21508925, 2019, 6, Downloaded from https://esajournals.onlinelibrary.wiley.com/doi/10.1002/ecs2.2757 by CAPES, Wiley Online Library on [28/02/2023].

See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Benthification, biotic homogenization behind the trophic
downgrading in altered ecosystems
~
LUIS ARTUR VALOES BEZERRA ,1,5, VANESSA MARIA RIBEIRO,2 MATHEUS OLIVEIRA FREITAS,2,3 LES KAUFMAN,4
ANDRE ANDRIAN PADIAL,1 AND JEAN RICARDO SIMOES
~ VITULE2,3
1
Laborat
orio de An
alise e S
ıntese em Biodiversidade (LASB), Programa de P
os-Graduacß~ao em Ecologia e Conservacß~ao, Universidade
Federal do Paran
a, Curitiba, Brazil
2
Laborat
orio de Ecologia e Conservacß~ao (LEC), Universidade Federal do Paran
a, Curitiba, Brazil
3
Programa de P
os-Graduacß~ao em Engenharia Ambiental (PPGEA), Universidade Federal do Paran
a, Curitiba, Brazil
4
Boston University Marine Program, 5 Cummington Mall, Boston, Massachusetts 02215 USA
5 
e Budejovice, Czech Republic
Biology Centre of the Czech Academy of Sciences, Institute of Hydrobiology, Cesk

Citation: Bezerra, L. A. V., V. M. Ribeiro, M. O. Freitas, L. Kaufman, A. A. Padial, and J. R. S. Vitule. 2019. Benthification,
biotic homogenization behind the trophic downgrading in altered ecosystems. Ecosphere 10(6):e02757. 10.1002/ecs2.2757

Abstract. Several dozen fish species have been introduced into Neotropical waters, causing significant biotic
changes that include deterministic predation interactions with ecosystem effects. In general, reservoirs are pre-
ferred over lakes as places for stocking policies, due to their artificial aspect, consequently promoting fish intro-
ductions. In a meta-analytic approach, we compared the biomass of plankton-feeding (and top predators) with
bottom-feeding fish species between reservoirs and lakes, considering the influence of invaders and trophic
levels. Among the 26 ecosystems (12 reservoirs and 14 lakes), there is a dominance of non-native bottom-feed-
ing species in artificial environments. We revealed a mechanism behind this dominance in a study case, a ten-
dency for trophic downgrading and biotic homogenization based on interactions between an expatriate
invasive centrarchid (sunfish family) predator from North America (the largemouth bass Micropterus salmoides)
and two omnivorous cichlids from Africa in novel environmental contexts in Brazil (the benthification process).
Both juvenile and adult largemouth bass are keystone predators in aquatic food webs in both their native and
introduced ranges. The omnivore–detritivore tilapiines Oreochromis spp. and the phytophagous–omnivorous
Coptodon spp. are species that exhibits strong generalist tendencies. Such species feed on the omnivore and
detritivore compartments, enhancing detritus cycling among a large variety of d13C sources. Since their con-
sumption is disproportionate, facilitation of other species occurs with multiplicative effects in the environment.
Interactions between invasive species, that is, when an invasive predator (bird, fish, or invertebrate) eats an
invasive prey (mainly fish and invertebrates), can serve to highlight biotic homogenization on fresh waters.

Key words: downgrading food webs; Ecopath with Ecosim; energy flow; food web modeling; invasional meltdown;
predation interactions; stable isotope; trophic homogenization.

Received 12 September 2018; revised 1 April 2019; accepted 8 April 2019. Corresponding Editor: Derek Tittensor.
Copyright: © 2019 The Authors. This is an open access article under the terms of the Creative Commons Attribution
License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.
E-mail: larturr@yahoo.com.br

INTRODUCTION likelihood of biotic homogenization in the

Anthropocene (McKinney and Lockwood 1999,
Large-scale environmental alterations such as Magurran et al. 2015). More than half of the
climate change (Blois et al. 2013), impoundments dammed basins in the world host at least one
in aquatic ecosystems (Rahel 2007, Johnson et al. non-native fish species (those introduced in arti-
2008), and introduction of non-natives (Leprieur ficial or natural environments outside their
et al. 2008, Vitule et al. 2015) increased the native range), while non-natives represent more

❖ www.esajournals.org 1 June 2019 ❖ Volume 10(6) ❖ Article e02757

 21508925, 2019, 6, Downloaded from https://esajournals.onlinelibrary.wiley.com/doi/10.1002/ecs2.2757 by CAPES, Wiley Online Library on [28/02/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
BEZERRA ET AL.

than a quarter of the fish species richness when The secondary invasion is a kind of facilitation
six hotspots of invasion are considered (Leprieur with mutual benefits for both or at least for one
et al. 2008). Not surprisingly, the highest propor- of the interacting species that encompasses a
tion of native and endemic fish species listed on broad range of biotic and abiotic phenomena
the IUCN Red List occur in dammed riverine (Pearson et al. 2018), which include predation of
basins (Leprieur et al. 2008). invader species by another invaders, if there is
Expansionist policies from the early 1970s until an ecosystem effect (O’Loughlin and Green
the end of the 20th century anticipated the con- 2017). If prey’s fitness increases with a decrease
struction of thousands of reservoirs in develop- in the intraguild competition, the predation
ing economies, with a focus on hydroelectric interaction creates benefits to both prey and
power and water storage. Such policies modified predator (Bezerra et al. 2018). Multiple interac-
most of the riverine basins in the world (Agos- tions among invaders at different trophic levels
tinho et al. 2016, Winemiller et al. 2016), and a could precipitate or accelerate biotic homoge-
demand for new inland fisheries has resulted in nization through cascading effects (Olden et al.
programs for the introduction of lentic pre- 2004, Bwanika et al. 2006, Rahel 2007, Johnson
adapted native and non-native species to reser- et al. 2008, Preston et al. 2012), which can cause
voirs (Rahel 2007). On average, 85% of the a complete and irreversible shift in the species
Neotropical reservoirs host less than 40 species assemblage by increasing the likelihood of addi-
(Agostinho et al. 2016) adapted to lentic condi- tional invasions (Simberloff and Von Holle 1999,
tions and in some degree dependent on nutrient Simberloff and Vitule 2014, Bezerra et al. 2018).
accumulation and recycling, much of them non- Food web models in aquatic environments indi-
native (Rahel 2007, Johnson et al. 2008, 2009, Gal- cate a large dependence on detritivores and
lardo et al. 2016). Such species have been intro- pelagic filter feeders to sustain upper trophic
duced for commercial and sports fisheries, and levels dominated by predatory fishes (Liu et al.
for aquaculture (Britton and Orsi 2012, Wine- 2007). Bottom-feeding species, mainly non-
miller et al. 2016). When introduced into reser- natives, lower the degree by which predators
voirs, as opposed to natural water bodies, non- and plankton-feeding species could limit the
natives are likely to find themselves among an consumption—the benthification (Mills et al.
impoverished assemblage of native species, a fac- 2003, Mayer et al. 2014) process—biological
tor that facilitates the invasion process (Johnson invasions, and ecosystem alterations toward the
et al. 2008, Petesse and Petrere 2012). Alien inva- dominance of bottom-feeding species and
ders in reservoirs exhibit all three of the notable trophic downgrading.
aspects of biological invasion: “population Here, we investigated the ongoing dominance
spread, community dominance, and range of bottom-feeding species in altered ecosystems
expansion” (Gurevitch et al. 2011). Multiple (reservoirs) in relation to natural lakes, specifi-
invaders are likely in this kind of environment, cally, if this dominance was linked to the pres-
and their interactions, as well as their influence ence of non-native species, based on the
over the native biota, are not fully understood benthification hypothesis. To this aim, we fol-
(Britton and Orsi 2012, Simberloff and Vitule lowed three approaches: (1) an analysis of meta-
2014). Therefore, reservoirs are among the best data for a comparison of 26 environments (12
places to observe human-induced effects (dam- reservoirs and 14 lakes) under differential influ-
ming and the introduction of non-natives) over ence of non-native species; (2) a trophic web
the natural fish biodiversity (Agostinho et al. model for an artificial environment in Brazil (Pas-
2016, Winemiller et al. 2016). The introduction of sau na reservoir) to represent interactions
largemouth bass (Micropterus salmoides) and between natives and non-natives, and its effects;
tilapiine cichlids of the genera Oreochromis is and (3) a calibration of this model with stable iso-
common example of species introduced world- tope analysis from the same environment, illus-
wide for human purposes, ranking them among trating how fish invasions could result in a class
the top 100 of the world’s worst global invaders of non-native predator–prey novel interaction
(Lowe et al. 2008), and interacting by predation powered by a destabilizing positive feedback
in their habitats. loop at the ecosystem level.

❖ www.esajournals.org 2 June 2019 ❖ Volume 10(6) ❖ Article e02757

 21508925, 2019, 6, Downloaded from https://esajournals.onlinelibrary.wiley.com/doi/10.1002/ecs2.2757 by CAPES, Wiley Online Library on [28/02/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
BEZERRA ET AL.

