This is a sel -a chi ed e sion o an o iginal a icle. This e sion
may di e om he o iginal in pagina ion and ypog aphic de ails.
Au ho (s):
Ti le:
Yea :
Ve sion:
Copy igh :
Righ s:
Righ s u l:
Please ci e he o iginal e sion:
CC BY 4.0
h ps://c ea i ecommons.o g/licenses/by/4.0/
Calib a ion o in si u chlo ophyll luo ome e s o o ganic ma e
© The Au ho (s) 2019
Published e sion
Kuha, Jonna; Jä inen, Ma ko; Salmi, Pauliina; Ka jalainen, Juha
Kuha, J., Jä inen, M., Salmi, P., & Ka jalainen, J. (2020). Calib a ion o in si u chlo ophyll
luo ome e s o o ganic ma e . Hyd obiologia, 847(21), 4377-4387.
h ps://doi.o g/10.1007/s10750-019-04086-z
2020
RESTORATION OF EUTROPHIC LAKES
Calib a ion o in si u chlo ophyll luo ome e s o o ganic
ma e
Jonna Kuha .Ma ko Ja
¨ inen .Pauliina Salmi .Juha Ka jalainen
Recei ed: 1 Feb ua y 2019 / Re ised: 16 Sep embe 2019 / Accep ed: 28 Sep embe 2019 / Published online: 5 No embe 2019
ÓThe Au ho (s) 2019
Abs ac O ganic ma e (OM) o he han li ing
phy oplank on is known o a ec luo ome ic in si u
assessmen s o chlo ophyll in lakes. Fo his eason,
calib a ing luo ome ic measu emen s o OM e o is
impo an . In his s udy, chlo ophyll (Chl) luo es-
cence was measu ed in si u in mul iple Finnish lakes
using wo sondes equipped wi h Chl luo ome e s
(ex.470/em.650–700 nm). OM abso bance (A
420
) was
measu ed om wa e samples, and one o he wo
sondes was also equipped wi h in si u luo ome e o
OM (ex.350/em.430 nm). The sonde wi h Chl and OM
luo ome e s was also deployed con inuously on an
au oma ed wa e quali y moni o ing s a ion on Lake
Konne esi. Fo da a om mul iple lakes, inclusion o
wa e colou es ima es in o he calib a ion model
imp o ed he p edic abili y o Chl assessmen s
ma kedly. When OM abso bance o in si u OM
luo escence was used in he calib a ion model,
p edic abili y be ween he in si u Chl and labo a o y
Chl a assessmen s was also enhanced. Howe e ,
co ec ion was no supe io o he one done wi h he
wa e colou es ima e. Ou esul s demons a ed ha
co ec ion wi h wa e colou assessmen s o in si u
measu emen s o OM luo escence o e s p ac ical
means o o e come he a ia ion due o OM when
assessing Chl in humic lakes in si u.
Keywo ds Au oma ed moni o ing Chlo ophyll a 
Fluo escence O ganic ma e Op ical senso s 
Wa e colou
In oduc ion
Phy oplank on biomass is widely used as an indica o
o eu ophica ion in he s a us assessmen o su ace
wa e s. Chlo ophyll aconcen a ion is used as a p oxy
o phy oplank on biomass, adi ionally quan i ied in
labo a o y om wa e samples by e hanol ex ac ion
ollowed by spec opho ome ic measu emen a
wa eleng hs 665 and 750 nm (Lo enzen, 1967); ISO
10260, 1992). The sho comings o his p o ocol a e
also well known: i equi es la ge sample olumes,
sample anspo a ion, s o age and handling. The e-
o e, spa ial and empo al co e age and ep esen a-
i eness o adi ional measu emen s a e gene ally
limi ed. Du ing he las wo decades, ield
Gues edi o s: Tom Jilbe , Raoul-Ma ie Cou u e,
B ian J. Huse & Kale i Salonen / Res o a ion o eu ophic
lakes: cu en p ac ices and u u e challenges
J. Kuha (&)P. Salmi J. Ka jalainen
Depa men o Biological and En i onmen al Science,
Uni e si y o Jy a
¨skyla
¨, PO Box 35, 40014 Jy a
¨skyla
¨,
Finland
e-mail: [email p o ec ed]
M. Ja
¨ inen
F eshwa e Cen e/Wa e In o ma ion Sys em Uni ,
Finnish En i onmen Ins i u e (SYKE), La oka anonkaa i
11, 00790 Helsinki, Finland
123
Hyd obiologia (2020) 847:4377–4387
h ps://doi.o g/10.1007/s10750-019-04086-z(0123456789().,- olV)(0123456789().,- olV)
spec o luo ome e s o measu e chlo ophyll in si u
ha e become inc easingly common wo ldwide (Mein-
son e al., 2015). In he ex , we use ‘‘Chl a’’ o e e o
chlo ophyll aconcen a ion measu ed in labo a o y, in
con as o ‘‘Chl’’ ha e e s o in si u luo escence o
chlo ophyll.
