Effects of landscape factors and hydroperiod on aquatic macroinvertebrates with different dispersal strategies in southern Brazil ponds

Abstract

Ponds represent discrete sites surrounded by a terrestrial landscape that is inhospitable for aquatic macroinvertebrates. Such ecosystems are appropriate for testing the influence of area, isolation, and correlated variables such as habitat diversity and hydroperiod, on aquatic macroinvertebrate communities. We tested three hypotheses in natural and non-impacted ponds located in a matrix of coastal dunes without human interference in southern Brazil: (1) area, habitat diversity, and hydroperiod increase the total macroinvertebrate richness and density, while isolation among ponds decreases the total macroinvertebrate richness and density; (2) the influence of environmental factors depends on the dispersal strategies of macroinvertebrates; and (3) macroinvertebrate composition is determined by the interaction of area, habitat diversity, hydroperiod, and isolation and differs among macroinvertebrates with different dispersal strategies. Fourteen ponds were sampled four times from October 2007 to August 2008 in the Lagoa do Peixe National Park coastline. Our results indicated that the pond area influenced only flying macroinvertebrate richness. The isolation and habitat diversity did not determine the richness, density, and composition of macroinvertebrates (total, flying, and non-flying). The richness, density, and composition of pond macroinvertebrates were determined mainly by hydroperiod. The mosaic created by the variation in pond hydroperiod would provide for a greater number of taxa with distinct dispersal capacities within the landscape. However, temporary ponds have disappeared at an alarming rate and there is a need to promote conservation of wetlands with different hydroperiods in southern Brazil.

Keywords:

• macroinvertebrate
• dispersal
• ponds
• species diversity
• hydroperiod
• habitat diversity

Introduction

The distribution of species in insular habitats, such as wetland systems, has been an important topic in modern ecology and conservation biology. Wetlands are important sites for biological conservation because they support rich biodiversity and have high productivity (Mitsch & Gosselink 2000). However, biodiversity in wetlands has been reduced throughout most of the world by agricultural, urban, and industrial development (Shine & Klemm 1999). The rapid degradation of wetlands and the insufficient status of scientific knowledge on patterns of species richness in these systems have led to the urgent need for ecological studies. However, most of these studies were developed in modified and fragmented wetlands inserted in a landscape changed by human activities. In this sense, the understanding of species composition and richness patterns in natural and non-impacted wetlands located in landscapes without direct human interference is just as important for wetland ecology and conservation planning.

Island biogeography and metapopulation theories have been used to characterize the spatial dynamics of communities in continental ecosystems such as forests and wetlands (Harris 1984; Hall et al. 2004). The theory of island biogeography (MacArthur & Wilson 1967) predicts that larger islands closer to the continent have higher species richness than smaller and more isolated ones. Increased richness with increasing area may be related to a reduced extinction rate (Tracy & George 1992). Another explanation is related to the greater capacity of colonization by individuals – an ecosystem's colonization potential increases with area (Ricklefs & Lovette 1999). On the other hand, the habitat diversity hypothesis states that the effect of area on species richness is solely through habitat diversity, and larger areas are more diverse because they contain more habitat types (Ricklefs & Lovette 1999). Moreover, increased isolation among ecosystems reduces immigration rate and increases the chance of species extinction (MacArthur & Wilson 1967). However, the influence of isolation on biological communities is directly associated with the organism's ability to reach colonizable habitats (Hanski 1999). In isolated environments, species with low dispersal ability will likely have lower chances of colonization and higher chances of extinction than species with high dispersal ability.

Freshwater habitats can be thought of as islands of water surrounded by land, and several studies have applied the equilibrium theory to freshwater organisms (Nilsson & Nilsson 1978; Oertli et al. 2002; Brose 2003). However, not all groups of organisms respond similarly to habitat size and isolation (Scheffer & van Geest 2006). In wetlands, there is a positive relationship between macroinvertebrate richness and area for specific taxonomic groups and for macroinvertebrates in general (Oertli et al. 2002; Sanderson et al. 2005; Stenert & Maltchik 2007; Studinski & Grubbs 2007). However, other studies have shown that this relationship was not significant for either Sphaeriidae and Coleoptera or the macroinvertebrate community (Oertli et al. 2002; Brose 2003; Hall et al. 2004; Batzer et al. 2004). These conflicting results show that there is not a consensus on the species-area relationship in wetland macroinvertebrates.

The effect of isolation on macroinvertebrate communities has received less attention. Wetland macroinvertebrates occur in habitats that represent discrete sites surrounded by an inhospitable terrestrial landscape. The impact of isolation on a macroinvertebrate community depends on the dispersal ability of the organisms (Morris 2012). Freshwater invertebrates achieve dispersal through a variety of mechanisms that can be broadly categorized as active, especially through aerial dispersion (adult insects), or passive, which may occur through transport by animal vectors or wind and often involves a specific desiccation-resistant stage in the life cycle (Bilton et al. 2001). Some studies found no significant relationship between aquatic and terrestrial invertebrate richness and habitat isolation (Brose 2003; Scheffer & van Geest 2006; Jonsson et al. 2009). However, other studies have shown that isolation is an important parameter for snail distribution (Brönmark 1985) and colonization pattern of predatory aquatic insects (Wilcox 2001).