Therefore, if reservoirs are preferred for intro- biomass of bottom feeding and plankton feeding
ductions, particularly of non-native fish (Rahel (or pelagic predators). We divided the biomass of
2007, Agostinho et al. 2016, Winemiller et al. each of these two groups by the mean biomass of
2016), and if bottom-feeding fish have prospered all groups in the ecosystem (units given in t/km2
in reservoirs, the success of non-natives and bot- or g/m2). Standard deviation (SD) was not an
tom-feeding fish could be related in some degree. input (equals to 0.5 to all groups). Therefore, the
In fact, non-native fish biomass or interactions effect size (standard mean difference, SMD) and
should be (1) especially prevalent in reservoirs SD bars were all proportional to the raw differ-
and scale-dependent (highlighting ecosystem ence in biomass between pelagic predators and
level vs. species-based approaches), (2) more com- bottom-feeding fish.
mon with other non-natives than expected with In parallel, we classified the ecosystems with
native prey and predators, and (3) linked to the high (H), intermediate (I), or low (L) influence of
distribution of trophic levels, so that the dynamics non-natives. In H, non-natives dominated the
of invaded, artificial systems can exhibit high fish community; in I, there was a similar compo-
stochasticity along with biotic homogenization. sition of natives and non-natives and a reported
ecosystem effect of the last, while in L, there were
METHODS no or few non-natives without a reported effect.
Finally, we joined habitat (lake or reservoir) and
Meta-data analytic approach non-native influence (H, I, L) as a single predic-
We examined the EcoBase (an electronic basis tor of the calculated effect sizes.
of Ecopath with Ecosim models) (available
online: http://sirs.agrocampus-ouest.fr/EcoBase) Study case on the Passa
una reservoir: study area
in a search for biomasses of fish species that fed The Passau  na reservoir (25°310 59.9″ S;
0
on benthic and pelagic sources (including ben- 049°23 16.5″ W) is one among many reservoirs
thopelagic, omnivore, and carnivore or plankton- built in tropical regions for supplying fresh water
feeding species) reported in food web models of for human use. The reservoir extends over
lentic fresh waters (lake or reservoir). We com- 10 km2 and is the primary water supply for Curi-
plemented the search with a model developed to tiba, one of the ten largest cities in Brazil with
the Passau na reservoir and Ecopath models pub- ~1.8 million inhabitants. It is relatively shallow,
lished in electronic libraries (Web of Science and with a mean depth of 6.5 m, a hydraulic resi-
Google Scholar). We only examined peer- dence time of about 290 d, and classified as
reviewed journals published in the English lan- mesotrophic (Carneiro et al. 2016). However, it
guage. To avoid pseudoreplication, we classified receives its water supply from small rivers in sur-
all fish groups at once in one of the two habit cat- rounding urban, agricultural, and industrial
egories (benthic feeding vs. plankton feeding). areas, a situation that can push water bodies
We searched the habit of all fish species, or in toward eutrophic conditions. Human activities
Froese and Pauly (2016) if the authors did not are in keeping with their multiple uses, including
mention the habit. We considered one food web irrigation, sports fishing, recreation, and water
model per published paper (the most recent) and storage (Carneiro et al. 2016). The region is in the
one effect per ecosystem (lakes or reservoirs). We C-climate zone and Cfb climate in Ko €ppen’s clas-
also examined only ecosystems in which fisheries sification for Brazil, with a mean temperature of
were allowed (except the Passau  na reservoir). If 17°C (see Alvares et al. 2013 for further details
there were two or more periods for a single on the climate).
model, we always used the most recent scenario.
For instance, we examined Lake Victoria at once, Fish and invertebrates: sampling and laboratory
in 2007. analyses
Fish were also sampled monthly from May
Meta-data approach: effect sizes 2011 to April 2012, in a combination of artificial
Since total biomass variated a lot from one baits and gillnets with panels ranging from 5 to
model to another (e.g., Lake Victoria vs. Pas- 40 m in length, 1.3–5 m in height, and a mesh
 na reservoir), we standardized the fish
sau size from 15 mm to 150 mm knot-to-knot.

❖ www.esajournals.org 3 June 2019 ❖ Volume 10(6) ❖ Article e02757

 21508925, 2019, 6, Downloaded from https://esajournals.onlinelibrary.wiley.com/doi/10.1002/ecs2.2757 by CAPES, Wiley Online Library on [28/02/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
BEZERRA ET AL.

Gillnets (on average nine sets per month) were adult Hopmal as leading the fit of other stanzas,
set in randomly stratified locations and soaked in the Multi-Stanza routine of the Ecopath model
for 24 h. Sampling with artificial baits was car- (see below; Christensen et al. 2008). We used this
ried out on 26 d, standardized with about eight routine as an application of the von Bertalanffy
hours of effort per day. Pelagic invertebrates growth function (VBGF) to approach the natural
were captured locally with plankton nets mortality (M) of middle stanzas while retrieving
(500 lm) or identified in the stomach contents of M and consumption of leading stanzas from the
fishes and recorded just as mesozooplankton— published literature (Froese and Pauly 2016). We
MesZoo (Cladocera and Copepoda). Insecta was managed the VBGF growth parameter (k) for
represented predominantly by Martarega sp. Micsal (from k = 0.3 to k = 0.9), in relation to
adults and Chaoborus sp. larvae. Hopmal (k = 0.3), and fixed the time in each
We oriented the sampling to represent the stanza (24 months) for both species. It brought
(macro) biota in specific ranges, near to the dam realism to the simulation, taking our captures
barrier, and nearshore (with 6.5 m of mean (representing a system with small Micsal and big
depth), aiming to preserve fresh tissues for stable Hopmal), and aligned with the literature, that
isotope analysis. Electrofishing was used to cap- reported a greater asymptotic length to Micsal.
ture fishes and invertebrates in a specific area
(nearshore and parallel to the dam) in a standard
transect about 2 km in length, with each sample The trophic web modeling: underlying processes
constituting 1.5 h of effort. The electrofishing behind domination of bottom feeding
occurred monthly in cold–dry (April 2011–Octo- The software Ecopath with Ecosim (EwE).—We
ber) and hot–wet (November 2011–March 2012) used the EwE program and its plug-in EcoTroph
periods. All fish specimens were identified to the (Christensen and Walters 2004, Gascuel and
lowest possible taxonomic level, measured (stan- Pauly 2009). In steady-state scenarios, the model
dard length, SL mm), and weighed (mass,  predicts the production (P) of a prey i is from its
0.1 g) in the laboratory. We also examined the predators j biomass (B), biomass accumulation
diet in the multivariate space, which tended to (BA), consumption to biomass (Q/B), and pro-
confirm the existence of three main ontogenetic duction to biomass (P/B), assuming a fraction of
functional groups. the prey’s B in the predator diet (DC)ij and an
Because adult fish are more resistant to elec- ecotrophic efficiency (EE)i, estimated by EwE
trofishing than juveniles, we captured adult life according to the equation:
stages of M. salmoides (Micsal) only with artificial
baits—maximum length in capture (Lmax):
 Xn

P Q
214 mm. Indeed, we used mainly nets to capture Bi ¼ Bj DCij þ BAj
the native predatory characoid Hoplias malabari- B i B
i¼1
 j
cus (Hopmal, Lmax: 520 mm). Even if both Lmax P
þ Bi ð1  EEi Þ (1)
were lower than expected in the literature, B i
970 mm and 650 mm, respectively (Froese and
Pauly 2016), Micsal was better represented by We had not used other factors impacting the
juveniles than Hopmal. Therefore, we classified production of i such as migration and fisheries
the first into three life stages or stanzas: juvenile because our purpose was to emphasize predation
1, J1 (smaller than 70 mm); juvenile 2, J2 (from interactions and because fisheries were prohib-
71 mm to 190 mm); and adult, A (larger than ited in the Passau na reservoir during the period
190 mm), while we placed Hopmal into two of simulation. We showed specific modeling
groups (juvenile, juv and adult), based on their steps in Appendix S1.
respective asymptotic length: 682.5 mm and Diet composition.—We identified stomach con-
1018.5 mm (Froese and Pauly 2016). tents of M. salmoides (Micsal) to the lowest possi-
Beyond patterns in the capture, we divided ble taxonomic level in the stereomicroscope while
species in stanzas for modeling purposes. We confirming identifications with specialists from
had differential information depending on spe- Cap~ao da Imbu  ia Natural History Museum in
cies life stages, so that we set juvenile Micsal and Curitiba, Brazil. First, we determined the

❖ www.esajournals.org 4 June 2019 ❖ Volume 10(6) ❖ Article e02757

 21508925, 2019, 6, Downloaded from https://esajournals.onlinelibrary.wiley.com/doi/10.1002/ecs2.2757 by CAPES, Wiley Online Library on [28/02/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
BEZERRA ET AL.

