inity

MO

DM

OD

MO

DM

OD

MO

DM

OD

MO

DM

OD

MO

DW

ave

expo

sure

MO

DM

OD

MO

DM

OD

MO

DM

OD

MO

DM

OD

MO

DD

epth

HIG

HH

IGH

HIG

HH

IGH

HIG

HH

IGH

HIG

HH

IGH

HIG

HLi

ght

LOW

LOW

LOW

LOW

LOW

LOW

LOW

LOW

LOW

Nut

rient

sM

OD

MO

DM

OD

MO

DM

OD

MO

DM

OD

MO

DM

OD

Oxy

gen

cont

ent

HIG

HH

IGH

HIG

HH

IGH

HIG

HH

IGH

MO

DH

IGH

MO

DTe

mpe

ratu

reH

IGH

HIG

HH

IGH

HIG

HH

IGH

HIG

HH

IGH

HIG

HH

IGH


46

A

PP

EN

DIX

1. P

AG

E 2

b/3

MET

HO

D/ E

QU

IPM

ENT

Bon

goD

emer

sal t

raw

l (TV

3 tra

wl)

Hyd

roac

hous

tic

sP

elag

ic tr

awl

ECO

NO

MY

Tim

e in

Fie

ld/s

ampl

e2h

not e

stim

ated

not e

stim

ated

not e

stim

ated

Tim

e in

Lab

/sam

ple

(not

incl

udin

g re

gist

ratio

n an

d nu

mer

ical

ana

lyse

s)2h

none

none

none

Labo

rato

ry a

naly

sis

INTE

ND

ED U

SE IN

BA

LAN

CE

Hab

itat m

apYE

SYE

SYE

SYE

SR

esp.

Var

in M

odel

-N

ON

ON

OE

nv. V

ar in

Mod

el -

YES

YES

YES

Mod

el V

alid

atio

n -

NO

NO

NO

Pla

nned

met

hod

for s

patia

l in

terp

olat

ion

- -

- -

Pla

nned

met

hod

for s

patia

l mod

ellin

g -

to c

ome

to c

ome

to c

ome

Dat

a pr

oces

sing

refe

renc

esto

com

eto

com

eto

com

eto

com

e

ESTI

MA

TIO

N O

F EN

VIR

ON

MEN

TAL

VAR

IAB

LES

AFF

ECTI

NG

DIS

TRIB

UTI

ON

OF

TAR

GET

SPE

CIE

SS

alin

ityH

IGH

HIG

HH

IGH

HIG

HW

ave

expo

sure

LOW

HIG

HH

IGH

HIG

HD

epth

HIG

HH

IGH

HIG

HH

IGH

Ligh

tM

OD

LOW

LOW

LOW

Nut

rient

sM

OD

LOW

LOW

LOW

Oxy

gen

cont

ent

HIG

HH

IGH

LOW

LOW

Tem

pera

ture

HIG

HM

OD

MO

DM

OD


47

AP

PE

ND

IX 1

. PA

GE

3a/

3M

ETH

OD

/ EQ

UIP

MEN

TD

rop

trap

Pus

hnet

Man

pow

ered

2m

eter

be

amtra

wl

Traw

l sa

mpl

ing

of

youn

g fis

h

Bea

ch S

eine

Div

ing

LIP

SW

hite

pla

tes

and

scoo

psG

ill n

et

surv

eys

SAM

PLIN

G D

ETA

ILS

Sam

plin

g st

rate

gyst

ratif

ied

or

rand

om w

ithin

m

uddy

/san

dy

subs

trate

s

stra

tifie

d or

ra

ndom

with

in

mud

dy/s

andy

su

bstra

tes

stra

tifie

d or

ra

ndom

with

in

mud

dy/s

andy

su

bstra

tes

aim

at h

igh

geog

raph

ic

cove

rage

rand

om w

ithin

sh

allo

w b

ays

of th

e st

udy

area

stra

tifie

d ra

ndom

stra

tifie

d ra

ndom

stra

tifie

d ra

ndom

stra

tifie

d ra

ndom

Sam

plin

g pe

riod

varie

s w

ith

stud

yva

ries

with

st

udy

varie

s w

ith

stud

yva

ries

with

st

udy

Jul-A

ugA

pr-J

unJu

l-Sep

Apr

-May

Aug

(mai

nly)