The most often cited factor that directly and indirectly affects pond macroinvertebrate communities is hydroperiod (duration, timing, and predictability), because it influences other abiotic (dissolved oxygen, pH, nutrients, salinity) and biotic (primary production, detritus breakdown, predation, competition) factors that affect invertebrates (Wissinger 1999). Larger, permanent wetlands supported a higher richness than smaller, intermittent ones because many macroinvertebrate species are not adapted to tolerating or escaping the dry phase (Williams 1996; Wissinger 1999). Moreover, macroinvertebrate communities in intermittent wetlands are distinct from those in permanent ones, because they need to have morphological, life history or dispersal mechanisms to enable survival during the dry phase (Wiggins et al. 1980; Williams 1996; Tarr et al. 2005). On the other hand, Wissinger et al. (2009) did not find temporary-habitat specialists and the macroinvertebrate species found in temporary communities were a nested subset of those in permanent communities.

Wetlands are considered to be ecological islands in a terrestrial matrix, and such systems are appropriate to test the influence of area, isolation, and correlated variables such as habitat diversity and hydroperiod on species richness, density, and composition (Heino 2002; Oertli et al. 2002; Jones et al. 2003; Stenert & Maltchik 2007; Stenert et al. 2008). Most of these studies were motivated by an increase in wetland fragmentation resulting from human activities (Hall et al. 2004; Guadagnin & Maltchik 2007). However, studies analyzing the species distribution patterns in non-impacted wetlands located in landscapes without direct human interference are still rare. Since landscape factors may affect aquatic macroinvertebrate richness, density, and composition in fragmented wetlands (Stenert & Maltchik 2007; Stenert et al. 2008), and macroinvertebrates comprise a heterogeneous group of organisms with different dispersal strategies (Bilton et al. 2001), we tested three hypotheses in natural and non-impacted ponds in a coastal dune landscape in southern Brazil: (1) higher area, habitat diversity, and longer hydroperiod increase the total macroinvertebrate richness and density while isolation among ponds decreases the total macroinvertebrate richness and density; (2) the influence of environmental factors depends on the dispersal strategies of macroinvertebrates; and (3) macroinvertebrate composition is determined by the interaction of area, habitat diversity, hydroperiod and isolation and differs among macroinvertebrates with different dispersal strategies.

Methods

Study area

The coastal plain in southern Brazil, which covers approximately 37,000 km² along a stretch of 640 km of the Atlantic seacoast, is one of the regions with the highest concentration of wetlands in southern Brazil (Maltchik 2003). The study area is located in the Lagoa do Peixe National Park (LPNP), which is the only conservation unit in southern Brazil included in the Ramsar Convention (Figure 1). The conservation unit has an area of 344 km² (31° 02′ – 31° 48′ S; 50° 77′ – 51° 15′ W) with a maximum length of 62 km and a mean width of 6 km.

Figure 1. Wetlands sampled in the coastline of Lagoa do Peixe National Park (LPNP), Rio Grande do Sul State, Brazil.

The region has a flat topography and moist subtropical climate. The absence of hills and the low altitude (<20 m above sea level) of the study area make the climatic conditions (precipitation and temperature) similar throughout the study region (Rambo 2000). The annual precipitation ranges between 1150 and 1450 mm with an annual mean of 1250 mm (Tagliani 1995). The mean temperatures range between 14.6 °C in winter and 22.2 °C in summer, and the mean annual temperature is 17.5 °C (Klein 1998).

Sample collection

This study was performed at the LPNP coastline (Figure 1), which is characterized by small ponds distributed along a sand dune landscape with typical dune herbaceous vegetation. The ponds were identified through satellite images. Fourteen ponds were randomly selected from a total of 35 in the LPNP coastal dune region (distance less than 200 m from the sea). Four samplings were carried out from October 2007 to August 2008, and each sampling location was recorded with a global position system (GPS) receiver (GPS III Plus, Garmin, Inc., http://garmin.com). The variation of physical and chemical water parameters of each pond (salinity, conductivity, TDS, pH, temperature, dissolved oxygen, turbidity, nitrate, soluble reactive phosphorus, oxidation reduction potential) during the studied period was measured by a portable meter (U-22, Horiba, http://www.horiba.com). The water depth of each pond was measured with a graduated staff at each sampling occasion.

