E alua ing common indica o s o he e ec s o ma ine 1
p o ec ed a eas on conse a ion and ca ches 2
3
1
In oduc ion33
Va ious o ms o spa ial managemen ha e been used o conse e and manage ma ine ecosys ems
34
ac oss cul u es h oughou human his o y (e.g. Johannes 2002). The 1990s ma ked a pe iod o
35
expanded in e es in bo h he science and use o spa ial managemen ools, speci ically he gene al
36
concep o “ma ine p o ec ed a eas” (MPAs) (Ca e al. 2019; Humph eys and Cla k 2020). MPAs
37
a e a eas o he ocean whe e human ac i i ies a e es ic ed o p ohibi ed o p omo e biodi e si y
38
conse a ion and po en ially achie e o he complemen a y objec i es (G o ud-Col e e al. 2021).
39
These objec i es can include p ese ing cul u al he i age, ebuilding popula ions inside p o ec ed
40
bo de s, p o iding conse a ion bene i s ou side in ished wa e s, and suppo ing su ounding
41
ishe ies (Gaines e al. 2010). In e na ional mo emen s such as he Mon eal–Kunming Global
42
Biodi e si y F amewo k’s call o p o ec ion o 30% o he oceans, and he Ag eemen unde
43
he Uni ed Na ions Con en ion on he Law o he Sea on he Conse a ion and Sus ainable Use
44
o Ma ine Biological Di e si y o A eas Beyond Na ional Ju isdic ion ( he High Seas T ea y o
45
BBNJ) ha e p o ided subs an ial momen um o he expanded use o MPAs in he wo ld’s oceans
46
in pu sui o hese goals.47
O e 9% o he ocean is now co e ed by MPAs (as o Oc obe 2025; Wo ld Da abase o P o ec ed
48
A eas), and he ecen global expansion o MPAs has been ma ched by g owing e o s o empi -
49
ically e alua e hei pe o mance (e.g. Osenbe g e al. 2011; Edga e al. 2014; Di Lo enzo e
50
al. 2020; Lynham and Villaseño -De bez 2024; Hop e al. 2024) . Since many o he bene i s
51
ha MPAs a e expec ed o deli e a e di icul o measu e di ec ly, e alua ions o MPA e ec s
52
o en ely on acking simple indica o s as p oxies o b oade MPA e ec s (Pelle ie 2011; Hop 53
e al. 2024). These indica o s enable assessmen s o be ca ied ou ac oss di e se con ex s, and
54
ha e become a ounda ional ool o a emp ing o unde s and MPA e ec s in bo h scien i ic and
55
policy se ings. Many indica o s compa e a ibu es wi hin MPAs o nea by unp o ec ed a eas
56
(i.e., esponse a ios; Smi h e al. (2024); Les e e al. (2009)), o om unp o ec ed a eas ha a e
57
nea o MPA bo de s compa ed o hose a he away (i.e., g adien s, Halpe n, Les e , and Kellne 58
4
2009). Mo e causally- obus app oaches inco po a e measu emen s bo h be o e and a e MPA
59
implemen a ion [i.e., be o e-a e -con ol-impac o BACI; Medo , Lynham, and Rayno (2022);
60
Lynham and Villaseño -De bez (2024); Ke , K i ze , and Cad in (2019); O ando e al. (2021)]. 61
While hese empi ical indica o s di e in hei complexi y, each is p esumably in ended o de e -
62
mine he e ec o he MPA on a pa icula obse ed me ic: highe biomass in MPAs compa ed
63
o unp o ec ed a eas is in e p e ed as conse a ion gains wi hin MPA bo de s (Smi h e al. 2024;
64
Les e e al. 2009), while g adien s in ishe ies ca ch o e o a e hough o signal e idence o
65
spillo e o adul o la al ish om he MPA o su ounding wa e s (Halpe n, Les e , and Kellne
66
2009; Medo , Lynham, and Rayno 2022; Lynham and Villaseño -De bez 2024; Robe s e al.
67
2001). Howe e , hese indica o esul s a e some imes hen assumed o also imply e idence o an
68
unobse ed e ec , such as changes in o al popula ion size o o al ishe ies ca ch. 69
The assump ion ha speci ic indica o s can s and in o b oade and ha de o obse e ecological
70
o social e ec s has no been o ou knowledge comp ehensi ely es ed, despi e hei widesp ead
71
use ( hough wo ks such as Hop e al. (2024), Ke , K i ze , and Cad in (2019), and Hilbo n e al.
72
(2024) e alua ed speci ic combina ions o indica o s and e ec s). Unde s anding he eliabili y o
73
indica o s, and he a iabili y o he unde lying MPA e ec s hey aim o measu e, is c i ical o
74
in e p e ing exis ing e alua ions, designing e ec i e moni o ing s a egies, and o e idence-based
75
applied ecological decision-making. 76
This pape add esses wo ques ions. We i s analysed whe he MPA e ec s on conse a ion
77
and ca ches a e likely o be a iable enough o jus i y he need o empi ical moni o ing. We
78
hen es ed he abili y o se e al commonly used empi ical indica o s o e lec he ac ual e ec s
79
o MPAs. We ound ha MPAs can ha e highly a iable e ec s on conse a ion and ca ches
80
depending on he speci ic social-ecological dynamics o he sys em in ques ion, and ha while
81
some indica o s eliably ack his a iabili y in some MPA e ec s, many do no . 82
5
Ma e ials and Me hods83
MPA Expe imen O e iew84
We used a simula ion amewo k o model bo h he a iabili y o MPA e ec s and he pe o mance
85
o empi ical MPA indica o s wi hin complex social–ecological sys ems. We simula ed a ishe y
86
sys em wi h a se o bio-economic ai s. We hen an an MPA expe imen on ha sys em by
87
gene a ing wo pai ed simula ions iden ical in e e y way, excep ha one sys em con ained a
88
no- ake MPA and he o he did no . We hen calcula ed he simula ed e ec s o he MPA on a
89
ange o objec i es based on he di e ences be ween he simula ion wi h an MPA ela i e o
90
he pai ed simula ion wi hou an MPA. Las ly, we calcula ed common empi ical indica o s o
91
MPA pe o mance om he simula ion wi h he MPA, and compa ed hese indica o s o he ue
92
simula ed e ec s. We hen epea ed his p ocess mul iple imes wi h di e en andomly gene a ed
93
ishe y s a es. The de ails o his p ocess a e explained below and illus a ed in Figu e 1.SeeSI
94
and Figu e S1 o a case s udy simula ion o hese p ocesses.95
Simula ing Fishe ies96
Fishe ies sys ems we e simula ed using he ma lin model (O ando e al. 2023; O ando 2025).
