Should we plant macrophytes? Restored habitat use by the fish community of Lake Apopka, Florida

ABSTRACT

Slagle ZJ, Allen MS. 2018. Should we plant macrophytes? Restored habitat use by the fish community of Lake Apopka, Florida. Lake Reserv Manage. 00:00–00.

Freshwater resources are impaired worldwide, and managers frequently use habitat restoration to mitigate anthropogenic disturbances to freshwater systems. Restoration attempts have not often been evaluated with respect to their benefits to the sportfish population, especially in lentic systems. Recent restoration techniques at Lake Apopka, Florida, have included planting of macrophytes, particularly spatterdock (Nuphar advena), to increase fish habitat, stabilize sediments, and sequester nutrients. We assessed fish community habitat use of recently restored habitat with comparison to both unrestored (non-vegetated) and natural vegetated littoral habitats. We compared relative abundance and species richness of the fish community across habitat treatments. We electrofished >30 transects in each of these 3 habitat treatments (i.e., natural, planted, and unrestored), collecting relative abundance data for all fishes and habitat data including vegetation species and percent area coverage (PAC). Natural and planted habitat held greater PAC of vegetation than unrestored sites. Natural habitat supported the greatest relative abundance (as indexed by catch per unit effort) and the greatest species richness of fishes, while planted habitats and unrestored habitats supported fewer relative numbers and lower richness. Planted habitats are being utilized by the fish community, but were not significantly different from unrestored sites. Increased diversity of macrophytes in planted habitats could increase the benefits for these populations. Managers should consider continued monitoring and evaluation of planted sites to increase our understanding of the benefits of habitat restoration.

KEYWORDS:

• Fish community
• habitat restoration
• habitat use
• Lake Apopka
• lake restoration
• macrophyte
• planting

Aquatic habitat restoration is growing in importance as the scientific community recognizes the impacts of anthropogenic habitat degradation. Impaired lake ecosystems are common across the globe due to factors such as excessive nutrient inputs, acidification, and water level stabilization (NRC 1992, Cooke et al. 2005). Habitat restoration efforts generally seek to reestablish important habitats, processes, and communities that have been negatively impacted by eutrophication and other threats. Often, lake managers seek to restore benefits from recreational freshwater fisheries (Lee and Jones 1991). These efforts require evaluation to inform future work, increase the beneficial effects from restoration projects, and wisely spend limited agency funding.

Many previous evaluations of fishery responses to habitat restoration have focused on lotic environments and salmonid habitat (reviewed by Roni et al. 2008; e.g., Mouton et al. 2012, Thomas et al. 2015, van Zyll de Jong and Cowx 2016), while fewer restoration projects on lentic systems have been evaluated. In Florida, some restoration monitoring efforts have lacked sufficient data to assess whether the restoration was successful (Dierberg et al. 1988, Dierberg and Williams 1989). Other researchers in Florida have evaluated fishery responses to drawdowns, dredging, macrophyte removal, and herbicide treatments (Colle and Shireman 1980, Moyer et al. 1995, Allen et al. 2003, Bonvechio and Bonvechio 2006, Jordan and Arrington 2014, Dotson et al. 2015), while macrophyte planting as a restoration technique has not been previously assessed. However, the Florida Fish and Wildlife Conservation Commission (FWC) has planted macrophytes in at least 13 lakes beginning in the 1980s (Dotson et al. 2015). Managers may plant macrophytes with the expectation that planting will increase fish habitat or population abundance, but these effects are unknown.

Macrophytes provide numerous benefits for aquatic ecosystems, and reduced macrophyte coverage reduces these benefits. Plants provide habitat for fish, invertebrates, and zooplankton (Savino and Stein 1982, Carpenter and Lodge 1986, Dibble et al. 1996, Petr 2000). Macrophyte beds act as wave breaks, limiting sediment resuspension and increasing water clarity (Kristensen et al. 1992, Gasith and Hoyer 1998, Burks et al. 2002, Van Donk and Van de Bund 2002). Macrophytes also sequester nutrients, thereby limiting phytoplankton growth (Blindow 1992, Van Donk and Van de Bund 2002, Phillips et al. 2016). Macrophytes can also improve aesthetics, reduce nuisance plant and algae growth, and generally moderate environmental disturbance (NRC 1992, Cooke et al. 2005).