The area of each pond was measured during the largest flood period (winter) with a Quickbird image extracted from GIS software (Google Earth Pro^TM 4.2.1). Habitat diversity was quantified by counting the number of distinct habitats of dominant vegetation. Six habitat types were identified: (1) submerged vegetation (e.g., Myriophyllum aquaticum, Potamogeton spp.); (2) floating-leaved plants (e.g., Ludwigia peploides, Salvinia spp.); (3) stands of Azolla; (4) rush-like plants (emergent plants taller than 30 cm with stem-like leaves, such as Androtrichum trigynum and Juncus effusus); (5) leafy emergent plants (emergent plants taller than 30 cm with branched stems and leaves, such as grasses and sedges); and (6) low emergent plants (herbaceous plants shorter than 30 cm such as decumbent grasses, Hydrocotyle bonariensis, Xyris jupicai, and Eriocaulon modestum). The habitat diversity was the cumulative number of vegetation habitats in the pond systems. The minimum size to consider a habitat was approximately 10% of the total area of each wetland. The pond isolation was measured by the minimum distance (edge–edge) to the nearest pond in the landscape (permanent or intermittent). The hydroperiod length of each pond corresponded to the number of sampling events with surface water (1–4). Permanent ponds maintained surface water in all sampling periods and intermittent ones did not have surface water for at least one sampling period.

Macroinvertebrate sampling

At each sampling occasion, the macroinvertebrates were collected with a dip-net (30 cm wide, 250-μm mesh) by kicking up the substrate and then sweeping above the disturbed area to capture dislodged or escaping macroinvertebrates (Rosenberg et al. 1997). Three random sweeps of 1 m each were carried out in each identified habitat, encompassing approximately 1 m² area sampled per habitat. The total number of sweeps per pond system ranged from 3 to 18, depending on the number of habitats (from 1 to 6). The sampling effort varied to ensure a representative measure of total richness and composition of each pond (Gotelli & Colwell 2001). Sweeps were pooled into 0.5-L plastic buckets and preserved in situ with 10% formaldehyde. In the laboratory, the samples were washed through a 250-μm sieve to remove leaves, stems, and other debris and the resulting material was preserved with 80% ethanol. Macroinvertebrates were identified to the level of family, according to Lopretto and Tell (1995), Merritt and Cummins (1996) and Fernández and Domínguez (2001). Macroinvertebrates were classified into two groups according to dispersal form: (1) flying macroinvertebrates, which included only insect families and (2) non-flying macroinvertebrates, which included crustaceans, molluscs, worms, and mites.

Data analyses

The richness and abundance of macroinvertebrates per pond were determined as the total number of families and mean number of individuals collected over the year (four samplings) at each pond, respectively. The statistical analyses were performed using rarefied richness and density of macroinvertebrates. Macroinvertebrate richness was rarefied to the smallest number of individuals from a pond. Rarefaction analyses were performed separately for total macroinvertebrate, flying and non-flying richness using 1000 interactions by the vegan package on the program R^TM statistical program version 2.14.2 (RDCT 2012). The macroinvertebrate density was calculated as the ratio between macroinvertebrate abundance and number of habitats in each pond. In other words, the total abundance found in each pond was divided by the number of habitats present in the pond to minimize the variation in the sampling effort among sites (one to six habitats). This correction was performed for total, flying, and non-flying abundance.

The correlation between environmental variables was tested using the Pearson's correlation (area, habitat diversity, isolation, and hydroperiod). First, simple linear regressions were performed to manually select which environmental variables (area, habitat diversity, isolation, and hydroperiod) should be incorporated into the multiple regression model (p < 0.25). Then, the influence of environmental variables selected on rarefied total richness and total density was analyzed by multiple linear regression with backward elimination (cut-off point at p = 0.10). The influence of environmental factors was also assessed separately for the two macroinvertebrate groups (flying and non-flying). Pond area and density were square-root transformed before statistical analyses to ensure the normality of the data set and to give less weight to the few dominant macroinvertebrate families. The analyses were performed on program R version 2.14.2, using the vegan package (RDCT 2012). These analyses were used to test the first and second hypotheses of our study.

Non-metric multidimensional scaling (NMDS) analysis was used to evaluate the variation of the total macroinvertebrate, flying, and non-flying composition among the coastal dune ponds. The analysis was performed with the Bray–Curtis dissimilarity index using two axes by the vegan package in program R version 2.14.2 (RDCT 2012). Three ordination models were performed to represent the total, flying, and non-flying macroinvertebrates. The NMDS was used to graphically represent the similarity of macroinvertebrate composition in two dimensions, but this analysis was not constrained by environmental variables. Thus, environmental variables (area, habitat diversity, isolation, and hydroperiod) were adjusted to the ordination model by the ‘envfit’ function of the vegan package (Oksanen et al. 2009). The correlation significance between variables and ordination axes (p < 0.05) was obtained by 1000 permutations. The data standardization ‘Wisconsin double standardization’ was performed before the NMDS analysis. Although this standardization is not equivalent to the rarefied richness for univariate data, it allowed us to reduce the influence of high values of abundance. It is indicated as a previous transformation in NMDS analysis when there are high levels of abundance in certain sites compared to others. Subsequently, we performed an indicator species analysis to verify the variation of each taxa between permanent and intermittent ponds, using the labdsv package in program R version 2.14.2 (RDCT 2012). The significance of the observed maximum indicator value (I_NDV_AL) for taxa was derived from 1000 permutations. These analyses were used to test the third hypothesis of our study.