97
ma lin simula es a use -speci ied numbe o age-s uc u ed ish popula ions ished by a use -
98
speci ied numbe o lee s on a simula ed seascape di ided in o “pa ches” o de ined a ea. Fo his
99
pape , we modeled a 21 by 21 g id o pa ches (441 o al), each 25 km
2
(5 km × 5 km), o a o al
100
seascape a ea o 11,025 km2.101
To explo e MPA e ec s and indica o s ac oss a ange o species, we modeled ou species
102
a che ypes de ined by ixed and a iable pa ame e s, b oadly based on a una (Thunnus albaca es),
103
sha k (Ca cha hinus alci o mis), g oupe (Epinephelus uscogu a us), and ee ish (Se iola
104
quinque adia a). Fixed pa ame e s included g ow h, mo ali y, ecundi y, ec ui men , and mo e-
105
men , wi h alues based on he li e a u e and ou judgemen (Table S1). We ixed hese o a oid
106
6
Figu e 1: Concep ual illus a ion o he p ocess o simula ing e ec s o MPAs, calcula ing
empi ical indica o s, and compa ing indica o s o e ec s. Colo ed polygons in S ep 2 indica e
habi a dis ibu ions o each simula ed species. See Me hods sec ion o de ailed explana ions o
hese s eps.
7
un ealis ic combina ions o , o example, g ow h and mo ali y (P ince e al. 2015). Each i e a ion
107
o he model also used d aws om a lis o a iable pa ame e s (Table S2).108
The model simula es mo emen ac oss he wo-dimensional seascape h ough passi e di usion
109
and ac i e axis along habi a g adien s, ollowing J. T. Tho son e al. (2021) and O ando (2025).
110
Mo emen a es o ec ui s and pos - ec ui s we e se so ha 95% o indi iduals emained wi hin
111
a gi en linea dis ance o hei o igin a e one yea . These dis ances we e g ouped in o low (2.5
112
km), medium (25 km), o high (250 km) dispe sal and assigned o ec ui and pos - ec ui s ages o
113
e lec a ange o mo emen dynamics (e.g. seden a y adul s and highly dispe sed la ae, and ice
114
e sa; Table S1). These alues a e no in ended as ue mo emen a es o he named species.115
Fo each simula ion we augmen ed ixed pa ame e s wi h andom d aws o a iable pa ame e s de-
116
sc ibing less-ce ain aspec s o he sys em (Table S2). Ecological ai s included he uni o mi y o
117
species-le el habi a , he co ela ion in habi a ac oss species, seasonal mo emen shi s, spawning
118
agg ega ions, he iming and s eng h o densi y dependence, and he magni ude and co ela ion
119
s uc u e o ec ui men de ia ion (Table S2). Va iable ai s also included ishe y cha ac e is ics
120
such as ishing p essu e on each species, species- and lee -speci ic economic alue, he p esence
121
o po s ha a ec spa ial choices, and lee selec i i y (Table S2).122
Fish popula ions can be he e ogeneously dis ibu ed in space, and species dis ibu ions can exhibi
123
posi i e, nega i e, o no co ela ions wi h o he ished species (J. T. Tho son and Ba ne 2017;
124
Ka p e al. 2025; B odie e al. 2021; Lopez e al. 2024). Following O ando e al. (2023) and
125
O ando (2025), o each species we andomly se a pa ame e
𝜅
desc ibing habi a he e ogenei y:
126
lowe
𝜅
yields smoo he habi a and highe
𝜅
yields mo e pa chy habi a . We also gene a ed a
127
habi a co ela ion ma ix among species. These
𝜅
alues and he co ela ion ma ix we e used o
128
cons uc a co a iance ma ix o a mul i a ia e no mal dis ibu ion, om which we d ew spa ial
129
habi a ields. These d aws p oduce species dis ibu ions ha a y ac oss simula ions in bo h hei
130
he e ogenei y and c oss-species co ela ion.131
Ou model also accoun s o some o he complex beha io s o ishing lee s. Flee s can adap
132
8
hei spa ial and empo al dis ibu ion in esponse o ishing oppo uni ies and egula ions. The
133
simula ed e ec s o an MPA a e sensi i e o whe he ishe s adjus o new egula ions, o example
134
by concen a ing e o along MPA bo de s, which can s ongly in luence ou comes (O ando
135
2025). 136
We allowed o di e en e o dynamics. Fo any gi en lee in a simula ion, o al ishing e o
137
ac oss pa ches was based on a andom selec ion o ei he Cons an E o o Open Access. Unde
138
Cons an E o , o al e o is ixed o e ime unless a policy such as an MPA is implemen ed. Un-
139
de Open Access, o al e o can expand o con ac om yea o yea based on p io p o i abili y
140
and unde s able condi ions e ol es owa ds a bionomic equilib ium wi h ze o p o i s (Cos ello e
141
al. 2016; Cab al e al. 2019). 142
Gi en o al e o , lee s alloca e e o ac oss pa ches based on ei he p o i s o e enue, andomly
143
chosen o each lee . When dynamics a e p o i -based, cos s can inc ease wi h dis ance om he
144
lee ’s home po . MPA implemen a ion can cause e o o edis ibu e o lea e he ishe y. Unde
145
displacemen , e o ha was inside he MPA is ealloca ed o pa ches ou side in he ime s ep
146
ollowing implemen a ion, acco ding o he lee ’s spa ial ule. Unde e o a i ion, e o ha
147
was inside he MPA exi s he ishe y. Unde Open Access, o al e o can subsequen ly adjus o e
148
ime. Toge he , hese dynamics allow lee s o adjus bo h he o al amoun o e o and i s spa ial
149
dis ibu ion in esponse o oppo uni ies and MPAs. 150
All else being equal, MPAs ha e g ea e e ec s in hea ily ished sys ems han in ligh ly ished
151
ones (Hilbo n e al. 2004). We simula ed ishing p essu e by andomly selec ing a baseline ishing
152
mo ali y a e o each species, held cons an in he absence o an MPA. Because he same ishing
153