Oth

er d

ata

colle

cted

with

in s

ame

stud

y -

- -

leng

th,

som

etom

es

age

vege

tatio

n so

met

imes

, te

mpe

ratu

re,

salin

ity,

turb

idity

dept

h, te

mp,

ve

geta

tion,

sa

linity

, nu

trien

ts,

turb

idity

dept

h, te

mp,

ve

geta

tion,

sa

linity

, tu

rbid

ity,

subs

trate

tem

p, s

al, v

eg

type

dept

h, te

mp,

se

cchi

ap

prox

, le

ngth

, age

, se

xD

ata

inte

rpre

tatio

n in

com

bina

tion

with

ot

her m

etho

ds -

- -

- -

- -

- -

Is th

e st

udy

desi

gned

to g

eogr

aphi

cally

cov

er th

e fo

llow

ing

grad

ient

s?S

alin

ityva

ries

varie

sva

ries

varie

sN

ON

ON

OYE

SN

OW

ave

expo

sure

varie

sva

ries

varie

sva

ries

NO

mod

erat

ely

YES

YES

NO

Dep

thva

ries

varie

sva

ries

varie

sN

OYE

SYE

SN

OYE

SLi

ght

varie

sva

ries

varie

sva

ries

NO

NO

som

etim

esN

ON

ON

utrie

nts

varie

sva

ries

varie

sva

ries

NO

NO

NO

NO

NO

Oxy

gen

cont

ent

Tem

pera

ture

varie

sva

ries

varie

sva

ries

NO

NO

NO

YES

NO

Doe

s th

e st

udy

cove

r are

as o

utsi

de th

e pr

imar

y di

strb

utio

n ar

ea o

fthe

targ

et s

peci

es?

SUPP

LEM

ENTA

L IN

FOR

MA

TIO

N


48

APP

EN

DIX

1. P

AG

E 3

b/3

MET

HO

D/ E

QU

IPM

ENT

Bon

go n

et, 6

0cm

di

amet

er w

ith

335

and

500µ

m

hi

Dem

ersa

l tra

wl (

TV3

traw

l)H

ydro

acho

ust

ics

Pel

agic

traw

l

SAM

PLIN

G D

ETA

ILS

Sam

plin

g st

rate

gygr

idde

pth

stra

tifie

dde

pth

stra

tifie

dde

pth

stra

tifie

d

Sam

plin

g pe

riod

3-5

surv

eys

per

year

Mar

-Oct

5-8

sur

veys

aro

und

the

year

5-8

sur

veys

ar

ound

the

year

5-8

sur

veys

aro

und

the

year

Oth

er d

ata

colle

cted

with

in s

ame

stud

yhy

drog

raph

y,

traw

l sur

vey

biom

ass,

le

ngth

freq

uenc

y,

wei

ght,

sex,

mat

urity

-bi

omas

s,

leng

th fr

eque

ncy,

w

eigh

t, se

x, m

atur

ity

Dat

a in

terp

reta

tion

in c

ombi

natio

n w

ith

othe

r met

hods

- -

-

Is th

e st

udy

desi

gned

to g

eogr

aphi

cally

cov

er th

e fo

llow

ing

grad

ient

s?S

alin

ityYE

SYE

SYE

SYE

SW

ave

expo

sure

NO

NO

NO

NO

Dep

thYE

SYE

SYE

SYE

SLi

ght

NO

NO

NO

NO

Nut

rient

sN

ON

ON

ON

OO

xyge

n co

nten

tYE

STe

mpe

ratu

reYE

SYE

SYE

SYE

SD

oes

the

stud

y co

ver a

reas

out

side

tYE

SN

OSU

PPLE

MEN

TAL

INFO

RM

ATI

ON

Add

ition

al g

ears

us

ed: I

KS

and

m

ultin

ets


49

10.2 Appendix 2. Methods for sampling macrofauna data

Appendix 2, page 1/2METHOD/ EQUIPMENT van VEEN grab van VEEN grab EKMAN grab Core sampler,

diver operatedCore sampler, remote

Information provided by JA /IAE JK /EMI JK /EMI JK /EMI JH /NERIPILOT AREA 4 4 4 4 1TARGET SPECIESSpecies/life stage all observed all observed all observed all observed all observedMobility type active (meters) active (meters) active (meters) active (meters) active (meters)Temporal variation seasonal-annual seasonal-annual seasonal-annual seasonal-annual seasonal-annual