Results

The environmental variables of permanent and intermittent ponds are presented in Table 1. All ponds were considered freshwater systems since water salinity ranged from 0% to 0.1% over the study period. A total of 35,060 individuals distributed among 56 macroinvertebrate taxa were collected during the study period (Table 2). While flying macroinvertebrates (aquatic insects) were represented by 14,657 individuals and 42 families, non-flying macroinvertebrates were represented by 20,403 individuals from 14 taxa (Table 2). The most abundant insect family was Chironomidae, which comprised 53.1% of the flying macroinvertebrates collected. The majority of the insect families comprised less than 1% of the flying macroinvertebrates collected (Table 2). Dogielinotidae and Hydrobiidae were the most abundant non-flying macroinvertebrate families and represented 59.7% and 19.4% of the individuals sampled in this group, respectively (Table 2). However, Hydrobiidae occurred only in one pond over the study period (Table 2).

Table 1. Physical and chemical variables (mean ± standard deviation) of 14 ponds studied during the four seasons in southern Brazil. The pond 13 was measured only in spring.

Pond	Depth (cm)	Temperature (°C)	EC (mS/cm)	pH	DO (mg/L)	Turbidity (NTU)	Salinity (%)	TDS (g/L)	ORP (mV)
1	11	19.4 ± 4.6	19.5 ± 0.9	6.4 ± 0.1	7.9 ± 2.3	18.5 ± 7.5	0.007 ± 0.006	0.12 ± 0.04	226.4 ± 111
2	51.4	20.4 ± 4.0	12.9 ± 0.8	6.4 ± 0.2	8.7 ± 0.7	21.6 ± 22.6	0.001 ±0.003	0.08 ± 0.01	280.6 ± 77.3
3	45.2	21.2 ± 3.4	11.4 ± 0.8	6.4 ± 0.2	9.6 ± 1.7	6.1 ± 8.7	0.000 ± 0.000	0.08 ± 0.01	266.6 ± 85.2
4	26.7	21.4 ± 2.6	12.2 ± 2.8	6.8 ± 0.3	9.9 ± 0.7	10.0 ± 6.1	0.003 ± 0.003	0.08 ± 0.02	298.2 ± 41.0
5	20	20.4 ± 2.6	13.8 ± 3.9	7.0 ± 0.3	10.4 ± 1.0	24.0 ± 28.3	0.005 ± 0.007	0.09 ± 0.03	321.5 ± 59.4
6	40.4	22.1 ± 3.1	10.9 ± 1.7	6.5 ± 0.2	9.5 ± 0.7	63.4 ± 70.3	0.001 ± 0.002	0.07 ± 0.01	270.2 ± 61.1
7	24.1	21.4 ± 2.7	8.6 ± 6.6	6.1 ± 0.2	9.3 ± 1.4	63.5 ± 17.1	0.006 ± 0.005	0.08 ± 0.03	215.3 ± 88.4
8	33.9	21.5 ± 3.3	18.0 ± 2.2	6.0 ± 0.2	8.3 ± 1.0	40.7 ± 38.8	0.010 ± 0.000	0.12 ± 0.02	203.1 ± 114
9	50	22.3 ± 5.3	16.3 ± 3.7	6.0 ± 0.2	8.6 ± 1.3	64.2 ± 109	0.006 ± 0.005	0.11 ± 0.03	221.8 ± 92.1
10	25	22.1 ± 5.9	14.7 ± 6.5	6.1 ± 0.3	8.4 ± 1.6	76.4 ± 77.9	0.005 ± 0.006	0.09 ± 0.03	240.9 ± 42.8
11	31.7	21.4 ± 4.8	14.4 ± 3.6	6.2 ± 0.1	7.9 ± 0.6	142 ± 228	0.007 ± 0.006	0.09 ± 0.02	197 ± 103
12	70	23.7 ± 5.4	12.9 ± 1.4	6.4 ± 0.1	9.9 ± 1.7	12.9 ± 7.9	0.002 ± 0.005	0.07 ± 0.02	251 ± 100
13	22.7	21.1	0.3	6.5	9.9	1	0.185	2.15	274.5
14	35	23.9 ± 5.9	22.6 ± 17.5	6.7 ± 0.6	10.5 ± 1.6	23.4 ± 35.5	0.013 ± 0.005	0.17 ± 0.07	256.1 ± 102

Note: EC = electrical conductivity; DO = dissolved oxygen; TDS = total dissolved solids, and ORP = oxidation reduction potential.