biovolume of items in the stomach contents in diet), in 100 Monte Carlo simulations (which
graduated plates, then translating into percent could mean 100 different steady states of the
representation by volume. We based the diet com- ecosystem) based on the intervals of confidence
position for M. salmoides on the stomach content related to the quality of each introduced data
of J1, J2, and adults. For diet information of the (i.e., the model Pedigree; see Appendix S1:
other components of the diet matrix, we consulted Table S2); then, we used the best trial to force tila-
Froese and Pauly (2016) or specific references that pia biomass in scenarios A and B.
investigated similar lentic and tropical reservoirs We did this procedure to represent the stochas-
(Lazzaro 1991, Luiz et al. 1998, Deus and Petrere- ticity into our modeling. The pedigree index was
Junior 2003, Delariva et al. 2013, Bezerra et al. an indicator of the degree to which native vs.
2018). We showed the general diet composition proxy data re-incorporated into Ecopath models
used as an input into the model in the supporting (Christensen et al. 2008). We adapted the stan-
information (Appendix S1: Table S1). dard values (Christensen and Walters 2004), cal-
Time-dynamic model: Ecosim.—We represented culated as follows: Biomass values estimated
the static model of EwE into a time-dynamic from other models or based on the literature
model (Ecosim; Christensen and Walters 2004). received a zero. Indirect methods, direct methods
The start year on EwE met the date of sampling with low precision, and direct methods with
(mostly on 2011), variating in two scenarios, for higher accuracy received scores of 0.4, 0.7, and 1,
the next 15 yr: In the first (scenario A), we respectively—inputs of PB, QB, and diet in the
increased the capture rate of tilapia linearly. In standards (Christensen and Walters 2004). Sto-
the second (scenario B), we decreased the capture mach analysis received 1 and literature-based
deterministically up to the point of collapse, that data 0 in the pedigree index.
is, biomass equals to zero (Appendix S1: Fig. S1). To improve the realism in our model, we
The change in relative biomass of a prey i (dBi/ included birds, known to make up part of the
dt) varies according to the master equation of food web, from an inventory in the Sıtios Novos
Ecosim: reservoir, CE, Brazil (Bezerra et al. 2018). The
geographic distance to the study area did not
dBi X X influence our simulation since we aimed to rep-
¼ gi Qij  Qij þ Ij  ðMOi þ Fi þ ei ÞBi resent trophic interactions, and because all the
dt
included birds are common in Neotropical areas
In the equation, gi is the net growth efficiency (Birdlife international initiative, data-
(production/consumption), Qij is the consump- zone.birdlife.org/home). Birds were the cor-
tion of prey i by predator(s) j (that includes the morant Phalacrocorax brasilianus, Ardeidae, and
vulnerabilities, Vij), MOi is the mortality due to other birds (uncommon but relevant groups, e.g.,
other factors, Fi is mortality due to fishing, and ei the kingfishers, halcyon, and Alcedinidae: Megac-
and ij are the emigration and immigration rates, eryle torquata and Chloroceryle americana). Except
respectively, that not variate in time in closed for some groups included in other birds, such
systems. At low vulnerabilities (Vij < 2), interac- bird species are found in freshwaters all over the
tions are subject to strong bottom-up regulation. world. Therefore, we presumed that its inclusion
At higher vulnerabilities (Vij > 4), the top-down as consumers added reality to our modeling of a
pressures predominate (Christensen et al. 2008, general scenario for reservoirs, beyond an incom-
Ahrens et al. 2012). Vulnerabilities were set ran- ing error associated with the sampling in the Pas-
domly (2 ≤ Vij ≤ 4) in a spreadsheet. sau na. For the same reason, we grouped other
Parametrization to the trophic web modeling and fish compartments in other piscivores, other
uncertainty.—The data used for the EwE analyses insectivores, other invertivores, other omnivores,
were biomass estimates of the community com- and other detritivores. The biomasses in other
ponents along with their diets, estimates of the P/ compartments were retrieved from specific refer-
B and Q/B ratios, and literature-based data for ences, considering Neotropical reservoirs (Angel-
ecotrophic efficiency. We used the routine Ecor- ini et al. 2006, Gubiani et al. 2011, 2012).
anger (Christensen and Walters 2004) to re-sam- Most of the fish compartments corresponded
ple from our input data (B, P/B, Q/B, EE, and to individual species or functional groups so that

❖ www.esajournals.org 5 June 2019 ❖ Volume 10(6) ❖ Article e02757

 21508925, 2019, 6, Downloaded from https://esajournals.onlinelibrary.wiley.com/doi/10.1002/ecs2.2757 by CAPES, Wiley Online Library on [28/02/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
BEZERRA ET AL.

we would be able to identify the role of non- predators, with 95% confidence intervals (diet
natives in the food web. In addition to matrix). Along with the other inputs (biomass,
M. salmoides, we represented tilapias in a single production, and consumption; please, see Meth-
compartment comprised of the genera Coptodon, ods: On the trophic web modeling: The software Eco-
Oreochromis, Tilapia, and their interspecific path with Ecosim (EwE), the diet determined the
hybrids. We included the common carp (Cypri- basic estimates of the EwE, which included
nus carpio) with other omnivores. effective trophic levels (TL), non-integer TL
sensu Odum (1969); Fdet, the contribution of
Ecosystem effects: support on stable isotope detritus biomass to each of the modeled group;
analysis and food web models and the omnivory index (OI), the variance on
While selecting individuals of similar size from the TL of prey of j that considers the diet frac-
the same collection site into composite samples tion of preys i to consumers (DCij; Christensen
before analysis, we removed a piece of white et al. 2008).
epaxial muscle tissue (~20 mg dry weight) above Finally, we combined the isotope signatures
the lateral line and below the dorsal fin. We iden- (individual-based predictors) with the outputs of
tified the invertebrate samples to the genus level, the EwE in a model (response variables) with the
later processed without subsamples. We dried aim to test for the best proxy of the ecosystem.
sampled tissues at 60°C in a drying oven, then To do this, we used d15N and d13C as a proxy to
measuring carbon (C) and nitrogen (N) stable the modeled TL, Fdet, and OI in general linear
isotopic ratios on separate subsamples of the models (GLM) with Poisson distributions
composite samples. We expressed isotopic values (pGLM), in three different models: TL or Fdet or
in delta (d) notation (Coplen 2011). Replicate OI ~ d15N and d13C. We carried all analyses in R
analyses of identical composites usually differed environment (R Core Team 2017).
by ≤0.1& for both C and N stable isotopes.
The sample size for each isotopic determina- RESULTS
tion was approximately 1 mg of d13C and 3 mg
of d15N. We did not obtain isotopic values for Analysis of meta-data: bottom feeding 9
J1 M. salmoides but from J2 individuals and plankton feeding
adults (72 individuals). We conducted isotopic We obtained 26 effect sizes (12 reservoirs and
analyses at the Boston University Stable Isotope 14 lakes), 23 negatives, while three positive
Laboratory (www.bu.edu/sil/index.htm). We did tended to be in lakes or a reservoir with a low
the flash combustion in tin boats at 1800°C in a influence of non-natives (Fig. 1, Table 1). The
Heraeus Elemental Analyzer, the separation of dominance of bottom-feeding fish occurred
the resulting CO2 and N2 gases from the helium mainly in reservoirs with high influence of non-
stream and cryogenic purification in a Finnigan natives (R2 = 28.06%; moderator HI effect
cold trap, before introduction into a Finnigan [QM] = 12.9; P = 0.02). The biomass of bottom-
Delta-S isotope ratio mass spectrometer. Finally, feeding fish was commonly restricted to lower
we measured against high-purity laboratory gas trophic levels (TL ≤ 3 or TL ≤ 3.5, if TL max ≥ 4),
standards that we calibrated with international contrastingly with the biomass plankton-feeding
standards (Pee Dee Belemnite [PDB] for d13C and species, generally with TL > 3 (or TL > 3.5, if TL
atmospheric air for d15N). max ≥ 4). However, the biomass of bottom feed-
For detritus (microzooplankton, seston, and lit- ing was not always lower, with a large effect size
ter) and mesozooplankton (MesZoo), we retrieved in lakes under intermediate influence non-
d13C and d15N values from Reis et al. (2017). natives (LI), that is, Annecy (France) and Lhema
Stable isotope assessments were in line with steps (Rwanda).
of a model-oriented design, predicting biomass
and diet composition for the main groups in the Ecosystem approach on the dominance of
food web (Christensen and Walters 2004). bottom-feeding and introduced species
We used the Stable Isotope Analysis in R, Overall, we captured 1489 individuals, repre-
SIAR (Parnell and Jackson 2008), to estimate the senting 20 species during the sampling program,
average contribution of the preys in the diet of and 28 compartments in the food web, that

❖ www.esajournals.org 6 June 2019 ❖ Volume 10(6) ❖ Article e02757

 21508925, 2019, 6, Downloaded from https://esajournals.onlinelibrary.wiley.com/doi/10.1002/ecs2.2757 by CAPES, Wiley Online Library on [28/02/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
BEZERRA ET AL.

Fig. 1. Analysis of meta-data on the difference in biomass of pelagic predators and bottom-feeding species in
reservoirs (R) or lakes (L), the first case letter in rows. The second letter indicates H, high (red); I, intermediate
(yellow); and L, low (blue) influence of non-native species. The H non-natives dominated the fish community, on
I, equally natives and non-natives, and, on L, few non-natives without ecosystem effects.

Table 1. Mixed-effect model, estimated with random effects and maximum likelihood (REML), of the analysis of
meta-data on the difference in the biomass of pelagic predators and bottom-feeding species in lakes or reser-
voirs with high, intermediate, and low influence of non-native species. Bolded text indicates the lowest estima-
tion (mean effect).

Influence Estimate SE Z P Lower CI Upper CI

Lake, high 0.58 0.36 1.61 0.11 1.29 0.13

Lake, intermediate 0.72 0.70 1.02 0.31 2.09 0.66
Lake, low 0.09 0.47 0.19 0.85 0.84 1.02
Reservoir, high 1.43 0.52 2.75 0.01 2.45 0.41
Reservoir, intermediate 0.33 0.51 0.65 0.52 1.34 0.67
Reservoir, low 0.27 0.68 0.40 0.69 1.06 1.59
Note: Confidence intervals (CI) at lower and upper 95% limits and residual heterogeneity/unaccounted variability (I2) are shown.

included 17 fish compartments, five inverte- categories J1, J2, and A, respectively. The native
brates, three birds, two primary producers, and fish species included the large predator Hoplias
detritus (Fig. 2, Table 2). We captured 105, 147, malabaricus (a characoid), the herbivorous catfish
and 122 individuals of M. salmoides from Hypostomus derbyi, and the omnivorous

❖ www.esajournals.org 7 June 2019 ❖ Volume 10(6) ❖ Article e02757

 21508925, 2019, 6, Downloaded from https://esajournals.onlinelibrary.wiley.com/doi/10.1002/ecs2.2757 by CAPES, Wiley Online Library on [28/02/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
BEZERRA ET AL.

Fig. 2. Trophic spectra (sensu Gascuel and Pauly 2009) of log B and log Bacc vs. trophic level (TL) relative to
the food web model of the Passau  na reservoir. Inset: keystone index (KI) vs. relative total impact (RI) of 28
groups scaled by biomass (bubble area), where RI > 0.6 and KI > zero are identified. Hopmal, Hoplias malabari-
cus; Micsal, Micropterus salmoides; and J, juveniles 1 and 2 and A, adults.

Geophagus brasiliensis (cichlid), Astyanax spp. biomass of tilapia decreased the M. salmoides bio-
(characin), Corydoras paleatus, and Rhamdia quelen mass on 0.1 units (Fig. 3A), but an exponential
(two more catfishes). Birds had the highest effec- increase led to a decrease (Fig. 3B).
tive trophic level (TL 3.76), followed by
M. salmoides A (TL 3.66), M. salmoides J2 (TL Support on stable isotope analysis: ecosystem
3.62), and H. malabaricus adults (TL 3.59) and effects of the bottom feeding
juveniles (TL 3.41). Top predators had the highest The d13C and the d15N values of tilapia were
accumulated biomasses, supported primarily by more disperse than M. salmoides (juvenile and
the biomass of fishes at intermediate trophic adult) and the native predator H. malabaricus,
levels (from TL 2.57, corresponding to C. palea- respectively (Fig. 4). Carbon and nitrogen signa-
tus, to Astyanax spp., TL 2.857 and tilapia, TL tures explained TL (Fig. 5A, B; P < 0.05) and
2.862). Detrital pathways contributed nearly Fdet (Fig. 5C), but not OI. The consumption of
twice the amount of assimilated food as primary d13C sources decreased with TL (Fig. 5A), while
production pathways. the d15N distribution increased, related to the
Keystone groups in the ecosystem were consumption on detritus (Fig. 5B, C), indepen-
M. salmoides adult (122 individuals) and J2, dently on the richness of items in the diet of a
H. malabaricus adult, and other piscivores, each given consumer.
with a keystone index (KI) greater than 1.0 and However, contrary to we expected from the
relative total impact (RI) close to 1.0. In the Eco- modeled TL, mean d13C, and mean d15N, values
sim, non-natives presented a differential response were highest for the benthic fishes C. paleatus
to the theoretical variation in the biomass of tilapia and H. derbyi (Fig. 6). Increasing TL correlated
(in gm2yr1), from 2011 to 2030. The native positively with the mean d15N variation (Spear-
H. malabaricus biomass remained almost unaltered man’s r = 0.76, P < 0.01), but not with the mean
in the two scenarios (Fig. 3A, B). A decrease in the d13C variation (Spearman’s r = 0.29, P = 0.35),

❖ www.esajournals.org 8 June 2019 ❖ Volume 10(6) ❖ Article e02757

 21508925, 2019, 6, Downloaded from https://esajournals.onlinelibrary.wiley.com/doi/10.1002/ecs2.2757 by CAPES, Wiley Online Library on [28/02/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
BEZERRA ET AL.

 na reservoir, from 2011 to 2012.

Table 2. Food web model of the Passau

Group Name TL B PB QB EE NE-GE RA RB

B Ardeidae 3.82 1.66 0.26 1.71 0.329 0.04 0.81 1.11

B Phalacrocorax brasilianus 3.69 0.04 0.11 0.77 0 0.04 0.82 0.51
F Micropterus salmoides A 3.66 0.16 0.69 21.5 0.09 0.01 0.96 16.51
F Micropterus salmoides J2 3.62 0.49 0.92 23.59 0.038 0.01 0.95 17.95
F Hoplias malabaricus A 3.59 3.46 0.9 5.7 0.138 0.04 0.8 3.66
F Hoplias malabaricus J 3.41 1.66 1.5 12.53 0.723 0.03 0.85 8.52
F Micropterus salmoides J1 3.33 0.61 1.03 34.83 0.131 0.01 0.96 26.83
F Other piscivorous 3.29 1.57 2.99 23.76 0.853 0.03 0.84 16.02
B Other Birds 3.13 1.73 0.19 1.28 0.769 0.04 0.81 0.83
F Rhamdia quelen 3.12 0.18 1.06 4.25 0.537 0.06 0.69 2.34
F Other insectivorous 3.06 2.33 4.55 23.23 0.91 0.05 0.76 14.03
F Tilapia 2.86 4.96 3.27 13.07 0.825 0.06 0.69 7.19
F Astyanax spp. 2.86 1.04 11.75 47 0.756 0.06 0.69 25.85
F Geophagus brasiliensis 2.85 1 2.13 8.53 0.894 0.06 0.69 4.69
F Other invertivorous 2.65 2.26 2.16 20.2 0.982 0.03 0.87 14
F Other omnivorous 2.64 2.56 6.23 39.47 0.748 0.04 0.8 25.35
F Corydoras paleatus 2.57 1.25 1.78 16.52 0.9 0.07 0.82 8.13
F Hypostomus derbyi 2.4 0.88 0.92 21.73 0.726 0.03 0.93 12.12
I Insecta 2.27 13.68 10.09 40.36 0.9 0.06 0.69 22.2
F Other detritivorous 2.23 2.31 5.29 44.72 0.922 0.03 0.85 30.49
I Other Crustacea 2.16 12.7 22.5 90 0.498 0.06 0.69 49.5
I Copepoda 2.11 8.88 24.18 96.70 0.161 0.06 0.69 53.19
F Other herbivorous 2 2.96 8.16 29.26 0.799 0.07 0.65 15.25
I Other Benthos 2 13.2 14.88 59.52 0.9 0.17 0.58 20.83
I Cladocera 2 29.64 18.99 75.94 0.289 0.06 0.69 41.77
D Detritus 1 23.75 0.452
P Periphyton 1 26.26 40 0.46
P Phytoplankton 1 31.66 183 0.29
Notes: Group identity abbreviations are as follows: B, birds; D, detritus; F, fish; I, invertebrates; and P, primary producers.
Other abbreviations (and units) are as follows: TL, trophic level; B, biomass (tonkm2year1); PB, relative production (year1);
QB, relative consumption (year1); EE, ecotrophic efficiency; NE-GE, net efficiency minus gross efficiency; RA, respiration/as-
similation; RB, respiration/biomass; J1 and J2, juveniles; and A, adults.

while TL was explained by both d15N and d13C, DISCUSSION

the isotopic variation, unrevealing a feedback of
redundant nitrogen food sources (see also the Our results suggest a generalized state of biotic
Appendix S1: Fig. S2). The only groups outside homogenization in lentic fresh waters under the
of the expected 95% confidence interval were the influence of damming and dominance of bottom-
intermediate consumers C. paleatus, H. derbyi, feeding species. The trophic downgrading is
and MesZoo. Indeed, the range of d13C isotope broadly described in terrestrial and aquatic envi-
values for H. derbyi was the greatest among all ronments (Estes et al. 2011), but the underlying
the groups tested. mechanisms are obscure, and their link with
Adults and J2, the classes represented in the invasional biology and secondary invasions was
stable isotope analysis, did not differ with not observed until now. Our study highlights a
respect to carbon sources (t = 0.81, df = 50.73, kind of biotic homogenization, ongoing with
P = 0.42), but did differ with respect to nitrogen trophic downgrading and the success of bottom
sources (t = 2.08, df = 52.76, P < 0.05; feeders, which characterize the benthification
Appendix S1: Fig. S3). Adults of M. salmoides process. In line with the study case of the Pas-
had a greater range of d15N than juveniles (see  na reservoir, the low trophic-level predomi-
sau
Appendix S1: Fig. S3), which was also reflected nance of bottom feeding in such dam
by TL. environments, Johnson et al. (2008) worldwide

❖ www.esajournals.org 9 June 2019 ❖ Volume 10(6) ❖ Article e02757

 21508925, 2019, 6, Downloaded from https://esajournals.onlinelibrary.wiley.com/doi/10.1002/ecs2.2757 by CAPES, Wiley Online Library on [28/02/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
BEZERRA ET AL.

Fig. 3. Ecosim model of the Micropterus salmoides adult (Micsal A), Hoplias malabaricus adult (Hopmal A), and
tilapia, after Monte Carlo simulations on input values of the Ecopath model in the Passau  na reservoir, from 2011
to 2026. Scenarios are expressed with a deterministic increase A or decrease B in the capture of tilapia (see
Appendix S1: Fig. S1).

show that damming can be related to the occur- probably their preferred prey in artificial reser-
rence of non-native, generalist prey and preda- voirs) and a variety of other detritivores, much
tors, which can trigger cascading effects toward more than other compartments (Starling et al.
benthic detrital pathways (see also Bezerra et al. 2002). More than species, we must understand
2018). these invaders as functional groups. Then, ben-
The prospected mechanisms of trophic homog- thification patterns could be expanded to the co-
enization described here (i.e., this dominance of occurrence of tilapia (or another invader with
bottom-feeding fish in low TL) can be attributed bottom-feeding habits) and other non-native
to the invader aspect of non-natives. Owing to its predators such as Cichla monoculus Agassiz, 1831
rapid somatic and population growth and signif- interacting with negative effects to native species,
icant resistance to environmental alteration such as Oligosarcus longirostris Menezes & Gery,
(Canonico et al. 2005, Deines et al. 2016), non- 1983 (Fig. 7). It was observed in less diverse
native such as tilapia seems highly adaptable to regions under drought effects, in which the avail-
the predation pressure (Bwanika et al. 2006) of able biomass of invaders in intermediate TLs was
M. salmoides and serves as the preferred food dominated by tilapia (Bezerra et al. 2018). Cas-
item in its diet. Indeed, tilapia foster detrital cading effects should be expected with an
cycling by consuming mosquitoes (larval phase, increase in allochthone detritus and autochthone

❖ www.esajournals.org 10 June 2019 ❖ Volume 10(6) ❖ Article e02757

 21508925, 2019, 6, Downloaded from https://esajournals.onlinelibrary.wiley.com/doi/10.1002/ecs2.2757 by CAPES, Wiley Online Library on [28/02/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
BEZERRA ET AL.

Fig. 4. (A) Stable isotope analysis in R (SIAR) plot with standard deviance of isotope values for Astsp (Astya-
nax spp.), Corpal (Corydoras paleatus), Geobra (Geophagus brasiliensis), Hypder (Hypostomus derbyi), Hopmal A
(Hoplias malabaricus adult), MesZoo (mesozooplankton, that are Cladocera and Copepoda), Micsal A (Micropterus
salmoides adult), Micsal J (Micropterus salmoides juveniles 1 and 2), and Rhaque (Rhamdia quelen). (B) A close-up of
(A) with Micsal A in solid lines and tilapia, Micsal J, and Hopmal A in dashed lines.

cycling by microorganisms and benthic inverte- predator native species, such as H. malabaricus
brates (Cole et al. 2006), which could be attribu- and R. quelen. Beyond serving as prey, intermedi-
ted to the biomass production on benthos ate consumers such as tilapia and the native spe-
(Cremona et al. 2016) (Fig. 8). cies G. brasiliensis, H. derbyi, and C. paleatus
Despite the stronger ecosystem impacts of stimulate nitrogen cycling in the food web of
M. salmoides adults, juveniles began feeding on reservoirs by the foraging on benthos (Lowe-
large individuals of tilapia and other bottom- McConnell 2000, Canonico et al. 2005, Bezerra
feeding species very early, relying heavily on tila- et al. 2018). The eutrophication of freshwaters is
pia (especially juvenile 2) as compared to other related to the presence of tilapiines (Starling
diet items, therefore triggering the cascading et al. 2002). The increase in TL with increasing
effects discussed here. Top predators had mas- carbon sources and, alternatively, no variation on
sive nitrogen accumulation rates and, conse- carbon sources due to increasing TL could high-
quently, higher TLs than adults of other top light the restriction to the consumption of

❖ www.esajournals.org 11 June 2019 ❖ Volume 10(6) ❖ Article e02757

 21508925, 2019, 6, Downloaded from https://esajournals.onlinelibrary.wiley.com/doi/10.1002/ecs2.2757 by CAPES, Wiley Online Library on [28/02/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
BEZERRA ET AL.

Fig. 5. Modeled ecosystem variables. Trophic level (TL) answered by mean d13C (A), mean d15N (B), and flow
from detritus (Fdet, C) according to the mean d15N values in Poisson-distributed general linear models (pGLM).
Only patterns (P < 0.05), from the prediction of ecosystem variables (y-axis) by individual-based proxies (x-axis),
are reported.

❖ www.esajournals.org 12 June 2019 ❖ Volume 10(6) ❖ Article e02757

 21508925, 2019, 6, Downloaded from https://esajournals.onlinelibrary.wiley.com/doi/10.1002/ecs2.2757 by CAPES, Wiley Online Library on [28/02/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
BEZERRA ET AL.

Fig. 6. (A) Mean d13C and mean d15N values ordered by trophic level (TL) according to the modeled groups:
detritus, MesZoo (mesozooplankton, Cladocera and Copepoda), Insecta, Hypder (Hypostomus derbyi), Corpal
(Corydoras paleatus), Geobra (Geophagus brasiliensis), Astsp (Astyanax spp.), Rhaque (Rhamdia quelen), Hopmal A
(Hoplias malabaricus adult), Micsal juv (Micropterus salmoides juveniles 1 and 2), and Micsal A (Micropterus sal-
moides adult). (B) Main axis represents flow from detritus (Fdet) and omnivory index (OI), and secondary axis
represents TL.

carbonated sources to low trophic levels. Their et al. 2011), among other sources, because of the
high d13C probably reflects heavy feeding upon presence of N2-fixing cyanobacteria and other N-
CH4-rich or recycled heterogeneous food rich sources (Darchambeau et al. 2014), which
sources, such as microorganisms, the remains of can be related to the N-based fertilizers
animals, and plants (Fish and Carpenter 1982, employed in aquaculture (Quesada et al. 2013)
Cremona et al. 2016). Carbon sources were only and to agricultural activities in surrounding
indirectly related to TLs and a better reflection of areas (Stein and Klotz 2016). Indeed, recent stud-
carbon sourcing (Post 2002). ies emphasize the role of the bacterial loop as
Commonly, fish are limited by nitrogen in arti- bottom-up support, instead of the phytoplankton
ficial environments (Cole et al. 2006, Holtgrieve by itself (Darchambeau et al. 2014). Considering

❖ www.esajournals.org 13 June 2019 ❖ Volume 10(6) ❖ Article e02757

 21508925, 2019, 6, Downloaded from https://esajournals.onlinelibrary.wiley.com/doi/10.1002/ecs2.2757 by CAPES, Wiley Online Library on [28/02/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
BEZERRA ET AL.
Colour online

Fig. 7. Snowball effects: In both scenarios (native/invader prey and predators), predation is largely dependent
on detritus cycling (intermediate consumers feeding on benthos and detritus), but the influence of invaders is dis-
proportional at the ecosystem level. Arrows and engines (in blue and black) indicate the intensity of processes
and effects size. Community effects could arise from interactions between individuals, without a noticeable
impact on the ecosystem, for instance, predation avoidance by aggregation. Otherwise, an increase in the mean
turbidity from this interaction indicates an ecosystem effect.

the general human pressure on lakes and reser- Tilapia and carp are considered responsible for
voirs, the deposition of reactive nitrogen could increasing detritivory rates associated with
be the cause for nitrogen limitation (Holtgrieve increasing habitat heterogeneity not only in
et al. 2011). The level of detritivory could be reservoirs but also in urban ecosystems in gen-
another point related to the tilapia’s effects. At eral (Ossola et al. 2016). The common carp
the ecosystem level, it could result in eutrophica- (C. carpio) is a common invader included in other
tion and a decrease in the weight at the capture omnivores in our approach (see Material and
of top predator fish (Attayde et al. 2011). Methods). Members of this group are no less

❖ www.esajournals.org 14 June 2019 ❖ Volume 10(6) ❖ Article e02757

 21508925, 2019, 6, Downloaded from https://esajournals.onlinelibrary.wiley.com/doi/10.1002/ecs2.2757 by CAPES, Wiley Online Library on [28/02/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
BEZERRA ET AL.
Colour online

Fig. 8. Blue to alternative green states toward benthification by the increasing influence of invaders and detri-
tivory at the ecosystem level. Less representative groups: Other birds, Other Insectivorous, Phalacrocorax brasil-
ianus, and Rhamdia quelen were omitted. Circles indicate biomasses, connect lines follow the diet matrix, and
warming colors indicate interaction strengthening (adapted from Bezerra et al. 2018).