OUTPUT VARIABLESName biomass abund., biomass abund., biomass abund., biomass abund., biomass

Unit gWW/m2 n/m2, gDW/m2 n/m2, gDW/m2 n/m2, gDW/m2 n/m2, gWW/m2Derived variable unit gWW/m2 gWW/m2 gWW/m2Positioning accuracy 20-100 (<50m) 3-20m 3-20m 3-20m 2-5 mMETHOD DETAILS AND REFERENCESSiev/mesh 0.5 mm 1 and 0.25 mm 0.25 mm 0.25 mmOriginal purpose of study monitoring monitoring monitoring monitoring Mapping /ground

truthingArea/volume of sample >4L and 0.1m2 >4L and 0.1m2 0.02m2 0.0083m2 0.0143m2Estimated accuracy moderate good good good moderateHow is evaluation achieved? Periodic

intercalibration exercises

Field intercalibrations



Method reference 1, 2 3 3 3 4ECONOMYTime in Field/sample not estimated 25 stations/week

*2 pers7-20 stations/day *1-2 pers

7-20 stations/day *1-2 pers

60-70 samples/day *3

Time in Lab/sample (not including registration and numerical analyses)

0.5 samples/ day *1pers

3 samples/ day *1pers

2 samples/day *1pers

2 samples/day *1pers

Laboratory analysis sorting, counting and weighing

sorting, counting and weighing




INTENDED USE IN BALANCEHabitat map YES YES YES YES YESResp.Var in Model not decided yet YES YES YES YESEnv. Var in Model Pot. Yes NO NO NO NOModel Validation Pot. Yes YES YES YES NOPlanned method for spatial interpolation

not decided yet kriging in well sampled areas

kriging in well sampled areas

kriging in well sampled areas

not decided yet

Planned method for spatial modelling

- - - - -

Data processing references - - - - -ESTIMATION OF ENVIRONMENTAL VARIABLES AFFECTING DISTRIBUTION OF TARGET SPECIESSalinity MODERATE LOW-HIGH LOW-HIGH LOW-HIGH MODERATEWave exposure MOD-HIGH MOD-HIGH HIGH HIGH MODERATEDepth MODERATE HIGH MOD-HIGH MOD-HIGH HIGHLight MODERATE LOW MOD-HIGH MOD-HIGH LOWNutrients MODERATE HIGH HIGH HIGH LOWOxygen content HIGH HIGH HIGH HIGH HIGHTemperature LOW-HIGH MOD-HIGH MODERATE MODERATE LOW


50

Appendix 2, page 2/2METHOD/ EQUIPMENT van VEEN grab van VEEN grab EKMAN grab Core sampler,

diver operatedCore sampler, remote

SAMPLING DETAILSSampling strategy arbitrary-

representative of basins

stratified by waterbodies and/or grid



transect -random

Sampling period May Late spring Ice free season Ice free season AprilPrincipal study area and size Approx

25.000km2Estonian coastline

Varies: 25 km2, or Estonian coastline

Varies: 25 km2, or Estonian coastline

Acc to GEUS, SGU output

Other data collected within same study

water transparency, standard hydrology and hydrochemistry

phytoplankton, zooplankton, bottom vegetation, sediment type, organic matter, temperature profiles, salinity, oxygen, water nutrients

vegetation, sediment type, sediment organic matter content

vegetation, sediment type

sediment type (phi scale)

Data interpretation in combination with other methods

plankton, nutrients, salinity, oxygen, grain size, depth

phytoplankton, nutrients, salinity, oxygen, sediment, human disturbances



sediment type

Is the study designed to geographically cover the following gradients?Salinity NO YES YES YES NOWave exposure NO YES YES YES NODepth YES YES YES YES YESLight NO NO YES YES NONutrients NO YES YES YES NOOxygen contentTemperature NO YES No NO NODoes the study cover areas outside the primary distrbution area ofthe target species?

