student presentations

reconsidering the term deep sea

what is usually meant by deep sea?

• common definition: deep sea = deeper than 200 m

• this boundary is historical, not ecological

• originates from the average depth of the north atlantic shelf

why is the 200 m boundary problematic?

• it is:

  • regions specific

  • eurocentric

  • based on early sampling limits

• humans now operate well below 200 m:

  • fishing

  • oil and gas

  • mining

• therefore, 200 m is no longer a functional limit

light does not stop at 200 m

• ligth decreases gradually with depth

• photosyntehsis can occur:

  • much shallower in turbid coastal waters

  • much deeper (>150-200 m) in clear ocean water

• light based zones are variable, not fixed

major environmental gradients extend well beyond 200 m

• temperature:

  • strong changes down to 500 m

• oxygen:

  • OMZz often at several hundred meters

• food supply:

  • most organic carbon found in top 1000 m

main conclusion

• the deep sea is not discrete dpeth zone

• the 200 m boundary is convenient but misleading

• a gradient based definition (light, temperature, oxygen, biology) is more accurate

machine learning traits and plankton

core idea

• plankton imaging + machine learning (ML) allows scientists to move beyond taxonomy and directly measure functional traits from images

• functional traits explain ecosystem function better than species identify alone

why functional traits matter

• traits describe how organisms function, not just who they are

• ecosystem processes (e.g. carbon cycling, productivity) depend on traits, not taxa

• many species → few key traits → simpler ecosystems descriptions

main bottleneck

• human annotation of plankton images is:

  • time consuming

  • expensive

  • the main limitation in plankton image analysis

• ML helps automate this process but still needs training data

measured vs inferred traits (important discussion)

measured traits

• size → master trait

  • length, area, volume

  • links to metabolism, feeding, growth

• shape / body plan

  • affects hydrodynamics and predator avoidance

• cell or body extensions

  • spines, setae, tentacles → feeding and defense

• trasnparency, opacity, colour

  • pigments → photosynthesis

  • opacity → guts, gonads, lipid sacs

inferred traits

• biovolume and biomass

  • estimated from size using empirical relationships

• feeding and metabolic rates

  • gut fullness, ingestion volume

  • scale with body size

• interactions

  • feeding, matign, parasitism, predation

• reproduction

  • egg sacs, clutch size

  • r vs k strategies

key ML approaches you need to recognize

• classification - label organisms or traits (e.g. with/without eggs)

• object detection - locate organism or trait

• segmentation - classify pixels

• deep regression - directly predict a numerical trait

• pose / keypoint estimation - body orientation, appendage position

evaluation first design

• define ecological question first

• then choose;

  • trait to measure

  • ML method

  • evaluation metric

• prevents wasted annotation effort

why this matters for marine biology

• enables:

  • trait based ecosystem models

  • better estimates of primary production and carbon export

  • scaling from individuals → population → ecosystems

• ML methods are data gnostic → transferable to other systems

arctic mercury cycling

why arctic mercury matters

• mercury (Hg) is a toxic pollutant that bioaccumulates and biomagnifies

• arctic wildlife and indigenous peoples have high Hg exposure despite few local sources

• arctic acts as a sink for global mercury, mainyl from lower latitudes

where arctic mercury comes from

• >98% of atmospheric Hg in the arctic comes from outside the region

• trasnported via:

  • atmosphere

  • ocean currents

  • rivers

• local anthropogenic emissions are negligible

key mercury forms (know this)

• Hg(0) - elemental, volatile, dominant in atmosphere

• Hg(II) - oxidised, deposits easily

• methymercury (MeHg) - most toxic, bioaccumulates in food webs

atmospheric processes

• spring atmospheric mercury depletion events (AMDEs)

  • Hg(0) oxidised → Hg(II)

  • driven by halogens (Br) from sea ice and snpw

• 40 to 90% of deposited Hg is re-emitted back to teh atmosphere

terrestrial mercury storage

• arctic soil and permafrost store huge legacy Hg pools

• main pathway

  • atmospheric Hg → vegetation → soil

  • permafrost Hg residence time 1000 years

• warming mobilises stored Hg via

  • permafrost thaw

  • wildfires

  • coastal erosion

river and coastal export

• rivers export Hg mainly during spring snowmelt

• coastal erosion releases Hg from:

  • ice rich permafrist

  • glacial sediment

• climate change -_> more old Hg released

marine mercury

• arctic ocean has higher surface Hg than other oceans

• sea ice:

  • limits air-sea exchange

  • traps Hg in ince and brine

• mercury trasnported out via currents to the atlantic ocean

food webs and humans

• MeHg enters foodd webs via phytoplankton

• biomagnifies → highest levels in:

  • polar bears

  • seals

  • whales

  • seabirds

• humans with traditional diets face higher health risks

climate change effects (very examinable)

• warming causes:  

  • more Hg release from land and ice

  • reduced sea ice → more Hg exchange

  • increased MeHg formation

• leads to greater ecological and human risk

stable isotope tracers

what are stable isotopes and why they matter

• isotopes = same element, different number of neutrons

• stable isotopes do not decay

• widely used isotopes in aquatic ecology:

  • C, N, O, S, Si

• used to trace sources, pathways, and processes in eocsystems

isotopic fractionation (key concept)

• fractionantion = separation of isotopes due to mass differences

• lighter isotopes react react faster than heavier ones

• two types:

  • equilibrium fractionation

  • kinetic fractionation

• all biological, chemical, and physical processes cause fractionation

isotopic fingerprints

• biological processes prefer lighter isotopes

• results in:

  • product (e.g. algae) → enriched in light isotope

  • remaining source → enriched in heavy isotope

• teh predictable difference = isotopic fingerprint

• fingerprints tell us:

  • what process occurred

  • where material cam from

nitrogen cycle application

• stable isotopes used to trace all N-cycle processes

• main source of marine nitrogen: nitrogen fixation

• main sink: denitrification

nitrogen fixation

• average marine

• N2-fixing organisms show little/no fractionation

• low isotope (N) → indicates nitrogen fixation

• high NO3- → indicates denitrification

primary production

• main source of organic carbon in aquatic systems

• 13C used to measure primary production

• enrichment experiments use HCO#-

• helps study effects of:

  • light

  • nutrients

  • temperature

  • species differences

carbon sources

• aloocthonous (terrestrial) carbon:

  • stirng fractionation

  • low C

• autochthnous (marine) carbon:

  • higher

• fossil fuels have a distinct C signal

• C can be used to track anthropogenic CO2 uptake by oceans

food webs and trophic position

• stable isotopes used to reconstruct food webs

• N increases with each trophic level C identifies carbon source

• aminoacid specific isotopes

  • separate baseline vs trophic enrichment

• integrates diet over time (not a snapshot)

pathways and coupling

• combining isotopes (C, N, S, Si) allows tracing:

  • nutrient uptake

  • turnover rates

  • pelagic bethic coupling

• useful for understanding biogeochemical pathways

isoscapes

• isotopic landscapes (maps of isotope ratios)

• link isotope to:

  • environemntal gradients biogeochemical processes

• used to track:

  • animal migration

  • nutrient sources

  • pollutant pathways

  • water mass movement

eDNA vs conventional biodiversity monitoring

what is eDNA

• environmental DNA (eDNA) = DNA extracted from water, soil, or air without capturing organisms

• used for biodiversity monitoring, species detection, and comunity assessment

why eDNA is important

• global biodiversity loss → need for better monitoring

• conventional methods limited by:

  • cost

  • time

  • taxonomic expertise

  • inaccessible areas

• eDNA offers a new monitoring tool, not a replacement

main advantages of eDNA

• higher sensitivity

  • detects rare, cryptic, small species

• higher efficiency

  • lower cost

  • less field time

  • non-invasive

• low taxonomic expertise needed

  • effective for:

    • endangered specie

    • invasive species

    • general biodiversity surveys

main limitations of eDNA

• can detect DNA from:

  • species not currently present

  • trasnported DNA (fasle positives)

• lower taxonomic resolution for some groups

  • due to incomplete reference databases

• abundance estimates are variable

  • semi-quantitative, not always reliable

how does eDNA compare to conventional methods?

• eDNA more likely to show:

  • higher species richness

  • higher detection probability

• eDNA usually shows

  • different community composition than conventional methods

• conventional methods:

  • better for accurate counts, biomass, size, condition

• conclusion:

  • eDNA provides a different perspective, not the same information

key results from the systematic review

• 400 comparitive studeis reviewed

• eDNA generally outperforms conventional methods for:

  • sensitivity (species richness, detection)

  • efficency (cost, effort)

• results are consistent across environments and taxa

• strong bias toward:

  • aquatic systems

  • global north

major knowledge gaps

• temporal monitoring

  • most studies sample only once

• global south

  • few studies in biodiversity rich tropical regions

• diversity metrics

  • over-reliance on species richness

  • limited ecosystem-level metrics

take home message

• eDNA methods are generally more sensitive and efficient than conventional methods, but they detect different communities and should be used as a complement rather than a replacement