Table 2. Total abundance, relative abundance (%), and frequency of occurrence (%) of macroinvertebrate taxa in ponds sampled in the coastline of Lagoa do Peixe National Park (LPNP), Rio Grande do Sul State, Brazil.

		Taxa			Ponds (%)
Class	Order	Non-flying macroinvertebrates	Total abundance	Relative abundance (%)	Permanent	Intermittent	Total
Malacostraca	Amphipoda	Dogielinotidae	12,186	59.7	100	80	92.9
	Isopoda	Janiridae	2	0.01	11.1	20	14.3
		Sphaeromatidae	1	0.00004	11.1	0	7.1
Gastropoda	Mesogastropoda	Hydrobiidae	3964	19.4	11.1	0	7.1
		Ampullaridae	646	3.2	55.6	80	64.3
	Pulmonata	Planorbidae	258	1.3	100	80	92.9
		Physidae	102	0.5	33.3	0	21.4
Arachnida	Acariformes	Acarina	1188	5.8	100	80	92.9
		Hydracarina	170	0.8	77.8	40	64.3
Hirudinea	Rhynchobdellida	Glossiphoniidae	678	3.3	100	60	85.7
	Arhynchobdellida	Hirudinidae	177	0.9	88.9	40	71.4
Turbellaria	Tricladida	Dugesiidae	555	2.7	77.8	40	64.3
Oligochaeta	Tubificida		282	1.4	88.9	80	85.7
Bivalvia	Veneroida	Sphaeriidae	194	1	33.3	40	35.7
		Total	20,403	100
		Flying macroinvertebrates
Insecta	Diptera	Chironomidae	7785	53.1	100	100	100
		Ephydridae	1113	7.6	44.4	20	35.7
		Tabanidae	230	1.6	55.6	80	64.3
		Culicidae	153	1	55.6	0	35.7
		Ceratopogonidae	49	0.3	66.7	40	57.1
		Tipulidae	38	0.3	55.6	60	57.1
		Thaumaleidae	20	0.1	22.2	20	21.4
		Stratiomyidae	13	0.1	33.3	0	21.4
		Dolichopodidae	9	0.1	11.1	20	14.3
		Sciomyzidae	2	0.01	11.1	20	14.3
	Trichoptera	Hydroptilidae	1456	9.9	100	60	85.7
		Polycentropodidae	168	1.1	44.4	0	28.6
		Leptoceridae	13	0.1	55.6	0	35.7
	Ephemeroptera	Caenidae	726	5	100	60	85.7
		Baetidae	375	2.6	100	60	85.7
	Odonata	Libellulidae	425	2.9	100	60	85.7
		Coenagrionidae	312	2.1	100	60	85.7
		Aeshnidae	46	0.3	66.7	40	57.1
		Lestidae	5	0.03	11.1	0	7.1
		Gomphidae	2	0.01	11.1	0	7.1
	Coleoptera	Dytiscidae	385	2.6	100	60	85.7
		Hydrophilidae	292	2.0	100	80	92.9
		Noteridae	160	1.1	77.8	60	71.4
		Staphilinidae	69	0.5	22.2	40	28.6
		Curcullionidae	68	0.5	88.9	60	78.6
		Elmidae	26	0.2	33.3	20	28.6
		Haliplidae	23	0.2	33.3	0	21.4
		Scirtidae	4	0.03	22.2	20	21.4
		Dryopidae	2	0.01	11.1	0	7.1
		Limnichidae	1	0.01	11.1	0	7.1
		Scarabaeidae	1	0.01	0	20	7.1
	Hemiptera	Pleidae	204	1.4	33.3	0	21.4
		Belostomatidae	92	0.6	77.8	40	64.3
		Corixidae	87	0.6	100	60	85.7
		Naucoridae	60	0.4	66.7	20	50
		Mesovellidae	50	0.3	33.3	20	28.6
		Veliidae	33	0.2	55.6	40	50
		Gelastocoridae	13	0.1	44.4	20	35.7
		Nepidae	6	0.04	33.3	0	21.4
		Notonectidae	6	0.04	33.3	0	21.4
		Hebridae	3	0.02	11.1	0	7.1
	Lepidoptera	Pyralidae	132	0.9	100	40	78.6
		Total	14,657	100

Table 3. Pond area, habitat diversity, isolation, hydroperiod, and macroinvertebrate (total, flying and non-flying) richness and abundance in permanent and intermittent ponds sampled in the coastline of Lagoa do Peixe National Park (LPNP), Rio Grande do Sul State, Brazil, over the annual seasons.