NO YES YES YESSUPPLEMENTAL INFORMATION 5, 6Footnotes: 1) HELCOM 1998

2) Dybern et al 19763) http://sea.helcom.fi/Monas/CombineManual2/PartC/CFrame.htm4) Kanneworf & Nicolaisen 1973, Jensen 19975) Sediment type is classified visually into broad categories (e.g. silt, clay, fine sand, medium sand, coarse sand, pebbles); homogenous/mixed; oxygen conditions estimated by colour 6) Data is normally not classified into assemblage types but instead is analysed using various multivariate statistics; the rationale behind it is the lack of clear assemblages but instead the presence of continuous changes of communities according to the changes in


51

10.3 Appendix 3. Methods for sampling zooplankton data

Appendix 3, page 1/2METHOD/ EQUIPMENT BONGO MULTINETS JUDAI NETInformation provided by RV /IFM-GEOMAR,

CM /DIFRESRV /IFM-GEOMAR, CM /DIFRES

RV /IFM-GEOMAR, CM /DIFRES

JA /IAE

PILOT AREA 2 2 2 4TARGET SPECIESSpecies/life stage copepods,

cladoceranscopepods, cladocerans

copepods, cladocerans

copepods, cladocerans

Mobility type active/passive active/passive active/passive active/passiveTemporal variation weeks weeks weeks weeksOUTPUT VARIABLESName abundance abundance abundance abundanceUnit n/m3 n/m3 n/m3 n/m3Derived variable unit biomass mg/m3 biomass mg/m3 biomass mg/m3 biomass mg/m3Positioning accuracy 50mMETHOD DETAILS AND REFERENCESOriginal purpose of study Monitoring Monitoring Monitoring MonitoringEquipment details Bongo net (30cm

diameter,150µm mesh size)

Multinet (0.5m2, 50µm mesh size)

Judai (36cm diameter, 160µm mesh size)

WP-2 net (100 microns)

Area/volume covered by each sample not estimated not estimated not estimated not estimated

Estimated accuracy GOOD GOOD GOOD GOODHow is evaluation achieved? subjectively subjectively subjectively subjectivelyMethod reference HELCOM,

COMBINEECONOMYTime in Field/sample 2h 2h 2h not estimatedTime in Lab/sample (not including registration and numerical analyses)

2h 2h 2h

Laboratory analysis counting and size estimations

INTENDED USE IN BALANCEHabitat map YES YES YES not decided yetResp.Var in Model YES YES YESEnv. Var in Model NO NO NOModel Validation NO NO NOPlanned method for spatial interpolation

not decided yet

Planned method for spatial modelling not decided yet

Data processing references


52

Appendix 3, page 2/2METHOD/ EQUIPMENT BONGO MULTINETS JUDAI NETESTIMATION OF ENVIRONMENTAL VARIABLES AFFECTING DISTRIBUTION OF TARGET SPECIESSalinity HIGH HIGH HIGH LOW/MODWave exposure LOW LOW LOW LOWDepth HIGH HIGH HIGH LOWLight MOD MOD MOD LOWNutrients LOW LOW LOW LOWOxygen contentTemperature HIGH HIGH HIGH MOD/HIGHSAMPLING DETAILSSampling strategy grid line line arbitrary-

representativeSampling period monthly monthly seasonally annuallyPrincipal study area, name and size ICES Sub-division

25, 26, 28ICES Sub-division 25, 26, 28

ICES Sub-division 25, 26, 28

Gulf of Riga

Other data collected within same study Hydrography Hydrography Hydrography Hydrography

Data interpretation in combination with other methods

hydrological and hydrochemical data

Is the study designed to geographically cover the following gradients?Salinity YES YES YES NOWave exposure NO NO NO NODepth NO YES YES NOLight NO NO NO NONutrients YES YES YES NOOxygen content YES YES YES NOTemperature YES YES YES NODoes the study cover areas outside the primary distrbution area ofthe target species?

