Download as pdf or txt
Download as pdf or txt
You are on page 1of 14

Ecology Letters, (2020) doi: 10.1111/ele.

13664

REVIEW AND
SYNTHESIS Uncovering patterns of freshwater positive interactions using
meta-analysis: Identifying the roles of common participants,
invasive species and environmental context

Abstract
Lindsey K. Albertson, Michael Positive interactions are sensitive to human activities, necessitating synthetic approaches to eluci-
J. MacDonald, Benjamin B. date broad patterns and predict future changes if these interactions are altered or lost. General
Tumolo, Michelle A. Briggs, understanding of freshwater positive interactions has been far outpaced by knowledge of these
Zachary Maguire, Sierra Quinn, important relationships in terrestrial and marine ecosystems. We conducted a global meta-analysis
Jose A. Sanchez-Ruiz, Jaris Veneros to evaluate the magnitude of positive interactions across freshwater habitats. In 340 studies, we
and Laura A. Burkle found substantial positive effects, with facilitators increasing beneficiaries by, on average, 81%
across all taxa and response variables. Mollusks in particular were commonly studied as both
Department of Ecology, Montana facilitators and beneficiaries. Amphibians were one group benefiting the most from positive inter-
State University, P.O. Box 173460, actions, yet few studies investigated amphibians. Invasive facilitators had stronger positive effects
Bozeman, MT 59717, USA
on beneficiaries than non-invasive facilitators. We compared positive effects between high- and
low-stress conditions and found no difference in the magnitude of benefit in the subset of studies
*Correspondence: E-mail: lindsey. that manipulated stressors. Future areas of research include understudied facilitators and benefi-
albertson@montana.edu ciaries, the stress gradient hypothesis, patterns across space or time and the influence of declining
taxa whose elimination would jeopardise fragile positive interaction networks. Freshwater positive
interactions occur among a wide range of taxa, influence populations, communities and ecosystem
processes and deserve further exploration.

Keywords
Ecosystem engineering, facilitation, habitat modification, mutualism, resource subsidy.

Ecology Letters (2020)

modification, protection and transport (Jones et al., 1994;


INTRODUCTION
Bertness and Leonard, 1997; Subalusky and Post, 2019).
Biotic interactions fundamentally influence populations, com- Understanding the relative strength of positive interactions
munities and ecosystem processes. A growing need to under- across mechanisms, facilitator and beneficiary identity, inva-
stand biotic interactions stems from their potential to be sion status, space and time has been a central goal of many
altered by human activities, including climate warming (Pre- researchers in this field over the past several decades. For
ston et al., 2008; Montoya and Raffaelli, 2010; HilleRisLam- example, differences in mechanisms of habitat modification
bers et al., 2013). Additionally, biodiversity loss and invasive (e.g. organismal structures such as coral reefs) versus resource
species alter biotic interactions, leading to cascading influences modification (e.g. breakdown of leaf material by organisms
on important ecosystem functions (Estes and Duggins, 1995; such as aquatic invertebrates) help explain the relative
Mooney and Cleland, 2001; Angelini et al., 2011). Although strength of positive interactions across space and time (Wright
positive biotic interactions play key roles (Brontstein, 2015), and Jones, 2006; Hastings et al., 2007; Tumolo et al., 2020).
they have received far less synthetic attention than negative Additionally, positive interactions have been documented as
biotic interactions such as competition and predation (Menge critical components of many pairwise or network interactions
and Sutherland, 1987; Gurevitch et al., 2001; Chesson and that can even modify competitive or predatory relationships
Kuang, 2008). At the turn of the century, several reviews (Losey and Denno, 1998; Kay and Schemske, 2004; Gross,
highlighted the importance of positive interactions (Fig. 1; 2008; Thomsen et al., 2010). Organisms that facilitate others,
Stachowicz, 2001; Bruno et al., 2003), yet many questions and especially through the mechanism of habitat modification (i.e.
areas for improved understanding still remain, including ecosystem engineers), are increasingly used for restoration and
quantifying positive interaction magnitude and investigating can be disproportionally influential when they are lost from
patterns across taxa, life-history stage and habitat type (Bert- ecosystems or displaced by invasive species that disrupt net-
ness and Callaway, 1994; Callaway and Walker, 1997; Jones work interactions (Crooks, 2002; Crain and Bertness, 2006;
et al., 1997; Choler et al., 2001; Trautz et al., 2017). Halpern et al., 2007; Renzi et al., 2019).
Positive biotic interactions such as facilitation, mutualism It has been challenging to establish general patterns and
and commensalism can be supported by several direct and context-dependency of positive interactions (Maestre et al.,
indirect mechanisms, including habitat modification, resource 2006; Silliman and He, 2018), but progress in our

© 2020 John Wiley & Sons Ltd


2 L. K. Albertson et al. Review And Synthesis

Figure 1 Number of publications in the data set


documenting a freshwater positive interaction by year
with a LOESS smoothing fit line. Dashed vertical
lines at 1994 and 2003 indicate years when influential
positive interactions papers were published (Bertness
and Callaway, 1994; Bruno et al., 2003).

understanding in terrestrial and marine environments has far Uncovering patterns of positive interactions in freshwaters
outpaced that in freshwater ecosystems (Bertness and Call- could be particularly useful for shedding light on general pat-
away, 1994; Callaway et al., 2002; Morris et al., 2007). In ter- terns of biotic interaction related to stress and environmental
restrial and marine ecosystems, positive interactions are well change. Specifically, the stress gradient hypothesis predicts that
recognised to regulate ecosystem function and biodiversity positive interactions will be relatively stronger in environments
and have provided important evidence to test central concepts with larger environmental stress or consumer pressure (Peck-
in ecology (Schemske et al., 2009; Angelini et al., 2011; He arsky, 1983; Peckarsky et al., 1990; Zanetell and Peckarsky,
et al., 2013; Bulleri et al., 2016). For example, positive interac- 1996; Bertness and Leonard, 1997; Moore, 2006). The stress gra-
tions in soils explain patterns of biodiversity and dispersal of dient hypothesis has garnered mixed support, and almost
microbes from above to below ground (Wardle, 2006). In entirely from examination within terrestrial and marine ecosys-
marine ecosystems, positive interactions are related to tems (Callaway et al., 2002; Lortie and Callaway, 2006; Maestre
increases in biodiversity (Hacker and Gaines, 1997). Addition- et al., 2006; Holmgren and Scheffer, 2010; Michalet et al., 2014).
ally, positive interactions in both marine and terrestrial In extreme cases, the positive outcome may be overwhelmed by
ecosystems can explain niche space expansion, which ulti- stressful physical conditions to the point when biotic interactions
mately has implications for ecosystem functions and services are minimally relevant. Like terrestrial and marine ecosystems,
(Bulleri et al., 2016). A recent review highlighted myriad ways freshwater ecosystems are also stressful and show large daily and
that positive interactions occur in freshwater ecosystems seasonal variations in natural hydrologic properties, water tem-
across plant and animal taxa (Silknetter et al., 2020). For peratures and sediment transport regimes (Resh et al., 1988;
example crayfish feeding and burrowing activity facilitates Lytle and Poff, 2004; Magoulick, 2014). As such, they represent
benthic algal species and macroinvertebrate density, often in ideal ecosystems with which to explore this important ecological
sediment-polluted streams (Creed, 1994; Albertson and concept. Additionally, freshwaters are experiencing significant
Daniels, 2016). Beavers create habitats that facilitate inverte- increases in anthropogenic-driven stressors (Carpenter et al.,
brates and amphibians, increasing density and biomass of 1998; Strayer, 2010; Strayer and Dudgeon, 2010; Jenny et al.,
some species (Hood and Larson, 2014; Anderson et al., 2015). 2020). These stressors include, but are not limited to pollution,
Structure building invertebrates such as net-spinning caddis- altered hydrology and species invasion, providing an opportu-
flies facilitate other benthic macroinvertebrates by providing nity to consider how the type of stressor might mediate the
novel areas of refuge from fast water velocity (Cardinale importance of positive interactions (Silliman and He, 2018). Yet
et al., 2001; Tumolo et al., 2019). However, a formal, quanti- the stress gradient hypothesis’s application to freshwater ecosys-
tative evaluation of the types and magnitude of positive inter- tems remains largely untested and underrepresented (but see
actions is still lacking. As such, there are still many unknown Luo et al., 2010; Fugere et al., 2012; Peoples et al., 2015). The
aspects of these important biotic interactions particularly in evaluation of freshwater positive interactions is thus necessary
freshwater ecosystems (Holomuzki et al., 2010; Silknetter for continued and improved broad understanding of this foun-
et al., 2020). dational ecological concept.

© 2020 John Wiley & Sons Ltd


Review And Synthesis Freshwater positive interactions 3

Here we used meta-analysis to test the overarching hypothe- Web of Science by only including articles in the categories in
ses that freshwater positive interactions are variable in magni- Appendix 1. The search was conducted on February 3, 2020,
tude across facilitator or beneficiary taxon, freshwater habitat for all years available in the Web of Science Core Collection
type, mechanism of benefit, facultative or obligate nature of (1900–2020). In addition to the systematic search, we
the relationship and invasive species status. We also tested extracted all publications from a recent review of positive
whether experiments that manipulated positive interactions in interactions in freshwater (Silknetter et al., 2020) and a read-
the presence of a stress gradient showed stronger positive ing list of publications compiled in 2018 by a local working
interactions in a more stressful treatment. These comparisons group on freshwater positive interactions. In total, our initial
were motivated by classical work highlighting variation that search resulted in 2,943 publications. We further refined this
results from species identity, invasion status, type of facilita- count by reading publication titles, abstracts or full text to
tion and stress (Jones et al., 1994; Callaway et al., 2002; assess whether they met our inclusion criteria (see below),
Strayer, 2010), as well as a recent review describing positive which resulted in 106 publications with extractable data.
interactions in freshwater ecosystems (Silknetter et al., 2020). We developed a systematic protocol to decide whether pub-
Response variables in our study ranged from the individual to lications would be included in our analysis, as well as proto-
ecosystem level, including growth, reproduction, density, bio- cols to decide whether to include multiple data points in cases
diversity and primary production. We quantitatively evaluated when a single publication presented multiple studies docu-
106 peer-reviewed publications from across the globe that menting a positive interaction (Nakagawa et al., 2017). To be
investigated many different groups of organisms that partici- included, the study had to document a positive interaction in
pate in biotic interactions that are positive for at least one of a natural freshwater ecosystem, and at least one of the species
the participants (Figs 1 and 2). Our analysis provides synthe- in the interaction had to be an obligate freshwater species.
sised information regarding patterns of positive interactions in For example a positive interaction between a fish and a terres-
freshwaters and identifies current knowledge gaps and future trial plant was included, whereas an interaction between two
research needs. riparian plants was excluded, and a study occurring in an
agricultural holding pond or flooded agricultural field was
excluded. The study must have quantified a measurable
METHODS response or consequence of the positive interaction. For
example we did not include studies that simply described the
Study selection
presence of a symbiosis between two species. We included
We performed a systematic search of the literature for studies studies that obtained results from experimental or observa-
that documented a positive response on a beneficiary in the tional methods but excluded studies with results only from
presence of a facilitator in freshwater ecosystems. We con- simulation modelling. To be included, the methodological
ducted the search using the online database Web of Science approach must have had, at minimum, two treatments, one
Core Collection for articles containing the topics ((“aquatic” with the facilitator (treatment group) and one without the
OR “stream$” OR “river$” OR “freshwater$” OR “wetland$” facilitator (control group). If these conditions were met, stud-
OR “lake$” OR “pond$” OR “vernal”) AND (“positive inter- ies were included if we could extract an estimate for the mean,
act*” OR “mutualis*” OR “commensal*” OR “facilitation” sample size and uncertainty around the mean for both groups.
OR “symbios*” OR “habitat ameliorat*” OR “stress amelio- The facilitator must have had positive effects on the benefi-
rat*” OR “ecosystem engineer*” OR “stress gradient*”) NOT ciary for the study to be included. If a study seeking positive
“blood stream”). The wildcards ($, *) represent unknown effects found neutral or negative effects, those data were not
characters and allow the search to capture articles containing included. Because we only included positive outcomes, our
alternative forms of words. This search was further refined in data set is biased towards positive effect size values. This

Figure 2 World map showing individual study


locations, number of studies per country and
examples of freshwater facilitator species commonly
studied in particular countries. Studied species from
left to right include the following: Zebra mussels,
beavers, golden mussels, common reed, California
blackworm and pondweed.

© 2020 John Wiley & Sons Ltd


4 L. K. Albertson et al. Review And Synthesis

approach was necessary to address our questions regarding cases, we pooled the treatment groups to derive a single-
patterns of positive interactions when they occur (see funnel weighted estimate for the mean as:
plot in Appendix 1, Figure S1). Untransformed data values n1 T1 þ n2 T2
must have been reported to be included. And finally, when Xpooled ¼
n1 þ n2
the authors reported a positive interaction at multiple time
points, we extracted the value for the last time point when the where T1 and T2 are the mean effect sizes for the first and sec-
positive interaction occurred (Higgins and Green, 2011). ond treatment, respectively, and n1 and n2 are the sample
sizes, and SD as:
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 
u
Data collection uðnT  1Þσ2 þ ðnT  1Þσ2 þ nT1 nT2 X2 þ X2  2XT XT
t 1 T1 2 T2 nT1 þnT2 T1 T2 1 2
SDpooled ¼
From each publication, we extracted individual studies by nT1 þ nT2  1
collecting means, standard errors and sample sizes for treat-
where σ T1 and σ T2 are the standard deviations for the first
ment and control groups when a beneficiary responded pos-
and second treatments. If more than two treatments shared a
itively to a facilitator. The presence or absence of a
control we applied this equation sequentially (treatments 1
facilitator could occur by experimental manipulation with
and 2 first, then treatment 3, etc.; Higgins and Green, 2011).
enclosures or exclosures, or through natural studies where
While this method results in a slight loss of information and
the facilitator was not present at the sampling site. Addi-
statistical power, we felt it was appropriate considering the
tionally, we extracted descriptive co-variates to describe
variables we averaged across were not part of our research
when, were and how positive interactions are studied in
questions, thus we retained the pertinent information of a
freshwater. These co-variates included the (1) the geographi-
facilitator being present (Cheung, 2005; Scammacca et al.,
cal location of the study, (2) the response variable measured
2014). We then calculated the within-study pooled variance
in the study, (3) the invasion status of the facilitator (inva-
which was used to weight individual studies to minimise the
sive, non-invasive) as defined by the paper’s authors, (4) the
influence of studies with low statistical power during the anal-
mechanism behind the positive interaction (habitat modifica-
ysis (Hedges and Olkin, 1985). We calculated the within-study
tion, resource modification, cleaning symbiosis, etc.), (5) the
pooled sampling variance derived by Hedges et al., 1999 as:
freshwater ecosystem (river, lake, vernal pool, etc.), (6) the  2  2
methodological approach of the study (field, mesocosm, lab- σ σ
oratory study), (7) the nature of the relationship between varðlnRRÞ ¼ T 2 þ C 2

nT X nC X T C
facilitator and beneficiary (obligate, facultative) and (8) type
of interaction (direct, indirect, both). We extracted estimates We modelled patterns of positive interactions using a three-
for means, uncertainty and sample sizes directly when pre- level meta-analytic model using the ‘rma.mv’ function from
sented in the main text or tables. If data were presented in the R package metaphor (Viechtbauer, 2010). The three-level
a figure, we used the imaging processing program ImageJ model accounts for non-independence at both the publication
along with the Figure Calibration plugin to extract mean level (RefID in our data set), as well as the manipulation level
and uncertainty estimates (Schindelin et al., 2012). When within a study as an individual publication could have multi-
uncertainty estimates were presented using a metric other ple manipulative treatments (Konstantopoulos, 2011). This
than the standard error, we transformed the estimate to a model structure allows for heterogeneity in the true effect sizes
standard error using equations in Appendix 2. All publica- between publications, as well as between experimental manip-
tions from which data were extracted and used for analysis ulations within publications by adding a random intercept at
can be found in Appendix 3. both levels of the hierarchy. We nested an experimental
manipulation inside RefID to account for publications that
Data analysis reported multiple effect sizes and multiple manipulations using
the syntax (c. 1 | RefID / Manipulation). We fit separate mod-
We calculated log-response ratios (lnRR) for each study as els to investigate how the positive interaction varied across
the standardised effect size for our meta-analysis (Hedges each of the following moderator variables: facilitator type,
et al., 1999). We chose this effect size because it is unitless, beneficiary type, facilitator invasion status, response variable,
and thus allows for comparisons across the broad range of facilitation mechanism, ecosystem and study methodological
species, response variables and units used in the studies we approach. Effect size estimates were weighted by the inverse
analysed. We calculated lnRR for each study using: of the marginal variance covariance matrix estimated by that
  model (Viechtbauer, 2010).
XT
lnRR ¼ ln  : We also analysed whether the effect of a positive interaction
XC
on a beneficiary varied across a stress gradient to evaluate
The lnRR ultimately quantifies the log proportional change support for the stress gradient hypothesis. The data set
between the treatment and control group (Lajeunesse, 2011). included 13 studies that experimentally manipulated stress and
In some studies, experiments compared multiple treatments measured a beneficiary response. Each study paired a benefi-
with a shared control group (e.g. the effect of a facilitator ciary response at a low-stress level with a response at a high-
taxon on a beneficiary with two treatments varying in benefi- stress level. When a publication reported data from a study or
ciary density), which violates the assumption of independent studies that manipulated more than two stress levels, we anal-
effect sizes for our models (Nakagawa et al., 2017). In these ysed only the highest and lowest stress values by coding the

© 2020 John Wiley & Sons Ltd


Review And Synthesis Freshwater positive interactions 5

stress level variable as ‘H’ or ‘L’ respectively. Because we studies respectively. The least commonly studied beneficiaries
drew multiple studies from the same publication, we modelled were fungus and birds, with only two and one studies involv-
the influence of stress on the response of a beneficiary species ing these groups respectively. In 73% of the studies, the facili-
using an identical nesting structure to account for non-inde- tator was a non-invasive species (248 out of 340).
pendence of effect size estimates and weighting as presented Studies quantified positive interactions by investigating
previously. many different response variables, mechanisms, freshwater
We first fit models with an intercept term to obtain the ecosystems and methodological approaches (Fig. 3c, e, f and
overall omnibus test, which indicates whether part of the h). We found that density was the most common response
heterogeneity in the true effect size is related to some of the variable used to quantify a positive interaction, which was
moderators in the model. If the overall omnibus test for a used in 34% of studies. Biomass and diversity were measured
model indicated there were differences in the effect sizes less frequently, but still made up a substantial proportion of
between the moderator variable groups (at α = 0.05), we refit the data set at 16% and 11% respectively. The least common
the model without an intercept term to test for differences response variables measured were population growth (1.7% of
between the moderator variable levels. We used the ‘glht’ studies) and boldness (0.6%). Habitat modifications were the
function from the R package multcomp (Hothorn et al., 2008) dominant mechanism for positive interactions studied at 70%,
to conduct all post hoc tests with a significance level of followed by resource modifications (26%). Cleaning symbioses
α = 0.05, after a Bonferroni p-value adjustment to account were studied least frequently as a mechanism and made up
for the family wise error rate when conducting multiple com- just 1.2% of studies. Studies occurred in river and lake ecosys-
parisons. We report selected pairwise comparisons in the tems 89% of the time; however, we found that positive inter-
results, however all pairwise comparisons can be found in actions have also been documented in wetlands, carnivorous
Appendix 4. We displayed effect size estimates of the lnRR in plants, vernal pools and tree holes. Field studies were most
figures, but we back-transformed response-ratios to present frequently used to investigate positive interactions (55%), fol-
the in-text results, which represent the percentage change in a lowed by mesocosm experiments (26%) and laboratory experi-
response with a facilitator present compared to a response ments (19%).
with no facilitator present. We used this back transformation Positive interactions also varied in the nature of the rela-
to aid in the interpretation of our modelled lnRR effect sizes tionship between the facilitator and beneficiary and type of
(Pustejovsky, 2018). All analyses were conducted using the R interaction (Fig. 3b and d). An overwhelming majority of
Statistical Computing Environment, version 4.0.0 (R Core studies (99%) involved facultative relationships between the
Team 2020). facilitator and beneficiary rather than an obligate relationship.
The three studies investigating obligate relationships involved
a cleaning symbiosis between a crayfish and an annelid worm.
RESULTS
Direct positive interactions were most common (227 studies),
We found that freshwater positive interaction papers have followed by indirect interactions (73 studies), and a combina-
been published consistently since the mid 1980s but increased tion of both direct and indirect (40 studies).
in the early 1990s following on the heels of seminal papers, We found minimal evidence that the facilitator group influ-
with the largest increase in publication rate around the year enced the strength of a positive interaction, and the omnibus
2000 (Fig. 1). Research occurred in 26 Countries, ranging test was not significant (QM12 = 20.73, P = 0.06; Fig. 4a).
from 117 studies occurring in the United States to one study However, the effect size for a positive interaction depended
occurring in both Panama and Spain (Fig. 2). For the 26 on the beneficiary group and the facilitator species’ invasion
countries where studies occurred, the per capita study count status (Fig. 4b and c; see Appendix 3 for all pairwise compar-
was, on average, 13.1 studies country-1. While 65 studies isons). The strength of a positive interaction was stronger for
occurred in Hungary, these were extracted from one publica- some beneficiary taxonomic groups compared to others
tion that reported effect sizes for multiple taxa across four dif- (QM12 = 38.78, P = 0.0001; Table S2). Across all responses
ferent facilitator groups (Bodis et al., 2014). Examples of measured, mollusks experienced a 36.87% greater positive
facilitators studied across the globe included beavers (Castor response compared to insects (95% CI [9.77%, 55.83%]), but
sp.), zebra mussels (Dreissena polymorpha), golden mussels no other beneficiary taxonomic groups differed in their posi-
(Limnoperna fortune), common reed (Phragmites australis), tive response. The strength of positive interactions also varied
pondweed (Potamogeton malaianus) and California blackworm based on the invasion status of the facilitator species (QM1 =
(Lumbriculus variegatus). Notably, no studies occurred in 8.69, P = 0.0032; Table S3). The strength of a positive inter-
Africa, Southeast Asia, the Middle East or Siberia. action was 25.85% smaller when a non-invasive group was
We found that studies investigated a wide range of taxo- the facilitator compared to an invasive group (95% CI
nomic groups as the facilitator or beneficiary in a positive [−39.22%, −9.54%]).
interaction and facilitators were both invasive and non-inva- The strength of a positive interaction also differed based on
sive (Fig. 3a and g). The most commonly studied freshwater the response variable measured (QM11 = 194.93, P < 0.0001;
facilitators were mollusks and plants with 143 studies and 55 Fig. 4D; Table S4). Compared to the two most frequently
studies respectively. The least commonly studied freshwater used response variables (density, biomass), we found little evi-
facilitators were periphyton and fungus with one study each. dence that positive interactions influenced diversity, which
For beneficiaries of positive interactions, we found that insects only increased by an estimated 10.86% (P = 0.1657). Compar-
and plants were the most studied occurring in 119 and 66 atively, a density response was 85.92% stronger (95% CI

© 2020 John Wiley & Sons Ltd


6 L. K. Albertson et al. Review And Synthesis

Figure 3 Lollipop plots displaying the number of studies across moderator variables. (a) Facilitator and beneficiary taxonomic groups; (b) Interaction type;
(c) Freshwater ecosystem where positive interaction occurred; (d) Nature of the species’ relationship; (e) Response variable used to measure a positive
interaction; (f) Mechanism supporting the positive interaction; (g) Invasion status of the facilitator and (h) Study methodological approach used to
investigate positive interactions.

[54.02%, 116.44%]) and a biomass response was 98.67% We found no evidence that positive interaction strengths are
stronger (95% CI [57.53%, 150.56%]) than a diversity modified by stress (QM1 = 0.66, P = 0.4169; Fig. 5). In this
response. Positive interaction magnitudes also varied based on subset of studies, consumer pressure and nutrients were the
the mechanism underpinning the interaction (QM3 = 8.80, stressors evaluated (Fig. 5). Compared to a control without
P = 0.0320; Fig. 4E; Table S5), as well as the study method- the facilitator, a positive response was 98% greater in the
ological approach (QM2 = 11.59, P = 0.0030; Fig. 4G). Habi- presence of a facilitator in high-stress conditions (P = 0.0902)
tat modifications were associated with a 33.67% greater and 118% greater in a low-stress condition (P = 0.0516), but
positive effect than those associated with a resource modifica- high and low-stress levels did not differ from one another.
tion (95% CI [3.85%, 72.05%]). No differences across ecosys-
tem types were detected (Fig. 4f; Table S6). Effect sizes for
DISCUSSION
positive interactions measured in the field were 44.06% stron-
ger than those measured in a laboratory environment (95% Positive interactions are important in natural ecosystems; how-
CI [10.08%, 88.52%]). Mesocosm studies were not different ever, we know far less about their prevalence, magnitude and
than field studies or laboratory studies (Table S7), and inter- responses to invasive species and stressors in freshwater ecosys-
action strengths did not vary based on the type of interaction tems compared to others. Our data set was comprised of positive
(Fig. 4h; Table S9) or the nature of the relationship between interaction papers that have been published consistently since
facilitator and beneficiary (Fig. 4i; Table S8). the 1980s. We observed a gradual increase in paper numbers

© 2020 John Wiley & Sons Ltd


Review And Synthesis Freshwater positive interactions 7

Figure 4 lnRR effect sizes for positive interaction strengths across moderator variables. (a) Facilitator taxonomic groups; (b) Beneficiary taxonomic groups;
(c) Invasion status of the facilitator; (d) Response variable used to measure a positive interaction; (e) Mechanism supporting the positive interaction; (f)
Freshwater ecosystem where positive interaction occurred; (g) Study methodological approach; (h) Interaction type and (i) Nature of species’ relationship.
P-value for the overall omnibus test is shown in the bottom right corner of each panel. Error bars are 95% confidence intervals. Red dashed line at
lnRR = 0 shows location of no effect.

through the mid-1990s, with the largest rate of increase in the invertebrates. Amphibian and biofilm beneficiaries showed the
early 2000s, a timeline that aligns well with the publication of largest magnitude positive responses, but these were not different
several seminal research papers and reviews highlighting the from one another or from other groups. These findings suggest
need for but understudied value of positive interactions in ecol- positive interactions are taxon-specific, but prevalent across
ogy (Bertness and Callaway, 1994; Jones et al., 1994; Stachowicz, kingdoms. It was encouraging to see that field studies showed
2001; Bruno et al., 2003). Rivers and lakes comprised the vast significant positive responses overall, suggesting that freshwater
majority of habitats studied in our data set, but no significant dif- ecologists are successfully detecting and quantifying positive
ferences were detected in the strength of positive interactions interactions in natural settings. We found no support for stron-
across any habitats, suggesting these biotic interactions are ubiq- ger effects of positive interactions in high stress treatments in the
uitously important. Freshwater positive interactions influenced few experiments that have tested a beneficiary response to the
beneficiaries in many ways, including increasing density, coloni- presence of a facilitator at various stressor levels. Our findings
sation rates, reproduction and survival. However, positive inter- contribute a quantitative evaluation of the scope of positive
actions had relatively small impacts on biodiversity, population interactions in freshwater ecosystems and generally concur with
growth and boldness. Non-invasive facilitators were more com- a recent review highlighting positive interaction research across
monly studied, but invasive facilitators had significantly stronger different freshwater facilitator and beneficiary taxa (Silknetter
positive effects. Mollusks benefited significantly more than et al., 2020).

© 2020 John Wiley & Sons Ltd


8 L. K. Albertson et al. Review And Synthesis

Figure 5 Weighted lnRR effect sizes and 95% confidence intervals for 13 studies that recorded a positive response in the presence of a facilitator in both
high (dark blue) and low stress (light blue) conditions. Effect sizes are weighted by the inverse of the marginal variance covariance matrix estimated by the
model and displayed by varying the size of the estimated mean lnRR point. Group means plotted as diamonds with width of diamond representing 95%
confidence interval. Red dashed line at lnRR = 0 shows location of no effect.

Our data set identified a total of 14 studied taxonomic Aquatic insects were the most studied beneficiary, a finding
groups as either facilitators or beneficiaries across a range of that is supported by decades of research using this particular
trophic positions and evolutionary histories. Mollusks were by group to investigate general patterns of disturbance, feeding
far the most commonly studied facilitators. Their sensitivity relationships, organic matter processing and biomonitoring in
to anthropogenic change and status as prolific invaders per- freshwater ecosystems (Karr, 1991; Wallace and Webster,
haps has driven the extensive interest in this particular group 1996; Covich et al., 1999). For example mayflies respond posi-
(Ram et al., 1992; Lydeard et al., 2004; Duchini et al., 2018). tively to habitat and predator refuge provided by mussels in
Mollusks often facilitate submerged macrophytes by reducing soft-sediment, lentic ecosystems (DeVanna et al., 2011). Lam-
algal blooms that might otherwise hinder macrophyte growth preys building nests alter physical substrate conditions and
by limiting light availability (He et al., 2014). They also facili- thereby increase density of benthic invertebrates (Hogg et al.,
tate macroinvertebrates by providing predator refugia and 2014). Mollusks were also commonly studied as beneficiaries.
increasing survival (Kobak et al., 2014). Other common facili- Production, colonisation and reproduction of beneficiaries all
tators were plants and insects. Plants and aquatic insects mod- showed the largest magnitude effects but were not studied as
ify freshwater physical processes such as erosion and sediment frequently as other response variables such as density and bio-
deposition in ways that create unique habitats or alter mass, suggesting there is room to understand these responses
resource availability for beneficiaries (Dodds, 1991; Fritz further. Although many studies show that positive interactions
et al., 2004; Reinhardt et al., 2010; Albertson and Allen, enhance biodiversity (Hacker and Gaines, 1997; Mulder et al.,
2015). Periphyton had the strongest facilitative effect but was 2001), our findings revealed that the magnitude of positive
studied much less frequently than crustaceans, fish, insects, interactions on diversity in freshwater is small. Perhaps posi-
plants and mollusks and the strong effect was demonstrated tive interactions in freshwater are realised more commonly
by a single study (Medina-Sanchez et al., 2004), highlighting through pairwise interactions specifically, and as such, no or
the need to further investigate this group’s status as a facilita- little effect on diversity per se is to be expected. That is, some
tor. species benefit while others do not, resulting in minimal
impacts on community diversity as a whole, a topic that has

© 2020 John Wiley & Sons Ltd


Review And Synthesis Freshwater positive interactions 9

received attention in marine and terrestrial ecosystems but substantial positive benefits to some species, it may be difficult
needs additional evaluation in freshwaters given our findings for managers to evaluate how and when to remove invasive
(Hoeksema and Bruna, 2000; Stanton, 2003). Alternatively, species if they confer positive effects for some but not all taxa.
research on freshwater positive interactions may be at a stage Invasive species that have positive effects that increase a bene-
in which researchers have yet to utilise or develop accurate ficiary may also not ultimately equate to desirable ecological
approaches for measuring community-level responses for outcomes when they increase the abundance or biomass of
detecting relationships between positive interactions and bio- beneficiaries beyond natural levels, drastically alter food web
diversity (Silknetter et al., 2020). dynamics and nutrient cycling, or even cause invasional melt-
The mechanism of positive interaction influenced how down, when synergistic interactions among invaders lead to
strongly the positive benefit was conferred, and habitat modi- accelerated impacts on native ecosystems (Simberloff and Von
fication more strongly benefited participants than resource Holle, 1999; Hall et al., 2006; Braga et al., 2018). Thus,
modification did. In freshwater research, ecosystem engineer- understanding the role and management of invasive facilita-
ing terminology has been utilised extensively to emphasise the tors poses an important contemporary ecological challenge.
strong links between organisms and physical habitat modifica- Despite numerous tests of the stress gradient hypothesis in
tion (Moore, 2006; Jones, 2012; Atkinson et al., 2018). This terrestrial and marine systems (He et al., 2013), we found only
link was strongly supported by our finding that habitat modi- a small number of quantitative experimental tests related to
fication was identified by researchers as the mechanism of stress and positive interactions in freshwaters. Our data set
positive interaction almost four times as often as resource included 13 studies that manipulated stress in an experimental
modification. However, many studies suggested that both setting, and these studies showed, on average, no difference in
habitat modification and resource modification were occur- positive interaction effect size between high- and low-stress
ring. In several cases, positive interactions were experienced treatments. Many of these studies did not explicitly aim to test
when one predator increased prey amount or ease of capture the stress gradient hypothesis, which predicts that the fre-
for another predator (Soluk and Richardson, 1997; Kelly quency or relative importance of positive interactions increase
et al., 2002). Along with habitat and resource modifications, with stress (Peckarsky et al., 1990; Bertness and Callaway,
cleaning symbioses are important mechanisms of freshwater 1994). Instead, studies in our data set investigated the magni-
positive interactions. Cleaning symbioses highlight the con- tude or strength of positive effect in high stress compared to
text-dependency of positive interactions. Crayfish, for example low-stress treatments (Bruno et al., 2003). Given that we did
have worms that benefit from eating parasites off of their gill not evaluate whether the frequency of positive interactions
structures, with certain crayfish species or life-stages benefiting increases as stress increases, our findings do not exhaustively
more than others (Thomas et al., 2016). The magnitude of indicate that the stress gradient hypothesis is not supported in
positive outcomes from cleaning symbioses was small overall, freshwater ecosystems. Rather, they highlight the need for
and future work might evaluate how they compare to well- much more work on this topic and the potential for future
recognised cleaning relationships in marine systems (Brown studies to evaluate outcomes when various stressors or various
et al., 2002; Schemske et al., 2009). Overall, a variety of mech- responses of positive effects are measured (He et al., 2013).
anisms were studied, but habitat modification was the most For example, multiple stressors are common in freshwater
common and strongest positive interaction. ecosystems (Heugens et al., 2001; Ormerod et al., 2010) and
Invasive facilitators had a stronger positive influence than can show antagonistic rather than synergistic influence. Per-
non-invasive facilitators. Perhaps, invasive facilitators occur in haps, this outcome dampens the positive influence of facilita-
high densities, have traits that lead to disproportionately large tors under stressful conditions. Furthermore, two types of
positive effects, or play a strong role in already-degraded, stressors included in the original stress gradient hypothesis,
stressed ecosystems where beneficiaries have more opportunity consumer pressure and physical stress, more often than not
to respond positively (MacDougall and Turkington, 2005; vary directly and not inversely, suggesting that coupled stres-
Ramus et al., 2017). A more explicit identification of invasive ses will increase at a faster rate and the importance of positive
facilitator traits may inform species distribution models that interactions will be further pertinent (Silliman and He, 2018).
account for positive biotic interactions (Maestre et al., 2009; Stress can be realised as both natural variation in abiotic con-
Kissling et al., 2012; Tikhonov et al., 2017; Wagner et al., ditions such as flow or predatory conditions such as high den-
2020). These traits may be particularly tied to habitat modifi- sities of enemies, as well as anthropogenic variation caused by
cation, the mechanism showing the strongest positive response damming, pollution and others (Silknetter et al., 2020). Defin-
in our data set (Crooks, 2002). Although a large portion of ing how stress is experienced by beneficiaries, as well as the
research on invasive species highlights their negative impacts thresholds where facilitation alleviates stress, has been a major
(Vitousek et al., 1997; Strayer, 2010; Rogosch and Olden, criticism and challenge for the stress gradient hypothesis
2020), the role of invasive species is complex, resulting in (Maestre et al., 2009; He and Bertness, 2014). Given the
simultaneous positive and negative influences when some spe- increasing degradation of water quantity and quality stem-
cies benefit but others do not (Rodriguez, 2006; Collins et al., ming from anthropogenic stressors, this topic deserves sub-
2017). For example invasive zebra mussels positively influence stantially further exploration (Vörösmarty et al., 2010; Bunn,
benthic primary producers by filtering water and improving 2016). Human activities are amplifying the number and mag-
clarity and their shells provide novel habitat to mayflies, nitude of stressors in natural ecosystems, making it increas-
whose populations increase in abundance (Cecala et al., 2008; ingly important to be able to define stress across
DeVanna et al., 2011). Because invasive facilitators do show environmental context and understand biotic interactions in

© 2020 John Wiley & Sons Ltd


10 L. K. Albertson et al. Review And Synthesis

the face of global change (Hammarlund and Harcombe, 2019; In conclusion, we used meta-analysis to elucidate patterns
Piccardi et al., 2019). of freshwater positive interactions and found that they are
Although an increase in the number of studies investigating widespread and participated in by many different taxonomic
positive interactions in freshwater jumped through the 1990s, groups. The similar strength of positive interaction across
perhaps aligning with a more general trend in ecology on the freshwater habitats was surprising given the large variability
heels of high-profile papers, there are still several important in physical constraints and organismal adaptations from those
areas for future research. Our literature search resulted in an different habitat types. Rare species or species of concern such
extensive 2,943 papers on the topic of positive interactions in as amphibians and mollusks may be most susceptible to alter-
freshwater ecosystems, but relatively few of these papers ulti- ation of positive interaction networks under global change
mately met the quantitative criteria of reporting a mean, error (Correa et al., 2015a). Critical areas for future research are to
and sample size required to be included in the meta-analysis, further investigate the role of positive interactions compared
highlighting the need for researchers to more thoroughly to negative interactions, the strength of positive interactions
report their experimental designs and results. A recent review in freshwater compared to terrestrial and marine, and to bet-
by Silknetter et al. (2020) outlined areas for future research in ter identify stressors and how they modify the strength of pos-
freshwater positive interactions, and the findings from our itive interactions. Our findings provide the first quantitative
data compilation align with these directives. For example, rel- synthesis of positive interactions in freshwaters, highlight the
atively few studies conducted work that measured responses ubiquity of freshwater positive interactions and identify
over space or time. Limited spatial and temporal measure- important research areas for future knowledge generation.
ments suggest that ecologists still cannot address important
questions of positive interactions in ecology because we are
ACKNOWLEDGEMENTS
not reporting work across scales (Jablonski, 2008; McGill,
2010). Negative interactions have seen important progress in We thank L. Calle, W. Cross and E. Scholl for discussion and
this area (McLaughlin and Roughgarden, 1992; Fausch, 1998; constructive feedback during manuscript development. Three
Chizzola et al., 2018; Weiss, 2019), but this topic will be an reviewers provided comments that greatly improved the
important future step for freshwaters, especially as spatial and manuscript. Funding was provided for LKA, MJM and BBT
temporal variation relates to stress gradients (He et al., 2013; in part by the National Science Foundation (DEB 1556684)
Tumolo et al., 2020). Effects of facilitators on populations and Montana State University. Funding for MAB was pro-
and communities is another important topic for future vided in part by The National Park Service (P17AC01089).
research. A species niche may be expanded by positive interac-
tions, but how these relationships vary across time or space
AUTHORSHIP
remains an active area of research (Chase and Myers, 2011;
Bulleri et al., 2016). Whether positive interactions maintain LKA conceived the original idea; All authors designed the
rare species in freshwater communities has not yet been inves- research and collected the data; MJM analysed the data;
tigated (Calatayud et al., 2020), but our findings suggest that LKA and MJM wrote the original paper draft; All authors
sensitive and increasingly rare but important groups such as contributed substantially to revisions.
amphibians, fish and mollusks can benefit substantially from
positive interactions (Pintor and Soluk, 2006; Whiles et al.,
PEER REVIEW
2006; Peoples et al., 2011). The vast majority of positive inter-
actions included in our data set were direct interactions, but The peer review history for this article is available at https://
indirect or some combination of the two interactions had just publons.com/publon/10.1111/ele.13664.
as large of an influence on beneficiaries, perhaps suggesting
that indirect positive interaction cascades that influence com-
DATA AVAILABILITY STATEMENT
munity-level metrics such as biodiversity are a possible new
frontier in freshwater science (Thomsen et al., 2010). And The data supporting the results are archived in Dryad
finally, the relative importance of positive interactions com- (https://doi.org/10.5061/dryad.zcrjdfn9b).
pared to negative interactions has not yet been established for
freshwater ecosystems. Although several reviews have high-
REFERENCES
lighted progress in our understanding of biotic interactions,
the relative importance of both negative and positive interac- Albertson, L.K. & Allen, D.C. (2015). Meta-analysis: Abundance,
tions and especially their context-dependency has yet to be behavior, and hydraulic energy shape biotic effects on sediment
transport in streams. Ecology, 96, 1329–1339.
identified (Thorp, 1986; Soluk and Collins, 1988; Mermillod-
Albertson, L.K. & Daniels, M.D. (2016). Effects of invasive crayfish on
Blondin and Rosenberg, 2006; Holomuzki et al., 2010; Correa fine sediment accumulation, gravel movement, and macroinvertebrate
et al., 2015b). This topic has provided some important communities. Freshw. Sci., 35, 644–653.
insights in terrestrial and marine ecosystems (Bertness et al., Anderson, N.L., Paszkowski, C.A. & Hood, G.A. (2015). Linking aquatic
1999; Stewart et al., 1999; Hay et al., 2004; Duchini et al., and terrestrial environments: can beaver canals serve as movement
2018; Prugh and Sivy, 2020). As such, an evaluation of the corridors for pond-breeding amphibians? Anim. Conserv., 18, 287–294.
relative magnitude of positive interactions in freshwater habi- Angelini, C., Altieri, A.H., Silliman, B.R., Bertness, M.D. (2011).
Interactions among foundation species and their consequences for
tats compared to negative interactions is an important next
step.

© 2020 John Wiley & Sons Ltd


Review And Synthesis Freshwater positive interactions 11

community organization, biodiversity, and conservation. Bioscience, 61, Collins, S.F., Nelson, K.A., DeBoom, C.S. & Wahl, D.H. (2017). The
782–789. facilitation of the native bluegill sunfish by the invasive bighead carp.
Atkinson, C.L., Allen, D.C., Davis, L. & Nickerson, Z.L. (2018). Freshw. Biol., 62, 1645–1654.
Incorporating ecogeomorphic feedbacks to better understand resiliency Correa, S.B., Araujo, J.K., Penha, J.M.F., da Cunha, C.N., Stevenson,
in streams: A review and directions forward. Geomorphology, 305, P.R. & Anderson, J.T. (2015a). Overfishing disrupts an ancient
123–140. mutualism between frugivorous fishes and plants in Neotropical
Bertness, M.D. & Callaway, R. (1994). Positive interactions in wetlands. Biol. Conserv., 191, 159–167.
communities. Trends Ecol. Evol., 9, 191–193. Correa, S.B., Costa-Pereira, R., Fleming, T., Goulding, M. & Anderson,
Bertness, M.D. & Leonard, G.H. (1997). The role of positive interactions J.T. (2015b). Neotropical fish–fruit interactions: eco-evolutionary
in communities: Lessons from intertidal habitats. Ecology, 78, 1976. dynamics and conservation. Biol. Rev., 90, 1263–1278.
Bertness, M.D., Leonard, G.H., Levine, J.M., Schmidt, P.R. & Ingraham, Covich, A.P., Palmer, M.A. & Crowl, T.A. (1999). The role of benthic
A.O. (1999). Testing the relative contribution of positive and negative invertebrate species in freshwater ecosystems. Bioscience, 49, 119.
interactions in rocky intertidal communities. Ecology, 80, 2711–2726. Crain, C.M. & Bertness, M.D. (2006). Ecosystem engineering across
Bodis, E., Toth, B., Szekeres, J., Borza, P. & Sousa, R. (2014). Empty environmental gradients: implications for conservation and
native and invasive bivalve shells as benthic habitat modifiers in a large management. Bioscience, 56, 211–218.
river. Limnologica, 49, 1–9. Creed, R.P. (1994). Direct and indirect effects of crayfish grazing in a
Braga, R.R., Gómez-Aparicio, L., Heger, T., Vitule, J.R.S. & Jeschke, stream community. Ecology, 75, 2091–2103.
J.M. (2018). Structuring evidence for invasional meltdown: broad Crooks, J.A. (2002). Characterizing ecosystem-level consequences of
support but with biases and gaps. Biol. Invasions, 20, 923–936. biological invasions: the role of ecosystem engineers. Oikos, 97,
Brontstein, J.L. (2015). Mutualism. Oxford, United Kingdom: Oxford 153–166.
University Press. DeVanna, K.M., Armenio, P.M., Barrett, C.A. & Mayer, C.M. (2011).
Brown, B.L., Creed, R.P. & Dobson, W.E. (2002). Branchiobdellid Invasive ecosystem engineers on soft sediment change the habitat
annelids and their crayfish hosts: are they engaged in a cleaning preferences of native mayflies and their availability to predators.
symbiosis? Oecologia, 132, 250–255. Freshw. Biol., 56, 2448–2458.
Bruno, J., Stachowicz, J.J. & Bertness, M.D. (2003). Inclusion of Dodds, W.K. (1991). Community interactions between the filamentous
facilitation into ecological theory. Trends Ecol. Evol., 18, 119–125. alga Cladophora glomerata (L.) Kuetzing, its epiphytes, and epiphyte
Bulleri, F., Bruno, J.F., Silliman, B.R. & Stachowicz, J.J. (2016). grazers. Oecologia, 85, 572–580.
Facilitation and the niche: implications for coexistence, range shifts and Duchini, D., Boltovskoy, D. & Sylvester, F. (2018). The invasive
ecosystem functioning. Funct. Ecol., 30, 70–78. freshwater bivalve Limnoperna fortunei in South America: multiannual
Bunn, S.E. (2016). Grand challenge for the future of freshwater changes in its predation and effects on associated benthic invertebrates.
ecosystems. Front. Environ. Sci., 4, 21. Hydrobiologia, 817, 431–446.
Calatayud, J., Andivia, E., Escudero, A., Melián, C.J., Bernardo-Madrid, Estes, J.A. & Duggins, D.O. (1995). Sea otters and kelp forests in Alaska:
R., Stoffel, M. et al. (2020). Positive associations among rare species Generality and variation in a community ecological paradigm. Ecol.
and their persistence in ecological assemblages. Nat. Ecol. Evol., 4, Monogr., 65, 75–100.
40–45. Fausch, K.D. (1998). Interspecific competition and juvenile Atlantic
Callaway, R.M., Brooker, R.W., Choler, P., Kikvidze, Z., Lortie, C.J., salmon ( Salmo salar ): on testing effects and evaluating the evidence
Michalet, R. et al. (2002). Positive interactions among alpine plants across scales. Can. J. Fish. Aquat. Sci., 55, 218–231.
increase with stress. Nature, 417, 844–848. Fritz, K.M., Gangloff, M.M. & Feminella, J.W. (2004). Habitat
Callaway, R.M. & Walker, L.R. (1997). Competition and facilitation: A modification by the stream macrophyte Justicia americana and its
synthetic approach to interactions in plant communities. Ecology, 78, effects on biota. Oecologia, 140, 388–397.
1958. Fugere, V., Andino, P., Espinosa, R., Anthelme, F., Jacobsen, D. &
Cardinale, B.J., Smith, C.M. & Palmer, M.A. (2001). The influence of Dangles, O. (2012). Testing the stress-gradient hypothesis with aquatic
initial colonization by hydropsychid caddisfly larvae on the detritivorous invertebrates: insights for biodiversity-ecosystem
development of stream invertebrate assemblages. Hydrobiologia, 455, functioning research. J. Anim. Ecol., 81, 1259–1267.
19–27. Gross, K. (2008). Positive interactions among competitors can produce
Carpenter, S.R., Caraco, N.F., Correll, D.L., Howarth, R.W., Sharpley, species-rich communities. Ecol. Lett., 11, 929–936.
A.N. & Smith, V.H. (1998). Nonpoint pollution of surface waters with Gurevitch, J., Curtis, P.S. & Jones, M.H. (2001). Meta-analysis in
phosphorus and nitrogen. Ecol. Appl., 8, 559–568. ecology. Adv. Ecol. Res., 32, 199–247.
Cecala, R.K., Mayer, C.M., Schulz, K.L. & Mills, E.L. (2008). Increased Hacker, S.D. & Gaines, S.D. (1997). Some implications of direct positive
Benthic Algal Primary Production in Response to the Invasive Zebra interactions for community species diversity. Ecology, 78(7), 1990–2003.
Mussel (Dreissena polymorpha) in a Productive Ecosystem, Oneida Hall, R.O., Dybdahl, M.F. & VanderLoop, M.C. (2006). Extremely high
Lake. New York. J. Integr. Plant Biol., 50, 1452–1466. secondary production of introduced snails in rivers. Ecol. Appl., 16,
Chase, J.M. & Myers, J.A. (2011). Disentangling the importance of 1121–1131.
ecological niches from stochastic processes across scales. Philos. Trans. Halpern, B.S., Silliman, B.R., Olden, J.D., Bruno, J.P. & Bertness, M.D.
R. Soc. B Biol. Sci., 366, 2351–2363. (2007). Incorporating positive interactions in aquatic restoration and
Chesson, P. & Kuang, J.J. (2008). The interaction between predation and conservation. Front. Ecol. Environ., 5, 153–160.
competition. Nature, 456, 235–238. Hammarlund, S.P. & Harcombe, W.R. (2019). Refining the stress
Cheung, M.W.L. (2005). Meta-analysis: A structural equation modeling gradient hypothesis in a microbial community. Proc. Natl. Acad. Sci.,
approach. New York, NY. Wiley. 116, 15760–15762.
Chizzola, M., Belton, L., Ganswindt, A., Greco, I., Hall, G., Swanepoel, Hastings, A., Byers, J.E., Crooks, J.A., Cuddington, K., Jones, C.G.,
L. et al. (2018). Landscape level effects of lion presence (Panthera leo) Lambrinos, J.G. et al. (2007). Ecosystem engineering in space and time.
on two contrasting prey species. Front. Ecol. Evol., 6, 191. Ecol. Lett., 10, 153–164.
Choler, P., Michalet, R. & Callaway, R.M. (2001). Facilitation and Hay, M.E., Parker, J.D., Burkepile, D.E., Caudill, C.C., Wilson, A.E.,
competition on gradients in alpine plant communities. Ecology, 82, Hallinan, Z.P. et al. (2004). Mutualisms and aquatic community
3295–3308. structure: The enemy of my enemy is my friend. Annu. Rev. Ecol. Evol.
Syst., 35(1), 175–197.

© 2020 John Wiley & Sons Ltd


12 L. K. Albertson et al. Review And Synthesis

He, H., Liu, X.B., Liu, X.L., Yu, J.L., Li, K.Y., Guan, B.H. et al. (2014). Kobak, J., Jermacz, L. & Plachocki, D. (2014). Effectiveness of zebra
Effects of cyanobacterial blooms on submerged macrophytes alleviated mussels to act as shelters from fish predators differs between native and
by the native Chinese bivalve Hyriopsis cumingii: A mesocosm invasive amphipod prey. Aquat. Ecol., 48, 397–408.
experiment study. Ecol. Eng., 71, 363–367. Konstantopoulos, S. (2011). Mixed effects and variance components
He, Q. & Bertness, M.D. (2014). Extreme stresses, niches, and positive estimation in three-level meta-analysis. Res. Synth. Methods, 2, 61–76.
species interactions along stress gradients. Ecology, 95, 1437–1443. Lajeunesse, M.J. (2011). On the meta-analysis of response ratios for
He, Q., Bertness, M.D. & Altieri, A.H. (2013). Global shifts towards studies with correlated and multi-group designs. Ecology, 92,
positive species interactions with increasing environmental stress. Ecol. 2049–2055.
Lett., 16, 695–706. Lortie, C.J. & Callaway, R.M. (2006). Re-analysis of meta-analysis:
Hedges, L.V., Gurevitch, J. & Curtis, P.S. (1999). The meta-analysis of support for the stress-gradient hypothesis. J. Ecol., 94, 7–16.
response ratios in experimental ecology. Ecology, 80, 1150–1156. Losey, J.E. & Denno, R.F. (1998). Positive predator-predator
Hedges, L.V. & Olkin, I. (1985). Statistical Methods for Meta-Analysis. interactions: Enhanced predation rates and synergistic suppression of
Academic Press, San Diego, CA. aphid populations. Ecology, 79, 2143–2152.
Heugens, E., Hendriks, A., Dekker, T., van Straalen, N. & Admiraal, W. Luo, W.B., Xie, Y.H., Chen, X.S., Li, F. & Qin, X.Y. (2010).
(2001). A review of the effects of multiple stressors on aquatic Competition and facilitation in three marsh plants in response to a
organisms and analysis of uncertainty factors for use in risk water-level gradient. Wetlands, 30, 525–530.
assessment. Crit. Rev. Toxicol., 31, 247–284. Lydeard, C., Cowie, R.H., Ponder, W.F., Bogan, A.E., Bouchet, P.,
Higgins, J. & Green, S. (2011). Cochrane Handbook for Systematic Clark, S.A. et al. (2004). The global decline of nonmarine mollusks.
Reviews of Interventions Version 5.1.0 [updated March 2011]. In: The Bioscience, 54, 321–330.
Cochrane Collaboration. Available from www.training.cochrane.org/ha Lytle, D.A. & Poff, N.L. (2004). Adaptation to natural flow regimes.
ndbook Trends Ecol. Evol., 19, 94–100.
HilleRisLambers, J., Harsch, M.A., Ettinger, A.K., Ford, K.R. & MacDougall, A.S. & Turkington, R. (2005). Are invasive species the
Theobald, E.J. (2013). How will biotic interactions influence climate drivers or passengers of change in degraded ecosystems? Ecology, 86,
change-induced range shifts? Ann. N. Y. Acad. Sci., 1297, 112–125. 42–55.
Hoeksema, J.D. & Bruna, E.M. (2000). Pursuing the big questions about Maestre, F.T., Callaway, R.M., Valladares, F. & Lortie, C.J. (2009).
interspecific mutualism: a review of theoretical approaches. Oecologia, Refining the stress-gradient hypothesis for competition and facilitation
125, 321–330. in plant communities. J. Ecol., 97, 199–205.
Hogg, R.S., Coghlan, S.M., Zydlewski, J. & Simon, K.S. (2014). Maestre, F.T., Valladares, F. & Reynolds, J.F. (2006). The stress-gradient
Anadromous sea lampreys (Petromyzon marinus) are ecosystem hypothesis does not fit all relationships between plant-plant interactions
engineers in a spawning tributary. Freshw. Biol., 59, 1294–1307. and abiotic stress: Further insights from arid environments. J. Ecol.,
Holmgren, M. & Scheffer, M. (2010). Strong facilitation in mild 94, 17–22.
environments: the stress gradient hypothesis revisited. J. Ecol., 98, Magoulick, D.D. (2014). Impacts of drought and crayfish invasion on
1269–1275. stream ecosystem structure and function. River Res. Appl., 30,
Holomuzki, J.R., Feminella, J.W. & Power, M.E. (2010). Biotic 1309–1317.
interactions in freshwater benthic habitats. J. North Am. Benthol. Soc., McGill, B.J. (2010). Matters of scale. Science (80-., ), 328, 575–576.
29, 220–244. McLaughlin, J.F. & Roughgarden, J. (1992). Predation across spatial
Hood, G.A. & Larson, D.G. (2014). Beaver-created habitat heterogeneity scales in heterogeneous environments. Theor. Popul. Biol., 41, 277–299.
influences aquatic invertebrate assemblages in boreal Canada. Wetlands, Medina-Sanchez, J.M., Villar-Argaiz, M. & Carrillo, P. (2004). Neither
34, 19–29. with nor without you: a complex algal control on bacterioplankton in a
Hothorn, T., Bretz, F. & Westfall, P. (2008). Simultaneous inference in high mountain lake. Limnol. Oceanogr., 49, 1722–1733.
general parametric models. Biometrical J., 50, 346–363. Menge, B.A. & Sutherland, J.P. (1987). Community regulation: Variation
Jablonski, D. (2008). Biotic interactions and macroevolution: Extensions in disturbance, competition, and predation in relation to environmental
and mismatches across scales and levels. Evolution, 62(4), 715–739. stress and recruitment. Am. Nat., 130, 730–757.
Jenny, J.-P., Anneville, O., Arnaud, F., Baulaz, Y., Bouffard, D., Mermillod-Blondin, F. & Rosenberg, R. (2006). Ecosystem engineering:
Domaizon, I. et al. (2020). Scientists’ Warning to Humanity: Rapid The impact of bioturbation on biogeochemical processes in marine and
degradation of the world’s large lakes. J. Great Lakes Res, 46(4), freshwater benthic habitats. Aquat. Sci., 68, 434–442.
686–702. Michalet, R., Le Bagousse-Pinguet, Y., Maalouf, J.-P. & Lortie, C.J.
Jones, C.G. (2012). Ecosystem engineers and geomorphological signatures (2014). Two alternatives to the stress-gradient hypothesis at the edge of
in landscapes. Geomorphology, 157, 75–87. life: the collapse of facilitation and the switch from facilitation to
Jones, C.G., Lawton, J.H. & Shachak, M. (1994). Organisms as competition. J. Veg. Sci., 25, 609–613.
ecosystem engineers. Oikos, 69, 373–386. Montoya, J.M. & Raffaelli, D. (2010). Climate change, biotic interactions
Jones, C.G., Lawton, J.H. & Shachak, M. (1997). Positive and negative and ecosystem services. Philos. Trans. R. Soc. B Biol. Sci., 365,
effects of organisms as physical ecosystem engineers. Ecology, 78, 2013–2018.
1946–1957. Mooney, H.A. & Cleland, E.E. (2001). The evolutionary impact of
Karr, J.R. (1991). Biological integrity: A long-neglected aspect of water invasive species. Proc. Natl Acad. Sci. USA., 98, 5446.
resource management. Ecol. Appl., 1, 66–84. Moore, J.W. (2006). Animal ecosystem engineers in streams. Bioscience,
Kay, K.M. & Schemske, D.W. (2004). Geographic patterns in plant- 56, 237–246.
pollinator mutualistic networks: Comment. Ecology, 85, 875–878. Morris, W.F., Hufbauer, R.A., Agrawal, A.A., Bever, J.D., Borowicz,
Kelly, D.W., Dick, J.T.A. & Montgomery, I. (2002). Predation on mayfly V.A., Gilbert, G.S. et al. (2007). Direct and interactive effects of
nymph, Baetis rhodani, by native and introduced Gammarus: direct enemies and mutualists on plant performance: a meta-analysis. Ecology,
effects and the facilitation of predation by salmonids. Freshw. Biol., 47, 88, 1021–1029.
1257–1268. Mulder, C.P., Uliassi, D.D. & Doak, D.F. (2001). Physical stress and
Kissling, W.D., Dormann, C.F., Groeneveld, J., Hickler, T., Kuhn, I., diversity-productivity relationships: the role of positive interactions.
McInerny, G.J. et al. (2012). Towards novel approaches to modelling Proc. Natl. Acad. Sci. USA, 98, 6704–6708.
biotic interactions in multispecies assemblages at large spatial extents. Nakagawa, S., Noble, D.W.A., Senior, A.M. & Lagisz, M. (2017). Meta-
J. Biogeogr., 39, 2163–2178. evaluation of meta-analysis: ten appraisal questions for biologists.
BMC Biol., 15, 18.

