All Rights for the image reserved to John Hope photostream from flickr, if you keep them,please keep the credit.

“I admit I was very envious each year when colleagues werenamed to The Best Lawyers in Virginia. I am thrilled to now have this singular honor among the rest.”





Abundance Occupancy relationships in North Sea Copepods

ventures into time and space


I'm currently working on publishing the work in my PhD thesis which will be in a (hopefully!) reputable publication near you soon. But here's a taster of what the work is about:




Detecting regime shifts in long-term copepod abundance-occupancy relationships.


The North Sea has been a hotbed for research into marine regime shifts which now appear to be embedded in a wider network of decadal regime shifts throughout the worlds oceans.


Attempts to detect regime shifts generally involve analysis of timeseries of abundance. Instead we approached the task from a macroecological perspective and searched for evidence of temporal heterogeneity in the relationship between local abundance and temporal occupancy.


Changes in AORs were indeed evident, and the time periods identified are consistent with those of other studies in the North Sea and beyond. What's more is the dynamics and regional population structure inferred by the characteristics of AORs. Interspecific differences are consistent with expectations of differences in range position and life history.




Life on the front:


incorporating meso-scale hydrographic structure into high-resolution models of copepod spatial distribution


The Continuous Plankton recorder represents one of the most valuable resources in marine pelagic ecology and beyond. It's great success lies in the low costs of the opportunistic nature of the survey. These very issues however pose problems in interpolation, through sparseness and spatial bias of data points. Interpolation has generally been coarse.


So I developed a species distribution modelling (SDM) approach to interpolation,  informed by high resolution satellite data. I included a number of exciting new variables derived form satellite data, representing hydrographic structure, in particular, maps of front intensity, distance to nearest front and qualitative descriptors of whether pixels lie on the warm or cold side of a front. I used a random forest algorithm to fit interspecific categorical models of copepod abundance in the North Sea and contrasted their outputs to maps generated through geostatistical interpolation, available through the WinCPR portal


The figure shows the distribution of samples, and maps generated by geostatistical interpolation (WinCPR), and SDM interpolation, first aggregated at the coarser WinCPR resolution (AGG) and at the higher resolution of the SDM input variables (HR). WinCPR and aggregated SDM models performed comparably but differences in the outputs are evident particularly in the central North Sea. What I find most exciting though is the interesting detail evident in the SDMHR outputs, where the effects of hydrography in the form of fronts, eddies and filaments are clearly visible.






Contrasting temporal hierarchies in copepod abundance-occupancy relationships: 


the importance of phenology


Once I had outputs from the SDM models I turned my sight to the whole of the North Sea, specifically aiming to explore the temporal hierarchy of copepod AORs in two species belonging to the same biogeographical association but with highly contrasting life history characteristics.  


First, I fit a typical inter-annual AOR using abundance and occupancy calculated at a yearly resolution (ie yearly mean local abundance vs yearly occupancy, top panel in figure).  I then calculated them at monthly resolutions and used a mixed modelling framework to fit both the fixed effects of the relationship, representing a typical intra-annual AOR, and the random effects, representing intra-annual variation around the relationship (middle panels). Finally, driven by very strong patterns in residuals of the intra-annual AOR models, I incorporated harmonic regression terms to account for species' phenological cycles. This allowed me to fit monthly AORs and observe their contrasting seasonal dynamics  (bottom panel).  


This analysis shows that higher level AORs are composites of lower level relationships and that phenology underlies such temporal hierarchies in regional population structure.  Life-history characteristics associated with contrasting phenological cycles are therefore also implicated in producing contrasting temporal AOR hierarchies.