NO NO NO -


53

10.4 Appendix 4. Methods for sampling phytobenthos data

App

endi

x 4,

pag

e 1/

3M

ETH

OD

/ EQ

UIP

MEN

TD

ivin

gS

kin

divi

ngD

rop-

Vid

eoS

atel

lite

data

Div

ing

UW

-vid

eoD

ivin

gIn

form

atio

n pr

ovid

ed b

yK

D /N

ER

IA

S /A

BF

MB

/ME

TSÄ

SW

/ME

TRIA

MI /

NIV

AM

I /N

IVA

MI /

NIV

API

LOT

AR

EA1

33

1-4

1B

altic

Sea

, not

K

atte

gatt

Bal

tic S

ea, n

ot

Kat

tega

ttTA

RG

ET S

PEC

IES

Spe

cies

/life

sta

geal

ldo

min

ant

dom

inan

tdo

min

ant

all

all

all

Mob

ility

type

sess

ilese

ssile

sess

ilese

ssile

sess

ilese

ssile

sess

ileTe

mpo

ral v

aria

tion

annu

al-

pere

nnia

lan

nual

-pe

renn

ial

annu

al-

pere

nnia

lan

nual

-pe

renn

ial

annu

al-

pere

nnia

lan

nual

-pe

renn

ial

annu

al-

pere

nnia

lO

UTP

UT

VAR

IAB

LES

Nam

eab

unda

nce

(l

)ab

unda

nce

(l

)ab

unda

nce

(l

)ab

unda

nce

(l

)ab

unda

nce

(l

)ab

unda

nce

(l

)ab

unda

nce

(l

)U

nit

% c

ont.

scal

ecl

ass

% o

rdin

al s

cale

% o

rdin

al s

cale

clas

s%

ord

inal

sca

le%

ord

inal

sca

le

Sca

le in

terv

als

1,2,

3…99

,100

%sc

aled

1-7

, 1-4

, or

0-4

0,1,

3,10

,20,

..90,

100%

0-20

, 20-

80, 8

0-10

0%0-

40,

1, 2

, 5, 1

0,

25,..

.75,

100

%0,

1, 2

, 5, 1

0,

25,..

.75,

100

%P

ositi

onin

g ac

cura

cy1-

2m3-

20m

5-1

0m2-

3m1-

5 m

1-5

m1-

5 m

MET

HO

D D

ETA

ILS

AN

D R

EFER

ENC

ESO

rigin

al p

urpo

se o

f stu

dyM

appi

ng

/Mon

itorin

gM

appi

ngM

appi

ngM

appi

ngM

onito

ring

Map

ping

Map

ping

Are

a/vo

lum

e co

vere

d by

eac

h sa

mpl

e25

m2

0.25

m2

and

20m

2m

in 2

0m2

1 sc

ene

60X6

0km

/res

ol.

10 m

100m

2 ob

serv

atio

n un

it30

-100

m

trans

ects

, ca

5-10

m w

idth

Est

imat

ed a

ccur

acy

Goo

dG

ood

Goo

dU

nder

ev

alua

tion

Goo

d -

-

How

is e

valu

atio

n ac

hiev

ed?

Rea

l tim

e te

levi

sion

from

di

ver t

o sh

ip

Fiel

d te

sts

with

in

deve

lopm

ent o

f m

anua

l

Div

ing

Gro

und

truth

ing

Inte

rcal

ibra

tion

betw

een

dive

rsN

ot e

valu

ated

Not

eva

luat

ed

Met

hod

refe

renc

eD

ahl e

t al 2

003,

K

raus

e-Je

nsen

20

05

Per

sson

&

Joha

nsso

n 20

05P

itkän

en 2

006

Wen

nber

g, in

pr

epM

oy e

t al 2

005

Kau

tsky

199

5


54

App

endi

x 4,

pag

e 2/

3M

ETH

OD

/ EQ

UIP

MEN

TD

ivin

gS

kin

divi

ngD

rop-

Vid

eoS

atel

lite

data

Div

ing

UW

-vid

eoD

ivin

gEC

ON

OM

YTi

me

in F

ield

/sam

ple

6-9d

ives

/day

14ha

/day

*2 p

ers

1,5k

m2/

d*2p

ers

(gro

und

truth

ing)

20 d

ays/

yr

Tim

e in

Lab

/sam

ple

(not

incl

udin

g re

gist

ratio

n an

d nu

mer

ical

ana

lyse

s)lit

tlelit

tle1

sqkm

/day

Labo

rato

ry a

naly

sis

som

e ta

xono

mic

va

lidat

ion

som

e ta

xono

mic

va

lidat

ion

vide

o an

alys

es,

som

e ta

xono

mic

va

lidat

ion

imag

e in

terp

reta

tion

som

e ta

xono

mic

va

lidat

ion

noso

rting

and

dry

w

eigh

t of

sam

ples

INTE

ND

ED U

SE IN

BA

LAN

CE

Hab

itat m

apYE

SYE

SYE

SYE

SYE

SYE

SN

OR

esp.