In areas where macrophytes are reduced in abundance or eliminated entirely, managers may choose to plant macrophytes. However, evaluation of macrophyte planting restoration efforts is rare, and very few references in the literature exist (Cooke et al. 2005, Strakosh et al. 2009)—especially in regard to the fish community use of restored habitat in comparison with natural and unrestored habitats. Managers also do not always consider what types of habitat different species of macrophytes provide; macrophyte species differ in structure (e.g., emergent vs. submerged, size of interstitial spaces) and therefore provide different benefits for different species and sizes of fishes (Gotceitas and Colgan 1990, Killgore et al. 1993, Stahr and Shoup 2015, Kuehne et al. 2016).

We assessed fish community habitat use of recent habitat restoration efforts with comparison to both unrestored and natural littoral habitats in a large degraded lake in Florida. We compared relative abundance and species richness across habitat treatments. We tested the hypothesis that planted habitats would support richer communities and greater numbers of fish than unrestored habitats, but expected naturally vegetated habitats to be most diverse and hold the most fish.

Methods

Study site

Lake Apopka is one of the major impaired water bodies in Florida. The lake is one of the largest in Florida at greater than 12,000 ha and is fed by a natural spring, rainfall, and storm water runoff (Fig. 1). During the early 1900s, Lake Apopka was a premier fishing destination (Schneider and Little 1973, Reddy and Graetz 1987, Schelske and Brezonik 1992). Anthropogenic nutrient additions into the lake, including from citrus groves, muck farms, and the Winter Garden sewage treatment plant, increased during the same time period (Clugston 1963, Schneider and Little 1973, Schelske et al. 2005). These increased nutrient discharges combined with alterations to water flow led to decreased water quality, greatly decreased macrophyte coverage, declining recreational fisheries, and the creation of thick muck layers that cover the bottom of the lake (Schneider and Little 1973, Reddy and Graetz 1987, Schelske and Brezonik 1992). These problems were first documented in 1947, and various restoration efforts began in the late 1970s (Schneider and Little 1973, Schelske et al. 2005).

Figure 1. Map of Lake Apopka, with locations of each transect origin. Natural sites were primarily located near Gourd Neck Springs, where most of the sand bottom in the lake is found.

A range of management strategies has been attempted to improve water quality and fish habitat in Lake Apopka including a marsh flow-way (Lowe et al. 1992, Coveney et al. 2002), macrophyte planting (Hoge et al. 2003), removal of undesirable fish (Schaus et al. 2010), and reduction of external nutrient inputs (Coveney et al. 2005). Results of current in-lake methods to improve water quality are thus far ambiguous; researchers are divided as to whether the current management strategies are reducing muck and improving water quality (Bachmann et al. 1999, Canfield et al. 2000, Lowe et al. 2001, Coveney et al. 2005, Hoyer et al. 2012). Macrophyte planting in Apopka has taken place since 1991 (most recently in 2012–2014) with the objectives of expanding fish and wildlife habitat and improving water quality (Hoge et al. 2003; Jaggers B, FWC, biologist, Jul 2015, pers. comm.). These plantings have shown survival and expansion in some cases in Lake Apopka (Coveney 2016), but it is unknown whether benefits to fish communities have occurred.

Site classification

We identified 3 different habitat treatments (i.e., natural, planted, or unrestored) based on several criteria (Fig. 2). Natural sites represented bottom substrate and plant species that typify the makeup of the littoral zone prior to the degradation of the lake; these species include giant bulrush (Schoenoplectus californicus), maidencane (Panicum hemitomon), and eelgrass (Vallisneria americana; Clugston 1963, Schneider and Little 1973). Planted sites were areas where the FWC had planted rows of spatterdock (Nuphar advena), identified by PVC pipes marking the beginning and end of each planted row and by visual inspection with FWC agents (Jaggers B, FWC, biologist, Jul 2015, pers. comm.). Unrestored sites substrate consisted primarily of organic detritus and were often unvegetated or were sparsely vegetated by cattail (Typha spp.; Hoge et al. 2003).

Figure 2. Examples of the 3 different habitat types sampled in this study, including (a) unrestored, (b) planted, and (c) natural. Unrestored sites either lacked vegetation or were dominated by cattail, while spatterdock dominated planted site vegetation. Natural site macrophytes were diverse and included giant bulrush, maidencane, and eelgrass.

We classified natural sites using a combination of sediment and flora. We sampled sediment using a transparent PVC pipe attached to a one-way valve that allows pressure to bleed off and a vertical sample of the substrate to be extracted (Davis and Steinman 1998). As we retrieved the sample, we inserted a rubber stopper into the bottom of the transparent PVC pipe (prior to removal from the water) to prevent substrate from escaping. In these samples, we identified the substrate as either sand (indicating a hard bottom similar to the original lake bottom) or muck and detritus (which makes up the majority of the lake bottom and limits a variety of plant life). We classified sites with sand as the dominant substrate as natural sites, while other substrates were classified as either planted or unrestored. We only classified a site as natural if the site had both sand bottom and the presence of at least one of three natural macrophyte species (maidencane, giant bulrush, or eelgrass) in the immediate vicinity (i.e., within ∼10 m).

Field sampling

We used timed boat electrofishing to sample the 3 different habitats. We conducted approximately 10 electrofishing samples per treatment in 3 seasons (fall 2015, winter 2015–2016, and spring 2016) for at least 30 samples per treatment. At each site, we electrofished using a boat-mounted electrofishing unit (GPP 9.0; Smith-Root, Inc., Vancouver, WA) with 2 boom probes and 2 netters. We electrofished for 300s of actual shock time for each sample, allowing the netters to occasionally pulse the current on and off to more effectively capture fish. We determined proper voltage, pulse width, and rate, and adjusted according to local lake conditions, targeting 4.0 amps. We identified all fish to species; we measured sportfish (e.g., Florida bass [Micropterus floridanus], bluegill [Lepomis macrochirus], and black crappie [Pomoxis nigromaculatus]) for total length (TL) and enumerated all other fishes.

We collected habitat samples at the beginning and end of each transect to serve as habitat covariates. We estimated the coverage and composition of aquatic macrophytes by throwing a 0.25 m² PVC quadrat in 3 haphazard locations adjacent to the boat: once off the bow and once off each side. We then identified the top 3 macrophyte species in each quadrat based on coverage according to Hoyer et al. (1996). We also visually estimated the total percent area coverage (PAC) in each quadrat. We followed these procedures at the beginning and end of each transect, for 6 quadrat samples per transect. We also visually estimated the area coverage of aquatic vegetation across the entire transect by averaging 3 or 4 field crew members' individual estimates of PAC to the approximate width of the electrofishing field. We also took photos at the beginning and end of each transect for reference, each looking toward the transect.

Analysis

We used generalized linear models (GLMs) to assess the relative abundance of fish species and the PAC of macrophytes. We assumed that catch per unit effort (CPUE) and PAC were distributed following a compound Poisson–gamma distribution (i.e., Tweedie distribution; Tweedie 1984). We chose this approach because the high abundance of zero counts in our data precluded the use of parametric statistics (i.e., the data were not normally distributed). The Tweedie distribution has the advantage of incorporating zero and nonzero data into a single model framework, and has application to fisheries CPUE data with many zero counts (Candy 2004, Shono 2008). We estimated the Tweedie index parameter p for each model using Fourier inversion (Dunn and Smyth 2008). We modeled total fish relative abundance and species relative abundance with GLMs; we compared the fit of each with a null model and likelihood ratio tests (LRTs; Anderson 2008, Bolker 2008). We then tested for significant differences between habitats using Tukey's range tests for each model (Tukey 1949). We also followed these same methods to model macrophyte metrics (e.g., transect PAC, mean quadrat PAC, and Simpson's index). We used Program R (R Core Team 2014) and the R package cplm (Compound Poisson Linear Models) for analyses (Zhang 2013).

We used Simpson's index and species richness estimators (Chao1 and Jackknife1) to compare diversity of fishes and macrophytes across the 3 habitat treatments. Simpson's index (D) is a commonly used measure of diversity that emphasizes dominance of a species within an assemblage (Simpson 1949, Magurran 2004). Simpson's index is reported as [−ln(D)] following Rosenzweig (1995) and represents the likelihood that 2 random individuals are the same species. A sample with a higher Simpson's has greater diversity (less dominated by one species) than a sample with a lower Simpson's index. We additionally used the Chao1 estimator (Chao 1987) to estimate the minimum species richness by habitat using our fish abundance data; this reduces bias compared to observed species richness by additionally estimating undetected species (Shen et al. 2003, Chao et al. 2009, Chiu et al. 2014). We used the Jackknife1 estimator of species richness for macrophytes, as it only requires presence/absence data (Heltshe and Forrester 1983, Magurran 2004). We used the R package vegan (Oksanen 2013) for diversity analyses.

Results

The 3 habitat treatments differed substantially in macrophyte cover and composition. In randomly placed quadrat throws, we only observed spatterdock in planted habitats and cattail in unrestored habitats (with 2 exceptions: one observation of frogs bit [Limnobium spongia] in a planted sample and one in an unrestored sample). We observed 8 species in natural habitats, including giant bulrush, eelgrass, maidencane, pickerelweed (Pontederia cordata), Egyptian paspalidium (Paspalidium geminatum), and cattail. GLMs for transect PAC, quadrat PAC, and Simpson's index were highly significant (LRTs; P < 0.01) compared to null models (i.e., plant coverage or diversity does not vary by habitat treatment). Natural habitats also had a greater mean Simpson's index, although natural and planted habitats were only marginally significantly different from each other (Tukey's range test; P = 0.054; Fig. 3). Planted and natural habitats had similar PAC and Simpson's index and were significantly different from unrestored habitats based on Tukey's range tests for transect PAC and quadrat PAC (P < 0.01; Fig. 3). Macrophyte species richness (Jackknife1 estimator ± SE) was greatest in natural habitats (8.97 ± 0.97) and similar across planted and unrestored habitats (2.97 ± 0.97 for both).

Figure 3. Mean of estimated plant coverage (PAC) in transect samples, quadrat samples (top panel), and Simpson's index (bottom panel) for macrophyte species encountered in random quadrat throws by habitat generated by a generalized linear model with the Tweedie distribution (Zhang 2013). Error bars depict the 95% confidence interval from the GLMs and different letters indicate significant differences between groups based on Tukey's range tests.

Natural habitat samples had greater relative numbers of fish compared to planted and unrestored habitat samples (Table 1). The median catch per unit effort (CPUE ± coefficient of variation, CV) for all fishes in natural habitats was 66.1 (±1.4) fish per hour, compared to 14.3 (±1.2) fish per hour in planted habitats and 8.1 (±1.3) fish per hour in unrestored habitats (Table 1). Additionally, natural habitats held much greater relative numbers of each sportfish compared to the other habitats (Table 1). These trends held true for the GLMs as well (Table 2). All models were highly significant compared to null models using likelihood ratio tests (P < 0.01). Mean fish-per-hour for all species (from GLMs) was significantly greater in natural habitat samples, while planted and unrestored habitats had similar relative abundances (Fig. 4). Mean relative abundance of each sportfish was greatest in natural habitats and lowest in both restored and unrestored habitats according to the individual species GLMs and Tukey's range tests.

Table 1. Median (±CV) fish caught per hour by habitat treatment and species. We failed to catch sportfish in many samples across habitats, but tended to catch more fish per hour in the natural treatment samples.

	Unrestored	Planted	Natural
Samples (n)	37	34	32
All species	8.1 (±1.3)	14.3 (±1.2)	66.1 (±1.4)
Largemouth bass	0 (±4.6)	0 (±2.5)	3.4 (±0.8)
Bluegill	0 (±2.9)	1.2 (±3.9)	30.4 (±0.8)
Black crappie	0 (±2.0)	1.2 (±1.1)	2.4 (±1.2)

Table 2. Mean (±95% C.I.) fish caught per hour estimated from generalized linear models by habitat treatment and species. Each habitat model was significant compared to a null model (CPUE being equal among all treatments) according to likelihood ratio tests.

	Unrestored	Planted	Natural
All species	20.98 (13.37–32.90)	25.07 (15.85–39.67)	89.57 (59.97–133.77)
Largemouth bass	0.25 (0.07–0.90)	0.83 (0.37–1.85)	3.70 (2.37–5.78)
Bluegill	4.62 (1.72–12.45)	7.17 (2.80–18.39)	32.82 (16.26–66.28)
Black crappie	1.72 (1.01–2.93)	2.34 (1.42–3.86)	4.87 (3.27–7.26)

Figure 4. Estimated mean relative abundance (CPUE) of all fish, bluegill, Florida bass, and black crappie divided by habitat treatment, generated by a generalized linear model with the Tweedie distribution (Zhang 2013). Error bars depict the 95% confidence interval from the GLMs and different letters indicate significant differences between groups based on Tukey's range tests.

Fish species richness was greatest in natural habitats, lower in planted habitats, and lowest in unrestored habitats, although species diversity did not follow the same trends. Mean (±SE) estimated minimum species richness (Chao1 estimator) was 25.6 (±3.4) species for natural habitats, 22.6 (±3.4) in planted habitats, and 17.8 (±1.8) in unrestored habitats (Table 3). Observed species richness followed these trends as well, but sample sizes (in this case, numbers of fish in each habitat) were vastly different and observed species richness is therefore biased. Simpson's indices for fish were similar across all habitats (Table 3), although unrestored habitats were slightly higher.

Table 3. Observed species richness, estimated minimum richness (Chao1 estimator), and estimated diversity (Simpson's index) for fish samples by habitat type. Species richness generally increased from unrestored to natural habitats, while diversity was slightly greater in unrestored habitats.

	Richness	Chao1	Simpson's index
	(n observed)	(mean ± SE)	(mean ± SE)
Unrestored	17	17.8 (±1.8)	0.57 (±0.04)
Planted	20	22.6 (±3.4)	0.47 (±0.07)
Natural	23	25.6 (±3.4)	0.45 (±0.09)

Discussion

We found natural habitats in Lake Apopka to contain the greatest relative abundance of fish and the greatest richness, while planted (i.e., restored) habitats and unrestored habitats generally held lower abundances and diversities. We also found that natural habitats were more diverse in flora. Our results were similar to Johnson and Crumpton (1998), who found greater biomass of fishes in “grass” and “bulrush” sites compared to “cattail” or “no vegetation” sites in Lake Apopka. Other studies have found greater diversity and abundance of fish in systems with a diverse macrophyte community compared to cattail-dominated systems (Gifford 2002). Cattail-dominated systems are additionally associated with decreased macroinvertebrate biomass and density compared to diverse communities (Schummer et al. 2012, Lawrence et al. 2016).

Differences in catchability between habitats could have affected our relative abundances, but our findings are robust to this possibility. Catchability (i.e., the proportion of fish caught out of the population in a specific area) differs across macrophyte coverages (Bayley and Austen 2002), and unrestored habitats had both lower fish abundance and lower macrophyte coverage compared to the other sampled habitats (Figs. 3 and 4). However, catchability of Florida bass and bluegill was actually greater in transects of lower macrophyte coverage (Bayley and Austen 2002), and we would expect to more easily catch these species in more open environments such as the unrestored habitat samples. Catchability of bluegill and Florida bass in Florida is relatively constant across a range of vegetation densities and seasons (Hangsleben et al. 2013), and fish abundance is typically greater in vegetated areas of lakes (Dibble et al. 1996). We attempted to control for catchability by using identical methods across all 3 habitats (e.g., 2 netters, similar speed and shock intensity, similar transect depths). We also sampled across 3 seasons, sampling within a short time period within seasons. We therefore believe that CPUE was a valid index of relative abundance in our study.

This study also suffers from spatial overlap and isolation across different treatments. Natural site samples were by necessity taken along the south shore of Lake Apopka (Fig. 1), the only area of the lake that consistently supports diverse flora; unrestored and restored samples were taken on the east and northeast shores of the lake, where restoration efforts were focused. We cannot say conclusively that differences in macrophytes are the sole driver of fish community richness and abundance, because natural sites differ in other physiochemical and biological properties that could affect the fish community (e.g., sand substrate, vs. muck and detritus for both unrestored and restored sites). Macrophytes' interactions with the water column create differences in dissolved oxygen, pH, and temperature from surrounding unvegetated areas (Killgore et al. 1993, Petr 2000). Fish respond to these changes differently depending on plant density and fish density (Miranda and Hodges 2000, Dunham et al. 2002, Wilson et al. 2015). We did not monitor water quality parameters across treatments. We observed broadly similar water depths across all 3 treatments, but we did not measure water depths in this study. Macrophyte habitat is one of many factors that drive fish habitat selection. Future studies should measure water quality parameters and other habitat characteristics to better elucidate differences between natural, restored, and unrestored sites.

Habitat may be limiting in Lake Apopka for the Florida bass population. Low levels of aquatic macrophytes may limit sportfish populations in larger lakes (Wiley et al. 1984, Hoyer and Canfield 1996a, 1996b). In particular, juvenile bass mortality could be increased through limited refugia (Canfield and Hoyer 1992, Killgore et al. 1993), and restoration of aquatic vegetation in reservoirs has been shown to increase densities of age-0 largemouth bass (Micropterus salmoides) in some systems (Strakosh et al. 2009, Stahr and Shoup 2015), while loss of macrophytes is associated with decreased juvenile abundance (Parsons et al. 2009). While response of the sportfish population will always be difficult and time consuming to measure in a system as large as Lake Apopka (Rogers and Allen 2012), continued habitat restoration efforts may increase the benefits for native sportfish populations by increasing refugia for juvenile sportfish.

Managers should consider planting mixed communities of macrophytes in addition to (or in combination with) established planted sites. Loss of plant biodiversity can cause large, cascading changes in the functionality of an ecosystem (Knops et al. 1999). We found that a diverse flora (i.e., more than one species) supported greater abundance and richness of fish species compared to the current planted monocultures of spatterdock. Mixed macrophyte communities also have greater resistance to invasions of undesirable and invasive species (Stenberg et al. 1997, Hoge et al. 2003), and planting mixed macrophyte communities could serve to extend the benefits from restoration measures such as drawdowns (where long-term benefits can be limited by rapid invasion of cattail; e.g., Moyer et al. 1995). Spatterdock is also a floating or emergent type of macrophyte; submersed macrophytes can provide varied benefits for fish communities (Gotceitas and Colgan 1990, Killgore et al. 1993, Stahr and Shoup 2015, Kuehne et al. 2016). The planting of mixed stands of macrophytes could have greater, longer-lasting benefits for the fish and invertebrate community compared to the status quo of planting spatterdock-only stands in Lake Apopka. Any additional plantings should also be monitored by management for survival and possible improvements to sportfish habitat.

Restoration activities, particularly macrophyte plantings, suffer from a lack of long-term evaluation, and wider cooperation between scientists and management organizations is needed (Minns et al. 1996). Future studies should consider evaluation of diverse planting vs. monoculture planting for improvements in ecosystem benefits, including greater numbers and diversity of fishes, macroinvertebrates, and greater resilience to invasion. Future studies should also be aware of the varied and unpredictable responses of fish communities to restoration efforts (Pine et al. 2009). Existing monitoring data should also be used for evaluation of restoration efforts in Lake Apopka (Spurgeon et al. 2016), although no long-term evaluation of macrophyte planting currently exists for Lake Apopka. With continued experimentation and evaluation, the most effective restoration practices can be identified and used to further improve the sportfish population, both in Lake Apopka and wherever habitat restoration techniques are employed.

Acknowledgments

We thank FWC for project funding and K. Zarada, J. Procopio, Y. Quintana, G. Scholten, M. Hosabettu, and D. Dreier for field support.

Additional information

Funding

Florida Fish and Wildlife Conservation Commission [15054].