Due o he ease o measu ing, Chl luo ome e s a e
equen ly used o s udy phy oplank on dis ibu ion,
ac i i y and popula ion dynamics in si u (P oc o &
Roesle , 2010; Zeng & Li, 2015). Fluo ome ic
quan i ica ion o Chl is gene ally cos -e ec i e and
allows equen obse a ions du ing sudden phenom-
ena such as mixing e en s o sho -li ed algal blooms
(Jennings e al., 2012; Klug e al., 2012). On he o he
hand, in e p e a ion o in si u Chl luo ome e da a is
no s aigh o wa d. Values yielded by he luo ome-
e s a e a bi a y and need o be calib a ed o, o
example, adi ional Chl aconcen a ions in o de o
in e p e he measu emen s, and calib a ion is a ec ed
by se e al sou ces o a ia ion. In addi ion o phy o-
plank on biomass, he in ensi y o Chl luo escence is
dependen on he pigmen composi ion and physio-
logical eac ions o phy oplank on (Williams &
B idges, 1964; Seppa
¨la
¨& Balode, 1998; Richa dson
e al., 2010). Physical and chemical condi ions in he
wa e also a ec he in si u Chl luo escence. Such
a iables include changes in he unde wa e ligh and
empe a u e condi ions, back sca e ing and abso p-
ion o ligh by o he pa icles han phy oplank on
(S ickland, 1968; Ca de e al., 1991; Se a e al.,
2009; Downing e al., 2012; Os owska, 2012). Mos
p ominen ly, Chl luo escence signal is a ec ed by
luo escence o o ganic ma e (OM) ela ed o some-
hing else han li ing phy oplank on (P oc o &
Roesle , 2010; Twiss, 2011).
P ac ices on how o con ol he sou ces o e o due
o OM a e unde a igo ous scien i ic discussion. OM
is a b oad e m, including DOM (dissol ed o ganic
ma e ) and POM (pa icula e o ganic ma e ). In
li e a u e, OM co ec ion has been conduc ed by using
concep s o soluble luo escence (Ca lson & Shapi o,
1981), yellow subs ances (Twiss, 2011), luo escen o
ch omopho ic dissol ed o ganic ma e (CDOM;
P oc o & Roesle , 2010) o humic subs ances (Ca l-
son & Shapi o, 1981). In his s udy, we use he gene al
e m OM o desc ibe o ganic ma e ial o he han
phy oplank on in he lake wa e , because we did no
ac iona e o cha ac e ize he di e en OM compo-
nen s. P e ious s udies ha e shown ha he OM signal
is gene ally linked o alloch honous dissol ed humic
subs ances in bo eal lakes (Ca lson & Shapi o, 1981;
Twiss, 2011). In he ma ine en i onmen , phy oplank-
on deg ada ion p oduc s may comp ise a la ge pa o
o al OM (Xing e al., 2017).
The a ie y o p oposed measu es o con ol he
OM signal is b oad, indica ing i s impo ance in Chl
luo escence p oblema ics. Ca lson & Shapi o (1981)
il e ed a sample wa e o sub ac OM backg ound
om Chl luo escence in No h Ame ican lakes and
Leppa
¨e al. (1995) ollowed he same p o ocol in
Finnish lakes. Twiss (2011) and P oc o & Roesle
(2010) used mul iple wa eband senso s o measu e
luo escen OM in lakes and o es ablish a calib a ion
model be ween Chl and OM, whe eas Xing e al.