mo ali y can ha e di e en e ec s depending on li e his o y and selec i i y, we summa ize ishing
154
p essu e using baseline “deple ion”, measu ed as p e-MPA biomass (B) di ided by un ished
155
biomass (B0), B/B0. Values nea ze o indica e hea ily ished popula ions, and alues nea one
156
indica e ligh ly ished popula ions. The dis ibu ion o baseline B/B0 ac oss simula ions is shown
157
in Figu e S25.158
9
The ou species a che ypes we e used in h ee bio-economic ishe y sys ems o inc easing com-
159
plexi y. The simple scena io is single-species, single- lee , ma ching he mos common way o
160
modeling MPA e ec s on one popula ion and one lee . The medium scena io is mul i-species,
161
single- lee , wi h all ou species in he same seascape, allowing posi i ely o nega i ely co ela ed
162
spa ial dis ibu ions and a iable ulne abili y. The complex scena io is mul i-species, mul i- lee ,
163
wi h wo lee s ha di e in selec i i y and ishing p essu e, allowing dynamics such as a species
164
being a ge ed by one lee and aken as byca ch by ano he .165
In o al, each simula ed ishe y is buil om combina ions o ixed and a iable bio-economic
166
pa ame e s (Table S1; Table S2) and he chosen complexi y le el. We p ojec ed each ishe y o 75
167
yea s o app oxima e p e-MPA equilib ium condi ions.168
Simula ing MPA E ec s169
We an each simula ed ishe y h ough an MPA expe imen whe e all a ibu es o he simula ed
170
ishe y we e held a iden ical alues, excep o he applica ion o an MPA. Ou MPA expe imen s
171
we e c ea ed om a ac o ial combina ion o MPA size ( ep esen ed as he pe cen o he su ace
172
a ea o he seascape co e ed in a ully p o ec ed – i.e., no- ake – MPA) and he placemen s a egy
173
o he MPA (A oid Fishing o Ta ge Fishing). MPA size anged om 5% o 60% o he seascape,
174
in inc emen s o 5%. Unde he Ta ge Fishing s a egy, pa ches a e placed in an MPA in descend-
175
ing o de o o al biomass caugh , unde he A oid Fishing s a egy, pa ches a e placed in an MPA176
in ascending o de o o al biomass caugh .177
We also explo ed how e ec s a ied depending on whe he he MPAs we e indi idual en i ies
178
(con iguous) o di ided ac oss a ne wo k (mosaic) (Gaines e al. 2010; Pons e al. 2022). Howe e ,
179
all esul s p esen ed he e used he mosaic s a egy since we es ed he sensi i i y o ou esul s
180
o using con iguous MPAs and ound no i had no meaning ul impac s, so omi ed hose uns o
181
educe compu a ional o e head.182
A e unning each ishe y o 75 yea s o each equilib ium condi ions, we hen an each ishe y
183
10
h ough each MPA scena io o 20 mo e yea s. 20 yea s was selec ed o gi e enough ime o MPA
184
e ec s o de elop, wi hou necessa ily gua an eeing equilib ium condi ions, e lec ing he eali y
185
o eal-wo ld MPA moni o ing p og ams. Running he models o mo e yea s had no meaning ul
186
e ec on ou esul s. A e 20 yea s, we calcula ed he ollowing MPA e ec s: 187
• To al biomass pe species inside MPA bo de s (Biomass Inside)188
• To al biomass pe species ou side MPA bo de s (Biomass Ou side)189
• To al biomass pe species (To al Biomass)190
• To al ca ch pe species and lee (Ca ch)191
The ue e ec s o he MPAs on each o hese me ics was calcula ed by compa ing he pe cen age
192
di e ence in he ele an alues a e 20 yea s pos bu n-in in he simula ion wi h he MPA ela i e
193
o he pai ed simula ion wi hou he MPA. An MPA e ec alue o 20% means ha he me ic
194
in ques ion was 20% highe in he simula ion wi h he MPA han he same me ic in he pai ed
195
simula ion wi hou he MPA. 196
Fil e ing Simula ions 197
While we ook s eps o es ic he simula ions o ealis ic scena ios (e.g. ixing co e li e his o y
198
ai s), some combina ions o pa ame e s s ill p oduced implausible esul s, which we e emo ed
199
om he inal analysis. We emo ed simula ions in which any species had ex emely low baseline
200
deple ion le els (less han 1% o un ished biomass), as while no impossible such ex eme le els o
201
deple ion a e a e in ma ine ish popula ions and p oduce ex eme esul s when conside ing MPA
202
e ec s on a pe cen age based scale. We emo ed any simula ions in which any species had o al
203
biomass alues less han one, o any ishing lee had ca ches less han one. These simula ions
204
we e emo ed since hey esul ed in imp obably high pe cen age e ec sizes. Pos - il e ing, he
205
combina ion o each simula ed ishe y wi h each MPA design esul ed in a o al o 10,896 unique
206
MPA expe imen s. 207
11
To al Biomass Ca ch
Biomass Inside Biomass Ou side
0-15%
15-30%
30-60%
0-15%
15-30%
30-60%
-100%
0%
100%
≥200%
-100%
0%
100%
≥200%
-100%
0%
100%
≥200%
-100%
0%
100%
≥200%
Seascape in MPA
MPA E ec
B/B0 >50% 25-50% 0-25% Quan ile 100% 80% 50%
Figu e 2: Dis ibu ion o simula ion esul s ac oss MPA e ec s (panels), p opo ion o seascape in
MPA (columns), and le el o baseline deple ion in he absence o MPAs (colo s, biomass B
di ided by un ished biomasss B0). Y-axis shows he “MPA e ec ”, he pe cen change in he
ou come in ques ion caused by he MPA. Biomass Inside e e s o o al indi idual species biomass
inside MPA bo de s. Biomass Ou side e e s o o al indi idual species biomass ou side MPA
bo de s. To al Biomass e e s o o al indi idual species biomass bo h inside and ou side MPA
bo de s. Ca ch e e s o o al ca ch pe species and lee ou side he MPA.
