S TATISTICAL ANALYSIS

Biological characteristics of the sampled snow crab

Size- and sex distribution

Size and sex distribution was graphed, as all other statistical analysis performed, by the use of the software SYSTAT. From this it is possible to get an overview of the sampled crabs, and identify potential size-groups.

Weight-width relationship

The weight-width relationship tells us something about growth pattern of the species, and can be used to compare conditions and feeding status. Easily graphed using carapace width (CW) as the independent variable (x) and wet weight in grams as the dependent variable (y) and log scales to visualize a linear relationship. This gives us an indication about the relation and differing trends among groups. As the size distribution of females and males are different (chapter 2.4) it is logical to analyze the sexes separately.

The relationship between weight and width can best be described by W = a CW^b, which in linear form translates to ln W = ln a + b ln CW, where a is the intercept and b is the slope. To analyze and compare weight-width relationship for multiple groups, regression analysis and analysis of covariance (ANCOVA) was used.

(1) Slope of regression lines

The ln weight - ln width regression model for winter (W) and autumn (A) crabs we estimated:

H0: SlopeA = SlopeW

H1: SlopeA ≠ SlopeW

Model:

ln W = constant + ln CW + Season + ln CW*Season

Where ln CW*Season expresses the slope of the regression line. Using α=0.05 H0 is accepted if p > 0.05 and the ln CW*Season (slope) is not statistically significant different for winter and autumn crabs, and this term should be removed from the model.

(2) Height of regression lines

It is of interest to see if the height of the regression lines is equal, that is if there is no difference in weight according to the same carapace width, provided that the slope is equal, i.e. if the crabs are heavier in one of the season, which might say something about the condition of the crabs:

H0: HeightA = HeightW

H1: HeightA ≠ HeightW

Model:

ln W = constant + ln CW + Season

Using α=0.05 H0 is accepted if p > 0.05, thus there is no statistically significant difference in weight of crabs of the same size between autumn and winter.

Morphology

The relationship between the right chela and CW of males is of interest to the maturation process and differentiation of life-stages as accounted for in chapter 2.4. The

chela-length/height/thickness was thus graphed against CW to shed light on development stages of the sampled crabs.

Diet study

For the diet study a present/absent methodology has been used. Two issues were of interest:

Number of food items consumed and which prey items had the highest occurrence, tested against variables “size-group” and “season”.

Number of prey items in snow crab diet

The number of prey groups consumed gives information on the diversity in feeding habits. It is of interest to investigate if the crabs consume a similar variety of prey items during autumn and winter, and if large and small crabs consume a similar number of prey items.

(1) Seasonal differences in number of prey groups consumed

It is of interest to test whether the mean number prey items consumed during autumn (A) is statistically significant different from mean number of prey items consumed during winter (W). In case of differences between size groups, these should be treated separately. As small crabs only appear for winter, seasons are only compared based on large crabs. Stated as hypotheses:

H0: Mean number of prey itemsA = Mean number of prey itemsW

µA - µB = 0

H1: Mean number of prey itemsA ≠ Mean number of prey itemsW

µA - µB ≠ 0

Where “A” is large male crabs caught during autumn and “W” is large male crabs caught during winter. Using α=0.05 H0 is accepted if p > 0.05, thus there is no statistically significant difference in mean number of consumed prey items.

(2) Differences in number of prey groups consumed for large and small crabs

It is of interest to test whether the mean number prey items consumed by large crabs is the same as for small crabs. As small crabs only appear during winter in this case, winter crabs should be the only crabs analyzed in this case. Stated as hypotheses:

H0: Mean number of prey itemsLarge = Mean number of prey itemsSmall

µLarge - µSmall = 0

H1: Mean number of prey itemsLarge ≠ Mean number of prey itemsSmall

µLarge - µSmall ≠ 0

Where “large” means large male crabs caught during winter and “small” is small male crabs caught during winter. Using α=0.05 H0 is accepted if p > 0.05, thus there is no statistically significant difference in mean number of consumed prey items.

Identified prey items in snow crab diet

The prey items consumed are what primary is of interest to assess the ecological importance of snow crab in the Barents Sea, which is one of the main research questions for this thesis. It is not only of relevance to investigate what snow crab as a whole consume, but also if there are differences within the population, with respect to size and season.

Effect on prey items from season and size class was tested with Fisher’s exact test, as it is categorical data of a present/absent nature in a cross tabulation, and with a small sample size (see e.g. Routledge, 2005).

Relationship among prey items

It is also of interest to see whether any prey groups appear together, and thus if there is any relationship with respect to species interactions. To test if they co-occur in any pattern gives us information to reveal bias in the stomach analysis as well as assessing feeding habits of the different size classes and among seasons. A cluster analysis was performed to illustrate the relationship between prey items.