(2017) used a single wa eband senso o luo escen
OM o co ec in si u Chl luo escence alues in deep
e ical p o iles om ma ine en i onmen . In si u Chl
luo ome e s a e becoming inc easingly common in
he moni o ing o inland wa e s, bu in e na ional
s anda ds o ield ope a ion a e s ill unde de elop-
men (e.g. La igne e al., 2012; C emella e al., 2018).
In he nea u u e, inc ease in wa e colou may
become mo e impo an due o b owning o bo eal
lakes (Mon ei h e al., 2007; E ans e al., 2012;
K i zbe g & Eks o
¨m, 2012). Co ec ions o Chl o
OM luo escence would hus become inc easingly
ele an in moni o ing.
In bo eal lakes, he connec ion be ween dissol ed
OM and wa e colou is ypically s ong and a he
uni o m due o he p esence o humic subs ances
(Ca lson & Shapi o, 1981; Hessen & T an ik, 1998;
Keski alo & Elo an a, 1999). Resul s by Leppa
¨e al.
(1995) indica ed good co ela ion be ween wa e
colou and backg ound luo escence in ou een lakes
in Eas e n Finland. Fo his eason, and he simplici y
o wa e colou measu emen , he main aim o his
s udy was o sc u inize wa e colou as a p oxy o OM
in co ec ion in humic lakes. In his s udy, we es ed an
in si u OM luo escence senso o co ec Chl esul s.
Majo luo escence peaks o humic subs ances in
wa e consis o exci a ion be ween 330 and 390 nm
and emission be ween 420 and 500 nm wa eleng hs
(Ma ilainen e al., 2011), and he OM senso deployed
in his s udy co e ed he same wa eleng h ange. We
hypo hesized ha co ec ing luo ome ic Chl da a
wi h (1) a e age o median wa e colou , (2) wa e
colou a he measu emen dep h assessed ia
abso bance (A
420
), o (3) in si u luo ome ic
123
4378 Hyd obiologia (2020) 847:4377–4387
measu emen s o OM imp o es, and equally well, he
p edic abili y be ween luo ome ic Chl and adi-
ional labo a o y Chl aassessmen . We aimed also o
ob ain in o ma ion o he seasonal OM a ia ion
wi hin one lake and o de e mine whe he con inuous
moni o ing o Chl could bene i om he OM
calib a ion.
Ma e ials and me hods
S udy lakes
This s udy was conduc ed on lakes loca ed in Cen al
and Sou he n Finland (N-Eu ope) a ying in size,
nu ien concen a ion and wa e colou (Table 1). To
unde s and lake-speci ic e ec s o di e en calib a-
ion models, Lakes Konne esi and Jy a
¨sja
¨ i in
Cen al Finland and Lakes Vesija
¨ i and Vanaja esi
in Sou he n Finland we e sc u inized sepa a ely. O
he lakes, Jy a
¨sja
¨ i and Vanaja esi ep esen small
and la ge humic lakes, espec i ely. Based on hei
ypical Chl aand o al phospho us concen a ions
(Table 1), hese lakes a e oligo-meso ophic. Lakes
Konne esi and Vesija
¨ i ep esen la ge clea wa e
lakes, he la e being oligo ophic and he o me a
mo e eu ophic lake.