18
Pe o mance o Empi ical Indica o s o MPA E ec s 315
The abili y o empi ical indica o s o ack MPA e ec s a ied widely (Figu e 3A). All h ee
316
inside-ou side indica o s had posi i e Spea man’s
𝜌
wi h Biomass Inside and To al Biomass,wi h
317
Biomass Densi y BACI ha ing he highes
𝜌
alues o bo h (
𝜌
= 0.64 and
𝜌
= 0.45 espec i ely),
318
ollowed by Biomass Densi y Response Ra io (
𝜌
= 0.56and
𝜌
= 0.38 espec i ely). All h ee
319
inside-ou side indica o s had nega i e Spea man’s
𝜌
alues wi h Biomass Ou side, meaning ha
320
simula ions wi h ela i ely highe biomass densi y esponse a ios we e associa ed wi h simula ions
321
wi h ela i ely lowe Biomass Ou side e ec s. 322
G adien indica o s exploi ing nea - a pa e ns a ound MPAs had much lowe Spea man’s
𝜌323
alues ac oss all e ec s (all
|𝜌| ≤ 0.2
), explaining almos none o he ank-le el a ia ion in any
324
o he e alua ed MPA e ec s. Among he nea - a indica o s, E o G adien s had he highes
325
co ela ion alues, (𝜌= 0.2 o Biomass Inside and 𝜌= 0.16 o To al Biomass) (Figu e 3). 326
None o he e alua ed indica o s we e meaning ully co ela ed wi h he e ec s o MPAs on o al
327
ca ches, hough all es ima ed co ela ions we e nega i e. Biomass Densi y BACI had he clea es
328
nega i e co ela ion wi h MPA e ec s on ca ch, wi h 𝜌 = −0.14 (Figu e 3). 329
The
𝜌
alues shown in Figu e 3indica e co ela ion be ween he indica o alue and he ou come
330
alue. These co ela ions ell us how eliably an indica o can be used o ank di e en MPAs in
331
e ms o pe o mance ela ed o a speci ic MPA e ec . We also examined how well aw indica o
332
alues ep esen ed aw MPA e ec s by calcula ing measu es o e o ( oo mean squa ed e o ) and
333
bias (mean absolu e e o ), whe e posi i e bias alues indica e ha , on a e age, indica o alues
334
we e highe han ue alues, and ice e sa. These me ics a e mo e use ul o de e mining how
335
well he alue o a gi en indica o a one MPA ansla es in o he alue o a gi en e ec s a he
336
same MPA. The a e age RMSE ac oss all indica o s was 93 pe cen age poin s, wi h an a e age
337
bias ac oss all indica o s o 13 pe cen age poin s. Mos indica o s we e posi i ely biased ela i e o
338
he ue e ec size, hough some indica o s we e nega i ely biased o Biomass Inside and To al
339
Biomass.340
19
All o he esul s shown in Figu e 3a e based compa isons a he le el o indi idual species and
341
whe e applicable lee s, o example compa ing he Biomass Densi y BACI alue o ee ish
342
o he ue e ec o he MPA on ee ish ca ches by an indi idual lee . We also an ou esul s
343
on o al alues, compa ing o example he Biomass Densi y BACI alue agg ega ed ac oss all
344
simula ed species o he ue e ec o he MPA on all ca ches ac oss all lee s (Figu e S31). Doing
345
so had no subs an ial impac on he co e esul s p esen ed in he body o he pape .346
20
0.16
0.38
0.39
-0.02
0.45
0.06
0.2
0.56
0.54
-0.02
0.64
0.05
-0.03
-0.49
-0.4
-0.05
-0.47
0
-0.08
-0.07
-0.11
-0.02
-0.14
-0.09
Biomass Densi y G adien
Biomass Densi y G adien BACI
E o G adien
Mean Leng h Response Ra io
Biomass Densi y Response Ra io
Biomass Densi y BACI
A) Spea man’s ρ
-1.0 -0.5 0.0 0.5 1.0
0.03
0.15
0.15
0
0.21
0
0.04
0.31
0.29
0
0.4
0
0
0.24
0.16
0
0.22
0
0.01
0.01
0.01
0
0.02
0.01
B) Spea man’s ρ²
0.00 0.25 0.50 0.75 1.00
109
90
55
69
86
65
118
77
88
100
69
96
116
128
42
58
125
53
143
144
77
87
143
84
Biomass Densi y G adien
Biomass Densi y G adien BACI
E o G adien
Mean Leng h Response Ra io
Biomass Densi y Response Ra io
Biomass Densi y BACI
Biomass Inside
Biomass Ou side
To al Biomass
Ca ch
C) RMSE
050100
20
28
-26
-27
27
-28
-2
7
-48
-48
5
-49
48
57
2
2
56
1
68
76
23
22
75
21
Biomass Inside
Biomass Ou side
To al Biomass
Ca ch
D) Bias
-25 0 25 50 75
Figu e 3: Spea man co ela ions 𝜌(A) and 𝜌2(B) be ween indica o s (y-axis) and MPA e ec s
(x-axis) a he le el o indi idual species and lee s ac oss all simula ions. Roo mean squa ed e o
(RMSE, C) and mean e o (Bias, D) be ween indica o s (y-axis) and e ec s (x-axis) a he le el o
indi idual species and lee s ac oss all simula ions. Plain ex y-axis labels indica es inside-ou side
indica o s, i alic y-axis labels indica e a nea - a indica o . Tex shows he alue o he me ic in
ques ion, also e lec ed in he backg ound colo o each cell acco ding o he associa ed colo ba
legend. 21
Cali o nia Response Ra io Case S udy347
Sub-se ing ou esul s o only Biomass Densi y Response Ra io alues ha ma ch he dis ibu ion
348
o esponse a ios o a ge ed species epo ed in Smi h e al. (2024) allows us o isualize he
349
concep s shown in Figu e 3. Simula ed esponse a ios we e posi i ely co ela ed wi h Biomass
350
Inside and To al Biomass (
𝜌 = 0.48
,
𝜌 = 0.37
), and nega i ely co ela ed wi h Biomass Ou side
351
and Ca ch (𝜌 = −0.28,𝜌 = −0.1) (Figu e 4).352
While Biomass Densi y Response Ra ios had a high
𝜌
alue wi h Biomass Inside, he e was sub-
353
s an ial a ia ion in his ela ionship, wi h low Biomass Densi y Response Ra ios in ac being
354
associa ed wi h la ge changes in Biomass Inside, and ice e sa. This was ampli ied o ela ion-
355
ships wi h lowe co ela ions, such as Biomass Densi y Response Ra ios and Ca ch, in which
356
Biomass Densi y Response Ra ios o 0.5 could be p oduced by simula ions wi h up o 100% in-
357
c eases in ca ch o 100% losses in ca ch, o commonly wi h no meaning ul change in ca ch a all
358
(Figu e 4).359
22
ρ=0.48
ρ² = 0.23
ρ=0.37
ρ² = 0.14
ρ=-0.28
ρ² = 0.08
ρ=-0.1
ρ² = 0.01
To al Biomass Ca ch
Biomass Inside Biomass Ou side
0.0 0.5 1.0 0.0 0.5 1.0
-50%
0%
50%
100%
150%
200%
0%
100%
200%
300%
0%
100%
200%
300%
0%
50%
100%
150%
200%
Simula ed Biomass Densi y Response Ra io
MPA E ec
# o Simula ions
25 50 75 100 125
Figu e 4: Simula ed Biomass Densi y Response Ra ios (x-axis) plo ed agains simula ed MPA
e ec s (y-axis), wi h dis ibu ion o esponse a ios ma ched o empi ical esponse a ios o
a ge ed in ish epo ed by Smi h e al. (2024). Cons ained o MPA sizes bes een 10-40% o
seascape and he scena ios wi h mul iple co-exis ing species. Tex shows Spea man
𝜌
co ela ions
and 𝜌2 alues.
23
Discussion360
Ou esul s show ha MPA e ec s on conse a ion and ca ches can be highly a iable in bo h
361
magni ude and di ec ion, unde sco ing he need o moni o ing and e alua ing MPAs a he han
362
assuming hei e ec s a p io i.363
Va iabili y o MPA E ec s364
MPAs almos always inc eased Biomass Inside hei bo de s (96% o simula ions), hough he
365
magni ude o hese gains a ied depending on a iables such as MPA size, placemen , and ishing
366
p essu e. Howe e , we also ound ha Biomass Ou side he MPA declined in 52% o ou simula-
367
ions, indica ing ha spillo e om inside MPAs was o en insu icien o o e come he e ec s
368
o concen a ed ishing e o ou side. O e all, inc eases in Biomass Inside MPAs we e ypically
369
la ge enough ha To al Biomass (inside and ou side) s ill inc eased in 88% o simula ions, despi e370
po en ial educ ions in Biomass Ou side.371
Ou esul ha MPAs can educe o al biomass is pe haps unin ui i e. Conside a scena io whe e
372
an MPA is placed on he co e habi a o one species, bu in doing so displaces ishing e o on o
373
he co e habi a o ano he species. This displacemen can esul in a ne inc ease in he o al
374
ishing mo ali y on his second species, and a subsequen decline in o al biomass. While hese
375
nega i e e ec s on o al biomass we e ela i ely a e (12% o simula ions), ou indings highligh
376
he po en ial o unin ended consequences o MPAs, pa icula ly when ished species and ishing
377
lee s ha e he e ogeneous dis ibu ions and beha io s Pons e al. (2022).378
While MPAs caused a ne inc ease in To al Biomass in 88% o simula ions, and a ne inc ease in
379
Biomass Ou side he MPA in 48% o simula ions, MPAs educed Ca ch in 73% o simula ions.
380
This shows ha in he majo i y o ou simula ions, MPAs caused a ade-o o be e conse a ion
381
bu lowe ca ches. This ade-o was no due o a lack o spillo e e ec s in ou simula ions, bu
382
a he ha in hose simula ions he amoun o spillo e om he MPA was no su icien o make
383
up o he loss in ishing g ounds caused by he p o ec ed a ea.384
24
Howe e , ou esul s also con i m ha MPAs can p oduce win-win ou comes whe e biomass and
385
ca ch bo h inc ease (21% o simula ions) mo e commonly when he ish popula ion in ques ion
386
would ha e been highly o e ished in he absence o an MPA ( 35% o simula ions whe e B/B0
387
≤
0.25), and he MPA size was no oo la ge ela i e o he mo emen dynamics o he ished
388
species (o oo small o ha e a meaning ul e ec ) (Figu e S29). Unde hese win-win condi ions,
389
he spillo e bene i s o he MPA ebuilding an o e ished popula ion we e su icien o o e come
390
o loss in ishing g ounds. 391
All else being equal MPAs had g ea e posi i e conse a ion e ec s in simula ions wi h la ge
392
MPAs and mo e o e ished (lowe B/B0) popula ions, and mo e nega i e ca ch e ec s in la ge
393
MPAs and less hea ily ished (highe B/B0) popula ions (Figu e 2). Howe e , hese a iables only
394
explained pa o he dis ibu ion o MPA e ec s, wi h MPAs o he same size p o ec ing equally
395
ished popula ions p oducing as ly di e en e ec s on conse a ion and ca ches depending on
396
o he s a e a iables ( Table S1; Table S2). Collec i ely, ou esul s demons a e ha he e ec s
397
o MPAs on conse a ion and ca ches a e a complex unc ion o many in e ac ing bio-economic
398
a iables including MPA size, ish mo emen ac oss di e en li e s ages, ishing p essu e, and he
399
esponses o ishing lee s o spa ial closu es. 400
Pe o mance o MPA Indica o s 401
Ou indings complemen pas s udies no ing ha common MPA indica o s can pe o m well bu
402
can also be misleading when aced he complex dynamics o ma ine social-ecological sys ems
403
(Hop e al. 2024; Ke , K i ze , and Cad in 2019; Hilbo n e al. 2024; Claude and Guide i 2010;
404
Fe a o, Sanchi ico, and Smi h 2018; Ch is ie e al. 2019). We ex end his li e a u e by p o iding a
405
comp ehensi e e alua ion o he pe o mance o mul iple indica o s as p oxies o a ange o MPA
406
e ec s unde a wide se o bio-economic s a es. 407
The mos eliable combina ion o indica o s and e ec s we e inside-ou side indica o s such as
408
Biomass Densi y BACI and Response Ra ios as p oxies o biomass e ec s o MPAs, wi h he
409
25
s onges co ela ions wi h Biomass Inside. Nume ous s udies ha e poin ed ou ha inside-ou side
410
indica o s a e likely o p o ide biased es ima es o he conse a ion e ec s o MPAs gi en uncon-
411
olled o iola ions o he co e assump ions o hese indica o s such as di e ences in baseline
412
habi a inside e sus ou side, ec ui men shocks, biological spillo e , and ishing lee esponses,
413
a e common in ma ine social-ecological sys ems (Fe a o, Sanchi ico, and Smi h 2018; La sen,
414
Meng, and Kendall 2019; Hilbo n e al. 2022; O ando e al. 2021; Hop e al. 2024). Ou esul s
415
show ha e en when aced wi h hese biasing ac o s, inside-ou side indica o s s ill could be ea-
416
sonably eliable p oxies o he ela i e conse a ion pe o mance o di e en MPAs, as e idenced
417
by he ela i ely high Spea man’s
𝜌
alues (Figu e 3). Howe e , hese inside-ou side indica o s
418
we e s ill gene ally imp ecise (high RMSE) and biased, indica ing ha one canno eliably in-
419
e p e he speci ic alues o an inside-ou side indica o as he nume ical e ec on an MPA on
420
conse a ion.421
None o indica o s e alua ed in his s udy eliably acked he e ec s o MPAs on ishe y ca ches
422
(Figu e 3). This highligh s an u gen need o esea ch on e ec i e indica o s o he e ec s o
423
MPAs on ishe y ca ches, as ou simula ion esul s indica e ha nega i e e ec s can be common.
424
Collec ion o ac ual da a on he economic pe o mance o ishe ies ( a he han indi ec indica o s
425
such as spillo e g adien s) could help p o ide be e es ima es o ca ch e ec s, hough inding sui -
426
able “con ol” ishe ies o isola e he causal e ec o MPAs on o al ca ch om o he exogenous
427
shocks (e.g. changes in uel p ices o ma ke demand) will be di icul .428
Nea - a indica o s exploi ing g adien s ou side o MPAs me hods ha e a long his o y in MPA
429
science (Halpe n, Les e , and Kellne 2009; Lynham and Villaseño -De bez 2024; Medo , Lyn-
430
ham, and Rayno 2022; Robe s e al. 2001; Di Lo enzo, Claude , and Guide i 2016; Me h a a
431
2020). Howe e , Nea – a indica o s ailed o ack any simula ed MPA e ec , despi e ou model
432
p oducing s ong g adien s in simula ed biomass and e o (Figu e S28). Bo h simula ion model-
433
ing and empi ical e idence ha e con i med ha some le el o spillo e will almos always occu
434
a one o mo e li e s ages gi en he mo emen dynamics o ma ine o ganisms (Cudney-Bueno
435
26
e al. 2009; Di Lo enzo e al. 2020; F anceschini, Lynham, and Madin 2024; Gaines e al. 2010;
436
Hilbo n e al. 2004; Hilbo n e al. 2024). The mo e ele an ques ion is wha does obse ing a
437
spillo e g adien ell us? Ou esul s show ha unde ou simula ed condi ions, he p esence o
438
ue spillo e g adien s in o example biomass densi y o ishing e o caused by an MPA, is no
439
gene ally necessa y no su icien e idence o he e ec s o MPAs on Biomass Inside,Biomass
440
Ou side,To al Biomass, o Ca ch.441
Ensembles o nea - a indica o s migh pe o m be e han any indi idual one (Ande son e al.
442
2017). The e a e also o he po en ial objec i es o MPAs ha may pe haps be eliably acked
443
by g adien -based me hods. Fo example, inc eases in he size and abundance o ish nea MPA
444
bo de s may be o high alue o ec ea ional ishe s (F anceschini, Lynham, and Madin 2024),
445
e en i hose inc eases on he bo de do no necessa ily equa e o commensu a e bene i s a he
446
scale o he o al popula ion o ishe y. 447
The poo pe o mance o nea - a indica o s as p oxies o b oade MPA e ec s in ou model
448
p esen s a majo challenge o e alua ing MPAs whe e only ishe y-dependen da a om ou side
449
MPAs a e a ailable. E alua ions o la ge high-seas MPAs c ea ed h ough e o s such as BBNJ
450
a e likely o depend solely on ishe ies da a om ou side he MPA (Medo , Lynham, and Rayno
451
2022; Lynham and Villaseño -De bez 2024), as he cos s o conduc ishe y-independen su eys in
452
la ge and emo e pelagic MPAs a e likely o be p ohibi i e. Hamp on e al. (2023) p o ides a good
453
example o in eg a ion o ishe y-dependen da a wi h p ocess-based models o es ima e he e ec s
454
o high-seas MPAs on una ishe ies. No el sou ces o da a such as he acous ic signals collec ed
455
by Fish Agg ega ing De ices used in una ishe ies could also be explo ed (Mo eno e al. 2019). 456
I is impo an o no e ha e en among he indica o s wi h ela i ely highe co ela ions (inside-
457
ou side), he sign o he co ela ion coe icien was no consis en . To illus a e, Biomass Densi y
458
Response Ra ios obse ed in a ge ed ish species a ound Cali o nia MPAs we e gene ally pos-
459
i i e, o en subs an ially so (Smi h e al. 2024). Ou simula ion esul s show ha while highe
460
esponse a ios we e gene ally p oduced by be e Biomass Inside ou comes, highe Biomass
461
27
doi:10.1093/icesjms/ sz014.615
La sen, Ashley E., Kyle Meng, and B uce E. Kendall. 2019. “Causal Analysis in Con ol–impac
616
Ecological S udies wi h Obse a ional Da a.” Me hods in Ecology and E olu ion 10 (7):
617
924–34. doi:h ps://doi.o g/10.1111/2041-210X.13190.618
Les e , S. E., B. S. Halpe n, K. G o ud-Col e , J. Lubchenco, B. I. Ru enbe g, S. D. Gaines, S.
