https://doi.org/10.47193/mas.3922026010403

ABSTRACT. The shing footprint, which reects humanity’s demand on marine ecosystems and

is closely linked to sheries sustainability, serves as the main environmental indicator for marine

resources. This study investigated the role of schooling as an indicator of human capital in shaping

the shing footprint in Greece over the period 1990-2022. The empirical analysis employed the

human capital and environmental degradation in shing grounds. The ndings reveal the threshold

RESUMEN. La huella pesquera, que reeja la demanda humana sobre los ecosistemas marinos y

The role of schooling in shaping the shing footprint in Greece: evidence

from an augmented ARDL approach

ORCID Pınar Karahan-Dursun https://orcid.org/0000-0002-8289-6570, Serkan Şengül https://orcid.org/0000-0001-9891-9477,

Şerif Canbay https://orcid.org/0000-0001-6141-7510

Ocean pollution is a complex mix of plastics, heavy

metals, agrochemicals, and industrial wastes that

threatens both ecosystems and human health. Mi-

croplastics, mercury, and persistent organic pollut-

ants not only disrupt the ocean’s ability to sequester

carbon but also enter marine food webs, weakening

(Landrigan et al. 2020).

These ecological threats are particularly notewor-

thy given the heavy reliance of societies on sher-

protein intake for about 40% of the global popu-

lation (FAO 2024). Accordingly, ensuring marine

ity of food supply and fostering economic stability.

Within this global context, Greece provides a

and Ionian Seas, Greece possesses one of the most

cording to the World Bank (2025), total sheries

production in Greece grew by 20.6% over the last

EU member states in 2022 (FAO 2025).

lenges directly linked to human capital, including

mental training, limited technological moderniza-

tion, and disparities in skills across coastal regions.

ical setting to explore how increases in education

and capacity-building may inuence environmental

pressure on marine ecosystems. The combination

scale sheries contexts facing similar challenges.

tems is represented by the shing grounds footprint,

incorporates the conservation and sustainable utili-

development (UN 2025). The shing grounds foot-

pact of human capital on the shing footprint in

2018; Danish et al. 2019; Ahmet et al. 2020a, 2020b;

ing us to capture how the contribution of human

capital to sheries sustainability differs below and

capital-shing footprint nexus was analyzed for the

mental Kuznets Curve (EKC) hypothesis and the

for Greece.

outcomes primarily by transforming knowledge,

Third, the institutional capacity channel stresses

2019; Ahmed et al. 2020a, 2020b; Khan 2020; Ganda 2022; Ayad 2023; Alsaleh et al. 2024; Dai et al. 2024; Teng et al. 2024).

port that schooling reduces CO2 emissions but in-

creases ecological footprint in Pakistan. Similarly,

Hondroyiannis et al. (2022) and Dai et al. (2024)

sociation of Southeast Asian Nations) economies,

indicate that schooling does not exert uniform envi-

ronmental effects and that its inuence depends on

tries, highlighting its productive dimension. Ayad

et al. (2024) and Teng et al. (2024) provide direct

evidence that schooling reduces shing footprint

(Gulf Cooperation Council) and G20 (the Group of

ships in Mediterranean countries, showing that low

stitutional capacity, technological adaptation, and

tensify economic activity and resource use, where-

as beyond a certain threshold it fosters sustainable

practices and reduces environmental degradation

(Çakar et al. 2021; Yıldırım et al. 2022).

Literature review

schooling –widely used as a core proxy for human

capital– is a fundamental determinant of environ-

such as ecological footprint, CO2 emissions, and

load capacity factor. However, empirical ndings

consistently reveal that the environmental effects of

schooling are context-dependent, varying across de-

velopment levels, institutional structures, and meth-

A prominent strand of the literature highlights

vironmental sustainability is not necessarily lin-

resource use, while higher schooling levels foster

technological upgrading, behavioral change, and

et al. (2021), Pata et al. (2023), Yılanci et al. (2023),

and Pata et al. (2024) demonstrate that economic

and unemployment inuence shing footprint dy-

(2023), Alsaleh et al. (2024), Ayad et al. (2024),

sheries and central role in the regional marine

the nonlinear effects of schooling on shing foot-

print or assessed the EKC and EPC hypotheses in

Greece. The present study lls this gap by exam-

shaping shing footprint dynamics in Greece over

LFFt = α0 + β1LHC + β2LHC2 + β3LY + εt (3)