In si u Chl measu emen s and Chl a labo a o y
analysis
In si u measu emen s on he selec ed lakes we e done
wi h wo sepa a e YSI6600 mul ipa ame e sondes
( ype V2-4; YSI Inc., Yellow Sp ings, OH, USA;
he ea e Sondes 1 and 2) ha we e equipped wi h Chl
luo ome e s (ex.470/em.650–700 nm) wi hou em-
pe a u e co ec ion. Sonde 1 was used o measu e Chl
luo escence on six o he s udy lakes including bo h
humic and clea wa e lakes (Table 1). Samples o he
labo a o y analysis o Chl awe e aken a he dep h o
1 m and se e al o he dep hs om he lakes measu ed
wi h Sonde 1 (see he link in Da a a ailabili y o
alues). Samples we e aken wi h a 4.4 l Limnos- ube
sample and s o ed in cool (6°C) and da k un il he
analysis wi hin 24 h a e he collec ion. Chl a(SFS-
ISO 10260:1992) was measu ed in he labo a o y a e
il a ion o 0.5–1.0 l o sample wa e h ough GF/C
il e s. Chl awas measu ed using he cold e hanol
ex ac ion me hod and wa e colou spec opho ome -
ically. Wa eleng hs 665 and 750 nm we e measu ed
Table 1 Loca ions and basic limnological cha ac e is ics o he s udy si es
Lake Coo dina es (*WGS84) A ea
(ha)
Dep h (m) To -P,
(lgl
-1
)
Chl a,
(lgl
-1
)
Wa e colou ,
(mg P l
-1
)
Sonde
La Lon Max Mean
Al aja
¨ i 62°18050.00400 25°43011.65500 210 16.5 3.8 29 16.5 80 1
Jy a
¨sja
¨ i 62°14011.67100 25°4602.86800 314 25.0 7.0 25 10.8 70 1
Konne esi 62°37057.36600 26°36016.50400 19,028 57.1 10.6 6 4.2 25 1
Ruokoja
¨ i 62°15032.90200 27°18041.2800 464 8.2 1.3 30 10.6 110 1
Vanaja esi 61°9019.71300 24°13044.98200 10,261 23.9 7.7 24 16.0 50 1
Vesija
¨ i 61°301.66200 25°3504.92600 6,471 40.0 6.1 27 9.6 10 1
Alasenja
¨ i 61°0046.56400 25°44035.2900 275 15.2 6.1 14 4.5 10 2
Alinen Rau ja
¨ i 61°11043.09800 25°601.24100 50 12.0 No da a 23 23.0 100 2
A kiomaanja
¨ i 61°3014.65900 25°44026.69200 208 20.2 5.1 17 5.3 25 2
Jou ja
¨ i 60°58036.40600 25°4206.56700 40 5.0 3.3 25 24.0 20 2
Me asja
¨ i 61°103.69500 25°4104.86800 24 2.6 1.5 31 19.3 50 2
Pa
¨a
¨ja
¨ i 61°3048.7600 25°7056.99800 1,352 85.0 14.8 12 5.5 70 2
Ruuhija
¨ i 61°1038.01100 26°0028.05400 573 18.7 5.6 18 10.8 30 2
Tyo
¨ ja
¨ i 60°59044.51500 25°2802.63500 56 7.0 1.5 23 No da a 50 2
Limnological da a om he da abase o Finnish En i onmen al Ins i u e (SYKE) o Dolman e al. (2015)
Sonde e e s o ei he o he wo mul ipa ame e sondes (Sonde 1 and Sonde 2) ha was deployed on he lake
To -P o al phospho us concen a ion, Chl a chlo ophyll aconcen a ion
123
Hyd obiologia (2020) 847:4377–4387 4379
wi h he Shimadzu UV-1800 spec opho ome e (Shi-
madzu Co., Kyo o, Japan) in 5-cm qua z cu e es.
Sonde 2 was used o measu e Chl luo escence in
eigh lakes, also including humic and clea wa e lakes
(Table 1). Ve ical p o iles we e eco ded, as o
Sonde 1, and in eg a ed (0–2 m) samples o Chl
acollec ed wi h a Limnos sample . Chl asamples
we e il e ed using Wha man GF/F il e s and s o ed in
a eeze (-20°C) un il measu emen wi hin
1–3 weeks a e sampling using a ho e hanol ex ac-
ion me hod and spec opho ome e (SFS-ISO 10260,
1992; Dolman e al., 2015). The me hods o Chl
aex ac ion we e di e en o he wo da ase s,
because he ex ac ions we e ca ied ou in wo
di e en ins i u es (Lammi Biological S a ion o
Uni e si y o Helsinki and Uni e si y o Jy a
¨skyla
¨).
Howe e , bo h me hods a e gene al and well-es ab-
lished p ocedu es in labo a o y analysis o Chl a.