						Macroinvertebrate richness						Macroinvertebrate abundance
	Season	Pond	Area (m^2)	Habitat diversity	Isolation (m)	Total	Rarefied	Flying	Rarefied	Non-flying	Rarefied	Total	Flying	Non-flying
Permanent
	Su, A, W, Sp	2	1680	4	104	38	16.9	27	12.3	11	2.3	8518	1712	6806
	Su, A, W, Sp	3	795	2	74	38	21.2	32	15.7	6	1.6	3162	2661	501
	Su, A, W, Sp	6	175	2	56	30	14.2	21	8.1	9	2.4	5293	3958	1335
	Su, A, W, Sp	7	700	2	25	35	15.3	23	9.5	12	1.8	3078	1141	1937
	Su, A, W, Sp	8	180	2	25	33	20.9	25	15.7	8	2.0	1293	609	684
	Su, A, W, Sp	9	120	2	30	21	14.2	14	9.3	7	1.4	1176	475	701
	Su, A, W, Sp	10	1078	2	30	33	13.4	23	12.7	10	1.2	3425	856	2569
	Su, A, W, Sp	12	228	1	400	27	14.6	18	12.4	9	1.9	3013	535	2478
	Su, A, W, Sp	14	1800	1	50	26	11.8	18	18.0	8	2.1	1076	85	991
Intermittent
	Sp, Su, W	1	152	2	423	20	9.9	15	4.9	5	2.3	1214	1120	94
	Sp, A, W	4	1200	2	140	30	19.4	21	11.1	9	2.5	963	818	145
	Sp, W	5	240	2	85	8	7.9	6	4.1	2	2.0	151	147	4
	Sp, Su, W	11	2431	3	15	32	17.9	21	15.6	11	1.5	2551	410	2141
	Sp	13	1650	1	59	7	7.0	2	1.9	5	2.7	147	130	17

Note: Sp = spring (October 2007); Su = summer (December 2007); A = autumn (May 2008); W = winter (August 2008).

The values for pond area, habitat diversity, isolation, and hydroperiod in permanent and intermittent ponds are presented in Table 3. The total richness ranged from 7 to 38 families per wetland over the study period (Table 3). While flying macroinvertebrate richness ranged from 2 to 32 families, non-flying richness ranged from 2 to 12 families in the ponds (Table 3). The total macroinvertebrate density varied from 147 to 8518 individuals per pond. While the flying density ranged from 85 to 3958 individuals per pond, non-flying density ranged from 4 to 6806 individuals per pond (Table 3). None of the environmental variables studied were correlated: area and habitat diversity (r = 0.271, p = 0.349), area and hydroperiod (r = −0.227, p = 0.436), area and isolation (r = −0.361, p = 0.205), habitat diversity and hydroperiod (r = 0.209, p = 0.473), habitat diversity and isolation (r = −0.211, p = 0.470), and hydroperiod and isolation (r = −0.036, p = 0.902).

The total macroinvertebrate richness was positively related to hydroperiod (r²_adj = 0.317, F_1,12 = 7.043, p = 0.021). The flying macroinvertebrate richness was positively related to the area and hydroperiod (r²_adj = 0.674, F_2,11 = 14.5, p < 0.001). However, non-flying macroinvertebrate richness was not related to any variable (all models with p > 0.05). Wetland isolation and habitat diversity were not associated with total, flying, and non-flying macroinvertebrate richness (p > 0.05).

Total and non-flying macroinvertebrate densities were positively related to hydroperiod (r²_adj = 0.317, F_1,12 = 5.957, p = 0.031 and r²_adj = 0.354, F_1,12 = 8.131, p = 0.014, respectively). Flying macroinvertebrate density was not associated with the environmental variables (all models with p > 0.05). Pond area, isolation, and habitat diversity were not associated with total, flying, and non-flying macroinvertebrate densities (p > 0.05). These results are related to the first and second hypotheses of our study. While the first hypothesis was partially corroborated, the second was accepted since the influence of environmental factors on richness and density varied between the distinct dispersal groups of macroinvertebrates.

The dissimilarity of the macroinvertebrate composition (Figure 2(a)), flying macroinvertebrate community (Figure 2(b)), and non-flying macroinvertebrate community (Figure 2(c)) were each represented by two axes in the multidimensional scaling. The gradient represented by the first NMDS axis separated permanent ponds from intermittent ones and the total, flying, and non-flying macroinvertebrate composition variations were related to hydroperiod (r² = 0.764, p = 0.004; r² = 0.822, p = 0.001; r² = 0.489, p = 0.037, respectively) (Figure 2). Wetland area, isolation and habitat diversity were not associated with total, flying, and non-flying macroinvertebrate composition variations. According to the species indicator analysis, Glossiphoniidae (I_NDV_AL = 0.9507, p = 0.003), Acarina (I_NDV_AL = 0.9225, p = 0.003), Corixidae (I_NDV_AL = 0.9011, p = 0.047), Libellulidae (I_NDV_AL = 0.8647, p = 0.005), Dytiscidae (I_NDV_AL = 0.8515, p = 0.036), Hydrophilidae (I_NDV_AL = 0.8122, p = 0.048), and Coenagrionidae (I_NDV_AL = 0.7534, p = 0.022) were associated with permanent ponds. No macroinvertebrate taxa were characteristic of intermittent ponds (p > 0.05). These results are related to the third hypothesis of our study, showing that the macroinvertebrate composition of total, flying, and non-flying macroinvertebrates was determined by pond hydroperiod.