619
Ai amé, and R. R. Wa ne . 2009. “Biological E ec s Wi hin No-Take Ma ine Rese es: A
620
Global Syn hesis.” Ma ine Ecology P og ess Se ies 384: 3346.621
Lopez, Jon, Shane G i i hs, B yan P. Wallace, Ve ónica Cáce es, Luz Helena Rod íguez,
622
Ma ino Ab ego, Joanna Al a o-Shigue o, e al. 2024. “Vulne abili y o he C i ically
623
Endange ed Lea he back Tu le o Fishe ies Byca ch in he Eas e n Paci ic Ocean. I. A
624
Machine-Lea ning Species Dis ibu ion Model.” Endange ed Species Resea ch 53 (Ma ch):
625
271–93. doi:10.3354/es 01288.626
Lo enzen, Kai. 2022. “Size- and Age-Dependen Na u al Mo ali y in Fish Popula ions: Biology,
627
Models, Implica ions, and a Gene alized Leng h-In e se Mo ali y Pa adigm.” Fishe ies
628
Resea ch 255 (No embe ): 106454. doi:10.1016/j. ish es.2022.106454.629
Lynham, John, and Juan Ca los Villaseño -De bez. 2024. “E idence o Spillo e Bene i s om
630
La ge-Scale Ma ine P o ec ed A eas o Pu se Seine Fishe ies.” Science 386 (6727): 1276–81.
631
doi:10.1126/science.adn1146.632
Maunde , Ma k N., Richa d B. De iso, Ku M. Schae e , Daniel W. Fulle , Alexand e M. Ai es-da-
633
Sil a, Ca olina V. Min e-Ve a, and S e en E. Campana. 2018. “The G ow h Cessa ion Model:
634
A G ow h Model o Species Showing a Nea Cessa ion in G ow h wi h Applica ion o Bigeye
635
Tuna (Thunnus Obesus).” Ma ine Biology 165 (4): 76. doi:10.1007/s00227-018-3336-9.636
McEl ea h, Richa d. 2020. S a is ical Re hinking: A Bayesian Cou se wi h Examples in and S an.
637
2nd ed. CRC Tex s in S a is ical Science. Boca Ra on: Taylo ; F ancis, CRC P ess.638
Medo , Sa ah, John Lynham, and Jenni e Rayno . 2022. “Spillo e Bene i s om he Wo ld’s
639
La ges Fully P o ec ed MPA.” Science 378 (6617): 313–16. doi:10.1126/science.abn0098.640
Me h a a, Elizabe h T. 2020. “Moni o ing Fishe ies Resou ces a O sho e Wind
641
34
Fa ms: BACI Vs. BAG Designs.” ICES Jou nal o Ma ine Science 77 (3): 890–900.
642
doi:10.1093/icesjms/ saa026.643
Mo eno, Gala, Guille mo Boy a, Igo Sanc is obal, Da id I ano, and Vic o Res epo. 2019. “To-
644
wa ds Acous ic Disc imina ion o T opical Tuna Associa ed wi h Fish Agg ega ing De ices.”
645
PLoS ONE 14 (6): e0216353. doi:10.1371/jou nal.pone.0216353.646
Nickols, Ke y J., J. Wilson Whi e, Dan Malone, Ma k H. Ca , Richa d M. S a , Ma issa L.
647
Baske , Alan Has ings, and Louis W. Bo s o d. 2019. “Se ing Ecological Expec a ions o
648
Adap i e Managemen o Ma ine P o ec ed A eas.” Jou nal o Applied Ecology 56 (10):
649
2376–85. doi:h ps://doi.o g/10.1111/1365-2664.13463.650
Osenbe g, C aig W., Je ey S. Shima, Sonja L. Mille , and Ad ian C. S ie . 2011. “ECOLOGY –
651
Assessing E ec s o Ma ine P o ec ed A eas: Con ounding in Space and Possible Solu ions.”
652
In, edi ed by Joachim Claude , 143–67. Ecology, Biodi e si y and Conse a ion. Camb idge:
653
Camb idge Uni e si y P ess. doi:10.1017/CBO9781139049382.010.654
O ando, Daniel. 2025. “P edic ed E ec s o Ma ine P o ec ed A eas on Conse a ion and Ca ches
655
A e Sensi i e o Model S uc u e.” Theo e ical Ecology 18 (1): 7. doi:10.1007/s12080-024-
656
00602-7.657
O ando, Daniel, Da cy B adley, Echelle Bu ns, Lennon Thomas, and James Tho son. 2023. “Sim-
658
ula ing Bene i s, Cos s and T ade-O s o Spa ial Managemen in Ma ine Social-Ecological
659
Sys ems.” Fish and Fishe ies 25 (2). doi:10.1111/ a .12804.660
O ando, Daniel, Jenni e E. Caselle, Ch is ophe Cos ello, Oli ie Deschenes, S e en D. Gaines,
661
Ray Hilbo n, and Owen Liu. 2021. “Assessing he Popula ion-Le el Conse a ion E ec s o
662
Ma ine P o ec ed A eas.” Conse a ion Biology 35 (6): 1861–70. doi:10.1111/cobi.13782.663
O ando, Daniel, Dawn Doughe y, and Jono R. Wilson. 2016. “Ma ke and Design Solu ions o
664
he Sho -Te m Economic Impac s o Ma ine Rese es.” Fish and Fishe ies 17 (4): 939–54.
