CTcoral is a co­oper­a­tion pro­jet with the MARUM institute in Bremen.

Con­tact per­son at MARUM: Jürgen Titschack

Frame­work-form­ing cold-wa­ter cor­als (CWC) are im­port­ant hab­itat en­gin­eers, which form reefs in the deep sea that present sig­ni­fic­ant biod­iversity hot­spots. Over time these reefs build large sea­floor obstacles, called coral mounds, which present im­port­ant car­bon sinks along con­tin­ental mar­gins. While our know­ledge on the en­vir­on­mental needs of CWC, the global dis­tri­bu­tion of CWC reefs and the form­a­tion of coral mounds con­tinu­ously in­creases, our un­der­stand­ing of the morpho­plas­ti­city of CWC re­gard­ing their cor­al­lite, colony and frame­work mor­pho­logy and its in­flu­ence on these sys­tems is still highly lim­ited due to the lack of ob­server-in­de­pend­ent quant­it­at­ive meth­od­o­lo­gies.

CTcoral aims to de­velop an auto­mated meth­od­o­logy to taxo­nom­ic­ally clas­sify and mor­pho­lo­gic­ally char­ac­ter­ise the most im­port­ant frame­work-form­ing CWC from 3-di­men­sional (3D) com­puted tomo­graphy (CT) scans. To achieve this goal, we will im­ple­ment ma­chine-learn­ing tech­niques that com­bine the ad­vant­ages of im­age- and shape-based clas­si­fic­a­tions. Since mar­ine skel­et­ons and shells are af­fected by bio­er­o­sion, which al­ters their shape and hence im­pacts the al­gorithm per­form­ance, we will fur­ther de­velop al­gorithms to seg­ment bio­er­o­sion and pro­duce bioerosion-corrected data­sets to com­pensate this ef­fect. Fur­ther­more, as the ac­cur­acy and con­fid­ence of the taxo­nomic clas­si­fic­a­tion and mor­pho­lo­gical char­ac­ter­isa­tion de­pend on coral frag­ment sizes and CT scan res­ol­u­tion, we will test to which de­gree the loss of in­form­a­tion by small frag­ment sizes and low res­ol­u­tion can be com­pensated by us­ing mul­tiple frag­ments from one sea­floor sample and by in­cor­por­at­ing ad­di­tional mor­pho­lo­gical cri­teria, re­spect­ively. The de­veloped meth­od­o­logy will be ap­plied to  ~370 CWC spe­ci­mens from all over the At­lantic Ocean to provide the first com­pre­hens­ive de­scrip­tion of the morpho­plas­ti­city of the most im­port­ant CWC.

CTcoral is an in­ter­dis­cip­lin­ary pro­ject that com­bines the ex­pert­ise of mar­ine geoscient­ists and com­puter sci­ent­ists. It forms the basis for a fol­low-up pro­posal that will con­cen­trate on the ap­plic­a­tion of the de­veloped meth­od­o­logy to in­vest­ig­ate en­vir­on­mental pref­er­ences of spe­cific cor­al­lite, colony and frame­work mor­pho­lo­gies and the in­flu­ences of coral com­munit­ies and their re­spect­ive mor­pho­logy on coral mound form­a­tion. The de­veloped meth­ods within CTcoral are ex­pec­ted to be of broad ap­plic­ab­il­ity to other or­gan­ism groups and will provide an im­port­ant con­tri­bu­tion on the road to­wards cy­ber­tax­onomy.