Hi, I'm still fairly new to statistics so please bear with me; I apologise for the lengthy question!
I'm just about to start writing up a small project for university (I have just started uni) and need some advice on what statistics to use on my data.
My research question is looking to see if there is a change in the plant species present (or lack of) across a 50m transect on vegetated shingle (starting from near the shoreline where the first plant species are found and continuing up the beaches topographical gradient, which hopefully will allow me to evaluate the effects of disturbance/stress factors on species and species richness).
I have completed four 50m transects (each transect is 50m apart), each containing in total seventeen 4 meter square (2m by 2m) quadrats placed at regular intervals along each transect with 1 metre between each quadrat. Data recorded is quadrat number, position along transect, average pebble size, plant species present, percentage of cover for bare ground/plant species in each quadrat and total species richness for each quadrat.
There is a very clear line of vegetation along the beach where I surveyed so each transect is roughly at the same starting position and follows a similar topography, this will hopefully allow me to reduce noise in the results and reach conclusions as to the effects of disturbance/stress/substrate size on plants found within a vegetated shingle habitat.
So, what I can't do is look for differences between transects but similarities; the differences examined will have to be between the quadrats within each seperate transect to see what changes occur over the transect distance.
The data is non-random so I assume that I'll have to use non-parametric tests such as Kruskal-Wallis but could do with some clarification as to what tests are possible with the data I have.
I have attached a Excel file of my raw data, it will need to be re-organised when I use Genstat to perform tests.
Once again, apologies for the amount that I have written here and thanks in advance for any advice.
Thanks Salvatore, many thanks for your reply. In regards to my non-random data, I was taught that parametric tests assume that the samples are randomly chosen and techniques such as transects produce non-random samples that suit non-parametric tests. Therefore I was assuming that parametric statistical tests such as a one way ANOVA were not appropriate for my data.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics