Effects of intensive agriculture and hydrological changes on macrophyte and macroinvertebrate assemblages in lowland riverine wetlands

The aim of this study was to investigate the effects of agricultural land use and periods of hydrological variability on the environmental variables, as well as macrophyte and macroinvertebrate assemblages in lowland riverine wetlands. We compared two wetlands with intensive agricultural land use against two others with extensive livestock considered references for the region during a normal and a dry ow period. Nutrient concentrations were signicantly higher in agricultural riverine wetlands. These wetlands exhibited higher relative coverage of oating anchored macrophytes and the absence of submerged vegetation. They showed signicantly lower taxonomic richness and density of macroinvertebrates and a higher relative abundance of scrapers and predators. Wetlands of both land uses had a lower total density of macroinvertebrates and a higher proportion of tolerant desiccation taxa in the dry period. Particular differences between land uses, such as lower dissolved oxygen concentrations and lower macroinvertebrate diversity in agricultural wetlands, were found during the dry period. These ndings indicate that the differences between land uses increased during the aforementioned period. This study provides evidence of the effects of the surrounding landscape and hydrologic periods in the environmental characteristics as well as the macrophyte and macroinvertebrate assemblages of the riverine wetlands studied.


Introduction
Wetlands are globally recognized as essential due to the many ecological functions and services that they provide to human society (Dixon and Wood 2007). They provide water quality protection (Verhoeven et  Riverine wetlands (RWs) are recognized as a major component of biodiversity in uvial ecosystems (Amoros et al. 2000;Tockner et al. 2002). They are areas where streams over ow and connect with the surrounding terrestrial environment (Ringuelet, 1962;Brinson et al. 2002) and where macrophyte assemblages are highly developed (Amoros et al. 2000). As it is known, macrophytes are key components of wetlands ecosystems (Rejmánková 2011). They remove a variety of pollutants from the water (Bonanno and Vymazal 2017), and increase the habitat complexity and heterogeneity of these ecosystems (Thomaz and Cunha 2010). Also, they provide several resources for other organisms, like microhabitats (Dudley, 1988;Warfe et al. 2008), shelter (Heck and Crowder, 1991; Thomaz and Cunha 2010) and food (Dvořák, 1996;Díaz-Valenzuela et al. 2016).
Within the wetlands biota, macroinvertebrates play an important role in the overall functioning of these ecosystems as they occupy a central position in the food web and in organic matter cycling and energy ow (Batzer et al. 1999). The reduction in water quality caused by agricultural land use also leads to decreases in macroinvertebrate richness and density, as well as the decrease in the abundance of sensitive groups (Genito et  The knowledge of biodiversity of RWs and the effects of land use and hydrological periods is necessary for ecosystem management and conservation plans. Our goal was to investigate the effects of intensive agricultural land use on the environmental variables and the macrophyte and macroinvertebrate assemblages of Pampean lowland RWs of Argentina, in time periods characterized by different hydrological conditions. We hypothesized that i) intensive agricultural land use induces changes in physical and chemical variables and in the structural and functional responses of macrophyte and macroinvertebrate assemblages; ii) an extraordinary dry period leads to variations in the assemblages and increases differences between land uses.

Study area
The study was conducted in the Pampean ecoregion of Buenos Aires, Argentina (Fig. 1). This region is a vast grassy plain that covers central Argentina. It has a humid and temperate climate, mean annual precipitation between 1000 and 1200 mm, and a mean annual temperature of 16°C (Hurtado et al. 2006).
Riverine wetlands of this region are threaten by agriculture that affects the water quality and alters the natural habitat (Gómez et al. 2016). In particular, RWs located in the Pampean plain of Argentina represent the low depressions of Pampean stream basins, characterized by abundant and diverse aquatic vegetation. The streams mentioned are characterized by an absence of riparian forest, low current velocities, and high nutrient levels (Rodrigues Capítulo et al. 2001;Feijoó and Lombardo 2007).
We selected two periurban RWs with intensive agricultural land use and other two with extensive livestock located on tributary streams of the freshwater section of the Río de la Plata estuary (Fig. 1, Online Resource: Table S1). The selection of this RWs was based on geohydrological and land use analysis To know the effects of hydrological variation, four samplings were performed on the RWs. Two samplings were carried out during a period of normal ow, which we call "normal period" (September and October, mean monthly cumulative precipitation of 80 mm and 61 mm respectively), and two samplings in a dry period (February and early March). The monthly accumulative precipitations of this period were 2 mm in February and 92 mm in March, with 82 mm accumulated in the ve days after the sampling date.
The average precipitation recorded was lower than the historical mean monthly accumulated precipitation for the last ten years (National Meteorological Service: February 177 mm and March 123 mm). This intense drought coincides with the "La Niña" phase of the ENSO phenomenon (Gómez et al. in press).
In each RW we selected a section of 50m in the center of the channel, depending on the ease of access, where the sampling was carried out. In that section, in each sampling occasion, physical and chemical variables and macrophyte assemblage descriptors were measured and macroinvertebrates were sampled (see below). During the dry period, the RWs studied were characterized by lower width, lower water velocity, and lower ow in the four wetlands studied (Online Resource: Table S1). The Carnaval, one of the agricultural RWs, was completely dry during the February campaign.

Page 5/27
The pH, temperature, dissolved oxygen concentration (DO), conductivity, and total dissolved solids were measured in situ in triplicate with a HORIBA Multiparameter U-10. Besides, on each sampling occasion, a water sample was collected for the analysis of nutrient concentrations (Total Nitrogen, Nitrates [N-NO 3

Macrophytes
The macrophytes species, along with their coverage and life forms, were recorded. Five evenly-spaced transects were performed perpendicular to the channel in the section of 50m in each wetland to calculate the macrophyte coverage. On each transect, the total coverage of each patch and species were estimated by measuring the length of the transect covered by the water and by each species (Feijoó and Menéndez 2009). In addition, total macrophyte coverage was calculated as the sum of total coverage values of present species. For this study, we classi ed the species as emergent, submerged, oating-leaved anchored, and free-oating (Cabrera, 1948; Lahitte et al. 2004).

Data analysis and statistical approach
The RWs along the physical and chemical variables were characterized by principal-component analysis (PCA). Before the analysis, the variables were standardized. The variables that presented little contribution, those that present low correlation with the rst and the second component, were removed to simplify analysis (Kassambara 2017). Differences in physical and chemical variables between land uses and hydrological periods and their interaction were assessed by two-way ANOVA with 'RW' nested within 'land use'. The variables that did not t with a normal distribution (temperature, conductivity, DO, Total Nitrogen and Total Phosphorus) were log-transformed. Pairwise comparisons on the main xed factors were performed using Tukey's post hoc tests. Model residuals were tested for normality using a Shapiro-Wilk test.
The total coverage of macrophytes and the relative coverage of each life form expressed as percentages were calculated to characterize the macrophyte assemblage. Also, the richness was estimated as the number of taxa present, and diversity was estimated by the Shannon-Wiener Index (Shannon and Weaver, 1949). In addition, to characterize the macroinvertebrate assemblage, density was expressed as the average number of individuals per square meter, and richness and diversity were estimated. We assigned each taxon to a functional feeding group (FFG) using available references (Cummins et al. Macrophyte coverage, diversity of both assemblages, as well as density and FFG of macroinvertebrates, were compared between land uses and hydrological periods by two-way ANOVA with 'RW' nested within 'land use'. The interaction between 'land use' and 'hydrological period' was also assessed. Density of macroinvertebrates data were log-transformed, whereas coverage of macrophytes and FFGs of macroinvertebrates were arcsine transformed to t with the normal distribution. Pairwise comparisons on the main xed factors were performed using Tukey's post hoc tests. Model residuals were tested for normality using a Shapiro-Wilk test. For count data (richness), we used generalized linear models (GLMs) with Poisson error distribution (link: log) for the same factors.
A Permutational Multivariate Analysis of Variance (PERMANOVA) was used to test differences in macroinvertebrate taxonomic composition between 'land use' (agriculture and livestock) and 'hydrological period' (normal and dry), with 'RWs' nested within 'land use'. The PERMANOVA was applied on a Bray-Curtis dissimilarity matrix calculated from the abundance data of macroinvertebrates. We also used a similarity percentage analysis (SIMPER) to identify taxa separating different land use categories and hydrological periods and to quantify the contribution of individual taxa to each category.
All statistical analysis were performed using the language environment R version 3.6.3 (R Development Core Team 2020) and the RStudio Team (2015). The packages used for the analysis were FactoMineR

Sites characterization
The rst two PCA axes explained 64.1% of the overall variance (Fig. 2). The rst axis (43.5%) illustrated the environmental differences between land uses. Agricultural RWs were characterized by higher nutrient concentration (total phosphorus, total nitrogen, soluble reactive phosphorus, and to a lesser extent, ammonium). By contrast, livestock RWs were characterized by higher values of pH, conductivity, and TDS and lower values of nutrient concentrations (Online Resource: Table S2, correlations between axis and variables). The second axis (20.6% of the total variance) was positively correlated with temperature, nitrate, and nitrite whereas was negatively correlated with ow and dissolved oxygen. This axis showed the difference between hydrological periods in livestock RWs, with respect to temperature, DO and ow. The dry period was characterized by higher temperature, lower ow and DO in comparison with the normal period.

Macrophytes
We recorded 15 species during the study, and the total coverage of macrophytes was always greater than 60% in the RWs studied ( Table 2) Table S3).

Macroinvertebrates
A total of 63 taxa of macroinvertebrates were collected in the RWs studied (Online Resource: Table S4). Taxa richness differed signi cantly between land uses (estimate = 0.39, zvalue = 2.68, p = 0.007), with an average of 13 taxa in agricultural RWs and an average of 20 taxa in livestock RWs (Fig. 3). Mean density also differed between land uses (F 1,37 = 31.26, p < 0.001), with agricultural RWs showing half the density of livestock RWs (Fig. 3). In addition, the density found during the dry period for the two land uses was half that recorded for the normal period (F 1,37 =34.40, p<0.001, Fig. 3). Diversity only showed signi cant differences between land uses in the dry period (F 1,37 = 5.48, p = 0.025), with higher values in livestock RWs than in agricultural RW (Online Resource: Table S5).