Wa e colou o he s udy lakes (colou
ypical
)
In his s udy, we i s es ed he epilimne ic wa e
colou o he s udy lakes as a pa ame e (he ea e
colou
ypical
) o co ec Chl measu emen s. Fo mos
o he s udy lakes, hese colou alues, also used o
lake ypology classi ica ion, we e a ailable om he
da abase o Finnish En i onmen Ins i u e (SYKE,
h ps://wwwp2.ympa is o. i/sc ip s/ki jaudu.asp, equi es
ee egis a ion and use o websi e ansla o ). Wa e
colou was analysed ei he wi h a Hellige/AVM Neo-
compa a o o spec opho ome ically (wa eleng h
410 nm) acco ding o SFS-EN-ISO 7887:2011. The
wa e colou o he me hods a e highly compa able.
Fo ou o he s udy lakes, colou
ypical
alue was
aken om Dolman e al. (2015), and we e de e mined
using spec opho ome ic analysis om 0.45 lm p e-
il e ed samples (SFS-EN-ISO 7887, 2011, 410 nm).
On si e wa e colou measu emen s (colou
Lab
)
F om he lakes measu ed wi h Sonde 1, samples o he
labo a o y analysis o wa e colou (colou
Lab
) we e
aken a he same dep hs as samples o labo a o y
analysis o Chl a(see he link in Da a a ailabili y o
alues). Samples we e aken as subsamples om he
4.4-l Limnos- ube sample and s o ed in cool and da k
condi ions un il he analysis wi hin 24 h a e he
collec ion. Wa e colou was measu ed in he labo a-
o y a e il a ion o 0.5–1.0 l o sample wa e
h ough GF/C il e s. Abso bance a 420 nm wa e-
leng h was measu ed wi h he Shimadzu UV-1800
spec opho ome e using 1-cm qua z cu e es and
con e ed o mg P l
-1
(SFS-EN-ISO 7887, 2011).
In si u luo escence o o ganic ma e
Sonde 1 was equipped wi h Cyclops-7 o ganic ma e
(OM) luo ome e (ex.350/em.430 nm; Tu ne
Designs Inc., Sunny ale, CA, USA). The OM luo-
ome e was calib a ed o wa e empe a u e as in
Wa as e al. (2011), o which he empe a u e
coe icien (q)o -0.009 was used o ans o m da a
o a e e ence empe a u e o 20°C (RFU
20
).
Con inuous wa e quali y moni o ing on Lake
Konne esi
To es he pe o mance o di e en calib a ion
me hods and o s udy seasonal a ia ion in Chl and
OM, Sonde 1 was con inuously also deployed in Lake
Konne esi du ing he open wa e season in 2013
(June–Oc obe ). One p o ile in e e y h ee hou s a
0.5-m s ep was eco ded om he 42-m-deep wa e
column. Daily a e ages o Chl and OM om 1.5 o
2.0 m dep h we e aken in o analysis. Su ace (0– m)
Chl ada a om 2013 we e collec ed by he egional
en i onmen al agency KES-ELY du ing he open
wa e pe iod (June–Oc obe , n= 5, Da abase o
Finnish En i onmen Ins i u e) and supplemen ed by
ou Chl asamplings (n= 4) om 1 m and 2 m dep hs
wi h he Limnos sample . Reg ession models
(Table 2a, b) de eloped o he mul iple lakes da ase
we e es ed o calib a e he con inuous Chl da a.
Compa ison o he calib a ion me hods
We compa ed Chl luo ome e da a and co esponding
Chl a esul s om labo a o y analyses and es ablished
calib a ion equa ions o he Chl luo ome e by i ing
a linea eg ession model. Then we added wa e colou
(colou
Typical
) in o he calib a ion equa ion. Fo Sonde
1, we also used colou
Lab
aken a he ime o p o ile
measu emen s and included o in si u OM luo ome e
da a in o he calib a ion models. Su ace (a e age o
1–2 m) Chl and OM luo ome e esul s om Lake
Konne esi (Cen al Finland) we e co ec ed wi h each
o he es ablished calib a ion equa ions, excluding
123
4380 Hyd obiologia (2020) 847:4377–4387

colou
Lab
, o illus a e he easibili y o he calib a ion
o au oma ed wa e quali y moni o ing.