critical than tilapias as ecosystem engineers introduced in more than 50 countries for sports
(Estes et al. 2011) and omnivore–planktivore spe- fishing (Jackson 2002, Froese and Pauly 2016),
cies in shallow reservoirs worldwide (Liu et al. including Neotropical reservoirs (Britton and
2007, Matsuzaki et al. 2009, Colvin et al. 2015). Orsi 2012). As a large and widespread fish,
An occasional spillover of these fish downstream M. salmoides reduced and/or extirpated the
threatens ecosystems given the role of native fish fauna in many places (Jackson 2002,
M. salmoides as a keystone group (Blois et al. Takamura 2007, Shelton et al. 2008, Ellender
2013, Pinnegar et al. 2014). This is of concern et al. 2011, Natsumeda et al. 2015), exerting a
since tilapiine cichlids, for example, members of stronger predation pressure than native preda-
the genera Coptodon, Oreochromis, and Tilapia, are tors (Alexander et al. 2014) or other non-natives
the most widespread group of non-native fishes (Trumpickas et al. 2011). Along with
worldwide (Canonico et al. 2005, Cassemiro M. salmoides, adverse ecological effects due to the
et al. 2018), presently cultured in most reservoirs introduction of tilapiine fishes have been
of South America (Lima Junior et al. 2018). Such reported throughout the world (Deines et al.
introductions are huge in Brazilian freshwaters 2016). Without resource limitation, that is, in the
since the 1970s (Ortega et al. 2015) and now presence of invaders such as tilapia, populations
achieved a critical stage in which they should be of M. salmoides could readily reach a density
monitored in water quality assessments equilibrium and optimal growth rates, with juve-
(Gondwe et al. 2011). niles and adults increasing in competitiveness
Given the spread of tilapia as the primary pro- with the native biota. Under drought effects, the
tein source in the fish aquaculture (Deines et al. consumption of apparently cul-de-sac species by
2016), the passion for M. salmoides and other bass tilapia is huge (Bezerra et al. 2018). It can result
fish in the sport fishery makes imminent their in positive feedbacks further benefiting tilapia
effects as co-occurring invaders in natural envi- and other invader predators and reinforcing
ronments. Indeed, M. salmoides is a top predator dominance and determinism within the system.

❖ www.esajournals.org 15 June 2019 ❖ Volume 10(6) ❖ Article e02757

 21508925, 2019, 6, Downloaded from https://esajournals.onlinelibrary.wiley.com/doi/10.1002/ecs2.2757 by CAPES, Wiley Online Library on [28/02/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
BEZERRA ET AL.

The dominance of cul-de-sac species was previ- the presence of native and non-native fish com-
ously justified in the absence of predators (Twar- munities under anthropogenic impacts resulted
dochleb and Olden 2016), but, under certain in non-deterministic patterns in lentic and lotic
circumstances, it can be the key to understand artificial environments (Liew et al. 2018). An
the ecosystem dynamics. extensive review of threats to South American
Overall, biotic homogenization and eutrophica- environments incorporates a more comprehen-
tion, related to widespread invaders, can be one sive collection of effects of non-native introduc-
of the immediate consequences of an invasional tions, biological invasions, and related aspects
meltdown (Simberloff and Von Holle 1999, Brit- (Bezerra et al. 2019).
ton and Orsi 2012, Hossain et al. 2013) in some
reservoirs. Nevertheless, available evidence for CONCLUSIONS
the original invasional meltdown hypothesis
(IMH) covers a range of processes, communities, We believe to have represented a tendency
and ecosystems (Gurevitch 2006, Jeschke et al. toward the dominance of bottom-feeding fish spe-
2012, Saul and Jeschke 2015). Positive feedbacks cies related to damming, introductions, and
from interacting invaders are among recognized trophic downgrading in altered ecosystems.
processes shaping the biodiversity, in the area Mechanisms behind the trophic homogenization
ecology and conservation biology (Gurevitch can be referred to lentic systems under intense
2006). Now, many studies focus on the facilitation pressure for introductions and that sample most
at a community level, but few evaluated feedback of their components locally. We controlled the
mechanisms with consequences to the ecosystem stochasticity with Bayesian statistics in all the pro-
(Gurevitch 2006, Jeschke et al. 2012). We hope to cedures employed (Mixed-effect model in the
have stimulated further investigations, particu- analysis of meta-data, Ecoranger in the Ecopath,
larly effects of the co-occurrence of damming, bio- and SIAR in the stable isotopic modeling), and
logical invasions, and benthification. Our results the resulting patterns of all three approaches were
point only to one consequence of this obnoxious well correlated. Considering the introduction of
engagement, biotic homogenization, which is tilapia and its predators, a future of non-native
compromising ecosystem services and threaten- species driving ecosystems is imminent in aquatic
ing the native fish fauna by increasing competi- environments (Ossola et al. 2016) where the
tion and predation (Weyl 2008, Sun et al. 2010, incoming economic problems are conspicuous.
Sanches et al. 2012) resulting in unpredictable The feedback loop highlighted here could shape
ecosystem alterations. an acceleration of the effects by itself, at the
We highlight that consequences of introduc- ecosystem level, thereby becoming inexorable.
tions and interactions among non-natives are
dependent on scale and habitat, as well as on the
ACKNOWLEDGMENTS
recipient community and novel interactions with
the non-native community. Undeniably, reser- This study was financed in part by the Coordenacß~ ao
voirs can also add to the local economy. An de Aperfeicßoamento de Pessoal de Nıvel Superior—
impoundment increases the water surface area, Brazil (CAPES)—Finance Code 001, and Fundacß~ ao da
and reservoirs are whiling to the introduction of Universidade Federal do Paran a—FUNPAR, CNPJ
game fish and top predators (Liew et al. 2018). 78.350.188/0001-95. J.R.S.V. and A.A.P. thank the Con-
New introductions in previously invaded sys- selho Nacional de Desenvolvimento Cientıfico e Tec-
tems can also alleviate the native community, gico (CNPq) for providing the research
nolo
such as suggested by antagonistic effects of intro- productivity grants (Project numbers: 310850/2012-6,
303776/2015-3, 307984/2015-0, 402828/2016-0). We are
duced crayfish on invader bullfrogs, in China
grateful for isotopic assessments made by the staff of
(Liu et al. 2018). However, the precautionary
Boston University Stable Isotope Laboratory Samples.
principle is a warranty on the secondary effects Thank you very much Dr. Peter Moyle (Department of
of new introductions. Then, we reinforce the uti- Wildlife, Fish, and Conservation Biology, University of
lization of this principle, before such enormous California) and Dr. James A. Nienow (Biology Depart-
interventions on natural and artificial environ- ment, Valdosta State University) for their English
ments. The literature is vast, a recent review of reviews and other appointments in early versions of the

❖ www.esajournals.org 16 June 2019 ❖ Volume 10(6) ❖ Article e02757

 21508925, 2019, 6, Downloaded from https://esajournals.onlinelibrary.wiley.com/doi/10.1002/ecs2.2757 by CAPES, Wiley Online Library on [28/02/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
BEZERRA ET AL.