Discussion
Our results suggest that agricultural land use had a strong in uence on the physical and chemical parameters, macrophyte structure, and macroinvertebrate metrics of the Pampean lowland RWs studied. The physical and chemical characteristics found for agricultural RWs denote water quality degradation of these wetlands in comparison with livestock RWs according to the rst studies of the region (Tarda et al. In accordance with our results, changes in macrophyte species resulting from agricultural land use were also recorded for USA wetlands (Gustafson and Wang 2002). Submerged macrophyte were absent and oating macrophyte coverage increased in agricultural lands, as documented in the context of increased nutrient load ( Feeding strategies of macroinvertebrates could also re ect the adaptation of species to stressors and form part of a uni ed measure across communities differing in taxonomic composition (Tomanova et al. 2006). The effect of agricultural land use was re ected in a greater relative abundance of scrapers and predators in RWs. The increase in the frequency of scraper feeding habits is expected to occur in nutrient-  (Gosselain et al. 2005). In this sense, the land use and hydrological period effects observed in the macrophyte assemblages could also contribute to the differences in the proportion of FFGs.
Much of the current knowledge about the ecological response of macroinvertebrates to droughts is related to the effects of largely predictable seasonal droughts rather than supra-seasonal events (Lake 2003). Species inhabiting temporal ecosystems that suffer predictable droughts have different resistance mechanisms, including tolerance to the deterioration of water quality conditions and the presence of desiccation-resistant life history stages (Bogan et al. 2017). Also, present resilience mechanisms as dispersal to rewetted habitats from refugia (Boulton and Lake 2008). On the other hand, faunal recovery from supra-seasonal droughts varies from one case to another (Lake 2003). Generally, species are vulnerable to these events as they are not adapted and cannot escape the disturbance events in time (Boulton 2003). The unusual absence of precipitation, the high evapotranspiration in summer, and the loss of connection with the groundwater in agricultural areas (Rodrigues Capítulo et al. 2020) led the Carnaval RW to be dry in the February campaign. The lower oxygen concentration in RWs surrounded by agricultural land indicated a higher effect of the dry period on the water characteristics of these RWs. According to Robinson et al. (2004), reduced ow commonly leads to decreases in dissolved-oxygen content, which is critical to the survival of many aquatic species and can also affect their distribution and abundance. This condition was associated with the lower macroinvertebrate diversity, the wide difference in taxa richness, and the higher dissimilarity in macroinvertebrate composition found in agricultural RWs in comparison with livestock RWs during the dry period.
In line with our ndings on the RWs studied, it is known that natural low ows cause decreases in invertebrate densities (Wood and Armitage 2004). Some authors suggested that this decrease occurs in response to changes in competition and predation because habitat area decreases and food quality and quantity are altered by ow reduction (Cowx et al. 1984;Wood et al. 2000). Furthermore, invertebrate community composition often changes in response to low or reduced ow in streams (Gore et al. 2001;Suren et al. 2003) and wetlands (Sim et al. 2013). The increase in the density of particular taxa associated with the dry period in macroinvertebrate assemblages was similar to that reported by other authors. Larned et al. (2007) also found Nematoda and Copepoda to be resistant to desiccation.
Desiccation-resistant stages are well-known in copepods (Dahms, 1995), and aquatic nematodes are believed to survive extended dry periods in a state of anhydrobiosis (Drummond et al. 2015). Moreover, Ceratopogonidae larvae appear almost immediately when surface ows resume, suggesting that they use the hyporheic zone as a refuge from surface drying (Stanley et al. 1994;Stubbington 2012). Conversely, H. curvispina and Caenis sp. decrease their abundance in the dry period. There are no previous reports on changes in the abundances of these taxa under drought conditions. However, other related species have been studied. Ladle and Bass (1981)  The results showed a synergetic effects of nutrient loading and hydrological disturbances in the RWs studied. Similar results were mentioned by Green et al. (2017) for the Doñana wetlands in Spain, a more complex system of wetlands than the studied. As mention these authors, decisions from informed policy makers can promote ecosystem resilience to global threats through local measures. The correct management of agricultural activities in the land surrounding a river, as part of integrated watershed management, is of paramount importance to the conservation of associated wetland water quality (Wang 2001). Therefore, unsustainable agricultural practices could be replaced with environmentally-friendly, ecological agriculture to preserve wetland ecosystems (Zou et al. 2018).

Conclusions
The ndings provide strong evidence of the importance that the surrounding landscape has in the environmental characteristics and macrophyte and macroinvertebrate assemblages of lowland RWs. We conclude that the effects of intensive agricultural land use on physico-chemical variables, mainly nutrient enrichment, contributed to the differences found in macrophyte life form coverage and macroinvertebrate assemblages. The differences between land uses were greater during a drought period, which indicates a higher sensitivity to different hydrological conditions in those wetlands surrounded by intensive agricultural systems in comparison with those surrounding by extensive livestock considered as reference for the region.
Despite the land use, the great macrophyte coverage in the RWs studied indicates the importance of these systems in the basin and the value of their conservation. However, the differences found in macroinvertebrate FFGs could also be in uenced by the composition of macrophyte assemblage, indicating the role and importance of this assemblage in lowland RWs. In this sense, the combined use of macrophytes and macroinvertebrate assemblages were a powerful tool for describing and assessing the studied riverine wetlands. The metrics evaluated were useful for studying the land use of the catchment and the hydrological conditions. This study provides valuable information to future conservation and management projects and to the scarce knowledge of this type of wetlands.    Table 1 for abbreviations.

Figure 3
Mean and standard error of macroinvertebrate taxa richness, density and Shannon diversity in the agricultural and livestock RWs in the two hydrological periods.