Calcula ions o modelling e iciency
(ME = 1 -P(y
o
-y
p
)
2
/P(y
o
-y
m
)
2
), whe e y
o
ep esen s obse ed alues, y
p
p edic ed alues, and
y
m
he mean o obse ed alues, and mean absolu e
pe cen age e o (MAE(%) = 100[P(|y
o
-y
p
|/y
o
|)]/
n,), whe e nis he numbe o pai s, we e conduc ed
acco ding o Maye & Bu le (1993) using Mic oso
Excel (2010, Mic oso Co., Redmond, WA, USA).
Linea eg essions and s a is ical es s we e conduc ed
wi h SPSS so wa e ( e sion 22.0, IBM Co., Chicago,
IL, USA). Resul s wi h P 0.05 we e epo ed
signi ican .
Resul s
In he lakes measu ed wi h Sonde 1 (see Table 1),
manually sampled Chl aconcen a ion a ied be ween
0.5 and 17.4 lgl
-1
, and in he lakes measu ed wi h
Sonde 2 be ween 3.6 and 43.7 lgl
-1
(Fig. 1). A e
Table 2 Uni a ia e and mul i a ia e linea eg ession models o he calib a ion o he Sonde 1 (a) and Sonde 2 (b) luo ome e s
using he labo a o y measu ed chlo ophyll a
a. Model equa ion o Sonde 1 nR
2
P
(1) Chl a= 1.717(Chl) -1.944 71 0.537 0.001
(2) Chl a= 3.024(Chl) -0.162(colou
ypical
)-1.877 71 0.914 0.001
(3) Chl a= 2.843(Chl) -0.133(colou
Lab
)-1.067 71 0.816 0.001
(4) Chl a= 2.288(Chl) -5.109(OM) ?3.613 71 0.879 0.001
b. Model equa ion o Sonde 2 nR
2
P
(1) Chl a= 0.811(Chl) -4.902 71 0.537 0.001
(2) Chl a= 1.076(Chl) -0.165(colou
ypical
)-2.938 71 0.914 0.001
(Chl a) concen a ions (lgl
-1
) and wa e colou (colou
ypical
, colou
Lab
,mgP l
-1
) and in si u luo escen o ganic ma e OM (RFU).
nnumbe o obse a ions (samplings on lakes) used o cons uc he models
Fig. 1 Linea eg ession (b oken line) be ween he obse ed
Chl aconcen a ions and es ima ed Chl aconcen a ions based
on he calib a ion wi h measu ed Chl ain he s udy lakes (model
1; see Table 2a, b). Solid line = 1:1 line. Do s—Sonde 1 da a;
open ma ke s—Sonde 2 da a; MAE mean absolu e pe cen age
e o (%), ME modelling e iciency (%)
123
Hyd obiologia (2020) 847:4377–4387 4381
calib a ing he luo ome e eadings only wi h he Chl
aconcen a ions (model 1, Table 2a, b), he eg ession
coe icien s, modelling e iciency (ME) and mean
absolu e pe cen age e o s (MAE) we e a he weak
(R
2
= 0.51 and 0.74, MAE = 89% and 38%, ME =
0.53 and 0.86, o Sondes 1 and 2, espec i ely,
Fig. 1).
The espec i e p edic abili ies o model 2, includ-
ing bo h Chl aand wa e colou (colou
ypical
)as
co ec ing a iables, we e no ably be e o bo h
sondes (R
2
= 0.81 and 0.92, MAE = 66% and 24%,
ME = 0.82 and 0.95, o Sondes 1 and 2, espec i ely,
Fig. 2).
Fo Sonde 1, we also cons uc ed models 3 and 4,
including Chl aand colou
Lab
o OM luo escence,
espec i ely. Model 3 yielded sligh ly be e p e-
dic abili y (R
2
= 0.85, MAE = 64%, ME = 0.86) han
he wa e colou (colou
ypical
, Fig. 3). Fo model 4,
in oducing OM luo escence as a calib a ion ac o
was ela i ely e ec i e (R
2
= 0.77, MAE = 65%,
ME = 0.78), bu no supe io o models 2 o 3. Chl
a alues yielded by di e en models did no di e
signi ican ly om each o he (Rela ed Samples Wil-
coxon Signed Rank Tes , P[0.2 o each combina-
ion pai o models).