manuscript. Thanks also to Dr. Daniel Simberloff and to biodiversity. Fish and Fisheries. https://doi.org/10.
Dr. Olaf Weyl for severe criticisms on previous versions 1111/faf.12365
of this paper. The authors have no conflict of interest to Blois, J. L., P. L. Zarnetske, M. C. Fitzpatrick, and S.
declare. Author contributions: LAVB, VMR, and JRSV Finnegan. 2013. Climate change and the past, pre-
noticed field evidence and developed the initial ideas sent, and future of biotic interactions. Science
about potential positive interactions between invasive 341:499–504.
fish in reservoirs. JRSV, VMR, MOF, and LK conceived Britton, J. R., and M. L. Orsi. 2012. Non-native fish in
the original idea of comparing largemouth bass isotopic aquaculture and sport fishing in Brazil: economic
signatures in reservoirs with their prey. VMR, MOF, benefits versus risks to fish diversity in the upper
and LK analyzed the feeding ecology data. LAVB wrote River Paran a Basin. Reviews in Fish Biology and
the first draft of the paper and suggested new methods Fisheries 22:555–565.
and a theoretical approach, including the analysis of Bwanika, G. N., L. J. Chapman, Y. Kizito, and J.
meta-data, the modeling, and the general aspect of ben- Balirwa. 2006. Cascading effects of introduced Nile
thification in altered ecosystems. AAP discussed previ- perch (Lates niloticus) on the foraging ecology of
ous and new analytical approaches, particularly Nile tilapia (Oreochromis niloticus). Ecology of
considering the analysis of meta-data. All the authors Freshwater Fish 15:470–481.
contributed substantially to the final form. Canonico, G. C., A. Arthington, J. K. McCrary, and M.
L. Thieme. 2005. The effects of introduced tilapias
on native biodiversity. Aquatic Conservation Mar-
LITERATURE CITED ine and Freshwater Ecosystems 15:463–483.
Carneiro, C., P. Kelderman, and K. Irvine. 2016. Assess-
Agostinho, A. A., L. C. Gomes, N. C. L. Santos, J. C. G. ment of phosphorus sediment–water exchange
Ortega, and F. M. Pelicice. 2016. Fish assemblages through water and mass budget in Passau  na Reser-
in Neotropical reservoirs: colonization patterns, voir (Paran a State, Brazil). Environmental Earth
impacts and management. Fisheries Research Sciences 75:564.
173:26–36. Cassemiro, F. A. S., D. Bailly, W. J. da Gracßa, and A. A.
Ahrens, R. N. M., C. J. Walters, and V. Christensen. Agostinho. 2018. The invasive potential of tilapias
2012. Foraging arena theory. Fish and Fisheries (Osteichthyes, Cichlidae) in the Americas. Hydro-
13:41–59. biologia 817:133–154.
Alexander, M. E., J. T. A. Dick, O. L. F. Weyl, T. B. Christensen, V., and C. J. Walters. 2004. Trade-offs in
Robinson, and D. M. Richardson. 2014. Existing ecosystem-scale optimization of fisheries manage-
and emerging high impact invasive species are ment policies. Bulletin of Marine Science 74:549–
characterized by higher functional responses than 562.
natives. Biology Letters 10:20130946. Christensen, V., C. J. Walters, D. Pauly, and R. Forrest.
Alvares, C. A., J. L. Stape, P. C. Sentelhas, J. L. De Mor- 2008. Ecopath with Ecosim version 6 user guide.
aes Goncßalves, and G. Sparovek. 2013. Ko €ppen’s Volume 281. Fish Centre, University of British
climate classification map for Brazil. Meteorologis- Columbia, Vancouver, British Columbia, Canada:
che Zeitschrift 22:711–728. 1–235.
Angelini, R., A. A. Agostinho, and L. C. Gomes. 2006. Cole, J. J., S. R. Carpenter, M. L. Pace, M. C. Van de
Modeling energy flow in a large Neotropical reser- Bogert, J. L. Kitchell, and J. R. Hodgson. 2006. Dif-
voir: a tool do evaluate fishing and stability. ferential support of lake food webs by three types
Neotropical Ichthyology 4:253–260. of terrestrial organic carbon. Ecology Letters 9:558–
Attayde, J. L., J. Brasil, and R. A. Menescal. 2011. 568.
Impacts of introducing Nile tilapia on the fisheries Colvin, M. E., C. L. Pierce, and T. W. Stewart. 2015. A
of a tropical reservoir in North-eastern Brazil. Fish- food web modeling analysis of a Midwestern, USA
eries Management and Ecology 18:437–443. eutrophic lake dominated by non-native common
Bezerra, L. A. V., R. Angelini, J. R. S. Vitule, M. Coll, carp and zebra mussels. Ecological Modelling
and J. I. Sanchez-Botero. 2018. Food web changes 312:26–40.
associated with drought and invasive species in a Coplen, T. B. 2011. Guidelines and recommended
tropical semiarid reservoir. Hydrobiologia 817: terms for expression of stable-isotope-ratio and
475–489. gas-ratio measurement results. Rapid Communica-
Bezerra, L. A. V., M. O. Freitas, V. S. Daga, T. V. T. tions in Mass Spectrometry 25:2538–2560.
Occhi, L. Faria, A. P. L. Costa, A. A. Padial, V. Cremona, F., A. Laas, L. Arvola, D. Pierson, P. No ~ges,
Prodocimo, and J. R. S. Vitule. 2019. A network and T. No ~ges. 2016. Numerical exploration of the
meta-analysis of threats to South American fish planktonic to benthic primary production ratios in

❖ www.esajournals.org 17 June 2019 ❖ Volume 10(6) ❖ Article e02757

 21508925, 2019, 6, Downloaded from https://esajournals.onlinelibrary.wiley.com/doi/10.1002/ecs2.2757 by CAPES, Wiley Online Library on [28/02/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
BEZERRA ET AL.

lakes of the Baltic Sea Catchment. Ecosystems Gurevitch, J., G. A. Fox, G. M. Wardle, Inderjit, and D.
19:1386–1400. Taub. 2011. Emergent insights from the synthesis
Darchambeau, F., H. Sarmento, and J. P. Descy. 2014. of conceptual frameworks for biological invasions.
Primary production in a tropical large lake: the role Ecology Letters 14:407–418.
of phytoplankton composition. Science of the Total Holtgrieve, G. W., et al. 2011. A coherent signature of
Environment 473–474:178–188. anthropogenic nitrogen deposition to remote
Deines, A. M., M. E. Wittmann, J. M. Deines, and D. watersheds of the northern hemisphere. Science
M. Lodge. 2016. Tradeoffs among ecosystem ser- 334:1545–1548.
vices associated with global tilapia introductions. Hossain, M. M., G. Perhar, G. B. Arhonditsis, T. Matsu-
Reviews in Fisheries Science & Aquaculture ishi, A. Goto, and M. Azuma. 2013. Examination of
24:178–191. the effects of largemouth bass (Micropterus sal-
Delariva, R. L., N. S. Hahn, and E. A. L. Kashiwaqui. moides) and bluegill (Lepomis macrochirus) on the
2013. Diet and trophic structure of the fish fauna in ecosystem attributes of lake Kawahara-oike, Naga-
a subtropical ecosystem: impoundment effects. saki, Japan. Ecological Informatics 18:149–161.
Neotropical Ichthyology 11:891–904. Jackson, D. A. 2002. Ecological effects of Micropterus
Deus, C. P., and M. Petrere-Junior. 2003. Seasonal diet introductions: the dark side of black bass. Pages
shifts of seven fish species in an Atlantic rainforest 221–232in American Fisheries Society Symposium
stream in Southeastern Brazil. Brazilian Journal of 31.
Biology 63:579–588. mez Aparicio, S. Haider, T. Heger, C.
Jeschke, J., L. Go
Ellender, B. R., O. L. F. Weyl, and E. R. Swartz. 2011. Lortie, P. Pysek, and D. Strayer. 2012. Support for
Invasion of a headwater stream by non-native major hypotheses in invasion biology is uneven
fishes in the Swartkops River System, South Africa. and declining. NeoBiota 14:1–20.
African Zoology 46:39–46. Johnson, P. T. J., J. D. Olden, C. T. Solomon, and M. J.
Estes, J. A., et al. 2011. Trophic downgrading of planet Vander Zanden. 2009. Interactions among inva-
earth. Science 333:301–306. ders: community and ecosystem effects of multiple
Fish, D., and S. R. Carpenter. 1982. Leaf litter and lar- invasive species in an experimental aquatic system.
val mosquito dynamics in tree-hole ecosystems. Oecologia 159:161–170.
Ecology 63:283–288. Johnson, P. T., J. D. Olden, and M. J. Vander Zanden.
Froese, R., and D. Pauly. 2016. Fishbase. FishBase. 2008. Dam invaders: Impoundments facilitate bio-
https://www.fishbase.org logical invasions into freshwaters. Frontiers in
Gallardo, B., M. Clavero, M. I. Sanchez, and M. Vil a. Ecology and Environment 6:357–363.
2016. Global ecological impacts of invasive species Lazzaro, X. 1991. Feeding convergence in South Amer-
in aquatic ecosystems. Global Change Biology ican and African zooplanktivorous cichlids Geopha-
22:151–163. gus brasiliensis and Tilapia rendalli. Environmental
Gascuel, D., and D. Pauly. 2009. EcoTroph: modelling Biology of Fishes 31:283–293.
marine ecosystem functioning and impact of fish- Leprieur, F., O. Beauchard, S. Blanchet, T. Oberdorff,
ing. Ecological Modelling 220:2885–2898. and S. Brosse. 2008. Fish invasions in the world’s
Gondwe, M. J. S., S. J. Guildford, and R. E. Hecky. river systems: when natural processes are blurred
2011. Carbon, nitrogen and phosphorus loadings by human activities. PLoS Biology 6:e28.
from tilapia fish cages in Lake Malawi and factors Liew, J. H., X. Giam, E. Clews, K. Y. W. Tan, H. H. Tan,
influencing their magnitude. Journal of Great Z. Y. Kho, and D. C. J. Yeo. 2018. Contrasting
Lakes Research 37:93–101. changes in freshwater fish assemblages and food
Gubiani, E. A., R. Angelini, L. C. G. Vieira, L. C. webs follow modification of tropical waterways.
Gomes, and A. A. Agostinho. 2011. Trophic models Ecology of Freshwater Fish 27:1114–1125.
in Neotropical reservoirs: testing hypotheses on Lima Junior, D. P., A. L. B. Magalh~ aes, F. M. Pelicice, J.
the relationship between aging and maturity. Eco- R. S. Vitule, V. M. Azevedo-Santos, M. L. Orsi, D.
logical Modelling 222:3838–3848. Simberloff, and A. A. Agostinho. 2018. Aquacul-
Gubiani, E. A., L. C. Gomes, and A. A. Agostinho. ture expansion in Brazilian freshwaters against the
2012. Estimates of population parameters and con- Aichi Biodiversity Targets. Ambio 47:427–440.
sumption/biomass ratio for fishes in reservoirs, Liu, Q.-G., Y. Chen, J.-L. Li, and L.-Q. Chen. 2007. The
Parana State, Brazil. Neotropical Ichthyology food web structure and ecosystem properties of a
10:177–188. filter-feeding carps dominated deep reservoir
Gurevitch, J. 2006. Commentary on Simberloff (2006): ecosystem. Ecological Modelling 203:279–289.
meltdowns, snowballs and positive feedbacks. Liu, X., et al. 2018. More invaders do not result in
Ecology Letters 9:919–921. heavier impacts: The effects of non-native bullfrogs