Var

in M

odel

YES

YES

pot y

espo

t yes

YES

YES

NO

Env

. Var

in M

odel

NO

YES

NO

pot y

esN

OYE

SN

OM

odel

Val

idat

ion

NO

YES

pot y

espo

t yes

NO

NO

NO

Pla

nned

met

hod

for s

patia

l in

terp

olat

ion

not d

ecid

ed y

etm

ovin

g w

indo

w

(som

e)A

rcG

IS to

pogr

id,

pot o

ther

sns

(ful

l co

vera

ge)

NO

NO

NO

Pla

nned

met

hod

for s

patia

l m

odel

ling

none

GR

AS

P,

regr

essi

on tr

eepo

t GR

AS

Ppo

t GR

AS

Ptim

e se

ries,

G

RAS

Pcr

iteria

ana

lysi

scr

iteria

ana

lysi

s

Dat

a pr

oces

sing

refe

renc

es -

Lehm

ann

et a

l 20

02B

ALM

AR

Wen

nber

g, in

pr

epLe

hman

n et

al

2002

Grip

et a

l, in

pr

epG

rip e

t al,

in

prep

ESTI

MA

TIO

N O

F EN

VIR

ON

MEN

TAL

VAR

IAB

LES

AFF

ECTI

NG

DIS

TRIB

UTI

ON

OF

TAR

GET

SPE

CIE

SS

alin

ityH

IGH

HIG

HH

IGH

HIG

HM

ED

--

Wav

e ex

posu

reH

IGH

HIG

HH

IGH

HIG

HH

IGH

HIG

HH

IGH

Dep

thH

IGH

HIG

HH

IGH

HIG

HH

IGH

HIG

HH

IGH

Ligh

tM

ED

-LO

WH

IGH

HIG

HH

IGH

HIG

HH

IGH

HIG

HN

utrie

nts

ME

DM

ED

HIG

HH

IGH

ME

DLO

WLO

WO

xyge

n co

nten

tLO

WLO

WLO

WLO

WLO

WLO

WLO

WTe

mpe

ratu

reLO

WLO

WLO

WLO

WM

ED

LOW

LOW


55

App

endi

x 4,

pag

e 3/

3M

ETH

OD

/ EQ

UIP

MEN

TD

ivin

gS

kin

divi

ngD

rop-

Vid

eoS

atel

lite

data

Div

ing

UW

-vid

eoD

ivin

gSA

MPL

ING

DET

AIL

SS

ampl

ing

stra

tegy

Stra

tifie

dR

ando

m,

stra

tifie

dG

rid, i

nter

vals

50

or 1

50 m

Grid

Tran

sect

0-3

0 m

de

pth

Tran

sect

s ov

er

dept

h gr

adie

nts

Tran

sect

s ov

er

dept

h gr

adie

nts

Sam

plin

g pe

riod

June

July

-Aug

Aug

-Sep

tA

ug-S

ept

June

Sum

mer

200

4-05

Sum

mer

200

4,

06P

rinci

pal s

tudy

are

a, a

nme

and

size

Acc

to G

EU

S,

SG

U o

utpu

t1-

25ha

40km

22

* 60

X 6

0 km

2P

arts

of

Nor

weg

ian

Ska

gerr

rak

coas

t, 20

0 km

co

astli

ne

Off-

shor

e ba

nks

alon

g th

e S

wed

ish

coas

t, no

t Kat

tega

tt or

S

kage

rrak

Off-

shor

e ba

nks

alon

g th

e S

wed

ish

coas

t, no

t Kat

tega

tt or

S

kage

rrak

Oth

er d

ata

colle

cted

with

in s

ame

stud

ysu

bstra

te ty

pe

clas

sde

pth,

su

bstra

te,

turb

idity

, sal

subs

trate

type

, de

pth

dept

h

Dat

a in

terp

reta

tion

in c

ombi

natio

n w

ith o

ther

met

hods

time

serie

sal

l ava

ilabl

e en

v va

rpo

t dep

th,

expo

sure

, su

rface

se

dim

ent

all a

vaila

ble

env

var

time

serie

s of

w

ater

che

mis

try,

tem

p et

c

Is th

e st

udy

desi

gned

to g

eogr

aphi

cally

cov

er th

e fo

llow

ing

grad

ient

s?S

alin

ityN

ON

OS

OM

EYE

SN

ON

ON

OW

ave

expo

sure

NO

SO

ME

YES

YES

SO

ME

NO

NO

Dep

thYE

SN

OYE

SS

OM

EYE

SYE

SYE

SLi

ght

YES

NO

YES

NO

YES

NO

NO

Nut

rient

sN

ON

OS

OM

ES

OM

EN

ON

ON

OO

xyge

n co

nten