The ela ionship be ween colou
Lab
and OM luo-
escence (RFU
20
) was loga i hmic (Fig. 4): OM
luo escence i s inc eased apidly along colou
Lab
,
bu in lakes wi h colou
Lab
[50 mg P l
-1
i le elled
o (Fig. 4).
Sc u iniza ion o he pe o mance o each model o
a sub-se o lakes co obo a ed he gene al o e iew
and demons a ed ha , p io o any OM co ec ion, he
a ia ion in he luo ome e eadings calib a ed only
wi h he Chl a a ied no ably om he Chl ameasu e-
men s e en inside one lake (Figs. 5,6). In humic
Lakes Jy a
¨sja
¨ i and Vanaja esi, he p edic abili y
be ween he wo was mode a e, bu mean absolu e
e o was high (R
2
= 0.88, MAE = 97%, ME = 0.83,
Fig. 5). In clea wa e lakes Konne esi and Vesija
¨ i,
he p edic abili y wi hou any OM co ec ion was
weake bu mean absolu e e o sligh ly lowe
(R
2
= 0.56, MAE = 86%, ME = 0.71, Fig. 6). How-
e e , when any o he OM co ec ion models (models
2–4 in Table 2a, b) was applied, he p edic abili ies
inc eased in all ou lakes. In Jy a
¨sja
¨ i and Vana-
ja esi, he eg ession coe icien s, mean absolu e
e o s and he modelling e iciencies o models 2–4
a ied only li le (R
2
= 0.93–0.97, MAE = 25–36%%,
ME = 0.93–0.97, Fig. 5). Simila ly, clea wa e lakes
Konne esi and Vesija
¨ i bene i ed om each o he
calib a ion me hods a he equally (R
2
= 0.74–0.76,
MAE = 74–79%, ME = 0.85–0.87, Fig. 6).
In he oligo ophic, humic Lake Konne esi, Chl
a a ied be ween 1.8 and 6.2 lgl
-1
in June–Oc obe
2013, colou
ypical
was 25 mg P l
-1
and OM
Fig. 2 Linea eg ession
(b oken line) be ween he
obse ed Chl
aconcen a ions and
es ima ed Chl
aconcen a ions based on
he calib a ion wi h
measu ed Chl aand wa e
colou es ima e
(colou
ypical
, Table 1).
Reg ession model 2 was
used he e (see Table 2a, b).
Solid line = 1:1 line. Do s—
Sonde 1 da a; open
ma ke s—Sonde 2 da a;
MAE mean absolu e
pe cen age e o (%), ME
modelling e iciency (%)
123
4382 Hyd obiologia (2020) 847:4377–4387
luo escence (RFU
20
) a ied be ween 0.95 and 58
(Fig. 7). When models 1, 2 and 4 (Table 2a) we e used
o co ec he au oma ed luo ome e moni o ing da a,
only li le di e ences be ween he modelled Chl
a esul s we e ound (Fig. 7). Howe e , he dec ease
o wa e colou in au umn 2013, moni o ed wi h in si u
OM, p oduced a pe iod wi h highe Chl luo escence
in Sep embe han he models using co ec ion wi h
only Chl ao Chl aand colou
ypical
(Fig. 7).
Discussion
Ou s udy demons a ed he use o wa e colou as an
es ima e o OM in con olling he OM luo escence in
Chl in si u luo ome e da a. The applicabili y o a
ypical colou alue o he lake in calib a ion was
alida ed by compa ing i wi h he calib a ions done
using simul aneously measu ed wa e colou and
in si u OM luo escence. As we expec ed, inclusion
o any o hese pa ame e s in o he calib a ion model
imp o ed he p edic abili y be ween he in si u Chl
luo escence and Chl a.