❖ www.esajournals.org 18 June 2019 ❖ Volume 10(6) ❖ Article e02757

 21508925, 2019, 6, Downloaded from https://esajournals.onlinelibrary.wiley.com/doi/10.1002/ecs2.2757 by CAPES, Wiley Online Library on [28/02/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
BEZERRA ET AL.

on native anurans are mitigated by high densities Ortega, J. C. G., H. F. Ju lio, L. C. Gomes, and A. A.
of non-native crayfish. Journal of Animal Ecology Agostinho. 2015. Fish farming as the main driver
87:850–862. of fish introductions in Neotropical reservoirs.
Lowe, S., M. Browne, S. Boudjelas, and M. De Poorter. Hydrobiologia 746:147–158.
2000. 100 of the world’s worst invasive alien spe- Ossola, A., A. K. Hahs, M. A. Nash, and S. J. Livesley.
cies: a selection from the global invasive species 2016. Habitat complexity enhances comminution
database. Invasive Species Specialist Group, Auck- and decomposition processes in urban ecosystems.
land, New Zealand. Ecosystems 19:927–941.
Lowe-McConnell, R. H. 2000. The roles of tilapias in Parnell, A., and A. Jackson. 2008. SIAR: stable isotope
ecosystems. Pages 129–162 in M. C. M. Beveridge analysis in R. R Foundation for Statistical Comput-
and B. J. McAndrew, editors. Tilapias: biology and ing, Vienna, Austria.
exploitation. Springer Netherlands, Dordrecht, The Pearson, D. E., Y. K. Ortega, J. Runyon, and J. L. Butler.
Netherlands. 2018. Secondary invasion re-redefined: the distinc-
Luiz, E. A., A. A. Agostinho, L. C. Gomes, and N. S. tion between invader-facilitated and invader-con-
Hahn. 1998. Ecologia tro fica de peixes em dois ria- tingent invasions as subclasses of secondary
chos da bacia do rio Parana. Revista Brasileira de invasion. Ecology and Evolution 8:5185–5187.
Biologia 58:273–285. Petesse, M. M. L., and M. Petrere. 2012. Tendency
Magurran, A. E., M. Dornelas, F. Moyes, N. J. Gotelli, towards homogenization in fish assemblages in the
and B. McGill. 2015. Rapid biotic homogenization cascade reservoir system of the Tiet^e river basin,
of marine fish assemblages. Nature Communica- Brazil. Ecological Engineering 48:109–116.
tions 6:8405. Pinnegar, J. K., M. T. Tomczak, and J. S. Link. 2014.
Matsuzaki, S. S., N. Usio, N. Takamura, and I. Washi- How to determine the likely indirect food-web con-
tani. 2009. Contrasting impacts of invasive engi- sequences of a newly introduced non-native spe-
neers on freshwater ecosystems: an experiment cies: a worked example. Ecological Modelling
and meta-analysis. Oecologia 158:673–686. 272:379–387.
Mayer, C. M., et al. 2014. The benthification of fresh- Post, D. M. 2002. Using stable isotopes to estimate
water lakes: exotic mussels turning ecosystems trophic position: models, methods, and assump-
upside down. Pages 575–586 in T. F. Nalepa and D. tions. Ecology 83:703–718.
W. Schloesser, editors. Quagga and zebra mussels: Preston, D. L., J. S. Henderson, and P. T. Johnson. 2012.
biology, impacts, and control. CRC Press, Boca Community ecology of invasions: direct and indi-
Raton, Florida, USA. rect effects of multiple invasive species on aquatic
McKinney, M. L., and J. L. Lockwood. 1999. Biotic communities. Ecology 93:1254–1261.
homogenization: a few winners replacing many Quesada, S. P., J. A. R. Paschoal, and F. G. R. Reyes.
losers in the next mass extinction. Trends in Ecol- 2013. Considerations on the aquaculture develop-
ogy and Evolution 14:450–453. ment and on the use of veterinary drugs: special
Mills, E. L., et al. 2003. Lake Ontario: food web issue for fluoroquinolones-A review. Journal of
dynamics in a changing ecosystem (1970–2000). Food Science 78:R1321–R1333.
Canadian Journal of Fisheries and Aquatic Sciences R Core Team. 2017. R: a language and environment for
60:471–490. statistical computing. R Foundation for Statistical
Natsumeda, T., N. Takamura, M. Nakagawa, Y. Computing, Vienna, Austria.
Kadono, T. Tanaka, and H. Mitsuhashi. 2015. Envi- Rahel, F. J. 2007. Biogeographic barriers, connectivity
ronmental and biotic characteristics to discriminate and homogenization of freshwater faunas: It’s a
farm ponds with and without exotic largemouth small world after all. Freshwater Biology 52:696–
bass and bluegill in western Japan. Limnology 710.
16:139–148. Reis, P. C. J., L. A. Martinelli, and F. A. R. Barbosa.
Odum, E. P. 1969. The strategy of ecosystem develop- 2017. Basal carbon sources and planktonic food
ment. Science 164:262–270. web in a tropical lake: an isotopic approach. Mar-
Olden, J. J. D., L. N. Poff, M. R. Douglas, M. E. Dou- ine and Freshwater Research 68:429–441.
glas, and K. D. Fausch. 2004. Ecological and evolu- Sanches, F. H. C., C. A. Miyai, T. M. CostaT^ a, R. A.
tionary consequences of biotic homogenization. Christofoletti, G. L. Volpato, R. E. Barreto, T. M.
Trends in Ecology and Evolution 19:18–24. Costa, R. A. Christofoletti, G. L. Volpato, and R. E.
O’Loughlin, L. S., and P. T. Green. 2017. Secondary Barreto. 2012. Aggressiveness overcomes body-size
invasion: when invasion success is contingent on effects in fights staged between invasive and native
other invaders altering the properties of recipient fish species with overlapping niches. PLoS ONE 7:
ecosystems. Ecology and Evolution 7:7628–7637. e29746.

❖ www.esajournals.org 19 June 2019 ❖ Volume 10(6) ❖ Article e02757

 21508925, 2019, 6, Downloaded from https://esajournals.onlinelibrary.wiley.com/doi/10.1002/ecs2.2757 by CAPES, Wiley Online Library on [28/02/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
BEZERRA ET AL.

Saul, W. C., and J. M. Jeschke. 2015. Eco-evolutionary Takamura, K. 2007. Performance as a fish predator of
experience in novel species interactions. Ecology largemouth bass [Micropterus salmoides (Lacepede)]
Letters 18:236–245. invading Japanese freshwaters: a review. Ecological
Shelton, J. M., J. A. Day, and C. L. Griffiths. 2008. Influ- Research 22:940–946.
ence of largemouth bass, Micropterus salmoides, on Trumpickas, J., N. E. Mandrak, and A. Ricciardi. 2011.
abundance and habitat selection of Cape galaxias, Nearshore fish assemblages associated with intro-
Galaxias zebratus, in a mountain stream in the Cape duced predatory fishes in lakes. Aquatic Conserva-
Floristic Region, South Africa. African Journal of tion Marine and Freshwater Ecosystems 21:338–
Aquatic Science 33:201–210. 347.
Simberloff, D., and J. R. S. Vitule. 2014. A call for an Twardochleb, L. A., and J. D. Olden. 2016. Non-native
end to calls for the end of invasion biology. Oikos Chinese mystery snail (Bellamya chinensis) supports
123:408–413. consumers in urban lake food webs. Ecosphere 7:
Simberloff, D., and B. Von Holle. 1999. Positive interac- e01293.
tions of nonindigenous species: Invasional melt- Vitule, J. R. S., V. M. Azevedo-Santos, V. S. Daga, D. P.
down? Biological Invasions 1:21–32. Lima-Ju  nior, A. L. B. Magalh~
aes, M. L. Orsi, F. M.
Starling, F., X. Lazzaro, C. Cavalcanti, and R. Moreira. Pelicice, and A. A. Agostinho. 2015. Brazil’s
2002. Contribution of omnivorous tilapia to eutroph- drought: Protec biodiversity. Science 347:1427–
ication of a shallow tropical reservoir: evidence from 1428.
a fish kill. Freshwater Biology 47:2443–2452. Weyl, O. L. F. 2008. Rapid invasion of a subtropical
Stein, L. Y., and M. G. Klotz. 2016. The nitrogen cycle. lake fishery in central Mozambique by Nile tilapia,
Current Biology 26:R94–R98. Oreochromis niloticus (Pisces: Cichlidae). Aquatic
Sun, W., S. Dong, X. Zhao, Z. Jie, H. Zhang, and L. Conservation Marine and Freshwater Ecosystems
Zhang. 2010. Effects of zooplankton refuge on the 18:839–851.
growth of tilapia (Oreochromis niloticus) and plank- Winemiller, K. O., et al. 2016. Balancing hydropower
ton dynamics in pond. Aquaculture International and biodiversity in the Amazon, Congo, and
18:647–655. Mekong. Science 351:128–129.

SUPPORTING INFORMATION
Additional Supporting Information may be found online at: http://onlinelibrary.wiley.com/doi/10.1002/ecs2.
2757/full

❖ www.esajournals.org 20 June 2019 ❖ Volume 10(6) ❖ Article e02757