tN

ON

ON

ON

ON

ON

ON

OTe

mpe

ratu

reN

ON

ON

ON

ON

ON

ON

OD

oes

the

stud

y co

ver a

reas

out

side

the

prim

ary

dist

rbut

ion

area

ofth

e ta

rget

spe

cies

?N

ON

ON

ON

ON

ON

ON

O


56

10.5 Appendix 5. Methods for sampling phytoplankton, hydrography & hydrochemistry data

Appendix 5, page 1/2METHOD/ EQUIPMENT CTD profiling CTD casts Water sampler HoseInformation provided by HHH /IFM-GEOMAR JA /IAE JA /IAE JA /IAE

PILOT AREA 2 4 4 4TARGET SPECIESSpecies/life stage HYDROGRAPHY HYDROGRAPHY HYDROCHEMISTRY PHYTOPLANKTON

Mobility type passive passive passive passiveTemporal variation days/weeks days/weeks days-months days/weeksOUTPUT VARIABLESName Temperature,

Salinity, Oxygen content

Temperature, Salinity, depth

N02, NO3, N-tot, NH4, PO4, P-tot, Si

species abundance

Unit °C, PSU, ml/l °C, PSU umol/L n/LDerived variable unit - - - biomass mg/m3Positioning accuracy 10m 50m 50m 50mMETHOD DETAILS AND REFERENCESOriginal purpose of study Mapping Monitoring Monitoring MonitoringEquipment details CTD-OTS80 and

CTD-ADMCTD Arop 500 MBL TPN (Transparent

plastic Nansen) with reverse thermometers

Area/volume covered by each sample 250 km**2 Integrated hose sample 0-10 m depth, subsample of 200mL preserved

Estimated accuracy GOOD GOODHow is evaluation achieved? - -Method reference Grasshof 1976 HELCOM 2001,

Ütermöhl 1958

ECONOMYTime in Field/sample 0.25 hours not estimated not estimatedTime in Lab/sample (not including registration and numerical analyses)

no no not estimated

Laboratory analysis no no Standard spectrophotometer

d C l L b

counting and size estimations

INTENDED USE IN BALANCEHabitat map COD/SPRAT map not decided yet not decided yetResp.Var in ModelEnv. Var in Model model input for

circulation modelModel ValidationPlanned method for spatial interpolation

not decided yet

Planned method for spatial modelling not decided yet

Data processing references


57

Appendix 5, page 2/2METHOD/ EQUIPMENT CTD profiling CTD casts Water sampler HoseESTIMATION OF ENVIRONMENTAL VARIABLES AFFECTING DISTRIBUTION OF TARGET SPECIESSalinity HIGH LOW/MODWave exposure LOWDepth HIGH LOWLight HIGHNutrients HIGHOxygen content HIGH MOD/HIGHTemperature LOWSAMPLING DETAILSSampling strategy grid representative representative arbitrary-

representativitySampling period monthly annuallyPrincipal study area, name and size Bornholm Basin

12000 km**2Other data collected within same study ichthyoplankton hydrology,

hydrochemistryData interpretation in combination with other methods

Manually analysed water samples

hydrological and hydrochemical data

Is the study designed to geographically cover the following gradients?Salinity YES NOWave exposure NO NODepth YES YES NOLight NO NONutrients NO NOOxygen content YES NOTemperature YES NODoes the study cover areas outside the primary distrbution area ofthe target species? -


58

tian möllmann & ole vestergaard emi: jonne kotta fgfri: antti … · 2008. 4. 30. · for...

Documents