A non-linea ela ionship be ween OM and wa e
colou is well documen ed (Wa as e al., 2011; Coble
e al., 2014). High concen a ions o dissol ed o ganic
ma e ial in he wa e column a ec he OM luo es-
cence measu emen s by supp essing he alues a
highe dissol ed OM concen a ions (Wa as e al.,
2011). We also obse ed his in ou s udy lakes
(Fig. 4), and i may explain why he in si u OM
calib a ion was no a supe io me hod o e he mo e
simplis ic wa e colou calib a ion. P oc o & Roesle
(2010) ou lined simila ly ha OM may lead o an
unde es ima ion o Chl aby abso bing exci a ion o
emission wa eleng hs o , on he o he hand, OM may
cause seemingly in ensi ied Chl emission by con-
ibu ing o he signal de ec ed by Chl luo ome e s. In
Fig. 3 Linea eg ession
(b oken line) be ween he
obse ed Chl
aconcen a ions and
es ima ed Chl
aconcen a ions based on
he calib a ion wi h
measu ed Chl aand
abso bance-based wa e
colou (A
420
, le panel,
model 3 in Table 2b) o
in si u OM luo escence
( igh panel, model 4 in
Table 2b). Solid line = 1:1
line; MAE mean absolu e
pe cen age e o (%), ME
modelling e iciency (%)
Fig. 4 Rela ionship be ween wa e colou and OM luo es-
cence (RFU
20
) in he s udy lakes
123
Hyd obiologia (2020) 847:4377–4387 4383
ou da a, he la e was ypical o mos o he s udied
lakes, excep o he eu ophic and mos clea wa e
lakes (colou B10 mg P l
-1
).
Ou esul s ag ee wi h p e ious s udies s a ing ha
wi hou OM calib a ion, he in e p e a ion o in si u
Chl esul s is gene ally misleading. P oc o & Roesle
(2010) collec ed wa e samples om a lake, pond,
s eam and a bog in Maine (USA) wi h a ying OM
le els and measu ed phy oplank on and OM luo es-
cence wi h he mul iple wa eband luo ome e s in he
labo a o y. They obse ed a linea ly inc easing OM
signal along he OM concen a ion in he dilu ion
se ies de e mined luo ome ically and spec opho o-
me ically. Based on he linea cong uence be ween
he OM and algae mul iexci e signals, hey s a ed ha
he quan i y o OM is o cen al impo ance. Goldman
e al. (2013) also showed signi ican o e es ima ion o
Chl concen a ion wi h inc eased OM concen a ions
in an es ua y. Leppa
¨e al. (1995) ound a high
backg ound luo escence in humic lakes o Sou h-
Eas e n Finland, caused by wa e colou , which lead o
ecommenda ions o he Chl luo ome y calib a ion
agains he OM backg ound.
In ou s udy, he imp o emen due o calib a ion
was no ha clea in he con inuously measu ed
luo ome e da a o Lake Konne esi. Howe e , in he
end o July 2013, OM luo escence dec eased and he
OM calib a ed Chl luo escence (model 4) inc eased
leading o a be e cong uence wi h he measu ed Chl
a, which migh indica e ha he con inuous on-line
Chl measu emen s could bene i om he OM cali-
b a ion (Fig. 7). Likely, he signi icance o in si u OM
co ec ion is emphasized, compa ed o labo a o y
analyses, when Chl is measu ed wi h a high equency
in eal ime. In Lake Konne esi, he au oma ed sys em
eco ded a dec ease o OM luo escence owa ds he
end o he summe , which is a known phenomenon in
bo eal lakes and associa ed wi h pho ochemical
Fig. 5 Pe o mance o he
calib a ion equa ions in
humic Lakes Jy a
¨sja
¨ i and
Vanaja esi (see Table 1 o
lake cha ac e is ics). B oken
line ep esen s linea
eg ession be ween he
obse ed Chl
aconcen a ions and
es ima ed Chl
aconcen a ions based on
he calib a ion wi h
measu ed Chl a(model 1,
see Table 2a, b o he
models) o measu ed Chl
aand wa e colou es ima e
(model 2), abso bance-
based wa e colou (A
420
,
model 3) o in si u OM
luo escence (model 4).
Solid line = 1:1 line; MAE
mean absolu e pe cen age
e o (%), ME modelling
e iciency (%)
123
4384 Hyd obiologia (2020) 847:4377–4387