Figure 2. (a) Total macroinvertebrate dissimilarity of permanent (squares) and intermittent (circles) ponds, (b) flying macroinvertebrate dissimilarity of permanent (squares) and intermittent (circles) ponds, and (c) non-flying macroinvertebrate dissimilarity of permanent (squares) and intermittent (circles) ponds.

Discussion

Hydroperiod was positively related to total and flying macroinvertebrate richness and total and non-flying densities. Several studies suggest that species richness declines as hydroperiod decreases (Schneider & Frost 1996; Williams 1996, 2006; Wissinger 1999; Whiles & Goldowitz 2005). The surface water reduction may be one of the main causes of lower richness and abundance of wetland macroinvertebrates because the dry phase imposes such rigorous environmental conditions that only a limited number of species can survive in them (Wiggins et al. 1980; Williams 1996; Wissinger 1999). However, the lack of relationship between non-flying macroinvertebrate richness and hydroperiod could be explained by their strategies for enduring or avoiding periods of desiccation since non-flying macroinvertebrates are resident organisms that, for the most part, have drought-resistant life stages that usually have no autonomous means of dispersal (Wiggins et al. 1980; Hall et al. 2004).

Macroinvertebrate composition also differed between permanent and intermittent ponds. Tarr et al. (2005), Collinson et al. (1995), and Sanderson et al. (2005) showed that macroinvertebrate taxa differed markedly along the hydroperiod gradient, with distinct genera dominating wetlands with varying hydroperiods. Furthermore, the species indicator analysis showed that some predator macroinvertebrate taxa such as Libellulidae, Coenagrionidae, Dytiscidae, Acarina, and Glossiphoniidae were observed mainly in ponds with longer hydroperiods in our study. Whiles and Goldowitz (2005) stated that larger predatory macroinvertebrate taxa become abundant in wetlands with longer hydroperiods. This group consists mainly of aquatic insects that have limited or no drought-resistant life stages and are capable of active movement between wetlands during their life cycle, usually as flying adults (Wiggins et al. 1980). Another study found a distinct macroinvertebrate fauna composition between permanent and intermittent wetlands in southern Brazil, and predator insect families such as Gyrinidae, Belostomatidae, Naucoridae, Notonectidae, Lestidae, Aeshnidae, Tabanidae, and Ceratopogonidae were associated with permanent wetlands (Stenert & Maltchik 2007).

However, no macroinvertebrate taxa were characteristic of studied intermittent ponds, indicating the absence of a macroinvertebrate composition typical for this type of hydroperiod. This lack of unique taxa in the intermittent ponds goes against the classical literature of temporary waters (Wiggins et al. 1980; Williams 1996, 2006). On the other hand, a recent study developed in New Zealand showed that invertebrate species in temporary communities are a nested subset of those in permanent communities with the absence of temporary-habitat specialists (Wissinger et al. 2009). Some explanations for this result were that the hydroperiod of the studied temporary habitats in New Zealand is seasonally unpredictable and that phylogenetic constraints on the fauna have favored generalists or restricted the evolution of specialist invertebrate taxa. Stenert and Maltchik (2007) also did not find typical macroinvertebrate taxa associated with intermittent wetlands. In the current study, the absence of a macroinvertebrate composition typical for intermittent ponds suggests that the taxa inhabiting intermittent ponds are a subgroup of those in permanent ones, reflecting a nested pattern of community composition. Moreover, this result may be partly interpreted in the light of the inherent biases associated with the taxonomic resolution limited to higher level such as family or order. However, family level is often adequate for revealing important characteristics of the natural ecosystems and community patterns (Furse et al. 1984; Heino & Soininen 2007; Melo & Hepp 2008).

Our study showed that pond area influenced only the richness of flying macroinvertebrates. Some studies also documented a significant relationship between insect communities and habitat size (Roth & Jackson 1987; Larson & House 1990). This result confirms that pond size may influence macroinvertebrate groups in different ways, and it may be associated with strategies for enduring or avoiding periods of desiccation (Wiggins et al. 1980). Insects are organisms that have limited or no drought-resistant life stages but they are capable of active dispersal between wetlands during their life cycle, usually as flying adults (Wiggins et al. 1980; Hall et al. 2004). Dispersal can be triggered by environmental conditions. For example, increased temperatures and falling water levels trigger dispersal in Heteroptera and Coleoptera (Bilton et al. 2001). Nevertheless, invertebrate species from most of the major groups, including insects, occur in small and temporary freshwater systems (Williams 2006). The intermittent ponds of North America are characterized by both crustaceans and insects (e.g., coleopterans, hemipterans, midges, mosquitoes, caddisflies) (Wiggins et al. 1980). In this sense, certain insect species specialized on small temporary wetlands because of increased food resources and lack of competition in these temporary systems.