© 2020 John Wiley & Sons Ltd


Review And Synthesis Freshwater positive interactions 13

Ormerod, S.J., Dobson, M., Hildrew, A.G. & Townsend, C.R. (2010). Silliman, B.R. & He, Q. (2018). Physical stress, consumer control, and
Multiple stressors in freshwater ecosystems. Freshw. Biol., 55, 1–4. new theory in ecology. Trends Ecol. Evol., 33, 492–503.
Peckarsky, B.L. (1983). Biotic interactions or abiotic limitations? A model Simberloff, D. & Von Holle, B. (1999). Positive Interactions of
of lotic community structure. In Dynamics of lotic ecosystems (eds Nonindigenous Species: Invasional Meltdown? Biol. Invasions, 1, 21–32.
Fontaine, T.D. & Bartell, S.M.). Ann Arbor Science Publications, Ann Soluk, D.A. & Collins, N.C. (1988). Synergistic interactions between fish
Arbor, pp. 303–323. and stoneflies - facilitation and interference among stream predators.
Peckarsky, B.L., Horn, S.C. & Statzner, B. (1990). Stonefly predation Oikos, 52, 94–100.
along a hydraulic gradient: a field test of the harsh-benign hypothesis. Soluk, D.A. & Richardson, J.S. (1997). The role of stoneflies in enhancing
Freshw. Biol., 24, 181–191. growth of trout: a test of the importance of predator-predator
Peoples, B.K., Blank, L.A. & Frimpong, E.A. (2015). Lotic cyprinid facilitation within a stream community. Oikos, 80, 214–219.
communities can be structured as nest webs and predicted by the stress- Stachowicz, J.J. (2001). Mutualism, facilitation, and the structure of
gradient hypothesis. J. Anim. Ecol., 84, 1666–1677. ecological communities. Bioscience, 51, 235–246.
Peoples, B.K., Tainer, M.B. & Frimpong, E.A. (2011). Bluehead chub Stanton, M.L. (2003). Interacting guilds: Moving beyond the pairwise
nesting activity: a potential mechanism of population persistence in perspective on mutualisms. Am. Nat., 162, S10–S23.
degraded stream habitats. Environ. Biol. Fishes, 90, 379–391. Stewart, T.W., Miner, J.G. & Lowe, R.L. (1999). A field experiment to
Piccardi, P., Vessman, B. & Mitri, S. (2019). Toxicity drives facilitation determine Dreissena and predator effects on zoobenthos in a nearshore,
between 4 bacterial species. Proc. Natl Acad. Sci. USA, 116, rocky habitat of western Lake Erie. J. North Am. Benthol. Soc., 18,
15979–15984. 488–498.
Pintor, L.M. & Soluk, D.A. (2006). Evaluating the non-consumptive, Strayer, D.L. (2010). Alien species in fresh waters: ecological effects,
positive effects of a predator in the persistence of an endangered interactions with other stressors, and prospects for the future. Freshw.
species. Biol. Conserv., 130, 584–591. Biol., 55, 152–174.
Preston, K.L., Rotenberry, J.T., Redak, R.A. & Allen, M.F. (2008). Strayer, D.L. & Dudgeon, D. (2010). Freshwater biodiversity
Habitat shifts of endangered species under altered climate conditions: conservation: Recent progress and future challenges. J. North Am.
Importance of biotic interactions. Glob. Chang. Biol., 14, 2501–2515. Benthol. Soc., 29, 344–358.
Prugh, L.R. & Sivy, K.J. (2020). Enemies with benefits: integrating Subalusky, A.L. & Post, D.M. (2019). Context dependency of animal
positive and negative interactions among terrestrial carnivores. Ecol. resource subsidies. Biol. Rev. Camb. Philos. Soc., 94, 517–538.
Lett., 23, 902–918. Thomas, M.J., Creed, R.P., Skelton, J. & Brown, B.L. (2016).
Pustejovsky, J.E. (2018). Using response ratios for meta-analyzing single- Ontogenetic shifts in a freshwater cleaning symbiosis: consequences for
case designs with behavioral outcomes. J. Sch. Psychol., 68, 99–112. hosts and their symbionts. Ecology, 97, 1507–1517.
Ram, J.L., Fong, P., Croll, R.P., Nichols, S.J. & Wall, D. (1992). The Thomsen, M.S., Wernberg, T., Altieri, A., Tuya, F., Gulbransen, D.,
zebra mussel (dreissena polymorpha), a new pest in North America: McGlathery, K.J. et al. (2010). Habitat cascades: The conceptual
Reproductive mechanisms as possible targets of control strategies. context and global relevance of facilitation cascades via habitat
Invertebr. Reprod. Dev., 22, 77–86. formation and modification. Integr. Comp. Biol., 50, 158–175.
R Core Team (2020). R: A language and environment for statistical Thorp, J.H. (1986). Two distinct roles for predators in freshwater
computing. www.R-project.org. assemblages. Oikos, 47, 75.
Ramus, A.P., Silliman, B.R., Thomsen, M.S. & Long, Z.T. (2017). An Tikhonov, G., Abrego, N., Dunson, D. & Ovaskainen, O. (2017). Using
invasive foundation species enhances multifunctionality in a coastal joint species distribution models for evaluating how species-to-species
ecosystem. Proc. Natl. Acad. Sci., 114, 8580–8585. associations depend on the environmental context. Methods Ecol. Evol.,
Reinhardt, L., Jerolmack, D., Cardinale, B.J., Vanacker, V. & Wright, J. 8, 443–452.
(2010). Dynamic interactions of life and its landscape: Feedbacks at the Trautz, A.C., Illangasekare, T.H. & Rodriguez-Iturbe, I. (2017). Role of
interface of geomorphology and ecology. Earth Surf. Process. co-occurring competition and facilitation in plant spacing
Landforms, 35, 78–101. hydrodynamics in water-limited environments. Proc. Natl. Acad. Sci.,
Renzi, J.J., He, Q. & Silliman, B.R. (2019). Harnessing positive species 114, 9379–9384.
interactions to enhance coastal wetland restoration. Frontiers in Tumolo, B.B., Albertson, L.K., Cross, W.F., Daniels, M.D. & Sklar, L.S.
Ecology and Evolution, 7, 1–14. (2019). Occupied and abandoned structures from ecosystem engineering
Resh, V.H., Brown, A.V., Covich, A.P., Gurtz, M.E., Li, H.W., Minshall, differentially facilitate stream community colonization. Ecosphere, 10,
G.W. et al. (1988). The role of disturbance in stream ecology. J. North e02734.
Am. Benthol. Soc., 7, 433–455. Tumolo, B.B., Calle, L., Anderson, H.E., Briggs, M., Carlson, S., Reinert,
Rodriguez, L.F. (2006). Can Invasive Species Facilitate Native Species? J.H. et al. (2020). Toward spatio-temporal delineation of positive
Evidence of How, When, and Why These Impacts Occur. Biol. interactions in ecology. Evol. Ecol. 10, 9026–9036.
Invasions, 8, 927–939. Viechtbauer, W. (2010). Conducting meta-analyses in R with the metafor
Rogosch, J.S. & Olden, J.D. (2020). Invaders induce coordinated isotopic package. J. Stat. Softw., 36, 48.
niche shifts in native fish species. Can. J. Fish. Aquat. Sci., 77, 1–11. Vitousek, P., DAntonio, C., Loope, L., Rejmanek, M. & Westbrooks,
Scammacca, N., Roberts, G. & Stuebing, K.K. (2014). Meta-analysis with R.(1997). Introduced species: A significant component of human-caused
complex research designs: Dealing with dependence from multiple global change. N. Z. J. Ecol., 21, 1–16.
measures and multiple group comparisons. Rev. Educ. Res., 84, Vörösmarty, C.J., McIntyre, P.B., Gessner, M.O., Dudgeon, D.,
328–364. Prusevich, A., Green, P. et al. (2010). Global threats to human water
Schemske, D.W., Mittelbach, G.G., Cornell, H.V., Sobel, J.M. & Roy, K. security and river biodiversity. Nature, 467, 555–561.
(2009). Is there a latitudinal gradient in the importance of biotic Wagner, T., Hansen, G.J.A., Schliep, E.M., Bethke, B.J., Honsey, A.E.,
interactions? Annu. Rev. Ecol. Evol. Syst., 40, 245–269. Jacobson, P.C. et al. (2020). Improved understanding and prediction of
Schindelin, J., Arganda-Carreras, I., Frise, E., Kaynig, V., Longair, M., freshwater fish communities through the use of joint species
Pietzsch, T. et al. (2012). Fiji: an open-source platform for biological- distribution models. Can. J. Fish. Aquat. Sci., 77, 1540–1551.
image analysis. Nat. Methods, 9, 676–682. Wallace, J.B. & Webster, J.R. (1996). The role of macroinvertebrates in
Silknetter, S., Creed, R.P., Brown, B.L., Frimpong, E.A., Skelton, J. & stream ecosystem function. Annu. Rev. Entomol., 41, 115–139.
Peoples, B.K. (2020). Positive biotic interactions in freshwaters: A Wardle, D.A. (2006). The influence of biotic interactions on soil
review and research directive. Freshw. Biol., 65, 811–832. biodiversity. Ecol. Lett., 9, 870–886.

© 2020 John Wiley & Sons Ltd


14 L. K. Albertson et al. Review And Synthesis

Weiss, L.C. (2019). Sensory Ecology of Predator-Induced Phenotypic SUPPORTING INFORMATION


Plasticity. Front. Behav. Neurosci., 12, 330.
Whiles, M.R., Lips, K.R., Pringle, C.M., Kilham, S.S., Bixby, R.J., Additional supporting information may be found online in
Brenes, R. et al. (2006). The effects of amphibian population declines the Supporting Information section at the end of the article.
on the structure and function of Neotropical stream ecosystems. Front.
Ecol. Environ., 4, 27–34.
Wright, J. & Jones, C.G. (2006). The concept of organisms as ecosystem
Editor, Jonathan Chase
engineers ten years on: Progress, limitations, and challenges. Bioscience,
56, 203–209. Manuscript received 9 July 2020
Zanetell, B. & Peckarsky, B.L. (1996). Stoneflies as ecological engineers- First decision made 28 November 2020
hungry predators reduce fine sediments in stream beds. Freshw. Biol., Manuscript accepted 20 November 2020
36, 569–577.

© 2020 John Wiley & Sons Ltd

You might also like