665
doi:10.1111/ a .12153.666
Pelle ie , Dominique. 2011. “INDICATORS – Cons uc ing and Valida ing Indica o s o he
667
E ec i eness o Ma ine P o ec ed A eas.” In, edi ed by Joachim Claude , 247–90. Camb idge:
668
35
Camb idge Uni e si y P ess. doi:10.1017/CBO9781139049382.014.669
Pons, Mai e, Jo dan T. Wa son, Daniel O ando, Sand a And aka, S ephanie B odie, And és
670
Domingo, Ma k Fi che , e al. 2022. “T ade-O s Be ween Byca ch and Ta ge Ca ches in
671
S a ic Ve sus Dynamic Fishe y Closu es.” P oceedings o he Na ional Academy o Sciences
672
119 (4). doi:10.1073/pnas.2114508119.673
P ince, Je emy, Ad ian Ho dyk, Sa ah R. Valencia, Neil Lone agan, and Kei h Sainsbu y. 2015.
674
“Re isi ing he Concep o Be e on–hol Li e-His o y In a ian s wi h he Aim o In o m-
675
ing Da a-Poo Fishe ies Assessmen .” ICES Jou nal o Ma ine Science 72 (1): 194–203.
676
doi:10.1093/icesjms/ su011.677
Pun , And é E. 2019. “Spa ial S ock Assessmen Me hods: A Viewpoin on Cu en Issues and
678
Assump ions.” Fishe ies Resea ch 213 (May): 132–43. doi:10.1016/j. ish es.2019.01.014.679
R Co e Team. 2024. R: A Language and En i onmen o S a is ical Compu ing. Vienna, Aus ia:
680
R Founda ion o S a is ical Compu ing. h p://www.R-p ojec .o g/.681
Robe s, Callum M., James A. Bohnsack, Fiona Gell, Julie P. Hawkins, and Rena a Good idge.
682
2001. “E ec s o Ma ine Rese es on Adjacen Fishe ies.” Science 294 (5548): 1920–23.
683
doi:10.1126/science.294.5548.1920.684
Smi h, Joshua G., Co i Lopazanski, Ch is ophe M. F ee, Julien B un, Cla issa Ande son, Ma k
685
H. Ca , Joachim Claude , e al. 2024. “Conse a ion Bene i s o a La ge Ma ine P o ec ed
686
A ea Ne wo k Tha Spans Mul iple Ecosys ems.” Conse a ion Biology n/a (n/a): e14435.
687
doi:10.1111/cobi.14435.688
Tho son, James T. 2019. “Guidance o Decisions Using he Vec o Au o eg essi e Spa io-
689
Tempo al (VAST) Package in S ock, Ecosys em, Habi a and Clima e Assessmen s.” Fishe ies
690
Resea ch 210 (Feb ua y): 143–61. doi:10.1016/j. ish es.2018.10.013.691
Tho son, James T., S e en J. Ba beaux, Daniel R. Goe hel, Kelly A. Kea ney, Edwa d A. Laman,
692
Julie K. Nielsen, Ma hew R. Siskey, Ke in Siwicke, and G an G. Thompson. 2021. “Es ima -
693
ing Fine-Scale Mo emen Ra es and Habi a P e e ences Using Mul iple Da a Sou ces.” Fish
694
and Fishe ies 22 (6): 1359–76. doi:10.1111/ a .12592.695
36
Tho son, James T., and Lewis A. K. Ba ne . 2017. “Compa ing Es ima es o Abundance T ends
696
and Dis ibu ion Shi s Using Single- and Mul ispecies Models o Fishes and Biogenic Habi-
697
a .” ICES Jou nal o Ma ine Science 74 (5): 1311–21. doi:10.1093/icesjms/ sw193.698
Tho son, James T., and Kaspe K is ensen. 2024. Spa io-Tempo al Models o Ecologis s. Fi s
699
edi ion. Chapman & Hall/CRC Applied En i onmen al S a is ics. Boca Ra on, FL: CRC P ess,
700
Taylo & F ancis G oup. 701
Tho son, James, and Kaspe K is ensen. 2024. Spa io-Tempo al Models o Ecologis s. New Yo k:
702
Chapman; Hall/CRC. doi:10.1201/9781003410294.703
T edennick, And ew T., Giles Hooke , S ephen P. Ellne , and Pe e B. Adle . 2021. “A P ac ical
704
Guide o Selec ing Models o Explo a ion, In e ence, and P edic ion in Ecology.” Ecology 102
705
(6): e03336. doi:10.1002/ecy.3336.706
Whi e, J. Wilson, Louis W. Bo s o d, Ma issa L. Baske , Lewis AK Ba ne , R. Je ey Ba , and
707
Alan Has ings. 2011. “Linking Models wi h Moni o ing Da a o Assessing Pe o mance
708
o No-Take Ma ine Rese es.” F on ie s in Ecology and he En i onmen 9 (7): 390–99.
709
doi:10.1890/100138.710
Whi e, J. Wilson, And ew Rassweile , Jameal F. Samhou i, Ad ian C. S ie , and C ow Whi e.
711
2014. “Ecologis s Should No Use S a is ical Signi icance Tes s o In e p e Simula ion Model
712
Resul s.” Oikos 123 (4): 385–88. doi:h ps://doi.o g/10.1111/j.1600-0706.2013.01073.x.713
Zuu , Alain F., ed. 2009. Mixed e ec s models and ex ensions in ecology wi h R. S a is ics o
714
biology and heal h. New Yo k, NY: Sp inge . 715
37