The lack of relationship between pond area and total and non-flying macroinvertebrate richness may be related to the reduced size of the study ponds, narrow pond area range (0.01–0.24 ha) and short hydroperiod of some ponds. Species richness varies independent of area in small fragments (Lomolino & Weiser 2001) and several studies, including a study in southern Brazil (Stenert & Maltchik 2007), have not supported the relationship between richness and area in intermittent wetlands (Lake et al. 1989; Schneider & Frost 1996; Batzer et al. 2004). The ephemeral nature of intermittent wetlands may hinder the equilibrium between colonization and extinction (Wiggins et al. 1980; Lake et al. 1989; Hall et al. 2004).

Isolation is another biogeographic attribute that is associated with species richness. Isolation (distance to nearest pond) did not influence the richness, density, and composition of total, flying, and non-flying macroinvertebrate communities. The lack of relationship between richness and isolation was also verified in other studies (e.g., Brose 2003; Scheffer & van Geest 2006; Jonsson et al. 2009). This discrepancy may be related to the distinct dispersal capacities of macroinvertebrates and the scale of active dispersal. Active dispersal is mostly limited to insects, and wingless taxa disperse when small dormant eggs or cysts are carried by wind, water, or animal vectors (Bilton et al. 2001; Morris 2012). The capacity for flight is highly variable. While some insect taxa, such as beetles, dragonflies, and true bugs, are good flyers and may travel hundreds or thousands of kilometers (Williams 1957; Feng et al. 2006; Wikelski et al. 2006), shorter dispersal distances (less than 1 km) have been reported in adult mosquitoes (Service 1993; Bilton et al. 2001). In our study, isolation varied from 15 to 423 m, and this distance gradient may have not been enough to recognize the limiting effects of isolation on macroinvertebrate communities, since flying macroinvertebrates represented the majority of macroinvertebrate taxa observed. Moreover, although winged taxa are able to disperse through active flight, dispersal is often facilitated by wind (Dingle 1972). On the other hand, passively dispersed taxa, such as molluscs, have dormant life stages with adaptations to facilitate vector mediated dispersal (Morris 2012). Also, an aspect that should be considered is that the nearby ponds may be playing a role as potential source populations.

The peculiar result about the distribution pattern of Hydrobiidae in our study, where 3964 individuals occurred in only one pond, can be related to the recruitment failure of the species in the studied enclosed ponds. Some studies with a Hydrobia species (H. ulvae) show that its rare populations in small enclosed lagoons appear to suffer prolonged recruitment failure; the hypothesis that it cannot be successful in these habitats because of heavy larval mortality seems plausible. This recruitment failure would place it at a disadvantage in lagoons and prevent it from dominating mixed Hydrobia populations, and its life history strategy would restrict it to open systems. Some apparently suitable lagoon habitats lack hydrobiids or possess only declining populations of H. ulvae (Fenchel & Kofoed 1976; Barnes & Gandolfi 1998; Barnes 1999). The species found in this pond could be suffering the same effect (recruitment failure) in the environment studied.

Pond area only influenced flying macroinvertebrate richness. Isolation and habitat diversity did not determine macroinvertebrate richness, density, and composition (total, flying and non-flying). Our study indicated that the richness, density, and composition of pond macroinvertebrates were determined mainly by hydroperiod. The surface water reduction in these ponds decreased macroinvertebrate richness and density, modifying their composition. The mosaic created by the variation in pond hydroperiod would provide for a greater number of taxa with distinct dispersal capacities within the landscape. In this sense, our results increased the knowledge of the macroinvertebrate community patterns in southern Brazil ponds, particularly located in protected and unmodified landscapes. However, temporary ponds have disappeared at an alarming rate because these shallow wetlands are extremely sensitive to pressure from human activities (Wissinger 1999). Therefore, it is necessary to promote the conservation of wetlands with different hydroperiods in southern Brazil, where more than 90% of wetland systems have already been lost and the remaining ones are still at high risk due to agricultural expansion.

Additional information

Funding

This work was supported by grants from Universidade do Vale do Rio dos Sinos – UNISINOS [grant number 02.00.023/00-0]; Conselho Nacional de Desenvolvimento Científico e Tecnológico – CNPq [grant number 52370695.2]. Leonardo Maltchik holds a Brazilian Research Council – CNPq Research Productivity [grant number 3009861/2010-6].