LFFt = α0 + β1LHC + β2LHC2 + β3LY + β4LURB + εt (4)

LFFt = α0 + β1LHC + β2LHC2 + β3LY + β4LY2

Network (GFN), while Y, URB, and UN are from

Each specication evaluates the inuence of

ity, reecting their different structural roles within

shows moderate dispersion over time, suggesting

educational dynamics that typically evolve over

sociations, particularly between human capital, ur-

multicollinearity concerns (Appendix, Table A1).

Moreover, correlations between the shing foot-

tween shing footprint and its socioeconomic de-

terminants, it is necessary to establish the order of

integration of each variable. Unit root testing is a

critical step in the empirical strategy because AR-

a mixture of I(0) and I(1) processes. Conventional

ultimately undermining the validity of subsequent

cointegration inference.

the Ng and Perron (2001) unit root test, which im-

Min 2.397895 9.640489 2.055789 1.947623 15.80165

Std. dev. 0.224332 0.132984 0.113283 0.396319 0.050293

statistically sound pre-testing of stationarity.

et al. (2001) assumes that the dependent variable

conducting an overall F-test on the lagged levels of

all variables and a t-test on the lagged level of the

on the lagged levels of the independent variables,

to (6), as follows:

exist (Sam et al. 2019). The null hypotheses for the

three test statistics for all models (equations 7-12)

are shown below:

i) Foverall test, H0: all δ coefcients on the lagged

level variables = 0.

ii) tDV test, H0: δ1 = 0

iii) FIDV test, H0: all δ coefcients on the lagged

level independent variables = 0.

The application of the AARDL method requires

that none of the variables are integrated of order

greater than one (Sam et al. 2019). Accordingly, the

empirical analysis started to investigate stationary

properties of the series using the Ng-Perron test

(Table 2).

For the Ng-Perron test, the null hypothesis for

the MZa and MZt statistics suggested that the se-

LURB and LUN (Table 2). Thus, results indicated

that all series except for LURB and LUN were not

tests, except for LURB and LUN. In conclusion,

integrated of order zero [I(0)].

conrming that none of the variables in the study

nicance level for Equation (3) (Table 4). Likewise,

evaluates the signicance of the lagged dependent

el for equation (6). Lastly, the estimated FIDV test

statistics, which evaluate the lagged values of the

independent variables, were statistically signicant

at 1% signicance level for equations (2), (4), and

(5), at 5% level for equation (1), and at 10% level

tion relationship among the variables, the long-run

cording to this estimated equation, a 1% increase

of these equations consistently revealed an inverted

EKC hypothesis for shing grounds, indicating an

LHC -7.937 -1.981 0.25 11.512

LHC2 -7.019 -1.871 0.267 12.985

LURB -62.994*** -5.482*** 0.087*** 2.027***

LUN -46.158*** -4.743*** 0.103*** 2.278***

ΔLFF -21.753*** -3.275*** 0.151*** 1.206***

ΔLHC -15.415*** -2.736*** 0.178** 1.738***

ΔLHC2 -15.419*** -2.742*** 0.178** 1.719***

Critical values (rst differences)

1% -13.8 -2.58 0.174 1.78

5% -8.1 -1.98 0.233 3.17

10% -5.7 -1.62 0.275 4.45

[gfn] glObal fOOtPrint netwOrk. 2025. [ac-

huang C, Zhang X, liu k. 2021. Effects of human

kaShem ma, rahman mm. 2020. Environmental

1016/j.scitotenv.2020.141594

wang C, liu X, li k, wei C, Xie m. 2025. In-

wOrld bank. 2025. World development indicators.