The distribution of biodiversity and the structure of ecological communities are the outcome of processes occurring at different scales (Hortal et al., 2010; 2012; Guisan and Rahbek, 2011). Indeed, understanding the mechanisms responsible for the maintenance of biodiversity depends on the ability to decompose the diversity variation that is associated with processes acting at different scales (Cottenie et al., 2003; Cottenie and De Meester, 2003; Legendre et al., 2009). Further, although many ecological studies focus on the drivers of species richness, the identities, abundance, phenotypes and phylogenetic relationships of these species are also critical determinants of the nature and strength of the relationships between diversity and a range of ecological functions (Edwards et al., 2013; Pavoine and Bonsall, 2011). In particular, functional diversity is a key aspect of biodiversity essential for understanding the processes behind community assembly and structure. Only species with certain traits can disperse, overcome environmental filters, establish and co-exist in a particular area and moment (Mittelbach and Schemske, 2015). Therefore, combining information from taxonomic and functional diversity allows ascertaining the processes dominating community assembly and structure, providing an effective way of assessing the effects of human interventions on biodiversity (Graco-Roza et al., 2021).

Microorganisms make out most of Earth’s diversity. Many can be found in freshwater systems, where diversity drivers acting at different scales are intertwined (Pajunen et al., 2017). Phytoplankton stand out as one of the most relevant primary producers in aquatic ecosystems. They are responsible for a great portion of global primary production and mediate the biogeochemical cycle of several elements, being the major source for silica, carbon, nitrogen and phosphorus uptake in some systems (Plus et al., 2015). Phytoplankton respond rapidly to environmental changes and disturbance gradients, being widely used as bioindicators for pollution (Wang et al., 2017). Their communities also respond to environmental gradients and spatial and historical processes (Stomp et al., 2011; Padisák et al., 2016; Ribeiro et al., 2018), as well as to biological interactions. These interactions include the control of phytoplankton biomass by macrophytes (through allelopathic effects that suppress phytoplankton; Vanderstukken et al., 2011) and by filter-feeding fish and zooplankton (through grazing and predation; Zhang et al., 2023) but also the positive contribution of the benthos to primary production and phytoplankton growth by resuspending sediments (Moncelon et al., 2022).

Importantly, phytoplankton communities are typically composed of species holding different functional traits, which are spatial and temporally distributed along a multi-dimensional setting (Edwards et al., 2013). The local environmental gradients within the water body (e.g. light irradiance, phosphate, nitrate) produce changes in the composition and functional structure of the community (Litchman and Klausmeier, 2008). Here, response traits may provide the link between environmental drivers and mechanisms of change, as they may respond to resource availability and other environmental conditions (Edwards et al., 2013). Importantly, as several phytoplankton traits (such as life form, presence of heterocysts or aerotopes, etc.) have direct effects on functional processes, the community changes associated to these responses also affect ecosystem functioning (Litchman et al., 2010; Thompson et al., 2015; Santos et al., 2015).

The generalized cosmopolitism of phytoplankton has led to consider changes in trophic conditions along the eutrophication–oligotrophication gradient as the main determinant of the distribution of this group (Pomati et al., 2012). However, while several studies indicate positive relationships between phytoplankton abundance and the amount of light and nutrients available (Lewis and Wurtsbaugh, 2008; Litchman et al., 2004; Ramdani et al., 2009), others show no evidence of the influence of these local environmental factors on phytoplankton communities (Beisner et al., 2006; Nabout et al., 2009). This may be related with the fact that, in freshwater systems, phytoplankton diversity is highly influenced by the spatial design of lakes and watersheds (Soininen et al., 2004). The relationship between taxon diversity and lake size differs widely among species (Scheffer et al., 2006), and it was weak (Søndergaard et al., 2005) or absent (De Marco et al., 2013) for phytoplankton in small ponds. In fact, several small waterbodies like ponds can support a more diverse microalgae community than a single large one (Bolgovics et al., 2019), calling for an important influence of landscape structure and spatial heterogeneity on phytoplankton diversity in freshwater systems. In a fragmented landscape, the limited connectivity among aquatic environments can lead to a decrease in the regional biodiversity able to colonize the ponds (Scheffer et al., 2006). In this context, the effect of processes acting at the landscape scale on the aquatic communities occurs indirectly via the influence that regional environmental conditions and/or the spatial structure of the landscape exert in the local physical and chemical conditions of waterbodies. Processes like riparian vegetation deforestation, hydrologic alterations or pollution cause changes in local physical and chemical conditions such as water temperature, erosion rates, light penetration, nutrient input and retention, flood magnitude and frequency, and/or heavy metal concentration, in turn affecting the population dynamics of freshwater organisms such as phytoplankton, zooplankton or benthos (Allan, 2004).

A consequence of the influence of landscape and local environmental conditions of water bodies on phytoplankton communities is that both factors can have significant effects on the functional diversity of their phytoplankton communities, which in turn may cause shifts in their functioning (Graco-Roza et al., 2021). For instance, phytoplankton body size has been related to both trophic status and lake isolation (Erdo¿an et al., 2021), although other studies have found that other traits, as well as functional diversity, are more associated to local factors than to landscape typology (Derot et al., 2020; Vadrucci et al., 2013). However, landscape management has a significant influence on waterbodies. Indeed, the increases in community biomass associated to high nutrient intake from agricultural practices often cause decreases in taxonomic and functional diversity (Da Silva et al., 2020; Pomati et al., 2012; Török et al., 2016). Further, the increase in phosphorus and nitrogen concentration in water bodies affected by the use of pesticides and fertilizers in the surrounding landscape cause decreases in phytoplankton functional evenness and functional richness, and increases in functional divergence, as eutrophication-tolerant species increase in importance (Wijewardene et al., 2021). Despite these evidences that certain trait combinations are filtered by both landscape and local factors with different degrees of importance, little is known about the scaling of such filtering in phytoplankton communities.

Here we evaluate the influence of several local and landscape factors on different aspects of phytoplankton diversity in agricultural ponds in the Brazilian Cerrado. We expected that local limnological variables such as pH, temperature, nutrients, and light would directly impact phytoplankton abundance. Specifically, increases in these variables should positively affect phytoplankton abundance, as they indicate favorable environmental conditions and resource availability (Padisák, 2004). Additionally, we hypothesized that pond characteristics (such as area, age, and perimeter) and landscape factors (such as presence of shadow, land use, and connectivity) would influence both local limnological variables and phytoplankton species richness, abundance, and functional diversity. Here, ponds with greater connectivity would show lower nutrient concentration as a result of the influence of the streams that provide such connectivity, which in this area are likely to have good water quality (Bichsel et al., 2016). Therefore, the connection between ponds and streams would indirectly and negatively affect the abundance of phytoplankton. In contrast, higher connectivity would have a direct and positive influence on species richness, as it facilitates colonization by new species (Oertli and Parris, 2019; but see Pęczuła and Szczurowska, 2013). Pond size could impact local limnological variables and indirectly affect phytoplankton abundance. Well-preserved ponds with greater shade and less intensive land use would have a positive influence on phytoplankton richness and functional diversity, while having the opposite effect on abundance. These preserved ponds would have lower productivity, resulting in lower dominance (Zhang et al., 2018). Furthermore, the age of the ponds would positively influence both phytoplankton richness and abundance (Tulsankar et al., 2020; Zimba et al., 2003). Lastly, we anticipated that variations in species richness and relative abundances of phytoplankton in local ponds would directly impact the functional diversity of phytoplankton within those ponds (Fig. 1).

Fig. 1.

Initial a priori Structural Equation Model, including the preliminary hypotheses on the relationships among variables. According to these hypotheses, land use (level of use and shadow), and pond morphology (area, age and isolation) should affect directly the limnological variables and species richness, and also indirectly all diversity metrics, including abundance and two different aspects of functional diversity; Functional Divergence (MTD) and Functional Evenness (FEve). See variable descriptions in Table 1.

(0.24MB).

There was a large variation between ponds in terms of morphometry, limnological conditions (such as conductivity or chloride) and productivity (e.g., chlorophyll-a). Considering the values of chlorophyll-a concentration, the studied ponds covered a wide range of levels of productivity, some being ultraoligotrophic (14 ponds ≤1.17 µg/L), while others oligotrophic (17 ponds, chl-a: 1.17–3.24 µg/L), mesotrophic (9 ponds, chl-a: 3.24–11.03 µg/L), eutrophic (3 ponds, chl-a: 11.03–30.55 µg/L) or supereutrophic (1 pond, chl-a: 30.55–69.07 µg/L). The number of land uses around the ponds varied from only one to three, with 11 ponds presenting all three uses, 18 presenting both agriculture and livestock uses, five presenting agriculture and fish-farming land uses, two presenting livestock and fish-farming uses, four presenting only agriculture and five only livestock.

In total, 300 phytoplankton species were found across the 45 ponds sampled (Supplementary Table S1). The most representative classes in terms of number of species were Bacillariophyceae, Conjugatophyceae (desmids), Trebouxophyceae, Chrysophyceae, Dinophyceae and Chlorophyceae, while the most abundant classes were Cyanophyceae, Treuboxiophyceae and Conjugatophyceae (Supplementary Table S2). Regarding species characteristics, the majority of the individuals were unicellular (68%) and filamentous (28%); most of them were neither flagellate (91.52%) nor mucilaginous (71.58%), and did not present silica (96.14%), heterocites (99%) and aerotopes (74.92%). Although FEve showed a low variation between ponds, species evenness varied widely, from 0.086 to 0.942. Abundance showed a high variation among the ponds (SD 40028 individuals/mL; Supplementary Table S3).

Two axes were selected in the PCA analysis performed on the local limnological variables, which explained respectively 27.3% and 22.1% of the variability in these variables (Fig. 3). The first axis (PC1) was strongly associated with water transparency (23.58% of contribution), conductivity (31.33% of contribution) and ammonia (16.69% of contribution), while the second axis (PC2) was more associated to chloride (31.85 % of contribution), temperature (24.11 % of contribution), water transparency (21.11 % of contribution) and dissolved oxygen (16.96 % of contribution).

According to the fitted SEM, connectivity was the most important correlate of phytoplankton abundance, followed by PC1 (more influenced by transparency, conductivity and ammonium). These variables together explained 32% of phytoplankton abundance in the ponds, while connectivity and pond area explained 13% of PC1 (i.e. conductivity, water transparency and ammonia), and PC2 (more influenced by chloride, temperature, water transparency and dissolved oxygen) held no significant effect on any aspect of the phytoplankton communities. The direct influence of connectivity on abundance was negative (r = −0.55), evidencing that the most connected ponds (i.e. the dams) hosted lower phytoplankton abundance. The local variables also had direct effects on abundance. Further, pond age had an impact on richness and pond area affected local variables. The limnological variables represented by PC1 (i.e., conductivity, water transparency and ammonia) were positively related with connectivity and negatively with area, meaning that the largest ponds showed lower values of conductivity, ammonia and water transparency. Further, the direct effects of abundance and pond age explained together 28% of species richness variations. In turn, species richness and PC1 explained 25% of the variability in functional evenness (i.e., FEve). Strikingly, abundance was positively associated to phytoplankton richness, whereas the association between richness and FEve was negative (i.e., richer ponds showed highly packed functional trait spaces). Functional divergence was not associated to any predictor or diversity metric (Fig. 4).

Fig. 4.

Structural Equation Model (SEM) depicting the main determinants of community diversity in the studied system of artificial ponds. This SEM combination is the most congruent with the observed data, with no support for deleting further paths. Note that local variables in the a priori model depicted in Fig. 1 are here summarized as the two axes of a PCA (see Fig. 3). Black arrows represent positive significant relationships and red arrows represent negative significant relationships (p < 0.05).

(0.27MB).

Our results show that the characteristics of both ponds and landscape are important direct and indirect drivers of the diversity of phytoplankton communities in artificial tropical ponds. Overall, the diversity and structure of pond communities were mediated by abundance; the richest ponds showed less functional evenness, and therefore higher functional redundancy. In turn, abundance depended on both pond connectivity and several local limnological variables. Joint effects of pond and landscape factors on phytoplankton communities have been already found for coastal and terrestrial lakes (Spatharis et al., 2019; Loewen et al., 2020), tropical floodplains (Moresco et al., 2020) or pond microcosm experiments (Pereira et al., 2018). Strikingly, our structural model presented higher explanatory power and is simpler (i.e., includes less variables) than most of these former works.

One of the most important results of this work is that abundance mediates species richness and, through it, functional evenness. Such preeminence of abundance on phytoplankton biodiversity dynamics is related with water quality gradients (Lavoie et al., 2014). Indeed, the high abundances found in several of the ponds studied point to the existence of more eutrophic environments. The species that become dominant in more eutrophic lakes often share similar traits (Reynolds, 1998). Therefore, eutrophic shallow lakes are typically dominated by one or few species from the same functional group (Borics et al., 2012), showing lower functional diversity and, in particular, lower trait evenness and a high degree of redundancy (as species are not sparsely distributed in the trait space). For example, most of the microalgae in the ponds we studied were unicellular or filamentous, so they show a high-rate surface/volume, which is an important trait in the efficiency of resource use (Reynolds, 2006), in their case determined by cell size and shape.

The level of use had no direct influence in the community metrics, however, several of the ponds studied are fish-farming (18 ponds). Fish can be an important factor in shaping the food web within these small ecosystems (Søndergaard et al., 2005). Therefore, the positive effect of phytoplankton abundance on species richness may be also related with the characteristics of the farm ponds themselves such as the use type. Phytoplankton diversity in small lakes and ponds can depends indirectly on surface area through its effects on fish and the structure of aquatic vegetation (Scheffer et al., 2006). Furthermore, previous research has shown that variations in zooplankton on shallow lakes were also correlated to factors such as fish abundance, macroinvertebrates and turbidity (Cottenie et al., 2001).

Abundance also mediated the effect of connectivity on phytoplankton communities. The connected ponds showed lower levels of abundance, which in turn limits species richness; also, these poorer communities were functionally even, composed by species with differing traits and little redundancy. Phytoplankton communities in lentic environments respond to factors acting at different scales, such as spatial structure, limnological conditions and watershed characteristics (Loewen et al., 2020). Indeed, the relevance of the spatial configuration of the landscape on phytoplankton structure depends on the study scale (Heino et al., 2015a, b), type of aquatic system (Heino et al., 2015c), eutrophication level, connectivity, environmental gradients or even hydrological season (Soininen, 2014; Izaguirre et al., 2016; Brasil et al., 2020). Dispersal limitations can decrease local diversity, so the biodiversity of pond communities is strongly linked to the connection and density of other ponds in the landscape (Oertli and Parris, 2019). However, it is possible that, in these studies, connectivity actually represents the influence of lotic systems. In our particular case, the connected ponds are dams, which may present constant or intermittent diluter and disturber effects from the river or stream that limit phytoplankton development. This contrasts with isolated ponds, which present more stable conditions, having less residence time than ditches or springs, thus favoring the buildup of more diverse communities through longer assembly processes. Indeed, similar effects have been observed on lake phytoplankton, where river connections (with low trophic level) can decrease phosphorus, conductivity and transparency, consequently diminishing richness and abundance (Pȩczuła and Szczurowska, 2013).

Light (water transparency), nutrients and conductivity were positively associated to abundance and Functional Evenness, reflecting the access and availability of resources to phytoplankton (Padisák, 2004). Factors like area, depth, residence time, water temperature, transparency, pH, conductivity, nitrogen and phosphorus determine phytoplankton abundance and functional structure on tropical reservoirs and lakes (Lewis et al., 2020; Santos et al., 2016). Phytoplankton biomass is commonly related with resource variation (Korhonen et al., 2011), and another study in temperate streams using SEMs found a positive effect of both water temperature and nutrients on phytoplankton biomass (Pajunen et al., 2017). In our study, ponds with higher conductivity and lower pH had poorer communities, a result that has been related with an increase in dominance and the consequent decrease in richness and functional evenness in progressively more acidic and eutrophic water bodies (Lau et al., 2017; Wang et al., 2017). However, the relationship between phytoplankton functional evenness and nutrients that we reported in the tropical farm ponds is the opposite to the one reported for temperate farm ponds (Wijewardene et al., 2021), where increasing functional evenness of microalgae has been related with increasing conductivity of rock pools in the warmer months (Aarnio and Soininen, 2021). The limited number of phytoplankton ecology studies in Cerrado farm ponds (but see, Bichsel et al., 2016; Carneiro et al., 2019; De Marco et al., 2013; Fonseca et al., 2018) prevents from determining whether these differences with temperate systems are characteristic of this biome, or merely restricted to the studied landscape.

Lake area was also an important determinant of phytoplankton diversity in our study system, as the largest ponds showed negative values of PC1, representing conductivity, ammonia and light, i.e. variables that can reflect water quality (Amengual-Morro et al., 2012). Another study from this same area showed that large ponds hold more stable phytoplankton communities because they can buffer the effect of heavy rains, while smaller ponds often harbor high levels of chlorophyll-a concentration (De Marco et al., 2013). Here is worth noticing that although land use variables did not affect pond communities in a direct way, landscape management can indirectly influence both nutrients and conductivity (Søndergaard et al., 2005). In any case, these results highlight the importance of phytoplankton as an indicator of water quality and anthropogenic impact (Shi et al., 2015), even when the effects of other aspects of land intensification are not evident.

Despite the typical complexity of ecological communities, changes in functional trait distributions along environmental gradients are often predictable (Götzenberger et al., 2012; Edwards et al., 2013). Here, the relationships between abundance and connectivity, pond size (area) and, indirectly, limnological variables, provided a mechanism to relate functional evenness with pond characteristics. Eutrophication would lead to ponds with higher abundances and richer communities that are formed by functionally highly redundant species. Evenness is known to decrease in the most diverse communities, which paradoxically can show reduced compositional turnover in response to environmental changes. It follows that even, but less abundant, communities maintain higher functional diversity and ecosystem functioning (Isbell et al., 2015). Here we must note that the traits used here (e.g.: MDL, aerotopes, siliceous structures; body form; mucilage) accounted for responses to both environmental gradients and interspecific interactions, since PC1 and Richness influenced directly the Functional Evenness (see also Litchman and Klausmeier, 2008; Santos et al., 2015).

Our results pointed to pond connectivity being the main determinant of phytoplankton abundance, species richness and functional diversity, together with other variables mediated by human activities that are associated to resource availability, such as light, nutrients and conductivity. This highlights the importance of human actions in the structure and functioning of artificial pond ecosystems, as the size and connectivity of man-made ponds reflect the needs and actions of landowners (Clifford and Heffernan, 2018). These anthropogenic effects operate through the regulation of phytoplankton abundance, which in turn mediates species richness, and through it, functional evenness. This effect should be taken into account for the construction and management of farm ponds in this region devoted to livestock herding, as the balance between eutrophication and functional evenness –and thus functioning– is related with the design of the ponds themselves. Therefore, conserving small farm ponds is crucial in order to preserve aquatic biodiversity in general, as these systems are also home to a range of organisms, such as fish, macrophytes, benthos, and zooplankton, which are intricately interconnected with phytoplankton, and their population dynamics are likely affected by comparable local and regional factors. Besides that, the complex relationships we unveiled highlight the importance of considering multiple facets of biodiversity for understanding community dynamics. Analyzing the interaction of factors that influence the dynamics of communities can contribute to the advance of limnological research in understanding the relationships between various stressors and determinants of freshwater communities.

Universidade Estadual de Goiás; Ramón Y Cajal program; Spanish Ministerio de Ciencia e Innovación; Universidad de Alcalá; CNPq; INCT in Ecology, Evolution and Biodiversity Conservation MCTIC/CNPq/FAPEG.

F.M.C., A.M.C.S. and J.H. designed the study, with N.G.M. and P.D.M.J.; F.M.C. and P.D.M.J. retrieved and processed the data; F.M.C., A.M.C.S. and N.G.M. analysed the data; F.M.C. wrote the paper, with A.M.C.S. and J.H.; all authors discussed the results, contributed substantially to revisions and approved the final version of the manuscript.

Data are available from the authors upon reasonable request.

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.