Download presentation
Presentation is loading. Please wait.
1
Stranding Patterns in Stenella spp.
Carissa Cabrera
2
Objective The goal of this project is to determine which variable influences strandings most significantly in 3 species of Stenella dolphins in the MHI from Age Season of stranding Island of stranding Sex Hypotheses & Predictions Ho: no parameters influence stranding patterns Ha: at least 1 parameter influences patterns Season? Methods: NMDS Spinner dolphin (Stenella longirostris) Striped dolphin (Stenella coerualba) Spotted dolphin (Stenella attenuata)
3
Dataset Description 3 species of stenella: 45 strandings
spinner (s. longirostris) striped (s. coerualba) spotted (s. attenuata) 45 strandings in the MHI Variables Age (calf, yearling, subadult, adult) Sex (male, female) Island (Big Island, Oahu, Maui, Kauai) Season (Winter, Spring, Summer Fall) M1: frequencies of strandings by species 16 samples: Combinations of island & season 24 species: Combinations of species + age + sex LSAF: longirostris, subadult, female CAF: coerualba, adult, female ACM: attenuata, calf, male M2: categorical by island (Big Island = 1, Maui = 2, Oahu = 3, Kauai = 4) categorical by season (winter = 1, spring = 2, summer = 3, fall = 4)
4
Dataset Processing Number of species discarded: Final sample size
Species without known answers to the variables in question eliminated Lanai Empty rows: Maui_Spring, Maui_Fall, Kauai_Spring, Kauai_Fall Empty columns: LYM, CCM, CCF, CYM, CYF, ACM, ACF, AYM, AYF, ASAF, AAF No outliers General Relativization by columns p = 1: since columns totals were different & I’m using Sorensen distance measure Final sample size 12 samples 12 species 2 environmental variables
5
Dataset Exploration Uncorrelated environmental variables because they are categorical Islands: Big island = 1 Maui = 2 Oahu = 3 Kauai = 4 Season: Winter = 1 Spring = 2 Summer = 3 Fall = 4 Skewness = 2+ with 75% empty cells Transformations are unable to resolve this, even log transform I want to explore patterns in the data Best to use a nonparametric ordination NMDS
6
Dataset Analysis: Setup
Data already relativized community data, 5 variables
7
Results Interpretation: Number of Axes
Criterion 1: stress reduced by at least 5 per axis Criterion 2: p value < 0.05 Final stress: with 5 dimensions Clarks Rule of Thumb: <5 is ideal
8
Results Interpretation: Scree Plot
Axis 1 & 2 & 3 insignificant because they fall within simulated data Axis 4 & 5 significant, falls below randomized data
9
Results Interpretation
Coefficient of determination axis 1 explains the most Orthogonality: axes are independent 41.0% of the variance is explained by 3 axes Axes 4? 5?
10
Results Interpretation
Axis 3: island gradient? Oahu high Maui middle Kauai & Big Island lower Axis 3: season gradient? Fall higher end Winter lower end Axis 1 & 3 Can’t overlay environmental variables with species because they aren’t quantitative Problematic Stress value <1 so conclusions can be drawn from the ordination, but contradicting criteria results imply caution Winter = 1 Spring = 2 Summer = 3 Fall = 4 Big island = 1 Maui = 2 Oahu = 3 Kauai = 4
11
Results Interpretation: strongest relationships
Adult female spinner dolphins exhibit a strong relationship with axis 3 Tau = 0.545 Strong negative relationship with axis 1 for adult male spinner dolphins Tau = Islands seem to be grouped together. Exception: 1 big island sample?
12
Discussion: the method
1. What do these results mean for the hypotheses/predictions you proposed? Using p values, I would accept Ho Using minimum stress, I would say there are significant patterns in the data, but not what was predicted Not season may be island? 2. What did this exercise teach you regarding your overarching analysis methods and objective? Emphasis of ordination as data exploration rather than hypothesis testing Despite stress values and axes significance, I would not feel comfortable drawing definite or even broad conclusions about my data
13
Discussion: next steps
What do you propose to do for your re-analysis? MRPP to evaluate whether strandings by groups of seasons are distinctly different from one another This is common in other species of marine mammals like California Sea Lions What would be the next steps for this study Quantitative environmental variables would create a more useful ordination plot to confirm if species vary by environmental variables Coordinates? Latitude A way to make stranding date quantitative? Date proportion of year?
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.